These two authors contributed equally to this work.
Cellular immune response
Redefining epithelial progenitor potential in the developing thymus
Article first published online: 13 AUG 2007
Copyright © 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
European Journal of Immunology
Volume 37, Issue 9, pages 2411–2418, September 2007
How to Cite
Rossi, Simona W., Chidgey, Ann P., Parnell, Sonia M., Jenkinson, William E., Scott, Hamish S., Boyd, Richard L., Jenkinson, Eric J. and Anderson, G. (2007), Redefining epithelial progenitor potential in the developing thymus. Eur. J. Immunol., 37: 2411–2418. doi: 10.1002/eji.200737275
- Issue published online: 27 AUG 2007
- Article first published online: 13 AUG 2007
- Manuscript Accepted: 26 JUN 2007
- Manuscript Revised: 18 MAY 2007
- Manuscript Received: 16 MAR 2007
- EU FP6 Thymaide Project
- MRC Programme
- Norwood Immunology
- Australian Stem Cell Centre
- Cell differentiation;
- Lymphoid organs;
Cortical and medullary epithelium represent specialised cell types that play key roles in thymocyte development, including positive and negative selection of the T cell repertoire. While recent evidence shows that these epithelial lineages share a common embryonic origin, the phenotype and possible persistence of such progenitor cells in the thymus at later stages of development remain controversial. Through use of a panel of reagents including the putative progenitor marker Mts24, we set out to redefine the stages in the development of thymic epithelium. In the early embryonic day (E)12 thymus anlagen we find that almost all epithelial cells are uniformly positive for Mts24 expression. In addition, while the thymus at later stages of development was found to contain distinct Mts24+ and Mts24– epithelial subsets, thymus grafting experiments show that both Mts24+ and Mts24– epithelial subsets share the ability to form organised cortical and medullary thymic microenvironments that support T cell development, a function shown previously to be lost in the Mts24– cells by E15 when lower cell doses were used. Our data help to clarify stages in thymic epithelial development and provide important information in relation to currently used markers of epithelial progenitors.
See accompanying commentary: http://dx.doi.org/10.1002/eji.200737709
epithelial cell adhesion molecule 1
fibroblast growth factor receptor 2
Establishment of specialised epithelial microenvironments in the thymus is essential for thymopoiesis and the generation of T cells 1, 2. This microenvironment provides key signals that support the recruitment of haemopoietic progenitors 3 as well as their subsequent development and selection. Thus, in the adult, the migration of T cell progenitors through thymic cortical regions to the subcapsular area is accompanied by their progression through the CD4–CD8– double-negative stages of development, involving differentiation, expansion and antigen receptor gene rearrangement 4, 5. Migration back through the thymic cortex towards the medulla is associated with the generation of immature CD4+CD8+ cortical thymocytes that, as a result of intrathymic selection events, give rise to functionally competent CD4+ and CD8+ cells 5.
This programme of T cell development involves sequential interactions with distinct thymic stromal cells that form specialised microenvironments in the thymus. Thymic stromal cells are heterogeneous, consisting of epithelial and mesenchymal cells, endothelial cells, as well as bone marrow-derived macrophages and dendritic cells 6. Of these cell types, thymic cortical and medullary epithelial cells have been shown to provide essential signals at multiple stages of thymocyte development. For example, MHC class II+ epithelial cells in the thymic cortex are known to be important in αβT cell receptor-mediated positive selection 7, 8, while medullary epithelial cells, in addition to dendritic cells, are important in ensuring tolerance to self antigens 9, 10.
The epithelial component of the thymus arises from the out-budding of endoderm in the third pharyngeal pouch that begins around day 10–11 of gestation 11. Recent advances have led to a better understanding of the genetic mechanisms regulating thymic epithelial cell development, with the identification of a role for a number of genes including FoxN1, Pax9, Hoxa3 and Eya1 11. However, despite these advances, the developmental stages and events that lead to formation of the distinct cortical and medullary epithelial subsets required to regulate T cell development are still not fully understood 12. Elucidating the early stages of thymic epithelial cell development has been hampered by the paucity of appropriate markers distinguishing epithelial subsets 13.
Recent studies have provided evidence for the existence of a common bipotent progenitor for cortical and medullary epithelium that is present in both the embryonic and neonatal thymus 14, 15. However, the phenotypic characteristics of thymic epithelial progenitors and whether or not they persist in the thymus at later stages remains controversial. In this regard, the antibody Mts24 has been used to define heterogeneity within thymic epithelium, with equal proportions of Mts24+ and Mts24– thymic epithelial cells being reported as early as embryonic day (E)12 of gestation 16, a stage where bipotent epithelial progenitors are known to be present 14. Moreover, the ability to give rise to functional thymic tissue containing both cortical and medullary epithelial lineages has been thought to be restricted to this Mts24+ subset 16, 17.
In this study, we have analysed the ontogeny and functional capacity of thymic stromal cell populations defined by the expression of Mts24, in the context of the pan-epithelial markers epithelial cell adhesion molecule 1 (EpCAM1) and cytokeratin to exclude the possible involvement of non-epithelial cells. In the E12 thymus we find that the majority (approximately 95%) of thymic epithelial cells are homogeneous with regard to expression of Mts24, and that heterogeneity of Mts24 expression in the thymic epithelial compartment is not clearly detectable until E14. Importantly, using higher thymic epithelial cell numbers to reveal more subtle functional properties, we show that ectopic grafting of reaggregate thymuses formed from either purified Mts24+ or purified Mts24– epithelial cells, isolated at successive stages of development, both retain the ability to form functional thymic tissue that is capable of supporting T cell development. We discuss our findings in relation to previous observations on thymic epithelial progenitors and current models of thymic epithelial development.
Mts24 fails to distinguish epithelial progenitor subsets in the E12 thymus
The formation of distinct cortical and medullary epithelial subsets in the thymus first occurs during embryonic development 1, 11 and we have recently shown that many of the epithelial cells in the E12 rudiment are bipotent progenitors able to give rise to both cortical and medullary epithelium 14. Expression of Mts24 has been associated with progenitor potential in the early thymus rudiment 16, 17 and has been reported to define a subset of cells with this potential in the E12 rudiment. However, it remains unclear whether these cells represent a specific subset of self-renewing progenitors or whether Mts24 is a developmental stage-specific marker initially expressed on all epithelial cells at E12.
To relate the expression of Mts24 to epithelial progenitor potential, we prepared frozen tissue sections of freshly dissected E12 thymic anlagen. Immunohistochemical staining using two different epithelial markers, pan-cytokeratin and EpCAM1 18, identifies the epithelial compartment of the E12 thymus as a central core surrounded by a non-epithelial capsule (Fig. 1A). Both these markers give a similar expression pattern suggesting that EpCAM1 expression is inclusive of all epithelial populations within the E12 rudiment. Analysis of staining with antibody Mts24 (Fig. 1A) also showed extensive labelling that was coincident with the pan-cytokeratin and EpCAM1. Thus, immunohistochemical analysis of tissue sections argues strongly that E12 thymic epithelial cells are homogeneous with respect to Mts24 expression.
To quantitate and provide direct confirmation of the overlap between pan-epithelial marker expression and Mts24 expression on a per cell basis in the E12 thymic rudiment, we disaggregated isolated thymus lobes and analysed the epithelial component by flow cytometry. Fig. 1B shows that the E12 thymus contains a distinct cytokeratin-positive population, while in agreement with immunohistochemical data (Fig. 1A), two-colour flow cytometric analysis showed that this cytokeratin-positive population is also uniformly positive for expression of EpCAM1 (Fig. 1B). We have also recently shown that those stromal cells in the E12 rudiment that are negative for epithelial markers are positive for the mesenchymal marker platelet-derived growth factor receptor α 19, providing confirmation that both the epithelial markers used here are inclusive of all epithelial cells in the E12 rudiment. Strikingly, when flow cytometric analysis of Mts24 expression was carried out in conjunction with either of these pan-epithelial markers we found that the vast majority of the epithelial cells present in the E12 thymus are Mts24+ (Fig. 1B). This confirms our histological findings and provides further evidence that Mts24 does not distinguish a specific progenitor subset within epithelial populations present at the E12 stage of thymus formation.
Heterogeneity of Mts24 expression in epithelial cells emerges during thymus development
In view of our findings that at E12 almost all epithelial cells are Mts24+, whereas it is known that all but a small minority of epithelial cells in the adult thymus lack Mts24 expression 17, we re-examined the emergence of Mts24-defined subsets in the epithelial compartment of the thymus at successive stages of ontogeny. Disaggregated suspensions from the thymuses of mouse embryos between E13 and E18 of gestation were co-labelled with Mts24 in conjunction with pan-cytokeratin or EpCAM1 as pan-epithelial markers to identify all thymic epithelial cells (Fig. 2A–C). Analysis of expression of Mts24 together with either pan-cytokeratin or EpCAM1 shows that while all epithelial cells at E13 of gestation are still Mts24+ (Fig. 2), a distinct population of Mts24– epithelial cells emerges at E14 (Fig. 2B, C). Thereafter, the proportion of Mts24+ cells declines until by E18 it represents only approximately 3–4% of the total epithelial population (Fig. 2B, C).
Both Mts24+ and Mts24– thymic epithelial cells generate functional thymic tissue
The ontogenetic data in the previous section show that epithelial cells in the E12 thymic rudiment are mostly Mts24+ whereas by E14, distinct Mts24+ and Mts24– populations have emerged. This indicates a potential precursor-product relationship between Mts24+ and Mts24– populations, and raises the possibility that the latter represent more developmentally advanced cells that have lost progenitor potential. Consistent with this, previous studies on the developing thymus have shown that reaggregates of purified Mts24+ cells are able to form functional thymic tissue when grafted ectopically into adult mice whereas reaggregates of age-matched Mts24– cells are not 16, 17. These observations have given rise to the notion that Mts24 expression marks cells with epithelial progenitor potential and that loss of Mts24 expression correlates with loss of this potential.
To relate this functional heterogeneity defined by Mts24 expression to the genetic characteristics of epithelial progenitor populations we used quantitative PCR to examine the expression of genes associated with progenitor status in both Mts24+ and Mts24– subsets of E14 thymic epithelium. Both populations were found to express similar levels of FoxN1 (Fig. 3D), a gene known to be important in thymic epithelial cell development 20, 21. Surprisingly, however, expression levels of p63 22 and bcrp1 19, genes associated with epithelial progenitors in various tissues, were comparable in both Mts24+ and Mts24– subsets of E14 thymic epithelium (Fig. 3E, F). In addition, expression of IL-7 (Fig. 3B) and fibroblast growth factor receptor 2 (FGFR2; Fig. 3C), the latter are a known regulator of proliferation of immature thymic epithelial cells 23, 24, were also both detectable in both Mts24+ and Mts24– E14 thymic epithelial cells. Collectively, these findings demonstrate the expression of a set of genes, characteristic of epithelial progenitor populations, in both Mts24+ and Mts24– thymic epithelial cells, despite the functional differences in progenitor potential previously reported to exist between these populations.
In the light of these observations, we sought to re-examine the developmental potential of Mts24+ and Mts24– populations when this was strictly correlated with a definitive epithelial phenotype. Thus EpCAM1+ epithelial cells from E14, E16 and E18 were separated into Mts24+ and Mts24– populations and used to form reaggregate thymus organ cultures that were transplanted under the kidney capsule of wt mice. After 4 wk, grafts were harvested and analysed for cortical and medullary epithelial differentiation and for their capacity to support T cell development.
Using reaggregates prepared from cells isolated from E14 and E16, thymus grafts were recovered at a similar frequency (Fig. 4A), were of a similar size (Fig. 4C) and supported the generation of similar numbers of thymocytes (Fig. 4B), irrespective of whether they were formed from the same starting number of Mts24+ or Mts24– epithelial cells. Moreover, flow cytometric analysis of thymocytes recovered from both Mts24+ (Fig. 4D) and Mts24– grafts (Fig. 4D) showed a normal programme of T cell development, while immunohistochemical analysis of frozen sections of grafts initiated from both Mts24+ or Mts24– epithelial cells revealed normal thymic architecture and the presence of cortical and medullary epithelial subsets, including Aire+ medullary epithelial cells, in both types of graft (Fig. 5).
Interestingly, although Mts24– epithelial cells were readily isolated from E18 thymus and were consistently found to generate functional thymic tissue upon transplantation (Fig. 4A–D), E18 Mts24+ epithelial cells failed to form into coherent RTOC for transplantation. On the single occasion when a coherent RTOC was formed from E18 Mts24+ epithelial cells and transplanted under the kidney capsule, analysis of the graft site and secondary lymphoid tissue of the recipient mouse for the presence of T cells provided no evidence for the formation of functional thymic tissue.
The establishment of differentiated epithelial microenvironments in the thymus is essential for the normal development of a functionally competent, self-MHC-restricted and self-tolerant T cell pool 1, 2, 5, 10. Recent studies have focused on the nature of the epithelial progenitors giving rise to the thymus, with evidence from clonal assays that bipotent progenitors giving rise to cortical and medullary lineages are present in the early thymic rudiment 14 and persist in the neonate 15. However, the phenotypic characterisation of these cells that would facilitate their isolation remains unclear.
Using a combination of the pan-epithelial markers EpCAM1 and cytokeratin, we show that the epithelial compartment of the E12 thymus is mostly (∼95%) homogeneous with regard to expression of Mts24. This marker has previously been reported to define a subset (approximately 50%) of epithelial cells in the E12 thymus with precursor activity being restricted to this subset even at this early stage 16. It seems unlikely that this discrepancy with our current findings is due to differences in the mouse strains used (in this study BALB/c as compared to C57BL/6, CBA F1 mice, used in the study of Bennett et al. 16) as we have also seen similar homogeneity in Mts24 expression in E12 thymic epithelial cells in C57BL/6 embryos (not shown).
An alternative explanation may lie in the fact that we carefully related our analysis of Mts24 expression and sorting of Mts24+/– cells to epithelial populations using pan-epithelial markers. In contrast, earlier studies appear to separate E12 thymic cells on the basis of Mts20+24+ and Mts20–24– phenotypes without simultaneous analysis of expression of a pan-epithelial marker 16. Thus Mts24– cells isolated from the E12 thymic rudiment in this way may be predominantly non-epithelial cells, explaining their inability to give thymus grafts on transplantation 16. This possibility is supported by our observation that most Mts24– stromal cells present in the E12 thymus do not express cytokeratin or EpCAM1 (Fig. 1) and are predominantly positive for platelet-derived growth factor receptor α, a marker of mesenchymal cells (25 and data not shown). We conclude that epithelial cells in the E12 rudiment are homogeneous with regard to Mts24 expression and that this marker does not distinguish progenitor from non-progenitor populations at this stage. This is consistent with our recent observations using clonal assays showing that many epithelial cells at E12 are progenitors able to give rise to both cortical and medullary epithelium 14 and suggests that the majority of epithelial cells have progenitor potential at this stage.
Although we were unable to confirm clear heterogeneity of Mts24 expression at E12, we observed the presence of a substantial Mts24– epithelial subset by E14, a population that increased in proportion thereafter until by E18 Mts24+ epithelial cells were in the minority. Our findings on the functional capacity of Mts24– and Mts24+ populations at these stages are relevant to the ongoing debate over the existence and identity of a thymic epithelial progenitor population that may persist beyond the initial stages of thymus formation during embryogenesis into later developmental stages and possibly into the adult. We clearly show that Mts24– cells from developmental stages as late as E18 can give rise to functional thymus grafts. These grafts were comparable in size and ability to produce T cells to those derived from the same number of age-matched Mts24+ cells. Whether this reflects comparable progenitor potential within both these populations involving ongoing renewal, as in epithelial populations in gut or skin, or the fact that both populations are relatively static but long-lived, remains to be determined.
Our current findings differ from those previously reported where Mts24– epithelial cells isolated from E15.5 thymus failed to produce thymus grafts, albeit using far fewer cells 17. This discrepancy may reflect differences in efficiency for thymus formation or proliferative potential between Mts24+ and Mts24– cells at this stage that are only evident when very small numbers of cells are grafted. However, our observation that matched numbers of Mts24+ and Mts24– gave rise to grafts of similar size and are able to support the generation of similar numbers of thymocytes, implying the presence of similar numbers of epithelial cells which are known to be limiting for thymocyte production 26, argues against any difference in proliferative potential or longevity in these populations.
Finally, our finding that by E18 Mts24+ cells appear to have little capacity to generate functional grafts suggests that by this developmental stage, expression of this marker may be mainly associated with more quiescent, mature cells rather than those with high-efficiency thymus-forming potential. In this regard, we have found that Mts24+ thymic epithelial cells undergo proliferation following sex steroid ablation, thereby contributing to the regeneration of the age-induced atrophic thymus (data not shown), and we have recently shown an association of Mts24 expression with keratinocytes that have greater progenitor capacity in terms of proliferation and gene expression profiles 27. In summary it is clear from the present study that thymic progenitor epithelium capacity is not exclusively contained within the Mts24+ population. While previous studies 17 show that in the mid-embryonic stage of development Mts24+ cells are more efficient than the Mts24– cells, the present study shows that this can be overcome with higher cell doses and that the situation is reversed by E18. This would be consistent with significant developmental changes in the cellular basis to thymic formation and maintenance with age. It will now be very important to resolve more precisely these underlying mechanisms in the post-natal thymus.
Materials and methods
BALB/c (H-2d) mice were used for this study. In grafting experiments, adult BALB/c mice were used as the recipients of reaggregate thymus tissue. To generate timed matings, males and females were placed in cages overnight. The following morning, females were checked for the presence of a vaginal plug, with the day of detection designated as E0.
The following antibodies were used for flow cytometric analysis of stromal cells: allophycocyanin-conjugated anti-EpCAM1 (18, clone G8.8; a gift of A. Farr and M. Kim), anti-pan-cytokeratin-FITC (clone C-11; Sigma) and biotinylated Mts24, the latter detected using streptavidin-PE. Antibodies used to analyse thymocytes were: anti-CD4-PE (clone GK1.5), anti-CD8-FITC (clone 53-6.7), from eBioscience. For immunohistochemistry, the additional antibodies were used: anti-cortical epithelium (cytokeratin 8), anti-cytokeratin 5 (Covance) and anti-Aire (clone B1/02-5H12-2; a kind gift of Dr. Hamish Scott).
Flow cytometric analysis of thymic epithelial cells during ontogeny
Freshly dissected thymuses from the indicated developmental stages were digested using trypsin as described 23. Cells were stained with anti-CD45.2-FITC (clone 104; eBioscience) to identify haemopoietic elements, and all data shown are gated on CD45– stromal cells. For analysis of expression of cytokeratin, cells were first labelled for surface antigens (Mts24 or EpCAM1 as appropriate), and then permeabilised using Permeafix (Johnson and Johnson) as described 23. Samples were analysed using a BD-LSR machine with forward and side scatter gates set so as to exclude non-viable cells 23.
Reaggregate thymus organ cultures and kidney capsule transplantation
To form reaggregate thymus cultures, EpCAM1+Mts24+ and EpCAM1+Mts24– epithelial subsets were prepared from thymus lobes of the indicated developmental stages, after prior depletion of CD45+ haemopoietic cells using DynaBeads 21. Cells were sorted to high purity (greater than 98%) using a MoFlo high-speed cell sorter. EpCAM1+Mts24+ or EpCAM1+Mts24– cells as appropriate were then used to form reaggregate cultures as described 28. To reveal any previously unidentified functional properties when as few as 2500 cells were used 17, typically 100 000 epithelial cells were used to form each reaggregate thymus. After 36 h culture, intact reaggregates were grafted under the kidney capsule of syngeneic BALB/c mice as described 14, 16, 17.
Freshly isolated E12 thymus lobes, or reaggregate thymus lobes harvested from recipient mice after 4 wk, were frozen in OCT embedding compound, and mounted onto cryostat chucks. Sections at 5 μM thickness were cut and stained as described 29.
Real-time PCR analysis
Freshly purified EpCAM1+Mts24+ and EpCAM1+Mts24– E14 epithelial subsets were snap-frozen in liquid nitrogen, and high-purity cDNA was obtained from purified mRNA 25, using µMacs One-step cDNA synthesis kit, according to the manufacturer's instructions (Miltenyi Biotec, Auburn, CA). Real-time PCR was performed using SYBR Green with primers specific for HPRT, FoxN1, BCRP1, p63, FGFR2 and IL-7. For normalization, the cDNA samples were diluted appropriately to give an equal expression value in a real-time PCR assay for HPRT. PCR reactions were carried out in triplicate in 15-μl volumes in reaction buffer containing 1× SensiMix QPCR SYBR Green Mix (Quantace) and 200 nM of each primer. After an initial denaturation step (95°C for 10 min), cycling was performed at 95°C for 15 s, 58–62°C for 20 s, and 72°C for 5 s (40 cycles).
Reaction amplification efficiency and the Ct values were obtained from the Rotor Gene 6.0 software (Corbett Research). For calculation of the relative expression values for each sample normalised to HPRT, a mathematical model (Pfaffl) that takes gene-dependent differences in the amplification efficiency into account was carried out as described 30. Increase of mRNA levels lower than twofold was not considered significant. Specific amplification was verified by melt curve analysis and also by fractionation of PCR products on a 3% agarose gel that were identified by fragment size.
Primer sequences and amplicon sizes are as follows: HPRT: forward 5′-CCAGCGTCGTGATTAGCGATG-3′, reverse 5′-ATAGCCCCCCTTGAGCACACAGAG-3′ (200 bp); FOXN1: forward 5′-CTCGTCGTTTGTGCCTGAC-3′, reverse 5′-TGCCTCTTGTAGGGGTGGAAA-3′ (243 bp); BCRP1: forward 5′-GAACTCCAGAGCCGTTAGGAC-3′, reverse 5′-CAGAATAGCATTAAGGCCAGGTT-3′ (166 bp); p63: forward 5′-GCCTGGACTATTTCACGACCC-3′, reverse 5′-GAACTGTTCAGGGATCTTCAG-3′ (98 bp); FGFR2: forward 5′-TTCTCCTAGTTACCCCGACAC-3′, reverse 5′-AGGCAGACAGGGTTCATAAGG-3′ (94 bp); IL-7: forward 5′-TTCCTCCACTGATCCTTGTTCT-3′, reverse 5′-AGCAGCTTCCTTTGTATCATCA-3′ (200 bp).
We thank A. Farr, and M. Kim for antibodies, and R. Bird for cell sorting. This work was supported by the EU FP6 Thymaide Project, an MRC Programme Grant to E.J.J. and G.A., Norwood Immunology and the Australian Stem Cell Centre.