Significant Quantitative and Qualitative Transition in Pituitary Stem / Progenitor Cells Occurs during the Postnatal Development of the Rat Anterior Pituitary

We reported recently that a pituitary-specific transcription factor PROP1 is present in SOX2-positive cells and disappears at the early stage of the transition from progenitor cell to committed cell during the embryonic development of the rat pituitary. In the present study, we examined the localisation and identification of SOX2-positive and PROP1/SOX2-positive cells in the neonatal and postnatal rat pituitaries by immunohistochemistry. Quantitative analysis of immunoreactive cells demonstrated that SOX2-positive pituitary stem/progenitor cells are not only predominantly localised in the marginal cell layer, but also are scattered in the parenchyma of the adult anterior lobe. In the marginal cell layer, the number of PROP1/SOX2-positive cells significantly decreased after postnatal day 15, indicating that a significant quantitative transition is triggered in the marginal cell layer during the first postnatal growth wave of the anterior pituitary. By contrast, other phenotypes of SOX2-positive stem/progenitor cells that express S100β appeared in the postnatal anterior pituitary. These data suggested that quantitative and qualitative transition occurs by acquisition of a novel mechanism in terminal differentiation in the postnatal development of the anterior pituitary.

GATA2, PROP1 (prophet of PIT1), PIT1 and SF1, have been discovered and characterised (1), although their precise functions in pituitary organogenesis are not yet fully understood.
In addition, non-endocrine cells have been found in the pituitary. One of the non-endocrine cells identified so far comprises a type of folliculo-stellate cell, which was first distinguished by electron microscopy (2) and named in light of its characteristic shape and follicle formation capacity (3). Although it was defined later by the presence of S100 protein, its functions in the pituitary and embryonic origin remain obscure (4,5). It is notable that folliculo-stellate cells lining the marginal cell layer facing the residual lumen of the Rathke's pouch have long been hypothesised as pituitary stem ⁄ progenitor cells (6)(7)(8)(9). Over the last decade, the marginal cell layer has often been proposed to be a pituitary stem ⁄ progenitor cell niche providing a cell remodelling system to compensate for pituitary cellular plasticity (10,11). Recent studies have demonstrated that cells expressing the gene for the HMG-box transcription factor SOX2, a marker of stem ⁄ progenitor cells, are present in the adult pituitary and have an ability to generate all of the pituitary cell types (12,13). Fauquier et al. (12) demonstrated that a small population of progenitor cells, which are present in the adult pituitary gland and express Sox2, have an ability to form pituispheres in culture and differentiate into all of the pituitary hormone-producing cell types. Chen et al. (13) observed that SOX2 positive cells are more abundant in the pituitary of early-postnatal mice at the age of the first pituitary growth wave (1-week-old) than in adult animals. Thus, SOX2 might have a key role in maintenance of stem ⁄ progenitor cells and ⁄ or differentiation of pituitary cell lineage. More recently, we presented immunohistochemical observations that a pituitary-specific factor PROP1 consistently coexists with SOX2 throughout the embryonic development of the pituitary (14).
Prop1 encodes a paired-like homeodomain transcription factor, and is a heritable responsive gene for the combined pituitary hormone deficiency in the Ames dwarf mice (Prop1-deficient dwarf mice) (15) and human patients showing absence or low levels of GH, PRL, TSH, LH and FSH (16)(17)(18), as well as ACTH (19). Longitudinal studies in Prop1-deficient dwarf mice from early embryogenesis through adulthood have demonstrated that PROP1 plays roles in morphogenesis, apoptosis and proliferation in the developing pituitary (20), as well as in cell migration in the expanding anterior lobe (21). We further observed that Prop1-expression is transient in PIT1-positive cells but absent in endocrine cells during embryonic development, suggesting that PROP1 exists in SOX2-positive stem ⁄ progenitor cells and disappears at the early stage of the transition of progenitor cells to committed cells in the embryonic development of the pituitary (14).
In the present study, we aimed to track quantitatively the cells expressing Prop1 and Sox2 in the postnatal development of the rat anterior pituitary by the immunohistochemical technique. Finally, we demonstrated that PROP1 is absent in any endocrine cells but consistently coexists with SOX2 in non-endocrine cells, most of which are S100-positive. Analysis of PROP1, SOX2 and S100b-positive cells in the anterior pituitary of S100b-green fluorescent protein (GFP) transgenic rat (22) demonstrated that significant quantitative and qualitative transition in Sox2-expressing stem ⁄ progenitor cells lining the marginal cell layer occurred in the postnatal development of the anterior pituitary.

Animals
Intact male Wistar-Imamichi strain rats, kept in a controlled environment, were used. S100b-GFP transgenic rat was generated by fusing the S100bpromoter to the reporter gene GFP (22). The present study was approved by the committee on animal experiments of the School of Agriculture, Meiji University.

Generation of antibody
Guinea pig anti-rat PROP1 antiserum was generated as described previously (14). Briefly, the cDNA of rat Prop1 corresponding to the C-terminal region (amino acid residues 126-223) (Fig. 1A) was cloned into pET32a vector (Novagen, Darmstadt, Germany) to generate the TrxA-His-tag fused protein.
After the fusion protein was separated by 12.5% sodium dodecyl sulphatepolyacrylamide gel electrophoresis (SDS-PAGE) and stained with Coomassie Brilliant Blue, the protein band corresponded to fusion protein of the C-terminal region of PROP1 was cut and applied on immunisation in guinea pig.
Verification of the generated anti-PROP1 antibody was accomplished by western blotting first (Fig. 1B) and then by immunohistochemistry (methods described below) and in situ hybridisation using mirror sections (8 lm thickness) of male rat pituitary (P5) (Fig. 1C). Western blotting was performed using cell lysate of Chinese hamster ovary cells transfected with expression vector pcDNA3.1 (Invitrogen, Carlsbad, CA, USA) fused with the full length cDNA of rat Prop1. After proteins were separated on a 10% SDS-PAGE, electro-blotting to a polyvinylidene fluoride membrane was carried out. The membrane was immunoreacted with anti-PROP1 antibody (dilution 1 : 5000) and then secondary antibody of horseradish peroxidase-conjugated donkey immunoglobulin (Ig)G against guinea pig IgG (dilution 1 : 25 000; Jackson . Recombinant rat C-terminal region (closed box; amino acid residues 126-223) was used to generate the antibody after purified by sodium dodecyl sulphate-polyacrylamide gel electrophoresis. HD indicates a homeodomain important for DNA binding (amino acid residues 66-125). (B). Western blotting was performed using cell lysate of Chinese hamster ovary cells which overexpressed rat PROP1 cDNA (+) and empty vector ()). (C). Immunohistochemistry with anti-PROP1 antibody (left) and in situ hybridisation of Prop1 mRNA (right) were performed using mirror sections of rat pituitary (P5). Merged image of immunohistochemistry and light microscopy is shown in a center panel. Cells retaining both protein and mRNA signals of Prop1 are indicated by asterisks (*). Scale bar = 10 lm.
ImmunoResearch, West Grove, PA, USA). Immunoreacted signals were visualised by TMB stabilised substrate for horseradish peroxides (Promega, Madison, WI, USA). In situ hybridisation was performed as described previously (23). Briefly, frozen sections were treated with protease K (1 lg ⁄ ml; 10 min at room temperature), fixed in paraformaldehyde (4% for 20 min at 4°C) and washed in phosphate buffer (pH 7.0). Digoxigenin (DIG)-labelled RNA probes of each of both strands for rat Prop-1 cDNA were synthesised using DIG RNA labeling Mix (Roche Diagnostics GmbH, Mannheim, Germany) and the AmpliScribe T3 High Yield Transcription Kit (Epicentre, Madison, WI, USA). Hybridisation and colour visualisation were performed according to the manufacturer's manual. Signals of in situ hybridisation were present in the cytosol of the cells whose nuclei were stained by immunohistochemistry (Fig. 1C).

Statistical analysis
Data from the real-time PCR were analysed using Student's t-test.

Prop1 is expressed in the postnatal anterior lobe
We previously reported that immunoreactive signal of PROP1 occupies all of the cells in the primordial pituitary Rathke's pouch at E13.5, then localises in the cells migrating to the expanding anterior lobe at E16.5 (14). In the present study, localisation of Prop1expressing cells was investigated in the rat pituitary from E18.5 to the postnatal period. At E18.5, PROP1-signals were predominantly localised in the marginal cell layer facing the anterior side of the lumen but not in the intermediate side and were scattered in the parenchyma of the anterior lobe ( Fig. 2). One day after at E19.5, in addition to lining the marginal cell layer, the stratified pattern of PROP1-signals in the parenchyma of the anterior lobe was observed and became more apparent at P0, followed by decreasing the number of PROP1-positive cells in the marginal cell layer at P5. In the sexually mature pituitary at P60, PROP1-signals were rarely present in the marginal cell layer, although a small number of signals were still observed in the parenchyma of the anterior lobe. This distribution pattern of PROP1-positive cells was mostly maintained at P600 (Fig. 2).

PROP1-positive cells consistently express Sox2
We previously observed that PROP1-positive cells definitely express Sox2 in the embryonic pituitaries (14). Double-immunostaining for PROP1 and SOX2 was performed for the postnatal pituitaries of P0, P5, P60 and P600 (Fig. 3). All of the PROP1-positive cells lining the marginal cell layer and stratifying the parenchyma of the anterior lobe at P0 and P5 were positive for SOX2 but not vice versa. At P60 and P600, it is notable that SOX2-positive cells were the major population in the marginal cell layer, where PROP1 signals had almost disappeared, and were scattered in the parenchyma of the anterior lobe where the number of PROP1-positive cells had gradually decreased.
Significant transition in localisation pattern of PROP1 ⁄ SOX2-positive cells in the marginal cell layer of the postnatal pituitary As observed above, the localisation pattern of PROP1-positive cells showed a significant transition. Hence, we performed longitudinal evaluation of the proportion of PROP1-, SOX2-and PROP1 ⁄ SOX2positive cells present in the marginal cell layer and the parenchyma of the anterior lobe from P5 to P600 (Table 1). PROP1-positive cells Transition of stem ⁄ progenitor cell in postnatal pituitary 935 accounted for 68.6% of the cells in the marginal cell layer and 20.8% of the cells in the parenchyma at P5. The values markedly decreased to 5.2% and 7.4%, respectively, at P60, followed by a further decrease to 1.2% and 3.9%, respectively, at P600. Interestingly, the SOX2-positive cells in the marginal cell layer showed a high proportion of 78-90% until P60 and still accounted for 43.8% at P600. On the other hand, the SOX2-positive cells in the parenchyma gradually decreased from 27.3% at P5 to 11.3% at P600.
The proportion of PROP1 ⁄ SOX2-positive cells in SOX2-positive cells was also determined in the marginal cell layer and the parenchyma (Table 1). Marked decrease in the marginal cell layer was observed from 87.9% at P5 to 5.8% at P60, whereas a modest decrease from 76.3% at P5 to 49.6% at P60 was observed in the parenchyma.

Prop1 expressed in non-endocrine cells in the adult anterior lobe
We applied double immunohistochemistry using anti-PROP1 and anti-hormone antibodies for the adult rat pituitaries to confirm the previous observation that PROP1 is absent in endocrine cells during the embryonic period (14). As shown in Fig. 4A, none of the endocrine cells coexisted with the PROP1 signal, which is the same as the observations made in the embryonic pituitary, indicating that PROP1-positive cells are non-endocrine cells. Because folliculo-stellate cell is known as a major non-endocrine cell type, immunohistochemistry of S100 protein, which is one of the molecular markers of folliculo-stellate cells, was investigated in the present study. The results obtained demonstrated that Prop1 was certainly expressed in S100-positive cells (Fig. 4B, arrow). In addition, we found that some cells were positive only for S100 or PROP1 (Fig. 4B, open and closed arrowheads, respectively). For additional confirmation, the immunohistochemistry of PROP1 was examined using the S100b-GFP transgenic rat (22) because S100b-positive cells of the transgenic rat are easily distinguishable by the S100b-promoter-derived expression of the reporter gene, GFP (Fig. 4C). Immunohistochemistry showed the same image as that of a normal rat using anti-S100 antibody (Fig. 4B). Thus, positive cells only for S100b or PROP1 (Fig. 4B, open and closed arrowheads, respectively) were also present.

Classification of non-endocrine cells in the adult anterior lobe
Classification of non-endocrine cells was examined by localisation of PROP1, SOX2 and S100b using the adult pituitary of S100b-GFP transgenic rats (22). Consequently, we observed five cell populations in the 60-day-old rat pituitary (Fig. 5A): Type1; SOX2-positive, Type2; SOX2 ⁄ PROP1-positive, Type3; SOX2 ⁄ PROP1 ⁄ S100b-positive, Type4; SOX2 ⁄ S100b-positive, and Type5; S100b-positive. The number of cells belonging to each cell type in the marginal cell layer and the parenchyma of the anterior lobe of P60 was counted for three and four areas, respectively, and the proportion is shown in Fig. 5. In the marginal cell layer, Type2 (SOX2 ⁄ PROP1positive) was only 1.4% and major cell types were Type1 (SOX2positive, 31.1%) and Type4 (SOX2 ⁄ S100b-positive, 45.9%) (Fig. 5A). On the other hand, in the parenchyma of the anterior lobe, Type3 (SOX2 ⁄ PROP1 ⁄ S100b-positive) was a major population, accounting for 59.7%, whereas the proportions of other cell types ranged from 6.6 to 12.8% (Fig. 5B). It is noteworthy that cells positive only for PROP1 were not present and all of the PROP1-positive cells expressed Sox2, together or not with S100b, as far as was examined. Classification of non-endocrine cells in the adult anterior lobe demonstrated that the distribution pattern of the SOX2-positive stem ⁄ progenitor cell population differs between the marginal cell layer and the parenchyma of the anterior lobe. However, we could not characterise a temporal transition in the distribution pattern because S100b-positive cells were first detected in the postnatal pituitary at P5, when analysed by the immunohistochemical technique (data not shown). Next, we performed quantitative real-time PCR to estimate the mRNA level of the S100b in the pituitary of the prenatal to postnatal period.

Expression profile of Prop1 and S100b in pituitary development
After the late stage of embryonic development, the expression of Prop1 was specific to the anterior lobe (Fig. 2) and it is known that the expression level of S100b is higher in the posterior lobe than that in the anterior lobe (22). For this reason, analysis of the anterior lobe was essential, although it was practically difficult to isolate the anterior lobe from the embryonic pituitary. Therefore, quantitative real-time PCR was performed using total RNAs prepared from the whole pituitaries of prenatal and neonatal periods (Fig. 6A) and from the anterior lobes of the postnatal period (Fig. 6B). The results obtained revealed that Prop1 was expressed from the prenatal through the postnatal period and its level was gradually decreased. On the other hand, S100b started to be expressed at around E21.5 and its expression level increased linearly after P15 in the anterior lobe. Because the expression level of S100b was five-fold higher in the posterior-intermediate lobe than in the anterior lobe at P5 (data not shown), the expression level in the anterior lobe at E21.5 and P0 might be low, resulting in a low S100b protein level and failure of detection by the immunoreaction. These data suggest that S100b-positive cells in the anterior lobe first appear at around P0.

Discussion
In the present study, we aimed to conduct a quantitative and qualitative analysis of pituitary stem ⁄ progenitor cells in the postnatal pituitary development by tracking down the cells expressing a stem ⁄ progenitor marker Sox2 and a pituitary-specific transcription factor Prop1 using immunochemical techniques. Consequently, we demonstrated that a significant transition in the proportion of PROP1 ⁄ SOX2-positive cells occurred in the marginal cell layer of the rat postnatal pituitary from the neonatal to the adult period, whereas only a gradual quantitative change in the parenchyma of the anterior lobe was observed. Thus, we suggested that a qualitative transition is triggered in the marginal cell layer within the second postnatal week. Our previous study (14) revealed that the expression of On the other hand, the expression of Prop1 started at E11.5 and eventually accounted for almost all of the cells composing the Rathke's pouch at E13.5. When pituitary organogenesis started to progress, unlike SOX2, PROP1-positive cells existed mainly in the marginal cell layer of the anterior side and in the parenchyma of the expanding anterior lobe; those in the intermediate lobe disappeared completely by the neonatal period (14). After birth, the proportion of PROP1-positive cells on the anterior side of the marginal cell layer decreased gradually from 69% at P5 to 27% at P15, and thereafter rapidly decreased to 5% until P60, whereas the proportion of SOX2-positive cells was maintained at 90% during the same period. Instead of the significant difference between SOX2 and  PROP1 in the marginal cell layer, in the parenchyma of the anterior lobe, the proportion of both SOX2-and PROP1-positive cells decreased gradually from 27.3% at P5 to 11.3% at P600 and from 20.8% at P5 to 3.9% at P600, respectively. Characteristically, the PROP1-positive cells consistently expressed SOX2 but not vice versa as far as examined from E11.5 to P600. These data obtained by double-immunohistochemical analysis using anti-PROP1 and anti-SOX2 antibodies confirmed two findings. First, pituitary stem ⁄ progenitor cells not only exist mostly in the marginal cell layer, but also in the parenchyma of the anterior lobe. Second, a qualitative transition might be triggered in the marginal cell layer on the anterior side during the first postnatal growth wave of the anterior pituitary, whereas only a gradual quantitative change is observed in the parenchyma of the anterior lobe (Fig. 7). The marginal cell layer has been proposed to contain a pituitary stem ⁄ progenitor cell niche in the postnatal pituitary (10)(11)(12)(13)(26)(27)(28)(29)(30). A tissue stem ⁄ progenitor cell is defined by its potency to generate differentiated progenies and its long-term self-renewal capacity. It was shown that Sox2-expressing cells are multipotent stem ⁄ progenitor cells to generate all of the major cell types in the adult mouse pituitary (12). Analysis of the side population, which is rich in cells with stem ⁄ progenitor-like features, revealed that non-Sca1 high fraction in the side population of the adult mouse pituitary clusters Sox2-expressing cells and possesses an ability to form pituispheres to be differentiated into all of the pituitary endocrine cell types (13). These Sox2-expressing stem ⁄ progenitor cells are localised in the marginal cell layer but are also scattered in the parenchyma of the anterior pituitary (12,13). , adrenocorticotrophic hormone (ACTH) and a subunit] and S100 proteins are visualised with fluorescein isothiocyanate (green). Visualisation of S100b was also performed using S100b-green fluorescent protein (GFP) transgenic rat of P60 (C). PROP1 signals localise in S100bpositive non-endocrine cells but not in any endocrine cells. Scale bar = 20 lm. An arrow indicates the PROP1 ⁄ S100b-positive cell. Open and closed arrowheads indicate the cell positive only for S100 or PROP1, respectively.
Transition of stem ⁄ progenitor cell in postnatal pituitary 939 stem ⁄ progenitor cells, a novel technique called telomapping, which visualises the length of telomeres, revealed that cells carrying the longest telomere localise not only mostly in the marginal cell layer, but also in the parenchyma of the anterior lobe. This again indicates the existence of pituitary stem ⁄ progenitor cells not only in the marginal cell layer, but also in the parenchyma of the anterior lobe (27). These data correspond to the localisation pattern of Sox2-expressing cells described in the present study.
In the present study, we further indicated that PROP1-positive cells consistently express Sox2 from the prenatal to the postnatal   5. Classification of non-endocrine cells with regard to the expression of Prop1, Sox2 and S100b. Immunohistochemistry of adult 60-day-old S100b-green fluorescent protein (GFP) transgenic rat was examined with double immunohistochemical staining of PROP1 (red) and SOX2 (blue) under S100b-promoterderived expression of GFP (green). Merged image is shown in the right panel. Scale bar = 20 lm. Expression pattern of Prop1, Sox2 and S100b was classified into five cell types (type 1 to type 5). The number of cells belonging to each cell type was counted in randomly selected one area (a, b and c) in each of three sections of two pituitaries for the marginal cell layer (A) and one area (a-d) in each of four sections of two pituitaries for the parenchyma of the anterior lobe (B). The proportion of each cell type was shown as a bar graph.
anti-mouse PROP1antibody gave signals likely in the cytoplasm and in the cytosol ⁄ nuclei, in contrast to our data indicating PROP1 signals are observed in the nuclei. In addition, in situ hybridisation demonstrated the presence of a Prop1 signal in the parenchyma of the mouse anterior lobe (15), supporting our data obtained by immunohistochemistry. These discrepancies might be caused by the difference in the properties of anti-PROP1 antibody used. Indeed, the antigens used are somewhat different between the C-terminal region 126-223 of rat PROP1 (present study) and the C-terminal region 151-223 of mouse PROP1 (Garcia-Lavandeira et al.). Identity in the C-terminal region 151-223 between mouse and rat is 84%, suggesting a possible difference in antigenicity, as well as in sensitivity.
Recently, Gleiberman et al. (26) hypothesised that the adult anterior pituitary is composed of two different terminally-differentiated endocrine cells originating from embryonic precursors and adult stem cells by the cell-lineage labelling technique using Nestin-Cre transgenic mouse. This hypothesis makes it possible to propose that both embryonic precursors and adult stem cells are mixed together in the marginal cell layer and the parenchyma of the anterior lobe from embryonic to postnatal development. Our observation that significant quantitative and qualitative transition in the marginal cell layer occurs during the first postnatal growth wave of the anterior pituitary suggests that a hierarchical change from embryonic precursors to adult stem cells may be triggered in the marginal cell layer during this stage but not in the parenchyma of the anterior lobe. However, Galichet et al. (31) presented a dissimilar result using similar Nestin-Cre transgenic mouse. The discrepancy may have arisen from the difference in the gene construct used. Nevertheless, we cannot exclude another possible explanation proposing that there is only one pool of Sox2-expressing stem ⁄ progenitor cells with a transitioning expression of Prop1 and S100b in the pituitary development. Further investigations are required to clarify this.
In our previous study (14) we proposed that PROP1 exists in Sox2-expressing stem ⁄ progenitor cells and disappears at the first stage of the transition of progenitor cells to committed cells and does not exist in any endocrine cells during embryonic development. In the present study, we demonstrate that PROP1 mostly exists in S100-positive folliculo-stellate cells but not in any endocrine cells during postnatal development. Although the majority of non-endocrine cells are folliculo-stellate cells, the existence of unidentified cell types in the adult anterior lobe is evident from the accumulating data, such as that obtained from nestin-immunoreactive cells (32). Folliculo-stellate cells themselves remain to be characterised clearly because they show heterogeneity in function, and their origin is unclear (4,5).
In recent years, a number of experimental studies have suggested that folliculo-stellate cells are candidates for pituitary stem ⁄ progenitor cells (10,11,33,34). Furthermore, some SOX2-positive stem ⁄ progenitor cells express S100 (12,27) and analysis of pituispheres revealed that S100-positive cells appear transiently on the way to generating terminally-differentiated endocrine cells (12). In Quantitative real-time polymerase chain reaction (PCR) was performed to estimate mRNA level of Prop1 and S100b using total RNAs extracted from the whole pituitaries at embryonic day (E)16.5 to postnatal day (P)0 (A) and from the anterior lobes at P5 to P60 (B). Each sample was measured in duplicate in two independent experiments, and data were calculated by comparative C T method to estimate the relative copy number to TAAT box binding protein (Tbp) as an internal standard. The data are presented as the mean AE SD of duplicate PCR in two independent experiments. Open and closed bars indicate Prop1 and S100b, respectively. Significance was examined by comparison with the value of E21.5 (A) or P5 (B). *P < 0.01 and **P < 0.05. Adult period P600 SOX2 SOX2/S100 SOX2/PROP1/S100 SOX2/S100, S100 Parenchyma Transition from PROP1 to S100 in stem ⁄ progenitor cells during pituitary development. Pituitary stem ⁄ progenitor cells express Sox2 in the early primordium and then start to co-express Prop1. During embryonic development of the pituitary, stem ⁄ progenitor cells line the marginal cell layer (MCL) and migrate in the parenchyma. After birth, transition of PROP1 to S100 progresses in the stem ⁄ progenitor cells at MCL and Sox2 ⁄ Prop1 ⁄ S100-expressing cells migrate in the parenchyma of the anterior pituitary where Sox2 ⁄ Prop1-expressing embryonic stem ⁄ progenitor cells still locate. the present study, we attempted to clarify the correlation among PROP1, SOX2 and S100b using adult S100b-GFP transgenic rats in which folliculo-stellate cells are visualised by GFP expressed under the control of S100b-promoter (22). We could classify non-endocrine cells into five groups by the expression pattern of PROP1, SOX2 and S100b: Type1, SOX2-positive; Type2, SOX2 ⁄ PROP1-positive; Type3, SOX2 ⁄ PROP1 ⁄ S100b-positive; Type4, SOX2 ⁄ S100b-positive; and Type5, S100b-positive. The proportion of each cell type was different in the marginal cell layer and the parenchyma of the anterior lobe. Eighty percent of the non-endocrine cells in the marginal cell layer were Type1 (SOX2-positive, 31%) and Type4 (SOX2 ⁄ S100b-positive, 46%). On the other hand, the majority of them in the parenchyma of the anterior lobe were Type3 (SOX2 ⁄ -PROP1 ⁄ S100b-positive, 60%). Because the expression of S100b started at around P0, SOX2 ⁄ S100b-positive cells would hypothetically be adult stem ⁄ progenitor cells. This might mean that SOX2-expressing adult stem ⁄ progenitor cells generate SOX2 ⁄ S100bpositive cells, followed by expressing Prop1 in the marginal cell layer of the adult pituitary. Then, most of these SOX2 ⁄ S100b ⁄ PROP1positive cells might migrate into the parenchyma of the anterior lobe and might compose major precursor-endocrine cells in the parenchyma, awaiting the next stimuli to be differentiated. On the other hand, in the parenchyma of the adult anterior lobe, Sox2expressing embryonic stem ⁄ progenitor cells are still present to generate SOX2 ⁄ PROP1-positive cells as precursor-endocrine cells. Thus, endocrine cells in the adult anterior lobe are composed of stem ⁄ progenitor cells with a different origin (Fig. 7). From another point of view, PROP1 exists transiently in PIT1-positive commitment cells, although the rate of coexistence declines gradually and reaches almost 0% at E21.5 (14). This coexistence of PROP1 and PIT1 completely disappears in the postnatal anterior lobe (data not shown), suggesting that PROP1 plays different roles or a different developmental system emerges and becomes dominant in the postnatal development. We have postulated the latter possibility, although the transition in the roles of PROP1 remain to be studied. Most recently, it was reported that prop1 knockdown in zebrafish caused abnormal morphology and gene expression in the early pituitary development, although gradual reverse was observed during late development, suggesting the existence of other fish-specific pathways downstream of prop1 (35). Although zebrafish is quite a different species from mammals, this result is suggestive for understanding the function of the PROP1 in embryonic and postnatal periods.

Primordium period
In summary, we have demonstrated that pituitary stem ⁄ progenitor cells are not only mostly present in the marginal cell layer, but also are scattered in the parenchyma of the anterior lobe. Moreover, a significant quantitative and qualitative transition occurred in the marginal cell layer during the first postnatal growth wave in the anterior pituitary. We propose that this transition might be a hierarchical change in pituitary stem ⁄ progenitor cells from those of embryonic to adult stages.