Features of Medullary Thymic Implicate Postnatal Development in Maintaining Epithelial Heterogeneity and Tissue-Restricted Antigen Expression This information is current as of September 23, 2021. Geoffrey O. Gillard and Andrew G. Farr J Immunol 2006; 176:5815-5824; ; doi: 10.4049/jimmunol.176.10.5815 http://www.jimmunol.org/content/176/10/5815 Downloaded from

References This article cites 63 articles, 25 of which you can access for free at: http://www.jimmunol.org/content/176/10/5815.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 23, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Features of Medullary Thymic Epithelium Implicate Postnatal Development in Maintaining Epithelial Heterogeneity and Tissue-Restricted Antigen Expression1

Geoffrey O. Gillard* and Andrew G. Farr2*†

Although putative thymic epithelial progenitor cells have been identified, the developmental potential of these cells, the extent of medullary thymic epithelium (mTEC) heterogeneity, and the mechanisms that mediate the expression of a wide range of periph- eral tissue-restricted Ags (TRAs) by mTECs remain poorly defined. Here we have defined several basic properties of the mTEC population that refine our understanding of these cells and impose important constraints for any model of mTEC differentiation and function. We report here that mTECs from adult mice are mitotically active, implying continual turnover, differentiation, and replacement of mTEC populations in the adult thymus. The mTEC population in adult thymus expresses transcription factors Downloaded from implicated in the maintenance of multipotential progenitor cell populations, suggesting that epithelial progenitors in the adult thymus may not be restricted to a thymic fate. mTECs also express multiple transcription factors required for the specification of multiple epithelial lineages in peripheral tissues. Thus, expression of some TRAs by mTECs may represent coordinated expression that reflects alternate programs of epithelial differentiation among mTECs. Analysis of TRA expression in individual and small pools of sorted mTECs show that mTECs are highly heterogeneous; each individual mTEC expresses a limited spectrum

of TRAs, and the frequency of mTECs that express any individual TRA is quite low (>0.4–2%). Collectively, these findings suggest http://www.jimmunol.org/ that the differentiation of mTECs can involve some of the developmental programs used by other epithelial lineages and that expression of some TRAs by mTECs may reflect this activity. The Journal of Immunology, 2006, 176: 5815–5824.

hymic epithelium (TE)3 is broadly divided into two com- that were previously thought to be expressed only by specialized partments, based on their phenotypic characteristics and cells in peripheral tissues (3–7). T their different roles in T cell development (reviewed in Understanding the basis for this epithelial heterogeneity has Refs. 1 and 2). The early stages of thymocyte commitment and been hampered by a lack of knowledge regarding TE differentia- development occur in the cortex, and the defining characteristic of tion. Although a putative progenitor population has been identified cortical thymic epithelial cells, beyond their position, is their abil- in the fetal thymus (8–10), their persistence in the postnatal thy- by guest on September 23, 2021 ity to mediate the positive selection of developing thymocytes. mus has not been established. Furthermore, the developmental po- Thymocytes that have successfully undergone positive selection tential of these putative progenitor cells is not known. At one ex- migrate into the medulla, where they encounter an array of self treme, they could be specified to a thymic fate and have a very Ags presented on medullary thymic epithelial cells (mTECs) and restricted developmental potential. Alternatively, they could pos- thymic dendritic cells. Immature thymocytes with a strong affinity sess multilineage potential that becomes restricted in response to for self Ags are negatively selected upon exposure to these Ags, proximal environmental cues. Understanding how TE differentia- ensuring that the mature T cell repertoire is tolerant of multiple self tion is controlled will be of great assistance in determining the Ags and tissues. Unexpectedly, the range of self Ags expressed in basis for mTEC heterogeneity and TRA expression by mTECs. the medulla by mTECs extends well beyond ubiquitous Ags to The purpose of this study was to define the properties of TE encompass a vast spectrum of Ags (tissue-restricted Ags (TRAs)) that would help refine testable models of mTEC differentiation. One property that has not been closely examined concerns the *Department of Immunology and †Department of Biological Structure, University of mitotic activity of TE. Compared with hematogenous cells in Washington, Seattle, WA 98195 the thymus, TE is relatively radiation resistant; this property Received for publication January 20, 2006. Accepted for publication March 1, 2006. allows for the generation of radiation bone marrow chimeras. The costs of publication of this article were defrayed in part by the payment of page That TE is relatively radiation resistant has led to the notion that charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. TE is composed predominantly of differentiated postmitotic 1 This work was supported by the National Institutes of Health Grants AI 24137 and cells, although the deleterious effect of conditioning regimens AI 059575. G.O.G. was supported in part by training grants from the National Insti- used in bone marrow transplantation suggests that epithelial tutes of Health and the Cancer Research Institute. turnover and maintenance may also be impaired by these treat- 2 Address correspondence and reprint requests to Dr. Andrew G. Farr, Department of Biological Structure, Box 357420, University of Washington, Seattle, WA 98195- ments (11, 12). The extent of turnover among TECs, and mTEC 7420. E-mail address: [email protected] in particular, may indicate the extent to which TEC maturation 3 Abbreviations used in this paper: TE, thymic epithelium; aire, autoimmune regula- is a postmitotic event. tor; CK, cytokeratin; Csnb, casein ␤; Csng, casein ␥; Csnk, casein ␬; Dppa3, devel- To account for mTEC heterogeneity, and the expression of some opmental pluripotency associated 3; Epcam, epithelial cell adhesion molecule; FoxN1, forkhead N1; Gip, gastric inhibitory ; Ins2, II; mTEC, medul- TRAs, we have previously suggested that some epithelial progen- lary thymic epithelial cell; Pdx1, pancreatic and duodenal 1; Ppy, pancre- itors in the postnatal thymus are multipotential and retain the abil- atic polypeptide; Pcp4, Purkinje cell protein 4; Sst, somatostatin; Spt1, salivary protein 1; Spt2, salivary protein 2; TRA, tissue-restricted Ag; CMF, Ca2ϩ and Mg2ϩ ity to follow differentiation programs of other epithelial lineages free; TBE, Tris-buffered EDTA; mup, major urinary protein; RT, reverse transcriptase. (13, 14). According to this view, subsets of mTECs would express

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00 5816 mTEC HETEROGENEITY AND TRA EXPRESSION

transcription factors typically associated with multipotential pro- on an Amnis Imagestream (Amnis) using the same dissociation method and genitor cells and/or those known to be involved in the specification primary Abs. Draq5 (Biostatus) was used to stain DNA. of other epithelial lineages. The absence of these transcription fac- Primary Abs were: monoclonal anti-murine epithelial cell adhesion mol- ecule (Epcam; clone G8.8) either directly conjugated to Alexa 647 (Mo- tors in mTECs would require a reassessment of this model. lecular Probes) or conjugated to digoxigenin; anti-CD45 PE (clone 30-F11; Finally, as a potential indicator of mTEC heterogeneity, the ex- eBiosciences); anti-MHC class II (clone Y3P, conjugated to digoxigenin; a pression of TRAs by these cells must be assessed with greater generous gift of A.Y. Rudensky); a mixture of purified anti-BP-1 Abs precision. Morphological and immunohistochemical data have (clones 6C3 and CDR1, directly conjugated to Alexa 647), CD4 FITC (BD Pharmingen); and CD8 APC (eBiosciences). Secondary Ab was anti- shown that expression of individual TRAs is restricted to scattered digoxigenin-FITC (Boehringer Mannheim). single cells (4, 5, 15) and small multicellular foci (16) of mTECs. These data support the notion that mTECs are a heterogeneous Isolation of RNA mixture of cells. In contrast, analyses of TRA expression by RNA samples in Trizol were vortexed and passed five times each through mTECs, using large pools of cells, have assumed that TRA-ex- three (18-, 22-, and 25-gauge) needles. RNA was then isolated using stan- pressing mTECs represent a homogeneous population. Use of dard Trizol isolation protocol and resuspended in 100 ␮lof1ϫ DNase Turbo buffer (Ambion). We added 100 U of SUPERasin (Ambion) to each large populations imposes an important limitation on interpreting sample, followed by5UofDNase I (Ambion). Each sample was incubated these results, because they could reflect either a transcriptional at 37°C for 30 min. The DNase-treated RNA was then reisolated using profile common to a homogeneous population of cells or the RNeasy columns (Qiagen), concentrated, and linearly amplified following summed expression of a heterogeneous mTEC population. Al- a modified Eberwine procedure as described in Ref. 17. From 3 to 5 ␮gof though the former interpretation is currently favored, our recent each amplified cRNA was used to generate dscDNA for PCR analysis with random primers and Arrayscript RT (Ambion) to generate the first strand. demonstration that TRAs associated with respiratory epithelium PCR analysis was conducted using primers for multiple molecules. All Downloaded from are preferentially expressed by a small subset of mTECs favors the primers were designed using Primer Express (Applied Biosystems) to have ␮ latter interpretation (16). A more detailed analysis of the pattern of a Tm of 60°C and to work well under the following conditions: 25- or 50- l TRA expression by mTECs would facilitate efforts to determine reactions with a final concentration of 10 mM Tris, 2.5 mM MgCl2, 0.2 mM dNTPs, 0.4 ␮M primers, and 0.625 U Taq DNA polymerase in buffer the mechanisms controlling this process. B (Promega). All primers, where possible, were designed to span introns. In this study, we report that mTECs in adult thymus are a mi- No polymerase controls were used to ensure cDNA specificity for ampli-

totically active population, raising the possibility that morpholog- cons that do not span introns. All primer sequences are available upon http://www.jimmunol.org/ ical heterogeneity and TRA expression by mTECs could be asso- request. Cycling conditions were 94°C for 5 min; cycles of 94°C for 30 s, ciated with ongoing epithelial differentiation and turnover. The 60°C for 30 s, 72°C for 45 s, 72°C for 10 min; 4°C hold. For cDNA from large mTEC populations, HPRT, Epcam, aire, and CD45 reactions were mTEC population expresses transcription factors that are associ- amplified through 28 cycles; all other PCR were amplified through 38–42 ated with multipotency, consistent with the possibility that epithe- cycles. PCR products were examined on 1.8% Tris-buffered EDTA (TBE) lial progenitor cells in thymus have a degree of developmental gels stained with Sybrsafe DNA gel stain (Invitrogen and Molecular plasticity. The observation that mTECs also express transcription Probes). factors involved in the differentiation of other epithelial cell lin- BrdU labeling studies eages, lineages that are represented by the spectrum of TRAs ex-

Age-matched cohorts of BALB/c mice (NCI) were maintained either on by guest on September 23, 2021 pressed within the mTEC population, raises the possibility that the BrdU drinking water (0.8 ␮g/ml; prepared and changed daily) or normal expression of some of these structural molecules by mTECs is drinking water in the University of Washington SPF facility for up to 5 wk. regulated in the same manner as in the parent tissue. Finally, RT- After 5 wk, experimental groups were returned to normal drinking water PCR analysis of TRA expression by small pools and individual for an additional 5 wk. At indicated intervals, 10 mice from each group were sacrificed, and thymi were dissociated to form a single-cell suspen- mTECs clearly demonstrated a high degree of heterogeneity within sion and enriched for epithelial cells by anti-CD45 bead depletion as de- the mTEC population. When analyzed on this scale, the pattern of scribed in Ref. 18. After bead depletion, the remaining cells were washed, TRA expression by mTECs appears to be more compatible with suspended in FBS, and spun onto aminoalkylsilane-coated slides with a developmental regulation rather than the chromosomal proximity cytocentrifuge. Slides were air dried overnight, fixed in a 75% acetone- and accessibility of TRA-encoding . 25% ethanol for 3 min at room temperature, rehydrated in PBS (pH 7.4), and then stained with primary Abs in PBS containing O.1% BSA ϩ 10% FBSfor1hatroom temperature. Slides were washed then stained with Materials and Methods secondary Abs (if biotin or digoxigenin conjugated; all fluorochrome-con- Mice jugated reagents and anti-BrdU mAbs were applied after HCl treatment), washed, and fixed at room temperature for 20 min in PBS containing 4% Adult B6 mice (5–10 wk) were obtained either directly from the National paraformaldehyde. Slides were washed three times in PBS before immer- Cancer Institute, from Charles River Laboratories, or from our colony sion in 4 N HCl for 30 min at room temperature. Slides were rinsed in PBS maintained at the University of Washington SPF facility. Age-matched 4- and transferred to 0.2 M sodium borate (pH 8.5) for 3 min to neutralize any to 6-wk-old BALB/c mice for BrdU labeling experiments were obtained residual acid. Slides were then incubated in PBS ϩ 0.1% BSA for 5 min, from the National Cancer Institute. All mice were handled and used in before addition of Alexa 488-conjugated or biotin-conjugated anti-BrdU accordance with protocols approved by the University of Washington’s mAb (Invitrogen and Molecular Probes) and all fluorochrome-conjugated Institutional Animal Care and Use Committee. secondary/tertiary Abs and/or staining reagents. Primary Abs were: G8.8 (purified or digoxigenin conjugate); 6C3 (anti- Sorting and flow cytometry BP-1; purified or digoxigenin conjugate); rabbit anti-pancytokeratin (CK; Enzymatic dissociation of thymi was performed as described (16). Briefly, Sigma-Aldrich); and biotin- or Alexa 488-conjugated anti-BrdU mAb (In- thymi were removed and diced in HBSS (Ca2ϩ- and Mg2ϩ-free; CMF) ϩ vitrogen Life Technologies and Molecular Probes). Secondary/tertiary Abs 2% FBS ϩ HEPES using forceps and scalpel, washed, and sequentially and reagents were: anti-rat Ig Alexa546 conjugate; anti-rat Ig Alexa 488 digested using collagenase D (Roche) in CMF HBSS followed by a mix- conjugate; streptavidin-Alexa 546 conjugate (Molecular Probes); anti- ture of collagenase and neutral dispase (Roche) in CMF HBSS to obtain a rabbit biotin conjugate (Sigma-Aldrich), anti-digoxigenin FITC conjugate single-cell suspension. Single-cell suspensions were treated with rat anti- (Roche). Fc␥R II/III mAb (clone 24G2) before staining. After staining, cells were Generation of pooled cDNAs from sorted mTECs washed twice in HBSS CMF ϩ 2% FBS and resuspended in the same containing 7-aminoactinomycin D (Molecular Probes) at a final concen- Generation of cDNA from pools of mTECs sorted from adult (6- to 12- tration of 2 ␮g/ml. Cells were sorted on either a BD Vantage or Aria cell wk-old B6 mice) thymus for TRA expression profile analysis was con- sorter directly into Trizol (Invitrogen; for analyzing large populations) or ducted according to the procedures detailed in Ref. 19. Briefly, cells in into lysis/reverse transcriptase (RT) buffer in 96-well plates for small pools ϳ0.5 ␮l were mixed in 4 ␮l of lysis/RT buffer (0.6 mM dNTP mix, 1ϫ first ␮ and single-cell analyses. Analysis of mTEC cell cycle status was performed strand buffer, 0.13 M first primer (ACTCACTATAGGGAGGCGGT24), The Journal of Immunology 5817

0.5 U/reaction SUPERasin RNase inhibitor (Ambion), 0.5 U/reaction Ri- showed significant accumulation of BrdU-labeled mTECs through bolock RNase inhibitor (Fermentas), 0.5% Nonidet P-40) heated to 70°C the 5-wk labeling period, followed by a corresponding decrease in ␮ for 3 min 45 s, and cooled on ice for 5 min before addition of 0.5 lofa the number of cells retaining the BrdU label after cessation of 1:1 mixture of lysis/RT buffer-SuperscriptII RT (200 U/␮l; Invitrogen Life Technologies). Plates were vortexed briefly, spun down, and incubated at BrdU treatment (Fig. 1B). The kinetics of the loss of the BrdU 37°C. After 30 min, samples were heated to 70°C for 10 min to inactivate label following cessation of BrdU administration mirrored the ki- the RT enzyme and chilled on ice. Next, 5 ␮lof2ϫ TdT mix (6 mM dATP, netics of the labeling phase. We conclude that the proliferation 3 mM CoCl ,2ϫ tailing buffer ϩ 25 U/reaction TdT (Roche) was mixed 2 documented here reflects epithelial turnover, because the adult thy- into each reaction and incubated at 37°C. After 15 min, samples were heated to 65°C for 15 min to inactivate TdT. Three parallel PCRs were run mus has already reached maximal size and is not undergoing pre- consisting of 3.3-␮l aliquots each of the TdT reactions added to 21.7 ␮lof pubertal expansion (21). primary PCR mix (0.9 mM dNTP mix, 10 ␮M first primer, 1ϫ HotStarTaq To independently confirm the existence of mitotically active ϫ Ϫ buffer, 0.5 Q solution containing 2.5 U HotStarTaq (Qiagen), and 0.125 CD45 Epcamhigh mTECs, we used the Amnis Imagestream in- U Pfu Turbo (Stratagene) per reaction), mixed, and run on a thermocycler using the following program: 94°C for 15 min; 1 cycle of 94°C for 45 s; strument to measure the DNA content of individual mTECs 50°C 2 min, 72°C for 3 min and 15 s; 36 cycles of 94°C for 45 s; 68°C for stained for CD45, Epcam, and the DNA stain Draq5. The Amnis 45 s; 72°C for 3 min; 72°C for 10 min; 4°C hold. Primary PCRs were Imagestream is a specialized instrument that combines the image- combined and purified using Qiaquick PCR purification columns (Qiagen) capturing capacity of a high throughput fluorescence microscope and eluted in a total volume of 100 ␮l. Secondary PCR were prepared using 40 ␮l of purified primary PCR products mixed with 60 ␮l of secondary to acquire real time images of each individual cell in multiple PCR mix (0.2 mM dNTP mix, 5 ␮M second primer (ACTCACTATAG channels, with the capacity to generate population analyses (sim- GGAGGCGGTTT), 1ϫ HotStarTaq buffer, 0.5ϫ Q solution, 2 U Taq ilar to a conventional flow cytometer) using the raw imaging data (Promega) and 0.25 U Pfu Turbo per reaction) and cycled with the fol- generated from large input populations. This instrument allowed Downloaded from lowing program: 94°C for 5 min; 32 cycles of 94°C for 30 s; 55°C for 45 s; 72°C for 3 min; 72°C for 10 min; 4°C hold. Secondary reactions were the simultaneous quantitation of signal intensity and visualization column purified and eluted in 100-␮l volumes. PCR analysis for multiple of the staining pattern for each individual cell that fell within a gate genes were performed using 2 ␮l of purified secondary PCR products in or population. We assessed the cell cycle status of individual standard 25-␮l PCR mixtures with 60°C annealing temperatures (38 cycles CD45ϪEpcamϩ mTECs isolated from unmanipulated adult B6 for all genes tested). Gene-specific PCR products were visualized on 1.8% thymus (Fig. 1C) using the DNA stain Draq5. A population profile TBE gels stained with Sybrsafe DNA stain. http://www.jimmunol.org/ Nested gene-specific PCR on pooled and individual sorted MHC IIhigh for mTECs based on the staining intensity of Draq5 demonstrated mTECs was performed using the method described in Refs. 20. Briefly, a main peak of cells that were quiescent and a heterogeneous pop- ϩ Ϫ MHC IIhigh Epcam CD45 cells from adult (6- to 12-wk-old B6 mice) ulation in the gated population that were Ͼ2N in DNA content or ␮ ϫ thymus were sorted directly into 4.5 l of lysis/RT buffer (1 RT buffer, were apoptotic (Fig. 1Ca). Representative images of individual dNTP, 12 ␮M each gene-specific RT primer ϩ RNase inhibitors ϩ 0.5% Nonidet P-40) on ice. After collection, samples were heated to 70°C for 3 cells representative of these populations include quiescent (2N min for 45 s and then cooled to 4°C for 3 min before addition of RT DNA content) mTECs (Fig. 1Cb), apoptotic mTECs (Fig. 1Cc), enzyme (0.5 ␮l of SSII-buffer, 1:1). Samples were mixed and incubated at and mTECs that have either synthesized DNA (Ͼ2N DNA con- 42°C for 90 min in a thermocycler. After RT reactions were complete, ␮ tent; Fig. 1Cd). Results of several experiments indicated that plates were transferred to ice for 5 min before addition of 45 l of the outer ϳ ϫ ␮ 1–2% of mTECs isolated from unmanipulated adult (6–10 wk) by guest on September 23, 2021 nest PCR mixture (1 PCR buffer, 2.5 mM MgCl2, 0.24 M gene-specific outer nesting primers, ϩ2.5 U Taq per reaction). Thermocycler conditions B6 thymus have synthesized DNA or recently divided. These re- for the outer nesting PCR: 94°C for 5 min; 36 cycles of 94°C for 30 s; 60°C sults are consistent with the BrdU labeling data and clearly indi- for 30 s; 72°C 1 min; 72°C extension for 10 min; 4°C hold. Inner nesting cate that there is ongoing mitotic activity within the mTEC pop- PCR were performed using 1 ␮l of the outer nest reaction added to 49 ␮l ϫ ulation of adult mice. It also appears that the rate of epithelial of individual inner nesting PCR mixture (1 PCR buffer, 2.5 mM MgCl2, 0.4 ␮M gene-specific inner nesting primers, 1.25 U Taq). Thermocycler turnover is rather low, which would be consistent with the relative conditions for the inner nesting PCR: 94°C for 5 min; 36 cycles of 94°C for radiation resistance of this population that allows the generation of 30 s; 60°C for 30 s; 72°C for 45 s; 72°C extension for 10 min; 4°C hold. radiation bone marrow chimeras. Inner nesting PCR products were visualized on 1.8% TBE gels stained with Sybrsafe DNA stain (Invitrogen Life Technologies and Molecular Probes). Results Sorted mTECs express molecules known to regulate thymic The mTEC population in adult mice is mitotically active and organogenesis and TEC development undergoing significant turnover Although ongoing epithelial turnover implies ongoing mTEC dif- One of the elements necessary for a developmental model of TEC ferentiation, mitosis could be attributed to self-renewal of adult heterogeneity and TRA expression is ongoing differentiation and mTECs. We wanted to determine whether the known molecular turnover of mTECs in adult thymus. We used two independent regulators of thymic organogenesis and mTEC differentiation are means to assess mitotic activity in mTECs, the first being a set of expressed within this mitotically active population. We generated 5 high Ϫ Ϫ BrdU labeling experiments. Cohorts of age-matched adult BALB/c cDNA from 1 to 2 ϫ 10 Epcam CD45 BP-1 mTECs isolated mice (4–6 wk old) received BrdU in their drinking water for a from adult B6 thymus. We first evaluated these cDNAs for the 5-wk pulse period and then returned to normal drinking water for expression of aire, Epcam, CD45, and a number of TRAs previ- a 5-wk chase period. At various intervals during this regimen, ously shown to be expressed by mTECs, including casein ␥ enzymatically dissociated thymi from test and control groups were (Csng), Purkinje cell protein 4 (Pcp4), and insulin II (Ins2) (3, 22). enriched for epithelial cells by anti-CD45 bead depletion (18), and These PCR results showed that all of these molecules (except then assessed for CK and BrdU content by immunofluorescence CD45) were expressed within the sorted mTEC population (Fig. microscopy (Fig. 1A). Triple staining experiments showed that al- 2A) and verified that we were examining a cell population com- most all epithelial cells recovered were mTECs (as defined by parable with those used in previous studies (3, 22). staining results with Abs to Epcam (a medullary marker), BP-1 (a We then assessed whether or not the known developmental reg- cortical marker), and CK; data not shown). Due to low recovery of ulators of thymic epithelial differentiation are expressed within the cortical TECs using this method, we were unable to draw any mTEC population. Although some, such as forkhead N1 (FoxN1), conclusions regarding the mitotic activity of this population. The Pax1, and lymphotoxin ␤ , have been shown to be ex- results of three independent experiments were consistent and pressed by postnatal thymic epithelium, the roles of Eya1, Hoxa3, 5818 mTEC HETEROGENEITY AND TRA EXPRESSION Downloaded from http://www.jimmunol.org/ by guest on September 23, 2021

FIGURE 1. mTECs are mitotically active in adult thymus. A and B, Results from studies where dividing cells were labeled in vivo by incorporation of BrdU. A, Representative cytospin of CD45-depleted thymic cells stained for BrdU (green) and CK (red). B, Quantitative analysis of the percentage of total CKϩ (epithelial) cells that have synthesized DNA during the labeling phase of the experiment (CKϩBrdUϩ). Results are representative of three separate experiments. No BrdU staining was observed in cells from age-matched control animals maintained on normal water throughout the experiment. A minimum of 700 epithelial (keratinϩ) cells was counted at each time point. C, Results of DNA content analyses of single sorted mTEC isolated from adult mice that was performed with the Amnis Imagestream instrument. a, Draq5 staining intensity histogram of EpcamϩCD45Ϫ cells. The subset of cells with high DNA staining intensity is indicated by gate R3. b, Representative image of an individual mTEC classified as quiescent (2N DNA content). c, mTEC considered to be apoptotic, based on the dense and blebbed nucleus. d, Individual mTEC with DNA content Ͼ2N and lacking nuclear features of apoptosis. A total of 34 of 1906 mTECs (1.8%) were scored as live, Ͼ2N DNA content cells. The scoring of EpcamϩCD45ϪDraq5high cells as apoptotic or mitotic was performed by Amnis personnel who lacked direct knowledge of the project, and results are representative of three experiments. and Pax9 in thymic epithelial development have been largely de- proposed the presence of multipotential progenitor epithelial cells fined by the impact of their targeted inactivation on thymic orga- in thymus (13, 14). Because the nature of and product-precursor nogenesis (reviewed in Ref. 23). Here we have determined that all relationships for mTEC subsets remain ill defined, we could not of these known regulators of thymic epithelial development were test this hypothesis directly. However, we could determine expressed in bulk, sorted populations of adult mTECs (Fig. 2B). whether or not some of the transcription factors previously dem- This expression pattern raised the possibility that Eya1, Hoxa3, onstrated to be characteristic of multipotential cell populations and Pax9 may continue to play a role in the development and were expressed within the mTEC population. The molecules we maturation of mTECs in the adult thymus. These results also dem- examined are largely restricted to multipotential cell populations onstrate that the molecular regulators of thymic organogenesis and (such as embryonic stem cells and blastocysts) and are either con- differentiation continue to be expressed within the adult mTEC sidered to be characteristic markers of these populations or have population and suggest that mTEC differentiation from a resident been directly implicated in the maintenance of pluripotentiality in progenitor population persists in adult thymus. the cells that express them. These include Stella (developmental pluripotency associated 3; Dppa3) (24), Ehox (embryonic ho- Molecules characteristic of multipotential cells are expressed by meobox) (25), Foxd3 (also known as Genesis) (26), and undiffer- mTECs entiated 1 (27). We determined that each of Based on the previous demonstration that epithelial foci in adult these molecules is expressed within the mTEC population (Fig. 3). thymus ultrastructurally and phenotypically resemble other We also evaluated the expression of , Nanog, and Oct4 endodermally derived epithelium (reviewed in Ref. 13), we have (Pouf51) by mTECs because they have been reported to compose The Journal of Immunology 5819

FIGURE 2. The molecules regulating early thymic development are also expressed by mTECs from adult thymus. PCR on a serial dilution (undiluted, 1/5, 1/25) of cDNA generated from 1 to 2 ϫ 105 sorted mTECs FIGURE 3. mTECs express a number of transcription factors charac- (EpcamϩCD45Ϫ) demonstrated that Epcam, aire, and a range of aire-de- teristic of multipotential progenitor cell populations. Bulk populations of pendent TRAs characteristic of multiple peripheral tissues (Csng, mam- sorted mTECs express a broad range of transcription factors characteristic mary gland; liver fatty acid binding protein (LFabp), liver; Gip, intestine; of multipotential progenitor cell populations. PCR on a serial dilution (un- Pcp4, neuronal; and Ins2, ) are expressed by this population (A). diluted, 1/5, 1/25) of cDNA from sorted EpcamϩCD45 mTECS demon-

Additional PCR analysis (B) shows that mTECs in adult thymus express strates that this population expresses the transcription factors Dppa3, em- Downloaded from multiple molecular regulators of thymic epithelial development: eyes ab- bryonic homeobox (Ehox), Genesis (Foxd3), and undifferentiated sent 1 (Eya1); FoxN1 (Whn); homebox a3 (Hoxa3); lymphotoxin ␤ recep- transcription factor-1 (Utf1). The combination of Nanog, octamer-binding tor (Lt␤R); paired box 1(Pax1); paired box 9 (Pax9). 4 (Oct4), and sex-determining region Y-box 2 (Sox2) have been charac- terized as the core transcriptional regulatory circuit in embryonic stem the core transcriptional regulatory circuit active in multipotential cells. Expression of these molecules was either restricted to, or highly cells (28). As shown in Fig. 3, each of these molecules was also enriched in mTECs as compared with sorted single-positive and double-

positive thymocytes. This preferential expression in mTECs also extends to http://www.jimmunol.org/ expressed within the mTEC population. In the case of Dppa3, ϩ FoxD3, and undifferentiated transcription factor 1, these results cortical TECs and CD45 DN/TN populations (data not shown). confirm array data that had identified these molecules as prefer- entially expressed by mTECs (29). Although it remains to be de- mTEC lineage (22, 29); however, other studies have suggested that termined whether or not the expression of this group of transcrip- mTECs are highly heterogeneous and that the expression of indi- tion factors can be attributed to a restricted subset of mTECs with vidual TRAs is highly restricted to relatively rare mTECs (4, 5, unique developmental potential, this work identifies new markers 15). To distinguish between these two possibilities, we assessed that may be useful for identifying mTEC progenitors and estab- the TRA expression profiles for small pools of sorted mTECs. lishing product-precursor relationships among mTEC subsets. Using the method of Chiang and Melton (19), we generated cDNA from small pools (20 cells/pool) of sorted Epcamhigh by guest on September 23, 2021 Given the reported function of these transcription factors in other Ϫ Ϫ cell populations referenced above, we hypothesize that their ex- CD45 BP-1 mTECs from the thymi of adult B6 mice. We an- alyzed the expression patterns of aire and 15 individual TRAs in pression will likely be restricted to a discrete subset of immature ϩ mTECs. pools of cells previously verified by PCR to be Epcam and MHC IIϩ. As shown in Fig. 5, each individual TRA, with the notable Multiple regulators of development in peripheral epithelial exception of casein ␤ (Csnb), was rarely expressed in the pools. tissues are expressed by mTECs The ubiquitous (and atypical) expression of Csnb suggests that this Having observed expression of multiple regulators of thymic de- molecule is regulated differently than the other TRAs we assessed velopment and transcription factors characteristic of multipotential and may indicate that Csnb is expressed as part of a global mTEC cells within the mitotically active mTEC population, we next expression profile. When all TRAs were considered, including wanted to determine whether mTECs also expressed transcription Csnb, the average number of TRAs expressed per pool was closest factors that regulate the differentiation of peripheral epithelial tis- to 2 of 15, with no pool expressing Ͼ4 of 15 TRAs. If Csnb is sues where TRAs are characteristically expressed. We determined omitted from the analysis, the average dropped to 1 of 14 TRAs that in addition to FoxA1 and FoxA2, which are centrally involved expressed per pool; no single pool expressed Ͼ3 of 14 TRAs, and in the development of a number of endodermal derivatives (30– 5 pools expressed 0 of 14 TRAs. 36), mTECs expressed other transcription factors that are critical The expression frequency for any individual TRA expressed in for the development of endodermal and neural tissues (Cdx1 (37, the pools (excepting Csnb) ranged from 0 of 13 to 3 of 13. If it is 38), Crx (39), Pdx1 (40–43), Hnf-6 (44–46), and Foxg1 (47, 48); assumed (given the relatively low frequency of expression for each Fig. 4A). Multiple TRAs characteristic of the tissues that these individual TRA) that positive expression of a TRA within a pool transcription factors regulate, including blue cone opsin (retina), reflects the presence of an individual mTEC that expresses that fatty-acid binding protein 1 (liver), gastric inhibitory polypeptide TRA, we could estimate the percentage of mTECs that make a (Gip; intestinal epithelium), Ins2 (pancreas), and major urinary given TRA. Based on the data presented here, the percentage of protein 1 (liver), were also expressed by mTECs (Fig. 2, Fig. 4B, mTECs that expressed a given TRA ranged from Ͻ0.4% Ϫ1% (0 and data not shown). These results show that the spectrum of of 260 cells to 3 of 260 cells). This estimate is consistent with TRAs expressed within the mTEC population extends beyond previously published estimates of the frequency of mTECs that structural or terminal TRAs to include upstream developmental express a given TRA (4, 15). mediators characteristic of peripheral epithelial tissues. Expression of individual TRAs by functionally mature mTECs is Expression of individual TRAs is restricted in mTECS also restricted to relatively rare cells Several studies have suggested that TRAs are broadly and promis- It has been proposed that mTECs progressively derepress TRA- cuously expressed by mature mTECs as a unique property of the containing loci as they mature, resulting in the highest levels of 5820 mTEC HETEROGENEITY AND TRA EXPRESSION

FIGURE 4. mTECs express a wide spectrum of transcription factors involved in the development of peripheral tissues. Bulk populations of sorted EpcamϩCD45Ϫ mTECs express a broad range of transcription fac- tors involved in the development of specialized cell types in peripheral tissues. A, Results of PCR on a serial dilution (undiluted, 1/5, 1/25) of cDNA from sorted mTECs established that a number of transcription fac- tors known to be involved in the development of peripheral epithelial cell FIGURE 5. Expression of TRAs is limited to rare cells within small Downloaded from lineages, including Cdx1 (embryonic positioning; intestinal epithelium), pools of sorted mTECs. Small pools of sorted mTECs (EpcamϩCD45Ϫ;20 Crx (retinal photoreceptor cells), FoxA1 and FoxA2 (Hnf-3␣ and Hnf-3␤, cells/pool) were assessed for the expression of aire and 15 TRAs. Pools multiple tissues), Hnf-6 (multiple tissues, including pancreas), and Pdx1 were validated using PCR for Epcam, MHC II, and HPRT expression. The (pancreatic organogenesis; transcription of Gip and Ins2). The expression expression profile of multiple TRAs in 13 pools is depicted. Dark squares, of these developmental transcription factors corresponds to the expression positive expression; light squares, lack of expression. Ags are grouped of downstream TRAs that are also found in mTECs (B), including blue loosely by tissue. The TRAs assessed include those known to be aire de- cone opsin (Bco; retina), glucagon (Gcg; pancreas), major urinary protein pendent (Csng, mammary; elastase 3b, pancreas; Gip, intestine; Ins2, pan- http://www.jimmunol.org/ 1 (Mup1; liver), and Spt1 (salivary gland). For additional examples of creas; Mup1, liver; Pcp4, neuronal; Spt1 and Spt2, salivary gland) and TRAs that are representative of the tissues specified by the transcription additional Ags characteristic of other peripheral epithelial cell lineages factors shown in A, see Fig. 2. (cystic fibrosis transmembrane conductance regulator homologue (CFTR), multiple; palate, lung, and nasal epithelium carcinoma associated (PLUNC), multiple; Csnb, Csnk, mammary gland; glucagon (Gcg), pan- aire and TRA expression in the most mature mTECs (29), with creatic polypeptide (PPP), somatostatin (Sst), pancreas; and thyroglobulin. The total number of TRAs expressed in a given pool is shown at the bottom high levels of MHC II protein expression used as the criterion for of the graph. Note that, despite the widespread expression of aire, a number mature mTEC. We generated amplified cDNA from small pools of aireϩ pools express few if any aire-dependent TRAs. (25 cells/pool) of sorted CD45ϪEpcamϩMHC IIhigh mTECs (Fig. by guest on September 23, 2021 6A) from adult B6 mice and evaluated these for the expression of 17 TRAs. Although the frequency of TRA expression was higher in the MHC IIhigh pools than in the pools of unfractionated cuously opened in mature mTECs, resulting in the coordinated CD45ϪEpcamϩ mTECs analyzed in Fig. 5, the expression profiles expression of these TRAs; the mechanisms proposed to account for all of these TRAs (again, with the exception of Csnb) indicate for this include run through transcription of the open locus (29). that the frequency of expression for each individual TRA is also Our analysis of TRA expression within small pools of sorted quite low in MHC IIhigh mTECs (Fig. 6B). A sample of the raw mTEC clearly showed that Csng and Csnk appear to be expressed data represented in Fig. 6B is shown in Fig. 6C. All nine pools independently from both Csnb and each other; indeed, we did not shown in Fig. 6 expressed aire and Csnb, whereas expression of find Csng and Csnk coexpressed within any pool (Figs. 5 and 6B). thyroglobulin was observed only in pool 8. All pools except pool In contrast, Csnb was highly expressed in mTECs and may be 3 were free of hematogenous cell contamination as determined by expressed by every mTEC or a large, as yet undefined subset of CD45 expression. When all TRAs were considered, including mTECs. Further single-cell analysis will be required to determine Csnb, the average number of TRAs expressed per pool was ϳ3of whether that is the case. 17; no pool expressed Ͼ6 of 17 TRAs. If Csnb is omitted from the Although we saw no evidence for coordinated expression of analysis, the average drops to 2 of 16 TRAs expressed per pool; no casein genes, we did observe that the pools that expressed multiple single pool expressed Ͼ5 of 16 TRAs, and 4 pools expressed 0 of TRAs coexpressed Gip, pancreatic polypeptide (Ppy), and soma- 16 TRAs. The expression frequency for any individual TRA in the tostatin (Sst). Although chromatin de-repression does not readily pools (excepting Csnb) ranged from 0 of 9 to 5 of 9. As above, we account for the coordinated expression of these genes (Gip and calculated that the percentage of MHC IIhigh mTECs expressing a Ppy are located over 6 Mb apart on 11, Sst is located given TRA ranges from Ͻ0.4 to 2% (0 of 225 cells to 5 of 225 on chromosome 16), developmental mechanisms may explain the cells). This value was also similar to previous estimates for the possible coordinated expression of these TRAs. The pancreatic percentage of mTECs that express an individual TRA. These re- transcription factor Pdx1 (pancreatic and duodenal homeobox1, or sults show that even in a population of mature mTECs, the fre- somatostatin transactivating factor-1) is known to directly regulate quency of mTECs that expressed any individual TRA was low. the transcription of both Gip (43) and Sst (40); the development of The expression patterns for the casein genes ␤, ␥, and ␬ (Csnb, the cells that express Ppy in pancreas is also dependent on Pdx1 Csng, and Csnk) that we observed within the pools are particularly (41). It will be important to determine the extent to which other noteworthy, given that these genes reside within the same locus on developmentally related genes are coordinately expressed in (49). Expression of all three by mTECs has been mTECs and the extent to which expression of developmentally put forth as a strong example of epigenetic derepression of TRAs related TRAs is coincident with the transcription factors that reg- (29), whereby the genes within the casein locus would be promis- ulate them in peripheral tissues. The Journal of Immunology 5821 Downloaded from

FIGURE 6. Expression of TRAs is limited to rare cells within small pools of sorted mature (MHC IIhigh) sorted mTECs. A, Sort profile used to obtain ϩ Ϫ high Epcam CD45 MHC II mTEC. B, Tabular representation of the expression of 17 TRA by 9 pools, each containing 25 of these mTECs using the same http://www.jimmunol.org/ format as in Fig. 5. The total number of TRAs expressed in a given pool is indicated at the bottom of each column. C, Representative sample of the raw data used to generate the chart. Only pools expressing Epcam, MHC II, and HPRT were included in the analyses. Tgn, Thyroglobulin.

Nested PCR analysis of cDNA from pools and sorted single Fig. 7, Gip was expressed by only 1 of 23 MHC IIϩ pools (Fig. 7A cells confirms the observed TRA expression patterns and data not shown), and 0 of 40 MHC IIϩ individual mTECs (Fig. We wanted to confirm the small pool data described above using 7B and data not shown). These results were consistent with the a different method (20) and extend our observations to the single- pool data in Figs. 5 and 6 and confirm that expression of an indi- cell level. To accomplish this, nested gene-specific PCR was per- vidual TRA is limited to rare individual mTECs. by guest on September 23, 2021 formed with cDNA from both small pools (25 cells/pool; Fig. 7A) ϩ Ϫ aire expression alone is not sufficient for the coexpression of and individual (Fig. 7B) sorted MHC IIhighEpcam CD45 cells. aire-dependent TRAs in sorted mTECs Although the use of nested PCR with cDNAs generated using gene-specific primers allowed for increased sensitivity, the spec- The expression of many of the TRAs assessed in Figs. 5 and 6 have trum of molecules that could be analyzed was significantly smaller. been shown to be aire dependent in mTECs, including Csng, elas- Amplified cDNA from all pools and single cells examined were tase 3B, Gip, iFABP, Ins2, major urinary protein (mup), Pcp4, and first validated via secondary PCR for expression of MHC II. We salivary protein 1 (Spt1; Ref. 22). Although almost all of these then tested these pools and single cells for the presence of tran- pools expressed aire (only pool 13 of the Epcamhigh mTECs did scripts for aire and the aire-dependent TRA Gip (22). As shown in not; see Fig. 5), few of these pools expressed aire-dependent TRAs, and none of the pools expressed all, or even a majority of them. No individual aire-dependent TRA was expressed in Ͼ5of the 21 pools that express aire. These results indicate that the ex- pression of aire in an mTEC was not sufficient for the simultaneous coexpression of aire-dependent TRAs. The data generated from the pools analyzed by gene-specific nesting (Fig. 7) was consistent with this finding. We found that MHC IIhighEpcamϩCD45Ϫ mTECs express aire at a relatively low frequency (5 of 40 single cells; Fig. 4B and data not shown). As a population, these MHC IIhigh mTECs have been previously shown to express aire at relatively high levels (29); our finding that only 5 of 40 single cells and 15 of 23 pools tested express aire shows that only a minority of mTECs within the MHC IIhigh population actively express aire, a frequency consistent with immunohisto- chemical staining patterns for aire observed in adult thymus (Ref. 50 and A. G. Farr and G. O. Gillard, unpublished observations). As FIGURE 7. Sorted MHC IIhigh mTECs are heterogeneous. Nested PCR was performed on pools (25 cells/pool, A) or individual (B) sorted MHC in our earlier experiments, expression of aire within an individual high cell or pool did not correspond well with Gip expression. Gip, an II mTECs to determine expression of MHC II, aire, and the aire-de- ϩ pendent TRA Gip. Data represent 12 of 24 pool samples (reverse tran- aire-dependent TRA, was expressed within only 1 of the 15 aire ϩ scriptase was not added to the first pool) and 11 of 40 single cells (MHC pools, and by none of the 5 aire single cells. Whether IIϪ wells were not included). the expression of Gip requires that aire be coexpressed within the 5822 mTEC HETEROGENEITY AND TRA EXPRESSION same cell still remains to be determined, because the frequency of that a number of transcription factors normally associated with Gip-expressing cells was so low that examination of a significantly multipotential cells, including Oct4, nanog, and Sox2, are ex- large cohort of sorted single cells will be required to address this pressed within the mTEC population. Defining the expression pat- question. tern of these transcription factors within the mTEC population may prove useful in parsing mTEC heterogeneity and facilitate the Discussion identification of the progenitor epithelial cell population in the In this study, we have defined several important and novel char- adult thymus. acteristics of thymic epithelium that are central to understanding That TECs can adopt alternative epithelial fates is supported by three elements of thymic epithelial biology that we believe are several lines of experimental evidence. In the absence of functional interrelated: the nature of thymic epithelial progenitor cells; the , the thymic rudiment of nude mice displays the morphology process of mTEC differentiation; and the generation of mTEC het- and organization of early fetal lung and a respiratory phenotype erogeneity. The properties of mTEC described here lend support to (63). One interpretation of this finding is that the progenitor pop- a developmental model for mTEC diversity previously proposed ulation in the nude thymus is multipotential and that foxn1 is re- (14), where this population is maintained by the steady state dif- quired to specify a thymic epithelial fate. There are also indications ferentiation of an epithelial progenitor population with broad de- that mTEC in normal mice adopt alternative developmental fates. velopmental potential within the adult thymus. This model predicts Cystic epithelial structures that are lined, in part, by ciliated epi- the proliferation and turnover of epithelial cells is ongoing in the thelial cells and display a wide range of respiratory system-related adult thymus, the presence of a subset of progenitor cells within genes, including plunc, surfactants, and CC10, are commonly the mTEC population that express the molecules characteristic of found in normal thymus (16). Foci of epithelial cells resembling Downloaded from multipotential progenitor cells, and the expression by mTECs of small thyroid follicles within the normal thymus have also been other transcription factors that are involved in the specification described (13). Although these latter instances may represent ex- and/or differentiation of other epithelial lineages. The heterogene- treme examples of the developmental potential of thymic epithelial ity of mTECs is reflected in the molecules they express; these progenitor cells where terminal differentiation has occurred, they molecules include TRAs. Although we cannot yet test this model highlight the potential plasticity of epithelial cells in the thymus.

directly, we have demonstrated here that the properties of mTECs Our results demonstrating the expression of peripheral develop- http://www.jimmunol.org/ are compatible with these proposed requirements. mental transcription factors by mTECs is significant in two re- First, generation of differentiated mTEC from a progenitor pop- spects. First, these findings sustantially expand the number of can- ulation predicts a level of epithelial turnover. We have shown that didate transcription factors that may be involved in mTEC proliferation of TEC continues in the postnatal thymus, albeit at a differentiation, thus providing additional tools for analyzing thy- relatively low level. A basal level of proliferation of mTECs in- mic epithelial heterogeneity. Second, accepting the premise that dicates that this population turns over and likely has the capacity mTEC heterogeneity arises as a consequence of differentiation and for self-renewal. On the basis of other differentiating systems, one that TRA expression is a reflection of this process, it is reasonable could envision low rate of turnover of the least mature progenitor to propose that conventional, lineage-restricted developmental cells, followed by a high rate of turnover and expansion of a transit transcriptional networks could regulate expression of downstream by guest on September 23, 2021 amplifying population, which then becomes postmitotic and ter- TRAs in mTECs. We have previously shown that Ttf1 (Nkx2.1), minally differentiates. The capacity of TECs to proliferate in re- which is centrally involved in lung and thyroid development, is sponse to exogenous keratinocyte growth factor in vivo and in expressed by mTEC (16), as are multiple lung- and thyroid-spe- vitro (51) suggest that this signaling pathway may play a critical cific TRAs (3, 16, 22, 29). We have shown here that mTEC also role in regulating TE turnover in postnatal thymus. express a number of transcription factors that play key roles in the The continuing differentiation of mTECs in the postnatal thymus differentiation of other peripheral tissues; the mTEC population presents alternative, but not mutually exclusive mechanisms for also expresses a number of TRAs characteristic of these peripheral the generation of mTEC heterogeneity and the regulation of TRA tissues. Because some of these transcription factors are required expression. Differentiating epithelial cells could access multiple for efficient expression of TRAs in peripheral tissues, their func- developmental pathways before specification to a terminal, per- tional inactivation might result in lacunar, lineage-restricted defi- haps thymic fate choice. The ability of immature cell populations ciencies in TRA expression by mTEC. The observation that mul- to express transcription factors characteristic of alternate lineages tiple Pdx1-dependent and lineage-related TRAs are coexpressed before commitment to a specified terminal lineage has been ob- within small pools of sorted mTEC supports this notion that tissue- served in multiple settings (20, 52–55) and may also be occurring specific transcriptional networks may be functioning in mTEC as in thymic epithelium. A multipotential thymic epithelial progenitor in their native peripheral epithelial tissues. population could respond to developmental cues within the thymic We favor the interpretation that expression of these tissue-spe- microenvironment to adopt alternate epithelial cell fates. Thymic cific transcription factors (and their downstream targets) results epithelium shares space with a complex mesenchymal compart- from conserved developmental mechanisms during mTEC differ- ment, together forming an environment replete with elements of entiation; however, an alternative interpretation is that the dere- several signaling pathways (bone morphogenetic protein (56, 57), pression of chromatin in mature mTECs leads to the promiscuous Wnt (58), fibroblast growth factor (51, 59), Notch (60), and expression of multiple developmental regulators (and their down- Hedgehog (61, 62)) that play critical roles in the specification and stream targets) by mature mTECs. Given the gaps in our knowl- differentiation of a range of other epithelial tissues. edge regarding the extent of mTEC heterogeneity, and the devel- Invoking multipotential epithelial progenitor cells as a basis for opmental relationships among mTEC populations, we cannot mTEC heterogeneity and TRA expression in the adult thymus is presently distinguish between these two interpretations. partially based on reports that described an epithelial progenitor Our examination of individual and small pools of sorted mTECs population in the fetal thymus (8, 9) and the assumption that these revealed that expression of a given TRA is limited to rare mTECs cells persist postnatally. mTEC progenitors that are multipotential and that individual mTECs do not express a broad range of TRAs. and not lineage restricted would be predicted to share the molec- These data strongly support the notion that mTECs represent a ular signature of other multipotential populations. We have shown diverse population and that TRA expression reflects that diversity. The Journal of Immunology 5823

Previous reports have examined TRA expression in pools of 104– 11. Min, D., P. A. Taylor, A. Panoskaltsis-Mortari, B. Chung, D. M. Danilenko, 106 mTECs (3, 22, 29); by using bulk populations, the diversity of C. Farrell, D. L. Lacey, B. R. Blazar, and K. I. Weinberg. 2002. Protection from thymic epithelial cell injury by keratinocyte growth factor: a new approach to the mTEC population is not apparent because the results represent improve thymic and peripheral T-cell reconstitution after bone marrow transplan- averaged expression from many cells. If the TRA expression re- tation. Blood 99: 4592–4600. 12. Rossi, S., B. R. Blazar, C. L. Farrell, D. M. Danilenko, D. L. Lacey, sults for the pools shown in Fig. 6 are summed, almost all of the K. I. Weinberg, W. Krenger, and G. A. Hollander. 2002. Keratinocyte growth TRAs examined were expressed (12 of 17 TRA expressed by 225 factor preserves normal thymopoiesis and thymic microenvironment during ex- MHC IIhigh mTEC). On the basis of such a summation, one might perimental graft-versus-host disease. Blood 100: 682–691. 13. Farr, A. G., J. L. Dooley, and M. Erickson. 2002. Organization of thymic med- conclude that the genes within the casein locus are coordinately ullary epithelial heterogeneity: implications for mechanisms of epithelial differ- expressed; examination of smaller samples reveals that these genes entiation. Immunol. Rev. 189: 20–27. are being independently regulated within individual mTECs. 14. Gillard, G. O., and A. G. Farr. 2005. Contrasting models of promiscuous gene These small pool/single-cell analyses also provide insight into expression by thymic epithelium. J. Exp. Med. 202: 15–19. 15. Avichezer, D., R. S. Grajewski, C. C. Chan, M. J. Mattapallil, P. B. Silver, the relationship between the expression of aire and aire-dependent J. A. Raber, G. I. Liou, B. Wiggert, G. M. Lewis, L. A. Donoso, and R. R. Caspi. TRAs. The poor correlation between aire and aire-dependent TRA 2003. An immunologically privileged retinal antigen elicits tolerance: major role expression indicates that aire expression is not sufficient for the for central selection mechanisms. J. Exp. Med. 198: 1665–1676. 16. Dooley, J., M. Erickson, and A. G. Farr. 2005. An organized medullary epithelial simultaneous coexpression of TRAs in mTEC. This may reflect a structure in the normal thymus expresses molecules of respiratory epithelium and requirement for other cofactors or indicate a temporal separation resembles the epithelial thymic rudiment of nude mice. J. Immunol. 175: between aire and TRA expression. It is possible that the impor- 4331–4337. 17. Scherer, A., A. Krause, J. R. Walker, S. E. Sutton, D. Seron, F. Raulf, and tance of aire in regulating some TRAs reflects a more conventional M. P. Cooke. 2003. Optimized protocol for linear RNA amplification and appli- regulatory transcription factor activity that influences the ability of cation to profiling of human renal biopsies. BioTechniques 34: Downloaded from 546–550, 552–544, 556. immature mTEC to initiate or sustain developmental hierarchies. 18. Anderson, G., E. J. Jenkinson, N. C. Moore, and J. J. Owen. 1993. MHC class Alternatively, aire might regulate the temporal pattern of mTEC II-positive epithelium and mesenchyme cells are both required for T-cell devel- turnover, differentiation, or commitment, which could, in turn, im- opment in the thymus. Nature 362: 70–73. 19. Chiang, M. K., and D. A. Melton. 2003. Single-cell transcript analysis of pan- pact the duration of TRA expression or the temporal window dur- creas development. Dev. Cell 4: 383–393. ing which immature mTECs could follow alternate lineage path- 20. Hu, M., D. Krause, M. Greaves, S. Sharkis, M. Dexter, C. Heyworth, and ways. In each of these scenarios, lack of aire would result in T. Enver. 1997. Multilineage gene expression precedes commitment in the he- http://www.jimmunol.org/ lacunar defects in mTEC differentiation and a reduced spectrum of mopoietic system. Genes Dev. 11: 774–785. 21. Kay, M. M. 1979. An overview of immune aging. Mech. Ageing Dev. 9: 39–59. TRA expression. 22. Anderson, M. S., E. S. Venanzi, L. Klein, Z. Chen, S. P. Berzins, S. J. Turley, Our data establish that the mTEC population is dynamic and H. von Boehmer, R. Bronson, A. Dierich, C. Benoist, and D. Mathis. 2002. diverse and that it expresses a wide range of transcription factors Projection of an immunological self shadow within the thymus by the aire pro- tein. Science 298: 1395–1401. known to regulate the developmental potential and lineage speci- 23. Blackburn, C. C., and N. R. Manley. 2004. Developing a new paradigm for fication of multiple cell types. These results provide the basis for thymus organogenesis. Nat. Rev. Immunol. 4: 278–289. a novel framework for understanding and investigating the related 24. Bowles, J., R. P. Teasdale, K. James, and P. Koopman. 2003. Dppa3 is a marker of pluripotency and has a human homologue that is expressed in germ cell tu- processes of thymic epithelial cell potential, differentiation, and mours. Cytogenet. Genome Res. 101: 261–265. function. 25. Jackson, M., J. W. Baird, J. Nichols, R. Wilkie, J. D. Ansell, G. Graham, and by guest on September 23, 2021 L. M. Forrester. 2003. Expression of a novel homeobox gene Ehox in trophoblast stem cells and pharyngeal pouch endoderm. Dev. Dyn. 228: 740–744. Acknowledgments 26. Hanna, L. A., R. K. Foreman, I. A. Tarasenko, D. S. Kessler, and P. A. Labosky. We thank Dr. A. Rudensky, Dr. M. Bevan, Dr. P. Greenberg, and M. 2002. Requirement for Foxd3 in maintaining pluripotent cells of the early mouse Erickson for helpful discussions and/or critical review of the manuscript, embryo. Genes Dev. 16: 2650–2661. and Dr. P. Morrissey and T. George at Amnis for use of the Image- 27. Okuda, A., A. Fukushima, M. Nishimoto, A. Orimo, T. Yamagishi, stream instrument and data analysis. Y. Nabeshima, M. Kuro-o, K. Boon, M. Keaveney, H. G. Stunnenberg, and M. Muramatsu. 1998. UTF1, a novel transcriptional coactivator expressed in pluripotent embryonic stem cells and extra-embryonic cells. EMBO J. 17: 2019– Disclosures 2032. The authors have no financial conflict of interest. 28. Boyer, L. A., T. I. Lee, M. F. Cole, S. E. Johnstone, S. S. Levine, J. P. Zucker, M. G. Guenther, R. M. Kumar, H. L. Murray, R. G. Jenner, et al. 2005. Core transcriptional regulatory circuitry in human embryonic stem cells. Cell 122: References 947–956. 1. Laufer, T. M., L. H. Glimcher, and D. Lo. 1999. Using thymus anatomy to dissect 29. Derbinski, J., J. Gabler, B. Brors, S. Tierling, S. Jonnakuty, M. Hergenhahn, T cell repertoire selection. Semin. Immunol. 11: 65–70. L. Peltonen, J. Walter, and B. Kyewski. 2005. Promiscuous gene expression in 2. Sprent, J., and H. Kishimoto. 2001. The thymus and central tolerance. Philos. thymic epithelial cells is regulated at multiple levels. J. Exp. Med. 202: 33–45. Trans. R. Soc. Lond. B 356: 609–616. 30. Weinstein, D. C., A. Ruiz i Altaba, W. S. Chen, P. Hoodless, V. R. Prezioso, 3. Derbinski, J., A. Schulte, B. Kyewski, and L. Klein. 2001. Promiscuous gene T. M. Jessell, and J. E. Darnell, Jr. 1994. The winged-helix transcription factor expression in medullary thymic epithelial cells mirrors the peripheral self. Nat. HNF-3␤ is required for notochord development in the mouse embryo. Cell 78: Immunol. 2: 1032–1039. 575–588. 4. Smith, K. M., D. C. Olson, R. Hirose, and D. Hanahan. 1997. Pancreatic gene 31. Behr, R., J. Brestelli, J. T. Fulmer, N. Miyawaki, T. R. Kleyman, and expression in rare cells of thymic medulla: evidence for functional contribution K. H. Kaestner. 2004. Mild nephrogenic diabetes insipidus caused by Foxa1 to T cell tolerance. Int. Immunol. 9: 1355–1365. deficiency. J. Biol. Chem. 279: 41936–41941. 5. Hanahan, D. 1998. Peripheral-antigen-expressing cells in thymic medulla: factors 32. Gao, N., K. Ishii, J. Mirosevich, S. Kuwajima, S. R. Oppenheimer, R. L. Roberts, in self-tolerance and autoimmunity. Curr. Opin. Immunol. 10: 656–662. M. Jiang, X. Yu, S. B. Shappell, R. M. Caprioli, M. Stoffel, S. W. Hayward, and 6. Klein, L., M. Klugmann, K. A. Nave, V. K. Tuohy, and B. Kyewski. 2000. R. J. Matusik. 2005. Forkhead box A1 regulates prostate ductal morphogenesis Shaping of the autoreactive T-cell repertoire by a splice variant of self protein and promotes epithelial cell maturation. Development 132: 3431–3443. expressed in thymic epithelial cells. Nat. Med. 6: 56–61. 7. Klein, L., T. Klein, U. Ruther, and B. Kyewski. 1998. CD4 T cell tolerance to 33. Lantz, K. A., and K. H. Kaestner. 2005. Winged-helix transcription factors and human C-reactive protein, an inducible serum protein, is mediated by medullary pancreatic development. Clin. Sci. 108: 195–204. thymic epithelium. J. Exp. Med. 188: 5–16. 34. Lee, C. S., J. R. Friedman, J. T. Fulmer, and K. H. Kaestner. 2005. The initiation 8. Bennett, A. R., A. Farley, N. F. Blair, J. Gordon, L. Sharp, and C. C. Blackburn. of liver development is dependent on Foxa transcription factors. Nature 435: 2002. Identification and characterization of thymic epithelial progenitor cells. 944–947. Immunity 16: 803–814. 35. Mirosevich, J., N. Gao, and R. J. Matusik. 2005. Expression of Foxa transcription 9. Gill, J., M. Malin, G. A. Hollander, and R. Boyd. 2002. Generation of a complete factors in the developing and adult murine prostate. Prostate 62: 339–352. thymic microenvironment by MTS24ϩ thymic epithelial cells. Nat. Immunol. 3: 36. Wan, H., S. Dingle, Y. Xu, V. Besnard, K. H. Kaestner, S. L. Ang, S. Wert, 635–642. M. T. Stahlman, and J. A. Whitsett. 2005. Compensatory roles of Foxa1 and 10. Ropke, C., P. Van Soest, P. P. Platenburg, and W. Van Ewijk. 1995. A common Foxa2 during lung morphogenesis. J. Biol. Chem. 280: 13809–13816. stem cell for murine cortical and medullary thymic epithelial cells? Dev. Immu- 37. Meyer, B. I., and P. Gruss. 1993. Mouse -1 expression during gastrulation. nol. 4: 149–156. Development 117: 191–203. 5824 mTEC HETEROGENEITY AND TRA EXPRESSION

38. Subramanian, V., B. Meyer, and G. S. Evans. 1998. The murine Cdx1 gene limiting role of Aire in thymic expression, clonal deletion, and organ-specific product localises to the proliferative compartment in the developing and regen- autoimmunity. J. Exp. Med. 200: 1015–1026. erating intestinal epithelium. Differentiation 64: 11–18. 51. Erickson, M., S. Morkowski, S. Lehar, G. Gillard, C. Beers, J. Dooley, 39. Furukawa, T., E. M. Morrow, T. Li, F. C. Davis, and C. L. Cepko. 1999. Reti- J. S. Rubin, A. Rudensky, and A. G. Farr. 2002. Regulation of thymic epithelium nopathy and attenuated circadian entrainment in Crx-deficient mice. Nat. Genet. by keratinocyte growth factor. Blood 100: 3269–3278. 23: 466–470. 52. Kiefer, J. C., and S. D. Hauschka. 2001. Myf-5 is transiently expressed in non- 40. Guz, Y., M. R. Montminy, R. Stein, J. Leonard, L. W. Gamer, C. V. Wright, and muscle mesoderm and exhibits dynamic regional changes within the preseg- G. Teitelman. 1995. Expression of murine STF-1, a putative insulin gene tran- mented mesoderm and somites I-IV. Dev. Biol. 232: 77–90. scription factor, in ␤ cells of pancreas, duodenal epithelium and pancreatic exo- 53. Miyamoto, T., H. Iwasaki, B. Reizis, M. Ye, T. Graf, I. L. Weissman, and crine and endocrine progenitors during ontogeny. Development 121: 11–18. K. Akashi. 2002. Myeloid or lymphoid promiscuity as a critical step in hemato- 41. Jonsson, J., U. Ahlgren, T. Edlund, and H. Edlund. 1995. IPF1, a homeodomain poietic lineage commitment. Dev. Cell 3: 137–147. protein with a dual function in pancreas development. Int. J. Dev. Biol. 39: 54. Rolink, A. G., C. Schaniel, L. Bruno, and F. Melchers. 2002. In vitro and in vivo 789–798. plasticity of Pax5-deficient pre-B I cells. Immunol. Lett. 82: 35–40. 42. Offield, M. F., T. L. Jetton, P. A. Labosky, M. Ray, R. W. Stein, M. A. Magnuson, 55. Akashi, K., X. He, J. Chen, H. Iwasaki, C. Niu, B. Steenhard, J. Zhang, J. Haug, B. L. Hogan, and C. V. Wright. 1996. PDX-1 is required for pancreatic outgrowth and L. Li. 2003. Transcriptional accessibility for genes of multiple tissues and and differentiation of the rostral . Development 122: 983–995. hematopoietic lineages is hierarchically controlled during early hematopoiesis. 43. Jepeal, L. I., Y. Fujitani, M. O. Boylan, C. N. Wilson, C. V. Wright, and Blood 101: 383–389. M. M. Wolfe. 2005. Cell-specific expression of glucose-dependent-insulinotropic 56. Tsai, P. T., R. A. Lee, and H. Wu. 2003. BMP4 acts upstream of FGF in mod- polypeptide is regulated by the transcription factor PDX-1. Endocrinology 146: ulating thymic stroma and regulating thymopoiesis. Blood 102: 3947–3953. 383–391. 57. Bleul, C. C., and T. Boehm. 2005. BMP signaling is required for normal thymus 44. Samadani, U., and R. H. Costa. 1996. The transcriptional activator hepatocyte development. J. Immunol. 175: 5213–5221. nuclear factor 6 regulates liver gene expression. Mol. Cell Biol. 16: 6273–6284. 58. Balciunaite, G., M. P. Keller, E. Balciunaite, L. Piali, S. Zuklys, Y. D. Mathieu, 45. Landry, C., F. Clotman, T. Hioki, H. Oda, J. J. Picard, F. P. Lemaigre, and J. Gill, R. Boyd, D. J. Sussman, and G. A. Hollander. 2002. Wnt glycoproteins G. G. Rousseau. 1997. HNF-6 is expressed in endoderm derivatives and nervous regulate the expression of FoxN1, the gene defective in nude mice. Nat. Immunol. system of the mouse embryo and participates to the cross-regulatory network of 3: 1102–1108. liver-enriched transcription factors. Dev. Biol. 192: 247–257. 59. Revest, J. M., R. K. Suniara, K. Kerr, J. J. Owen, and C. Dickson. 2001. Devel- Downloaded from 46. Rausa, F., U. Samadani, H. Ye, L. Lim, C. F. Fletcher, N. A. Jenkins, opment of the thymus requires signaling through the fibroblast growth factor N. G. Copeland, and R. H. Costa. 1997. The cut-homeodomain transcriptional receptor R2-IIIb. J. Immunol. 167: 1954–1961. activator HNF-6 is coexpressed with its target gene HNF-3␤ in the developing 60. Anderson, G., J. Pongracz, S. Parnell, and E. J. Jenkinson. 2001. Notch ligand- murine liver and pancreas. Dev. Biol. 192: 228–246. bearing thymic epithelial cells initiate and sustain Notch signaling in thymocytes 47. Xuan, S., C. A. Baptista, G. Balas, W. Tao, V. C. Soares, and E. Lai. 1995. independently of T cell receptor signaling. Eur. J. Immunol. 31: 3349–3354. Winged helix transcription factor BF-1 is essential for the development of the 61. Outram, S. V., A. Varas, C. V. Pepicelli, and T. Crompton. 2000. Hedgehog cerebral hemispheres. Neuron 14: 1141–1152. signaling regulates differentiation from double-negative to double-positive thy-

48. Hebert, J. M., and S. K. McConnell. 2000. Targeting of cre to the Foxg1 (BF-1) mocyte. Immunity 13: 187–197. http://www.jimmunol.org/ locus mediates loxP recombination in the telencephalon and other developing 62. Sacedon, R., A. Varas, C. Hernandez-Lopez, C. Gutierrez-deFrias, T. Crompton, head structures. Dev. Biol. 222: 296–306. A. G. Zapata, and A. Vicente. 2003. Expression of hedgehog in the 49. Rijnkels, M., D. A. Wheeler, H. A. de Boer, and F. R. Pieper. 1997. Structure and human thymus. J. Histochem. Cytochem. 51: 1557–1566. expression of the mouse casein gene locus. Mamm. Genome 8: 9–15. 63. Dooley, J., M. Erickson, H. Roelink, and A. G. Farr. 2005. Nude thymic rudiment 50. Liston, A., D. H. Gray, S. Lesage, A. L. Fletcher, J. Wilson, K. E. Webster, lacking functional foxn1 resembles respiratory epithelium. Dev. Dyn. 233: H. S. Scott, R. L. Boyd, L. Peltonen, and C. C. Goodnow. 2004. Gene dosage- 1605–1612. by guest on September 23, 2021