β-catenin is required in the neural crest and mesencephalon for pituitary gland organogenesis

The pituitary gland is a highly vascularized tissue that requires coordinated interactions between the neural ectoderm, oral ectoderm, and head mesenchyme during development for proper physiological function. The interactions between the neural ectoderm and oral ectoderm, especially the role of the pituitary organizer in shaping the pituitary precursor, Rathke’s pouch, are well described. However, less is known about the role of head mesenchyme in pituitary organogenesis. The head mesenchyme is derived from definitive mesoderm and neural crest, but the relative contributions of these tissues to the mesenchyme adjacent to the pituitary are not known. We carried out lineage tracing experiments using two neural crest-specific mouse cre lines, Wnt1-cre and P0-cre, and determined that the head mesenchyme rostral to the pituitary gland is neural crest derived. To assess the role of the neural crest in pituitary development we ablated it, using Wnt1-cre to delete Ctnnb1 (β-catenin), which is required for neural crest development. The Wnt1-cre is active in the neural ectoderm, principally in the mesencephalon, but also in the posterior diencephalon. Loss of β-catenin in this domain causes a rostral shift in the ventral diencephalon, including the pituitary organizer, resulting in pituitary dysmorphology. The neural crest deficient embryos have abnormally dilated pituitary vasculature due to a loss of neural crest derived pericytes. β-catenin in the Wnt1 expression domain, including the neural crest, plays a critical role in regulation of pituitary gland growth, development, and vascularization.


Background
In comparison to invertebrate chordates, vertebrates are characterized by an elaborate head with an expanded brain, elaborate sensory organs, and a craniofacial skeleton. These modifications from the ancestral chordate where enabled in part by the appearance of neural crest cells, neurogenic placodes, and muscularized hypomeres, which are derived from definitive mesoderm, in the ancestral vertebrate [1,2]. These three tissues, along with the neural ectoderm, interact in diverse ways to form the elaborate head structures of vertebrates, including sensory organs and craniofacial skeletal components [3,4]. The pituitary gland also appears with the formation of vertebrates and is partially derived from the adenohypophyseal placode [5]. The contributions of the adenohypophyseal placode to pituitary development are well described [5,6]; however, much less is known about neural crest and definitive mesoderm contributions to pituitary organogenesis.
Rathke's pouch, the precursor to the pituitary anterior and intermediate lobes, is surrounded by cranial head mesenchyme and is in close proximity to the rostral end of the notochord and pre-chordal plate, which have important signaling functions in early head development [7]. Chick explant studies demonstrated that co-cultures of the ventral diencephalon and Rathke's pouch could only induce differentiation of corticotropes when mesenchyme was included in the culture [8]. Additionally, explants of chick notochord can cause surface ectoderm to invaginate, forming a structure similar to Rathke's pouch [8]. These experiments suggest that chick head mesenchyme plays a role in supporting cell differentiation in the anterior lobe. The permissive mesenchymal signal for anterior lobe cell specification has not been identified. Several secreted factors expressed in mouse pituitary adjacent mesenchyme, including Chordin, Noggin, Nbl1, and Fstl1, are candidates for a mesenchymal signal in pituitary organogenesis [9][10][11]. Foxd1 is a forkhead domain transcription factor expressed in the pituitary adjacent mesenchyme, and Foxd1 −/− mice have increased proliferation of anterior lobe cells and decreased LHβ expression. These results suggest that Foxd1 may regulate the expression of a mesenchymal signal necessary for pituitary development [12].
The cranial mesenchyme contributes to the hypophyseal portal system, which is a network of blood vessels that surrounds and invades the pituitary gland, enabling delivery of releasing hormones from the hypothalamus to the pituitary anterior lobe and from the pituitary gland to target organs in the body. The head mesenchyme is comprised of both definitive mesoderm, migrating through the primitive streak, and neural crest, which migrates away from the dorsal side of the neural tube and throughout the body [4]. The neural crest forms much of the peripheral nervous system, which is consistent with its ectodermal origin, but within the head it contributes to tissues, such as bones, muscles, and dermis of the skin, which are usually derived from definitive mesoderm [13]. In recognition of the diversity of tissues generated by the neural crest, it is frequently described as a fourth germ layer [14]. The developmental plasticity of the neural crest results from the maintenance of a gene regulatory network, characteristic of pluripotent blastula cells, in the neural crest lineage [15]. The cranial vasculature reflects the dual contributions of definitive mesoderm and neural crest; endothelial cells are derived from definitive mesoderm and the neural crest forms the pericytes and smooth muscle that wrap around and regulate the endothelial cells [16][17][18]. To define the contributions of head mesenchyme to pituitary organogenesis, we utilized a genetic model to examine neural crest contributions to pituitary gland organogenesis. β-CATENIN is a key component of canonical Wnt signaling, and loss of βcatenin in the neural crest lineage is known to cause apoptosis of migrating neural crest cells, leading to severe head malformations [19]. We present here an analysis of pituitary gland organogenesis in the absence of neural crest, and uncovered a critical role for neural crest contributions to the pituitary vasculature.

Results
Tg(Wnt1-cre)11Rth (abbreviated here as Wnt1-cre) and C57BL/6 J-Tg(P0-Cre)94Imeg (P0-cre) are frequently used to fate map and generate tissue specific loss-of-function alleles in the neural crest [20,21]. To determine the contributions of the neural crest to head mesenchyme near the developing pituitary gland we crossed both the Wnt1cre and P0-cre lines with the reporter lines B6.Gt(RO-SA)26Sor tm4(ACTB-tdTomato,-EGFP)Luo (Rosa mT/mG ) and Gt (ROSA)26Sor tm1Sho (Rosa stopLacZ ) to determine neural crest contributions to the pituitary adjacent head mesenchyme [22,23]. At embryonic day of development 8.5 (e8.5) both Wnt1-cre and P0-cre mediate recombination in the migrating neural crest cells ( Fig. 1a and b). The Wnt1-cre also mediates recombination in the midbrain, a domain of cre activity that is not observed in the P0-cre. The midbrain activity of Wnt1-cre is also observed at e12.5, but it is not confined to the midbrain, as LacZ is expressed in the posterior diencephalon and in the anterior hindbrain ( Fig. 1c). At e12.5 the head mesenchyme rostral to Rathke's pouch expresses LacZ, indicating that it is derived from the neural crest ( Fig. 1d and e). The head mesenchyme caudal to Rathke's pouch is devoid of X-gal staining, suggesting that this tissue is derived from definitive mesoderm [7,24,25]. A sharp boundary between the neural crest derived head mesenchyme and the definitive mesenchyme is observed ventral to Rathke's pouch in midsagittal sections (Fig. 1d). Rathke's pouch is located at the rostral end of the notochord, and chick explants studies demonstrate that the notochord may be involved in Rathke's pouch formation [8,26,27]. At more lateral sections the boundary between neural crest and definitive mesoderm is not as sharp, as the definitive mesoderm extends past the lateral edge of Rathke's pouch (Fig. 1e). At e14.5 X-gal stained cells are detected in the forming pituitary anterior lobe of Wnt1-cre; Rosa stopLacZ embryos, suggesting that the neural crest derived mesenchyme is invading the oral ectoderm-derived Rathke's pouch tissue (Fig. 1f). By e18.5 Wnt1-cre expressing cells have made a significant contribution to the pituitary gland (Fig. 1g). Lineage tracing in P0-cre; Rosa stopLacZ embryos at e12.5 also demonstrates a sharp boundary between neural crest and definitive mesoderm at mid-sagittal locations (Fig. 1h). Although Wnt1-cre activity in the midbrain and neural crest and P0-cre activity in the neural crest mimic endogenous Wnt1 and P0 expression, we also observed ectopic activity in the ventral diencephalon for both cre lines ( Fig. 1d and h), and cre activity in Rathke's pouch for the P0-cre (Fig. 1h).
Previous work established that Wnt1-cre mediated deletion of β-catenin causes neural crest cells to undergo apoptosis during migration, which provides a method to generate neural crest deficient embryos [19]. Because pituitary development was not analyzed in these embryos, we crossed Wnt1-cre with B6.129-Ctnnb1 tm2Kem /KnwJ (referred to here as β-cat fx/fx ) and then generated Wnt1cre;β-cat fx/fx embryos to determine if the neural crest  head mesenchyme is important for pituitary organogenesis. Indeed, Wnt1-cre;β-cat fx/fx embryos do have abnormalities in pituitary development ( Fig. 2a-i). At early stages of Rathke's pouch formation (e10.5) a larger domain of oral ectoderm is recruited into Rathke's pouch, as marked by immunostaining for the critical LIM homeodomain transcription factor, LHX3 ( Fig. 2b and c). At e12.5 the enlarged, mutant Rathke's pouch includes more oral ectoderm tissue than typical, extending more rostrally than normal (Fig. 1e). The enlarged domain of presumptive pituitary tissue becomes highly dysmorphic (Fig. 2f ). The abnormal pituitary extends through the cartilage plate that underlies the pituitary gland at e18.5 and can form an anterior lobe like structure that projects into the oral cavity at more rostral locations ( Fig. 2h and i).
The number and distribution of hormone expressing cell types in the pituitary anterior lobe in both wild type and Wnt1cre; β-cat fx/fx embryos was examined by immunostaining for pituitary hormones at e18.5 ( Fig. 3a-h). No obvious difference in the quantity or distribution of anterior lobe cell types was detected between the wild type and mutant embryos, despite the highly dysmorphic, mutant pituitary gland.
The dysmorphic pituitary of Wnt1cre; β-cat fx/fx embryos resembles those of other genetically modified mouse embryos that have an expansion of the pituitary organizer in the ventral diencephalon [6]. The pituitary organizer is characterized in part by the expression of Fgf10 and Bmp4, which are necessary for the induction and proliferation of Rathke's pouch [28,29]. We examined Bmp4 and Fgf10 expression in wild type and Wnt1cre; β-cat fx/fx embryos by in situ hybridization and found that the mutant embryos have expanded expression domains of both morphogenetic proteins (Fig. 4ad). The expression domain of the transcription factor SIX6 typically exists in the ventral diencephalon rostral to the pituitary organizer. Six6 expression is shifted rostrally from Rathke's pouch in the mutant embryos ( Fig. 4ef ). These results demonstrate that the patterning within the ventral diencephalon is disrupted in Wnt1cre; β-cat fx/fx embryos. Excess BMP and FGF signaling from the pituitary organizer likely contributes to the recruitment of additional oral ectoderm into Rathke's pouch [9,[30][31][32].
Wnt1-cre is expected to drive recombination in multiple tissues where loss of β-catenin could affect pituitary organogenesis, including the neural crest and ventral diencephalon. We examined β-CATENIN expression by immunohistochemistry to determine the efficiency of cre recombination in these tissues ( Fig. 5a-f ). At e12.5 the Wnt1-cre; β-cat fx/fx embryos do not express β-CATENIN β-cat fx/fx e10.5 embryos d Wild type e12.5 e & f Two separate Wnt1-cre; β-cat fx/fx e12.5 embryos gi Hemotoxylin and eosin staining on e18.5 coronal sections g Wild type h & i Two sections from the same Wnt1-cre; β-cat fx/fx embryo, with I representing a more rostral location. All scale bars equal 100 μm in the residual neural crest derived mesenchyme rostral to Rathke's pouch, indicating that Wnt1-cre efficiently recombines the conditional β-catenin allele in this tissue ( Fig. 5d and e). Wnt1-cre is not uniformly active within the ventral diencephalon, and only small clumps of cells exhibit loss of β-CATENIN ( Fig. 5d and f ). This level of β-catenin ablation is unlikely to alter ventral diencephalon patterning. However, we may be underestimating the degree of β-catenin ablation as the protein persists in the adherens junctions for many hours following recombination at the β-catenin locus [33]. In addition, the presence of β-CATENIN protein in the adherens junctions does not maintain canonical Wnt signaling after cre mediated recombination [33]. Therefore, we cannot rule out the possibility that β-catenin ablation in the infundibular domain may result in an expansion of the pituitary organizer.
We sought to recapitulate the Wnt1-cre; β-cat fx/fx phenotype using the P0-cre, which is also active in the neural crest and has some ectopic activity in the ventral diencephalon, to confirm that loss of β-catenin in these domains causes the pituitary phenotype. P0-cre; β-cat fx/fx embryos have severe craniofacial malformations compared to wild type littermates consistent with a neural crest deficiency (Additional file 1: Figure S1), and similar in phenotype to Wnt1-cre; β-cat fx/fx embryos [19].   However, unlike Wnt1-cre; β-cat fx/fx embryos, the pituitary morphology of P0-cre; β-cat fx/fx mutant embryos is very similar to wild type despite the absence of neural crest derived cranial mesenchyme rostral to Rathke's pouch, as evidenced by the absence of the cartilage plate rostral to the pituitary gland ( Fig. 6a and b). Using immunostaining we determined that the P0-cre efficiently recombines the conditional β-catenin allele in the neural crest, but not in the ventral diencephalon at e12.5 ( Fig. 6g-i). Within Rathke's pouch, variable, small patches of β-CATENIN deficient cells are observed, but β-CATENIN is not completely eliminated (Fig. 6j). Thus, both Wnt1-cre and P0-cre are similarly effective in deleting β-catenin in the neural crest, but the P0-cre does not delete β-catenin in the ventral diencephalon. Therefore, the Wnt1-cre; β-cat fx/fx dysmorphic pituitary is not caused by the loss of neural crest cells.
The P0-cre can drive recombination in Rathke's pouch, and β-CATENIN can influence the activity of the transcription factors PROP1, PIT1 (POU1F1), PITX2, and SF1 (NR5A1) in the pouch and its derivatives [34][35][36][37]. Therefore, we sought to determine if the ectopic, mosaic loss of β-CATENIN in the P0-cre; β-cat fx/fx pouch could alter anterior lobe cell specification. No significant changes in hormone expressing cell types were observed in the mutant embryos ( Fig. 7ah), indicating that the mosaic loss of β-CATENIN in Rathke's pouch does not disrupt pituitary cell specification.
Wnt signaling in the ventral diencephalon is known to regulate the expression domain of the pituitary organizer [30,31]. While the mosaic loss of β-catenin in the ventral diencephalon observed with Wnt1-cre is unlikely to alter pituitary organizer activity, we sought to inactivate β-catenin in the ventral diencephalon to determine if canonical Wnt signaling mediates pituitary organizer expression. We crossed Shh tm1(EGFP/cre)Cjt (Shh-cre) with Rosa stopLacZ and stained embryo sections with X-gal. We detected staining in the ventral diencephalon and the oral ectoderm at e10.5 in a pattern that recapitulates the endogenous expression of Shh (Additional file 2: Figure  S2a) [38]. We next crossed the Shh-cre with β-cat fx/fx to determine if the loss of β-catenin in the ventral diencephalon would lead to an expanded pituitary organizer and an expansion of Rathke's pouch. At e12.5 the pituitaries of Shh-cre; β-cat fx/fx embryos were indistinguishable from wild type littermates (Additional file 2: Figure S2b-c). The lack of β-catenin in the Shh expressing domain of the fore and hind limbs leads to disruptions in digit formation, as expected (Additional file 2: Figure S2d-f ). Previous studies have demonstrated that β-catenin is necessary for maintaining limb mesenchyme and Fgf10 expression in the apical ectodermal ridge [39]. We used immunostaining to determine whether the Shh-cre effectively deleted β-CATENIN in the ventral diencephalon. Although there were patches of cells that were negative for β-CATENIN immunostaining, the majority of cells in the ventral diencephalon retained β-CATENIN expression (Additional file 2: Figure S2g-i). We concluded that unlike in the limb bud, the Shh-cre is unable to efficiently recombine the β-cat fx/fx allele in the ventral diencephalon, leaving canonical Wnt signaling largely intact. This result also confirms that the mosaic loss of β-catenin in the ventral diencephalon is insufficient for producing the dysmorphic pituitaries of Wnt1-cre; β-cat fx/fx embryos.
There are many precedents for variability in cre-mediated deletion, including parent-of-origin effects, episome formation, number of conditional alleles, and genetic background differences, any of which could affect the activity of the transgenic cre lines used in this study [40][41][42][43]. However, the three cre lines used here demonstrate that β-catenin deletion in the neural crest and mosaic β-catenin deletion in the ventral diencephalon do not generate a dysmorphic pituitary gland. The major difference in activity between the cre lines is the expression of Wnt1-cre in the midbrain.
We tested the possibility that loss of β-catenin in the midbrain results in patterning disruptions of the ventral diencephalon by examining additional markers of the posterior ventral diencephalon. Our fate mapping studies suggest that the Wnt1-cre is also active in the posterior diencephalon (Fig. 1c). Immunoflourescence for β-CATENIN on e11.5 wild type and Wnt1-cre; β-catenin fx/ fx embryos demonstrates that β-CATENIN expression is lost in many cells near the boundary of the midbrain and posterior diencephalon (Fig. 8a-d). Otx2 is expressed in the forebrain and midbrain and is necessary for forming the isthmic organizer at the midbrain/hindbrain boundary [44][45][46]. During Rathke's pouch induction, OTX2 is expressed in the infundibulum and more posterior regions of the ventral diencephalon, but is excluded from the neural ectoderm rostral to the infundibulum [47]. At e11.5 OTX2 is strongly expressed in the posterior diencephalon, but has a lower level of expression in the infundibulum (Fig. 1e and f ). In Wnt1cre; β-cat fx/fx embryos OTX2 is strongly expressed in the infundibulum and a domain of weaker OTX2 expression is observed rostral to the infundibulum (Fig. 8g and h). Lef1 is expressed in the premamillary region of the developing hypothalamus [48]. This domain of LEF1 expression is not contiguous with the infundibulum (Fig. 8i and j). In Wnt1-cre; β-cat fx/fx embryos the boundary of LEF1 expression is displaced rostrally and corresponds to the caudal edge of the infundibulum (Fig. 8k and l). These results demonstrate that the patterning of the entire posterior ventral diencephalon, including the pituitary organizer, is shifted rostrally when β-catenin is deleted by the Wnt1-cre. The pattern of LacZ expression in the e18.5 Wnt1-cre; Rosa stopLacZ pituitary suggests that the neural crest contributes to the pituitary vasculature (Fig. 1f ). Therefore, we examined the pituitary vasculature in both models of neural crest deficient embryos using immunostaining for CD31 (PECAM), an endothelial cell marker (Fig. 9a-h). The blood vessels in the pituitary glands of both the Wnt1-cre; β-cat fx/fx and P0-cre; β-cat fx/fx embryos are abnormally dilated at e18.5 (Fig. 9a-f ). Endothelial cells appear to be migrating into the forming pituitary anterior lobe at e14.5 (Fig. 9g). Neither this pattern nor the size of the forming endothelium is disrupted in P0-cre; β-cat fx/fx embryos, suggesting that the early migration of endothelial cells into the pituitary anterior lobe is unaffected by the loss of the neural crest (Fig. 9h). Pericytes are cells that wrap around the endothelium and regulate blood vessel diameter, and the pericytes of the head are neural crest-derived [17,18]. Pericytes express PDGFRβ at e18.5 in wild type mouse embryos (Fig. 9i). This pericyte marker is absent in P0-cre; β-cat fx/fx embryos, confirming that the loss of neural crest cells results in a loss of pituitary gland pericytes (Fig. 9j). PDGFβ −/− or PDGFRβ −/− mice develop dilated and leaky vasculature throughout the body that is similar to the vascular defects we observed in the pituitary glands of the P0-cre; β-cat fx/fx embryos [49][50][51].

Discussion
The developing pituitary gland is surrounded by head mesenchyme, which has a dual origin from definitive mesoderm and neural crest [4,24]. Using two different neural crest expressing transgenic cre lines, we have shown that at axial levels the head mesenchyme on the rostral side of Rathke's pouch is of neural crest origin. This result is in agreement with previous fate mapping experiments of the neural crest derived head mesenchyme [25]. The mesenchyme on the caudal side is derived from definitive mesoderm; the rostral end of the notochord and the prechordal mesoderm is located on the caudal side of Rathke's pouch [24,26,27]. At more lateral levels the boundary between neural crest and definitive mesoderm is still present, but the definitive mesoderm can migrate to the rostral side of Rathke's pouch. The separation between head mesenchyme of neural crest and definitive mesoderm origins on either side of the pituitary gland presents intriguing possibilities for differential effects of these mesenchymal populations on pituitary gland organogenesis. For instance, stimulation of Notch signaling in the ventral diencephalon reduces Fgf10 expression in the pituitary organizer, which reduces Lhx3 expression in Rathke's pouch and increases apoptosis [52]. Lhx3 expression is reduced preferentially on the caudal side of Rathke's pouch and apoptosis is preferentially increased [52]. The differential effects on Rathke's pouch may be caused by signals from the definitive mesoderm, which are not expressed in the neural crest derived mesoderm.
Mesenchyme is known to be important for pituitary induction [8,12], but the specific factors involved and source of mesenchyme were unexplored. To address the role of neural crest derived mesenchyme on pituitary organogenesis we examined mouse embryos deficient in neural crest. Intriguingly, ablation of neural crest using two different promoters to drive cre-mediated excision of β-catenin produced two different pituitary phenotypes. Wnt1-cre mediated deletion of β-catenin in the neural crest produced Rathke's pouch and pituitary anterior lobe dysmorphology, without an effect on cell specification, while P0-cre mediated deletion had no effect at all, despite equally effective neural crest cell ablation.
The dysmorphic pituitary caused by Wnt1-cre is likely attributable to expansion of the pituitary organizer, including excess Bmp4 and Fgf10 expression, leading to additional oral ectoderm being induced to form Rathke's pouch. This phenotype is similar to other models of pituitary organizer expansion that result in an enlarged Rathke's pouch [9,[30][31][32].
We considered three possibilities for how Wnt1-cre mediated inactivation of β-catenin could result in a dysmorphic pituitary, and tested these possibilities by using the Shh-cre and P0-cre. The data from each experiment is summarized in Table 1. The first possibility we considered is that the neural crest pattern the ventral diencephalon, including establishing the rostral boundary of the pituitary organizer, and that neural crest deficient embryos result in a pituitary organizer boundary shift. The neural crest regulate patterning of the forebrain and midbrain [53][54][55], and could potentially affect patterning in the ventral diencephalon. However, P0-cre, which effectively ablates β-catenin in the neural crest, does not induce expansion of Rathke's pouch. Therefore, the loss of neural crest cells is unlikely to cause the expansion of the pituitary organizer.
The second possibility is that modest, ectopic deletion of β-CATENIN in the ventral diencephalon causes expansion of the pituitary organizer. TCF7L2 (TCF4) is a transcription factor that is activated by β-CATENIN, and Tcf7l2 −/− and Wnt5a −/− embryos both have an expansion of the pituitary organizer that leads to a dysmorphic pituitary [30,31,56,57]. These precedents support the idea that the mosaic deletion of β-catenin in a few patches within the ventral diencephalon could be recapitulating the Wnt5a and Tcf7l2 phenotypes. However, Wnt1-cre and Shh-cre mediate similar mosaic, ectopic β-CATENIN deletion in the ventral diencephalon, but only Wnt1-cre causes a phenotype. In addition, β-catenin is only ablated in a few cells of this tissue by either cre strain. Taken together, these facts argue against the possibility that the pituitary phenotype is caused by deletion of β-CATENIN in the ventral diencephalon.
The third possibility is that Wnt1-cre mediated deletion of β-catenin in the midbrain results in a rostral shift of the pituitary organizer. The major difference in (See figure on previous page.) Fig. 8 ad Immunostaining for β-CATENIN (green) on e11.5 sagittal sections, counterstained with DAPI (blue). a and b Wild type embryo. Arrow in a indicates the level of the infundibulum. Boxed region is magnified in b. c and d Wnt1-cre; β-cat fx/fx embryo. Arrow in c indicates the level of the infundibulum. Boxed region is magnified in d. Arrow in d indicates cells at the posterior diencephalon and midbrain boarder deficient for β-CATENIN. eh Immunostaining for OTX2 (green) on e11.5 sagittal sections, counterstained with DAPI (blue). e and f Wild type embryo. Arrow in e indicates the level of the infundibulum. Boxed region is magnified in f. Arrow in f indicates a low level of OTX2 expression. g and h Wnt1-cre; β-cat fx/fx embryo. Arrow in g indicates the level of the infundibulum. Boxed region is magnified in h. Arrow in h indicates a low level of OTX2 expression. il Immunostaining for LEF1 (green) on e11.5 sagittal sections, counterstained with DAPI (blue). (i and j) Wild type embryo. Arrow in i indicates the level of the infundibulum. Boxed region is magnified in j. Arrow in j indicates the boundary of LEF1 expression. k and L Wnt1-cre; β-cat fx/fx embryo. Arrow in k indicates the level of the infundibulum. Boxed region is magnified in l. Arrow in L indicates the boundary of LEF1 expression. All scale bars equal 100 μm Fig. 9 ah Immunostaining for PECAM (green), counterstained with DAPI (blue). a Wild type e17.5 cryosection, coronal orientation b Wnt1-cre; β-cat fx/fx e17.5 cryosection, coronal orientation c & e Wild type e18.5 paraffin section, coronal orientation. Boxed area in c is magnified in e. d & f P0-cre; β-cat fx/fx e18.5 paraffin section, coronal orientation. Boxed area in d is magnified in f. Blood cells are autofluorescent (pink). g Wild type e14.5 paraffin section, sagittal orientation. h P0-cre; β-cat fx/fx e14.5 paraffin section, sagittal orientation. i & j Immunostaining for PDGFRβ (green), counterstained with DAPI (blue) on e17.5 paraffin, coronal sections. i Wild type j P0-cre; β-cat fx/fx , blood cells are autofluorescent (pink). All scale bars equal 100 μm expression between the Wnt1-cre and the P0-cre is the expression of the Wnt1-cre in the midbrain. Our Wnt1-cre fate mapping studies demonstrate that the Wnt1-cre is also active in the posterior diencephalon. When β-catenin is deleted in the midbrain and posterior diencephalon the patterning of the entire ventral diencephalon is shifted rostrally, including the pituitary organizer, resulting in a dysmorphic Rathke's pouch. Previous studies have demonstrated that loss of Wnt1 or β-catenin in the mesencephalon causes a loss of midbrain identity and expansion of the hindbrain [19,58,59]. Wnt1 plays a critical role in establishing the isthmic organizer at the boundary between the between the midbrain and hindbrain [60]. Initial characterization of Wnt1 −/− embryos did not identify a disruption in forebrain patterning. Examination of images in a previous characterization of Wnt1 −/− embryos [59], suggests that Rathke's pouch is expanded in these embryos. Our data indicates that canonical WNT signaling is not only necessary for midbrain-hindbrain boundary formation, but also proper patterning of the ventral diencephalon.
In comparison to the ectodermal structures of the pituitary gland, very little is known about the contributions of the head mesenchyme to pituitary development. Chick explant studies demonstrate a permissive role of mesenchyme in pituitary cell specification, and a potential inductive role of the notochord in Rathke's pouch formation [8]. A mesenchymal signaling component is also supported by genetic studies in mice, where loss of Foxd1 in the mesenchyme results in an increase in pituitary anterior lobe cell number and a decrease in LHβ expression [12]. It is not known what the actual mesodermal signal is that regulates pituitary development. The BMP antagonist CHORDIN is expressed in the prechordal mesoderm of mouse embryos, adjacent to the caudal side of Rathke's pouch [11]. The prechordal plate also expresses NOGGIN, and mice that are Chrd −/− ; Nog +/− develop holoprosencephaly, indicative of midline patterning defects [61]. The Chrd −/− ; Nog +/− embryos do not express Nkx2.1 in the ventral diencephalon, which leads to a loss of Fgf8 expression in the pituitary organizer and loss of Rathke's pouch [61]. These results demonstrate that the prechordal plate is necessary for establishing the pituitary organizer in the ventral diencephalon, which then induces Rathke's pouch. Direct effects of prechordal plate signaling on development of the pituitary anterior lobe are unknown.
To begin determining what functions the head mesenchyme might play in pituitary organogenesis, we fate mapped the neural crest derivatives near the pituitary gland. Our results unequivocally demonstrate that the neural crest produce the head mesenchyme on the rostral side of Rathke's pouch, and form a sharp boundary with definitive mesoderm on the caudal side of Rathke's pouch. While this boundary has been observed surrounding the eye, our results refine the position of the boundary in relation to the pituitary gland with a sharp demarcation occurring at the site of Rathke's pouch invagination [24]. It is intriguing to speculate that Rathke's pouch may provide guidance cues to neural crest cell migration. Characterization of the embryonic pituitary transcriptome identified 74 genes with gene ontology (GO) terms related to cell adhesion and migration, including genes with known roles in neural crest migration [62]. Analysis of these candidate genes may reveal differential expression within Rathke's pouch necessary for directing migrating neural crest cells to the rostral side.
A critical aspect of pituitary development is the delamination of cells from the tightly packed epithelium at the ventral side of Rathke's pouch, in a rostral direction, as they take on the shape typical of glandular cells, and form the rudimentary anterior lobe. This commences at approximately e12.5 and continues through gestation. Some neural crest derived mesenchyme cells invade the forming anterior lobe at e14.5. This is the same time that vascular invasion is initiated [63][64][65]. Indeed, at later stages of pituitary development the neural crest derived mesenchyme exhibits a pattern reminiscent of vascular tissue. In the vertebrate head, pericytes, regulatory cells which wrap around endothelial cells, are derived from the neural crest [17,18]. The neural crest cells invading the anterior lobe are likely forming the pericytes of the pituitary gland. The vasculature of both Wnt1-cre; β-cat fx/fx and P0-cre; β-cat fx/fx embryos is dysmorphic and dilated, which is attributable to the lack of regulatory pericytes, as marked by PDGFRβ. Dilated and leaky vasculature results when pericytes are not recruited to the endothelial cells [49,50]. The vasculature of the Wnt1-cre; β-cat fx/fx and P0-cre; β-cat fx/fx pituitaries recapitulates this phenotype and demonstrates a requirement for neural crest derived pericytes in pituitary vasculature formation.
Neither the Wnt1-cre; β-cat fx/fx or P0-cre; β-cat fx/fx embryos display a disruption in pituitary anterior lobe cell specification. Therefore, the neural crest derived head mesenchyme is not necessary for normal anterior lobe cell specification. The axial mesoderm on the caudal side of Rathke's pouch is a rich source for developmental signals, which are known to pattern surrounding tissues, including the neural tube and paraxial mesoderm. Rathke's pouch sits at a dynamic site at the interface of definitive mesoderm, neural crest, neural ectoderm, and oral ectoderm. Determining the precise molecular mechanisms that each of these tissues plays in pituitary organogenesis in mouse and other vertebrate model systems will help in determining how this vertebrate specific organ may have evolved.

Conclusions
Using both the Wnt1-cre and the P0-cre to lineage trace the neural crest, we determined that the mesenchyme on the rostral side of Rathke's pouch is neural crest in origin. This rostral mesenchyme contributes to the vasculature of the pituitary gland. Deletion of a conditional null allele of β-catenin in the neural crest lineage using both the Wnt1-cre and P0-cre generates mouse embryos that are deficient in neural crest cells [19]. These embryos have dysmorphic pituitary blood vessels and lack PDGFRβ labeled pericytes. The Wnt1-cre is also active in the mesencephalon, while the P0-cre is not. Loss of βcatenin in the mesencephalon causes a loss of midbrain structures [19]. In addition, we have determined that loss of β-catenin in the mesencephalon and posterior diencephalon is the likely cause of a rostral shift in the expression domain of Bmp4 and Fgf10 in the ventral diencephalon. The altered expression domain of these morphogenetic proteins results in a highly dysmorphic Rathke's pouch. Therefore, β-catenin is required in the Wnt1 expression domain, including the neural crest, for the proper specification of pituitary gland growth, development, and vascularization.

Mice
The Institutional Committee on the Use and Care of Animals for the University of Michigan and the University of South Carolina approved all experiments using mice (protocol number PRO00004640 at the University of Michigan to SAC and protocol number 2106-100665-012213 at the University of South Carolina to SWD). The Wnt1-cre, Shh -cre, β-cat fx/fx , Rosa stopLacZ , and Rosa mT/mG mice were obtained from Jackson Laboratory [19][20][21][22][23].
The P0-cre mice were obtained from Dr. Ken-ichi Yamamura [21]. Mice were housed in specific pathogen free conditions with automatic watering, ventilated cages, and fed ad libitum. Genotyping of mice was performed as previously reported [19][20][21][22][23]. Observation of copulation plugs in female mice was used to detect mating. Noon of the day the copulation plug was detected was established as e0.5.

Lineage tracing
β-Galactosidase activity from the Rosa stopLacZ allele was detected with X-gal staining in cryosections and embryo whole mounts. For cryosections both cre positive and cre negative Rosa stopLacz embryos were fixed in 4 % paraformaldehyde in LacZ fix for 15 min on ice, washed with LacZ wash, equilibrated in 30 % sucrose, and embedded in O.C.T. Compound (Tissue-Tek) in a dry ice/ethanol bath [66]. Frozen sections were cut on a cryostat at 16 μm, stained with X-gal, and counterstained with Neutral Red [66]. The skin and skull were removed from cre positive and cre negative Rosa stopLacz embryo whole mounts to allow penetration of X-gal stain. Tissues were fixed for 30 min in 4 % paraformaldehyde in LacZ fix at 4 C, washed three times for 30 min in LacZ wash, and stained with X-gal stain overnight at 37 C. The pituitary was isolated following X-gal staining and photographed on a Leica M125 microscope with a color camera. Fluorescent emission from tdTomato and eGFP was visualized using cre positive and cre negative Rosa mT/mG embryos on a Leica M125 microscope with fluorescence capabilities and photographed using a Retiga 2000R digital camera.

Ethics approval
The Institutional Committee on the Use and Care of Animals for the University of Michigan and the University of South Carolina approved all experiments using mice (protocol number PRO00004640 at the University of Michigan to SAC and protocol number 2106-100665-012213 at the University of South Carolina to SWD).

Consent for publication
Not applicable.

Availability of data and material
The data supporting the results of this manuscript are included in the body of the manuscript and as supplemental data.