In Thymic Epithelial Differentiation Aire

The Journal of Immunology

Aire-Dependent Alterations in Medullary Thymic Epithelium Indicate a Role for Aire in Thymic Epithelial Differentiation1

Geoffrey O. Gillard,2* James Dooley,† Matthew Erickson,† Leena Peltonen,§ and Andrew G. Farr3*†‡

The prevalent view of thymic epithelial differentiation and Aire activity holds that Aire functions in terminally differentiated medullary thymic epithelial cells (MTECs) to derepress the expression of structural tissue-restricted Ags, including pancreatic endocrine hormones. An alternative view of these processes has proposed that Aire functions to regulate the differentiation of immature thymic epithelial cells, thereby affecting tissue-restricted Ag expression and negative selection. In this study, we demonstrate that Aire impacts several aspects of murine MTECs and provide support for this second model. Expression of transcription factors associated with developmental plasticity of progenitor cells, Nanog, Oct4, and Sox2, by MTECs was Aire dependent. Similarly, the transcription factors that regulate pancreatic development and the expression of pancreatic hormones are also expressed by wild-type MTECs in an Aire-dependent manner. The altered transcriptional profiles in Aire-deficient MTECs were accompanied by changes in the organization and composition of the medullary epithelial compartment, including a reduction in the medullary compartment defined by keratin (K) 14 expression, altered patterns of K5 and K8 expression, and more prominent epithelial cysts. These findings implicate Aire in the regulation of MTEC differentiation and the organization of the medullary thymic compartment and are compatible with a role for Aire in thymic epithelium differentiation. The Journal of Immunology, 2007, 178: 3007–3015.

lthough the importance of the thymic environment in The commonly held view of Aire function in MTECs is based governing the Ag specificity of the T cell repertoire and on the prevalent model of TRA expression by these cells, whereby A establishing self tolerance is well-established, the mech- terminally differentiated MTECs exhibit stochastic derepression of anistic bases for these processes remain poorly understood. The gene expression as a consequence of the opening and transcription discovery that mutations in a single gene, the autoimmune regu- of multiple chromatin domains mediated by a novel chromatin lator (AIRE), cause an autoimmune polyglandular syndrome in hu- remodeling mechanism (3). From this perspective, Aire would mans (1) led to the demonstration in mice that Aire activity is control expression of TRAs in mature MTECs by functioning as a required for the expression of a diverse subset of tissue-restricted “randomizer” of gene expression (2). However, this model does 4 Ags (TRAs) by murine medullary thymic epithelial cells not account for the TRAs whose expression by MTECs is Aire (MTECs), and that Aire contributes to the ability of cells within the independent (2, 4) or for the impaired negative selection of OVA- thymic microenvironment to enforce negative selection (2). specific T cells in AireϪ/Ϫ mice bearing insulin promoter-driven OVA, where Aire deficiency does not affect levels of OVA trans- gene expression (5). Furthermore, the high level of mitotic activity high *Department of Immunology, †Department of Biological Structure, and ‡Institute for in the MHC class II subset of MTECs (6) is difficult to recon- Stem Cell and Regenerative Medicine, University of Washington, Seattle, WA 98195; cile with the view that this population consists of terminally dif- and §Department of Medical Genetics, University of Helsinki, Department of Mo- ferentiated MTECs (4). lecular Medicine, National Public Health Institute, Biomedicum, Helsinki, Finland, and The Broad Institute, Massachusetts Institute of Technology, Boston, MA An alternative model proposes that TRA expression reflects the Received for publication September 20, 2006. Accepted for publication December transcriptional activity of immature MTECs undergoing differen- 12, 2006. tiation (7). Recent demonstrations that the thymic epithelium (TE) The costs of publication of this article were defrayed in part by the payment of page is a dynamic population undergoing significant turnover in the charges. This article must therefore be hereby marked advertisement in accordance adult thymus (6, 8, 9) indicate that the MTECs in the adult thymus with 18 U.S.C. Section 1734 solely to indicate this fact. must represent a collection of cells at different stages of differen- 1 This work was supported by the National Institutes of Health (Grants AI24137 and AI 059575). G.O.G. was supported in part by training grants from the National In- tiation, including immature progenitor cells. According to this stitutes of Health and the Cancer Research Institute. L.P. received support from the view, TRA expression could reflect the transcriptional activity of European Union-funded project EURAPS (LSHM-CT-2005-005223) and the Center of Excellence of Complex Disease Genetics of the Academy of Finland. relatively undifferentiated epithelial cells that have not yet become specified to a thymic fate or could reflect alternative programs of 2 Current address: Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Boston, MA 02215. epithelial differentiation that yield MTECs sharing phenotypic 3 Address correspondence and reprint requests to Dr. Andrew G. Farr, Department of properties with other epithelial derivatives (10). This model pro- Biology Structure, Box 35-7420, University of Washington, Seattle, WA 98195-7420. poses a very different role for Aire in the thymus, where Aire E-mail address: [email protected] would be involved in the differentiation of MTECs, and where the 4 Abbreviations used in this paper: TRA, tissue-restricted Ag; MTEC, medullary thy- Ϫ/Ϫ ϩ ϩ altered spectrum of TRA expression observed in Aire mice mic epithelial cell; TE, thymic epithelium; WT, wild type; CMF, Ca2 - and Mg2 - free; RQ, relative quantity; Gip, glucose-dependent insulinotrophic peptide; LTR, would be a consequence of this altered differentiation program. lymphotoxin receptor; Epcam, epithelial cell adhesion molecule; HPRT, hypoxan- These two models make very different predictions regarding the thine phosphoribosyltransferase; UEA, U. europeus agglutinin. impact of Aire deficiency on MTEC differentiation. If Aire has a Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 role limited to modulating the transcriptional activity of terminally www.jimmunol.org 3008 Aire DEFICIENCY ALTERS MEDULLARY TE

Table I. Contraction of the medullary epithelial compartment in the AireϪ/Ϫ thymusa

Age at Analysis, Aire Medulla-Cortex Medulla as % of No. Sections Thymic Area in Weeks Genotype Ratio Whole Thymus Analyzed Analyzed (mm2)

4 ϩ/ϩ 0.191 Ϯ 0.047 15.95 ϩ 3.15 28 119.5 Ϫ/Ϫ 0.125 Ϯ 0.030 11.09 ϩ 2.33 30 161.1 8 ϩ/ϩ 0.315 Ϯ 0.067 23.62 ϩ 4.20 19 84.3 Ϫ/Ϫ 0.161 Ϯ 0.045 13.58 ϩ 3.47 23 120.6 10 ϩ/ϩ 0.172 Ϯ 0.040 14.58 ϩ 2.89 21 50.0 Ϫ/Ϫ 0.092 Ϯ 0.030 8.36 ϩ 2.41 23 63.8 12 ϩ/ϩ 0.174 Ϯ 0.058 14.64 ϩ 4.06 14 53.7 Ϫ/Ϫ 0.109 Ϯ 0.023 9.77 ϩ 1.86 15 71.6 12 ϩ/ϩ 0.183 Ϯ 0.040 15.53 ϩ 2.72 22 90.1 Ϫ/Ϫ 0.091 Ϯ 0.039 8.41 ϩ 3.25 19 84.7 16 ϩ/ϩ 0.234 Ϯ 0.049 18.89 ϩ 3.28 21 57.3 Ϫ/Ϫ 0.094 Ϯ 0.029 8.64 ϩ 2.15 21 62.8

a Sections were processed as described in Materials and Methods to quantify areas of 3G10 reactivity. Values for “Med- ullary-Cortex Ratio” and “Medulla as % of Whole Thymus” represent mean values Ϯ SEM derived from the indicated number of sections. Differences between pooled Aireϩ/ϩ and AireϪ/Ϫ samples (n ϭ 6) or between age-matched samples were statistically signiﬁcant ( p Ͻ 0.05 and p Ͻ 0.001, respectively).

differentiated MTECs, Aire deficiency should not impact the over- Materials and Methods all organization or composition of the MTEC compartment, as Mice proximal aspects of their differentiation program would be Aire Aire-deficient mice were generated as previously described (11) and main- independent. Furthermore, a direct role for Aire regulation of TRA tained on a B6/129 background. An independently generated line of Aire- expression in the thymus would obviate the requirement for the deficient mice (2) maintained on a C57BL/6 background was obtained transcriptional regulation that normally controls their expression in from The Jackson Laboratory. Adult C57BL/6 mice were obtained from corresponding extrathymic tissues. In contrast, the developmental Charles River Laboratories or from our colony. All mice were maintained in the University of Washington specific pathogen-free facility and used in model predicts that Aire deficiency would be associated with a accordance with protocols approved by the University of Washington In- more global perturbation of the medullary epithelial compartment stitutional Animal Care and Use Committee. because it would impact MTEC differentiation and/or survival. If Immunohistochemistry the transcriptional hierarchies that regulate extrathymic epithelial Immunohistochemistry was performed on thymi from multiple littermates differentiation/TRA expression are conserved by MTECs, the de- Ϫ Ϫ of both lines of Aire / mice and appropriate age-matched control (wild- velopmental model further predicts that MTECs would also ex- type (WT) littermates or C567BL/6) mice as previously described. Re- press these transcriptional factors in addition to structural TRAs. agents used in these analyses included G8.8 (available through the Devel- Accordingly, the selective reduction of TRA expression by opmental Studies Hybridoma Bank; www.uiowa.edu/ϳdshbwww/), 3G10 MTECs in AireϪ/Ϫ mice would be accompanied by reduced expres- ERTR4, ERTR5, Troma1 (Developmental Studies Hybridoma Bank), polyclonal rabbit anti-K5 (Covance), anti-MHC class II, anti-CD40, 10.1.1, sion of corresponding upstream regulatory transcription factors. biotinylated Ulex europeus (Vector Laboratories). In this study, we have tested some of these corollary hypotheses For immunofluorescence, the following secondary Abs were used: goat regarding Aire function and TE differentiation. We report here that anti-rat IgG Alexa 488, goat anti-rat Alexa 546, goat anti-rabbit IgG Alexa 546, and streptavidin-Alexa 488 (all from Molecular Probes). For three- Nanog, Oct 4, and Sox2, transcription factors critical in maintain- color analysis, goat Abs specific for rat IgM ␮-chain or rat IgG ␥-chains ing the multipotentiality of progenitor cells and candidates for ex- (Pierce) were conjugated with Alexa 647 or Alexa 488, respectively, ac- pression by epithelial progenitor cells in the thymus, are expressed cording to the manufacturer’s protocol. For immunoperoxidase detection of by MTEC in a highly Aire-dependent manner. We also show that, 3G10, a three-step procedure was used, where unconjugated 3G10 Abs were detected by sequential application of goat anti-rat IgM ␮-chain Abs in addition to pancreatic endocrine TRAs, such as insulin or glu- (Pierce) modified by N-hydroxysuccinimide-digoxigenin (Boehringer cagon, MTECs from normal thymi also express many of the reg- Mannheim), followed by peroxidase-conjugated sheep anti-digoxigenin Ј Ј ulatory transcription factors that are centrally involved in the de- F(ab )2 Ab (Boehringer Mannheim) and color development with 3,3 dia- minobenzidine (Sigma-Aldrich). velopment of the endocrine and exocrine compartments of the pancreas. Aire deficiency leads to dramatic reductions in the ex- Morphometry pression of most of these pancreatic regulatory transcription fac- Thymic sections were collected at 70-␮m intervals through both lobes of tors, suggesting that Aire may play a role in regulating MTEC age-matched Aireϩ/ϩ and AireϪ/Ϫ thymi. After processing to demonstrate expression of pancreatic TRAs in a more conventional manner, 3G10 staining as previously described (12), the sections were photo- graphed at ϫ25 magnification and images were printed at the same mag- perhaps by acting as a proximal “master” control of more distal nification. Thymic lobe profiles were cut out and weighted, and then the regulatory gene hierarchies in developing MTECs or by impacting medullary compartment identified by 3G10 staining was cut out and the the differentiation and/or survival of the cells that express these resulting pieces were also weighed. The weight of the paper was calibrated to image area by weighing squares of paper that corresponded to areas genes. In this study, we have also documented alterations in the defined by a stage micrometer. Numbers of sections analyzed and the area Ϫ Ϫ organization and character of medullary epithelium in the Aire / that these sections encompassed are listed in Table I. Statistical signifi- thymus. These data establish that the impact of Aire deficiency cance was determined by the Wilcoxon rank-sum test. extends beyond the lack of expression of a subset of TRAs in Sorting and flow cytometry mature MTEC and effects changes in the thymic environment that Enzymatic dissociation of thymi was performed as described previously would be compatible with a role for Aire in MTEC differentiation (8). Briefly, thymi from 4- to 10-wk, age- and gender-matched mice were and/or survival. diced in HBSS (Ca2ϩ and Mg2ϩ-free (CMF)) plus 2% FBS plus 10 mM The Journal of Immunology 3009

FIGURE 2. Aire-deficient MTEC fail to express multiple pancreatic endocrine hormone genes. A, RT-PCR characterization of MTEC population used for this analysis. The flow cytometric characterization of these cells has been described (8). Consistent with their medullary character, these cells express K5, MHC class II, and Aire. Aire signal was undetectable in samples from AireϪ/Ϫ mice. Sample dilutions: undiluted, 1/5 and 1/25. B, Semiquantitative RT-PCR analysis of pancreatic endocrine gene products expressed by Aireϩ/ϩ and Ϫ/Ϫ MTEC. a Gcg, Glucagon; Ins2, insulin 2; Ppy, pancreatic polypeptide; SST, somatostatin. Samples were normalized to HPRT (see A) or Epcam, which gave equivalent results (data not shown). FIGURE 1. Transcription factors associated with multipotentiality are Sample dilutions: undiluted, 1/5 and 1/25. C, Quantitative real-time PCR expressed by MTEC in an Aire-dependent manner. A, Flow cytometric ϩ/ϩ ϩ analyses of pancreatic endocrine gene produces expressed by Aire and separation of enzymatically dissociated G8.8 thymic stromal cells from Ϫ Ϫ Aire / MTEC. Samples labeled as in B. Data represent three independent Aireϩ/ϩ and AireϪ/Ϫ thymi. Profiles have been gated on CD45Ϫ cells. samples. SD for any sample was Ͻ2% from the average value. Although the representation of G8.8ϩCD45Ϫ epithelial cells from these two sources appear roughly equivalent, variability between experiments and the low sensitivity of directly conjugated anti-cortical reagent Abs do not allow conclusions to be drawn regarding the relative representation of TE subsets. B, Semiquantitative RT-PCR analyses of sorted Aireϩ/ϩ and Alexa 647. FITC-conjugated anti-digoxigenin Abs (Boehringer Mann- AireϪ/Ϫ MTEC. Sample dilutions: undiluted, 1/5 and 1/25. C, Quantitative heim) were used to detect digoxigenin-modified Abs. real-time PCR analysis of sorted Aireϩ/ϩ and AireϪ/Ϫ MTEC. Expression levels were normalized to Epcam and are represented as the percentage of Isolation of RNA WT expression. Data represent three independent samples. SD for any Techniques for RNA isolation, amplification, and PCR analyses have been sample was Ͻ2% from the average value. D, Semiquantitative RT-PCR described previously (8). Primer sequences are available upon request. De- analyses of sorted Aireϩ/ϩ and AireϪ/Ϫ MTEC. Sample dilutions: undi- termination of relative quantity (RQ) values has been described previously luted, 1/5 and 1/25. LtBr, Lymphotoxin ␤ receptor. (13) and SDs of real time data were calculated according to the manufacturer’s instruction (www.appliedbiosystems.com).

HEPES buffer, washed by gravity sedimentation, and then digested with Results collagenase D (Roche) in CMF HBSS, followed by a mixture of collagenase Aire-deficient MTECs express reduced levels of transcription and neutral dispase (Roche) to obtain a single-cell suspension. Single-cell factors that confer multipotentiality suspensions were treated with rat anti-FcR II/III mAb (clone 24G2) before Ab labeling to reduce nonspecific Ab binding. After staining, cells were The presence of a progenitor MTEC population in the adult thy- washed twice in HBSS CMF and suspended in the same containing 7-ami- mus has been inferred by demonstrations of cells with progenitor noactinomycin D (Molecular Probes), at a final concentration of 2 ␮g/ml. activity in the fetal thymus (14–16) and recently demonstrated in Cells were sorted directly into TRIzol (Invitrogen Life Technologies). Pri- mary Abs included G8.8 (antiepithelial cell adhesion molecule (Epcam)) adult thymus (17), although the full developmental potential of conjugated with digoxigenin, a PE-conjugated anti-CD45 (eBioscience), a these cells has not been determined. We have previously estab- mixture of purified anti-BP-1 Abs (clones 6C3 and CDR1) conjugated to lished that the Nanog, Oct4, and Sox2 core transcriptional factors 3010 Aire DEFICIENCY ALTERS MEDULLARY TE

FIGURE 3. Aire dependence of the transcription factor hierarchy control- ling pancreatic endocrine development. A, A schematic diagram of pancreatic development and some of the transcription factors that regulate that process. The pancreatic phenotypes of mutant mice lacking these transcription factors are listed. All of the indicated transcription factors are also expressed by Aireϩ/ϩ MTEC. B, Semiquantitative RT-PCR analyses of pancreatic transcription factors expressed by Aireϩ/ϩ and AireϪ/Ϫ sorted MTEC (CD45ϪCDR1ϪEpcamϩ). Samples were undiluted, 1/5, and 1/25. C, Quantitative real-time PCR analysis of pancreatic transcription factors expressed by Aireϩ/ϩ and AireϪ/Ϫ sorted MTEC (CD45ϪCDR1ϪEpcamϩ) processed and presented as in Fig. 1. Data represent three independent samples. SD of any sample was Ͻ2% from the average value.

of multipotentiality (18, 19), are expressed within the MTEC pop- dependent manner (Fig. 2, B and C). The endocrine pancreatic ulation and have proposed this to reflect the presence of a devel- compartment was represented by glucagon, pancreatic polypep- opmentally flexible progenitor epithelial population in the adult tide, somatostatin, and glucose-dependent insulinotrophic peptide thymus (8). To explore the hypothesis that Aire could influence TE (Gip). Gip is Aire dependent and is regulated in the intestine/gas- differentiation, we determined the impact of Aire deficiency on the trointestinal cells by Pdx1, a transcription factor required for pan- expression of these transcription factors by MTECs. Medullary creatic development. Normalizing levels of pancreatic gene ex- epithelial cells were isolated from enzymatically dissociated thymi pression to levels of hypoxanthine phosphoribosyltransferase by flow cytometry. A representative set of Aireϩ/ϩ and AireϪ/Ϫ (HPRT) or Epcam (a reliable marker of MTEC), we found that thymic samples is shown in Fig. 1A. Although this approach al- Aire-deficient MTEC expressed virtually no pancreatic polypep- lowed recovery of a defined subset of MTEC, variability of cell tide, somatostatin, or Gip; the levels of glucagon and insulin2 ex- recovery and variable efficiency of MTEC recovery do not allow pressed by MTEC were reduced to ϳ40 and 10% of WT levels, conclusions regarding relative representation of TE subsets based respectively. on this approach with our current methodologies. That being said, If the general mechanism that regulates expression of pancreatic we did not observe any reproducible differences between Aireϩ/ϩ TRAs by pancreatic epithelial cells is conserved in MTEC, the and AireϪ/Ϫ thymi prepared in this manner. Fig. 1, B and C,de- regulatory hierarchy of transcription factors that control pancreatic picts results of semiquantitative and real-time PCR of MTEC cD- endocrine development should also be expressed by MTEC. Some NAs from enzymatically dissociated Aireϩ/ϩ and AireϪ/Ϫ thymi. of the transcription factors that are critical for pancreatic develop- These results confirmed the expression of Nanog, Oct4, and Sox2 ment are shown in Fig. 3A and are described below. Pdx1, along ϩ ϩ by Aire / MTECs (8) and revealed a dramatic reduction in their with Ptf1a, is required for pancreatic organogenesis; Pdx1 also has Ϫ Ϫ expression within the Aire / MTEC population. We performed a critical role in later stages of endocrine cell development, par- similar analyses of the expression of other transcription factors ticularly for the development and function of mature ␤ or ␦ cells previously implicated in thymic organogenesis (FoxN1, Pax1, in the islets (22–24). Expression of Ptf1a is also required at later Pax9, and lymphotoxin-␤ receptor (LT␤R); reviewed in Refs. 20 stages in the differentiation of exocrine lineages (25). Neurogenin and 21) or in the differentiation of multiple epithelial lineages 3 is a transcription factor that is required for the development of all (Foxa1 and Foxa2; referenced in Ref. 8) and determined that four endocrine lineages in pancreas (26), while NeuroD1 expres- ϩ ϩ Ϫ Ϫ Aire / and Aire / MTECs demonstrated comparable expression is required for the establishment of mature endocrine cells, sion levels of these transcription factors (Fig. 1D). Based on results particularly ␤ cells (27). The main phenotype in Nkx2.2- and of the dilution series analyses, real-time PCR analyses were not Nkx6.1-deficient mice is a severe decrease in the production of performed for these latter groups of molecules. insulin, which has been attributed to impaired maturation of ␤ cells (28, 29). The paired-box transcription factors, Pax4 and Pax6, also Aire deficiency leads to severe reductions in the pancreatic play important roles in the specification of the four endocrine lin- transcriptional pathway eages. Pax6 expression by endocrine progenitor cells is important The properties of the MTEC populations analyzed here have been for the formation of all endocrine lineages, particularly ␣ cells described previously (8) and are also shown in Fig. 2A. Confirming (30), while Pax4 is required for the formation of ␤ and ␦ cells (31). previous reports (2, 4), real-time quantitative PCR analyses of We analyzed cDNAs generated from bulk populations of cDNA derived from sorted Aireϩ/ϩ and AireϪ/Ϫ MTEC showed EpcamhighCD45Ϫ MTECs sorted from groups of age-matched that that MTEC express a number of genes characteristically as- adult Aireϩ/ϩ mice for the expression of a subset of these pan- sociated with the endocrine portion of the pancreas in an Aire- creatic developmental transcription factors. As shown in Fig. 3B, The Journal of Immunology 3011

Ptf1a, Ngn3, NeuroD1, Nkx2.2, Nkx6.1, Pax4, and Pax6. A similar analysis of sorted MTEC from AireϪ/Ϫ mice revealed that expression of most of these pancreatic transcription factors was dramat- ically reduced in Aire-deficient MTEC, with the exceptions of Ptf1a and NeuroD1, which displayed Ͻ2-fold differences. These semiquantitative RT-PCR results were confirmed by real-time PCR analyses (Fig. 2C), where expression of Pdx1, Nkx2.2, Nkx6.1, Pax4, and Pax6 by AireϪ/Ϫ MTEC was reduced to ϳ5% of WT levels. It is noteworthy that severe reductions in Pdx1 expression by the Aire-deficient MTEC population was mirrored by severe reductions in the expression of three endocrine structural genes, Ins, Sst, and Gip, that are directly regulated by Pdx1 and whose promoters are binding targets for Pdx1 (32–34). These results demonstrate that the genes affected by Aire deficiency in MTEC extend beyond structural products of peripheral lineages to include the regulatory elements that control expression of these genes in peripheral tissues.

The organization and composition of the medullary thymic epithelial compartment is impacted by Aire deficiency Previous analyses of the AireϪ/Ϫ thymus concluded that the organization was unremarkable, based on distinct cortical and medullary compartments in histologic samples (2, 11, 35). Here, we have used immunohistochemistry to analyze the organization and composition of the Aire-deficient thymic stromal compartment in more detail. In the Aireϩ/ϩ thymus, high levels of Epcam expression delin- eated the medullary epithelial compartment (Fig. 4, A and E). Within this medullary compartment, additional heterogeneity in the Aireϩ/ϩ thymus was demonstrated by reactivity with the fucose-specific U. europeus agglutinin (UEA) (36) (Fig. 4, B and E) or with the mAb 10.1.1, which reacts with a subset of MTEC (37) (Fig. 4G). The staining patterns observed with UEA (Fig. 4B) and 10.1.1 (Fig. 4H)intheAireϪ/Ϫ thymus indicated that the repre- FIGURE 4. Immunohistochemical analyses of MTEC heterogeneity in sentation of MTEC heterogeneity was fundamentally altered. Al- the Aireϩ/ϩ (A, C, E, and G) and AireϪ/Ϫ (B, D, F, and H) thymus. A and ϩ/ϩ though the representation of MTEC that expressed low levels of B, Immunofluorescence of Epcam expression (red) in Aire (A) and ϩ/ϩ Ϫ/Ϫ AireϪ/Ϫ (B) thymus. C and D, Immunofluorescence of UEA binding UEA staining were comparable in Aire and mice, the clus- (green) in Aireϩ/ϩ (C) and AireϪ/Ϫ (D) thymus. E and F, Merged images ters of MTEC strongly labeled with this lectin was not detected in Ϫ/Ϫ of Epcam expression (red) and UEA binding (green) in Aireϩ/ϩ (E) and the Aire thymus (compare Fig. 4, C and D). Similarly, the Ϫ Ϫ AireϪ/Ϫ (F) thymus. G and H, Merged images of Epcam expression (red) Aire / thymus displayed a marked reduction in the frequency of and 10.1.1 binding (green) in Aireϩ/ϩ (G) and AireϪ/Ϫ (H) thymus. All 10.1.1ϩ cells (compare Fig. 4, G and H). micrographs same magnification. Bar, 225 ␮m. The mAb 3G10, which reacts with medullary TE and has been provisionally identified as an anti-K14 mAb (12) defines the major we found that normal MTEC expressed many of the transcription population of MTEC in both WT and AireϪ/Ϫ thymus. The prom- factors that are involved in the specification and/or differentiation inence of epithelial cysts and the less dense medullary compart- of endocrine and exocrine pancreatic lineages, including Pdx1, ment in the AireϪ/Ϫ thymus is evident in thymic sections labeled

FIGURE 5. Immunoperoxidase demonstration of the 3G10ϩ compartment of the Aireϩ/ϩ (A and B) and AireϪ/Ϫ (C and D) thymus. A, Low magnification view of Aireϩ/ϩ thymus. Multiple islands of medullary epithelium are evident. A corticomedullary epithelial cyst is indicated by an asterisk. B, Higher magnification of Aireϩ/ϩ thymus to demonstrate the dense and uniform distribution of 3G10ϩ cells within the medulla. Epithelial cyst is indicated by asterisk. C, Low magnification view of AireϪ/Ϫ thymus. Multiple islands of medullary epithelium are evident. Several corticomedullary epithelial cysts are indicated by asterisks. D, Higher magnification of AireϪ/Ϫ thymus to demonstrate a looser and more heterogeneous distribution of 3G10ϩ cells and expansion of the cystic epithelial compartment (indicated by asterisks) within the medulla. A and C, Bar: 225 ␮m; B and D, bar: 775 ␮m. 3012 Aire DEFICIENCY ALTERS MEDULLARY TE

Table II. Reduced expression of keratin 14 message by AireϪ/Ϫ MTECa thymi examined in this manner and was statistically significant in all instances ( p Ͻ 0.05 between Aireϩ/ϩ and Ϫ/Ϫ groups and p Ͻ Keratin Set Aireϩ/ϩ AireϪ/Ϫ 0.001 between each age-matched pair). These alterations were ev- 8 1 1.00 1.28 ident in mice younger than 12 wk of age (when autoimmune ac- Ϫ Ϫ 2 1.00 1.22 tivity is first detected in Aire / mice; Ref. 2), and as early as 1 wk 14 1 1.00 0.24 of age (data not shown), so it is unlikely that the alterations of the 2 1.00 0.61 medullary compartment are secondary to the onset of autoimmune Ϫ Ϫ ϩ ϩ ϩ Ϫ Ϫ a Sorted CD45 CDR1 Epcam TE cells from Aire / and / mice were pro- phenomena. Real-time PCR analyses of K8 and K14 expression cessed to generate cDNA and then analyzed with real-time RT-PCR to determine high Ϫ levels of K8 and K14 message expression. RQ values were determined, setting the in two independent sorted populations of Epcam CDR1 cycle threshold of the Aireϩ/ϩ sample to 1. CD45ϪMTEC from AireϪ/Ϫ and ϩ/ϩ thymi confirmed the trends in keratin expression observed immunohistochemically. Matched cDNA samples were first normalized to the expression of Epcam with 3G10 (compare Fig. 5, A and B with C and D). Morphometric analyses of multiple sections of Aireϩ/ϩ and Ϫ/Ϫ thymic tissue that and then RQ values were obtained for K8 and K14. In both sets of Ϫ/Ϫ were processed to demonstrate 3G10 with peroxidase-labeled Abs sorted MTEC, the Aire cells displayed a smaller RQ value for revealed a marked alteration of the medullary compartment of K14, indicating a fundamental alteration in the character of the Ϫ/Ϫ AireϪ/Ϫ thymus. As shown in Table I, AireϪ/Ϫ thymi had a smaller Aire MTEC (Table II). 3G10ϩ medullary compartment, defined either as a cortical/med- To further assess the impact of Aire deficiency on thymic epi- ullary ratio or as the percent contribution of the medullary area to thelial heterogeneity, we examined the expression of K5 and K8 by total thymic area. The reduction of this MTEC compartment in the thymic epithelial cells. It had been previously reported that K8 AireϪ/Ϫ thymus was apparent in the six pairs of Aireϩ/ϩ and Ϫ/Ϫ expression is largely restricted to cortical epithelial cells, with a

FIGURE 6. A–C, Low magnification demonstration of K5 (red) and K8 (green) in Aireϩ/ϩ (A–F) and AireϪ/Ϫ (G–L) thymus. A, K8; B, K5; C, merge. Bar, 200 ␮m. D–F, High magnification demonstration of K8 (green) and K5 (red) in Aireϩ/ϩ thymus. D, K8; E, K5; F, merge. Bar, 60 ␮m. G–I, Low magnification demonstration of K8 (green) and K5 (red) in AireϪ/Ϫ thymus. Asterisks denote cysts. G, K8; H, K5; I, merge. Bar, 200 ␮m. J–L, High magnification demonstration of K8 (green) and K5 (red) in AireϪ/Ϫ thymus. J, K8; K, K5; L, merge. Bar, 60 ␮m. The Journal of Immunology 3013

FIGURE 7. Demonstration of K5 (red), K8 (green), and UEA (blue) in Aireϩ/ϩ (A–D) and AireϪ/Ϫ (E–H) thymus. Globular K5Ϫ,K8ϩ, UEAϩ MTEC are indicated by arrows. Glob- ular K5Ϫ,K8ϩ, UEAϪ MTEC are indicated by arrowheads. Globular K5ϩ,K8ϩ MTEC found in the AireϪ/Ϫ thymus are indicated by double-headed arrows. This last population variably displays low levels of UEA binding.

small subset of K8ϩ cells within the medulla. In contrast, K5 ex- Discussion pression was found to be largely restricted to the medulla with a There has been a long-standing assumption, arising from the rel- ϩ ϩ small subset of K5 K8 cells at the corticomedullary boundary ative radiation resistance of TE, that the epithelium in the adult and scattered in the cortex. Based on the expression patterns of thymus is a largely postmitotic population with little turnover. Ac- these cytokeratins under different experimental conditions, it has ϩ ϩ cordingly, the derepression model of promiscuous gene expression been proposed that the K5 K8 thymic epithelial cells represent has been reasonably proposed as a logical mechanism to account a progenitor population that give rise to more differentiated thymic for TRA expression by a static population of MTEC. However, epithelial cells that expressed either K5 or K8 (38–40). We con- recent studies have demonstrated continued MTEC turnover in the firmed the widespread expression of K5 throughout the medulla adult thymus (6, 8, 9), indicating that the medullary compartment and cells coexpressing K5 and K8 at the corticomedullary junction represents a steady state of MTEC at different stages of differen- or scattered in the cortex (Fig. 6, A–F). We also confirmed the universal expression of K8 by cortical cells and a small subset of tiation and implies the presence of a resident epithelial progenitor medullary TE that possessed a K5ϪK8ϩ phenotype with a globular population in the adult thymus. Importantly, it was demonstrated high profile (Fig. 6, A–F). In contrast to previous reports, we found that the MHC class II population of MTEC, which have been variable but widespread expression of K8 by medullary TE (Fig. 6, proposed to represent mature MTEC expressing the highest levels A and D). Many of the more stellate medullary TE cells that ex- of TRAs and Aire (4), also display a high level of proliferation and pressed high levels of K5 variably expressed K8 as well (compare have been proposed to represent a transit-amplifying subset of Fig. 6, A and D with B and E). MTEC (6). The representation of this MTEC subset in several The pattern of K5/K8 expression by MTEC in the AireϪ/Ϫ thy- mouse “knockout” models also suggest that the MHChigh mus highlighted a number of unique features of the medullary (CD80high) population represents an early stage of MTEC dif- epithelial compartment. In contrast to the stellate medullary epi- ferentiation (6). The mitotic activity of MTEC, together with ϩ ϩ thelial cells in the Aire / thymus that were joined by fairly broad the finding that MTEC express a broad spectrum of transcrip- cellular processes and that formed a fairly dense and uniform net- tion factors that have been implicated in the specification and Ϫ/Ϫ work (Fig. 6, C and F), Aire MTEC were often quite spindly differentiation of multiple epithelial lineages (8), form the basis and were joined to each other by thin and sometimes long pro- for an alternative model of TRA expression by MTEC, where ϩ cesses, leaving larger areas of medulla devoid of cytokeratin ep- this property would reflect the activity of immature MTEC un- ithelial cells (Fig. 6, I and L). Furthermore, the frequency of dergoing differentiation (7). From this perspective, the MHChigh MTEC that were globular, “rounded-up” without any visible cel- (CD80high) MTEC population would represent immature, rather lular projections, and expressing high levels of K8 were more than terminally differentiated, MTEC. prominent in the medullary compartment of AireϪ/Ϫ thymi (com- The demonstration here that MTEC express both the transcrip- pare Fig. 6, D and J). tion factors critical for pancreatic development and the endocrine Two populations of globular K5ϪK8ϩ MTEC defined by their gene products they regulate in extrathymic tissues raises the pos- reactivity with UEA are present within the normal thymic medulla Ϫ ϩ ϩ sibility that MTECs use conserved transcriptional hierarchies for (Ref. 38 and Fig. 7, A–D), where K5 K8 UEA MTEC and K5ϪK8ϩUEAϪ MTEC are indicated by arrows and arrowheads, the expression of TRAs and that the molecular mechanisms reg- respectively. Both of these populations of MTEC were also present ulating expression of pancreatic TRAs by MTEC are very similar in the AireϪ/Ϫ thymus (Fig. 7, E–H). The globular MTEC ex- to that in the pancreas. This possibility would be congruent with pressed K8 and low levels of K5 that were found in the AireϪ/Ϫ previous evidence for coordinated expression of lineage-related thymus are indicated by double-headed arrows in Fig. 7, E–H. TRAs, where single-cell RT-PCR analyses of isolated MTEC re- Occasionally, this population of MTEC also displayed reactivity vealed coordinated expression of developmentally related, Pdx1- with UEA. The demonstration here of K8ϩK5ϪUEAϩ and dependent endocrine Ags that did not possess chromosomal prox- K8ϩK5ϪUEAϪ subsets of globular MTEC represents a greater imity (8). Furthermore, the Aire dependence of transcription degree of heterogeneity than that originally described for this pop- factors crucial for endocrine pancreatic differentiation/function in ulation (38). MTECs suggests that the reduction of pancreatic TRA expression 3014 Aire DEFICIENCY ALTERS MEDULLARY TE by AireϪ/Ϫ MTECs may be secondary to disruption of this tran- ithelium (48), perhaps reflecting an alternate fate choice by third scriptional hierarchy. Additional single-cell transcriptional analy- pharyngeal pouch endoderm incapable of expressing functional ses, along with assessment of the spectrum of pancreatic TRAs Foxn1. expressed by MTEC from mice lacking these tissue-restricted reg- The alterations in MTEC composition/organization observed in ulatory transcription factors, should clarify this issue and are in the AireϪ/Ϫ thymus compared with the RelBϪ/Ϫ thymus indicate progress. that the general program of MTEC differentiation is more subtly The proposition that Aire plays a central role in MTEC differ- impacted by Aire deficiency. From this perspective, the Aire de- entiation is based in part on the contraction of the medullary com- pendence of nanog, Oct4, and Sox2 expression by MTEC is in- partment and reduced MTEC heterogeneity in the AireϪ/Ϫ thymus. triguing. If the activity of these transcription factors in MTEC This phenotype of the AireϪ/Ϫ thymus suggests that Aire influ- recapitulates their role in maintaining the pluripotentiality of stem ences the dynamic process of MTEC differentiation and thereby cells, their lack of expression by AireϪ/Ϫ MTEC may alter the affects the composition of MTEC. Although many aspects of thy- extent of MTEC heterogeneity or the range of TRAs they express, mic epithelial differentiation are poorly understood, it is clear that perhaps by reducing the spectrum of developmental programs that the NF-␬B-signaling pathway plays a central role. The profound developing MTEC can undertake or by changing the kinetics of reduction of the MTEC compartment in the RelBϪ/Ϫ thymus en- differentiation processes. One possibility is that Aire impacts TRA compasses a dramatic loss of Aire expression and a paucity of expression by influencing the tempo or efficiency of MTEC dif- UEAϩMTEC (41, 42), while deficiencies of signaling pathways ferentiation. If the expression of some TRAs occurs transiently that regulate RelB expression variably recapitulate the RelBϪ/Ϫ during proximal stages of MTEC differentiation, an accelerated thymic phenotype. Thymi from mice deficient in TNFR-associated tempo of MTEC differentiation due to Aire deficiency would re- factor 6 (43), NF-␬B-inducing kinase (44), or I␬B kinase ␣ (45) duce the permissive period for TRA expression and reduce the closely resemble the RelBϪ/Ϫ thymus, while the thymic phenotype frequency of MTECs expressing Aire-dependent TRAs at a given of LT␣R/LT␤R-deficient mice is less severe (46, 47) and time. Precocious or accelerated MTEC differentiation could also CD40Ϫ/Ϫ thymi appear to be normal (referenced in Ref. 44). The result in a reduced burst size of differentiating MTECs and thus Aire-deficient thymic phenotype is certainly milder that that ob- account for the contracted 3G10ϩ MTEC population observed in served in the RelBϪ/Ϫ, TNFR-associated factor 6Ϫ/Ϫ, NF-␬B-in- the Aire-deficient thymus. It is likely that this reduced MTEC com- ducing kinaseϪ/Ϫ,orI␬B kinase ␣Ϫ/Ϫ mice and bears some sim- partment would be less efficient in supporting negative selection in ilarity to the thymic phenotype of mice lacking LT␣RorLT␤R. general, and negative selection of TRA-reactive thymocytes in par- The looser organization of the contracted medullary compartment ticular, due to the relatively rare expression of any individual TRA of the AireϪ/Ϫ thymus may account for more efficient enzymatic (3, 8, 49). In this regard, the contraction of medullary epithelium dissociation AireϪ/Ϫ thymic tissue that we have observed (G. O. in the Aire-deficient thymus could provide a mechanistic explana- Gillard and A. G. Farr, unpublished observations). It is worth not- tion for the impaired negative selection of OVA-specific trans- ing that these alterations of the thymic stromal compartment de- genic T cells in Aire-deficient mice that is independent of OVA scribed here are not accompanied by significant perturbations in transcription levels (5). the representation of thymocytes subsets defined by flow cytomet- The alterations in thymic medullary epithelial composition and ric analyses (11). organization seen in the AireϪ/Ϫ thymus clearly indicate that the The contracted medullary compartment of the AireϪ/Ϫ thymus activity of Aire in MTECs extends beyond its previously ascribed lacked the confluent stellate MTEC seen in the normal thymus and role as a regulator of TRA expression in mature MTECs. These had increased representation of globular “rounded-up” MTEC ex- alterations in epithelial composition and organization are not ac- pressing high levels of K8. The variable and widespread coexpres- counted for by derepression models as presently constituted, but sion of K5 and K8 within the medullary compartment that we are consistent with a developmental model for Aire function. Al- observed in both the Aireϩ/ϩ and AireϪ/Ϫ thymus is difficult to though the transcriptional data presented here can be interpreted to reconcile with the view that K5ϩK8ϩ MTEC represent a precursor reflect a higher order of transcriptional derepression effected by population and that K8 expression is largely restricted to cortical Aire, and thus cannot exclude a role for Aire as a “randomizer” of TE (38–40). We feel this discrepancy reflects the sensitivity of gene expression, they provide circumstantial evidence that is con- detection because the same primary Abs were used in both studies. sistent with an alternative model of Aire function that can be ex- The Alexa fluorochromes used in this study provide stronger flu- perimentally tested. The novel properties of MTECs and the thy- orescent signals than the FITC or Texas Red dyes used in the mic consequences of Aire deficiency that are documented here previous reports. We have observed a similar pattern of K8 ex- need to be accounted for when developing and refining models for pression by medullary TE with a three-step immunoperoxidase TE differentiation and Aire function in the thymus. method (unconjugated primary Ab, digoxygenin-conjugated anti- Ј rat IgG Abs, and a peroxidase-conjugated anti-digoxygenin F(ab )2 Acknowledgments Ab; data not shown). Thus, while it is clear that the distribution We thank Drs. A. Rudensky, A. Liston, and M. Bevan for discussions and morphology of MTEC expressing these two keratins are per- and comments. Ϫ/Ϫ turbed in the Aire thymus and likely reflect an altered differ- Disclosures entiation program, we feel the developmental relationship between The authors have no financial conflict of interest. the TE populations defined by K5 and K8 expression is not clear. The increased prevalence of cystic epithelial structures in the References Ϫ/Ϫ Aire thymus is another indication that MTEC differentiation 1. Bjorses, P., J. Aaltonen, N. Horelli-Kuitunen, M. L. Yaspo, and L. Peltonen. has been affected. This view is based on our previous observation 1998. Gene defect behind APECED: a new clue to autoimmunity. Hum. Mol. that this small epithelial compartment in the normal thymus (10) Genet. 7: 1547–1553. 2. Anderson, M. S., E. S. Venanzi, L. Klein, Z. Chen, S. P. Berzins, S. J. Turley, resembles the predominant epithelial compartment in the H. von Boehmer, R. Bronson, A. Dierich, C. Benoist, and D. Mathis. 2002. Foxn1Ϫ/Ϫ thymus. The profound proximal defect in epithelial dif- Projection of an immunological self shadow within the thymus by the Aire pro- Ϫ/Ϫ tein. Science 298: 1395–1401. ferentiation in the Foxn1 thymus leads to accumulation of 3. Kyewski, B., and J. Derbinski. 2004. Self-representation in the thymus: an ex- “thymic” epithelium with phenotypic properties of respiratory ep- tended view. Nat. Rev. Immunol. 4: 688–698. The Journal of Immunology 3015

4. Derbinski, J., J. Gabler, B. Brors, S. Tierling, S. Jonnakuty, M. Hergenhahn, 28. Sussel, L., J. Kalamaras, D. J. Hartigan-O’Connor, J. J. Meneses, R. A. Pedersen, L. Peltonen, J. Walter, and B. Kyewski. 2005. Promiscuous gene expression in J. L. Rubenstein, and M. S. German. 1998. Mice lacking the homeodomain tran- thymic epithelial cells is regulated at multiple levels. J. Exp. Med. 202: 33–45. scription factor Nkx2.2 have diabetes due to arrested differentiation of pancreatic 5. Anderson, M. S., E. S. Venanzi, Z. Chen, S. P. Berzins, C. Benoist, and ␤ cells. Development 125: 2213–2221. D. Mathis. 2005. The cellular mechanism of Aire control of T cell tolerance. 29. Sander, M., L. Sussel, J. Conners, D. Scheel, J. Kalamaras, F. Dela Cruz, Immunity 23: 227–239. V. Schwitzgebel, A. Hayes-Jordan, and M. German. 2000. Homeobox gene 6. Gray, D. H., N. Seach, T. Ueno, M. Milton, A. Liston, A. M. Lew, C. C. Nkx6.1 lies downstream of Nkx2.2 in the major pathway of ␤-cell formation in Goodnow, and R. L. Boyd. 2006. Developmental kinetics, turnover, and stimu- the pancreas. Development 127: 5533–5540. latory capacity of thymic epithelial cells. Blood 108: 3777–3785. 30. Sander, M., A. Neubuser, J. Kalamaras, H. C. Ee, G. R. Martin, and M. S. 7. Gillard, G. O., and A. G. Farr. 2005. Contrasting models of promiscuous gene German. 1997. Genetic analysis reveals that PAX6 is required for normal tran- expression by thymic epithelium. J. Exp. Med. 202: 15–19. scription of pancreatic hormone genes and islet development. Genes Dev. 11: 8. Gillard, G. O., and A. G. Farr. 2006. Features of medullary thymic epithelium 1662–1673. implicate postnatal development in maintaining epithelial heterogeneity and TRA 31. Sosa-Pineda, B., K. Chowdhury, M. Torres, G. Oliver, and P. Gruss. 1997. The expression. J. Immunol. 176: 5815–5824. Pax4 gene is essential for differentiation of insulin-producing ␤ cells in the mam- 9. Yang, S. J., S. Ahn, C. S. Park, K. L. Holmes, J. Westrup, C. H. Chang, and malian pancreas. Nature 386: 399–402. M. G. Kim. 2006. The quantitative assessment of MHC II on thymic epithelium: 32. Wu, H., W. M. MacFarlane, M. Tadayyon, J. R. Arch, R. F. James, and K. implications in cortical thymocyte development. Int. Immunol. 18: 729–739. Docherty. 1999. Insulin stimulates pancreatic-duodenal homoeobox factor-1 ␤ 10. Dooley, J., M. Erickson, and A. G. Farr. 2005. An organized medullary epithelial (PDX1) DNA-binding activity and insulin promoter activity in pancreatic cells. structure in the normal thymus expresses molecules of respiratory epithelium and Biochem. J. 344(Pt. 3): 813–818. resembles the epithelial thymic rudiment of nude mice. J. Immunol. 175: 33. Andersen, F. G., J. Jensen, R. S. Heller, H. V. Petersen, L. I. Larsson, 4331–4337. O. D. Madsen, and P. Serup. 1999. Pax6 and Pdx1 form a functional complex on 11. Ramsey, C., O. Winqvist, L. Puhakka, M. Halonen, A. Moro, O. Kampe, the rat somatostatin gene upstream enhancer. FEBS Lett. 445: 315–320. P. Eskelin, M. Pelto-Huikko, and L. Peltonen. 2002. Aire deficient mice develop 34. Jepeal, L. I., Y. Fujitani, M. O. Boylan, C. N. Wilson, C. V. Wright, and multiple features of APECED phenotype and show altered immune response. M. M. Wolfe. 2005. Cell-specific expression of glucose-dependent-insulinotropic Hum. Mol. Genet. 11: 397–409. polypeptide is regulated by the transcription factor PDX-1. Endocrinology 146: 12. Anderson, M., S. K. Anderson, and A. G. Farr. 2000. Thymic vasculature: or- 383–391. ganizer of the medullary epithelial compartment? Int. Immunol. 12: 1105–1110. 35. Zuklys, S., G. Balciunaite, A. Agarwal, E. Fasler-Kan, E. Palmer, and G. A. 13. Dooley, J., M. Erickson, G. O. Gillard, and A. G. Farr. 2006. Cervical thymus in Hollander. 2000. Normal thymic architecture and negative selection are as- the mouse. J. Immunol. 176: 6484–6490. sociated with Aire expression, the gene defective in the autoimmune-polyen- J. Immunol. 14. Bennett, A. R., A. Farley, N. F. Blair, J. Gordon, L. Sharp, and C. C. Blackburn. docrinopathy-candidiasis-ectodermal dystrophy (APECED). 165: 2002. Identification and characterization of thymic epithelial progenitor cells. 1976–1983. Immunity 16: 803–814. 36. Farr, A. G., and S. K. Anderson. 1985. Epithelial heterogeneity in the murine thymus: fucose-specific lectins bind medullary epithelial cells. J. Immunol. 134: 15. Gill, J., M. Malin, G. A. Hollander, and R. Boyd. 2002. Generation of a complete ϩ 2971–2977. thymic microenvironment by MTS24 thymic epithelial cells. Nat. Immunol. 3: 37. Farr, A., A. Nelson, S. Hosier, and A. Kim. 1993. A novel cytokine-responsive 635–642. cell surface glycoprotein defines a subset of medullary thymic epithelium in situ. 16. Rossi, S. W., W. E. Jenkinson, G. Anderson, and E. J. Jenkinson. 2006. Clonal J. Immunol. 150: 1160–1171. analysis reveals a common progenitor for thymic cortical and medullary epithe- 38. Klug, D. B., C. Carter, E. Crouch, D. Roop, C. J. Conti, and E. R. Richie. 1998. lium. Nature 441: 988–991. Interdependence of cortical thymic epithelial cell differentiation and T-lineage 17. Bleul, C. C., T. Corbeaux, A. Reuter, P. Fisch, J. S. Monting, and T. Boehm. commitment. Proc. Natl. Acad. Sci. USA 95: 11822–11827. 2006. Formation of a functional thymus initiated by a postnatal epithelial pro- 39. Klug, D. B., E. Crouch, C. Carter, L. Coghlan, C. J. Conti, and E. R. Richie. 2000. genitor cell. Nature 441: 992–996. Transgenic expression of cyclin D1 in thymic epithelial precursors promotes 18. Boyer, L. A., T. I. Lee, M. F. Cole, S. E. Johnstone, S. S. Levine, J. P. Zucker, epithelial and T cell development. J. Immunol. 164: 1881–1888. M. G. Guenther, R. M. Kumar, H. L. Murray, R. G. Jenner, et al. 2005. Core 40. Klug, D. B., C. Carter, I. B. Gimenez-Conti, and E. R. Richie. 2002. Cutting transcriptional regulatory circuitry in human embryonic stem cells. Cell 122: edge: thymocyte-independent and thymocyte-dependent phases of epithelial pat- 947–956. terning in the fetal thymus. J. Immunol. 169: 2842–2845. 19. Loh, Y. H., Q. Wu, J. L. Chew, V. B. Vega, W. Zhang, X. Chen, G. Bourque, 41. Heino, M., P. Peterson, N. Sillanpaa, S. Guerin, L. Wu, G. Anderson, H. S. Scott, J. George, B. Leong, J. Liu, et al. 2006. The Oct4 and Nanog transcription net- S. E. Antonarakis, J. Kudoh, N. Shimizu, et al. 2000. RNA and protein expression work regulates pluripotency in mouse embryonic stem cells. Nat. Genet. 38: of the murine autoimmune regulator gene (Aire) in normal, RelB-deficient and in 431–440. NOD mouse. Eur. J. Immunol. 30: 1884–1893. 20. Manley, N. R. 2000. Thymus organogenesis and molecular mechanisms of thy- 42. Burkly, L., C. Hession, L. Ogata, C. Reilly, L. A. Marconi, D. Olson, R. Tizard, mic epithelial cell differentiation. Semin. Immunol. 12: 421–428. R. Cate, and D. Lo. 1995. Expression of relB is required for the development of 21. Blackburn, C. C., and N. R. Manley. 2004. Developing a new paradigm for thymic medulla and dendritic cells. Nature 373: 531–536. thymus organogenesis. Nat. Rev. Immunol. 4: 278–289. 43. Akiyama, T., S. Maeda, S. Yamane, K. Ogino, M. Kasai, F. Kajiura, M. 22. Offield, M. F., T. L. Jetton, P. A. Labosky, M. Ray, R. W. Stein, M. A. Magnuson, Matsumoto, and J. Inoue. 2005. Dependence of self-tolerance on TRAF6-directed B. L. Hogan, and C. V. Wright. 1996. PDX-1 is required for pancreatic outgrowth development of thymic stroma. Science 308: 248–251. and differentiation of the rostral duodenum. Development 122: 983–995. 44. Kajiura, F., S. Sun, T. Nomura, K. Izumi, T. Ueno, Y. Bando, N. Kuroda, H. Han, 23. Jonsson, J., U. Ahlgren, T. Edlund, and H. Edlund. 1995. IPF1, a homeodomain Y. Li, A. Matsushima, et al. 2004. NF-␬B-inducing kinase establishes self-tol- protein with a dual function in pancreas development. Int. J. Dev. Biol. 39: erance in a thymic stroma-dependent manner. J. Immunol. 172: 2067–2075. 789–798. 45. Kinoshita, D., F. Hirota, T. Kaisho, M. Kasai, K. Izumi, Y. Bando, Y. Mouri, 24. Kawaguchi, Y., B. Cooper, M. Gannon, M. Ray, R. J. MacDonald, and A. Matsushima, S. Niki, H. Han, et al. 2006. Essential role of I␬B kinase ␣ in C. V. Wright. 2002. The role of the transcriptional regulator Ptf1a in converting thymic organogenesis required for the establishment of self-tolerance. J. Immu- intestinal to pancreatic progenitors. Nat. Genet. 32: 128–134. nol. 176: 3995–4002. 25. Krapp, A., M. Knofler, S. Frutiger, G. J. Hughes, O. Hagenbuchle, and 46. Boehm, T., S. Scheu, K. Pfeffer, and C. C. Bleul. 2003. Thymic medullary epi- P. K. Wellauer. 1996. The p48 DNA-binding subunit of transcription factor PTF1 thelial cell differentiation, thymocyte emigration, and the control of autoimmu- is a new exocrine pancreas-specific basic helix-loop-helix protein. EMBO J. 15: nity require lympho-epithelial cross talk via LT␤R. J. Exp. Med. 198: 757–769. 4317–4329. 47. Chin, R. K., J. C. Lo, O. Kim, S. E. Blink, P. A. Christiansen, P. Peterson, 26. Gradwohl, G., A. Dierich, M. LeMeur, and F. Guillemot. 2000. neurogenin3 is Y. Wang, C. Ware, and Y. X. Fu. 2003. Lymphotoxin pathway directs thymic required for the development of the four endocrine cell lineages of the pancreas. Aire expression. Nat. Immunol. 4: 1121–1127. Proc. Natl. Acad. Sci. USA 97: 1607–1611. 48. Dooley, J., M. Erickson, H. Roelink, and A. G. Farr. 2005. Nude thymic rudiment 27. Naya, F. J., H. P. Huang, Y. Qiu, H. Mutoh, F. J. DeMayo, A. B. Leiter, and lacking functional foxn1 resembles respiratory epithelium. Dev. Dyn. 233: M. J. Tsai. 1997. Diabetes, defective pancreatic morphogenesis, and abnormal 1605–1612. enteroendocrine differentiation in ␤2/neuroD-deficient mice. Genes Dev. 11: 49. Hanahan, D. 1998. Peripheral-antigen-expressing cells in thymic medulla: factors 2323–2334. in self-tolerance and autoimmunity. Curr. Opin. Immunol. 10: 656–662.