Thymus produces educated T cells that can react with peptide antigen loaded on a self major histocompatibility complex (MHC) but not with self antigens. Developing thymocytes are guided and selected in the microenvironment of thymic stromal cells. Among the stromal cells, thymic epithelial cells (TECs) are major components that plays important roles of thymocyte differentiation in the separate compartments, the cortex and the medulla.

TECs also play critical roles during thymic organogenesis as shown in Foxn1 mutant mice, in which early TEC differentiation is abrogated, functional thymus lacks and, therefore, no T cell is present (Nehls et al., 1996). In mice, initial thymic structure begins to form with the TEC precursor cells originated from the third pharyngeal pouch around fetal day 10.5 (Blackburn and Manley, 2004; Gill et al., 2003; Rodewald, 2008; Su et al., 2001). At this stage, fetal thymus does not show clear medullary compartmentalization yet although the cells of medullary thymic epithelial cells (mTECs) in nature are found (Roberts et al., 2012). Fetal thymus begins to express the cortex-specific markers such as CDR1 in addition to general epithelial markers, EpCAM and MHCII (Ahn et al., 2008; Boehm, 2008; Lee et al., 2012; Yang et al., 2005). Later, thymus undergoes atrophy by aging after puberty and/or by damaging insults such as radiation or stress hormones (Blackburn et al., 2002; Cheng et al., 2010; Gill et al., 2003). However, thymus can also be rejuvenated by removing steroid sex hormones or removing organs that produce sex hormones (Berzins et al., 2002; Lynch et al., 2009; Sutherland et al., 2005). The functional thymic epithelial stem cell (sTEC) in the adult or aged animals were identified (Blackburn et al., 2002; Rodewald et al., 2001; Swann and Boehm, 2007; Ucar et al., 2014; Wong et al., 2014). It is important to identify the molecular marker present in the sTEC to understand the mechanism of thymic regeneration and to translate into the clinic for the recovery of important cellular immunity.

There has been much evidence that cortical TEC (cTEC) and mTEC are derived from the single precursor TECs (pTEC) or sTEC (Bleul et al., 2006; Rossi et al., 2006). While pTEC can be bipotent or specific lineage- committed (Park et al., 2013; Ucar et al., 2014), TEC development may be more progressive without instant commitment to a specific lineage (Alves et al., 2014). The original specific antibodies used for the identification of sTECs are MTS24 and MTS20 (Bennett et al., 2002; Gill et al., 2002). These TEC stem cells were located in the small medullary islets of very young thymus or corticomedullary junction of the adult thymus (Rodewald et al., 2001). The cytokeratin K5 and K8 are also important molecules for the identification of sTECs (Klug et al., 1998; 2002). It was also proposed that TEC stem cells reside in the MTS10⁺ cells in the medullary area.

It has been considered that Foxn1 might be the master key transcription factor controlling TEC differentiation (Chen et al., 2009; Cheng et al., 2010; Corbeaux et al., 2010; Manley and Condie, 2010; Nowell et al., 2011; Ucar et al., 2014). Some other transcription factors such as Pax1/9 and Hoxa3, and signaling molecules such as Shh, Wnt, Bmp and Fgf were also identified from the studies using the mouse lines with gene ablation (Hollander et al., 2006; Manley and Condie, 2010). Those molecules function from the stage of third pharyngeal pouch to the initial state of thymus formation. Foxn1 appears to be important for the survival and proliferation of committed TECs at the stage of thymic organ maintenance. Wnt4 is responsible for the expression of Foxn1 (Balciunaite et al., 2002).

However, the presence of sTEC without Foxn1 expression was recently shown by Bruno Keywisky’s group by using a new feature for stem cells to be able to form spheres in 3D culture (Ucar et al., 2014). Methods which can isolate stem cells based on their functionality will provide thrust for the studies on the initiation of thymic organogenesis at the molecular level and on the detailed processes on how it behaves. Our understanding of sTEC is still in a primitive state. One of the distinguished common features of stem cells, either from the specific organs or even from cancer cells, is called “side population” (SP) in flow cytometry (Golebiewska et al., 2011; Zhou et al., 2001). The cells in the side population emit both blue and red fluorescence from the DNA staining dye, Hoechst33342. This phenomenon is mediated by the ABC transporters and can be blocked by the inhibitor (Golebiewska et al., 2011; Zhou et al., 2001). Therefore, it is very useful to identify stem cells when no well-characterized stem cell marker is available.

TSCOT (Slc46A2/Ly110) is a gene encoding cTEC-specific membrane protein (Ahn et al., 2008; Chen et al., 2000; Kim et al., 2000; Yang et al., 2005), isolated from the cDNA library of SCID thymus and of fetal thymic stroma (Kim et al., 1998; Park, 1997). Its expression peaks at the early stage of thymic development and reduces when the thymus is more mature (Ahn et al., 2008; Kim et al., 2000; Lee et al., 2012; Yang et al., 2005). When LacZ reporter is inserted in the TSCOT locus, β-galactosidase expression was found in the whole thymus of new born but only in the cortex and corticomedullary junction of adults (Ahn et al., 2008). When hooked to the promoter fragments (9.1 kb), evolutionarily conserved sequences located in the upstream of the coding sequence, reporter EGFP expression copied the expression pattern of endogenous gene while a shorter promoter fragment (3.1 Kb) revealed unexpected expression in the medulla at the adult stage of transgenic mice (Chen et al., 2000; Lee et al., 2012). The Cre recombinase under the control of TSCOT promoters resulted in expression of EGFP and β-galactosidase in the bipotent pTEC by the deletion of loxP sequences harbored in the ROSA locus of the transgenic mouse lines (Park et al., 2013). Therefore, the unique restricted pattern of the TSCOT expression is of high value to study TEC differentiation and thymic organogenesis.

Expression of TSCOT has also been noticed in the male epididymal duct in conventional Northern blotting and immunohistochemistry (Obermann et al., 2003), and the TSCOT locus has been assigned in a susceptibility of cervical carcinoma by human genetic analyses (Engelmark et al., 2006; 2008). In the current era of bioinformatics, there has been many systemic data accumulating in the public database and available for analysis.

In this study, we took advantage of public database and bioinformatics tools and performed genetic profiling in addition to classical methodologies. We show TSCOT is expressed prior to bipotent pTEC, at the side population stage of thymic epithelial cells, and also even in differentiating ES cells. Its expression does not depend of Foxn1. TSCOT expression and its roles in other epithelial tissues like skin and lung are discussed.

Expression profiling using public database

The data sets were obtained from Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/), and the extension changed as txt files to analyze in the GENESIS program (version 1.7.6) released by Graz University of Technology Institute for Genomics and Bioinformatics. Genes and the probes used are shown in Table 1. Multiple sample data are averaged before the final analysis. The normalized data of the genes were sorted by similarity to TSCOT genes or calculated by using Hierarchical Clustering to generate heatmaps. Some of the GEO data sets are drawn as graphs and calculated P values with a two-tailed-T test in GraphPad Prism (version 6.0c).

Mice

The mouse lines TDLacZ (Ahn et al., 2008), 3.1T-EGFP (Chen et al., 2000), and 9.1T-NE (Lee et al., 2012) were maintained in the Laboratory of Molecular and Cellular Immunology Animal Facility of Inha University, Korea. All animal studies are in compliance with the Use of Laboratory Animals under the proper protocols. The protocols were approved by the Committees on the Ethics of Animal Experiments of NIH (LCMI Protocol 8) and Inha University (Protocol LMCI-2). Fetal mice were obtained from timed mating. The presence of a vaginal plug was considered at E0.5.

For genotyping, tail samples were extracted and used for a polymerase chain reaction with primers for the TDLacZ locus: Neo primer (ACCGCTATCAGGACATAGCGTTGG), 1C12 F1 (TTACTCAAAGTGATGCTGGACTGG), 1C12 B2 (CCGAGGGTTCCTTGGTACATTC), and the EGFP locus: EGFP-F (GCCACAAGTTCAGCGTGTCC), EGFP-R (GCTTCTGTTGGGGTCTTTGC), using the red Extract-N-Amp Tissue PCR kit (Sigma).

Automatic cell counting

A fluorescence-based, automatic cell counter (Luna-FL, Logos Biosystems) was used to measure accurately the numbers of cells including thymic epithelial cells. The contamination from red blood cells could be automatically excluded because this system enumerates only nucleated cells.

Thymic stromal cell preparation

A single cell suspension was prepared as described (Lee et al., 2012). Briefly, thymic tissues or deoxyguanosine treated fetal thymic organ culture were treated with 0.25% trypsin (Invitrogen) for about 20 min, in the presence of DNase I (Sigma), and washed with phosphate buffered saline (PBS) containing 10% fetal bovine serum (FBS). For further purification of TEC, the single cell suspension was isolated using magnetic bead cell sorting after incubating with anti-Fc mAb 2.4G2 and anti-mouse CD45 microbeads (Milteny Biotec) for 20 min at 4°C.

Flow cytometry

Monoclonal antibodies used in the staining of cells include anti-MHCII (I-A^b), anti-CD45 (Ly-5), and anti-Sca-1. The antibodies were purchased from Caltag or from BD PharMingen. Anti-aminopeptidase A (CDR-1) and anti-EpCAM (G8.8) were prepared in the Custom Antibody Services Facility, NIAID, NIH. Biotinylated UEA-1 was purchased from Vector Laboratories.

Cells were washed in cold FACS buffer (PBS + 1% BSA), subsequently stained on ice with the primary and the secondary antibodies, then analyzed on FACSCalibur or FACSAriaII with two lasers in the presence of 1?2 μg/ml of propidium iodide (PI). Anti-Fc, 2.4G2 antibody was included in all flow cytometry staining to block Fc receptor. For side population analysis, a 1 × 10⁶ dissociated single cell suspension of fetal thymic organ culture (FD14.5) were incubated for an hour at 4°C in the presence of 5 μg/ml Hoechst 33342 dissolved in Hanks balanced salt solution. For verification of the side population, verapamil 0.25 mM was included. After washing at 4°C, cells were resuspended and examined by a flow cytometer equipped with a UV laser (FACSAriaII). For multicolor staining with SP analysis, cells were prestained with selected antibodies including homemade mAb CLVE (Yang et al., 2005). Negative control of TSCOT staining was carried out with all the same combination of antibodies except mAb CLVE. Analyses were done using the FlowJo program (http://flowjo.com).

RT-PCR

Sorted 1000 cells were used for RNA preparation. cDNA was generated with Superscript III and RT-PCR was carried out with the primers for TSCOT: F84 (5-CAGTCTTCCAATAACCTGCTTTGGCCT-3) and B83 (5-CGATTCCATGTGCCCCATTG-3) to amplify a 310 bp fragment and for GAPDH (Ahn et al., 2008;Kim et al., 2000). The primers for TSCOT are located in the separate exons with one intron and RT^? control sampled did not show any band in the gel.

Histostaining and microscopy

The immunofluorescence and X-gal staining the sections is described (Lee et al., 2012). An isolated thymus was washed in PBS and fixed in 1% para-formaldehyde, 0.2% glutaraldehyde, 0.02% NP-40, 1 mM MgCl₂ in PBS for 1 or 2 h on ice and was embedded in Tissue Freezing Medium (Triangle Biomedical Sciences, USA). The 4 μm sections were fixed for 2 min in 1% formaldehyde, 0.2% glutaraldehyde, 0.02% NP-40 1 mM NaCl, then incubated with X-gal solution (1 part X-gal 40 μg/ml in dimethyl formamide, in 40 parts 2 mM MgCl₂, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide in PBS) at 37°C for 48 h. For the detection of EGFP for fetal thymus sections, confocal microscopy was performed on the frozen sections in NIAID confocal facility (Leica SP2).

Gene profiling analysis verifies the tissue-specific TSCOT expression

In order to study the expression pattern of TSCOT at the genome level, we first used Google to identify any data and downloaded from the public database (http://www.ncbi.nlm.nih.gov/geo/) that shows the differential expression pattern (GSE7905). From the GEO database, the fetal tissue-specific expression was first examined (Fig. 1A). TSCOT expression was found only in the human fetal thymus, not in the fetal brain, fetal liver, nor fetal kidney. FOXN1 expression showed a similar but not an identical pattern. PAX9 and HOXA3 that are previously associated with the third pharyngeal pouch formation are even more different from the TSCOT pattern. In human adult tissues (GSE14938), TSCOT was found in the thymus and skin. In addition, it was present in lung and epididymis at lower levels (Fig. 1B). FOXN1 expression is also strongly expressed in the skin and thymus. However, it is not strongly expressed in any of the tissues that TSCOT is expressed at in the lower levels. Instead, FOXN1 is expressed in the liver, stomach and placenta. These suggest that TSCOT and FOXN1 may not be strongly associated in differentiated adult tissues.

Expression profiles of TSCOT and selected genes are investigated for the expression in the isolated mouse thymic epithelial cells (GSE56928). The genes for profiling (Table 2) are selected based on the literature which contains the information on the expression of the genes in the thymic epithelium or third pharyngeal pouch (references in Table 2). As shown in Fig. 1C, expression patterns are clustered as six different groups. A group contains TSCOT, Wnt4, Meis1, Gas1, Pax1, Isl1, Pax9, Six1, Pbx1, IL7, Bcl2, Bmp4, Dab2, FoxG1, Six4 and Hoxa3. These genes are expressed in both cTEC and mTEC^lo. Among them, TSCOT, Wnt4, Meis1, and Pax1 showed the strongest expression in the cTEC of the youngest mouse. Our earlier study on the expression kinetics (Kim et al., 2000) is consistent with these results. In this group, Pax1, Pax 9, Six1, Meis1 and Hoxa3 are the genes involved in the pouch stages (Manley et al., 2004). Wnt4 and Bmp4 were shown to be involved in the thymic organogenesis at the upstream of Foxn1 (Bleul and Boehm, 2005). It is interesting to note that Dab2 is a Wnt inhibitor. Gas1, Bcl2 and IL7 are the genes involved in the general cell cycle and survival. Group B contains Psmb11 (β5t) and Ly75 (NLDC202/DEC205) that are genuine cTEC-specific genes. Group C (Wint5a, Wnt5b, Eya1, Fgf8, Notch3 and Dkk3) includes genes that are expressed higher in the later stages of cTEC and mTEC^lo. Eya1 is known for roles in the third pharyngeal pouch (Gordon and Manley, 2011; Wei and Condie, 2011). However, its expression profile is somewhat different from the other genes involved in the same stage. Next, group D contains Foxn1 and Crip3 (TLP) that show the expression in cTEC and mTEC^hi. Here again, it clearly shows a deviation of expression pattern between TSCOT and Foxn1. Group E genes (Lif, Tert, and Fgf7) show the highest expression in the mTEC^lo or mTEC^hi. Given the known functions of Tert high expression in less divided cells, this result suggests that mTEC^lo may be found in more immature cells. The last group, group F contains HoxC13, Osm, IL6, CD248 (Endosialin), Tbx1, Aire, Shh, Sox2, and Grhl3. Those genes show the highest expression in the mTEC^hi population. HoxC13 regulates Foxn1 expression, and three genes, Osm, IL6, and CD248 are involved in thymic atrophy and involution. The roles of Tbx1 and Shh in mTEC^hi are not completely understood yet except that they are known for involvement in pouch formation.

The expression of TSCOT in mTEC^lo is not so surprising since it was found in the corticomedullary junction of young adult thymus where precursor or stem cells for thymic epithelium resides. When 4 week old thymic stromal cells were investigated by flow cytometry, the CD45^? population contains transitional cells with both cortical and medullary markers (CDR1⁺UEA-1⁺). Those cells are included in the TSCOT⁺ gated cells beside CDR1⁺UEA-1^? cTECs (Fig. 1D).

From these analyses, it was concluded that TSCOT is expressed in cTEC and undifferentiated and/or precursor mTEC. These expression profiles are common among the genes involved in early thymic organogenesis.

TSCOT is expressed in all TEC-committed stromal cells in fetal thymus

Next, we investigated the expression of TSCOT and reporters at the fetal stages in the different mouse models that we have previously characterized for the postnatal stages. The β-galactosidase reporter expression in TDLacZ thymus is restricted in the thymus as two dots at FD11 (Ahn et al., 2008). Figure 2A shows the β-galactosidase expression in the thymic sections at FD14.5. Expression of β-galactosidase is evenly distributed in the whole thymus, indicating most, if not all, the thymic epithelial cells at this stage express β-galactosidase. Another reporter mouse line, 3.1T-EGFP, which expresses EGFP in all TECs at the newborn stage (Park et al., 2013), showed EGFP expression earlier during fetal stages (Fig. 2B). At FD14 and 17, EGFP expression is also evenly distributed in the whole thymus. These results are consistent with the conclusion we previously described in which TSCOT is expressed in the pTEC stage (Park et al., 2013).

Fetal thymic stromal cells from normal C57BL/6 mouse (FD14) were analyzed for TSCOT expression with specific mAb CLVE (Yang et al., 2005). At this stage, EpCAM⁺ cells were all TSCOT⁺ (data not shown). When CD45^? stromal cells were displayed for CD31 as an endothelial lineage marker along with MHCII, it became clear that TSCOT expression is present in all MHCII⁺ cells and CD31^? MHCII^? cells (Fig. 2C). Only CD31⁺MHCII^? cells of endothelial lineage were TSCOT^?. From these results, it is concluded that endothelial cells either lost TSCOT expression due to lineage commitment from the common stem cell or originated from other type of precursor cells that do not express TSCOT.

TSCOT is expressed in the side population of TEC preparation

By using the TEC preparation from the deoxyguanosine treated FTOC of FD14.5, the presence of SP was tested with Hoechst 33342. In Fig. 3A, SP, which is ABC transporter sensitive, is clearly visible. When the inhibitor Verapamil was included, SP had decreased to 0.21% from 1.45%. Side population analyses were also applied with the TEC preparation using the same type culture of fetal thymus from 9.1T-NE mouse that shows EGFP expression patterns in the same way as endogenous TSCOT (Lee et al., 2012). As shown in Fig. 3B, a portion of the SP of 9.1T-NE TECs expresses EGFP when compared with that of normal C57BL/6 TEC preparation. In contrast, the side population of 3.1T-EGFP TEC population did not show any EGFP expression (data not shown).

When SP analysis was carried out along with antibody staining, TSCOT expression in SP and the major population (MP) were also clear. SP, either MHCII negative or positive, showed specific TSCOT expression with mAb CLVE. In addition, 84% of the SP cells and 95% of the MP cells were TSCOT⁺ (Fig. 3C). The next experiment was to verify TSCOT expression at the RNA level using a sorted side population of TECs prepared from normal C57BL/6. RT-PCR using the RNA prepared from the sorted cells clearly showed TSCOT expression in SP and in MP (Fig. 3D).

Many different types of stem cells express Sca-1 marker and a recent report mentioned that a TEC progenitor population is Sca-1⁺ (Golebiewska et al., 2011). We also tested expression of Sca-1 in fetal TEC preparation (Fig. 3E). Sca-1⁺ populations are present in both EpCAM⁺MHCII⁺ and EpCAM^?MHCII^? populations and most of them are TSCOT⁺. From these results, fetal TEC contains significant portion of sTECs that are TSCOT⁺.

TSCOT is expressed in differentiating embryonic stem cells

We searched the available data on TSCOT expression in the embryonic stem cell (ES) population. Two GEO sets of data (GSE14503 and GSE9440) contain an expression profile of T3 ES cell culture, embryonic body formation, and differentiating T3 ES cell into pancreatic islet-like cell clusters or fibroblasts (Fig. 4A). Expression of TSCOT is found in embryonic bodies but not in the undifferentiated ES nor in differentiated pancreatic islets and fibroblasts. TSCOT expression is clustered with FGF8, FOXN1, IL22 and FGF7. In addition, WNT5A, SHH, ISL1 and GRHL3 (Get1) are also in the neighboring group with the expression pattern that is only transiently expressed. Grhl3 is particularly interesting since knock out (KO) mouse skin show reduced TSCOT expression (see later).

The expression profile of TSCOT in the data from GSE3749 shows that the J1 ES cell line transiently expresses TSCOT when it is differentiated by removing leukemia inhibitory factor (LIF) (Fig. 4B).

When the data from various side populations were specifically searched for, SP of mouse mammary gland cells showed an increased TSCOT expression in the SP at a lower significance (P = 0.0732) (Fig. 4C). In some other SPs of mammary epithelium, studies did not reveal any significant difference between SP and non SP (data not shown).

From the results above, we concluded that TSCOT expression is initiated in differentiating ES cells and remained in some tissue committed SP stem cells.

TSCOT expression is independent of FOXN1 but depend on IRF6 and GRHL3

Because FOXN1 was considered as a putative TEC key transcription factor, we searched for TSCOT expression in the remaining thymic rudiment of nude mouse. In RNA prepared from several tissues, TSCOT expression was found in the thymic rudiment of nude mice (Fig. 5A). RNA samples that were not treated with reverse transcriptase did not generate any bands (data not shown). This result is consistent with the conclusion derived from the gene profiling analyses.

In order to find the putative regulatory factors, differential TSCOT expressions were also examined in the skins of mouse lines with various mutations (Figs. 5B and 5C). In IRF6 KO mouse (GSE5800), TSCOT expression had reduced along with, HoxC13, Fgf7, Tbx1, and Hoxa3 while Notch3, Foxn1, Grhl3, and Wnt4 had increased. The opposite expression profiles of TSCOT and Foxn1 in IRF6 KO skin suggest that these genes are independently regulated (Fig. 5B). It is interesting to note that the binding sites of IRF6 are located in the regulatory regions of Grhl3 (Botti et al., 2011; de la Garza et al., 2012). More interestingly, TSCOT expression is also reduced in the skin of GRHL3 KO mice (GSE7381) (Fig. 5C). To investigate the actual involvement of those transcription factors will require more investigation. The effects of various cytokines for the TSCOT expression in the human epidermal keratinocytes (GSE7216) can also be visualized in Figure 5D. TSCOT expression is down regulated by IL1b, IL22 and IL24, but not by KGF, IFNγ, IL19, IL20 and IL26d (Fig. 5D) in the keratinocytes.

These results suggest a regulatory mechanism for TSCOT expression by the transcription factors, such as IRF6 and GRHL3, and by cytokines, but not by FOXN1.

Is TSCOT a tumor suppressor?

We also researched TSCOT expression in the lung development. Human fetal lungs at various stages between 54?154 days (GSE14334) are clustered in Fig. 6A. The genes that are expressed in a similar way to those of TSCOT are GRHL3, FOXN1, PAX9, and CRIP3. Those genes are known to be upregulated during the fetal lung developmental process (Kho et al., 2010). Therefore, it is likely that those transcription factors are involved in the positive regulator of TSCOT in the lung. The general patterns of IRF6, SHH, ISL1, and SOX2 are downregulated during lung development, the opposite pattern to that of TSCOT. This suggests that those genes are potentially involved in the negative regulation of TSCOT in lung development.

In Fig. 6B, the cluster analysis of 20 genes coexpressed in the same fashion as TSCOT is shown in Table 3. To our surprise, TSCOT expression is clearly missing in three types of lung cancers, suggesting TSCOT may function as a tumor suppressor for lung cancer. The top genes clustered for the similar expression profiles are listed in Table III. As shown, many of the genes show possible tumor suppressor phenotypes (references in Table 3).

Using bioinformatics approaches and conventional molecular and cellular methods, we showed that TSCOT expression is turned on in TEC at the stem cell stage, and even prior to commitment of TEC lineages. In addition, we identified putative regulatory transcription factors and cytokines during thymus, skin and lung development.

TSCOT expression is turned on early thymic organogenesis and some other epithelial tissues

During last several years, gene expression profiling at the genome scale are accumulated in the databases and accessible to the public. We took advantage of this advancement for the study of thymic organogenesis and TEC lineage differentiation using TSCOT as a lineage cell type specific marker and other known genes. From this approach, we learned that our previous studies are verified and we can get a lot more information than conventional experiments that are difficult to perform due to the limitation of small numbers of cells present in the actual organ.

In our previous studies, we asserted that TSCOT is TEC lineage specific and expressed in the cTEC and bipotent pTEC (Ahn et al., 2008; Kim et al., 2000; Park et al., 2013). Tissue specificity in thymus and expression in the limited TEC lineages are verified by the genome scale data analysis of expression profiling (Figs. 1, 4, and 5). Furthermore, we learned that more tissues such as skin and lung also express TSCOT. The transcription factors involved in the thymic organogenesis may be also involved in skin and lung development (Figs. 1A?1C, 5B, 5C, and 6A). In addition, kinetic profiling of expression of TSCOT during cTEC lineage development has also been verified (Fig. 1C). TSCOT expression is highest in the youngest cTEC as we described earlier (Kim et al., 2000). TSCOT expression in mTEC^lo provides an interpretation that transitional cells found in the postnatal TEC preparation with UEA-1⁺CDR1^lo cells (Fig. 1D) are most likely the same kind. Earlier findings of sTECs in the medullary islet (Rodewald et al., 2001) and in the cortical medullary junction (Ahn et al., 2008) are consistent with the idea that TEC stem cells may overlap or share early mTEC features.

To further investigate cells at earlier stages than pTEC expression, we first utilized a functional SP analysis and showed that SP of fetal TEC preparation expresses TSCOT (Fig. 3). We like to call those cells sTEC for SP and stem TECs. In the SP of other type of tissues, such as mouse mammary glands, SP showed a slightly higher TSCOT expression than non SP (Fig. 4C). Our search for TSCOT expression in ES cells produced interesting results that TSCOT is induced in the differentiating embryonic bodies or ES cell cultures without LIF. TSCOT expression is off when the cells are differentiated into pancreatic epithelium or fibroblasts (Fig. 4A), and in the endothelial lineage (Fig. 2C). These results support the idea that TSCOT is expressed at the stem cell stage during certain organogenesis.

Given the concept that TSCOT is expressed in sTEC, its expressions in the skin and lung are not very surprising. They all express epithelial markers such as EpCAM and Keratins. There are cases that these cell lineages are actually interconvertible under certain circumstances. The stem cell preparation from TEC can be differentiated into skin type keratinocyte (Bonfanti et al., 2010) and thymic epithelium of nude mouse has shown to have lung epithelial morphologies (Dooley et al., 2005). This phenomenon can be interpreted as that of a reprograming of gene expression, transforming stem cells which are committed to one organ type, into another at the level of master gene expression.

A schematic model in Figure 7 summarizes the findings of the expression profile during organogenesis. TSCOT expression turned on in early uncommitted ES cells may be maintained in thymus, skin, and lung. In other organs, TSCOT expression is now turned off. The stem cell for the thymic organogenesis and the precursor TEC (pTEC) maintains TSCOT expression. The cTEC and mTEC^lo cells are cells that express TSCOT, and mature mTEC^hi cells lose TSCOT expression. In lung, TSCOT expression is turned on but tumorigenesis will turn off TSCOT expression.

Modulation of TSCOT gene expression was revealed by gene expression profiling

By cluster analyses with selected transcription and soluble factors that are shown to be involved during the thymic organogenesis, we were able to identify multiple putative positive or negative regulators. IRF6 and GRHL3 are putative positive regulators for TSCOT expression in the skin (Figs. 5B and 5C). It has been reported that GRHL3 is located downstream of IRF6 in human keratinocytes (Botti et al., 2011; de la Garza et al., 2012; Malik et al., 2010). TSCOT expression potentially is regulated by IL1b, IL22, and IL24 in negative fashion (Fig. 5D). These results suggest that TSCOT is controlled by IRF6 and GRHL3, and IL1b, IL22, and IL24 in human keratinocyte. In contrast, in human fetal lung development, both TSCOT and GRHL3 expressions are upregulated together while IRF6 expression is downregulated (Fig. 6A). These phenomena suggest the complex network of expression regulation and/or cross checking regulation of genes during different epithelial tissue development.

TSCOT may be a new member of the tumor suppressors

Besides the structural features of a transporter that appeared containing primary amino acid sequences (Kim et al., 2000), it is still unclear what biological and biochemical functions that TSCOT plays. TSCOT is a member of Slc46A, and another member, Slc46A1, has been characterized as a proton coupled folate transporter (Diop-Bove et al., 2013). The heavily hydrophobic nature of the TSCOT amino acid composition and a simple twelve membrane spanning feature, with the presence of a central inner loop in the absence of ATP binding domain, suggests that TSCOT may transport small hydrophobic molecules. We proposed earlier that it may function in the survival of TECs based on the expression (Kim et al., 2000).

It is exciting to find that TSCOT expression in the lung disappear in three types of lung cancers, large lung cell carcinoma, lung adenocarcinoma, and squamous lung carcinoma (Fig. 6B). This expression strongly implies TSCOT may function as a type of tumor suppressor. This supports the fact that TSCOT also function in the same way for the genetic type of cervical cancer susceptibility proposed (Engelmark et al., 2006; 2008). In fact, the Human TSCOT locus (9q32) was mapped to the susceptibility of cervical cancer through a SNP polymorphism study (Engel mark et al., 2006; 2008). It may function as a necessary component to maintain normal epithelium. When it is missing in lung epithelium, carcinogenesis progresses without hindrance. Other genes expressed in a similar fashion (Fig. 6) also show the functionality in tumor suppressors as described in Table III.