Recirculating Foxp3⁺ regulatory T cells are restimulated in the thymus under Aire control

Abstract

Thymically-derived Foxp3⁺ regulatory T cells (T_reg) critically control immunological tolerance. These cells are generated in the medulla through high affinity interactions with medullary thymic epithelial cells (mTEC) expressing the Autoimmune regulator (Aire). Recent advances have revealed that thymic T_reg contain not only developing but also recirculating cells from the periphery. Although Aire is implicated in the generation of Foxp3⁺ T_reg, its role in the biology of recirculating T_reg remains elusive. Here, we show that Aire regulates the suppressive signature of recirculating T_reg independently of the remodeling of the medullary 3D organization throughout life where T_reg reside. Accordingly, the adoptive transfer of peripheral Foxp3⁺ T_reg in Aire^KO recipients led to an impaired suppressive signature upon their entry into the thymus. Furthermore, recirculating T_reg from Aire^KO mice failed to attenuate the severity of multiorgan autoimmunity, demonstrating that their suppressive function is altered. Using bone marrow chimeras, we reveal that mTEC-specific expression of Aire controls the suppressive signature of recirculating T_reg. Finally, mature mTEC lacking Aire were inefficient in stimulating peripheral T_reg both in polyclonal and antigen-specific co-culture assays. Overall, this study demonstrates that Aire confers to mTEC the ability to restimulate recirculating T_reg, unravelling a novel function for this master regulator in T_reg biology.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-022-04328-9.

Introduction

CD25⁺Foxp3⁺ regulatory T cells (T_reg) constitute a distinct subset of CD4⁺ T cells endowed with suppressive functions. By maintaining immune tolerance, T_reg critically prevent the emergence of autoimmune and inflammatory diseases. The vast majority of T_reg is produced in the thymic medulla where they arise from two distinct developmental programs involving CD25⁺Foxp3⁻ (CD25⁺ T_regP) and CD25⁻Foxp3^lo (Foxp3^lo T_regP) precursors [1–4]. For a long-time, thymic CD25⁺Foxp3⁺ T_reg were thought to exclusively correspond to developing cells. However, recent studies have shown that circulating CD25⁺Foxp3⁺ T_reg have the ability to migrate back into the thymus [5–7]. Thus, thymic CD25⁺Foxp3⁺ T_reg is a heterogeneous population containing both developing and recirculating cells [3]. Interestingly, recirculating T_reg show suppressive properties similar to their splenic counterparts [7]. They exhibit an activated and differentiated phenotype and were found to negatively regulate the de novo generation of CD25⁺Foxp3⁺ T_reg through IL-2 consumption [5].

The recirculation of peripheral Foxp3⁺ T_reg in the thymus is mediated at least by two chemokine receptors, CCR6 and CXCR4 [5, 8]. Interestingly, the entry of peripheral Ccr6^−/− T_reg into the thymus is impaired, suggesting that CCR6, expressed by effector/memory T cells, is involved in the recirculation of Foxp3⁺ T_reg from the periphery to this organ [8]. Furthermore, the expression of the CCR6 ligand, CCL20, is regulated by the Autoimmune regulator (Aire) [8]. Aire is mainly expressed by a subset of medullary thymic epithelial cells (mTEC) commonly called mTEC^hi, characterized by a CD80^hiMHCII^hi phenotype. Furthermore, upon recirculation into the thymus, a subset of B cells has been described to upregulate Aire [9]. Independent reports have described that CD25⁺Foxp3⁺ T_reg are reduced in the thymus of Aire^KO mice [10, 11]. Accordingly, Aire expression was proposed to direct autoreactive CD4⁺ thymocytes into the T_reg cell lineage [12]. Interestingly, in the perinatal life, Aire promotes the generation of a distinct population of Foxp3⁺ T_reg, which persists in adults and prevents the emergence of autoimmunity [13]. In C57BL/6 mice, Aire-deficiency is associated with mild autoimmunity characterized by lymphocyte infiltrations and autoantibody production targeting several organs such as the pancreas, eyes, salivary glands, liver and lungs [14–17]. In human, autosomal recessive mutations in the Aire gene induce a life-threatening pathology called autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy (APECED), also known as autoimmune polyendocrine syndrome-type 1 (APS-1) [18, 19]. Interestingly, numbers, activation and suppressive functions of Foxp3⁺ T_reg are altered in APECED patients [20–22].

Although thymic Foxp3⁺ T_reg have been found to be more heterogeneous than previously thought [3], this emerging population of recirculating peripheral T_reg remains poorly described. In particular, the mechanisms that control their suppressive phenotype in the thymus are unknown. In this study, we show that Aire regulates the suppressive properties of recirculating CCR6⁺ T_reg, independently of the remodeling of the 3D organization of the thymic medulla where they reside [5]. High-throughput sequencing revealed that recirculating CCR6⁺ T_reg in the thymus of Aire^KO mice have impaired expression of several genes associated with T_reg suppressive functions and helper T cell polarization. Accordingly, recirculating CCR6⁺ T_reg from Aire^KO mice failed to attenuate the severity of multiorgan autoimmunity, demonstrating that their suppressive activity is impaired. Moreover, the adoptive transfer of splenic Foxp3⁺ T_reg in Aire^KO recipients resulted in an impaired suppressive signature upon their recirculation in the thymus. Using bone marrow (BM) chimeras, we demonstrate that Aire expressed by mTEC rather than by thymic B cells is responsible for the suppressive properties of recirculating CCR6⁺ T_reg. Finally, recirculating T_reg were found in close contact with Aire⁺ mTEC and restimulated in an antigen-specific manner. Altogether, our data reveal that Aire confers to mTEC the capacity to control the restimulation of recirculating CCR6⁺ T_reg in the thymus.

Materials and methods

Mice

CD45.1 WT (B6.SJL-PtprcaPepcb/BoyCrl, Stock n°002,014, Charles River), CD45.2 WT (Stock n°000,664, Charles River), CD45.1/2 WT, CD45.2 Aire^KO [23], Rag2^KO [24], CD45.1 Foxp3^eGFP mice [25], OTII [26] and Rip-mOVA [27] were on a C57BL/6 J background. Rip-mOVA × OTII mice were backcrossed on a Rag2^KO background. All mice were maintained under specific pathogen-free conditions at the Centre d’Immunologie de Marseille–Luminy (CIML, France). Standard food and water were given ad libitum. Males and females were used at the age of 5 days, 9 days, 6 weeks and 1 year. Chimeras were generated between 6 and 10 weeks of age. All experiments were done in accordance with National and European laws for laboratory animal welfare (EEC Council Directive 2010/63/UE) and the Marseille Ethical Committee for Animal experimentation no. 14.

BM chimeras

BM chimeras were generated by injecting intravenously (i.v.) 5.10⁶ BM cells from CD45.1 Foxp3^eGFP mice into lethally irradiated CD45.2 Aire^WT or Aire^KO recipients (two doses of 500 rads, 8 h apart, X-ray using a RS-2 000 Irradiator; Rad Source Technologies). Similarly, 5.10⁶ BM cells from either CD45.2 Aire^WT or Aire^KO mice were injected into lethally irradiated CD45.1/2 WT recipients. Mice were analyzed 6-weeks post-reconstitution.

Foxp3⁺ T_reg adoptive transfer

5.10⁵ cell-sorted splenic congenic CD45.1 CD4⁺CD25⁺Foxp3^eGFP T_reg were i.v injected into sublethally irradiated CD45.2 Aire^WT or Aire^KO recipient mice (one dose of 500 rads, X-ray using a RS-2 000 Irradiator; Rad Source Technologies). Mice were analyzed 1-week post-T_reg adoptive transfer.

Multiorgan autoimmunity experiments

Rag2^KO recipients were injected i.v. with 3.10⁶ CD4⁺CD25⁺ T_reg-depleted splenocytes purified from CD45.1 WT mice. Four weeks later, 15.10⁴ splenic CD4⁺CD25^hi T_reg or 10.10⁴ recirculating thymic CCR6⁺CD4⁺CD25⁺ T_reg from Aire^WT or Aire^KO mice were adoptively transferred. Mice that did not receive any T_reg were used as controls. Peripheral tissues were harvested and analyzed three weeks later.

Cell isolation

Thymic T_reg, splenic T_reg and T_reg-depleted splenocytes were isolated by scratching the thymus or the spleen on a 70-μm mesh. Red blood cells were lysed with RBC lysis buffer (eBioscience). Thymic and splenic T_reg cells were pre-enriched by depleting CD8⁺ and CD19⁺ cells using biotinylated anti-CD8α (clone 53.6.7; BD Biosciences) and anti-CD19 (clone 1D3; BD Biosciences) antibodies with antibiotin microbeads by AutoMACS using the Deplete program (Miltenyi Biotec). Recirculating thymic T_reg were sorted as CCR6⁺CD4⁺CD8⁻CD25⁺ cells and splenic T_reg as CD4⁺CD25^hi cells using a FACSAria III cell sorter (BD Biosciences). Splenocytes were depleted of CD4⁺CD25⁺ T_reg using a FACSAria III cell sorter (BD Biosciences). Thymus were digested at 37 °C in HBSS medium containing Liberase TM (50 µg/ml; Roche) and DNase I (100 μg/ml; Roche) until complete tissue digestion. Total mTEC (CD45⁻EpCAM⁺UEA-1⁺Ly51^−/lo) or mTEC^hi (CD45⁻EpCAM⁺UEA-1⁺Ly51^−/loCD80^hi) were pre-enriched by depleting CD45 hematopoietic cells using anti-CD45 magnetic beads (Miltenyi Biotec) by AutoMACS with the DepleteS program and sorted using a FACSAria III cell sorter (BD Biosciences).

In vitro co-culture assays

3.10⁴ cell-sorted splenic CD4⁺CD25^hi T_reg from Aire^WT or Aire^KO mice were co-cultured for 24 h with 6.10³ CD45⁻EpCAM⁺UEA-1⁺Ly51^−/lo mTEC from Aire^WT or Aire^KO mice in RPMI (ThermoFisher) supplemented with 10% FBS (Sigma Aldrich), L-glutamine (2 mM, ThermoFisher), sodium pyruvate (1 mM, ThermoFisher), 2-mercaptoethanol (2 × 10^–5 M, ThermoFisher), penicillin (100 IU/ml, ThermoFisher), streptomycin (100 μg /ml, ThermoFisher) and mouse IL-2 (40 ng/ml, Immunotools). For antigen-specific co-culture assays, 5.10³ cell-sorted splenic CD4⁺CD25^hi T_reg from Rip-mOVA x OTII x Rag2^KO mice were co-cultured for 24 h with 1.10³ CD45⁻EpCAM⁺Ly51^−/loAire^eGFP mTEC from Aire^het or Aire^KO mice previously loaded with OVA_323-339 peptide (5 µg/ml, Polypeptide group) for 2 h.

Flow cytometry

Cells were stained with standard procedures using antibodies listed in Table S2. For intracellular staining with anti-Foxp3, anti-GITRL and anti-OX40L antibodies, cells were fixed, permeabilized and stained with the Foxp3 staining kit according to the manufacturer’s instructions (eBioscience). Stained cells were analyzed with FACSCanto II and LSR II (BD Biosciences) and data were analyzed using FlowJo software (Tree Star).

Quantitative RT-PCR

The total RNA was extracted with TRIzol (Invitrogen) and cDNA was synthesized with random oligo dT primers and Superscript II reverse transcriptase (Invitrogen). Quantitative PCR was performed with SYBR Premix Ex Taq Master Mix (Takara) on an ABI 7500 fast real-time PCR system (Applied Biosystem). The results were normalized to actin mRNA expression. A list of primer sequences is provided in Table S3.

RNA-sequencing experiments

5.10⁴ CCR6⁺ T_reg were cell-sorted from the thymus of 6-week-old Aire^WT and Aire^KO mice. Two biological replicates were prepared for each condition. The total RNA was extracted using the RNeasy Micro Kit (Qiagen) and treated with DNase I. RNA-seq libraries were prepared using the TruSeq Stranded mRNA kit (Illumina) and sequenced with the Illumina HiSeq 2000 machine to generate datasets of single-end 50 bp reads. The reads were mapped to the mouse reference genome (mm10) using TopHat2 (version 2.0.12) [28], then counted using Cufflinks or Cuffdiff (version 2.2.1) [29, 30] and the mm10 genome GTF gene annotation file. In addition to read counting, Cuffdiff performs between-sample normalization and was used to calculate the differential gene expression and its statistical significance in Aire^WT vs Aire^KO T_reg by using the default “pooled” dispersion method that applies to experiments having few (≥ 2) biological replicates per condition. Expression levels generated by Cufflinks, as fragments per kilobase of transcript per million mapped reads (FPKM), were processed by the Matrix2png program [31] to generate heatmaps of gene expression levels. The dataset generated in this study are available in the Gene Expression Omnibus (GEO) database under accession number GSE188419. The expression of Tnfsf4 and Tnfsf18 was analyzed in RNA-sequencing dataset from Aire^KO mTEC^hi (GSE87133).

Immunofluorescence staining

Thymi were collected from 9-day-, 6-week- and 1-year-old Aire^WT or Aire^KO mice and fat tissues removed to avoid any interference with the 3D reconstitution process. Organs were longitudinally included in O.C.T (Sakura Finetek), frozen at -80 °C and cut in 20-µm-thick slices. Thymic sections were fixed with 2% paraformaldehyde, then saturated with 3% BSA and 0.01% Triton X100 in 0.1 M Tris HCl buffer. Sections were next stained for 45 min with rabbit anti-keratin 14 or with anti-Aire Alexa Fluor 488 (5H12; eBioscience), anti-Foxp3 PE (FJK-16 s; eBioscience) and anti-CD73 Alexa Fluor 647 (TY/11.8; BioLegend) in hybridization buffer (1% BSA and 0.02% Triton X100 in 0.1 M Tris–HCl, pH 7.4). Keratin 14 staining was revealed with Cy3-conjugated anti-rabbit (Invitrogen). Sections were counterstained with DAPI (1 µL/ml, BioLegend) and mounted with Mowiol (Calbiochem). For 3D reconstruction, images were acquired with a slide scanner (Panoramic SCAN II; 3D Histech). Confocal microscopy was performed with a LSM 780 confocal microscope (Carl Zeiss Microscopy).

Detection of immune infiltrates and autoantibodies

Peripheral tissues were fixed in buffered 10% formalin solution. 4 µm-thick paraffin-embedded sections were counterstained with hematoxylin and eosin. Autoantibody production was assessed by immunostaining organ sections from Rag2^KO mice with the sera (1/80) of analyzed mice. Autoantibodies were revealed with FITC-conjugated goat anti-mouse IgG. Sections were counterstained with DAPI and mounted with Mowiol (Calbiochem). All images were acquired with a slide scanner (Panoramic SCAN II; 3D Histech) and analyzed with ImageJ software (National Institutes of Health) to compute the mean fluorescence intensity of each image.

Thymic 3D reconstitution

For 3D reconstitution, images from the entire thymus of 9-day-, 6-week- and 1-year-old Aire^WT and Aire^KO mice were processed with For3D software as previously described [32, 33]. Briefly, images were smoothed by median and Gaussian filtering and medulla volumes were determined using ImageJ and Matlab (The Mathworks) software. Medullary islets identified in the 3D structures were measured individually using ImageJ and were color-coded using Imaris (Bitplane).

Statistical analysis

Statistics were performed with GraphPad Prism 9.1 software. Normal distribution of the data was assessed using d'Agostino–Pearson omnibus normality test. Statistical significance was then assessed using unpaired Student’s t test for two normal distributions, Mann–Whitney test for two non-normal distributions or Kruskal–Wallis test for more than two distributions. *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001. All bar graphs show mean ± SEM, unless mentioned.

Results

Aire regulates the recirculation of peripheral CD25⁺Foxp3⁺ T_reg in the thymus, independently of the remodeling of the 3D organization of the medulla throughout life

Considering the importance of Aire in the induction of self-tolerance [34], we first assessed whether it could be involved in the topology of the thymic medulla. For this, we compared WT (Aire^WT mice) with Aire-deficient mice in which the first exon of the Aire locus was replaced by the sequence of enhanced green fluorescent protein (eGFP) (Aire^KO mice) [23]. Thymic sections of 9-day-, 6-week- and 1-year-old, either from Aire^WT or Aire^KO mice, were stained with the keratin-14 mTEC-specific marker [35] and reconstructed with our in-house dedicated “Full organ reconstruction in 3D” (For3D) software [33] (Fig. 1A–C and Movies S1–6). Interestingly, the number of medullary islets in Aire^WT mice diminished between 9 days and 6 weeks of age while it increased between 6 weeks and 1 year of age (Fig. 1D). Furthermore, the total and individual medullary volumes increased between 9 days and 6 weeks while it decreased between 6 weeks and 1 year (Fig. 1E, F). Nevertheless, the central compartment of ~ 1 mm³ was observed at all ages analyzed (Fig. 1F) These observations reveal that the medulla organization is dynamic throughout life. In comparison to Aire^WT mice, the thymi of Aire^KO mice show a similar dynamic medulla organization characterized by normal numbers of medullary islets and total medullary volumes with the presence of a major compartment (Fig. 1A–C, F). These results indicate that the dynamic remodeling of the 3D organization of the thymic medulla throughout life is not regulated by Aire.

Fig. 1.

The medullary topology is dynamic throughout life, independently of Aire expression. A–C Representative images of thymic sections stained for keratin 14 (red) and counterstained with DAPI (blue) (upper panel). For3D reconstruction of thymic lobes from 9-day- (A), 6-week- (B) and 1-year-old (C) Aire^WT and Aire^KO mice, using Matlab (middle panel) and Imaris to depict medullary regions according to their volumes from cyan (smallest) to magenta (largest) (lower panel). Axes are graduated in millimeters (mm). Scale bar, 1 mm. The asterisk denotes the central medulla. D,E Histograms show the number of medullary islets (D) and the total medullary volume (E) derived from two thymic lobes for each condition. F The graph shows the volumes of each medullary islet derived from two thymic lobes for each condition from individual mice measured by For3D. The dashed circle denotes the central medulla of each lobe. Horizontal lines represent the geometric mean and SD. ****p < 0.0001 using Kruskal–Wallis test for (F)

Because medulla formation correlates with Foxp3⁺ T_reg emergence during ontogeny [36], we next analyzed whether Aire deficiency could affect their development in neonates of 5 days of age. In contrast to CD25⁺ T_regP, frequencies and numbers of Foxp3^lo T_regP and mature CD25⁺Foxp3⁺ T_reg were diminished in the thymus of 5-day-old Aire^KO mice as compared to their respective counterparts (Fig. 2A). This decrease was also observed in 6-week- and 1-year-old Aire^KO mice (Fig. 2B, C). Since mature CD25⁺Foxp3⁺ T_reg contain both developing and recirculating cells in the adult thymus, we used the key thymus-homing chemokine receptor CCR6 to distinguish developing (CCR6⁻) and recirculating (CCR6⁺) mature T_reg [6]. Interestingly, both numbers of CCR6⁻ and CCR6⁺ mature T_reg were reduced in Aire^KO mice compared to Aire^WT mice at 6 weeks and 1 year of age (Fig. 2D, E). Furthermore, 1-year-old Aire^WT mice showed a marked reduction in numbers of CCR6⁻ and CCR6⁺ mature CD25⁺Foxp3⁺ T_reg as compared to 6-week-old Aire^WT mice, which reflects the effect of thymic involution on T_reg cells in normal conditions. To date, two chemokine receptors, CCR6 and CXCR4 have been implicated in the recirculation of peripheral T_reg into the thymus [5, 8]. Strikingly, the expression of their respective ligands, Ccl20 and Cxcl12, was substantially reduced in Aire^KO mTEC^hi (Fig. 2F), consistently with the altered recirculation of peripheral T_reg in the thymus. Altogether, these results show that Aire controls both the development and recirculation of Foxp3⁺ T_reg throughout life.

Fig. 2.

Developing CCR6⁻ and recirculating CCR6⁺ T_reg are reduced in Aire^KO mice throughout life. A Flow cytometry profiles, frequencies and numbers of CD25⁺ T_regP, Foxp3^lo T_regP and CD25⁺Foxp3⁺ T_reg analyzed in CD4⁺ SP thymocytes from the thymus of 5-day-old Aire^WT and Aire^KO mice. The data are derived from 3 independent experiments (n = 2–4 mice per group and per experiment). B, C Frequencies and numbers of CD25⁺ T_regP, Foxp3^lo T_regP and CD25⁺Foxp3⁺ T_reg in the thymus of 6-week- (B) and 1-year- (C) old Aire^WT and Aire^KO mice. D, E Flow cytometry profiles (D), frequencies and numbers (E) of CCR6⁻ and CCR6⁺ cells in CD25⁺Foxp3⁺ T_reg from 6-week- and 1-year-old Aire^WT and Aire^KO mice. The data are derived from at least 3 independent experiments (n = 3–4 mice per group and per experiment). F The expression of Ccl20 and Cxcl12 was measured by qPCR in purified mTEC^hi (EpCAM⁺UEA-1⁺Ly51^−/loCD80^hi) from 6-week-old Aire^WT and Aire^KO mice. Bar graphs show ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 using unpaired Student’s t test (A, B, C, E) and two-tailed Mann–Whitney test for (F)

Thymic CCR6⁺ T_reg from Aire^KO mice show an impaired effector and suppressive phenotype

Because Aire controls the recirculation of CCR6⁺ T_reg, we made the hypothesis that it could also control their suppressive properties. To test this hypothesis, we first measured the expression level of several genes associated with T_reg suppressive functions in purified CCR6⁺CD4⁺CD25⁺ single-positive (SP) thymocytes that correspond to CD25⁺Foxp3⁺ T_reg (Fig. S1). Interestingly, whereas Foxp3 level was normal, the expression of Klrg1, a marker of terminally differentiated T_reg [37], was reduced in CCR6⁺ T_reg from 6-week-old Aire^KO mice (Fig. 3A). Accordingly, the expression of genes encoding for the inhibitory cytokine Il10, the cytolytic molecules Gzmb and Fasl, Lag3-associated with dendritic cell modulation, as well as the ectoenzymes Entpd1 (CD39) and Nt5e (CD73), implicated in target cell metabolic disruption, was reduced in CCR6⁺ T_reg of Aire^KO mice. A similar altered suppressive signature was also observed in CCR6⁺ T_reg purified from the thymus of 1-year-old Aire^KO mice (Fig. S2).

Fig. 3.

The suppressive signature of recirculating CCR6⁺ T_reg is impaired in the thymus of Aire^KO mice. A The expression level of Foxp3, Klrg1, Il10, Gzmb, Fasl, Lag3, Entpd1 and Nt5e was measured by qPCR in thymic CCR6⁺ T_reg from 6-week-old Aire^WT (n = 6) and Aire^KO (n = 9) mice. Bar graphs show mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001 using two-tailed Mann–Whitney test. B Scatterplot representations of log10 gene expression levels (FPKM) in recirculating CCR6⁺ T_reg from Aire^WT and Aire^KO mice. For representation purposes, expression values of genes below 0.01 were assigned to 0.01. Genes with fold difference > 2 and p-adj < 0.05 were considered as up- or down-regulated genes (red and green dots, respectively). RNA-seq was performed on 2 independent biological replicates derived from two mice. C Heatmap of selected genes down-regulated in recirculating CCR6⁺ Aire^KO T_reg (Fold Change > 2) compared to their CCR6⁺ Aire^WT counterparts and involved in T_reg suppressive functions, Th–like T_reg, helper T cell polarization, fat and muscle T_reg. Two biological replicates are shown for each condition

To further determine the impact of Aire on the functional properties of CCR6⁺ T_reg, we analyzed their gene expression profile by high-throughput RNA sequencing (Fig. 3B, C). Genes showing a significant variation (p ≤ 0.05) in gene expression between Aire^WT and Aire^KO CCR6⁺ T_reg with a fold change difference > 2 or < 0.5 were considered as up- and down-regulated, respectively. We identified a total of 2 634 upregulated genes reaching significance for 1060 of them (Cuffdiff p < 0.05) in Aire^WT CCR6⁺ T_reg as compared to their Aire^KO counterparts (Fig. 3B). Moreover, only 1 175 genes were downregulated with 301 of them reaching significance (Cuffdiff p < 0.05). Thus, the expression of Aire in the thymus upregulates three times more genes than it downregulates in recirculating CCR6⁺ T_reg. In accordance with the altered T_reg suppressive signature observed by qPCR (Fig. 3A), we found that Il10, Gzmb, Fasl, Lag3, Entpd1 and Nt5e were downregulated in CCR6⁺ T_reg of Aire^KO mice (Fig. 3C and Table S1). Furthermore, the expression of Prdm1 (Blimp-1), which characterizes effector Th-like T_reg [38] and the terminally differentiated markers Klrg1 and Tigit [37, 39] was reduced in CCR6⁺ T_reg of Aire^KO mice compared to their Aire^WT counterparts. Strikingly, CCR6⁺ T_reg of Aire^KO mice also expressed lower levels of several genes associated with their suppressive signature such as Ctla4 and Lgals1 (galectin-1) implicated respectively in dendritic cell modulation and target cell apoptosis [40, 41]. The expression of Tbx21 and Irf4, encoding for transcription factors associated with Th1- and Th2-like T_reg, as well as Pparg and Id2 genes associated with fat-resident effector T_reg [42] was also diminished. Accordingly, the expression of the chemokine receptors Cxcr3 of Th1-like, Ccr4 and Ccr8 of Th2-like as well as Ccr1 and Ccr2 of fat-resident T_reg [43], implicated in effector T_reg migration to the inflammatory site, was reduced in CCR6⁺ Aire^KO T_reg. Altogether, these results indicate that Aire expression is crucial for the effector and suppressive properties of recirculating CCR6⁺ T_reg in the thymus.

Recirculating CCR6⁺ T_reg from Aire^KO mice fail to attenuate the severity of multiorgan autoimmunity

Because numbers and suppressive signature of recirculating CCR6⁺ T_reg were reduced in the thymus of Aire^KO mice, we analyzed whether these defects would be also observed in the periphery. Blood and splenic CD4⁺Foxp3⁺ T_reg from 6-week- and 1-year-old Aire^KO mice showed similar frequencies, numbers and expression levels of suppressive genes as compared to their Aire^WT counterparts (Fig. S3). We next assessed the ability of peripheral and thymic CCR6⁺ T_reg from Aire^KO mice to dampen the severity of multiorgan autoimmunity. To this end, CD4⁺CD25⁺ T_reg-depleted splenocytes from CD45.1 WT mice were transferred into Rag2^KO lymphopenic recipients. Four weeks later, splenic T_reg or thymic CCR6⁺ T_reg purified from either Aire^WT or Aire^KO mice were adoptively transferred into these recipients (Figs. 4A and S4A). Signs of autoimmunity in peripheral tissues were visualized by histology and quantified by flow cytometry three weeks later. Rag2^KO mice that did not receive any T_reg were used as controls. Accordingly with the normal suppressive signature of splenic T_reg from Aire^KO mice (Fig. S3), Rag2^KO mice adoptively transferred with these cells show tissue infiltration levels similar to mice transferred with Aire^WT splenic T_reg (Fig. 4B, C). In marked contrast to mice that received Aire^WT thymic CCR6⁺ T_reg, mice adoptively transferred with thymic CCR6⁺ T_reg from Aire^KO mice failed to attenuate T-cell infiltration in several peripheral tissues despite similar numbers of CD45.2 donor Foxp3⁺ T_reg in lymph nodes (Fig. 4D, E and S4B, C). Examination of CD45.1 donor cell infiltration by flow cytometry revealed that 100% of Rag2^KO mice that received Aire^KO thymic CCR6⁺ T_reg showed a high infiltration level in the pancreas and salivary glands, 83% in eyes, 33% in the lung and liver as well as 16% in the kidney (Fig. 4E). In contrast, only 16% of Rag2^KO mice transferred with Aire^WT thymic CCR6⁺ T_reg showed a high infiltration in the liver. Furthermore, flow cytometry analysis showed that thymic CCR6⁺ T_reg of Aire^KO mice were unable to prevent the infiltration of CD45.1 CD4⁺ and CD8⁺ T cells in the pancreas and eyes as well as CD45.1 CD8⁺ T cells in salivary glands (Fig. S4D). Finally, immunostaining of Rag2^KO tissue sections with sera from these mice revealed higher levels of autoantibodies against the pancreas and salivary glands in mice transferred with Aire^KO thymic CCR6⁺ T_reg than in mice injected with Aire^WT thymic CCR6⁺ T_reg (Fig. 4F, G). Thus, thymic CCR6⁺ T_reg of Aire^KO mice failed to attenuate the severity of multiorgan autoimmunity, demonstrating that their suppressive activity was impaired.

Fig. 4.

The adoptive transfer of recirculating CCR6⁺ T_reg of the Aire^KO thymus fails to protect from multi-organ autoimmunity. A Experimental setup: Rag2^KO recipients were adoptively transferred with CD4⁺CD25⁺ T_reg-depleted CD45.1 WT splenocytes. Four weeks later, they were injected with splenic T_reg or thymic CCR6⁺ T_reg derived from Aire^WT or Aire^KO mice. Three weeks after T_reg adoptive transfer, peripheral tissues were examined for immune infiltration by histology and quantified by flow cytometry. Rag2^KO recipients injected only with CD4⁺CD25⁺ T_reg-depleted CD45.1 WT splenocytes were used as controls. B-E Representative photographies of peripheral tissue sections derived from mice transferred with splenic T_reg (B) or thymic CCR6⁺ T_reg (D) and counterstained with hematoxylin/eosin. Diagrams represent organ infiltration levels by CD45.1 donor cells measured by flow cytometry upon splenic T_reg (C) or thymic CCR6⁺ T_reg (E) transfer. Infiltration levels were normalized to the infiltration observed in controls. Dark and light violet in diagram represent high and low infiltrations. Each diagram represents one individual mouse. Scale bar, 150 µm for (B), 500 µm for (D). F, G Sera from mice transferred with CCR6⁺ T_reg were tested for the presence of autoantibodies (green) against pancreas (F) and salivary glands (G) of Rag2^KO mice. Nuclei were counterstained with DAPI (blue). Secondary antibodies (II Abs) alone were used as controls. Scale bar, 1 mm. Histograms show mean fluorescence intensity for each condition. Data are derived from 2 to 3 independent experiments (n = 3–5 mice per group and per experiment). Bar graphs show mean ± SEM, ****p < 0.0001 using unpaired Student’s t test for (F, G)

Aire expression in the thymic stroma controls the suppressive signature of recirculating CCR6⁺ T_reg

Since Aire expression is not restricted to mTEC but was also found in thymic B cells [9], we then investigated its respective contribution in the stromal and hematopoietic compartments to control the recirculation and suppressive properties of CCR6⁺ T_reg. To determine the role of Aire in hematopoietic cells, we generated BM chimeras by reconstituting lethally irradiated CD45.1/2 WT recipients with either CD45.2 Aire^WT or Aire^KO BM cells (Fig. S5A). Six weeks later, Aire^KO BM chimeras did not show major defects in total CD19⁺B220⁺ B cells, neither in the IgD⁻ and IgD⁺ B cell subsets, both described to express Aire [9] (Fig. S5B, C). Overall, frequencies and numbers of CD25⁺ T_regP, Foxp3^lo T_regP, CCR6⁻ and CCR6⁺ mature CD25⁺Foxp3⁺ T_reg were also normal (Fig. S5D, E). We next cell-sorted CCR6⁺ T_reg from the thymus of these BM chimeras and analyzed their suppressive signature. Recirculating T_reg from both chimeras exhibited a similar expression of suppressive genes that was altered in CCR6⁺ T_reg of Aire^KO mice (Fig. S5F). Thus, Aire expression in thymic B cells is unlikely involved in T_reg development, recirculation and suppressive signature.

We then analyzed whether Aire expression in stromal cells controls CCR6⁺ T_reg functional suppressive properties. To this end, we generated BM chimeras in which lethally irradiated CD45.2 Aire^WT or Aire^KO mice were reconstituted with CD45.1 Foxp3^eGFP BM cells (Fig. 5A). Six weeks later, similar numbers of CD4⁺ SP thymocytes of CD45.1 donor origin were observed in Aire^WT and Aire^KO recipients (Fig. 5B). Although numbers of CD25⁺ T_regP and Foxp3^lo T_regP were also similar in both groups, frequencies and numbers of mature CD25⁺Foxp3⁺ T_reg were specifically reduced in Aire^KO chimeras (Fig. 5C). This defect was attributable to diminished frequencies and numbers of CCR6⁺ T_reg (Fig. 5D). Recirculating CCR6⁺ mature CD25⁺Foxp3⁺ T_reg of CD45.1 origin were then cell-sorted from the thymus of both chimeras and analyzed for the expression of several genes associated with T_reg effector functions. Whereas Foxp3 level was normal, the expression of Klrg1, Il10, Tgfb1, Gzmb, Fasl, Lag3, Entpd1 and Nt5e was reduced in Aire^KO chimeras as compared to control chimeras (Fig. 5E). Altogether, these results show that whereas Aire in hematopoietic cells is dispensable, its specific expression in stromal cells controls both recirculation and suppressive properties of CCR6⁺ T_reg.

Fig. 5.

Aire deficiency in the thymic stroma impairs the recirculation and the suppressive signature of CCR6⁺ T_reg. A Experimental setup: lethally irradiated CD45.2 Aire^WT or Aire^KO recipients were reconstituted with BM cells from CD45.1 Foxp3^eGFP mice. Six weeks later, thymic T_reg subsets of CD45.1 origin were analyzed by flow cytometry. CCR6⁺ T_reg were cell-sorted to measure the expression levels of genes associated with their suppressive functions. B Flow cytometry profiles, frequencies and numbers of CD4⁺ SP thymocytes of CD45.1 origin in the thymus of BM chimeric mice. C, D Flow cytometry profiles and numbers of CD25⁺ T_regP, Foxp3^lo T_regP and total CD25⁺Foxp3⁺ cells (C) as well as of CCR6⁻ and CCR6⁺ cells in CD25⁺Foxp3⁺ T_reg (D) of CD45.1 origin in the thymus of BM chimeras. Data are derived from 2 independent experiments (n = 4 mice per group and per experiment). E The expression level of Foxp3, Klrg1, Il10, Tgfb1, Gzmb, Fasl, Lag3, Entpd1 and Nt5e was measured by qPCR in purified CCR6⁺CD25⁺Foxp3⁺ T_reg from CD45.1 Foxp3^eGFP → Aire^WT (n = 8) and CD45.1 Foxp3^eGFP → Aire^KO (n = 9) chimeras. Data are derived from 2 independent experiments (n = 4–5 mice per group and per experiment). Bar graphs show mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 using unpaired Student’s t test for (B-D) and two-tailed Mann–Whitney test for (E)

Antigen-specific restimulation of recirculating T_reg by Aire⁺ mTEC

To further decipher the impact of Aire-expressing mTEC in the suppressive signature of recirculating T_reg, we used an in vitro co-culture setup of peripheral T_reg and mTEC purified from Aire^WT or Aire^KO mice. Compared with Aire^WT splenic T_reg co-cultured with Aire^WT mTEC, the expression of Tgfb1, Gzmb, Fasl, Entpd1 and Nt5e was reduced in Aire^WT splenic T_reg co-cultured with Aire^KO mTEC (Fig. 6A). Moreover, Aire^KO splenic T_reg co-cultured with Aire^WT mTEC upregulated the expression of these genes in contrast to Aire^KO splenic T_reg co-cultured with Aire^KO mTEC. Thus, as compared to their Aire^WT counterparts, Aire^KO mTEC failed to enhance suppressive signatures of both Aire^WT and Aire^KO splenic T_reg.

Fig. 6.

Aire⁺ mTEC control the activation and suppressive signature of recirculating T_reg through antigen-dependent contact. A The expression level of Tgfb1, Gzmb, Fasl, Entpd1 and Nt5e was measured by qPCR in splenic T_reg from Aire^WT or Aire^KO mice co-cultured with Aire^WT or Aire^KO mTEC. Data are derived from 2 independent experiments. B Experimental setup: purified splenic T_reg from CD45.1 Foxp3^eGFP mice were adoptively transferred i.v. into sublethally irradiated CD45.2 Aire^WT or Aire^KO recipients. Adoptively transferred CD45.1 Foxp3^eGFP T_reg were cell-sorted from the thymus of recipient mice one week later and their suppressive signature was analyzed by qPCR. C Flow cytometry profiles, frequencies and numbers of CD45.1 Foxp3^eGFP donor T_reg observed in the thymus of CD45.2 Aire^WT or Aire^KO recipients. D The expression level of Il10, Tgfb1, Gzmb, Fasl, Lag3 and Nt5e was measured by qPCR in CD45.1 Foxp3^eGFP donor T_reg transferred into Aire^WT (n = 4) or Aire^KO (n = 4) recipients. E Representative images of WT thymic sections stained for Aire (green), Foxp3 (red) and CD73 (white). Yellow and white arrowheads denote developing CD73⁻ and recirculating CD73⁺ T_reg, respectively. Scale bar, 30 µm. F Flow cytometry profiles of CD69 activation of splenic OTII T_reg from Rip-mOVA x OTII x Rag2^KO mice co-cultured with OVA_323-339-loaded Aire^het (Aire^eGFP/WT) or Aire^KO (Aire^eGFP/eGFP) mTEC^hi after 24 h later. The histogram shows the frequencies of CD69⁺ T_reg normalized to the activation of OTII T_reg co-cultured with Aire^het mTEC^hi. G The expression level of Il10, Tgfb1, Gzmb, Fasl, Lag3, Entpd1 and Nt5e was measured by qPCR in splenic OTII T_reg from Rip-mOVA x OTII x Rag2^KO mice co-cultured with OVA_323-339-loaded Aire^het (Aire^eGFP/WT; n = 6) or Aire^KO (Aire^eGFP/eGFP; n = 6) mTEC^hi. Data are derived from 2 independent experiments. Bar graphs show mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 using two-tailed Mann–Whitney test for (A, C, D, F, G)

We then analyzed the in vivo effect of the Aire^KO thymic stroma in regulating the suppressive signature of peripheral WT T_reg. To this end, splenic T_reg purified from CD45.1 Foxp3^eGFP congenic mice were adoptively transferred into either Aire^WT or Aire^KO recipients (Fig. 6B). One week later, we observed reduced frequencies and numbers of donor CD45.1 Foxp3^eGFP T_reg in the thymus of Aire^KO mice as compared to Aire^WT mice (Fig. 6C). Moreover, the remaining donor Foxp3^eGFP T_reg that recirculated back to the thymus exhibited an impaired suppressive signature (Fig. 6D). These results not only confirm that Aire favors the recirculation of peripheral T_reg in the thymus but also highlight its key role in enhancing their suppressive properties.

Thus, we made the hypothesis that recirculating T_reg could be restimulated by Aire⁺ mTEC. To test this, WT thymic sections were first stained for Aire, Foxp3 and CD73, the latter being a reliable marker of recirculating T_reg [1, 44]. Of note, anti-CCR6 antibody was not used in this experiment because it failed to give any signal in our hands. Interestingly, we observed that both developing CD73⁻ and recirculating CD73⁺ T_reg were found in close proximity to Aire⁺ mTEC (Fig. 6E). These results suggest that recirculating T_reg could be restimulated by establishing antigen-specific contacts with Aire⁺ mTEC. To test this hypothesis, splenic OTII T_reg from Rip-mOVA x OTII mice were co-cultured with OVA_323-339-loaded Aire^het (Aire^eGFP/WT) or Aire^KO (Aire^eGFP/eGFP) mTEC^hi (Fig. S6). When compared with unstimulated OVA-specific T_reg, Aire^KO mTEC^hi were able to activate OTII T_reg but to a lesser extent than Aire^het mTEC^hi (Fig. 6F). These results demonstrate that Aire⁺ mTEC can stimulate peripheral T_reg in an antigen-specific manner and this ability is impaired in the absence of Aire. Furthermore, as compared to OTII T_reg co-cultured with Aire^het mTEC^hi, the expression of Il10, Tgfb1, Gzmb, Fasl, Lag3, Entpd1 and Nt5e was reduced in OTII T_reg co-cultured with Aire^KO mTEC^hi (Fig. 6G), indicating that this antigen-specific restimulation is much less efficient. We thus analyzed MHCII expression in mTEC from Aire^KO mice. In accordance with a previous study [45] and the pro-apoptotic role of Aire [46], we found increased frequencies of MHCII⁺ mTEC in these mice (Fig. S7A). We then made the hypothesis that costimulatory signals may be implicated in the phenotype observed. Two ligands of the Tumor Necrosis Factor Superfamily (TNFSF) have been described to be constitutively expressed by mTEC and participate in thymic T_reg development [47]. Interestingly, we found that mTEC^hi expressed reduced levels of Tnfsf4 (OX40L, Aire^WT: 2.32 FPKM vs Aire^KO: 0.37) and Tnfsf18 (GITRL, Aire^WT: 2.92 vs Aire^KO: 0.81) in Aire^KO mice (Fig. S7B). This result was also confirmed at the protein level by flow cytometry (Fig. S7C). These observations are consistent with the reduced cellularity of Foxp3^lo T_regP and CD25⁺Foxp3⁺ T_reg in Aire^KO mice (Fig. 2A–C). Further investigations are needed to decipher the role of these two TNFSF ligands in the biology of recirculating T_reg in the thymus. Altogether, our data reveal that Aire⁺ mTEC, through antigen-specific restimulation, are responsible for the strong suppressive signature of recirculating T_reg.

Discussion

Our study demonstrates that Aire expression by mTEC promotes the suppressive properties of recirculating CCR6⁺ T_reg independently of the dynamic remodeling of the medullary 3D organization, where recirculating T_reg reside. We previously described that the medulla of 6-week-old young adult mice is complex with a large central compartment surrounded by hundreds of individual islets [32]. Nevertheless, determining whether this topology varies throughout life and whether Aire is implicated remained open issues. Interestingly, we found that 9-day- and 6-week-old WT mice show ~ 1000 and ~ 400 medullary islets, respectively. Given that individual islets arise from a single progenitor [48], these observations suggest that during thymic development, they grow and fuse together, leading to a reduced number of islets in young adult mice as compared to neonates. In 1-year-old mice, the number of medullary islets increases to reach ~ 1000 islets, as observed in neonates. Considering that the medulla topology is governed by crosstalk with autoreactive CD4⁺ thymocytes [32, 49–52], the high islet number observed in aged mice could be due to a suboptimal cellular crosstalk due to reduced cellularity of CD4⁺ thymocytes linked to age-related thymic involution. Although Aire plays multiple roles in T-cell tolerance induction [34], our results show that it does not shape the 3D organization of the thymic medulla. This is consistent with the fact that the clonal deletion of autoreactive CD4⁺ thymocytes is impaired in Aire^KO mice [45, 53], which consequently leads to an effective medulla organization.

Several studies have shown that mTEC are implicated in Foxp3⁺ T_reg development [54–56]. Nevertheless, the specific role of Aire in thymic T_reg heterogeneity throughout life remained to be defined. Interestingly, we found that Aire controls the cellularity of Foxp3^lo T_regP and mature CD25⁺Foxp3⁺ T_reg in 5-day-, 6-week- and 1-year-old mice. This is illustrated by decreased numbers of Foxp3^lo T_regP and CD25⁺Foxp3⁺ T_reg in Aire^KO mice, which could not be due to impaired medulla organization since no defect was observed at this level. Given that CD25⁺ T_regP and Foxp3^lo T_regP show distinct developmental pathways that give rise to CD25⁺Foxp3⁺ T_reg with non-overlapping regulatory activities [1], our results provide new insights in the role of Aire in the emergence of Foxp3^lo T_regP. Furthermore, Aire regulates the pool of recirculating CCR6⁺ T_reg throughout life. Interestingly, CCR6⁺ T_reg from Aire^KO mice expressed reduced levels of several genes associated with their polarization and suppressive functions. Importantly, they express normal levels of Foxp3, indicating that they remain engaged in the T_reg cell lineage. In accordance with the defective suppressive signature of CCR6⁺ T_reg from Aire^KO mice, we found that the adoptive transfer of these cells failed to attenuate the severity of multiorgan autoimmunity. In contrast to their thymic counterparts, splenic T_reg of Aire^KO mice show a protection similar to Aire^WT splenic T_reg, consistently with their normal suppressive signature. Altogether, these results indicate that Aire is crucial for the suppressive functions of recirculating CCR6⁺ T_reg in the thymus.

Although Aire is expressed by a subset of recirculating thymic B cells [9], we found that its absence in hematopoietic cells had no impact neither in T_reg development nor in CCR6⁺ T_reg recirculation and suppressive signature. Considering the weak expression of Aire in thymic B cells as compared to mTEC, it is not surprising that Aire deficiency in hematopoietic cells does not control the cellularity and suppressive signature of CCR6⁺ T_reg. In marked contrast, beyond controlling the recirculation of CCR6⁺ T_reg, BM chimeras in Aire^KO recipients revealed that Aire expression in stromal cells is responsible for their highly suppressive phenotype.

Interestingly, recirculating Foxp3⁺ T_reg were observed in close proximity to Aire⁺ mTEC, similarly to developing Foxp3⁺ T_reg. The reduced expression of Ccl20 and Cxcl12, likely responsible for the lower amount of recirculating T_reg in these mice [5, 8], could contribute to a lesser stimulation of recirculating T_reg by mTEC. However, in vitro co-culture experiments suggest that defective chemoattraction is unlikely responsible for the phenotype observed in Aire^KO mice. Furthermore, in vitro co-culture assays revealed that Aire⁺ mTEC were capable to activate peripheral T_reg in an antigen-specific manner. Our results suggest that peripheral T_reg could regulate de novo T_reg development not only by competing for IL-2 [5, 57] but also for the cognate self-antigen. In the absence of Aire, in vitro co-culture assays and in vivo adoptive transfer experiments demonstrate that recirculating T_reg activation and suppressive signature were altered. Therefore, our results unravel that Aire⁺ mTEC control the activated and differentiated phenotype of recirculating T_reg upon their entry into the thymus. A possible explanation could be that Aire controls co-stimulation signals that are responsible for maintaining the effector phenotype of recirculating T_reg. In addition, Aire could also modulate indirectly CCR6⁺ T_reg suppressive properties through other mechanisms such as the medullary positioning of XCR1⁺ type 1 conventional dendritic cells by controlling the production of the chemokine XCL1 [11]. Moreover, further investigations are required to determine the fate of CCR6⁺ T_reg in the thymic medulla. Three possibilities can be envisaged: CCR6⁺ T_reg (1) become long-term resident cells, (2) migrate back to the periphery or (3) die by apoptosis.

This study ameliorates our understanding on recirculating T_reg in the thymus, which remain poorly described to date. In summary, it identifies that Aire controls the suppressive properties of recirculating CCR6⁺ T_reg in the thymus. It also assigns a new role for Aire in conferring to mTEC the aptitude to restimulate recirculating T_reg. Thus, this study furthers our understanding on the mechanisms allowing recirculating T_reg to fine-tune de novo T_reg production. Finally, our results are expected to contribute to a better understanding of T_reg deficiencies observed in the human pathology APECED.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JC, AB, JCS, LC and MI conducted the experiments, analyzed and interpreted the data. MG and AS analyzed the data. JC, AB, JCS and MI wrote the manuscript. MI initiated, supervised and conceived the study.

Funding

This work was supported by institutional grants from INSERM, CNRS and Aix-Marseille Université.  The Immune Tolerance and T-Cell Differentiation laboratory received funding from the ARC Foundation (PJA20171206491 to M.I.), CoPoC-proof of concept (MAT-PI-17326-A-01 to M.I.), a prematuration grant from A*MIDEX, a French “Investissements d'avenir” program (LTalpha-Treg to M.I.) and Agence Nationale de la Recherche (grant ANR-19-CE18-0021–01, RANKLthym to M.I.). We also acknowledge financial support from France Bio Imaging (ANR-10-INBS-04–01) and France Génomique national infrastructure, funded as part of the "Investissements d'Avenir" program managed by the ANR (ANR-10-INBS-0009). J.C. and A.B. were supported by a PhD fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche et de l’Innovation (MESRI).

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files. The dataset generated in this study are available in the Gene Expression Omnibus (GEO) database under accession number GSE188419.

Code availability

Not applicable.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

All experiments were done in accordance with National and European laws for laboratory animal welfare (EEC Council Directive 2010/63/UE) and the Marseille Ethical Committee for Animal experimentation.

Consent to participate

Not applicable.

Consent to publication

Not applicable.