Several groups generated comprehensive panels of monoclonal antibodies (mAbs) recognizing distinct elements of thymic epithelium (see refs. 18 and 55-57). According to their reactivity, these antibodies have been grouped into markers recognizing pan-epithelium, subcapsular and septum structures, fibroblasts, cortical epithelium, medullary epithelium, Hassall's corpuscles (enigmatic structures in the medullary zones), endothelium, and "miscellaneous" structures. Most of these reagents are reactive to intracellular antigens and are therefore useful to describe thymus morphology under various physiological and pathological conditions, but are of limited use to purify viable subsets by cell surface phenotypes.
Some of these markers have been used to observe changes in thymus architecture in T-cell developmental mutants, and attempts have been made to correlate a block in T-cell development with a block in TEC development (46). It should be noted, however, that such correlation may be very indirect, and these systems are likely to be too complex to provide definitive and conclusive results.
3.1. In Vivo Reconstruction of a Functional Thymus Environment From Purified Epithelial Cells: Functional Properties of Thymic Epithelial Cell Grafts
Thymic epithelium has become experimentally accessible with the introduction of methods to dissociate and reassemble thymic epithelium in vitro (39). Using this approach, the requirements for the formation of a functional thymus could be examined in vitro. In this type of experiment, input of either stromal cells or progenitor cell types can be varied (e.g., see refs. 19, 30, 58, and 59). Both RFTOCs and FTOCs cannot be cultured for periods extending longer than approx 12 d. Thus, such thymus cultures represent only a "transient thymus"; consequently, the architecture of RFTOCs maintained in vitro does not resemble normal thymic structures (31,60).
Transplantation of cultured thymic epithelial fragments (61) or fused thymic tissue fragments derived from the third pharyngeal pouch (62) was reported some time ago. In both cases, a functional thymus structure was restored in vivo. Specifically, these grafts were colonized by lymphocyte progenitors from the host that developed into functional T cells. In these experiments, however, the input of thymic epithelial cells could not be varied qualitatively because the tissue fragments could not be manipulated at the level of a cell suspension. Such manipulation is only possible if thymic epithelium is first dissociated, then reassembled, and subsequently grafted into a recipient mouse.
The methodology and the initial results from RFTOC grafting experiments were first reported in 1996 (63) and were specified in 2000 (60). In these experiments, alymphoid RFTOCs were grafted under the kidney capsules of host mice. The recipients were either immunocompromised mice, and thus incompetent to reject allogeneic grafts (e.g., nude or recombination-activating genes [RAG]-deficient hosts), or histocompatible mice bearing a congenic marker. The congenic marker is required to identify the origin of thymocytes in the graft (host bone marrow origin vs "carryover" within the graft). The structural and functional in vivo properties after transplantation of reaggregates of thymic epithelium have yielded the following novel results:
1. When RFTOCs, assembled in vitro from purified fetal thymic epithelium, are cultured as alymphoid thymic "organoids" for several days, these structures develop the rigidity required for handling and subsequent transplantation.
2. RFTOC grafts can attract host (bone marrow)-derived T-cell progenitors from the circulation.
3. T-cell development proceeds along the well-defined stages also found in an endogenous thymus.
4. The fidelity of negative and positive selection is normal when "provoked" in TCR-transgenic recipient mice, in which negative and positive selection depend on MHC molecules absent from the host, but present exclusively on thymic epithelium in the graft.
Thus, RFTOC transplantation uncovered that purified thymic epithelium, starting from a single-cell suspension, has the remarkable capacity to self-reorganize into a structurally and functionally competent microenvironment promoting T-cell development in vivo (31,60,63-65).
3.2. Phenotype of Thymic Epithelial Cells With the Potential to Generate a Functional Thymus In Vivo
In vivo reconstruction of a functional thymus environment was originally performed using aggregates of either CD45- MHC class II+ epithelial cells (60) or all CD45- cells (31). CD45- MHC class II- epithelial cells were not sufficient. The fact that thymus epithelium formation potential resides in the CD45- MHC class II+ thymic epithelium (31,60,63) has been confirmed and extended in recent reports (64,65).
Gill and colleagues (64) showed that a functional thymus can be generated from a major subset of CD45- MHC class II+ epithelial cells, defined as MTS24+. In a similar study, Bennett and colleagues (65) defined thymus-forming cells as MTS20+MTS24+. It should be noted that CD45- MHC class II+ MTS24+ and MTS20+MTS24+ cells are essentially identical populations (64). Moreover, given that around 50% of CD45- MHC class II+ are MTS24+, and the other half of MHC class II+ are MTS24- (64), this population is enriched only by a factor of approx 2 compared to CD45- MHC class II+ epithelium.
Despite the fact that no clonal assays were performed, these two reports claimed evidence for an identification and characterization of thymic epithelial progenitor cells (64,65). Although it is not impossible that CD45-MHC class II+MTS24+ cells contain some epithelial progenitors, this conclusion is clearly not proven by the data shown in these reports. The grafts were assembled from bulk populations, and without single cell readouts, bulk experiments cannot yield information on precursor activities. Moreover, ratios of cell numbers (input vs output) have not been determined, probably owing to the difficulties in retrieving epithelial cells quantitatively from the grafts. Therefore, no attempts were made by Gill et al. or Bennett et al. to demonstrate an increase in epithelial cellularity, which is expected from precursor activity (64,65).
3.3. Reorganization of Thymic Epithelium in the Graft
RFTOC maintained in vitro lack recognizable medulla-cortex architecture. Indeed, they appear to be randomly organized. In marked contrast, his-tological examination of RFTOC grafts reveals a striking reappearance of proper medulla-cortex organization in vivo (31,60). Thus, in vivo, but not in vitro, RFTOCs can "self-reorganize" into a functional thymic architecture with clear medulla-cortex boundaries.
This finding raised the interesting question of how the thymus can reestablish its key morphological pattern (i.e., the division into distinct epithelia characterized as medulla and cortex) once the original pattern is destroyed by enzymatic digestion. Given that in vitro reaggregation occurred from a single-cell suspension, initially yielding a random structure in vitro, medulla-cortex organization could take place via segregation and clustering of preexisting medullary epithelial cells ("sorting out"). Alternatively, growth of single progenitors or stem cells might contribute to formation of distinct thymic compartments, such as cortex and medulla. These possibilities have been raised and experimentally addressed by Rodewald and colleagues (31).
In principle, two types of experiments, summarized below, were performed to distinguish between these possibilities. In one set of experiments, reaggregates were assembled from mixtures of thymic epithelium isolated from two mouse strains differing in their MHC class II haplotypes (C57BL/ 6 [I-Ab]) and BALB/c [I-Ad]). Such mixed reaggregates were transplanted, and analysis of MHC class II expression could trace the origin of the epithelium to either of the two donors. In the other set of experiments, chimeric mice were generated by injection of embryonic stem (ES) cells into blastocysts using a combination of ES cells and blastocysts that, again, differed in MHC class II. The results from both types of experiments are summarized below.
3.3.1. Evidence for Epithelial Stem/Progenitor Cells in Thymus Organogenesis From Mixed Thymic Epithelial Cell Grafts
The principles of the generation, transplantation, and analysis of mixed reaggregates of thymic epithelium are shown in Fig. 2. Fetal thymus was enzymatically digested, and thymus epithelium was purified by cell sorting or by depletion of CD45+ (hematopoietic cells) from the single-cell suspension. At the stage of isolation (fetal d 15 or 16), medullary and cortical epi-
thelium were already separated into clearly defined areas, as schematically depicted in Fig. 2A,B. The donor mouse strains used were C57BL/6 (I-Ab) and BALB/c (I-Ad). Mixed reaggregates showed a random distribution of I-Ab+ (C57BL/6-derived) and I-Ad+ (BALB/c-derived) epithelium, as determined by staining with antibodies specific for I-Ab or I-Ad. Moreover, medulla-specific (MTS10) and cortex-specific (MTS44) antibodies (56) revealed that RFTOCs kept in vitro contained individual cells stained with either of these reagents, but no separable cortex or medulla pattern (schematically depicted in Fig. 2C).
Mixed reaggregates were transplanted under the kidney capsule of MHC class II-deficient C57BL/6 mice (66) (Fig. 2D). MHC class II-deficient recipients were chosen because class II molecules, in addition to thymic epithelium, are also expressed on bone marrow-derived dendritic cells, and their presence in the thymus might "complicate" an analysis of the epithelial composition and origin of the grafts. Specifically, in the absence of MHC class II expression on hematopoietic cells, all class II expression is exclusively confined to thymic epithelium of the grafted type.
In vivo, RFTOC grafts were colonized by host bone marrow-derived pro-T cells. Intrathymic T-cell development in such class II+ grafts included the generation of mature CD4+CD8- thymocytes. This population, owing to lack of class II expression in the host, failed to develop in the endogenous host thymus (66-68). Ex vivo histological analyses of tissue sections from RFTOC grafts showed that medullary areas reappeared. Double staining for the donor origin using antibodies specific for anti-I-Ab and anti-I-Ad demonstrated, surprisingly, that distinct medullary areas (islets) were derived from mutually exclusive donor epithelia; that is, these epithelial structures were of either I-Ab or I-Ad origin, but not mixed (depicted in Fig. 2E). Furthermore, immunofluorescence analyses using cytokeratin-specific antibodies proved that the lineage of these MHC class II+ areas was epithelial. These experiments led to the intriguing conclusions that (1) medullary areas can be reestablished in thymic epithelial grafts, and (2) these areas are derived from either of the two, but not both, donor origins. The latter observation strongly
Fig. 2. (continued) chimeric thymic epithelium. After culture, mixed reaggregated thymus grafts were implanted into MHC class II-deficient mice. Grafts developed in vivo into a fully functional ectopic thymus as shown by the generation of all thymocyte subsets and the MHC class II-dependent CD4+CD8- single-positive thy-mocytes. Remarkably, medulla-cortex reorganization occurred in vivo, and medullary zones segregated into islets of Balb/c or C57Bl/6, but not mixed, origin. This single origin of each islet indicates stem/progenitor cell activity in the thymus.
suggests that medullary epithelial structures can arise from single stem/progenitor cells by formation of medullary islets (31).
3.3.2. Evidence for Epithelial Stem/Progenitor Cells in Thymus Organogenesis From MHC-Chimeric Mice In Vivo
The data obtained from RFTOC transplantation experiments uncovered the remarkable property of thymic epithelium to reestablish proper architecture starting from an initially disrupted, randomly arranged composition formed in vitro. Another set of experiments has been reported in which aspects of thymus epithelial morphogenesis were studied more directly in vivo. The applied strategy was based on the fact that each cell in an organ of chimeric mice, generated by injection of ES cells into genetically distinct blastocysts, can originate from either the ES cell or the blastocyst. The quantitative contribution of cells derived from the ES cell or blastocyst to an individual tissue in an individual mouse can vary from animal to animal.
To identify mice in which the proportions of ES- and blastocyst-derived tissues were approximately comparable (balanced mice), the contribution of tissues derived from ES cells vs tissues derived from blastocysts was analyzed in an ectodermal (skin), a mesodermal (muscle), and an endodermal (liver) tissue. Mice were typed in these tissues using microsatellite markers specific for ES or blastocyst genomic deoxyribonucleic acid (DNA). Large numbers of chimeras were generated by injection of ES cells into MHC-mismatched blastocysts. The ES cells used were either from CBA (I-Ak) or BALB/c (I-Ad) mice, and the blastocysts were from C57BL/6 (I-Ab) mice. Using this approach, mice were identified in which epithelium derived from both ES cells and blastocysts contributed comparably to thymus formation (31).
Chimeric thymi were analyzed in detail by three-color histology using antibodies specific for the two donor epithelia (anti-I-Ak and anti-I-Ab) and specific for pan-thymic epithelial cytokeratin. Thymi were also examined by staining with the antibody MTS10, which recognizes medullary zones. Interestingly, these studies revealed that, in a physiologically developed thymus, medullary areas are composed of individual epithelial islets, each derived from either the ES cell or the blastocyst.
Histological measurements led to the calculation that these epithelial islets vary in diameter from a minimum of 60 x 40 to a maximum of 170 x 170 ^m. Each cell cluster harbors between 5 and 45 epithelial cells in a two-dimensional lattice. Serial sectioning of an entire mouse thymus lobe demonstrated that, in the mouse, one thymus lobe harbors about 300 medullary areas. Each area can include several islets. Therefore, it was estimated that one lobe contains approx 900 islets (31). The morphological "isletlike" char acter of medullary epithelium was noted in an earlier study (21). The islet character of medullary epithelium is most apparent in the juvenile thymus. In contrast, medullary epithelial islets tend to be larger in the adult thymus; here, the medulla can appear more confluent. Collectively, these experiments provided the first evidence that at least part of the thymic epithelium is composed of individual islets.
As summarized above, medulla-cortex compartmentalization has been thought for a long time to occur via invagination of an endodermal into an ectodermal epithelial sheet at the third pharyngeal pouch and cleft, respectively. Despite the fact that epithelial stem or progenitor cells have been invoked in thymus development, based on marker studies, no experimental evidence for such cells had been obtained. Data from chimeric mice, as well as data from RFTOC grafts, have provided the first evidence for an involvement of epithelial stem or progenitor cells in thymus morphogenesis. However, it should be noted that these experiments provide the first evidence for stem/progenitor cells in thymus organogenesis, but such cells were not physically purified by phenotype. Isolation of highly enriched thymic epithelial stem/progenitor cells will be required before their prospective developmental behavior can be studied.
Was this article helpful?