Autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED), a monogenic recessive disease characterized by autoimmunity against multiple tissues, offers a unique possibility to study the breakdown of self-tolerance in humans. It is caused by mutations in the autoimmune regulator gene (AIRE), which encodes a transcriptional regulator. Work using Aire−/− mice suggests that Aire induces ectopic expression of peripheral Ags and promotes their presentation in the thymus. We have explored reasons for the difference between the comparatively mild phenotype of Aire-deficient mice and human APECED patients. We provide evidence that, unlike in the Aire−/− mice, in the patients a key mediator of active tolerance, the CD4+CD25+ regulatory T (Treg) cell subset is impaired. This was shown by significantly decreased expression of FOXP3 mRNA and protein, decreased function, and alterations in TCR repertoire. Also, in the normal human thymus a concentric accumulation of AIRE+ cells was seen around thymic Hassall’s corpuscles, suggesting that in the patients these cells may be involved in the observed Treg cell failure. In Aire−/− mice the expression of FoxP3 was normal and even increased in target tissues in parallel with the lymphocyte infiltration process. Our results suggest that a Treg cell defect is involved in the pathogenesis of APECED and emphasize the importance of active tolerance mechanisms in preventing human autoimmunity.
Autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED),3 or autoimmune polyendocrine syndrome I, is a rare, recessively inherited, monogenic disease enriched in populations such as the Finns, Sardinians, and Iranian Jews. It is characterized by autoimmune attack against multiple tissues, especially endocrine organs. In a Finnish cohort the final prevalence was estimated to be 88% for hypoparathyroidism, 84% for Addison’s disease, and 33% for diabetes (1). Other manifestations include gastric parietal cell atrophy, gonadal failure, and, in practically all patients, chronic mucocutaneous candidiasis. The onset of the disease usually takes place in early childhood, and most patients are symptomatic by the age of 10. The disease is caused by loss-of-function mutations in the autoimmune regulator gene (AIRE) (2, 3), structurally and functionally a transcriptional regulator that also contains a putative E3 ubiquitin ligase domain. AIRE is strongly expressed in thymic medullary epithelial cells (MECs) and at lower levels in many peripheral tissues, including spleen and lymph nodes (4, 5).
Because of its monogenic background, APECED offers a unique possibility to study the pathogenesis of organ-specific autoimmunity in humans. Work using Aire−/− mice suggests that Aire up-regulates the expression of peripheral tissue-specific genes, including prominent autoantigens such as insulin, in the thymic MECs, a process termed ectopic transcription. In Aire−/− mice the presentation of this set of peripheral Ags to developing T cells is disrupted, impairing negative selection and, in a transgenic model, allowing high-affinity autoreactive T cells to mature and migrate to the periphery (6, 7). More recently, it was shown that Aire also down-regulates many genes (8, 9), and appears to have direct effects on the Ag presentation machinery in the thymic MECs (10). Ubiquitin ligase activity by AIRE has also been reported (11), but these findings were subsequently challenged (12). The relative importance of these various putative functions of AIRE to the maintenance of tolerance is not known.
The Aire−/− mice display signs of immunological dysregulation, T cell hyperreactivity, and perturbations of TCR repertoire, and by the age of 3 mo most of them have autoantibodies and lymphocyte infiltrates in multiple organs (6, 13, 14). However, in striking contrast to AIRE−/− humans who invariably develop clinical autoimmunity, the Aire−/− mice remain clinically healthy or in a few cases develop manifestations not seen in human APECED. These manifestations include an inflammation of lacrimal glands in one mouse strain (14) and inflammation of the exocrine pancreas in Aire−/− NOD mice (15). Simultaneous expression of an Ag targeted to pancreatic islets and a transgenic TCR specific to it triggers diabetes in Aire−/− mice, but thymic expression of the model Ag was shown to be Aire-independent (10). Thymic expression of the lacrimal gland autoantigen was also not affected by Aire-deficiency, and in both cases the pathogenetic mechanism leading to autoimmunity remains unclear. These discrepancies raise questions as to what extent the murine model really reflects the pathogenesis of APECED and what factors are behind the obvious clinical differences between AIRE−/− humans and mice.
Another process potentially affected by loss-of-function mutations in AIRE is the development of CD4+CD25+ regulatory T (Treg) cells (16), which might be dependent on the thymic presentation of peripheral autoantigens. In mice, the deletion of these important mediators of active tolerance induces an organ-specific autoimmune syndrome that can be associated with endocrinopathies such as thyroiditis, adrenalitis, and diabetes (16). In humans, an impairment of Treg cells has recently been reported in autoimmune polyendocrine syndrome II (17), a syndrome of unknown and probably heterogeneous background, and Treg cells have also been explored in other autoimmune diseases (18). In Aire−/− mice, Treg cells develop in normal numbers, possess a normal phenotype, and have been reported to function normally (7, 10, 14). In contrast, a cross between scurfy mice, which lack Treg cells, and Aire−/− mice results in rapid lethality, showing that these two defects have a synergistic effect on murine autoimmunity (19).
In this study we show that in APECED patients the CD4+CD25+ Treg cell population is impaired. Our results extend the work done using Aire−/− mice to humans and suggest that the failure of Treg cells is important for the progression to clinical autoimmunity.
Materials and Methods
We studied 26 patients (14 women) with a mean age of 39.8 years (range 26–60) and 26 healthy controls (15 women) with a mean age of 40.2 years (range 23–65). APECED diagnosis was verified by sequencing the 14 exons and the exon-intron boundaries of the AIRE gene as described (20). The most common disease components were mucocutaneous candidiasis (26 of 26), Addison’s disease (22 of 26), hypoparathyroidism (19 of 26), hypogonadism (13 of 26), hypothyroidism (9 of 26), and diabetes (6 of 26); 21 were homozygous for the Finn-major mutation R257X. The patients received appropriate hormone replacement therapy of endocrine deficiencies at physiological doses, and in hypoparathyroidism normocalcemia was maintained with dihydrotachysterol and calcium. Because these treatments aim to restore normal hormonal levels, they should not have significant immunomodulatory effects. None of the patients received systemic immunosuppressive treatment at the time of the sampling. The HLA typing of our patients has been previously published (20). The HLA-matched control (Fig. 4⇓) was identified among donors studied in 1996 for acute viral infection (21) and was healthy at the time of the sampling for the current study. None of the subjects was pregnant, had acute infections, or received vaccinations at the time of the sampling. All subjects gave written informed consent, and the study was approved by the ethical committee of the Helsinki University Hospital (Helsinki, Finland).
Generation of C57BL/6 Aire−/− mice by targeted disruption of the murine Aire gene has been described (13). The controls were parental, syngeneic wild-type (WT) mice. The mice were kept in a specific pathogen-free barrier at the National Public Health Institute, Helsinki, Finland, and sacrificed at the age of 5 or 6 mo. Perfused organs were frozen in liquid nitrogen and homogenized in TriPure reagent (Roche) using an Ultra-Turrax apparatus (Janke & Kunkel). RNA isolation was conducted using an RNeasy midi kit (Qiagen). The study was approved by the ethical committee of animal use at the National Public Health Institute and the University of Helsinki (Helsinki, Finland).
Cell isolation and culture
PBMC were isolated by Ficoll-Hypaque gradient centrifugation (Amersham Biosciences) and cryopreserved in 90% FCS and 10% DMSO. Normal CD25high Treg cells were isolated from buffy coats obtained from healthy blood donors (Finnish Red Cross Blood Service, Helsinki, Finland) and used to test the ability of patient T cells to respond to Treg cell-mediated signals. Cell sorting and RNA extraction were performed on freshly isolated cells. PBMC were cultured in duplicate in a 96-well plate for 5 days at 2 × 105l-glutamine, 50 μM 2-ME, 100 μg/ml streptomycin, and 100 U/ml penicillin (supplements from Sigma-Aldrich). The PBMC were pulsed for the last 6 h with [3H]thymidine (1 μCi/well; Amersham Biosciences), harvested with a Skatron harvester, and analyzed with a liquid scintillation counter (Wallac). Background in nonstimulated wells was subtracted to obtain the specific proliferative response. Anti-CD3 mAb (50 μg/ml) was immobilized by incubating it in the culture well for 1 h, after which unbound mAb was washed away. Heat-killed Candida albicans (2.5 μg/ml) was prepared by incubating yeast cells in PBS at 56°C for 30 min. PHA (Sigma-Aldrich) was used at 5 μg/ml and human insulin (Orion Pharma) at 10, 2, or 0.4 μg/ml.
Cell separation, flow cytometry, and immunohistochemistry
22). Anti-CD25 mAb-coated Dynabeads (Dynal) were used to isolate CD25high cells for cell culture experiments. The captured cells were detached by overnight incubation at 37°C. The purity was determined with flow cytometry and typically ranged between 70 and 90%. Thymic tissue was obtained from children undergoing cardiac surgery. The removed tissue was immediately frozen in liquid nitrogen or fixed in formalin. For immunofluorescence, acetone-fixed (at −20°C for 20 min), 5-μm cryostat sections were stained with either goat anti-human AIRE (Santa Cruz Biotechnology) or anti-human AIRE mAb followed by appropriate FITC-labeled second-step reagents. For immunohistochemistry, 5-μm sections of the paraffin-embedded, formalin-fixed tissues were dewaxed, heated by microwave oven in a buffer containing 1 mM EDTA in 10 mM Tris-HCl (pH 9.0) for 10 min for Ag retrieval, and stained with the monoclonal anti-AIRE Abs and goat anti-mouse IgG Abs conjugated with HRP using the EnVision+ protocol (DakoCytomation). The densities of the AIRE+ cells in the thymic medulla at or in the near vicinity of Hassall’s corpuscles and at more remote medullary areas were estimated in sections stained by the indirect immunofluorescence or immunohistochemistry techniques. An Olympus BX50 microscope equipped with an Orca IIIm charge-coupled device camera (Hamamatsu Photonics) was used for analysis.
PCR and TCR repertoire analysis
TCR repertoire analysis was performed as described (2324). For quantitation of similarity, we adapted the method published by Gorochov et al. (25). The areas of all the individual peaks in a Vβ-Cβ profile were expressed as a percentage of the combined area of that profile and compared with corresponding values of the cord blood TCR repertoire. The absolute values of the differences of all the individual peaks were summed, divided by two, and subtracted from 100, giving the similarity of the analyzed repertoire with the cord blood repertoire. A fully overlapping repertoire would give a value of 100, whereas a value of 0 would signify a totally different and nonoverlapping repertoire.
DNA was isolated using a QIAamp blood kit (Qiagen), and the amount of TCR excision circles was estimated by a quantitative PCR-ELISA, as described (26).
Error bars indicate SD. p values were calculated with a two-tailed Student t test and correlations were determined using Pearson’s correlation coefficient, with a limit of p < 0.05 for significance. For p values of AIRE+ cell densities in the different areas of the thymus, the Mann-Whitney U test was used.
Suppressive function of CD4+CD25+ cells in APECED patients
In humans, CD25+ Treg cells form a subset of roughly 5% of all CD4+ T cells and have been reported to be those with the highest expression of CD25 (hereafter designated CD25high) (22, 27, 28). There was no difference in the frequency of CD25high cells or the intensity of CD25 expression between APECED patients and controls (Table I⇓). Therefore, with a coculture assay of isolated CD25high and CD25− cells we tested whether a functional impairment could be observed in the CD25high Treg cells of the patients. Unfractionated cells from patients and controls proliferated vigorously and equally well in response to polyclonal stimulation with immobilized anti-CD3 mAb or PHA. In controls at a ratio of one CD25high cell per three CD25− cells this response was diminished by 20–50%, whereas at the ratio of 1:1 it decreased by 85–95% (Fig. 1⇓). In contrast, coculture with CD25high cells from the patients failed to significantly suppress the response of CD25− cells to anti-CD3 mAb or PHA, consistent with a diminished suppressive capacity of their CD25high population. However, CD25high cells from both patients and controls were equally capable of suppressing the response to C. albicans, possibly because this is an Ag-specific response and therefore more limited than the polyclonal stimulation by PHA or anti-CD3 mAb. This result also showed that T cells from the APECED patients were not intrinsically resistant to suppressive signals from Treg cells. To further exclude this possibility, we isolated CD25high Treg cells from healthy blood donors and cultured them together with T cells from APECED patients or controls at a ratio of one per one cell. Treg cells obtained from a healthy donor were able to suppress anti-CD3-induced proliferative responses equally efficiently in patients and controls (a decrease of 79.5 ± 12.2 and 73.8 ± 23.2%, respectively). An attempt to test the effect of CD25high cells on responses against insulin, a peripheral Ag up-regulated by Aire in the murine thymus (6), was unsuccessful because insulin did not elicit a measurable proliferative response in any of the controls or patients.
Activation status of CD25high cells in APECED patients
The results described above suggested that, despite a normal frequency of CD25high cells, the Treg cell population in the patients was impaired. However, because CD25 is not expressed exclusively by Treg cells, it was important to exclude the possibility that an increased frequency of activated T cells in the patients could be responsible for the findings. There was no significant difference in the expression of HLA-DR or CD62L, markers associated with activation status, on CD3+ cells from the patients or controls. We then determined the expression of these markers in the CD25high population and also that of CD45RO, recently reported to differentiate CD25+ Treg cells from nonregulatory CD25+ T cells (28). Again, no significant differences between the patients and controls were observed (Table I⇑).
Next, we analyzed the Candida-specific proliferative response of CD25high and CD25− cells. Both patients and controls had an equally strong response to heat-killed C. albicans, reflecting the ubiquitous nature of this opportunistic pathogen. Because the patients had a chronic candidiasis, we reasoned that if the CD25high population in the patients consisted of activated cells, a significant fraction of the CD25high cells should be Candida-specific. In this case, depletion of CD25high cells would result in a decreased response to Candida, but no such diminution of the response was observed (Table I⇑). Depletion of CD25high cells had also no significant effect on polyclonal, anti-CD3-induced responses. Finally, we determined the TCR excision circle content of PBL from patients and controls. These extrachromosomal circles are not replicated along the chromosomal DNA, leading to their dilution during mitosis. Although variation was high in both groups, the TCR excision circle level in the patients was not significantly different from that of the controls (data not shown), indicating that the circulating cells had a roughly similar proliferative history.
Expression of FOXP3 in APECED patients
The most reliable Treg cell marker is FOXP3, a transcription factor expressed by >95% of human CD25high cells and closely associated with both the development and suppressive function of Treg cells (22, 28, 29). PBMC from the controls had 2-fold more FOXP3 mRNA than PBMC from the patients (Fig. 2⇓A). We then sorted the CD25high cells from six randomly chosen patients and six controls. As expected, FOXP3 mRNA was expressed at a much higher level in the CD25high cells than in the total PBMC in both patients and controls. However, CD25high cells from the controls expressed, on the average, four times more FOXP3 than CD25high cells from the patients (Fig. 2⇓B). Thus, the patients expressed consistently less FOXP3 mRNA, both in the CD25high subset and in the total T cell population.
To confirm the mRNA results, we next performed intracellular staining of FOXP3 protein followed by flow cytometric analysis. In the patients, the frequency of FOXP3+ cells was significantly lower than in the controls, although there was variation within both groups and the distributions overlapped (Fig. 2⇑C). Importantly, analysis of the CD25high cells in the patients showed that this population could not be divided into a FOXP3+ and FOXP3− fraction. Instead, the level of FOXP3 staining in the patients appeared as a single population but with decreased mean fluorescence intensity when compared with the controls (Fig. 2⇑, D and E). We also analyzed FOXP3 expression on CD45RA− cells (28) with similar results (Fig. 2⇑F). These findings further rule out a significant dilution of the CD25high population by activated cells.
TCR repertoire of CD25high cells in APECED patients
To obtain a more detailed view of the Treg cell defect in the patients, we analyzed the TCR β-chain repertoire of CD25high cells. It has been reported that Treg cells express a highly diverse TCR repertoire (30), a finding that our analysis of healthy controls confirmed (data not shown). Because HLA type can influence TCR V gene usage and distort results obtained from small numbers of cells, we first searched among a set of HLA-typed donors for a healthy control matched at the HLA class II locus with any of our patients. One such match was identified, a female (58 years old) with a shared class II HLA-type DRB1*0301; 01, DQA1*0101; 0501, DQB1*0201; 0501, DPB1*0401. The patient was a 26-year-old female, homozygous for R257X mutation in the AIRE gene, with candidiasis and parathyroid, adrenal, and ovarian failure as disease manifestations.
To obtain a reliable comparison, we sorted the same number of CD25high cells from the patient and the control and also adjusted the samples by quantitative β-actin amplification. The total TCR β-chain repertoire was amplified by using 24 different Vβ -specific primers followed by electrophoretic analysis. This method typically produces 6–10 bands for each V-gene, corresponding to in-frame rearrangements with different CDR3 lengths, and allows both a global view of the repertoire as well as sensitive detection of subtle changes (23). Reflecting the polyclonal nature of the Treg cell repertoire, the CD25high cells expressed most Vβ genes even though the sample size was only 50,000 sorted cells. However, when the patient was compared with the control a significant difference in the TCR diversity was observed. In the control the CDR3 length distribution revealed clonal expansions in most Vβ genes, while in the patient the repertoire was much less skewed (Fig. 3⇓A). The average number of different CDR3 lengths within individual Vβ genes was likewise significantly higher in the patient (Fig. 3⇓B). Finally, a comparison of CD25high cells from the APECED patient and the control to an average naive repertoire derived from 11 cord blood samples showed that the CD25high cell repertoire of the APECED patient was significantly closer to the naive repertoire than the repertoire of the control (Fig. 3⇓C).
To ascertain that these differences were not unique to the two individuals studied, we analyzed in a similar manner the repertoire of CD25high cells from four HLA-nonmatched patients and four controls. All samples were adjusted to 50,000 sorted cells. Comparison of the CDR3 length distribution confirmed the analysis of HLA-matched patient and control (Fig. 3⇑D). The control repertoires had several prominent expansions in most Vβ genes, while the patient repertoires were clearly less focused and closer to the average naive repertoire. Thus, the CD25high cell repertoire of the patients was more polyclonal and closer to a naive repertoire than that of the controls, suggesting a failure to select or shape a normal Treg cell repertoire.
Expression of AIRE in human thymus
The expression of AIRE in the thymic MECs is believed to play a role in the negative selection of autoreactive T cells (4, 5, 6, 7), but another structure, the epithelial Hassall’s corpuscle, seems to be important in the positive selection of human CD25+ Treg cells (31). We therefore examined the relationship of AIRE+ cells and Hassall’s corpuscles in four different thymic tissues. Thymic tissues stained by immunofluorescence or immunohistochemical techniques with a monoclonal anti-human AIRE showed AIRE+ cells dispersed within the thymic medulla and at the margins of Hassall’s corpuscles, although the corpuscles themselves were mostly AIRE− (Fig. 4⇓). The median density of AIRE+ cells at or in the near vicinity of Hassall’s corpuscles was 68.4 cells/mm2 (median range, 19.5–146.5), whereas the mean density in the more remote medullary areas was 26.1 cells/mm2 (median range, 13.0–78.1; p = 0.003). In total, 12 visual fields (3.6 mm2; magnification, ×200) were counted. Similar results were also obtained by using polyclonal anti-AIRE Abs (data not shown).
Expression of FoxP3 in Aire−/− mice
It has been reported that in Aire−/− mice CD4+CD25+ Treg cells develop in normal numbers and also function normally. To test whether a failure of Treg cell development takes place in the mouse strain mimicking the most common AIRE mutation (R257X; see Ref. 13), we analyzed FoxP3 mRNA levels in six Aire−/− and six WT mice but found no significant difference in either the thymus or the spleen (data not shown). Salivary glands, one of the target tissues of the autoimmune process, contained more T cells in the Aire−/− mice than in the WT mice as shown by a higher level of TCR Cα mRNA. In addition to these infiltrating T cells, the salivary glands from the Aire−/− mice also had a significantly higher level of FoxP3 mRNA than salivary glands from the WT mice. Importantly, there was a highly significant correlation between the level of TCR Cα and FoxP3 expression (Fig. 5⇓), indicating that the amount of FoxP3-expressing Treg cells increased in parallel with the infiltration process. Together with the previously published reports (7, 10, 14), these data indicate that there is no Treg cell defect comparable to that observed in APECED patients in the Aire−/− mice, either systemically or locally in the target organs.
The current view on the pathogenesis of APECED, based on studies in Aire−/− mice, holds that in the absence of Aire-induced ectopic transcription of peripheral Ags, negative selection in the thymus is impaired, leading to an increased frequency of autoreactive cells and autoimmunity (6, 7). However, the phenotype of Aire−/− mice is conspicuously mild, perhaps because Treg cells keep the autoreactive T cells in check. A detailed analysis of Treg cells in two Aire−/− mouse strains has shown that they develop and function normally (7, 10, 14), and our findings in the originally described Aire−/− strain are in agreement. Interestingly, the expression of FoxP3 in salivary glands increased in line with the immunopathological process, suggesting that the tissue was infiltrated by both inflammatory and Treg cells. This finding has a parallel in a recent study of NOD mice in which Treg cells were shown to participate in early insulitis and prevent the onset of T cell-mediated tissue destruction (32). Also, it has been suggested that the immunopathology developing in Aire−/− NOD mice might reflect the combined effect of failed central tolerance and defects of regulation intrinsic to the NOD strain (15), and a cross between Aire−/− and FoxP3−/− mice proved rapidly lethal (19). It is tempting to speculate that the difference in Treg cell function and the participation of Treg cells in the murine tissue infiltrates might explain why Aire−/− mice remain clinically healthy while AIRE−/− humans invariably develop APECED.
When studying abnormalities of the immune system, it can be difficult to separate the primary causative events from the secondary consequences of immunopathology. This is especially true of human disease, where the onset of the autoimmune process usually precedes clinical symptoms, often by several years. We therefore took care to exclude the possibility that the observed differences between patients and controls could be due to an increased frequency of activated CD25+ T cells diluting the Treg cells. Several lines of evidence make this unlikely. There was no significant difference in the expression pattern of CD25 or in the expression of several activation markers, either in the total T cell population or in the CD25high subset. This included CD45RO, a marker recently shown to distinguish regulatory from nonregulatory CD25+ cells in humans (28). The CD25high population was not enriched in Candida-specific T cells, and the proliferative history of circulating T cells was similar in both patients and controls. Importantly, even though FOXP3 levels were lower in the patients, intracellular FOXP3 staining showed that the CD25high population in the patients could not be divided into FOXP3− cells and FOXP3+ Treg cells. This result indicates that the CD25high cells of the patients were not measurably diluted by FOXP3−-activated cells that would be responsible for the observed differences. It therefore seems that the disease process in APECED, as in several other organ-specific autoimmune diseases, is restricted to the affected tissues and does not induce measurable systemic T cell activation (33).
How then does the lack of functional AIRE lead to a defect of Treg cells? Relatively little is known about the development of Treg cells in humans, but in transgenic mice the expression of an agonist peptide on thymic epithelium is sufficient for Treg cell generation (34). Other data indicate that thymic development of FOXP3+ Treg cells is driven by self-Ags, and the mature population is autoreactive (16, 29, 35, 36). The selecting Ags are thus likely to be the same ones that mediate negative selection of clones with a higher affinity. Because loss-of-function mutations in AIRE affect ectopic transcription in the thymus, at least in the mouse, the Treg cell defect in APECED patients might be a consequence of the lack of peripheral Ags in the thymus. Immature thymocytes with an intermediate affinity TCR would normally be positively selected to form the natural, Ag-specific Treg cell population but, in the absence of AIRE, do not receive a TCR-mediated signal and fail to mature. In this case the resultant Treg cell defect would be restricted to Ags regulated by AIRE, which might explain the organ specificity of the human APECED.
However, given the recently revealed complexities of the Aire-deficient phenotype in mice, the failure of ectopic transcription is by no means the only possibility. If AIRE-deficiency indeed leads to a generalized defect of thymic Ag presentation (10), the development of Treg cells could also be more generally affected. The failure of Treg cells isolated from APECED patients to suppress polyclonal responses induced by anti-CD3 mAb or PHA might well be an argument in favor of such a systemic defect not limited to any specific AIRE-controlled peripheral Ag. The TCR repertoire of Treg cells from patients was also quite different from that of the controls and in many ways less focused and closer to a naive repertoire. This marked difference in the clonal distribution of Treg cells in patients and controls suggests a wider aberration, because an Ag-specific failure would be expected to manifest as narrow holes in an otherwise normal repertoire. Also, AIRE is expressed in peripheral tissues (4, 5), albeit at lower levels than in the thymus, and it may be associated with the differentiation of dendritic cells or their interaction with T cells (37, 38, 39). Disruption of dendritic cell function could well contribute to the kind of defect observed in our analysis of TCR repertoire. The abnormal Treg cell repertoire in APECED patients might therefore reflect the failure of both thymic selection and peripheral activation and expansion of Treg cells.
It was recently shown that the epithelial cells at the margins of the Hassall’s corpuscles produce thymic stromal lymphopoietin, which induces the generation of CD11c+ dendritic cells (31). These dendritic cells, in turn, support the development of human Treg cells and, in the normal human thymus, CD25+ Treg cells are located exclusively in the medullary areas surrounding Hassall’s corpuscles, colocalizing with the thymic stromal lymphopoietin-producing epithelial cells. Interestingly, AIRE+ cells were also found around the Hassall’s corpuscles, so the loss of AIRE function at this location may be behind the observed broad Treg cell impairment in APECED patients. Especially suggestive is the fact that Hassall’s corpuscles are abundant in humans but rare in the murine thymus (40). Although it is indirect evidence, this interspecies difference might provide a mechanistic explanation for the difference in Treg cell function and clinical phenotype in Aire-deficient mice and APECED patients.
In conclusion, our data indicate that Treg cell failure is an important factor in the development of APECED. The presence of autoreactive T cells in healthy individuals is a well-known phenomenon and in itself is insufficient to break tolerance (41). Even in the absence of AIRE-regulated ectopic transcription of peripheral Ags in the thymus and a concomitant impairment of negative selection, no disease occurs without a defect in the Treg cell population. The well-defined monogenic background of APECED and analysis of the functional properties of the AIRE protein will help in further elucidating the balance between autoreactivity and immunoregulation in human autoimmunity.
We thank M. Schoultz and R. Väisänen for technical assistance, A. Banham and P. Peterson for the gift of mAb, and S. Meri and I. Ulmanen for comments on the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by the Academy of Finland, the Ahokas Foundation, the Helsinki University Science Foundation, research funds of the Helsinki University Central Hospital, the Sigrid Juselius Foundation, the Päivikki and Sakari Sohlberg Foundation, the Maud Kuistila Foundation, Finska Läkaresällskapet, and the European Union’s Sixth Framework Programme (European Association of Plastic Surgeons) Project LSHM-CT-2005-005223.
↵2 Address correspondence and reprint requests to Dr. T. Petteri Arstila, Haartman Institute, Department of Immunology, PB21, 00014 University of Helsinki, Finland. E-mail address:
↵3 Abbreviations used in this paper: APECED, autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy; AIRE, autoimmune regulator; Treg, regulatory T; MEC, thymic medullary epithelial cell; WT, wild type.
- Received April 19, 2006.
- Accepted October 26, 2006.
- Copyright © 2007 by The American Association of Immunologists