Dispensable Role of Aire in CD11c⁺ Conventional Dendritic Cells for Antigen Presentation and Shaping the Transcriptome

Deficiency for Aire results in the development of organ-specific autoimmune disease associated with chronic mucocutaneous candidiasis, called autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy (APECED) (OMIM 240300) (1–4). The mechanisms underlying the development of these immunological symptoms, however, have not been fully understood. Aire is predominantly expressed from medullary thymic epithelial cells (mTECs), and transcriptomic analyses using Aire-deficient mTECs demonstrated that Aire controls a wide variety of genes; additionally, much attention has been paid to the genes of tissue-restricted Ags (TRAs) because reduced expression of TRAs in Aire-deficient mTECs could result in the impaired elimination of autoreactive T cells and/or the impaired production of regulatory T cells in an Ag-specific manner (5–8). Besides mTECs, Aire is expressed from APCs in the periphery, named extrathymic Aire-expressing cells (eTACs), and their complementing role to the thymic tolerance mechanisms has been proposed (9). Interestingly, eTACs seem to be a heterogeneous population, and Aire-expressing group 3 innate lymphoid cell (ILC3)–like cells were found besides the conventional dendritic cell (cDC) type of eTACs (10). Indeed, a recent single-cell multiomics approach revealed that there are two major classes of eTACs: a migratory dendritic cell (DC) type of eTACs and eTACs expressing retinoic acid receptor–related orphan receptor γt (RORγt), with the latter corresponding to the Aire-expressing ILC3-like cells (10) and having been named Janus cells (11). The exact identity of Aire-expressing ILC3-like cells, however, is now under debate (12–14) (see Results). Recently, the essential role of Aire in the ILC3-like–type eTACs has been demonstrated for the efficient host defense against Candida albicans through eliciting the Th17-mediated immune response (15). Furthermore, ILC3-like–type eTACs have been demonstrated to have broad gene expression, including a range of TRAs, with homology to the gene signatures from mTECs (11). Despite the intense characterization of ILC3-like–type eTACs, the role of the cDC type of Aire-expressing cells (cDC-type eTACs) in the protection against C. albicans and the production of the transcriptome relevant to the peripheral tolerance remained unknown. Furthermore, phenotypic comparisons between the two types of eTACs using the conventional flow cytometric analysis and immunohistochemical analysis have not been performed.

To gain better insights into the phenotypes of eTACs and to address the role of Aire within them, we have used an Aire-reporter mouse strain in which endogenous Aire was replaced by the human AIRE and GFP-Flag tag (AGF hereafter) fusion protein (Aire/AGF-knockin [KI]) (16). Of note, this Aire-reporter molecule was produced as authentic Aire protein in the form of nuclear dots: this was in marked contrast to the other previously generated Aire-reporter mice in which GFP labeled the whole Aire-expressing cells (i.e., both the cytoplasm and nucleus) (11, 17). Accordingly, Aire/AGF-KI had overcome the molecular gap between the Aire protein (in the nuclei) and reporter molecule (GFP from the whole cells) seen in the conventional Aire-reporter mice by making GFP part of the Aire nuclear dots. This strain faithfully detected not only Aire-expressing mTECs but also eTACs both at a steady state and upon antigenic challenges. More importantly, this strain allows us to distinguish DCs committed to expressing Aire from Aire-nonexpressing DCs in both the presence and absence of functional Aire protein. This was because the AGF reporter molecule showed no function of Aire according to their disease phenotypes (16): mice harboring a combination of the Aire-null allele and AGF allele (i.e., −/AGF allele) on a NOD background showed a typical autoimmune attack against pancreatic acinar cells, a characteristic of Aire-deficient mice on a NOD background (18, 19). In contrast, heterozygous Aire/AGF-KI (+/KI) did not show any Aire-deficient phenotypes, indicating that the AGF reporter molecule had no dominant-negative effect on the wild-type (WT) allele. This allowed us to use heterozygous Aire/AGF-KI (+/KI) as Aire-sufficient mice. With this high-fidelity Aire-reporter strain, we have characterized two types of eTACs, that is, cDC-type and ILC3-like–type eTACs. In the current study, we particularly focused on the cell-intrinsic role of Aire in the cDC-type of eTACs in induction of the Th17 immune response against C. albicans and the production of transcriptome from them. Our studies have illuminated the distinct role of Aire in cDC-type eTACs and ILC3-like–type eTACs by demonstrating the dispensable role of Aire in the cDC-type eTACS.

Aire/AGF-KI mice (16) and Aire-knockout (KO) mice (20), both on C57BL/6 backgrounds, were generated by homologous recombination in embryonic stem cells as previously described. Mice were maintained under pathogen-free conditions and treated in accordance with the Guidelines for Animal Experimentation of Tokushima University School of Medicine.

The inguinal, axillary, and cervical lymph nodes (LNs) were digested with collagenase D (Roche)/DNase I (Roche) in RPMI 1640 medium containing 10% FBS, 1% nonessential amino acids solution (Life Technologies), 1% sodium pyruvate (Life Technologies), 20 mM HEPES (Life Technologies), 0.1 mg/ml penicillin-streptomycin mixed solution (Nacalai Tesque) and 50 μM 2-ME (Sigma-Aldrich), hereafter referred to as R10, at room temperature for 30 min. The suspension was homogenized by pipetting up and down to release the cells including the DCs. Then, LN cells were suspended in PBS containing 1% BSA and 5 mM EDTA and incubated with anti-CD16/32 (clone 96) mAb for Fc blocking followed by staining with mAbs against CD45 (clone 30-F11), CD11c (clone N418), CD11b (clone M1/70), CD8 (clone 53-6.7), and I-A/I-E (clone M5/114.15.2). mAbs against CD16/32, CD45, CD11c, CD11b, and CD8 were purchased from BioLegend. A mAb against I-A/I-E was purchased from eBioscience.

Bone marrow (BM) cells were prepared as previously described (21). In brief, BM cells from femurs of mice were harvested with R10 and RBCs were lysed in RBC lysis buffer by hypotonic shock. BM cells adjusted to 1 × 10⁶ cells/ml in the R10 were supplemented with 20 ng/ml recombinant murine GM-CSF (Miltenyi Biotec) and 10 ng/ml recombinant murine IL-4 (PeproTech). Then, 2 × 10⁶ BM cells per well in 12-well plates (Thermo Fisher Scientific) were cultured at 37°C in a 5% CO₂ environment for 60 h. Then, bone marrow-derived dendritic cells (BMDCs) were stained with the following mAbs for flow cytometric analysis. Cells were incubated with anti-CD16/32 (clone 96) mAb for Fc blocking followed by staining with B220 (clone RA3-6B2) and mAbs against CD11c (clone N418), CD11b (clone M1/70), CD24 (clone M1/69), and I-A/I-E (clone M5/114.15.2). B220 and anti-CD24 mAbs were purchased from BioLegend.

Flow cytometric analysis was performed using a FACSAria II (BD Biosciences) as described previously (17). For the staining of cell surface molecules, mAbs against CD80 (clone 16-10A1), CD86 (clone GL1), ICOS ligand (ICOS-L; clone HK5.3), CD40 (clone 3/23), CD127 (clone A7R34), PD-L1 (clone B7-H1), CCR7 (clone 4B12), CD200 (clone OX2), epithelial cell adhesion molecule 1 (EpCAM; clone G8.8), CD14 (clone Sa14-2), CD64 (clone X54-5/7.1), and Gr-1 (clone RB6-8C5) purchased from BioLegend were used. For intercellular staining, mAb against RORγt (clone Q31-378) purchased from BD Biosciences was used after fixation and permeabilization with an eBioscience Foxp3/transcription factor staining buffer set (eBioscience). AGF⁺ and AGF⁻ DCs isolated from LNs (LN-DCs) gated for CD45⁺CD11c⁺MHC class II (MHC-II)⁺ fraction were sorted by a FACSAria II. AGF⁺ BMDCs gated for the B220⁻CD24⁺CD11b^−/low fraction and AGF⁻ BMDCs gated for the B220⁻CD11c⁺MHC-II⁺ fraction were sorted by a FACSAria II. Sorted cells were examined with immunocytochemical analysis, a T cell proliferation assay, and RNA sequencing (RNA-seq). Data were analyzed using FlowJo 10.8.1 software (Tree Star).

Immunohistochemical staining was performed as described previously (22). The inguinal LNs from Aire/AGF-KI were harvested and embedded in OCT compound (Sakura Finetek) and frozen at −80°C. Cryosections (6 μm) prepared from frozen tissue were fixed with acetone. The samples were blocked with 1% BSA containing PBS at room temperature for 30 min followed by staining with anti-GFP polyclonal Ab, anti-CD11c (clone N418) mAb, and anti-RORγt (clone B2D) mAb at room temperature for 90 min. The samples were then washed with PBS five times and nuclei were stained with DAPI (Sigma-Aldrich). The samples were mounted with ProLong Gold mounting medium (Life Technologies). The polyclonal Ab against GFP and mAbs against CD11c and RORγt were purchased from Invitrogen, BioLegend, and eBioscience, respectively. The sections were examined for fluorescence microscopy analysis using a BZ-X810 all-in-one fluorescence microscope and BZ-X810 analyzer software (Keyence, Osaka, Japan).

FACS-sorted LN-DCs and BMDCs were attached to the glass slides and fixed with acetone after the cells dried. Then, the samples were blocked with 1% BSA containing PBS at room temperature for 30 min followed by staining with anti-GFP polyclonal Ab and anti-Aire mAb (clone 5H12) (Invitrogen). The samples were then washed with PBS five times and nuclei were stained with DAPI (Sigma-Aldrich). Samples were mounted with mounting medium.

CFSE labeling and a T cell proliferation assay were performed as described previously (23). Splenic CD4⁺ T cells from OT-II mice (OVA-specific TCR transgenic mice) (24) were collected with magnetic cell sorting (Miltenyi Biotec) followed by labeling with CFSE. Then, CSFE-labeled OT-II CD4⁺ T cells were cultured with FACS-sorted DCs at a 10:1 ratio in the presence of OVA protein for 72 h. CFSE dilutions in OT-II CD4⁺ T cells were assayed by a FACSAria II (BD Biosciences).

A 100 μg/ml C. albicans mycelial membrane protein (MP) fraction from C. albicans SC5314 (25) was prepared as previously described (26). Then, 200 μl of a 1:1 emulsion of C. albicans mycelial MP fraction and CFA (Candida/CFA) was injected into mice (s.c.). Alternatively, a 1:1 emulsion of PBS and CFA (PBS/CFA) was injected into the control mice. Mice immunized with Candida/CFA or PBS/CFA were sacrificed on day 6. CD4⁺ T cells from LN and spleen were collected with MACS CD4 MicroBeads (Miltenyi Biotec) according to the manufacturer’s instructions. AGF^+/− DCs and CD4⁺ T cells from Candida/CFA-injected mice were cultured with a 1:2 ratio in the presence or absence of a 2.5 μg/ml C. albicans mycelial MP fraction for 6 d. Then, samples were collected and restimulated with 500 ng/ml ionomycin (Sigma-Aldrich) and 50 ng/ml PMA (Sigma-Aldrich) in the presence of 10 μg/ml brefeldin A (Sigma-Aldrich) and inhibiting the cytokine secretion for 4 h at 37°C. Cells were suspended in PBS containing 1% BSA and 5 mM EDTA and incubated with anti-CD16/32 (clone 96) mAb for Fc blocking followed by staining with anti-CD4 mAb (GK1.5). Then, the cells were fixed and permeabilized with an eBioscience Foxp3/transcription factor staining buffer set. Intracellular cytokines were stained with anti–IFN-γ (XMG1.2) and anti–IL-17A (TC11-18H10.1). Flow cytometric analysis was performed using a FACSAria II (BD Biosciences).

C. albicans SC5314 was cultured on a YPD agar plate (1% yeast extract, 2% Bacto peptone, 2% glucose, and 1.5% agar) for 48 h at 37°C as previously described (26). We performed an intragastric injection of C. albicans (2 × 10⁸ cells/mice) for in vivo priming. For control mice, PBS was injected. Ten days after intragastric injection of C. albicans, CD4⁺ T cells from LN and spleen were collected with MACS CD4 MicroBeads (Miltenyi Biotec) according to the manufacturer’s instructions. AGF^+/− BMDCs and CD4⁺ T cells from C. albicans intragastrically injected mice were cocultured with a 1:2 ratio in the presence or absence of 1.25 × 10⁶ cells/ml of heat-killed C. albicans for 6 d in 96-well plates. Restimulation of CD4⁺ T cells and flow cytometric analysis were performed as described above.

Total RNA was extracted from FACS-sorted CD11c⁺AGF⁺ LN-DCs and CD11c⁻AGF⁺ LN-DCs on CD45⁺lineage⁻ (CD3, CD19, B220, and Gr-1) with ReliaPrep RNA miniprep systems (Promega) and converted to cDNA with a high-capacity cDNA reverse transcription (RT) kit (Thermo Fisher Scientific) according to the manufacturer’s protocols and guidelines. mRNA expression analysis was performed using a QuantiTect probe PCR kit (Qiagen) on a Thermal Cycler Dice real-time system II (Takara Bio). The following primers and dual-labeled probes (FAM/TAMRA) were used: Aire primers, forward, 5′-GGAGGATTCTCTTTAAGGACTACAA-3′, reverse, 5′-CTGGTTTAGGTCCACATCTTTTGG-3′, probe, 5′-TACAGCCGCCTGCATAGCATCCTGGA-3′; Rorc primers, forward, 5′-CTGCCCCATTGACCGAACC-3′, reverse, 5′-GCCAAACTTGACAGCATCTCG-3′, probe, 5′-AACCGATGCCAGCATTGCCGCCT-3′; Hprt primers, forward, 5′-TGAAGAGCTACTGTAATGATCAGTCAAC-3′, reverse, 5′-AGCAAGCTTGCAACCTTAACCA-3′, probe, 5′-GGGTTAAGCAGTACAGCCCCA-3′.

RNA-seq libraries were prepared by applying a random displacement amplification sequencing method for single-cell RNA-seq (27). Specifically, AGF⁺ and AGF⁻ DCs from LNs were sorted for the CD45⁺CD11c⁺MHC-II⁺ fraction by FACSAria II. AGF⁺ BMDCs in a B220⁻CD24⁺CD11b^−/low fraction and AGF⁻ BMDCs in a B220⁻CD11c⁺MHC-II⁺ fraction were also sorted into a tube containing lysis buffer (buffer TCL, Qiagen). Total RNAs from the cells were isolated using RNAClean XP beads (Beckman Coulter) directly from the lysis buffer and were eluted into 10 μl of water containing RNase inhibitor (4 U/μl RNasin Plus RNase inhibitor, Promega). The RNA was denatured by incubating at 65°C for 5 min, then placement on ice and treatment with DNase I (0.1 U/μl DNase I, amplification grade, Invitrogen) in 0.25× RT buffer (PrimeScript RT reagent kit, Takara Bio) at 30°C for 15 min. Then, cDNA was synthesized in 30 μl of 1× RT buffer containing RT enzyme mix (1.5 μl of PrimeScript RT enzyme mix I, Takara Bio), oligo(dT)-RT primer (0.2 μM Oligo(dT)18 primer, Thermo Scientific), T4 gene 32 protein (0.033 mg/ml, NEB), and first NSR (not-so-random) primer mix for mice (3.3 μM, Sigma-Aldrich custom DNA oligonucleotides) by incubating for 10 min at 25°C, 10 min at 30°C, 30 min at 37°C, 5 min at 50°C, and 5 min at 94°C. Next, the second strand was synthesized by adding 5 μl of 10× NEB buffer 2 (NEB), 1.25 μl of 2′-deoxynucleoside 5′-triphosphate mix (10 mM each nucleotide, NEB), 1.5 μl of Klenow fragment (3′→5′ exo⁻, NEB), 5 μl of second NSR primer for mice (100 μM, Sigma-Aldrich custom DNA oligonucleotides), and 7.25 μl of distilled water and incubating for 60 min at 16°C followed by 10 min at 70°C. The ds-cDNA was purified using the AMPure XP beads (Beckman Coulter) and quantified using Qubit (Qubit dsDNA HS assay kit, Invitrogen). Then, ds-cDNA <1 ng was applied for Tn5 tagmentation in a 25-μl reaction with a homemade Tn5 enzyme prepared as previously described (28, 29). DNA was purified using AMPure XP beads and amplified by 10–15 cycles of PCR using indexed primers and KAPA HiFi DNA polymerase (Roche). The libraries were purified using AMPure XP beads and pooled for next-generation sequencing on an Illumina HiSeq system.

After trimming low-quality reads and adapter sequences by fastp (30), short reads were mapped to mm10 by star (version 2.7.3a) (31) employing an Ensembl GTF file for mm10 downloaded from https://www.ensembl.org. Unmapped reads and reads mapped with a low-quality score (mapping quality <5) were removed using samtools (version 1.9) (32), and duplicated reads were removed by Picard MarkDuplicates (http://broadinstitute.github.io/picard). Mapped reads on each transcript were counted using htseq-count (version 0.11.2) (33) with -stranded = no option and an Ensembl GTF file for mm10. Raw read counts on transcripts were processed using R with the edgeR package (version 3.20.9) (34) for normalization and to identify differentially expressed genes. The FASTQ data of mTEC^high were extracted from the Gene Expression Omnibus database under accession number GSE222285 and processed by the same analytic pipeline as DC samples. TRAs were defined as genes expressed in less than three tissues. Heatmaps were visualized using R with the ComplexHeatmap package (version 2.12.0) (35). Gene Ontology (GO) analysis and gene set enrichment analysis of a gene set was performed using R with the clusterProfiler package (version 3.6.0) (36). The dataset associated with this project has been submitted to the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE211006 and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE222285).

Statistical comparisons within two groups were examined with a two-tailed Student t test. One-way ANOVA was used for more than three groups. A p value <0.05 was considered statistically significant.

To characterize the Aire-expressing cells in the periphery, eTACs, we used heterozygous Aire/AGF-KI (16). Enzymatic digestion of LNs showed the GFP signals from two distinct populations after gating for CD3⁻CD19⁻B220⁻Gr-1⁻ cells: major CD45⁺AGF⁺ hematopoietic cells and minor CD45⁻AGF⁺ stromal cells (Fig. 1A, lower left). When CD45⁺AGF⁺ cells were evaluated for CD11c expression, we observed a CD11c^high population (Fig. 1A, lower middle, squared in red) and a CD11c^−/low population (Fig. 1A, lower middle, squared in black). The signal intensity of GFP from the former was weaker compared with the latter. When the same CD45⁺AGF⁺ cells were evaluated for RORγt expression, we observed a major RORγt^{low to int} population (Fig. 1A, lower right, squared in red) and a minor RORγt^high population (Fig. 1A, lower right, squared in black): the signal intensity of GFP from the former was weaker than the latter. When the CD45⁺AGF⁺ cells were evaluated for their simultaneous expression of CD11c and RORγt, three major populations emerged: CD11c⁺RORγt⁻, CD11c⁻RORγt⁺, and CD11c⁻RORγt⁻ cells (Fig. 1B, middle). In contrast, CD45⁻AGF⁺ cells were mostly CD11c⁻RORγt⁻ (Fig. 1B, right). We focused on the CD45⁺ hematopoietic cells in this study.

CD11c⁺RORγt⁻ cells (∼60%) dominated CD11c⁻RORγt⁺ cells (∼10%) among the CD45⁺AGF⁺ cells in numbers (Fig. 1C), the ratio of which was within the range of the previous report: the ratio between the Aire⁺Rorc⁻ cells and Aire^highRorc⁺ cells revealed by the single-cell multiomics was ∼8:1 (11). The authors originally reported that both Aire⁺Rorc⁻ and Aire^highRorc⁺ cells had high transcriptional similarity to migratory DCs (11). However, their exact identity, especially for the latter, is now under debate because RORγt⁺ APCs have been reported to play an important role in the generation of commensal-specific peripheral regulatory T cells (12–14). Based on these circumstances, we tentatively named Aire-expressing CD11c⁺RORγt⁻ cells and Aire-expressing CD11c⁻RORγt⁺ cells in the LNs CD11c⁺ Aire-DCs and RORγt⁺ Aire-APCs, respectively. Real-time PCR analyses demonstrated that the expression level of Aire was ∼3-fold higher in RORγt⁺ Aire-APCs compared with CD11c⁺ Aire-DCs (Fig. 1D, left), consistent with the stronger GFP signals from RORγt⁺ Aire-APCs compared with those from the CD11c⁺ Aire-DCs (Fig. 1A). The expression level of Rorc was much higher in RORγt⁺ Aire-APCs compared with CD11c⁺ Aire-DCs, as expected (Fig. 1D, right).

Both CD11c⁺ Aire-DCs and RORγt⁺ Aire-APCs expressed AGF nuclear dots that showed the colocalization with endogenous Aire by immunocytochemical analysis (Fig. 1E), although the numbers of AGF nuclear dots were much fewer compared with those from mTECs (R. Miyazawa and Mitsuru Matsumoto, unpublished observation). Furthermore, we did not detect clear AGF dots in many CD11c⁺ Aire-DCs and in RORγt⁺ Aire-APCs probably due to the relatively weak Aire expression. Morphologically, we did not notice any clear difference between CD11c⁺ Aire-DCs and CD11c⁻ Aire-APCs (i.e., RORγt⁺ Aire-APCs) when stained with May–Giemsa, both showing a rather narrow cytoplasm typical for cDCs (Fig. 1E, rightmost). We also evaluated the location of Aire-expressing APCs within the LNs (LN-DCs or LN-APCs in short hereafter) by immunohistochemistry. CD11c⁺ Aire-DCs were located predominantly within the T cell areas whereas RORγt⁺ Aire-APCs existed mainly in the interfollicular spaces (Fig. 1F). Similar to the CD11c⁺ Aire-DCs in the LNs, Aire-expressing marginal zone DCs in the spleen were reported outside the B cell follicles (37). Although CD11c⁺ Aire-DCs dominated RORγt⁺ Aire-APCs in number (Fig. 1C), RORγt⁺ Aire-APCs in the interfollicular spaces looked more prominent compared with CD11c⁺ Aire-DCs in the T cell areas due to the stronger GFP signals (Fig. 1F). The location of CD11c⁺ Aire-DCs and RORγt⁺ Aire-APCs in LNs was not different between heterozygous (+/KI) and homozygous Aire/AGF-KI (KI/KI, in which functional Aire protein is absent) (16). The results suggested that Aire in APCs does not have a major impact on their location in a cell-autonomous fashion (Supplemental Fig. 1A).

Because CD11c⁺ Aire-DCs formed a major population of Aire-expressing cells in the LNs and RORγt⁺ Aire-APCs were characterized in the other studies (10, 12–15), we focused on the characterization of CD11c⁺ Aire-DCs. LN suspensions gated for the CD45⁺CD11c⁺MHC-II^high fraction contained the Aire⁺ DCs (corresponding to the CD11c⁺ Aire-DCs but not to RORγt⁺ Aire-APCs because of the CD11c⁺ gating strategy) (Fig. 2A, left). They expressed Sirpα at the intermediate level (Fig. 2A, right). To test which type of DCs (i.e., CD8⁺CD11b⁻ [cDC1s] or CD8⁻CD11b⁺ [cDC2s]) (38) express Aire, we divided the CD45⁺CD11c⁺MHC-II^high fraction into Aire⁺ (AGF⁺) and Aire⁻ (AGF⁻) DCs (Fig. 2B). Although both Aire⁺ and Aire⁻ DCs had more CD8⁻CD11b⁺ type than CD8⁺CD11b⁻ type, it was less evident in the Aire⁺ DCs (the ratio of CD8⁺CD11b⁻ versus CD8⁻CD11b⁺ was ∼1:3) compared with Aire⁻ DCs (which was an ∼1:7 ratio) (Fig. 2B). Thus, CD11c⁺ Aire-DCs were composed of both cDC1s and cDC2s and they were slightly skewed toward cDC1s when compared with Aire-nonexpressing DCs.

Because Aire deficiency confers high susceptibility to C. albicans infection of the mouth and esophagus, we compared the numbers of CD11c⁺ Aire-DCs in the cervical LNs with those in the LNs from other portions (i.e., inguinal and axillary LNs). We found that cervical LNs contained significantly higher percentages of CD11c⁺ Aire-DCs compared with LNs from the other portions (Fig. 2C, left). Higher frequencies of CD11c⁺ Aire-DCs in the cervical LNs than in other peripheral LNs remained constant over age, which was also observed in the absence of Aire (KI/KI) (Supplemental Fig. 1B). Cervical LNs contained a lower proportion of CD8⁺CD11b⁻ cDC1 type of CD11c⁺ Aire-DCs compared with other peripheral LNs (Fig. 2C, right).

Besides cDC1/cDC2 classification, DCs in the LNs can be divided into migratory DCs and resident DCs depending on their expression of the chemokine receptor CCR7 (39, 40). When LN-DCs from WT mice were stained with mAbs against MHC-II and CD11c, CCR7^high migratory DCs showed slightly higher expression of MHC-II with lower expression of CD11c compared with CCR7^low resident DCs (Fig. 2D). Conversely, CCR7^low resident DCs showed a slightly lower expression of MHC-II with higher expression of CD11c compared with CCR7^high migratory DCs. We examined to which type CD11c⁺ Aire-DCs correspond. We found that almost half of the CD11c⁺ Aire-DCs (gated for CD45⁺CD11c⁺MHC-II^high) showed CCR7^high expression (migratory type) and the remainder showed CCR7^low expression (resident type) (Fig. 2E, left), indicating that CD11c⁺ Aire-DCs did not solely belong to either type. Instead, CD11c⁺ Aire-DCs contained both migratory and resident types. Thus, all of the non-ILC3-like–type eTACs did not show a migratory phenotype in contrast to the nomenclature proposed based on the single-cell multiomics analysis (11). AGF⁻ DCs also contained both migratory and resident types at a similar ratio (Fig. 2E, right).

We then examined the expression of surface molecules including those important for Ag presentation. The expression level of CD86 was slightly higher in CD11c⁺ Aire-DCs (in red) compared with those from Aire⁻ DCs (in blue) (the same gating strategy for Aire⁻ DCs was used as in collecting the CD11c⁺ Aire-DCs except for AGF expression) (Fig. 2F, upper). Although less evident, the expression levels of CD40 and ICOS-L were also higher in CD11c⁺ Aire-DCs than in Aire⁻ DCs. Because the initial study reported several molecules unique to eTACs in humans (41), we also evaluated the expression of these surface molecules. Consistent with the previous reports in both humans (41) and mice (11), we observed higher expression of CD127, PD-L1, and CD200 from CD11c⁺ Aire-DCs compared with Aire⁻ DCs (Fig. 2F, lower). Interestingly, the expression levels of EpCAM were higher in CD11c⁺ Aire-DCs than in the Aire⁻ DCs, which could be one possible reason for the initial confusion about the eTACs as a stromal origin (9).

Knowing that DCs are the major component of eTACs, we tried to induce the BMDCs that express Aire using Aire/AGF-KI. We cultured BM cells isolated from the heterozygous Aire/AGF-KI supplemented with several cytokines to produce the AGF⁺ BMDCs. Among the several cytokines tested, we could induce AGF⁺CD103⁻ BMDCs most efficiently using the combination of GM-CSF plus IL-4 and/or GM-CSF plus IL-4 plus Flt3L (Supplemental Fig. 1C, upper). AGF signals from the BM cells appeared as early as 2 d after the culture, and it was almost the peak of the AGF expression: ∼5–8% of the cells expressed AGF on day 2, and AGF signals gradually declined on day 4 (∼2.5%) followed by day 6 (∼2.0%). Because the addition of Flt3L did not improve the efficiency of producing AGF⁺ BMDCs significantly, subsequent experiments were performed without Flt3L, that is, a combination of GM-CSF and IL-4 (Supplemental Fig. 1C).

We characterized Aire⁺ BM-DCs more in detail. Because cultured BMDCs can be classified into three major populations depending on their expression of CD11b and CD24 (CD11b⁻CD24^med/+, CD11b⁺CD24⁺, and CD11b⁺CD24^low/−) (42), we first examined which fraction contained the Aire⁺ BMDCs (Fig. 3A). Aire⁺ BMDCs were present almost exclusively in the CD11b⁻CD24^med/+ fraction. Aire⁺ BMDCs scarcely expressed CD11c, and their RORγt expression level was low to intermediate (i.e., CD11c^−/lowRORγt^{low to int}) (Fig. 3B). This phenotype of Aire⁺ BMDCs does not represent simply either CD11c⁺ Aire-DCs or RORγt⁺ Aire-APCs from LNs. Rather, Aire⁺ BM-DCs showed a hybrid of the characteristics representing the two types of Aire⁺ APCs in that they expressed neither CD11c nor RORγt at high levels (see (Fig. 1A). Importantly, however, note that Aire⁺ BMDCs are myeloid but not ILCs in light of the fact that they were expanded upon GM-CSF stimulation.

We then evaluated the expression of molecules involved in the Ag presentation from Aire⁺ BMDCs that appeared in the B220⁻CD11b⁻CD24^med/+ fraction from heterozygous Aire/AGF-KI (+/KI). Aire⁺ BMDCs homogenously expressed MHC-II at high levels (Fig. 3C, top in red). In contrast, Aire⁻ BM-DCs in the same fraction expressed MHC-II at variable levels ranging from negative to high, although MHC-II⁻ cells dominated the MHC-II⁺ cells (Fig. 3C, top in blue). Expression levels of CD80, CD86, ICOS-L, and CD40 were also higher in Aire⁺ BM-DCs compared with Aire⁻ BM-DCs. Of note, the expression levels of these molecules (i.e., CD80, CD86, ICOS-L, and CD40) were more evident even when compared with those from the primary Aire⁺ DCs in LNs (compare (Fig. 3C, top with (Fig. 2F, upper). Expression of CD127, CCR7, and CD200 was also higher in Aire⁺ BMDCs compared with Aire⁻ counterparts, although the expression levels of PD-L1 and EpCAM were indistinguishable between Aire⁺ and Aire⁻ BM DCs (Fig. 3C, middle). Expression of the markers for monocytes/macrophages (CD14, CD64, and Gr-1) was negative for the Aire⁺ BMDCs (Fig. 3C, bottom). Aire⁺ BMDCs expressed a single dot of AGF protein in most cases that were colocalized with endogenous Aire (Fig. 3D), although we did not detect clear AGF dots in many Aire⁺ BMDCs (R. Miyazawa and Mitsuru Matsumoto, unpublished observation). Morphologically, Aire⁺ BMDCs showed broader cytoplasm exhibiting a brighter Giemsa staining compared with Aire⁻ BMDCs (Fig. 3D, rightmost).

Although Aire⁺ BMDCs expressed higher levels of the molecules for Ag presentation compared with their Aire⁻ counterpart that appeared in the CD11b⁻CD24^med/+ fraction (Fig. 3A), we found that Aire expression was not a prerequisite for the expression of the molecules related to the Ag presentation. When the expression of these molecules (MHC-II, CD80, CD86, ICOS-L, and CD40) was assessed from the cultured AGF⁺ BMDCs established from homozygous Aire/AGF-KI (in which there is no functional Aire protein) (Fig. 3E, upper in blue), it was comparable to those from Aire-sufficient heterozygous Aire/AGF-KI (Fig. 3E, upper in red). Thus, Aire may not be required for promoting the ability of Ag presentation (see below). In contrast, expression of CD127 and CD200 was slightly higher in Aire-sufficient AGF⁺ BMDCs from heterozygous Aire/AGF-KI compared with that from Aire-deficient AGF⁺ BMDCs from homozygous Aire/AGF-KI (Fig. 3E, lower). Thus, expression of CD127 and CD200 might be a unique feature of Aire-expressing DCs as we have seen in LN-DCs (Fig. 2F, lower).

We then examined the Ag-presenting capacity of Aire⁺ DCs, that is, CD11c⁺ Aire-DCs and Aire⁺ BMDCs, using the TCR-transgenic system (24). For CD11c⁺ Aire-DCs, we mixed CFSE-labeled transgenic CD4⁺OT-II T cells specific for OVA peptide with AGF⁺CD11c⁺MHC-II^high cells isolated from the LNs (corresponding to the CD11c⁺ Aire-DCs) at a 10:1 ratio. When OVA protein was added into the culture (10 μg/ml), we observed the proliferation of OT-II T cells (Fig. 4A, uppermost), indicating that CD11c⁺ Aire-DCs have the capacity for Ag processing and presentation. Importantly, however, note that AGF⁻CD11c⁺MHC-II^high cells (the same gating strategy was used as in collecting the CD11c⁺ Aire-DCs except for AGF expression) also induced the proliferation of OT-II T cells in the presence of OVA to a similar degree (Fig. 4A, third upper). More importantly, when AGF⁺CD11c⁺MHC-II^high cells (corresponding to the CD11c⁺ Aire-DCs on the Aire-sufficient background) were isolated from the Aire-deficient background (−/KI) (called Aire^less-AGF cells hereafter), they could also induce the proliferation of OT-II T cells in the presence of OVA protein (Fig. 4A, second upper). As a negative control, we prepared the wells in which only the OT-II T cells and OVA protein were present without any APCs in the culture (Fig. 4a, bottom). The results indicated that although CD11c⁺ Aire-DCs function as APCs, Aire was not required for this action.

We also examined the Ag-presenting capacity of Aire⁺ BMDCs. We mixed CFSE-labeled transgenic OT-II T cells with AGF⁺ BMDCs at a 10:1 ratio. When OVA protein was added into the culture raging from 1 to 100 μg/ml, we observed the proliferation of OT-II T cells in a dose-dependent manner (Fig. 4B, top). However, CD11c^highMHC-II^high BMDCs showing no Aire expression also induced a comparable degree of proliferation of OT-II T cells in the presence of OVA (Fig. 4B, bottom). Similar to the case for CD11c⁺ Aire-DCs, AGF⁺ BMDCs lacking functional Aire protein established from homozygous Aire/AGF-KI (KI/KI) also induced the proliferation of OT-II T cells (Fig. 4B, middle). The results indicated that although Aire⁺ BMDCs function as APCs, Aire was not required for this action as we have seen in the CD11c⁺ Aire-DCs from the LNs (Fig. 4A).

It has been recently demonstrated that RORγt⁺ Aire-APCs play important roles in the protection against C. albicans through eliciting the Th17-mediated immune response, and Aire in RORγt⁺ Aire-APCs is indispensable for this action (15). We therefore asked whether Aire in CD11c⁺ Aire-DCs has a similar role. We first confirmed that Aire plays an important role in the host defense against C. albicans using the mouse model on a C57BL/6 background. When the mouths of the animals were swabbed with live C. albicans (26), Aire-KO showed higher clinical scores compared with WT mice (Supplemental Fig. 2A, 2B). CFU from the tongues (Supplemental Fig. 2C) and histological scores (Supplemental Fig. 2D, 2E) were also higher in Aire-KO compared with WT mice. The results indicated that Aire-KO on a C57BL/6 background is suitable for the assessment of the host defense against C. albicans. To address whether higher susceptibility to oral infection of C. albicans in Aire-KO involves any role of Aire in the CD11c⁺ Aire-DCs, mice were in vivo primed by an MP fraction of mycelial cells of C. albicans (26) with CFA injected s.c. Six days later, CD4⁺ T cells were isolated from the LNs and spleen. CD4⁺ T cells primed by s.c. injection of the MP fraction of mycelial cells were then cocultured with CD11c⁺ Aire-DCs from heterozygous Aire/AGF-KI. Significant induction of CD4⁺ Th17 cells was observed only in the presence of the recall Ag during the culture (Fig. 5A, 5B). Of note, when the primed CD4⁺ T cells were cocultured with CD11c⁺AGF⁺ DCs isolated from homozygous Aire/AGF-KI in which there is no functional Aire protein, we also observed similar induction of CD4⁺ Th17 cells (Fig. 5B, 5D). Thus, CD11c⁺ Aire-DCs could function as APCs in the induction of the Th17-mediated immune response against C. albicans, and Aire in CD11c⁺ Aire-DCs was not required for this action. This result was in great contrast to the role of Aire in RORγt⁺ Aire-APCs (15). Interestingly, we also observed the induction of CD4⁺ Th17 cells even when we used Aire-nonexpressing CD11c⁺AGF⁻ DCs from both heterozygous and homozygous Aire/AGF-KI (Supplemental Fig. 3A, 3B, 3D). The results suggested that the ability to induce C. albicans–specific Th17 immune response may not be confined to Aire⁺ DCs among cDCs.

We also examined the role of Aire in induction of the C. albicans–specific Th17 immune response using Aire⁺ BMDCs. In this case, mice were primed with an intragastric injection of live C. albicans (26). Ten days later, CD4⁺ T cells primed with live C. albicans were isolated from the LNs and spleen, and they were cocultured with Aire⁺ BMDCs in the presence of heat-killed C. albicans. We observed higher induction of CD4⁺ Th17 cells in the presence of recall Ag compared with no recall Ag during the culture on Aire-sufficient background (+/KI; (Fig. 6A, 6B). The Th17 response was observed only when mice were primed with live C. albicans, indicating that Aire⁺ BMDCs could elicit the C. albicans–specific adoptive Th17 immune response in this model. We then examined whether AGF⁺ BMDCs isolated from homozygous Aire/AGF-KI in which there is no functional Aire protein could induce a CD4⁺ Th17 response. We observed similar induction of CD4⁺ Th17 cells even when AGF⁺ BMDCs lacked functional Aire protein (KI/KI; (Fig. 6B, 6D). The results indicated that Aire in AGF⁺ BMDCs was not required for this action as we have seen in the experiments using CD11c⁺ Aire-DCs. In contrast, we did not see any significant induction of CD4⁺ Th1 cells against C. albicans in this protocol (Figs. 5C, 5E, 6C, 6E, Supplemental Fig. 3C, 3E).

The data so far indicated that CD11c⁺ Aire-DCs isolated from LNs as well as the cultured Aire⁺ BMDCs could function as efficient APCs. In contrast, we noticed that Aire itself was not required for these actions. Given that Aire plays an important role in shaping the transcriptome in mTECs, we next examined the role of Aire in the production of transcriptome within CD11c⁺ Aire-DCs. We isolated CD45⁺CD11c⁺MHC-II⁺ cells from the LN suspensions using Aire-sufficient (+/KI) and Aire-deficient mice (−/KI) and divided them into the AGF⁺ and AGF⁻ fractions. In this study, the AGF⁺ fraction from CD45⁺CD11c⁺MHC-II⁺ cells corresponded to CD11c⁺ Aire-DCs but not to RORγt⁺ Aire-APCs because we gated for CD11c⁺ cells for the isolation. We prepared Aire⁺ BMDCs on both Aire-sufficient (+/KI) and Aire-deficient backgrounds (−/KI) for investigating their transcriptomes (GSE211006). When the transcriptomes from CD11c⁺ Aire-DCs, Aire (AGF)⁻ DCs, Aire⁺ BMDCs, and mTEC^high on both Aire-sufficient and Aire-deficient backgrounds were compared, there was much difference between eTACs (CD11c⁺ Aire-DCs and Aire⁺ BM-DCs) and mTEC^high (Fig. 7A). Indeed, the transcriptome from Aire⁺ BMDCs was closer to that from the CD11c⁺ Aire-DCs compared with that from mTEC^high irrespective of the presence or absence of Aire by the principal component analysis (Supplemental Fig. 4A). We compared the expression levels of Aire between eTACs and mTEC^high and found that eTACs expressed Aire as low as one-seventh to one-eighth of mTECs by trimmed mean of M value (TMM) normalized counts (Fig. 7A, upper columns). Of note, we did not observe major changes in the transcriptome in DCs even in the absence of Aire unlike in the case of mTECs (Fig. 7A, lower heatmap).

When the transcriptome from AGF⁺ cells (from +/KI) and that from Aire^less-AGF cells (from −/KI) were compared in CD11c⁺ Aire-DCs, we found that Aire and Fn1 were upregulated whereas Dnmt3l was downregulated, and no other genes were significantly changed among 15,037 genes detected (fold change >2 or <0.5; false discovery rate [FDR] <0.05) (Fig. 7B, left). In this case, alteration in the expression of Aire and Dnmt3l was due to the gene-targeting event, and Fn1 was the only bona fide differentially expressed gene. Similarly, when we compared the transcriptome between Aire⁺ BMDCs and Aire^less-AGF BMDCs, Aire and Dnmt3l were the only genes that showed significant alteration (Fig. 7B, right). Thus, Aire deficiency did not affect the gene expression profiles in both CD11c⁺ Aire-DCs from LNs and cultured Aire⁺ BMDCs. When we applied Aire-dependent genes extracted from mTEC^high (WT versus Aire-KO, fold-change >2 and FDR <0.05) to the transcriptome of CD11c⁺ Aire-DCs, we observed no difference in their distribution from all the other genes (Fig. 7C, Supplemental Fig. 4B: Wilcoxon rank sum test p value = 0.1683 and χ² test p value = 0.2531, respectively). The results indicated that Aire-dependent genes in mTEC^high were not Aire-dependent in CD11c⁺ Aire-DCs.

To compare the role of Aire in gene control in CD11c⁺ Aire-DCs with that in mTECs more globally, we examined the correlation of Aire dependency defined by the fold change–fold change plot between CD11c⁺ Aire-DCs and mTEC^high. No significant correlation was found between CD11c⁺ Aire-DCs and mTECs in their gene expression induced by Aire (Fig. 7D, left). Because Aire in mTECs controls the transcriptome of TRAs for self-tolerance, we also focused on the expression of TRAs from CD11c⁺ Aire-DCs. We found that only a single TRA gene, Fn1, was significantly upregulated (fold-change >2 and FDR <0.05) in AGF⁺ cells compared with Aire^less-AGF cells for CD11c⁺ Aire-DCs (Fig. 7D, right). In contrast, the expression of TRAs from mTEC^high was dramatically up regulated (fold change >2 and FDR <0.05) in the presence of Aire (Fig. 7D, right), and there was no overlap with the TRA genes upregulated in CD11c⁺ Aire-DCs. Thus, Aire in CD11c⁺ Aire-DCs does not have a major impact on the production of the transcriptome including that for TRAs, unlike Aire in mTECs.

Given that loss of Aire in CD11c⁺ Aire-DCs showed no clear alterations in the Ag-presenting capacity for exogenous Ags (i.e., OVA and C. albicans) as well as in the production of the transcriptome, we investigated the identity of CD11c⁺ Aire-DCs by comparing the transcriptome between AGF⁺CD11c⁺ Aire-DCs and Aire (AGF)⁻ DCs. We observed 810 upregulated (fold change >2 and FDR <0.05) and 858 downregulated (fold change <0.5 and FDR <0.05) genes from CD11c⁺ Aire-DCs when compared with those from Aire⁻ DCs on an Aire-sufficient background (Fig. 7E). Rather unexpectedly, GO analysis of the upregulated genes in CD11c⁺ Aire-DCs showed the terms mostly related to the cellular development and general function rather than the immunological function (Supplemental Fig. 4C, left). Conversely, GO terms associated with immunological function were entirely enriched in Aire⁻ DCs (Supplemental Fig. 4C, right). Consistent with this finding, gene set enrichment analysis confirmed that genes belonging to the GO term “activation of immune response” were upregulated in Aire⁻ DCs (Supplemental Fig. 4D). The same transcriptomic changes between AGF⁺CD11c⁺ Aire-DCs and Aire⁻ DCs were observed even on an Aire-deficient background (Supplemental Fig. 4E). In contrast, expression levels of costimulatory molecules such as Cd80 and Cd86 were higher in AGF⁺CD11c⁺ Aire-DCs compared with Aire⁻ DCs (Fig. 7F), consistent with the results of flow cytometric analysis at least for CD86 (Fig. 2F, upper). Furthermore, the expression of costimulatory molecules was higher in Aire⁺ BMDCs compared with Aire⁻ BMDCs, consistent with the flow cytometric analysis (Fig. 3C). The results suggested that Aire-DCs are in the “mature” stage characterized by the high expression of Cd80 and Cd86, whereas genes related to immunogenic GO were more prominent in Aire⁻ DCs. The fact that the expression of Aire and the identity of CD11c⁺ Aire-DCs were not linked, as exemplified by the results shown in Supplemental Fig. 4E, may strengthen our conclusion on the dispensable role of Aire in CD11c⁺ Aire-DCs from a different aspect. Finally, we found that Aire⁺ BMDCs showed high expression of genes related to the cell cycle (Supplemental Fig. 4F), probably due to the massive expansion upon supplementation with cytokines.

A recent detailed single-cell multiomics approach has clearly demonstrated that there are two classes of eTACs (11): migratory DC-type eTACs and ILC3-like eTACs expressing RORγt (10). We have confirmed the presence of two types of Aire-expressing APCs in the periphery, cDC-type eTACs and ILC3-like–type eTACs, using a high-fidelity Aire-reporter mouse strain coupled with conventional flow cytometric and immunohistochemical analyses. These two types of eTACs showed different expression levels of Aire and a distinct location in the LNs. Although the single-cell multiomics approach has suggested that both classes of eTACs are DCs, and that migratory DC-type eTACs have the characteristics of migratory DCs with high CCR7 expression (11), our flow cytometric analysis has revealed that CD11c⁺ Aire-DCs were not confined to the migratory DCs. Instead, CD11c⁺ Aire-DCs were composed of both CCR7^high migratory DCs and CCR7^low resident DCs at a roughly equal ratio (Fig. 2E) (39, 40). We therefore classified Aire⁺ DCs into the cDC-type Aire⁺ cells (cDC-type eTACs) and ILC3-like–type Aire⁺ cells (ILC3-like–type eTACs) in this study. One possible reason for the phenotypic difference between our study and the single-cell multiomics study might originate from the different Aire-reporter strains used: the straight KI of the Aire-reporter molecule in our study (16) versus BAC transgenic mice in the other study (11). The difference in the form of reporter molecules for cell sorting (authentic Aire-GFP nuclear dots in our study versus GFP from the whole cell in the other study) might be another possible reason for the discrepancy.

Although the indispensable role of Aire in ILC3-like–type eTACs in the control of C. albicans was clearly demonstrated in the recent report (15), the exact role of Aire in cDC-type eTACs in the Ag presentation and production of the transcriptome has not been investigated until the current study. So far, we found that Aire in cDC-type eTACs has no major impact on the Ag-presenting capacity for protein Ag (OVA) as well as for the infectious agent of C. albicans. Given that more CD11c⁺ Aire-DCs were present in the cervical LNs (in which the immune response against oral and esophageal candidiasis might be taking place) than in other peripheral LNs, this was further unexpected for the following reasons. First, cDC-type eTACs dominated ILC3-like–type eTACs in number by several fold, although the expression level of Aire from each cell in cDC-type eTACs was lower compared with that from ILC3-like–type eTACs. Second, cDC-type eTACs were predominantly located in the T cell areas within LNs where Ag-specific T cells should have better access to APCs (Fig. 1F). In contrast, ILC3-like–type eTACs were mainly located in the interfollicular spaces of the B cells where interaction with T cells in situ may not be so frequent at least at a steady state. Therefore, higher expression of C. albicans–sensing receptors such as Galectin-3, Dectin-1, and TLR2 from ILC3-like–type eTACs (15) might have overcome their quantitative and anatomical disadvantages for the immune response against C. albicans. Indeed, transcriptomic analyses of ILC3-like–type eTACs between WT and Aire-KO were performed after repeated challenges with heat-killed C. albicans (15), and Aire-dependent expression of cell adhesion molecules, costimulatory molecules, and proinflammatory responses might have together contributed to the host defense by the ILC3-like–type eTACs upon Candida infection. In this regard, unidentified immune functions of cDC-type eTACs by employing different experimental models may be required. Furthermore, elucidation of the distinct role of cDC-type eTACs and ILC3-like–type eTACs beyond the host defense against C. albicans will help our better understanding of the Aire-mediated immune homeostasis in the periphery.

For the assessment of the immune response against C. albicans, we have used the MP fraction of mycelial cells from C. albicans in some experiments because mice primed with this fraction promoted the differentiation of CD4⁺ T cells into Th17 (26). Furthermore, Th17 cells differentiated in response to this fraction could prevent murine oral candidiasis. Although we could induce the immune response against C. albicans using cDC-type eTACs with this Candida MP Ag, we observed no major role of Aire in cDC-type eTACs for this action. We speculate that Aire in ILC3-like–type eTACs plays a role in shaping the unique function of this population through capturing and presenting Ags from C. albicans in an Aire-dependent manner. This protective function of ILC3-like–type eTACs was not compensated by the cDC-type eTACs. Accordingly, identifying the effective T cell Ags from C. albicans presented by ILC3-like–type eTACs is an important issue to develop effective immunotherapy against candidiasis. Nevertheless, whether Aire-mediated thymic tolerance is additionally contributing to the host defense against C. albicans requires further study. This is because Aire deficiency results in the production of neutralizing autoantibodies against IL-17A, IL-17F, and IL-22, which also play a role during the Candida infection (43, 44). Conversely, whether ILC3-like–type eTACs are involved in peripheral self-tolerance besides the protection against C. albicans is another interesting question. Interestingly, ILC3-like–type eTACs showed the transcriptional and regulatory homology to mTECs revealed by single-cell multiomics (11).

eTACs have been originally demonstrated to express a diverse array of distinct self-antigens, the spectrum of which is complementing those of mTECs (9). eTACs are capable of interacting with and deleting naive autoreactive T cells (9) or inactivating cognate T cells (45), a seemingly analogous way by which Aire⁺ mTECs contribute to thymic self-tolerance. Importantly, however, note that the tolerogenic role of eTACs has been demonstrated by the transgenic mouse models: pancreatic self-antigens (i.e., BDC2.5 [45], hybrid peptides between chromogranin A and insulin [46], and islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP) [11]) were driven by the Aire promoter in these animal models. By crossing with corresponding TCR-transgenic mice, the authors demonstrated that the presentation of the model self-antigens from the eTACs was relevant to the control of diabetes development. Although these studies suggested the significance of the expression of self-antigens from eTACs, the outcome of Aire deficiency in eTACs has not been tested. Similarly, although eTACs have been demonstrated to play a role in maternal–fetal tolerance (47), the study was performed by the ablation of eTACs using Aire-promotor–driven diphtheria toxin receptor coupled with diphtheria toxin injection. Thus, although eTACs may constitute an important component of peripheral self-tolerance, the central question remains unanswered of whether Aire has any cell-intrinsic role in the function of eTACs for establishing self-tolerance. Further work is required to answer this question by comparing the cellular heterogeneity between WT and Aire-KO at a single-cell level (48).

So far, we have no evidence for the functional importance of Aire in cDC-type eTACs for Ag presentation and the production of the transcriptome. In this regard, one study characterizing the eTACs from human tonsils merits attention (41). The study concluded that the expression of Aire was transient, rather than stable, and was associated with the differentiation to a mature phenotype of eTACs. The study also concluded that Aire expression within eTACs was not associated with an enriched expression of TRAs. Based on the results of our present study, we favor this notion. In this case, the expression of Aire might be a bystander phenomenon during the differentiation into mature DCs rather than Aire being responsible for acquiring the capacity for Ag presentation in cDC-type eTACs. Alternatively, antigenic stimuli may change the significance of Aire in cDC-type eTACs that were not tested in our work. The relative importance between cDC-type eTACs and ILC3-like–type eTACs in immune control under various conditions needs further study.

The authors have no financial conflicts of interest.