Abstract
Aire in medullary thymic epithelial cells (mTECs) plays an important role in the establishment of self-tolerance. Because Aire+ mTECs appear to be a limited subset, they may constitute a unique lineage(s) among mTECs. An alternative possibility is that all mTECs are committed to express Aire in principle, but Aire expression by individual mTECs is conditional. To investigate this issue, we established a novel Aire reporter strain in which endogenous Aire is replaced by the human AIRE-GFP-Flag tag (Aire/hAGF-knockin) fusion gene. The hAGF reporter protein was produced and retained very efficiently within mTECs as authentic Aire nuclear dot protein. Remarkably, snapshot analysis revealed that mTECs expressing hAGF accounted for >95% of mature mTECs, suggesting that Aire expression does not represent a particular mTEC lineage(s). We confirmed this by generating Aire/diphtheria toxin receptor–knockin mice in which long-term ablation of Aire+ mTECs by diphtheria toxin treatment resulted in the loss of most mature mTECs beyond the proportion of those apparently expressing Aire. These results suggest that Aire expression is inherent to all mTECs but may occur at particular stage(s) and/or cellular states during their differentiation, thus accounting for the broad impact of Aire on the promiscuous gene expression of mTECs.
Introduction
Medullary thymic epithelial cells (mTECs) are the central source of a variety of tissue-restricted Ags (TRAs) required for the elimination of autoreactive T cells (1, 2). Although Aire in mTECs was demonstrated to play an essential role in this tolerogenic activity, the exact cellular mechanisms of TRA gene expression controlled by Aire have not been fully elucidated (3).
mTECs develop from progenitors whose identity is still incompletely defined (4–6). Currently, only a limited number of tracer molecules is available for mapping the process of mTEC development; these include CD80 and MHC class II (MHC-II). CD80lowMHC-IIlow immature mTECs (mTEClow) develop into CD80highMHC-IIhigh mature mTECs (mTEChigh), and Aire expression is confined to the mTEChigh population (7, 8). Typically, 30–40% of all mTECs, or 40–50% of the mTEChigh population, express Aire that is detectable by snapshot analysis using flow cytometry with the anti-Aire mAb (7, 9) or by using Aire reporter mice (8, 10). Because Aire is expressed by a rather limited subset of mTECs, it is generally considered that Aire+ mTECs represent a unique lineage(s) that are specialized for TRA gene expression. However, it remains unclear why a rather limited subset of mTECs with defective Aire expression in Aire-deficient mice dramatically lack the promiscuous gene expression evident in the mTEC population as a whole (3, 11, 12). In this regard, it is noteworthy that we do not know how precisely Aire+ mTECs can be defined using available detection systems. For example, it is possible that these systems are not sufficiently sensitive for detecting Aire and that bona fide Aire+ mTECs account for a larger fraction of mTECs than is currently appreciated. If this is the case, it may alter our understanding of TRA gene expression by mTECs under the strong control of Aire.
To investigate this issue, we created a novel Aire reporter strain in which endogenous Aire is replaced by the human AIRE-GFP-Flag tag (hAGF) fusion protein. This Aire reporter molecule was produced as authentic Aire protein in the form of nuclear dots, in marked contrast to previously generated Aire reporter strains in which individual Aire+ mTECs were homogeneously (i.e., both the cytoplasm and nucleus) labeled with fluorescent protein (e.g., GFP). Because of this distribution of GFP throughout the cell, a kinetic difference between Aire expression (in the nucleus) and the GFP signal (in the whole cell) was inevitable in these Aire reporter strains (8, 10). Furthermore, differences in protein kinetics between Aire and GFP seem to have been disadvantageous for the detection of Aire+ mTECs with high sensitivity and specificity. Our novel Aire reporter strain (Aire/hAGF-knockin [KI]) overcame these problems by making GFP part of the Aire/AIRE nuclear dot protein. This approach markedly improved the sensitivity and specificity of Aire+ mTEC detection. Our studies using this novel Aire reporter strain suggested that most, if not all, mTECs are committed to Aire expression during their differentiation program. Our findings would well account for the overall picture of TRA gene expression by mTECs that is critically dependent on the function of Aire as a single transcription factor.
Materials and Methods
Mice
Aire/hAGF-KI and Aire/diphtheria toxin receptor (DTR)-KI mice were generated by homologous recombination in embryonic stem (ES) cells established from C57BL/6 mice using the same design as the generation of Aire/GFP-KI mice (8), with the exception of the KI cassette. After the targeted cells had been injected into morula-stage embryos, the resulting chimeric male mice were mated with C57BL/6 females (CLEA Japan) to establish germline transmission. The mice were then crossed with a transgenic line expressing the general deleter Cre recombinase to remove the neor gene cassette. The protocols used in this study were in accordance with the Guidelines for Animal Experimentation of Tokushima University School of Medicine.
Immunohistochemistry
Immunohistochemical analysis of the thymus using rat anti-mouse Aire mAb (clone RF33-1), rabbit polyclonal anti-GFP Ab (Invitrogen), and anti-EpCAM mAb (BD) was performed as described previously (13).
Thymic epithelial cell preparation and flow cytometric analysis
Preparation of TECs and flow cytometric analysis with a FACSAria II (BD) and a Gallios (Beckman Coulter) were performed as described previously (13). The mAbs used were anti-CD45, anti-EpCAM, anti-IAb, and anti-CD80 (all from eBioscience). UEA-1 was from Vector Laboratories. Rat anti-Aire mAb (clone RF33-1) was prepared in Mitsuru Matsumoto’s laboratory.
Real-time PCR
RNA was extracted from FACS-sorted mTEChigh cells with RNeasy Mini Kits (QIAGEN) and made into cDNA with a SuperScript VILO cDNA Synthesis Kit (Thermo Fisher), in accordance with the manufacturer’s instructions. Real-time PCR for quantification of salivary protein 1, C-reactive protein, and Hprt was performed as described previously (14).
Fetal thymus organ culture
Fetal thymus organ culture (FTOC) was performed as described previously (13). Thymic lobes were cultured in media containing recombinant RANKL (1 μg/ml; Oriental Yeast) and/or DT (500 ng/ml; Sigma) for 7 d.
Statistical analysis
All results are expressed as mean ± SEM. Statistical analysis was performed using the Student two-tailed unpaired t test for comparisons between two groups. Differences were considered significant if p values were ≤0.05.
Results
Generation of mice expressing hAGF fusion protein in Aire+ mTECs
To develop a better system for monitoring Aire+ mTECs, we used hAGF as an Aire reporter molecule after confirming that the hAGF gene was expressed as authentic Aire protein in the form of nuclear dots when transfected into the cultured cells (15) (H. Nishijima and M. Matsumoto, unpublished observations). We then established Aire/hAGF-KI mice in which expression of the hAGF gene is under the transcriptional control of endogenous Aire (Fig. 1A). Modification of the Aire locus was minimized by inserting an hAGF gene cassette between exon 1 and exon 2, as we used previously for generating Aire/GFP-KI mice (8). In this case, the hAGF-containing human AIRE module was originally expected to compensate for the loss of endogenous Aire in the targeted allele. However, mice harboring a combination of the Aire-null allele and hAGF allele (i.e., -/hAGF allele) backcrossed onto NOD for five to seven generations showed typical autoimmune attack against pancreatic acinar cells (Supplemental Fig. 1, Supplemental Table I), a characteristic of Aire-deficient mice on a NOD background (16, 17). In contrast, severe pneumonitis, another typical lethal autoimmune phenotype of Aire-deficient NOD mice, was not evident in these mice (Supplemental Fig. 1, Supplemental Table I), and they survived much longer than Aire-deficient NOD mice (data not shown). These results suggested that hAGF was not fully functional in vivo; in fact, it proved to be hypomorphic in comparison with authentic Aire protein when subjected to a stringent assessment for autoimmunity. Consistent with these observations, mTEChigh isolated from homozygous Aire/hAGF-KI mice (i.e., hAGF/hAGF allele) showed reduced expression of Aire-dependent TRA of salivary protein 1 but not Aire-independent TRA of C-reactive protein (Supplemental Fig. 2). We also observed reduced expression of salivary protein 1 in heterozygous Aire/hAGF-KI mice (i.e., +/hAGF allele), probably reflecting a gene-dosage effect of Aire, as reported previously (18). However, these functionally hypomorphic features of hAGF do not preclude the use of Aire/hAGF-KI mice as an Aire reporter strain to monitor Aire expression, because the hAGF allele had no obvious adverse effect on the differentiation program of Aire+ mTECs (see below).
Establishment of Aire/hAGF-KI mice. (A) Targeted insertion of the hAGF gene into the Aire locus by homologous recombination. Southern blot analysis of genomic DNA extracted from targeted ES cells (right panels). DNA was digested with EcoRV and SpeI (for the 3′ probe) or BglII (for the 5′ probe) and hybridized with the probes shown in red. Aire+/hAgf-neo mice were crossed with a Cre recombinase–expressing transgenic line to remove the neor gene cassette (bottom). (B) Cells expressing hAGF (green) were located within thymic medulla stained with anti-EpCAM mAb (red) by immunohistochemistry of a thymus section from an Aire/hAGF-KI mouse. Scale bar, 100 μm. One representative experiment from a total of four repeats is shown. (C) Concomitant expression of hAGF (green) and endogenous mouse Aire (red), as assessed by immunohistochemistry. Scale bar, 10 μm. One representative experiment from a total of four repeats is shown.
Using immunohistochemistry, we first examined whether GFP expression by Aire/hAGF-KI thymus reflects endogenous Aire expression. We observed many GFP signal dots within the epithelial cell adhesion molecule 1 (EpCAM)+ medullary region (Fig. 1B). In marked contrast to the previously generated Aire reporter strains in which GFP itself was used as a surrogate of Aire expression, and, consequently GFP expression was observed throughout (i.e., cytoplasm and nucleus) each Aire-expressing mTEC (8, 10), we observed many mTECs in which hAGF expression was confined to the nucleus as dot-like foci (Fig. 1B, 1C). Although the size of each dot tended to be bigger than authentic Aire nuclear dots, probably as a result of the high abundance of hAGF protein within the cell, the hAGF signals almost completely overlapped those of murine Aire (Fig. 1C), suggesting possible heterologous dimerization between hAGF and endogenous Aire. In contrast, most of the CD11c+ dendritic cells in the thymus were hAGF− (H. Kawano, H. Nishijima, and M. Matsumoto, unpublished observations), suggesting that Aire expression by thymic dendritic cells is negligible in comparison with that of mTECs, as we reported previously (8).
Most mTECs are committed to Aire expression
We first investigated Aire expression in adult wild-type C57BL/6 mice using conventional flow cytometry. Use of a rat anti-Aire mAb (clone RF33-1) produced in Mitsuru Matsumoto’s laboratory revealed that Aire+ cells accounted for 30–40% of all mTECs showing CD80high expression (Fig. 2A, top left panel) or MHC-IIhigh expression (Fig. 2A, middle left panel) after gating for CD45−EpCAM+UEA-1+ cells (mTECs) from the enzymatically digested stromal component; for convenience, a cutoff value for Aire expression was set using CD45−EpCAM+UEA-1− cells (cortical TECs [cTECs]) from the same animal (Fig. 2A, bottom left panel) or mTECs similarly isolated from Aire-deficient mice (e.g., Fig. 2D, lower right panel). However, it was noteworthy that there was no clear boundary between the Aire+ and Aire− cell populations in these analyses (Fig. 2A, left panels). Similar FACS profiles were obtained using another rat anti-Aire mAb (clone 5H12) (9) that has been widely used for the detection of Aire (H. Kawano and M. Matsumoto, unpublished observations). Use of the anti-Aire mAb revealed that mTECs isolated from adult heterozygous Aire/hAGF-KI mice had FACS profiles and percentages of Aire+ mTECs that were similar to those in wild-type mice (Fig. 2A, right panels), indicating that replacement of endogenous Aire with hAGF had no discernible effect on the expression of endogenous Aire.
Commitment of most mTECs to Aire expression. (A) Thymic stromal cells from adult wild-type mice (+/+) and heterozygous Aire/hAGF-KI mice (+/hAGF) were isolated enzymatically and evaluated for expression of Aire together with CD80 and MHC-II using flow cytometry after gating for CD45−EpCAM+UEA-1+ cells (mTECs) (upper and middle panels). FACS profile from CD45−EpCAM+UEA-1− cells (cTECs) isolated from the same wild-type mice is shown (lower left panel) to indicate a cutoff value for Aire expression. Aire+ cells were present as mTECs showing CD80high expression (upper panels) or MHC-IIhigh expression (middle panels). Similar percentages of Aire+ mTECs were observed between wild-type mice and heterozygous Aire/hAGF-KI mice using the anti-Aire mAb. Percentages of cells in the indicated regions are included. One representative experiment from a total of four repeats is shown. (B) Percentages of hAGF+ mTECs far exceed those of Aire+ mTECs detected by the anti-Aire mAb in heterozygous Aire/hAGF-KI mice. Flow cytometric analysis was performed using the same samples as in (A). One representative experiment from a total of four repeats is shown. (C) Coexpression of Aire and hAGF reporter in mTECs isolated from heterozygous Aire/hAGF-KI mice (upper right panel). FACS profile from cTECs isolated from wild-type mice is shown (lower left panel) to indicate a cutoff value for Aire expression. (D) More hAGF+ mTECs were observed in homozygous Aire/hAGF-KI mice (hAGF/hAGF) (upper right panel) than in heterozygous Aire/hAGF-KI mice (upper left panel). Approximately 80% of total mTECs were hAGF+ in homozygous Aire/hAGF-KI mice, accounting for >95% of the mTEChigh population. hAGF+ cells from homozygous Aire/hAGF-KI mice were negative for endogenous Aire (lower right panel). One representative experiment from a total of two repeats is shown.
Upon detailed inspection of TEC composition, the percentages of mTECs in the UEA-1+ population among the CD45−EpCAM+ total TEC fraction were indistinguishable between wild-type mice and heterozygous Aire/hAGF-KI mice (Supplemental Fig. 3). However, a significant increase in mTEChigh, together with a concomitant decrease in mTEClow, was observed in heterozygous Aire/hAGF-KI mice (Supplemental Fig. 3). Because Aire-deficient mice show an increase in mTEChigh (19), we suspect that the increase in mTEChigh in heterozygous Aire/hAGF-KI mice was also a reflection of the gene-dosage effect of Aire due to hypomorphic function of hAGF. Taken together, these results suggest that hAGF is a suitable surrogate marker of endogenous Aire, at least for monitoring of its expression.
Remarkably, upon detection of GFP (hAGF) signals from heterozygous Aire/hAGF-KI mice using flow cytometry, we found that ∼65% of the total mTEC population were hAGF+ (Fig. 2B, right panels); this accounted for up to 80% (e.g., 65.6/15.3 + 65.6 = 0.81) of the CD80high (mTEChigh) population, which was much higher than the percentage of Aire+ mTECs detected using anti-Aire mAbs (i.e., 45–55% of the mTEChigh population in Fig. 2A). Consistent with immunohistochemical analysis, a large proportion of hAGF+ cells coexpressed endogenous Aire (Fig. 2C, right panel), although many hAGF+Aire− mTECs were present if the cutoff value determined using cTECs from wild-type mice (Fig. 2C, lower left panel) or Aire-deficient mTECs (data not shown) was applied. These results suggested that detection of Aire+ mTECs with the anti-Aire mAb underestimated the population of bona fide Aire-expressing mTECs.
We also analyzed hAGF+ cells from homozygous Aire/hAGF-KI mice in which endogenous Aire was replaced by hAGF in both alleles. We found that >80% of total mTECs were hAGF+ (Fig. 2D, upper right panel), accounting for >95% (i.e., 80.2/2.9 + 80.2 = 0.97) of the mTEChigh population. The greater proportion of hAGF+ cells showing homozygous rather than heterozygous Aire/hAGF-KI expression may be due to the gene-dosage effect of hAGF. Indeed, we observed stronger GFP signals in the former than in the latter (compare the intensity of the GFP signals in Fig. 2D, upper panels). Alternatively, it is possible that the monoallelic expression of the Aire locus in some cells contributed to the phenomenon that we observed. However, it is important to note that expression of the reporter molecule from both alleles (i.e., homozygous Aire/hAGF-KI) represents a more physiological pattern of Aire expression compared with that from a hemiallele (heterozygous Aire/hAGF-KI). hAGF+ cells from homozygous Aire/hAGF-KI mice were negative for endogenous Aire, as expected (Fig. 2D, lower right panel). Thus, use of a novel Aire reporter strain to increase the sensitivity of detection of Aire+ mTECs suggested that Aire is most likely expressed by all mature mTECs.
Because the percentages of Aire+ mTECs detected by the anti-Aire mAb showed no major difference between wild-type mice and heterozygous Aire/hAGF-KI mice (Fig. 2A), the differences between the percentages of Aire+ mTECs defined by the anti-Aire mAb and Aire reporter (hAGF) expression in heterozygous Aire/hAGF-KI mice appeared to arise from the varying sensitivity in Aire detection. However, the specificity of hAGF+ cells as a surrogate for Aire+ mTECs needs to be validated. For this purpose, we used an anti-RANKL mAb that blocks de novo differentiation of mTECs from an immature state to a mature stage that includes Aire+ mTECs (20).
In vivo single-shot administration of the anti-RANKL mAb decreased the number of Aire+ mTECs detected by the anti-Aire mAb in both wild-type mice and heterozygous Aire/hAGF-KI mice as early as 4 d after injection, which was followed by a greater reduction 8 d after injection (Fig. 3A). When the kinetic properties of the disappearance of Aire+ mTECs at different time points were assessed (at days 4 and 8 after injection of anti-RANKL mAb), they were similar between wild-type mice (half-life = 3.2 d) and Aire/hAGF-KI mice (half-life = 3.7 d) (Fig. 3B, upper panel). These results further support the notion that replacement of endogenous Aire with hAGF had no discernible effect on the mTEC-differentiation program or endogenous Aire expression, although we acknowledge that a gene-dosage effect of Aire through the hypomorphic function of hAGF was evident (i.e., reduced TRA gene expression and increase in mTEChigh).
Validation of Aire reporter molecule expression as a surrogate of endogenous Aire expression from mTECs of Aire/hAGF-KI mice. (A) Disappearance of Aire+ mTECs from wild-type mice (+/+) (upper panels) and Aire/hAGF-KI mice (+/hAGF) (lower panels) upon treatment with anti-RANKL mAb was monitored 4 d (middle panels) and 8 d (right panels) after injection. PBS was injected as a control (left panels). Thymic stromal cells were gated for CD45−EpCAM+UEA-1+ cells. (B) Kinetics of the disappearance of Aire+ mTECs from wild-type mice (left panel) and of Aire+ mTECs (upper right panel) and hAGF (GFP)+ mTECs (lower right panel) from Aire/hAGF-KI mice upon treatment with the anti-RANKL mAb described in (A). Each symbol corresponds to an individual mouse. The half-life of Aire+ mTECs from wild-type mice detected by the anti-Aire mAb was calculated as 3.2 d (an exponential trend line is given, R2 = +0.83). Data are from a total of four experiments using 13 mice. The half-life of Aire+ mTECs from Aire/hAGF-KI mice, as detected by the anti-Aire mAb (upper right panel) and hAGF reporter (lower right panel), was calculated as 3.7 d (R2 = +0.79) and 5.8 d (R2 = +0.91), respectively. Data are from a total of four experiments using 13 mice.
Of note, the decay process of hAGF detected as a GFP signal (half-life was 5.8 d) was slower than that of endogenous Aire detected by the anti-Aire mAb in Aire/hAGF-KI mice (half-life = 3.7 d) (Fig. 3B, right panel), which explains, at least in part, the greater sensitivity of the detection of Aire+ mTECs by the hAGF reporter compared with the anti-Aire mAb. Taken together, the present findings suggest that our novel Aire reporter strain has achieved both improved sensitivity and high specificity for monitoring Aire+ mTECs and revealed the presence of an unexpectedly large population of Aire-expressing mTECs that had not been recognized previously using the anti-Aire mAb or other Aire reporter strains.
Long-term ablation of Aire+ mTECs results in dramatic loss of mature mTECs
Given that Aire+ mTECs do not represent a particular lineage(s) and that, in fact, most mTECs are committed to Aire expression during their differentiation program, it would be expected that continuous ablation of Aire+ mTECs results in the loss of most mature mTECs. We tested this hypothesis by establishing another strain of mice in which the DTR fused with GFP (21) was targeted to Aire+ mTECs. We generated Aire/DTR-KI mice using the same design as that for generating Aire/hAGF-KI mice through homologous recombination in ES cells (Fig. 4A). Immunohistochemical analysis of thymi from heterozygous Aire/DTR-KI mice demonstrated GFP signals in the cytoplasm and nucleus of Aire+ mTECs (Fig. 4B, 4C), as we observed for Aire/GFP-KI mice (8). Flow cytometric analysis of thymi from Aire/DTR-KI mice also showed a pattern of GFP expression similar to that from Aire/GFP-KI mice (Fig. 5A, untreated) (8). GFP+ mTECs were confined to the CD80high population (Fig. 5A, untreated upper), and costaining of mTECs with anti-Aire mAb from heterozygous Aire/DTR-KI mice demonstrated three types of cells, GFP−Aire+, GFP+Aire+, and GFP+Aire− mTECs (Fig. 5A, untreated, lower left panel), most likely due to a kinetic difference in the synthesis and degradation of GFP and Aire protein, as discussed above.
Establishment of Aire/DTR-KI mice. (A) Targeted insertion of the DTR-GFP fusion gene into the Aire locus by homologous recombination. Southern blot analysis and removal of the neor gene cassette were performed as described for Aire/hAGF-KI mice. (B) Cells expressing DTR-GFP fusion protein (green) were located within thymic medulla stained with anti-EpCAM mAb (red) by immunohistochemistry of a thymus section from an Aire/DTR-KI mouse. Scale bar, 100 μm. One representative experiment from a total of four repeats is shown. (C) Concomitant expression of DTR-GFP (green) and endogenous mouse Aire (red) assessed by immunohistochemistry of a thymus section from an Aire/DTR-KI mouse. Scale bar, 10 μm. One representative experiment from a total of four repeats is shown.
Long-term ablation of Aire+ mTECs in vivo results in loss of most mature mTECs beyond the fraction showing apparent Aire expression. (A) Single injection of 500 ng of DT i.p. into adult heterozygous Aire/DTR-KI mice caused almost complete loss of GFP+ mTECs and a dramatic decrease in Aire+ mTECs 2 d after treatment, followed by recovery at day 4. One representative experiment from a total of more than four repeats is shown. (B) Multiple i.p. injections of 500 ng of DT (every day for 5 d) into adult heterozygous Aire/DTR-KI mice effectively ablated GFP+ and/or Aire+ mTECs 1 d after treatment, while maintaining a significant proportion of the Aire−mTEChigh population. Thereafter, both Aire (DTR)+ mTECs and Aire−mTEChigh cells recovered by day 3. One representative experiment from a total of three repeats is shown. (C) Repeated injection of DT, but not PBS, every 3 d (six injections total) resulted in dramatic loss of mature mTECs beyond the fraction apparently expressing Aire. One representative experiment from a total of more than four repeats is shown.
We first injected 500 ng of DT i.p. as a single shot into adult heterozygous Aire/DTR-KI mice. We observed a nearly complete loss of GFP+ mTECs (Fig. 5A, Day 2, upper middle panel) and a dramatic decrease in Aire+ mTECs (Fig. 5A, Day 2, lower middle panel) 2 d after DT treatment. However, we started to see recovery of these cells at day 4 after injection (Fig. 5A, Day 4, right panels). We also injected DT every day for 5 d and analyzed the composition of mTECs to study the relationship between Aire+ mTECs (exiting in the mTEChigh fraction) and the Aire− mTEChigh population (Fig. 5B). Similarly to a single DT injection, we observed a dramatic decrease in GFP+ and/or Aire+ mTECs 1 d after multiple DT injections (Fig. 5B, Day 1, lower panel). In this case, we assume that the absence of Aire (DTR)+ mTECs had continued for 5 d based on the results obtained from a single DT injection (Fig. 5A). Nevertheless, a significant proportion (i.e., >50%) of the mTEChigh population not expressing Aire (Aire−mTEChigh) was present: the percentages of Aire−mTEChigh cells were similar to those after a single DT injection (compare Fig. 5A, middle panels and Fig. 5B, second panels from the left), and few GFP+ and/or Aire+ mTECs were noted in both cases. These results suggested that the presence of Aire+ mTECs was not an absolute requirement for maintenance of the Aire−mTEChigh population (see Discussion). Similarly to a single DT injection (Fig. 5A), we observed recovery of Aire (DTR)+ mTECs after multiple DT injections (Fig. 5B, Days 2 and 3).
Based on this successful ablation of Aire+ mTECs by short-term DT treatment, followed by a rather rapid recovery, we tried to ablate Aire+ mTECs over a longer period by injecting DT every 3 d for a total of six injections. We found that this treatment resulted in the dramatic loss of CD80high mTECs (Fig. 5C, Multiple DT, upper right panel), including all three cell types (GFP+Aire−, GFP+Aire+, and GFP−Aire+ mTECs) (Fig. 5C, Multiple DT, lower right panel). It is noteworthy that far fewer mTECs were recovered after treatment of Aire/DTR-KI mice with DT over this longer period in comparison with control mice, as evidenced by the intensity of the dots in the FACS profiles (Fig. 5C). Post-Aire mTECs that reside in the mTEChigh-intermediate population (19, 22) had been eliminated from the thymus during long-term ablation of Aire+ mTECs, and the remaining small population of CD80highGFP− mTECs was considered to be mature mTECs just prior to the onset of Aire expression (pre-Aire mTECs). A similarly dramatic decrease in mature mTECs after long-term ablation of Aire+ mTECs was noticed in another Aire/DTR-KI mouse generated through bacterial artificial chromosome transgenesis, although the reason for this phenomenon has not been pursued in depth (22).
Using FTOC, we further confirmed that the loss of most mature mTECs in Aire/DTR-KI mice resulting from continuous administration of DT in vivo is attributable to the commitment of all mTECs to Aire expression during their differentiation program. Embryonic thymi taken from wild-type mice and heterozygous Aire/DTR-KI mice at embryonic day 16.5 were cultured for 7 d without eliminating the thymocytes. For both wild-type mice and heterozygous Aire/DTR-KI mice, we were able to harvest more mTECs when embryonic thymi were cultured in the presence of recombinant RANKL (1 μg/ml) versus medium alone (Fig. 6). In contrast, when the medium was supplemented with DT (500 ng/ml), we observed a complete loss of GFP+ mTECs in heterozygous Aire/DTR-KI mice, as expected, together with a slight decrease in the mTEChigh population. When RANKL and DT were added together, we observed a dramatic loss of mature mTECs: very few CD80highGFP− pre-Aire mTECs remained in Aire/DTR-KI mice. We interpret this to be the result of efficient promotion of immature Aire− mTECs toward Aire-expressing mature mTECs by RANKL, with concomitant acquisition of sensitivity to DT through Aire/DTR expression. These results further support our view that all mTECs are, in principle, committed to Aire expression during their differentiation.
Promotion of immature mTECs by RANKL, followed by ablation of Aire-expressing mTECs induces the loss of most mature mTECs in FTOC. (A) Embryonic thymi taken from wild-type (+/+) and heterozygous Aire/DTR-KI (+/DTR) mice at embryonic day 16.5 were cultured for 7 d without eliminating the thymocytes (top panels). For both wild-type mice and Aire/DTR-KI mice, the addition of recombinant RANKL (1 μg/ml) promoted mTEC differentiation, together with an increase in the number of mTECs, compared with those cultured in medium alone. Addition of DT (500 ng/ml) caused complete loss of GFP+ mTECs, together with a slight decrease in the mTEChigh population, in Aire/DTR-KI mice. Addition of RANKL plus DT resulted in the loss of most mature mTECs in Aire/DTR-KI mice (bottom panels). One representative experiment from a total of three repeats is shown. (B) Changes in the percentages of the mTEChigh population (irrespective of GFP expression in Aire/DTR-KI mice) with each of the treatments shown in (A). One representative experiment from a total of three repeats is shown.
Discussion
We demonstrated that Aire expression is an inherent property of mTECs during their differentiation program. We consider that the variable levels of Aire expression evident among the mTEC population at any given time point, at least as revealed by the anti-Aire mAb, represent the heterogeneity of mTECs at various stage(s) of differentiation and/or in different cellular states. This view of Aire expression by mTECs may provide a novel insight into the roles of Aire in TRA gene expression. Given that Aire is strongly expressed at particular stage(s) of mTEC differentiation and that Aire can promote (8) or inhibit (23) the terminal differentiation of mTECs, lack of Aire might create mTECs at synchronized developmental stage(s) on a global scale. Because the spectrum of TRAs expressed by mTECs depends on their developmental status (22, 24, 25), any developmental shift toward a particular stage(s) due to a lack of Aire would result in a defect in the promiscuous gene expression that normally relies on the heterogeneity of mTECs in a developmental state. Similarly, if Aire has a role in adjusting the functional state to one that is optimal for TRA gene expression (26, 27), it is conceivable that loss of Aire in all mTECs within a certain developmental window for Aire expression would also result in a defect in promiscuous gene expression. Thus, based on the fact that all individual mTECs express Aire at least once in their lifetime, a defect in a single transcription factor (i.e., Aire) resulting in a global alteration in TRA gene expression in the total mTEC population might not be so unlikely.
In our series of experiments using Aire/DTR-KI mice, continuous ablation of Aire+ mTECs resulted in the loss of most mature mTECs, and we interpreted this as indicating that all mTECs go through a phase of Aire expression. However, it is also possible that the disappearance of most mature mTECs as a result of continuous treatment with DT was due to the requirement of Aire+ mTECs for maintenance of the Aire−mTEChigh population. In particular, only a few CD80highGFP− cells, probably representing “pre-Aire mTECs,” remained after quite efficient ablation of Aire+ mTECs upon addition of DT plus RANKL in FTOC. However, multiple injections of DT into Aire/DTR-KI mice in the short term created a situation in which a significant proportion of mTEChigh were present, despite the fact that Aire+ mTECs had been relatively sparse for a few days (demonstrated in Fig. 5B). Therefore, we consider that the disappearance of most mature mTECs as a result of DT treatment was not a bystander effect resulting from loss of Aire+ mTECs. Instead, together with the results of a complementary experiment using Aire/hAGF-KI mice, we suggest that Aire expression is inherent to all mTECs during their differentiation program.
We cannot rule out the possibility that Aire expression may also depend on mTECs being in a particular cellular state and that treatment with RANKL in the FTOC experiments may have created this cellular state prone to Aire/DTR expression, independently of any developmental cues. Similarly, the factor(s) that control Aire expression in mTECs in vivo, beyond its regulation by the developmental program, are still unclear (28), although interaction between self-antigens from mTECs and corresponding autoreactive T cells during negative selection was reported to promote Aire expression in mTECs (29).
Although we demonstrated that mTECs expressing Aire do not constitute a unique lineage(s) among mTECs, it is possible that Aire+ mTECs can be divided into different lineages based on their developmental and/or functional properties that are yet to be identified. Further studies are required to illuminate the complete picture of the developmental processes and/or mechanisms underlying the cellular heterogeneity of mTECs.
Disclosures
The authors have no financial conflicts of interest.
Footnotes
This work was supported in part by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science and the Ministry of Education, Culture, Sports, Science and Technology of Japan and by the Japan Agency for Medical Research and Development-Core Research for Evolutional Science and Technology (to M.M.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- cTEC
- cortical TEC
- DT
- diphtheria toxin
- DTR
- diphtheria toxin receptor
- EpCAM
- epithelial cell adhesion molecule 1
- ES
- embryonic stem
- FTOC
- fetal thymus organ culture
- hAGF
- human AIRE-GFP-Flag tag
- KI
- knockin
- MHC-II
- MHC class II
- mTEC
- medullary thymic epithelial cell
- TEC
- thymic epithelial cell
- TRA
- tissue-restricted Ag.
- Received April 29, 2015.
- Accepted October 2, 2015.
- Copyright © 2015 by The American Association of Immunologists, Inc.