Sustained Notch signaling in progenitors is required for sequential emergence of distinct cell lineages during organogenesis

Abstract

Mammalian organogenesis results from the concerted actions of signaling pathways in progenitor cells that induce a hierarchy of regulated transcription factors critical for organ and cell type determination. Here we demonstrate that sustained Notch activity is required for the temporal maintenance of specific cohorts of proliferating progenitors, which underlies the ability to specify late-arising cell lineages during pituitary organogenesis. Conditional deletion of Rbp-J, which encodes the major mediator of the Notch pathway, leads to premature differentiation of progenitor cells, a phenotype recapitulated by loss of the basic helix-loop-helix (bHLH) factor Hes1, as well as a conversion of the late (Pit1) lineage into the early (corticotrope) lineage. Notch signaling is required for maintaining expression of the tissue-specific paired-like homeodomain transcription factor, Prop1, which is required for generation of the Pit1 lineage. Attenuation of Notch signaling is necessary for terminal differentiation in post-mitotic Pit1⁺ cells, and the Notch-repressed Pit1 target gene, Math3, is specifically required for maturation and proliferation of the GH-producing somatotrope. Thus, sustained Notch signaling in progenitor cells is required to prevent conversion of the late-arising cell lineages to early-born cell lineages, permitting specification of diverse cell types, a strategy likely to be widely used in mammalian organogenesis.

Notch signaling is an evolutionarily conserved mechanism that regulates proliferation, apoptosis, cell fate determination, and morphogenesis in organisms ranging from nematodes to humans (for reviews, see Lewis 1998; Artavanis-Tsakonas et al. 1999). Notch signaling is mediated by the interaction between the Notch receptor and its ligands Delta and Serrate. Both receptor and ligand are cell-surface transmembrane proteins that contain extracellular arrays of epidermal growth factor (EGF) repeats. Specific EGF repeats mediate direct interaction between ligand and receptor. Mammals have four Notch receptors (Notch1, Notch2, Notch3, and Notch4) and five ligands (Delta-like1, Delta-like3, Delta-like4 [homologs of Delta], and Jagged1 and Jagged2 [homologs of Serrate]). Upon ligand binding, Notch receptors undergo successive proteolytic cleavages that lead to the release of the Notch intracellular domain (ICD) and subsequent nuclear translocation. Once in the nucleus, the ICD forms a complex with the Rbp-J DNA-binding protein, which is the primary mediator of Notch signaling, and the Mastermind coactivator; converts Rbp-J from a transcriptional repressor to a transcriptional activator; and induces transcription of target genes such as members of the Hairy enhancer of split (Hes) family of basic helix- loop-helix (bHLH) DNA-binding transcription factors Hes1 or Hes5 and the Hes-related protein (Herp) family (for review, see Iso et al. 2003). In the Drosophila nervous system, Notch regulates a process of lateral inhibition, whereby a single neuron differentiates within a field of similar precursor cells. During lateral inhibition, cells expressing high levels of Notch ligand commit to neural differentiation and activate Notch signaling in their neighbors, thereby preventing them from adopting the same fate. Though Notch is best known for its role in lateral inhibition, activation of this pathway also promotes cell fate (Irvine 1999; Gaiano et al. 2000; Grandbarbe et al. 2003). Recent studies have shown that Notch signaling regulates a broad range of patterning processes during embryonic and postnatal development (Hitoshi et al. 2002; Krebs et al. 2003, Raya et al. 2003; Burns et al. 2005; Crosnier et al. 2005; Duncan et al. 2005; Fre et al. 2005; van Es et al. 2005; for reviews, see Lai 2004; Yoon and Gaiano 2005).

Coordination of signals from different pathways is essential for cell fate specification during animal development. The pituitary gland provides an excellent model system to study signaling events in organogenesis. The mature pituitary gland contains six hormone-producing cell types, including corticotropes secreting adrenocorticotrophic hormone (ACTH), a proteolytic product of proopiomelanocortin (POMC), thyrothopes secreting thyroid-stimulating hormone (TSH), somatotropes secreting growth hormone (GH), lactotropes secreting prolactin (PRL), gonadtropes secreting luteinizing hormone (LH) and follicle-stimulating hormone (FSH), and melanotropes secreting melanocyte-stimulating hormone (MSH), a cleaved product of POMC. These cells derive from a common primordium, Rathke's pouch, and appear in a defined temporal and spatial fashion (Japon et al. 1994; for reviews, see Watkins-Chow and Camper 1998; Sheng and Westphal 1999; Dasen and Rosenfeld 2001; Rizzoti and Lovell-Badge 2005). Three of these cell types—thyrothopes, somatotropes, and lactotropes—differentiate from Pit1-expressing precursors and depend on the function of Pit1, a tissue-specific POU-class homeodomain transcription factor (Camper et al. 1990; Li et al. 1990). The expression of Pit1 is positively regulated by the concerted efforts of the paired-like homeodomain transcription factor, Prophet of Pit1 (Prop1), and the Wnt/β-catenin signaling pathway (Gage et al. 1996; Sornson et al. 1996; Olson et al. 2006). Mutations in Pit1 or Prop1 result in a failure of Pit1 lineages differentiation, leading to a postnatal dwarf phenotype (Camper et al. 1990; Li et al. 1990; Ward et al. 2005). Differentiation of corticotropes is dependent on the T-box transcription factor, Tbx19, and regulated by other transcription factors and signaling events (Lamolet et al. 2001; Liu et al. 2001; Chesnokova and Melmed 2002; Pulichino et al. 2003).

Multiple signaling pathways converge to dictate molecular events underlying pituitary development. Loss-of-function and gain-of-function genetic studies together with ex vivo organ culture experiments have identified essential roles of FGF signaling emanating from the ventral diencephalon in proliferation and survival of progenitor cells comprising Rathke's pouch, as well as in establishing a dorsal to ventral gradient within the nascent pituitary gland (Ericson et al. 1998; Treier et al. 1998; Ohuchi et al. 2000; Revest et al. 2001). In addition, whereas BMP4 signaling is required for the pituitary organ commitment, BMP2 establishes a ventral to dorsal gradient necessary for cell type determination (Ericson et al. 1998; Takuma et al. 1998; Treier et al. 1998). Moreover, SHH signaling also exerts critical roles in supporting proliferation and terminal differentiation (Treier et al. 2001; Herzog et al. 2003; Sbrogna et al. 2003). Here, we report that the evolutionarily conserved Notch signaling pathway functions in the early phases of pituitary organogenesis to prevent premature differentiation of progenitors and drive specification of Pit1 precursors, which would otherwise emerge as the POMC-expressing corticotropes. Attenuation of Notch signaling at a later phase of pituitary development is required for the proper cell type terminal differentiation. These findings establish an additional role of the Notch signaling pathway in controlling the emergence of distinct precursor subtypes.

Results

Core components of the Notch signaling pathway during pituitary development

To explore the potential function of the Notch signaling pathway during pituitary development, in situ hybridization was carried out to examine the expression pattern of the known mammalian Notch ligands and receptors, Dll1, Dll3, Dll4, Jag1, Jag2, and Notch1-Notch4, respectively, as well as the direct downstream targets of Notch signaling, Hes1, Hes5, and Hey1. Jag1, Dll1, Notch2, and Notch3, as well as Hes1 and Hey1, were expressed in the Rathke's pouch, indicating that Notch signaling is active during early pituitary development (Figs. 1, 2). At embryonic day 9.5 (E9.5), Jag1, Notch2, Notch3, and Hes1 are expressed in the invaginating oral ectoderm. Between E10.5 and E12.5, Jag1, Dll1, Notch2, and Notch3 are detected throughout Rathke's pouch, with Hes1 demonstrating a restriction from the ventral-most region by E12.5. By the onset of Pit1 expression at E13.5, Dll1, Notch2, Notch3, and Hes1 expression have begun to be down-regulated in the perspective anterior pituitary, while persisting in the perilumenal cells. By contrast, Jag1 expression appears largely restricted to mesenchymal cells lining the invaginated Rathke's pouch. At later stages of pituitary development for E14.5 and E17.5, Dll1, Notch2, Notch3, and Hes1 were detectable only in cells adjacent to the lumen. These expression patterns reveal that both the ligands and receptors of the Notch pathway are expressed in early stages of pituitary development, and that subsequent down-regulation of expression correlates well with the onset of the pituitary gland maturation, consistent with an inhibitory role for Notch signaling in regulating cell terminal differentiation in the developing pituitary. While Dll3 is not expressed in early stages of Rathke's pouch formation, its expression is detectable at E14.5 and E17.5 in the intermediate lobe as reported previously (Supplementary Fig. 1), although no defect in melanotrope differentiation was observed in Dll3 ^-/- embryos (Raetzman et al. 2004).

FIGURE 1.

In situ analysis of expression pattern of core components of the Notch signaling pathway during pituitary development. Two ligands (Dll1, Jag1), two receptors (Notch2, Notch3), and the downstream target of Notch/Rbp-J signaling (Hes1) are expressed in the oral ectoderm and the Rathke's pouch (RP) by E12.5. At E13.5, Dll1, Jag1, Notch2, Notch3, and Hes1 expression are down-regulated in the anterior pituitary (AP) and are further confined to the perilumenal cells. At E14.5, Jag1 is expressed in the mesenchymal cells surrounding and within the pituitary gland. (IL) Intermediate lobe.

FIGURE 2.

Notch signaling represses premature corticotrope differentiation via its downstream target Hes1. (A) Hes1 and Hey1 are down-regulated in Rbp-J ^f/f, Pitx1-Cre mice at E11.5. (B) Premature corticotrope differentiation is indicated by POMC as well as Tbx19 expression in Rbp-J ^f/f, Pitx1-Cre mice at E12.5. Expression of bHLH genes, including Mash1, NeuroD1, and Math3, are up-regulated but restricted in the ventral region of Rathke's pouch. Corticotrope premature differentiation is evident in Hes1 ^{- / -} mice, accompanied by up-regulation of Tbx19, Mash1, and NeuroD1. However, there is no pronounced ectopic expression of Math3. The posterior lobe of pituitary of Hes1 ^{- / -} mice is absent. Double-immunofluorescence staining of POMC and Ki67 in E12.5 embryos showed increased POMC in Rbp-J ^f/f, Pitx1-Cre and Hes1 ^{- / -} embryos in comparison with their respective littermate controls. POMC⁺ cells are Ki67^-. Cells surrounding the lumen remain proliferative, while more cells at the ventral region of the pouch in mutant embryos exit the cell cycle and are negative for Ki67 staining.

Notch activity-dependent Hes1 expression controls the timing of corticotropes differentiation

The expression of multiple Notch receptors and ligands during early stages of pituitary development suggested that analysis of the endogenous functions of Notch signaling would be most effective through manipulating the expression of the DNA-binding protein Rbp-J, the primary mediator of Notch signaling. We therefore deleted the Rbp-J gene in Rathke's pouch by crossing mice containing a floxed Rbp-J allele (Tanigaki et al. 2002) with transgenic mice expressing the Cre recombinase under control of the Pitx1 promoter (Pitx1-Cre). This Cre allele exhibited efficient Cre-recombinase activity starting at E9.0 in all progenitors of Rathke's pouch (Olson et al. 2006). Quantitative RT-PCR of mRNA from micro-dissected E12.5 pituitaries revealed that floxed Rpb-J exons 6 and 7 were down-regulated fourfold in Rbp-J ^f/f, Pitx1-Cre embryos in comparison with Rbp-J ^f/+, Pitx1-Cre mice (data not shown). In situ hybridization readily detected a decrease in Hes1 and Hey1 expression at E11.5, suggesting that Pitx1-Cre can effectively mediate floxed Rbp-J recombination in pituitary premordium, and Hes1 and Hey1 are downstream targets of the Notch signaling in the pituitary (Fig. 2). Because Hes1 is required in multiple tissues for proper development, we also probed the function of Hes1 in pituitary organogenesis in order to compare the effects of loss of Notch activity and the loss of a downstream target of Notch signaling. Interestingly, Rbp-J ^f/f, Pitx1-Cre embryos isolated at E12.5 exhibited similar phenotypes to those observed in Hes1 ^{- / -} embryos (Fig. 2). In situ hybridization and immunofluorescence staining revealed premature differentiation of corticotropes at the most ventral region of the pouch, with concomitant up-regulation of Tbx19/Tpit, the function of which is necessary for corticotrope terminal differentiation (Pulichino et al. 2003). Expression of bHLH factors such as Mash1, NeuroD1, and Math3 were also expanded or ectopically up-regulated, except that no apparent up-regulation of Math3 was observed in Hes1 ^{- / -}. Interestingly, activation of these bHLH factors was restrained in the ventral-most region of the pouch where differentiated corticotropes were observed, suggesting that ablation of Notch activity, or relief from Hes1 repression, is insufficient to activate their expression and that additional components are required.

In the absence of either Notch activity or Hes1, there is a decreased number of proliferating pituitary progenitors and an increased number of cells exiting the cell cycle, as demonstrated by the proliferation marker Ki67 staining and bromodeoxyuridine (BrdU) labeling (Fig. 2; Supplementary Fig. 2). The cells in the vicinity of the lumen remained proliferative, whereas a higher proportion of cells in the caudomedial region ceased proliferation as compared with control littermates. These differences are not due to enhanced cell death as no significant differences in apoptosis were observed in both the Rbp-J ^f/f/Pitx1-Cre and the Hes1 ^-/- mice, demonstrated by staining for cleaved caspase-3 (data not shown). Double staining of Ki67 and POMC demonstrated that differentiated POMC⁺ corticotropes are Ki67^-, therefore excluding the possibility that the increased number of corticotropes is a consequence of increased proliferation.

Taken together, these data demonstrate that Notch activation is required to prevent premature differentiation of corticotropes, and that Hes1 is a primary target mediating the Notch pathway in the control of the timing of corticotrope differentiation. In addition, the presence of proliferating progenitors in the absence of Notch activity demonstrates that Notch signaling is not the sole determinant in maintaining the progenitor cell status during pituitary development.

Notch signaling is required for the commitment of Pit1⁺ precursor cell fate by regulating Prop1 expression

During pituitary development, the next cell lineage commitment following the initial determination of corticotropes is characterized by the appearance of Pit1 transcripts at E13.5. In the Rbp-J ^f/f, Pitx1-Cre mice, Pit1 expression fails to be initiated (Fig. 3A), and at E17.5, when the lateral region of the anterior pituitary would be normally populated with Pit1⁺ cells, most of which would be GH⁺ somatotropes, only a few Pit1⁺ cells could be identified in the Rbp-J ^f/f, Pitx1-Cre mice by either in situ hybridization or immunofluorescence labeling. Consequently, only a few GH⁺ somatotropes, and almost no TSHβ⁺ thyrotropes, could be detected (Fig. 3A). Instead, at E14.5 and E17.5, the anterior pituitary was composed predominately of POMC⁺ corticotropes (Fig. 3C; Supplementary Fig. 3). Gonadotrope differentiation, which occurs independently of Pit1 expression, was not significantly affected, as judged by expression of SF1 and LHβ (Fig. 3A). These results demonstrate that Notch signaling is necessary to induce the onset of Pit1 lineage commitment, and, in its absence, the progenitors take on a corticotrope fate at the expense of the Pit1 lineage. By contrast, analysis of the Hes1 ^-/- mice revealed that the ontogeny of initial Pit1 induction, terminal differentiation of Pit1 lineages, occurred normally (Fig. 3B; Supplementary Fig. 4). These data suggest that down-regulation of Hes1 does not account for the defects in Pit1 induction observed in Rbp-J mutant embryos, and that Notch signaling may control Pit1⁺ lineage commitment by regulating downstream targets other than Hes1.

FIGURE 3.

Notch signaling is required for Pit1 lineage commitment. (A) Pit1 expression is absent in Rbp-J ^f/f, Pitx1-Cre mutant embryos at E13.5. At E17.5, Rbp-J ^f/f, Pitx1-Cre mutant embryos, Pit1 expression, and differentiation of Pit1 lineages thyrotropes (TSHβ) and somatotropes (GH) are impaired, while gonadotrope commitment, indicated by SF1 expression at E14.5 and LH expression at E17.5, is not affected in Rbp-J ^f/f, Pitx1-Cre mutant embryos. (B) Pit1 induction occurs normally in E13.5 Hes1 ^-/- embryos. (C) Double-immunofluorescence staining of POMC (green) and Ki67 (red), or Pit1 (green) and Ki67 (red) at E14.5 pituitaries shows that the anterior pituitary of the Rbp-J ^f/f, Pitx1-Cre mutant is populated with an increased number of corticotropes and is devoid of Pit1⁺ cells. The differentiated cells are not proliferative, as indicated by Ki67 staining. Cells surrounding the lumen are Ki67⁺ in both Rbp-J ^f/f, Pitx-Cre mutant embryos and littermate controls. (D) D2R expression at the intermediate lobe (IL) is increased in Rbp-J ^f/f, Pitx1-Cre mutant embryos at E17.5. Dual-immunofluorescence labeling of POMC (green) and Ki67 (red) at E17.5 pituitaries shows almost all POMC⁺ cells in IL are Ki67^-.

Because Pit1 activation is regulated by the concerted actions of Prop1 and the Wnt/β-catenin signaling pathway, we sought to examine the expression of Prop1 as well as Axin2, a direct downstream target of Wnt/β-catenin signaling and therefore serving as an indicator of signaling activity in this pathway (Olson et al. 2006). In situ hybridization analyses revealed no significant differences in Axin2 expression in the Rbp-J ^f/f, Pitx1-Cre mice as compared with heterozygous littermates, implying that the Wnt/β-catenin signaling remained intact in the absence of Notch activity (Fig. 4B). Prop1 expression, conversely, was markedly diminished by E12.5, when levels of Prop1 mRNA normally peak. We further noted that at E11.5, when Prop1 is initially expressed at lower levels, there was no obvious difference in the Prop1 expression profile exhibited by the mutant and the control (Fig. 4A), suggesting that Notch activity is required for the up-regulation of Prop1 at E12.5, but not the initiation of Prop1 expression. By comparison, Prop1 transcripts in Hes1 ^-/- mice at E12.5 were not significantly affected (Fig. 4B), consistent with the model that Rbp-J may modulate Prop1 expression directly.

FIGURE 4.

Prop1 is a direct target of Notch signaling. (A) Prop1 expression is significantly down-regulated in Rbp-J ^f/f, Pitx1-Cre mutant embryos at E12.5 but is unchanged at E11.5. (B) Axin2 expression in Rbp-J ^f/f, Pitx1-Cre mutant embryos and Prop1 expression in Hes1 ^-/- at E12.5 are not significantly changed. (C) Genomic DNA sequences of mouse and human Prop1 were compared using VISTA. (Red) Promoters, (yellow) UTRs, (blue) exons, (pink) introns. Two evolutionarily conserved regions, the promoter and the first intron, show >75% homology. A putative Rbp-J-binding site is identified in the first intron. (D) A ³²P-labeled 25-bp oligonucleotide encompassing the putative Rbp-J-binding site was incubated in the absence (lane 1) or in the presence (lanes 2-5) of in vitro translated Rbp-J and the competitors. Unlabeled oligonucleotides (lane 3), equivalent oligonucleotides where the putative Rbp-J-binding site was mutated (lane 4), or the oligonucleotides containing a Rbp-J-binding site from Epstein-Barr virus C promoter region (lane 5) were used as competitors at 100× molar excess. The arrow indicates the shifted probe caused by Rbp-J binding, and the arrowhead indicates free probe. (E) Quantitative ChIP assay of E12.5 pituitaries using anti-Rbp-J showed recruitment of Rbp-J to the first intron of Prop1. The promoter regions of the Hes1 and GHRHR were used as the positive and negative controls, respectively. (F,G) Transient transfection of 2-kb Prop1-Luc (F) or Prop1in-Luc (G) with control siRNA or siRNA specific to Rbp-J.

To further investigate the molecular mechanism underlying the genetic relationship between Notch activity and Prop1 expression, an in silico search utilizing VISTA (http://rvista.dcode.org) was performed for conserved noncoding regions in the Prop1 genes among six mammalian species: mouse, rat, dog, sheep, bovine, and human. Two highly evolutionarily conserved regions (>75%) were identified, in the promoter and the first intron, respectively. A search for conserved regulatory elements in these regions identified a consensus binding site for Rbp-J within the first intron (Fig. 4C). In order to examine whether this putative binding site might be recognized by Rbp-J, we performed electrophoretic mobility shift assays (EMSA) using synthetic oligonucleotides representing the putative binding site and flanking regions, and Rbp-J produced by in vitro transcription and translation. Rbp-J bound efficiently to the Prop1 intron in vitro, and this binding could be competed with itself or a known Rbp-J-binding site, but not with oligonucleotides in which the putative recognition sites had been mutated (Fig. 4D).

In order to estimate whether Rbp-J is recruited to the Prop1 gene in vivo, we performed chromatin immuno-precipitation (ChIP) analysis from E12.5 dissected pituitaries using antiserum specific for Rbp-J (Chu and Bresnick 2004). The results revealed recruitment of Rbp-J to the promoter of Hes1, as well as to the first intron of Prop1, suggesting that Prop1 is a direct target of Notch signaling activity in the pituitary (Fig. 4E). Consistent with these in vivo findings, transient cotransfection reporter assays in the pituitary cell line GHFT1 employing either a 2-kb fragment including the promoter and first intron of Prop1, or the evolutionarily conserved intron region in conjunction with a heterologous minimal promoter, demonstrate that reporter activity was inhibited by small interfering RNA (siRNA) directed against Rbp-J (Fig. 4F,G). These data support a mechanistic role of Rbp-j in direct regulation of Prop1.

Notch activity is necessary to suppress melanotrope cell fate

While the ventromedial regions of Rathke's pouch differentiate into Pit1⁺ precursors, cells originating from the dorsal regions of the pouch continue to proliferate, eventually differentiating into Tbx19⁺, Mash1⁺, D2R⁺ (dopamine D2 receptor), POMC⁺ intermediate lobe melanotropes (Meador-Woodruff et al. 1989; Japon et al. 1994; Lamolet et al. 2001; Liu et al. 2001; Pulichino et al. 2003). In situ hybridization analyses of D2R, and immunostaining of POMC at E17.5, revealed an increase in the overall number of melanotropes in the intermediate lobe of the Rbp-J ^f/f, Pitx1-Cre mice as compared with control littermates (Fig. 3D). Double-immunofluorescence staining of POMC and Ki67 revealed that intermediate lobe POMC⁺ cells were negative for Ki67 staining, ruling out the possibility that expansion of the melanotrope population was caused by increased proliferation. By comparison, Ki67⁺ cells in the intermediate lobe were significantly diminished in Rbp-J^f/f, Pitx1-Cre mice, suggesting that more progenitors were differentiated to melanotrope cell fate in the absence of Notch activity. Intriguingly, the intermediate lobe of the Hes1 ^-/- embryo is virtually absent (Supplementary Fig. 4). This is likely due to a nonpituitary effect and is currently under investigation.

Ectopic Notch signaling prevents terminal cell differentiation

Notch activity, as indicated by Hes1 expression, is down-regulated as cells undergo lineage commitment, suggesting that active Notch signaling may interfere with terminal differentiation. In order to test this hypothesis directly, we employed a transgenic mouse model engineered to sustain Notch signaling in Pit1-expressing cells by expressing epitope-tagged Notch1 ICD (NICD) under the control of a 15-kb 5′ flanking region of the Pit1 promoter, which has been shown to activate target gene expression in three distinct Pit1-dependent lineages. Transgenic mice exhibit a postnatal dwarf phenotype, weighing one-half of wild-type littermates at 1 mo of age; they remain smaller than wild-type litter-mates throughout life. Analyses of adult transgenic and wild-type littermates revealed hypoplasia of the anterior pituitary in transgenic mice, with prominent reduction of somatotrope, thyrotrope, and lactotrope populations (data not shown). These observations demonstrate that sustained expression of activated Notch in Pit1⁺ precursors inhibited terminal differentiation of three Pit1-dependent lineages.

Examination of ectopic NICD expression by immunohistochemistry using an epitope-specific antibody demonstrated a spatial and temporal pattern of expression in the anterior pituitary almost identical to that of endogenous Pit1 expression, initiated at E13.5, and sustained thereafter (Fig. 5A). Dual immunofluorescence labeling with anti-Pit1 and anti-HA revealed colocalization of Pit1 and NICD (Supplementary Fig. 3). Ectopic NICD expression in the Pit1 lineage did not affect apoptosis, as assessed by a TUNEL assay or cleaved caspase 3 immunostaining (data not shown), nor did it affect cell proliferation as assayed by BrdU labeling and Ki-67 staining at E17.5 (Fig. 5B; data not shown). In fact, at E13.5, when Pit1 is initially expressed, double-immunostaining analysis using anti-Pit1 and anti-Ki67 showed that almost all Pit1-expressing cell were Ki67-negative, suggesting that Pit1⁺ cells were not proliferating at this embryonic stage. Double-immunofluorescence labeling of Pit1 and Ki67 in E14.5 and E17.5 transgenic mice and wild-type controls showed that there was no significant difference in the number of Pit1⁺/Ki67⁺ cells. Thus, NICD expression in Pit1⁺ cells did not affect proliferation or survival of the Pit1 lineages (Fig. 5B). However, in situ hybridization analysis of E17.5 pituitaries from transgenic and wild-type littermates revealed a significant decrease of somatotropes and thyrotropes, as measured by GH, GHRHR (growth hormone-releasing hormone receptor), and TSHβ expression (Fig. 5C), with only occasional GH-positive cells and TSHβ-positive cells observed at E17.5 pituitary in transgenic mice. To determine the origin of these cells, double-immunofluorescence labeling with antiserum specific for HA epitope and either GH or TSHβ revealed that the residual GH⁺ or TSHβ⁺ cells in transgenic pituitaries were HA-negative, indicating they were derived from Pit1⁺ precursors that had failed to efficiently express the Pit1-NICD transgene (Fig. 5D). These results suggest that ectopic expression of NICD blocks terminal differentiation and that relief from the Notch repression is prerequisite for cells to undergo terminal differentiation. Pit1-independent corticotropes and gonadotrope lineages, expressing POMC and LHβ, respectively, were not significantly affected by Pit1-NICD expression (Fig. 5C), suggesting that the effects of Notch signaling on inhibition of terminal differentiation occurred in a cell-autonomous manner.

FIGURE 5.

(A) Transient transfection in the pituitary GC cell line shows that the Pit1-NICD transgene construct can stimulate Hes1-Luc activity. Immunohistochemical staining with anti-HA antibody to detect transgene expression in E17.5 transgenic pituitary. Transgene expression is detectable as early as E13.5 when Pit1 expression begins. (B) Double-immunofluorescence staining of Pit1 (green) and Ki-67 (red) in wild-type and transgenic pituitary at E14.5 and E17.5 showed almost all proliferating cells are Pit1-negative at these stages and ectopic NICD expression in Pit1 lineage did not induce proliferation. (C) Prolonged activation of Notch signaling inhibits terminal cell differentiation of Pit1 lineages. In situ analysis of the transgenic pituitaries was performed with the GH, TSHβ, POMC, LHβ, and GHRHR. Differentiation of corticotropes and gonadotropes is less affected. (D) Double-immunofluorescence staining of HA (red) and terminal differentiation markers GH (left, green) and TSHβ (right, green) in pituitaries of transgenic mice showed that differentiated cells did not express the transgene.

In Snell mice, which harbor a point mutation in the Pit1 gene that abolishes its transcriptional function, three Pit1 lineages are absent while gonadtropes are markedly increased (Dasen et al. 1999), and the latter phenotype was not observed in the NICD transgenic mice. Thus, our results implicated that NICD blocks lineage differentiation likely by interfering with pathways either parallel to or downstream from the Pit1 action.

Ectopic NICD expression down-regulates proneural bHLH expression

Because Hes1 and Hes5 are well established Notch targets in other tissues that in turn inhibit the expression or antagonize the functions of cells specifying bHLH proteins, we examined the expression of Hes1 and Hes5, as well as Mash1 and NeuroD1, which are expressed during pituitary development (Liu et al. 2001; Lamolet et al. 2004). In situ hybridization analysis demonstrated up-regulation of Hes1 and ectopic expression of Hes5 at E14.5 pituitaries of Pit1-NICD transgenic mice (Fig. 6A), whereas Hes1 expression is almost undetectable in the anterior pituitary of wild-type embryos at this stage of development and Hes5 is normally not expressed in pituitary (Figs. 1, 7A). These results indicate that the down-regulation of Hes1 expression in the wild-type anterior pituitary is most likely due to down-regulation of Notch receptors and ligands, and that Rbp-J itself apparently functions as a transcription repressor at later stages of pituitary development, consistent with the observation that overexpression of a dominant-negative form of Rbp-J (Rbp-J + engrailed repressor domain) in Pit1⁺ cells did not interfere with terminal differentiation of Pit1 lineages (X. Zhu, unpubl.)

FIGURE 6.

(A) In situ analysis of bHLHs expression in E14.5 pituitaries of transgenic mice and wild-type control. Both Hes1 and Hes5 were induced by ectopic NICD expression while Mash1, NeuroD1, and Math3 were significantly repressed. (B) Math3 expression ontogeny during pituitary development. Expression of Math3 begins at E13.5 in the caudomedial region of the anterior pituitary and persists in the anterior lobe of the adult pituitary. (C) Expression of Math3 is down-regulated in Snell mice at E17.5. (D) Alignment of the promoter regions of human and mouse Math3 using VISTA identified an evolutionarily conserved region in the promoter with >75% homology. A putative binding site for Pit1 lies in this region. (E) ChIP from adult pituitaries using anti-Pit1 shows Pit1 is recruited to the Math3 as well as GHRHR promoter regions.

FIGURE 7.

(A) Math3 is required for somatotrope maturation. In situ analysis in E17.5 pituitaries of Math3 ^-/- and littermate controls showed that somatotrope markers GH and GHRHR are down-regulated in Math3 ^-/-. Immunofluorescence staining showed GH protein is undetectable at E17.5 in Math3 ^-/- embryos. GHRHR expression remained undetectable in postnatal Math3 ^-/- pituitaries. (B) Quantitative RT-PCR of Pit1, Prl, TSHβ, GHRHR, and GH at E17.5 pituitaries showed significant and specific down-regulation of GH and GHRHR mRNA. (C) Model of Notch signaling in pituitary development. Notch-regulated Hes1 expression maintains self-renewal of the Ki67⁺ progenitor and prevents precocious corticotrope differentiation. Notch activity promotes Prop1 up-regulation at E12.5 and drives progenitors to adopt the fate of Pit1⁺ precursors. Mash1 and Math3 are required for proper development of thyrotropes and somatotropes, respectively.

As might be expected, sustained induction of Hes1 and Hes5 in Pit1-NICD transgenic mice results in significant down-regulation of Mash1 and NeuroD1 in the anterior pituitaries of transgenic mice (Fig. 6A). It has been reported recently that NeuroD1 is required for the early corticotrope differentiation but not lineage commitment (Lamolet et al. 2004), and in related studies, we have found that Mash1 exerts roles in terminal differentiation of thyrotropes, gonadtropes, and corticotropes (our unpublished data). Intrigued by these findings, we sought to explore if other proneural bHLH factors are expressed in the pituitary and, if so, whether they are also down-regulated by Pit1-NICD expression. A semiquantitative RT-PCR comparing E14.5 wild-type and transgenic pituitaries was performed to characterize the potential expression of a panel of proneural bHLH factors and Math3 was identified, expression of which was markedly inhibited in transgenic embryos in comparison with wild-type littermates (Fig. 6A).

Math3 is required for somatotrope maturation and function

Targeted disruption of Math3 leads to cerebellar defects, with increased apoptosis in the external granular layer as well as postnatal growth retardation (Tomita et al. 2000). However, the underlying molecular mechanism(s) regulated by Math3 during postnatal growth remains largely unknown. Because Math3 is expressed in the pituitary, and the postnatal dwarfism phenotype in Math3 ^-/- mice resembles those of the Pit1-NICD and Snell mice, we probed Math3 functions in pituitary development. In situ hybridization analysis revealed that Math3 expression in the anterior pituitary begins at E13.5 and persists throughout adulthood (Fig. 6B). The spatio-temporal expression profile of Math3 is reminiscent of that of Pit1. In silico comparison of the promoter regions of Math3 from mouse and human using VISTA identified a consensus Pit1-binding site within a highly conserved region (>75%) in the promoter (Fig. 6D), suggesting that Math3 is likely a direct downstream target of Pit1. ChIP analysis of adult mouse pituitaries revealed recruitment of Pit1 to the Math3 promoter, as well as to the promoter of GHRHR, a known target of Pit1 (Fig. 6E). Moreover, Math3 expression in the pituitary was significantly reduced in Snell mice (Fig. 6C). Together, these data suggest that Math3 is a direct downstream target of Pit1 and uncover a developmentally regulated program whereby Math3 is repressed by Notch signaling and activated by the tissue-specific POU domain transcription factor Pit1. To our knowledge, in addition to Pit1 itself, Math3 is the first identified transcription factor regulated by Pit1.

Examination of pituitaries from P8 and P15 Math3 ^-/- and wild-type littermates revealed a hypoplastic anterior pituitary with decreased numbers of somatotropes in Math3 ^-/- mice (data not shown). At E17.5, expression of POMC, LHβ, Pit1, αGSU, and TSHβ were not affected in Math3 ^-/- mice (Fig. 7B; data not shown). However, expression of GHRHR was almost completely abolished. The number of somatotropes expressing GH was markedly decreased and, more strikingly, even fewer GH⁺ cells could be detected by immunostaining (data not shown), suggesting that Math3 is required in somatotropes for expression of GHRHR and GH at embryonic stages. Quantitative RT-PCR assay confirmed reduced levels of total GH mRNA in E17.5 Math3 ^-/- pituitaries, and further studies revealed no aberrant mRNA splicing of GH mRNA that might lead to defective GH protein synthesis (data not shown). The molecular mechanisms of the apparent discrepancy between the levels of GH mRNA and protein have clear implications that will be the subject of future investigation. However, the block of somatotrope maturation in Math3 ^-/- mice is transient and recovers postnatally, because in P8 and P15 Math3 ^-/- pituitaries, GH⁺ somatotropes, although obviously present in reduced cell number, were readily detectable by immunostaining, suggesting that Math3 is necessary for the proper onset of somatotrope specification. Expression of GHRHR, in contrast, remained minimal (Fig. 7A). The postnatal dwarf phenotype in Math3 ^-/- mice is likely attributable to the reduced number of somatotropes, as is observed in little mice, which harbor a point mutation in the GHRHR gene, rendering somatotropes incapable of responding to the hypothalamic trophic factor GHRH (Lin et al. 1993).

Discussion

The distinct cell types in the pituitary gland are generated in a temporal and spatial fashion. We have shown here that Notch activity operates in a precise temporal window during pituitary development. In early developmental stages, Notch signaling prevents premature differentiation by regulating Hes1 expression. Perhaps more importantly, Notch signaling plays an essential role in controlling the lineage commitment of Pit1⁺ precursors, without which they would “switch” to an alternative cell fate. It does so, at least in large part, by directly regulating the expression of Prop1, a gene that is necessary for the genesis of Pit1⁺ precursors. In the late phases of pituitary development, however, Notch activity is dramatically attenuated in Pit1⁺ cells, in part because expression of a subset of bHLH factors that are otherwise negatively regulated by active Notch signaling is required for terminal differentiation of Pit1 lineages. One of these bHLH factors, Math3, is itself a downstream transcription target of Pit1, and is crucial for maturation and expansion of somatotropes through regulation of GHRHR expression.

Notch activation controls formation of Pit1 precursors

We have shown that both ligands and receptors of the Notch signaling pathway are expressed in proliferating progenitors in Rathke's pouch during early pituitary development. These cells transduce Notch signaling, as indicated by the loss of the Notch downstream effectors Hes1 and Hey1 when Notch activity is impeded by deletion of the Rbp-J gene. As cells within the Rathke's pouch exit the cell cycle, migrate ventrally and laterally out of the proliferative zone, and undergo lineage commitment and subsequent terminal differentiation, Notch signaling is turned off, and remaining Notch activity can be detected in only periluminal cells. Thus, the proliferative zone in pituitary development to a certain extent is similar to the ventricular zone in cerebral cortex development. In that system, Notch signaling is required to maintain the progenitor pools and inhibit neuronal differentiation (for review, see Yoon and Gaiano 2005). Due to the great inherent redundancy within the Delta/ Notch pathway and the pleiotropic defects associated with deletion of some of the Notch pathway components, blocking the Notch signaling results in premature differentiation and depletion of progenitor pools, precluding the analysis of Notch function in later-born cell lineage determination.

We have observed that in Rathke's pouch when Rbp-J is conditionally removed in progenitor cells, there is no immediate complete conversion of progenitor cells into post-mitotic corticotropes. Instead, the Ki67⁺ proliferative zone remains, which is capable of supporting the differentiation of gonadotropes and intermediate lobe melanotropes, implying that the progenitor cells retain a certain degree of pluripotency. However, Pit1⁺ precursors are almost completely missing, and instead are converted into corticotropes. Our data suggest a model in which Notch signaling controls the formation of diverse precursor subtypes from a progenitor pool. In this model, progenitors receiving no or only a short pulse of Notch signaling progress to the first-born cell lineage corticotropes, while those receiving longer durations of Notch signaling are suggested to undergo irreversible changes in gene expression and/or epigenetic status, such that they are competent to assume a later-born cell fate such as the Pit1 lineage precursor. Consistent with this model is our finding that Notch signaling is essential for the maintenance of Prop1 expression, providing a molecular mechanism underlying Notch signaling-dependent commitment of Pit1 precursors. This proposed model is likely generally employed in other developmental contexts. It has been shown that in the ventricular zone during mammalian cerebral cortex development, radial glial cells that have been temporarily subjected to Notch activation skip the early-born cell fate and differentiate into later-born upper-layer neurons (Mizutani and Saito 2005). In a detailed analysis of the kinetics and mechanisms of T-lineage differentiation in an in vitro culture system, continuous Notch signaling is obligatory to maintain development along the T-cell pathway (Taghon et al. 2005).

In addition to the well-characterized Notch targets Hes1 and Hey1, we have identified Prop1 as a Notch downstream target. However, their expression exhibits different dynamics, with Hes1 and Hey1 detectable at E9.5 in Rathke's pouch—indicative of active Notch signaling (Fig. 1; data not shown)—whereas Prop1 is not expressed until E11.5, suggesting that Notch activity is not sufficient to induce Prop1 expression (Sornson et al. 1996) and that initiation of Prop1 expression is independent of the Notch activity. However, the sustained expression of Prop1 is Notch-dependent. These data are consistent with the view that expression of Notch targets is dependent on the developmental status and appears to be target gene specific (Umesono et al. 2002; Anthony et al. 2005; Taghon et al. 2005). Premature expression of Prop1 in Rathke's pouch proves to be deleterious, leading to agenesis of the anterior pituitary gland, probably by inhibiting the endogenous function of Hesx1 (Dasen et al. 2001). Prolonged expression of Prop1 interferes with anterior pituitary cell differentiation (Cushman et al. 2001). Therefore, identifying the critical factors initiating Prop1 gene expression will provide further insights into the integration of the Notch pathway with other developmental programs. It has been reported recently that Prop1 is required for Notch2 protein expression in the pituitary (Raetzman et al. 2004). However, our in situ hybridization analyses of Prop1 ^-/- embryos showed that expression of Notch2 and Hes1, which is dependent on Notch signaling, is not significantly affected (Raetzman et al. 2004; our unpublished data), suggesting that deletion of Prop1 is not sufficient to down-regulate Notch signaling during pituitary development.

Down-regulation of Notch activity at a later stage is necessary for terminal differentiation

Notch signaling is down-regulated as cells exit the cell cycle and undergo terminal differentiation. We show that forced expression of NICD in Pit1⁺ post-mitotic cells is sufficient to prevent them from differentiating into hormone-producing cells, consistent with the role of Notch signaling in maintaining progenitor fate. These data highlight the importance of properly controlled Notch signaling during development and raise a fundamental question as to the regulation of Notch activity. Our expression data show that Notch activity in the pituitary is closely correlated with the expression of Notch receptors and ligands, implying a key component of this regulation would be their own expression, although other cellular mechanisms may also be involved (for review, see Schweisguth 2004). A recent study in Zebrafish retinal neurogenesis has demonstrated that a mutation in histone deacetylase 1 (Hdac 1) results in a defect in the transition from proliferating progenitor cells to post-mitotic neurons due to a failure to suppress Notch and Wnt pathways. Although this mechanism may be cell type specific, it suggests that transcription corepressors of Rbp-J may feedback to regulate Notch activity (Yamaguchi et al. 2005).

NICD expression in Pit1⁺ post-mitotic cells leads to induction of Hes1, Hes5, and Hey1 and repression of Mash1, NeuroD1, and Math3, implicating their functions in Pit1 lineage terminal differentiation. Indeed, Mash1 executes roles in differentiation of thyrotrope as well as corticotrope, melanotrope, and gonadotrope (our unpublished data). In this study, we demonstrate that Math3 is critical for maturation and expansion of somatotropes by regulating the expression of GHRHR. Thus, using the Pit1-NICD transgenic mouse as a genetic approach, we have uncovered bHLH factors as new components involved in Pit1 lineage differentiation. Deletion of neither Mash1 nor Math3 alone account for the full spectrum of defects in the Pit1-NICD transgenic mouse and, therefore, whether they exert overlapping roles during pituitary development, as has been demonstrated in neurogenesis (Tomita et al. 2000), is the subject of ongoing investigation.

Materials and methods

Mice

A DNA-encoded mouse Notch1 intracellular domain, amino acids 1744-2183, was generously provided by Dr. R. Kopan (Washington University, St. Louis, MO) (Schroeter et al. 1998). A hemagglutinin (HA) tag was added at the C terminus of NICD cDNA. The Notch1-ICD-HA ORF was inserted between a rabbit 0.65-kb β-globin intron and a 0.63-kb poly(A) fragment of the human growth hormone gene at the 3′ end. The 15-kb Pit1 promoter was inserted 5′ of this cassette. Transgenic mice were generated as described in Treier et al. (1998), and transgenic animals were genotyped by PCR using primers 5′-GCAACGT GCTGGTTATTGTGC-3′ and 5′-CGGTCTGTCTGGTTGTG CAAGCTG-3′. The transgenic line was maintained on a CB6F1 background as a heterozygote. The male transgenic animals are dwarf and fertile while the females are dwarf and sterile. Other mice used in this study—floxed Rbp-J, Hes1 knockout, and Math3 knockout—have been described previously (Ishibashi et al. 1995; Tomita et al. 2000; Tanigaki et al. 2002). The Rbp-J ^{floxed/floxed}, Pitx1-Cre mutant embryos were obtained by crossing Rbp-J ^{floxed/floxed} mice with mice heterozygeous for floxed Rbp-J and Pitx1-Cre (Olson et al. 2006).

In situ hybridization and immunofluorescence

In situ hybridization and immunofluorescence were carried out as previously described (Simmons et al. 1990). Mouse embryos from E9.5 to E17.5 were fixed in 10% neutral formalin, penetrated with 20% sucrose in PBS, and embedded in OCT compound. Serial 16-μm sections were hybridized with ³⁵S-labeled antisense RNA probes. The probes used in this study were either purchased as EST clones from ATCC or generated by RT- PCR from various tissues and verified by sequencing. For immunofluorescence staining, the sections were boiled for 10 min in 10 mM citrate buffer (pH 6.0) to retrieve antigens and stained with mouse mAbs against the HA tag (Babco, 1:400), Ki-67 (Pharmingen, 1:50), BrdU (ICN Biomedicals, 1:20), rabbit polyclonal antibodies against GH (DAKO, 1:200), TSHβ (National Hormone and Pituitary Program, National Institute of Diabetes and Digestive and Kidney Diseases, rabbit, 1:400), ACTH (Sigma, 1:100), and Pit-1 (1:200). Secondary peroxidase, Alexa Fluor 488-, and Alexa Fluor 594-conjugated antibodies were from Jackson ImmunoResearch and Molecular Probes. Slides were coverslipped in Vectashield Mounting Medium with DAPI (Vector Laboratories). The results were analyzed on a Zeiss Axioplan2 microscope with a Hamamatsu camera, and pictures were superimposed in Adobe Photoshop.

EMSA, transfection, ChIP, and quantitative PCR

EMSA experiments were performed as previously described (Reizis and Leder 2002). Double-stranded oligonucleotides were labeled with γ-³²P-ATP. Rbp-J was transcribed and translated using the TNT Quick Coupled Transcription/Translation Systems (Promega). In vitro translated proteins were incubated with 1× binding buffer (25 mM Tris at pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 5% glycerol), DNA competitors, and 1 μg of poly-dIdC for 15 min on ice prior to adding probe. Probe was added to the reaction and allowed to bind for 20 min at room temperature, and then protein-DNA complexes were resolved by electrophoresis. Oligonucleotides used for EMSA are 5′-CTTGAGCTCGTGGGAAAGGCTTGCC-3′, 5′-GGCAAG CCTTTCCCACGAGCTCAAG-3′ (Prop1 intron); 5′-CTTGA GCTCGTGaacAAGGCTTGCC-3′, 5′-GGCAAGCCTTgttCA CGAGCTCAAG-3′, (Prop1 intron with mutations); and 5′-AAACACGCCGTGGGAAAAAATTTGG-3′, 5′-CCAAATT TTTTCCCACGGCGTGTTT-3′ (Rbp-J-binding site from Epstein-Barr virus C promoter region). Transient transfection of pituitary cell lines GC and GHFT1 using SuperFect (Qiagen) or Lipofectamine 2000 (Invitrogen) was performed according to the manufacturers’ instructions. ChIPs were performed as previously described (Wang et al. 2006) with some modifications. The pituitaries were dissected from wild-type E12.5 embryos or adult mice and cross-linked with 2% formaldehyde for 20 min at room temperature. Aliquots (1 μL) of 30 μL of purified DNA fragments were analyzed by quantitative real-time PCR. For RT-PCR analysis, total RNA was isolated from dissected pituitary using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized using SuperScript II (Invitrogen). Quantitative RT-PCR using Sybr Green was performed on an Mx3000P QPCR System (Stratagene). Primers sequences are available on request.