Pax1 is expressed during development of the thymus epithelium and is required for normal T-cell maturation

Development of the thymic epithelium and T cell development are intimately linked by multiple interactions between thymic stromal cell and developing thymocytes (van Ewijk, 1991; Ritter and Boyd, 1993; Boyd et al., 1993). For T cell maturation (Faas et al., 1993; Jenkinson et al., 1992; Hueber et al., 1992; Muller et al., 1993) as well as positive and negative selection of thymocytes (Jenkinson et al., 1992; Kaye and Ellenberger, 1992; Vukmanovic et al., 1992, 1994; Tanaka et al., 1993; Hugo et al., 1994) interactions between the immature thymocytes and different stromal cell lines are required. During this process the stroma provides adhesive sites by expression of specific adhesion molecules (Sawada et al., 1992; Lannes Vieira et al., 1993; Patel and Haynes, 1993), present (self-) antigenes via MHC molecules (Anderson et al., 1993; Markowitz et al., 1993) and secretes cytokines, growth and differentiation factors required for T cell maturation (Moll et al., 1992; Palacios and Nishikawa, 1992; Faas et al., 1993, Moore et al., 1993; Farr et al., 1993; Wolf and Cohen, 1992; Kasai et al., 1993; Takahama, 1994). The maturation pattern of thymocytes is precisely defined by characteristic precursor-offspring relationships and controlled by sequential interactions between the thymocytes and different stromal cell types (van Ewijk, 1991). The intrathymic migration route of thymocytes mirrors these interactions. Starting at the corticomedullary junction, prothymocytes (CD2^lo, thy-1^lo, CD4⁻8⁻ TCRαβ) reach the outer cortex through thymic parenchyma. Then, they turn back and, while moving through the cortex, express new surface proteins (CD4⁺8⁺TCRαβ^lo). After acquiring the capacity to recognize self-MHC (CD4⁺8⁺TCRαβ^hi, CD69⁺) they finally complete their developmental journey as mature T cells (CD4⁺ or CD8⁺ TCRαβ^hi) in the medulla.

In a reciprocal fashion, thymocytes also influence the development of the thymic stroma. However, much less is known about the origin and the maturation pattern of the different epithelial cell types during thymic ontogeny and in the adult thymus. The thymus stroma is composed of epithelial cells of both ectodermal and endodermal origin and of bone marrow-derived cells and has been extensively characterized both at the morphological level and for the expression of distinct cell surface markers (van Ewijk, 1991; Boyd et al., 1993). During embryogenesis, endoderm of the third pharyngeal pouch fuses with ectoderm of the third branchial cleft to form the early epithelial thymic primordium (van Ewijk, 1991; Boyd et al., 1993). The development of this primordium depends on interactions with the surrounding neural crest-derived mesenchyme (Bockman and Kirby, 1984). Between days 11 and 12 of gestation in the mouse, the thymus primordium is populated by lymphoid progenitor cells. As a result of interactions between these cells and the thymic epithelium, cortical and medullary regions can be distinguished from day 13 p.c. onwards (Boyd et al., 1993). Recent results have demonstrated that the differentiation of both cortical and medullary cell types depends on interactions with thymocytes of specific maturation stages (Holländer et al., 1995). Towards the end of gestation most of the stromal components of the thymus are thought to be fully developed (van Ewijk, 1991; Boyd et al., 1993).

Thymus development is affected at different stages in a variety of mouse mutants and human disorders. In the mouse mutant strain nude a mutation in the forkhead domain containing transcription factor whn results in a defective ectodermal contribution to the thymus primordium, leading to a cystic thymus rudiment which is unable to attract lymphocytes (Nehls et al., 1994; Cardier and Haumont, 1980). Human patients with DiGeorge syndrome completely lack a thymus or possess only a small defective rudiment, probably caused by a neural crest defect interfering with the normal interaction between the epithelial thymus primordium and its surrounding neural crestderived mesenchyme (Bockman and Kirby, 1984; DiGeorge, 1968). In scid mice and several transgenic strains with an early block in T cell development missing interactions between thymocytes and thymus epithelium result in impaired differentiation of thymus epithelial cell types (Boyd et al., 1993). So far no mutation which specifically affects the endodermderived cells of the thymus epithelium has been described.

In this report we describe the expression pattern of the transcriptional regulatory protein Pax1 in thymus epithelial cells of endodermal origin and show that mutations in Pax1 result in reduction of thymus size and altered T cell development. The murine Pax1 gene was originally identified by homology to the Drosophila paired and gooseberry genes (Deutsch et al., 1988). During mouse embryogenesis the Pax1 gene product plays an important role in the development of the vertebral column and the pectoral girdle as has been determined by analyzing an allelic series of mouse mutants with alterations in the Pax1 gene (Deutsch et al., 1988; Balling et al., 1988; Koseki et al., 1993; Wallin et al., 1994; Dietrich and Gruss 1995). Here we show, that the Pax1 gene product is expressed first by endodermal precursor cells of the thymus epithelium in the pharyngeal pouches. It is further maintained in thymic epithelial cells during ontogeny and later on in the adult thymus by a small subset of cortical epithelial cells. Analyzing the thymus and T cell development in Pax1 mutant mice we demonstrate that Pax1 is required for proper thymus development.

undulated (un) mice were purchased from the Jackson Laboratory. Undulated short-tail (Un^s) and undulated-extensive (un^ex) mutant mice were kindly provided from Dr A. M. Malashenko, Krosnogorsk, Russia and Dr J. L. Cruickshank, Leeds, England, respectively. These mutants were backcrossed onto the C57BL/6 strain. N1 to N5 generations were used for analysis. Embryos were recovered on days 9.5 to 18.5 p.c. where day 0.5 was 12 a.m. on the day of detection of the vaginal plug. On average, litters were born on day 19 p.c. As controls, +/+ or un/+ littermates were used.

Pax1 mutant embryos were genotyped by PCR analysis on embryonic DNA prepared from yolk sacs or the embryo proper as described by Wallin et al. (1994).

Total cellular RNA was isolated from adult mouse hematopoetic organs and day 12.5 p.c. embryos by the guanidine/caesium chloride method. RNase protection assays were performed as described by Koseki et al. (1993).

Pax1 riboprobes were generated from a HincII-SacI paired box fragment using either ³⁵S-UTP or digoxigenin-11-UTP labelling for in situ hybridization on sections or in whole embryos, respectively. Fixation, hybridization and subsequent detection procedures were essentially as described by Kessel and Gruss (1991); Rosen and Beddington (1993).

For immunostainings, embryos or isolated thymi were fixed in 3% acetic acid in absolute ethanol at 4°C over-night, incubated first in absolute ethanol and then in xylene, both twice for 30 minutes, followed by a 1:1 mixture of xylene:paraffin for 30 minutes at 55°C once, infiltrated with paraffin by three incubations for 1 hour at 55°C and embedded. The generation of the Pax1-specific antiserum has been described previously (Chalepakis et al., 1991).

Pax1 immunostaining was done on deparaffinized sections that were first bleached in 0.3% H₂O₂ in methanol for 30 minutes, washed in PBS, blocked with 10% normal goat serum in PBS for 60 minutes and then incubated with the antiserum diluted 1:200 in the blocking solution. Detection was made with peroxidase-conjugated goat-antirabbit IgG (Sigma), and DAB was used as chromogen.

For histological analysis, thymi were fixed in Bouin’s solution and embedded in paraffin. 7 μm sections were stained in haematoxylineosin.

Thymocyte suspensions were prepared by squeezing the cells through a fine metal grid. Aggregates were removed by filtration through a nylon mesh. Flow cytometry was performed by using the following biotinylated or FITC or PE-labelled mAbs: H57-597 (from Pharmingen) GK1.5 for CD4; 53-6.7 for CD8 (both from Becton-Dickinson, Palo Alto, CA); RM2-5 for CD2; H1.2F3 for CD69 (Pharmingen); 53.7.3 for thy-1.2 (Becton Dickinson); ER-TR4, ER-TR5 and anti-IA^b (AF6-120.1, Pharmingen, San Diego, CA) to analyze thymic stroma. Biotinylated mAbs were counterstained with a streptavidin-PE conjugate (Southern Biotechnology Ass., Birmingham, AL) for two-color fluorescence or with streptavidin-TRI-color (MEDAC, Hamburg, Germany). The analyses were performed using a Becton-Dickinson FACScan and LYSYS II software.

Expression of Pax1 RNA in the embryonic thymus has been described previously (Deutsch et al., 1988). Northern blot analysis of RNA isolated from adult thymus also demonstrated the presence of Pax1 transcripts in the adult organ (data not shown). In order to determine whether Pax1 expression could also be detected in other lymphatic organs, RNAse protection assays were performed. As shown in Fig. 1, only the thymus but not bone marrow, lymph nodes, spleen or Peyer’s patches contained detectable levels of Pax1 RNA.

To localize the site of Pax1 expression in the thymus more precisely and to study the expression pattern of the Pax1 gene during thymic ontogeny we performed a series of in situ hybridization and immunohistochemical analyses. By whole-mount in situ hybridization Pax1 expression could be detected in the endoderm of the foregut region as early as day 8.5 (data not shown). At day 9.5, transcripts were present in the first two pharyngeal pouches and by day 10.5 p.c., clear expression was also observed in the third pharyngeal pouch (Fig. 2).

Immunostaining demonstrated that Pax1 protein is present in the endoderm of the first, second and third pharyngeal pouches at day 10.5 (Fig. 3a). The third domain is known to fuse with ectodermal cells of the third branchial cleft to form the early thymic epithelial primordium which is then from day 11 onwards populated by lymphoid progenitor cells (van Ewijk, 1991; Boyd et al., 1993). Immigration of these cells did not change the Pax1 expression pattern since a large number of epithelial cells are Pax1⁺ in the thymus anlage of day 12.5 embryos (d12.5; Fig. 3b). At this stage the Pax1⁺ cells are evenly distributed throughout the thymus primordium. At day 14.5 the two major Pax1 expression domains are found in the vertebral column and the thymus (Fig. 3d). At this stage, a distinction between cortical and medullary regions can be made, indicating for the first time that the Pax1-expressing cells become confined to the developing thymic cortex (Fig. 3c). Pax1 expression by cortical epithelial cells continues to be maintained up to the adult thymus. In sections of thymi from 6 week old mice, expression could still be detected immuno-histochemically in a subpopulation of cortical stromal cells, whereas the medulla did not contain any Pax1 positive cells (Fig. 4). The Pax1⁺ cells were found to be dispersed through-out the cortex and also in subcapsular locations. The main difference in Pax1 expression pattern between the developing thymus of day 14 and the adult thymus resides in the relative number of Pax1⁺ cells: at day 14, a significant fraction of thymic epithelial cells show Pax1 expression. In adult mice however, Pax1 expression is seen only in a small number of cortical epithelial cells (Fig. 4) that neither express MHC II molecules nor the thymic epithelial cell surface protein stained by the mAb ER-TR4 (data not shown). These results suggest that Pax1 is expressed mainly by thymic epithelial precursor cells and not by their more differentiated offspring.

To investigate the contribution of the Pax1 gene product to thymic development we studied three different Pax1 mouse mutants, undulated (un), undulated-extensive (un^ex) and Undulated short-tail (Un^s). un and un^ex are recessive mutations and homozygotes are viable and fertile, whereas Un^s, with a complete deletion of the Pax1 gene, is a perinatal homozygous lethal. All undulated mutants have a smaller thymus which in adult un/un and un^ex/un^ex mice is about two thirds of a normal thymus. In the more strongly affected genotypes, Un^s/un and Un^s/Un^s, the reduction of the thymus was more apparent, and the organs did not reach more than half the normal size. Histological examination did not reveal any obvious abnormalities except the appearance of large cysts (Fig. 5) and hematoxylin-eosin stainings showed an apparently normal gross morphology with cortical and medullary regions (data not shown).

Thymic epithelial cells and developing thymocytes are in a close symbiotic relationship (van Ewijk, 1991; Boyd et al., 1993). Since the altered or lacking Pax1 protein resulted in a smaller thymus, we analyzed, by flow cytometry, whether these changes would also affect thymocyte maturation. Thymi of the +/+, Un^s/+, un^ex/+, un^ex/un^ex and Un^s/Un^s genotypes were isolated from embryos at different stages of development (day 17.5, 18.5, 19 and 21 p.c.) and from adult +/+ un/+ un^ex/+, un/un and un^ex/un^ex mice. Since Un^s/Un^s mice are not viable after birth and since adult un/un and un^ex/un^ex mice might be stressed because of their defects in vertebral column development, we focussed our analysis on thymic development during embryogenesis and in d19.5 Un^s/Un^s and wild-type mice. At day 17.5, thymocyte numbers were found to be reduced by a factor of 5 in the Un^s/Un^s homozygous mutants (Table 1) and by a factor of 2 in Un^s/+ embryos. A twofold reduction was also found in un/un and un^ex/un^ex embryos and mice (data not shown). Phenotypic analyses of the major thymocyte subsets of all undulated mutants at day 17.5 revealed a slight, however not significant, increase in the contribution of the CD4⁻8⁻ thymocyte compartment, which consists mainly of the most immature T cells of the whole thymocyte population (Table 1). The contribution of the more mature CD4⁺8⁺, CD4⁺ and CD8⁺ subsets was equivalently decreased (Table 1). In d19.5 Un^s/Un^s mice, the contribution of the CD4⁻8⁻ compartment was significantly increased to 33.5% as compared to 13.1% in wild-type mice. At the same time the contribution of the more mature CD4⁺8⁺ subset was reduced to 64.7% instead of 81.3% in control littermates (Table 1). Thymocytes are positively and negatively selected by interacting with thymic epthelial cells and we therefore analyzed whether the changed Pax1 expression also affects T cell education. Positive selection is phenotypically characterized by the expression of the early activation marker CD69 (Swat et al., 1993). Triple staining with anti-CD4, anti-CD8, and CD69 antibodies in d19.5 p.c. Un^s/Un^s embryos displayed a twofold reduction in the relative numbers of positively selected CD4⁺8⁺CD69⁺ subsets (Table 1, Fig. 6D,E) corresponding to a 2- to 8-fold reduction in absolute numbers (Table 1). Thus, in 19-day old Un^s/Un^s embryos the composition of the different thymocyte subsets was more similar to d17.5 p.c. wild-type embryos than to their 19- to 21-day old normal counterparts. This result shows that the generation of mature thymocytes is impaired in Pax1⁻ mice. This impairment could be due to lower numbers of thymocyte precursor cells which are characterized by low surface levels of thy-1. To our surprise, we could not identify these cells in the thymus of d19.5 Un^s/Un^s mice since all thymocytes expressed elevated levels of thy-1 (Fig. 6A,C). The upregulation of thy-1 expression was reproducibly found in Un^s/Un^s mutants where thymocytes expressed about 2-4× higher levels of this glycoprotein than control littermates (Fig. 6A).

Expression of Pax1 in epithelial cells of the thymus primordium up to the adult stage, and thymus defects in Pax1 mutant mice indicate an important role for Pax1 in thymus development and thymus function in the adult. Early on, Pax1⁺ cells are found in the pharyngeal pouch endoderm. After fusion with ectoderm from the third branchial cleft, Pax1-expressing cells are evenly distributed within the developing thymus epithelial primordium and later on become restricted to the cortical region. As in other parts of the embryo Pax1 is excluded from the ectoderm, its expression in the developing thymus is probably in cells of endodermal origin. This expression pattern therefore suggests an intimate mixing of ectodermal and endodermal cells within the early thymus primordium and also that endoderm contributes to the cortical region, whereas earlier histological studies of thymus development had proposed a purely ectodermal origin for the cortex (Cardier and Haumont, 1980). Likewise the observed phenotype of Pax1 mutant mice suggests a defect in all or part of the endoderm-derived cells within the thymus epithelium.

The Pax1 gene product functions as a transcriptional activator. The genes regulated by the Pax1 protein during thymus development and in adult cortical cells are so far not known. Neither is the specific role of the cortical subpopulation expressing Pax1 for thymus function understood. One might, however, speculate that Pax1 expression marks a set of precursor cells that can differentiate further into specific thymic stroma cell types. To our knowledge no antibody marker with an identical staining pattern has been described. The only expression pattern showing some similarity is the one of t, a SV40 T antigen controlled by the growth hormone releasing factor promoter (Moll et al., 1992a,b). In this case, the transgene-positive cells are – like the Pax1⁺ cells – found exclusively in the cortex, but also the transgene-positive cells are clustered at the cortico-medullary junction. In axial skeleton development Pax1 expression is required for early steps of sclerotome development and is downregulated when sclerotome cells differentiate into chondrocytes. By analogy, in cortical thymic epithelial cells Pax1 expression could also promote early steps of the differentiation of cortical thymic epithelial cell subsets. Due to a lack of Pax1 function mutant mice could suffer from a deficiency in the differentiation of one or more cortical stromal cell types.

This hypothesis was corroborated by the analysis of a series of Pax1 mutants in which a decrease in thymus size as well as a reduced number of thymocytes was found in mutant animals. Since thymic stromal cells foster T cell development (van Ewijk, 1991; Boyd et al., 1993), this phenotype might reflect a corresponding decrease in the number of thymic stromal cells that originate from Pax1⁺ precursors or changes in the expression of proteins that interact with T cells during thymocyte maturation. Each of these defects would cause a reduced capacity of the thymus to support T cell development in the cortical phase. Since we were unable to find another marker that would identify Pax1⁺ cells, this hypothesis remains to be proven. Whatever the defects in the thymic stroma of Pax1 mutants are, they lead to an accumulation of immature CD4⁻8⁻ cells and an increased expression of thy-1. As Un^s/Un^s homozygotes die at birth it remains unknown, whether the observed delay in T cell maturation would be corrected postnatally. Two functions are linked to the glycoprotein thy-1: thy-1 is involved in adhesion of thymocytes to thymic epithelial cells (Hueber et al., 1992; Irlin and Peled, 1992; Kroczek et al., 1986) and, thy-1 crosslinking induces, in combination with signalling through the T cell receptor, cell death of CD4⁺8⁺ thymocytes (Kroczek et al., 1986; Gunter et al., 1987). The latter function has been linked to negative selection of T cells (He et al., 1991). Thus, the increased expression of thy-1 in Pax1⁻ mice might make thymocytes more susceptible to cell death, which would result in lower numbers of more mature thymocyte subsets.

In conclusion our studies demonstrate that the expression of the Pax1 gene is not only important for sclerotome (Koseki et al., 1993,Wallin et al., 1994) but also for thymus development. A dual involvement in thymus and skeletal development, similar to that of Pax1, has also been established for another transcriptional activator, the c-fos gene product. In the c-fos case, however, the long bones and not the vertebral column are affected besides the thymus.

We thank Rüdiger Fritsch for the Pax1 antibody, Petra Fiedler and Bärbel Strack for excellent technical assistance, and Uli Birsner for oligonucleotide synthesis. J. Wallin was supported by an EMBO fellowship and by the Swedish Medical Research Council. H. K. was supported by the Human Frontier Science Program. The work was supported by the Max-Planck-Society and the Deutsche Forschungsgemeinschaft.