Lack of the ventral anterior homeodomain transcription factor VAX1 leads to induction of a second pituitary

The pituitary is a crucial hormone release site that is derived from an evagination of the oral ectoderm, termed Rathke's pouch, and a part of the diencephalic neuroectoderm, termed infundibulum. The developmental interactions between these tissues are regulated by many factors, including transcription factors such as Pitx1 and Pitx2 and their target, Lhx3, which are specifically expressed in the oral ectoderm/Rathke's pouch; others, such as Six3/6, Sox2, Hesx1 and Pax6, which are found in both oral and neuroectoderm; and growth factors such as fibroblast growth factor 8 and 10 (FGF8, FGF10), bone morphogenic proteins 2 and 4 (BMP2/4), sonic hedgehog (SHH), NOTCH, and WNT, which are expressed in the neural or oral ectoderm, or in both tissues (Sheng et al., 1996; Treier et al., 1998; Suh et al., 2002; Charles et al., 2005; Zhu et al., 2007; Davis et al., 2009; Kelberman et al., 2009). Not surprisingly, transcriptional and signaling pathways intersect during pituitary development. For example, loss of TCF7L2, a high-mobility group transcription factor that serves as mediator of canonical WNT signaling, leads to a rostral expansion of FGF and BMP activities in the neuroectoderm, a rostral expansion of Six6 expression in the oral ectoderm, and an increased recruitment of cells into the pituitary (Brinkmeier et al., 2007).

Here, we demonstrate a novel role for the ventral homeodomain protein VAX1 in pituitary development. In the mouse, Vax1 is initially expressed in most of the ventral neuroectoderm and then is found in the telencephalon, the ventral diencephalon and the optic stalk (Hallonet et al., 1998; Bertuzzi et al., 1999). Mice with mutations in Vax1 show prominent perturbations in the development of the optic chiasm and die at birth due to cleft palate (Bertuzzi et al., 1999). As shown here, they also display a second, more rostrally located, Rathke's pouch. This second pouch is completely separated from the primary pouch, matures into an ectopic adenohypophysis containing all pertinent cell lineages and exhibits posterior pituitary markers, indicating the presence of an ectopic neurohypophysis. Given that Vax1 is not expressed in the oral ectoderm or the pituitary primordium, it probably exerts its effects indirectly. In fact, our results show that the major role of Vax1 in pituitary development is to limit the zone in which FGF10 can act as a pituitary-inducing signal, thereby assuring that a single, correctly positioned pituitary is formed.

The Vax1 knockout allele, Vax1^tm1Grl, here referred to as Vax1⁻, has been described previously (Bertuzzi et al., 1999). The allele was kept on a mixed C57BL/6;129S1/Sv background.

All histological analyses were performed according to Bharti et al. (Bharti et al., 2008). Antibodies are shown in Table 1. The following in situ probes were used: Vax1 (Bertuzzi et al., 1999); Pit1, Pomc, Lhx3 (Sheng et al., 1996); Six3 (Oliver et al., 1995); Id3 (Jones et al., 2006); Six6, Pitx2, Tcf7L2, Fgf10, Bmp4, Gli1, Shh and Ptch1 (all from Open Biosystems, Huntsville, Al). Tcf7L2 encodes a high-mobility-group transcription factor that is also called Tcf4 (T-cell specific factor 4; MGI:1202879), but the symbol Tcf4 is officially assigned to another gene (transcription factor 4; MGI:98506), which encodes a bHLH protein.

For ChIP assays, ventral forebrain tissue from six to eight embryos (E10.5-E11, wild-type or Vax1^−/−) was processed as described previously (Skuntz et al., 2009), using a rabbit anti-VAX1 antiserum (see Table 1) for immunoprecipitation. For sequences of primers used for ChIP assays and reporter construct cloning, see Table S1 in the supplementary material. Reporter assays were performed as described previously (Bharti et al., 2008) (see legend to Fig. S4 in the supplementary material).

As shown previously, in the mouse, Vax1 expression starts at embryonic day (E) 8 (Hallonet et al., 1998). At both E11.5 and E13.5, Vax1 is expressed in the diencephalic neuroepithelium but excluded from the infundibulum and the oral ectoderm or Rathke's pouch (Fig. 1A,B).

While analyzing Vax1 mutant brain sections at E13.5, we noted the presence of an ectopic pouch-like structure underneath the diencephalic neuroectoderm rostral to Rathke's pouch (Fig. 1C,D; arrow in Fig. 1D). Interestingly, this pouch-like structure expressed the pituitary progenitor markers Lhx3 and Pitx2 (Fig. 1E-H; arrows in F,H), as did the normal Rathke's pouch in either wild type or Vax1 mutants. It also expressed pro-opiomelanocortin (Pomc), which normally marks corticotropes, one of the first anterior pituitary cell types to differentiate at E13.5 (Japon et al., 1994; Pulichino et al., 2003) (Fig. 1I,J; arrow in J). These results suggest that the loss of Vax1 causes the induction of a second, more rostrally located Rathke's pouch.

To test whether the second pouch would mature into an ectopic adenohypophysis and eventually lead to the formation of a full second pituitary, we performed additional histological and gene expression studies at later stages of development. With increasing developmental time, the second pouch grew to a larger structure that in some embryos showed a bipartite lumen (Fig. 2B, arrow and arrowhead). It continued to express Lhx3 and Pitx2, as well as Six3 (Fig. 2C-F, see Fig. S1A,B in the supplementary material). Pit1, a regulator of somatotropes, thyrotropes and lactotropes, that normally starts to be expressed at E13.5 (Dasen et al., 1999; Nica et al., 2004), was also found in the second pouch at E13.5 (not shown) and continued to be expressed at this site at E17.5 (Fig. 2G,H). The expression of Pomc continued through E15.5 (see Fig. S1C,D in the supplementary material) and beyond (not shown), and one of its products, adreno-corticotropic hormone (ACTH), started to be expressed at E13.5 (not shown) and continued through E15.5 and E17.5 (see Fig. S1E-G in the supplementary material; Fig. 2I,J). Moreover, additional markers were expressed in the second pouch in a temporal sequence reminiscent of that found in the primary pouch (Japon et al., 1994; Pulichino et al., 2003). They included thyroid-stimulating hormone (TSH), starting at E15.5 (not shown) and continuing through E17.5; growth hormone (GH) at E17.5; and luteinizing hormone (LH) at E17.5 (see Fig. S1H-L in the supplementary material; Fig. 2K,L). Interestingly, the temporal similarities in gene expression extended to the anatomical locations of the respective markers. For example, ACTH-positive cells, which are normally found at the periphery of the primary pouch, were also located at the periphery of the second pouch (Fig. 2J). Corresponding anatomical locations were also found for the centrally located GH- and TSH-positive cells (Fig. 2L, see Fig. S1I in the supplementary material) and the ventrally located LH-positive cells (see Fig. S1L in the supplementary material). Importantly, the primary pouch in Vax1 mutants, though slightly dysmorphic, displayed temporal and spatial gene expression profiles and histological characteristics that are comparable with those in wild type.

To test whether the formation of an ectopic anterior pituitary was followed by development of a corresponding posterior pituitary, we then examined the expression of the carrier proteins neurophysin (NP) I and II and the hormone vasopressin (VP). In wild type, these markers label neurons extending caudally from the hypothalamus to the posterior pituitary lobe (Zhu et al., 2007). In Vax1 mutants, we found these markers to extend caudally towards the primary pituitary and rostrally towards the second pituitary (Fig. 2M-T; see Fig. S1M,N in the supplementary material). The results suggest that in the absence of Vax1, the ectopically formed anterior pituitary is associated with the development of a second posterior pituitary.

The above results show that a second pituitary develops in Vax1 mutants but they do not address the initial steps, and hence the underlying molecular mechanisms, that lead to this ectopic pituitary. In fact, the first sign of an abnormality was seen at E10.5, when we observed a slight rostral shift of the rostral expression boundaries of Lhx3 and Pitx2 (Fig. 3A-D). The first anatomical alteration was not seen before E11.5, when the rostral oral ectoderm showed a thickening (Fig. 3E,F, compare insets showing magnifications) that now ectopically expressed the above two markers (Fig. 3G-J) as well as an additional Rathke's pouch marker, Six6 (Fig. 3K,L). Furthermore, two targets and mediators of SHH signaling, Gli1 (Fig. 3M,N) and Ptch1 (see Fig. S2A,B in the supplementary material) (Treier et al., 2001), were also expressed at the ectopic rostral location, suggesting that SHH signaling was intact at this site. Moreover, Shh, which normally is excluded from the invaginating primary pouch, was also excluded from the second pouch (arrows in Fig. S2D,F in the supplementary material), but was present in the remainder of the rostral oral ectoderm (arrowheads in Fig. S2D,F in the supplementary material). The oral ectodermal thickening was probably due to increased cell proliferation as the number of phosphohistone H3-positive cells was increased in the oral ectoderm, but not in the primary Rathke's pouch of the mutants (Fig. 3O,P; oral ectoderm marked with white lines, rostral areas magnified in insets, Rathke's pouch marked with white stippled lines; see Fig. 3Q for quantitation). The increase in the number of phosphohistone H3-positive cells was consistent with an increase in the total number of oral ectodermal nuclei counted after DAPI staining (289±53 in wild type compared with 369±83 in the mutant; average of cell counts from three mid-sagittal sections of three different embryos each). Because no changes were found by TUNEL labeling (not shown), the oral ectodermal thickening was unlikely to be due to reduced rates in cell death. Furthermore, serial frontal sections labeled for Lhx3 and Pitx2 showed a label-free area of at least 160 μm between the labeled caudal and rostral sections, indicating a total separation of the two pouches even in the parasagittal area (see Fig. S2G-L in the supplementary material). The above observations confirmed that the formation of the second pouch was initiated from a separate region of the rostral oral ectoderm.

It has been observed previously that embryos lacking Tcf7L2 show a rostral expansion of the expression of Bmp4 and Fgf10, both of which encoding pituitary inducing signals (Brinkmeier et al., 2007). Tcf7L2 expression, however, was not changed in the neuroectoderm of Vax1 mutants (Fig. 4A,B, arrowheads). Consistent with this finding, the expression of Bmp4 and Fgf10 remained largely unchanged in the caudal neuroectoderm, as did that of Id3, a Bmp4 target, and phosphoSMAD (pSMAD), which serve as BMP effectors (see Fig. S3A-H in the supplementary material; Fig. 4C,D). Furthermore, Bmp4 was also not induced in the neuroectoderm at the site of the second pouch (arrowhead in Fig. S3A,B in the supplementary material; Fig. 4C,D). Consistent with this observation, Id3 and pSMAD were also not specifically expressed in the neuroectoderm or the underlying oral ectoderm at this site (arrowheads and arrows in Fig. S3C-F in the supplementary material), in contrast to Fgf10, which was strongly expressed at this location after E11.5 (see Fig. S3G,H in the supplementary material; Fig. 4E-H, arrowheads). At this time point, strong expression of the FGF effector proteins pERK1/2 extended caudally and rostrally in the neuroectoderm and the underlying oral ectoderm (Fig. 4I,J, arrows pointing to the oral ectoderm). Bmp4, however, was only seen at the site of the second pouch (see Fig. S3I,J in the supplementary material) after the Fgf10 signal became prominent and the pouch was already induced. This suggests that Bmp4 was not needed for inducing the second pouch, although it may be involved in its later development.

The above observations suggested that, in the neuroectoderm, Vax1 might regulate Fgf10 directly rather than through Tcf7L2. To analyze whether VAX1 protein binds putative regulatory elements in the Fgf10 promoter, we performed chromatin immunoprecipitation (ChIP) assays using a VAX1 antiserum and chromatin extracts prepared from mouse E10.5 ventral forebrains, a stage at which the absence of Vax1 does not yet cause major alterations in gene expression or anatomy. Sequences within ~5 kb of the Fgf10 upstream region, containing potential conserved homeodomain-binding sites, were amplified (Odenwald et al., 2005; Mui et al., 2005) (see Fig. S4A,B in the supplementary material). As shown in Fig. 4K, ChIP assay showed amplification only of amplicons that contained highly conserved consensus homeodomain binding sites (amplicons ‘b’ and ‘e’ containing at least one ATTAA/TTAAT sequence), and not of amplicons that contained only the core part (ATTA,TAAT) of this sequence (amplicons ‘a’, ‘c’, ‘d’) (see Fig. S4A,B in the supplementary material). The fact that extracts from Vax1 mutant embryos gave no signals indicates that the ChIP assays were specific. To confirm that these sites were functional, we co-transfected luciferase reporter constructs containing Fgf10 promoter fragments with or without 131 bp representing amplicon ‘e’ (HD6-9 in Fig. S4B in the supplementary material) into HEK293 cells. These cells were chosen because they share neuroepithelial/neuronal characteristics (Shaw et al., 2002). As shown in Fig. S4C in the supplementary material, the HD6-9 deletion construct gave a higher signal compared with that containing HD6-9, as it did in a fibroblastic (NIH3T3) and a retinal pigment epithelial (ARPE19) cell line (not shown). This indicated that this region has repressive functions. As expected, Vax1 co-transfection repressed the full-length construct but interestingly also the deletion construct, conceivably because this deletion construct still contained a consensus HD-binding site (see Fig. S4A,B in the supplementary material).

As schematically depicted in Fig. 4L, our findings lead to a model of pituitary biogenesis in which Vax1 plays an important role as a repressor of Fgf10 expression in the neuroectoderm rostral to the infundibulum, thereby restricting its expression to the infundibulum. When Vax1 is absent, Fgf10 is prominently expressed at the site of second pouch induction. Nevertheless, this Fgf10 expression is not contiguous between primary and second pouch, conceivably because of the continuous expression of Tcf7L2, the absence of which, as mentioned, leads to an expansion of Fgf10 expression (Brinkmeier et al., 2007). This interpretation is consistent with the fact that expression of constitutively active β-catenin, which is known to act in conjunction with Tcf factors, suppresses the Fgf10 promoter (not shown). Finally, it is worth mentioning that the phenotype of the Vax1 mutation in mice shares similarities with septo-optic dysplasia in humans, which is characterized by malformed optic discs, a missing septum pellucidum and pituitary defects (Mutlu et al., 2004; Kelberman and Dattani, 2006; Mazzanti et al., 2009; Kelberman et al., 2009). This disorder is genetically heterogeneous and in a subset of individuals has previously been associated with mutations in HESX1. Although Hesx1 expression was unchanged in Vax1 mutant mice (not shown), the human homolog VAX1 remains a candidate gene for pituitary duplications in humans in which no HESX1 mutations have been found.

We thank the National Hormone Pituitary Program, NIDDK, NIH, for antibodies; Drs Heiner Westphal, Peter Gruss, and Mathew Kelley for in situ probes; Vinish Saini for Fig. 4L art; and Drs Susan Wray and Harold Gainer for helpful comments on the manuscript. This work was supported by the Intramural Research Program of the NIH, NINDS. Deposited in PMC for release after 12 months.