Genetic interaction between the homeobox transcription factors HESX1 and SIX3 is required for normal pituitary development

Abstract

Hesx1 has been shown to be essential for normal pituitary development. The homeobox gene Six3 is expressed in the developing pituitary gland during mouse development but its function in this tissue has been precluded by the fact that in the Six3-deficient embryos the pituitary gland is not induced. To gain insights into the function of Six3 during pituitary development we have generated Six3^+/−; Hesx1^Cre/+ double heterozygous mice. Strikingly, these mice show marked dwarfism, which is first detectable around weaning, and die by the 5^th-6^th week of age. Thyroid and gonad development is also impaired in these animals. Analysis of Six3^+/−; Hesx1^Cre/+ compound embryos indicates that hypopituitarism is the likely cause of these defects since pituitary development is severely impaired in these mutants. Similar to the Hesx1-deficient embryos, Rathke’s pouch is initially expanded in Six3^+/−; Hesx1^Cre/+ compound embryos due to an increase in cell proliferation. Subsequently, the anterior pituitary gland appears bifurcated, dysmorphic and occasionally ectopically misplaced in the nasopharyngeal cavity, but cell differentiation is unaffected. Our research has revealed a role for Six3 in normal pituitary development, which has likely been conserved during evolution as SIX3 is also expressed in the pituitary gland of the human embryo.

Introduction

The pituitary gland is considered a master regulator of homeostasis in vertebrates as it regulates a multitude of vital physiological functions such as metabolism, growth, fertility and stress response. The mature pituitary gland is composed of three lobes; the anterior and intermediate lobes (anterior pituitary, AP) and the posterior lobe (posterior pituitary, PP), which arise from different embryonic tissues (Cushman and Camper, 2001; Zhu et al., 2007). The anterior pituitary forms early in mouse development at 8.5 days post coitum (dpc) through a thickening of the roof of the oral ectoderm (pituitary placode) that contacts the floor of the ventral diencephalon. By 9.5 dpc, this thickened oral ectoderm tissue evaginates and forms Rathke’s pouch, the primordium of the AP. The mature AP houses six hormone-producing cell types: corticotropes, thyrotropes, somatotropes, lactotropes, gonadotropes and melanotropes, which produce adrenocorticotrophic hormone (ACTH), thyroid-stimulating hormone (TSH), growth hormone (GH), prolactin (PRL), luteinising hormone (LH) as well as follicle stimulating hormone (FSH), and melanocyte stimulating hormone (MSH), respectively. Once secreted into the blood stream, these hormones act upon their target organs (adrenal glands, thyroid, bone, mammary glands, gonads and skin) to control essential body functions. On the other hand, the PP is of neural origin and arises from the infundibulum, a ventral evagination of the floor of the diencephalon, which establishes intimate contact with Rathke’s pouch that is essential for normal development and function of the AP. The PP is devoid of hormone producing cells, but instead contains the axonal projections from the hypothalamic nuclei, which produce the hormones oxytosin (OT) and vasopressin (ADH). Unlike the AP, the PP is highly innervated, but weakly vascularised. Failure to produce one (usually Isolated Growth Hormone Deficiency, IGHD) or several of these hormones (Combined Pituitary Hormone Deficiency, CPHD) causes hypopituitarism in mice and humans (Procter et al., 1998; Kelberman and Dattani, 2007).

HESX1 is a paired-like homeobox transcription factor that is essential for normal forebrain and pituitary formation in both mice and humans. Hesx1 expression is very dynamic and transcripts can be detected transiently within the anterior forebrain from 7.5-8.5 dpc, and in Rathke’s pouch from 8.5-13.5 dpc (Thomas and Beddington, 1996; Hermesz et al., 1996). Hesx1^−/− embryos exhibit variable degrees of anterior forebrain defects, brought about by a cell fate transformation of anterior to posterior forebrain, possibly as a consequence of the ectopic activation of the Wnt/β-catenin signalling pathway in the anterior forebrain (Dattani et al., 1998; Martinez-Barbera et al., 2000; Andoniadou et al., 2007). Hesx1^−/− mutants develop dysmorphic pituitaries, which are enlarged, bifurcated, and often misplaced in the nasopharyngeal cavity. In spite of the very dramatic morphological defects, cell lineage determination and terminal hormone-producing cell type differentiation appears normal (Dasen et al., 2001; E. Sajedi and J.P. Martinez-Barbera, unpublished). Hesx1^−/− mice are not viable and die perinatally (Dattani et al., 1998; Andoniadou et al., 2007). Mutations in HESX1 have been associated with human phenotypes affecting midline telencephalic commissural tracts (corpus callosum and anterior commissure), the optic nerves and the pituitary gland, such as septo-optic dysplasia (SOD) and CPHD (Kelberman and Dattani, 2007).

SIX3 is a homeobox transcription factor related to the Drosophila sine oculis gene (Oliver et al., 1995). Six3 and Hesx1 expression domains partly overlap during early mouse development, in particular within the anterior forebrain and Rathke’s pouch. However, Six3 expression persists for longer in these tissues compared with Hesx1. Targeted disruption of the Six3 locus has demonstrated the requirement for Six3 in normal forebrain formation (Lagutin et al., 2003). In fact, Six3^−/− embryos show forebrain defects, which are comparable to those observed in Hesx1^−/− mutants, although they are more severe and fully penetrant. As in the Hesx1-deficient embryos, there is a posterior transformation of the anterior forebrain, which is likely to be due to the activation of Wnt/β-catenin signalling in these cells. Mutations in human SIX3 have been linked to holoprosencephaly (Wallis et al., 1999). So far, it has not been possible to determine a role for Six3 during pituitary development as the forebrain defects extend to the region of the ventral diencephalon responsible for the initial induction of Rathke’s pouch.

The similarities between the Hesx1 and Six3 expression patterns, and the common molecular mechanism underlying the forebrain defects of targeted mutant mice for these genes, prompted us to investigate the possibility that they might interact during normal anterior forebrain and Rathke’s pouch development. Interestingly, we found that Six3^+/−;Hesx1^Cre/+ compound heterozygous animals exhibit severe growth retardation after weaning, with additional gonadal and thyroid gland defects, resulting in a lethal phenotype. We present evidence that whilst the differentiation programme in the pituitary gland occurs correctly, cell proliferation of AP progenitors is enhanced at 12.5 dpc, which leads to hypertrophy and dysmorphogenesis of the AP in Six3^+/−;Hesx1^Cre/+ compound embryos. Molecular analysis suggests that this enhanced proliferation may be due to an up-regulation of Wnt/β-catenin signalling in the pituitary gland in these embryos. Our results demonstrate a novel role for Six3 in normal pituitary morphogenesis.

Materials and Methods

Mice

Hesx1^Cre/+ mice carry a null allele in which the entire Hesx1 open reading frame is replaced by the Cre recombinase coding sequence. Hesx1^Cre/Cre embryos show the same forebrain and pituitary defects to those observed in Hesx1^−/− embryos previously described (Dattani et al., 1998; Martinez-Barbera et al., 2000; Andoniadou et al., 2007). Hesx1^Cre/+ mice were backcrossed to the C57BL6/J background for approximately 6-7 generations. Six3^+/− mice, originating on a CD1 background, were kindly provided by Dr. G. Oliver (Memphis, USA) (Lagutin et al., 2003). We crossed Six3^+/− founders with Hesx1^Cre/+ to generate Six3^+/−; Hesx1^Cre/+ mice. The growth study (Fig. 1) used animals from this first filial generation (F1). Animals were weighed weekly from weaning (3^rd week) for four weeks to generate a growth curve for each genotype. Subsequently, the colony was expanded using Six3^+/− stud males from F1, which were crossed to C57BL6/J females. Therefore, although still on a mixed background, these Six3^+/− animals incorporated more C57BL6/J genetic background in subsequent generations. For the embryological studies, animals from the first 2-4 generations were used. It is worth mentioning that we have observed an increase in the severity of the phenotype in subsequent generations, i.e. a proportion of double mutants die before weaning. Therefore, as in the Hesx1-deficient embryos (Andoniadou et al., 2007), the genetic background has an effect on the phenotype of the Six3^+/−; Hesx1^Cre/+ compound embryos. Genotyping of mice and embryos was carried out by PCR on DNA purified from tail tip biopsies from mice and embryos using specific primers (Lagutin et al., 2003; Andoniadou et al., 2007). Primers and PCR protocols are available upon request.

Figure 1.

Severe growth retardation and hypoplasia of the testis and thyroid gland in Six3^+/−;Hesx1^Cre/+ mice. (A) Six3^+/−;Hesx1^Cre/+ compound heterozygote showing growth retardation and smaller body size compared with wild-type, Six3^+/− and Hesx1^Cre/+ littermates at four weeks of age. (B) Growth graphs of Six3^+/−;Hesx1^Cre/+ animals (crosses), wild-type (triangles), Six3^+/− (squares) or Hesx1^Cre/+ (circles). Note the growth retardation and early lethality in Six3^+/−;Hesx1^Cre/+ animals when compared with other genotypes (P<0.001, one-way ANOVA).(C-D) Transverse sections through the testis stained with hematoxylin-eosin of wild-type (C) and compound heterozygous (D) mice. Note that the Six3^+/−;Hesx1^Cre/+ testis has small and hypocellular seminiferous tubules compared with the wild-type control (arrows in C,D). (E-H) Transverse sections through the thyroid gland stained with hematoxylin-eosin of wild-type (E and G) and Six3^+/−;Hesx1^Cre/+ (F and H) animals. Note the abnormal histology in the Six3^+/−;Hesx1^Cre/+ animal, which shows an evident reduction in the number of follicles, when compared with the wild-type (arrows). Photographs of histological analysis are representative of three embryos per genotype.

In situ hybridisation, histology, cell death and proliferation analyses

Preparation of mouse tissues and embryos, and in situ hybridisation on paraffin sections, were performed as described (Dasen et al., 2001; Andoniadou et al, 2007). Cell death in the pituitary gland was analysed by TUNEL staining on wax sections as described (Martinez-Barbera et al., 2002). Proliferation analyses were performed using 6 alternate sagittal sections per embryo through the pituitary gland of at least three embryos per genotype. The mitotic index of the luminal cells represents the percentage of proliferating cells from the total number of luminal cells (DAPI positive). Anti-phosphorylated-histone H3 (1:1,000) (rabbit polyclonal; Upstate) and anti-cyclin D2 (1:250) (Rabbit anti-mouse; Autogen Bioclear) were used for these studies.

Quantitative real-time PCR

Total RNA was extracted from 12.5 dpc pituitary glands of wild-type, Hesx1^Cre/+, Hesx1^Cre/Cre, Six3^+/− and Six3^+/−;Hesx1^Cre/+ embryos using the RNeasy Micro kit (Qiagen), including DNase I treatment. Reverse transcription was performed using the Omniscript RT kit (Qiagen) using random hexamers (Promega). Real-time qRT-PCR was performed using the SYBR GreenER reagent (Invitrogen) on an Applied Biosystems 7500 thermal cycler. The cycling steps were performed as follows: one cycle of 50°C for 2 min, one of 95°C for 10 min, 40 cycles of 95°C for 15 s then 57°C for 1 min. Reactions were run in triplicate and repeated for several independent samples for each genotype (n=3-6 for Hesx1, Lef1, Axin2, Pitx2, Cyclin D1, Cyclin D2 and c-Myc). Fold changes in gene expression were determined using the ΔΔCT method, using the 7500 Fast System software (Applied Biosystems), by normalizing expression of targets to endogenous Gapdh. Data values were presented as averages for each genotype and the standard error of the mean (SEM) was calculated to determine the error bars (±SEM). Forward and reverse primers used are shown in Supplementary Table S1.

Analysis of protein-protein interactions

Plasmids expressing GAL4-HESX1, SIX3-VP16 or TLE1-VP16 were generated by cloning the specific coding regions into the pM and pVP16 vectors (Clontech). A plasmid containing the entire mouse Six3 coding region was kindly provided by Dr. G. Oliver (Memphis, USA). The plasmid expressing untagged HESX1 and the reporter plasmids p-P3-luciferase and p-Gal4BS-luciferase have been previously described (Sajedi et al., 2007). The integrity of the regions encoding the fusion proteins was confirmed by DNA sequencing. Cell transfection and immunoprecipitation experiments were performed as described (Sajedi et al., 2007).

Results

Growth retardation, hypoplasia of the thyroid gland and gonads and premature lethality in Six3^+/−;Hesx1^Cre/+ compound mice

Intercrosses between Six3^+/− and Hesx1^Cre/+ animals produced offspring that were phenotypically normal until weaning (3^rd postnatal week). Genotyping of embryos revealed the expected Mendelian mode of inheritance for all genotypes, whereas there was a slight decrease in the proportion of Six3^+/−;Hesx1^Cre/+ pups at weaning (Table 1). Although this decrease was not statistically significant, it is notable that a small number of Six3^+/−;Hesx1^Cre/+ mice were found dead within the first three weeks after birth (see Materials and Methods). Soon after weaning, it became evident that a large proportion (~87 %) of Six3^+/−;Hesx1^Cre/+ mice failed to gain weight, resulting in animals with severe growth retardation compared to the wild-type, Six3^+/− or Hesx1^Cre/+ littermates (Fig. 1A). Both male (n=5) and female (n=6) Six3^+/−;Hesx1^Cre/+ mice showed severe growth deficiency and weighed around four times less than their Six3^+/−, Hesx1^Cre/+ or wild-type littermates at five weeks of age. Eventually, these animals died prematurely around the 5^th-6^th week of age (Fig.1B).

Table 1.

Genotypes of embryos and mice from Six3^+/− and Hesx1^Cre/+ intercrosses.

	Genotypes
Stage
	+/+	Six3 +/−	Hesx1 Cre/+	Six3+/− ;Hesx1Cre/+
10.5 dpc	13	10	8	10
11.5 dpc	12	9	8	7
12.5 dpc	15	19	17	20
13.5 dpc	12	10	12	13
15.5 dpc	6	7	5	8
17.5 dpc	14	16	14	13
Total embryos (%)	72 (25.9%)a	71 (25.5%)a	64 (23%)a	71 (25.6%)a
3 weeks postnatal	22	21	22	17
8 weeks postnatal	22	21	22	3b

^aChi-square test showed no statistically significant deviation from the expected 25% ratios.

^bThe reduced survival of Six3^+/−;Hesx1^Cre/+ at 8 weeks is statistically significant (p<0.01, Z-test).

Severe growth retardation resulting in dwarfism has been observed in mouse lines with compromised pituitary development and function, such as in the Ames Dwarf (Prop1) and little (Ghrhr) mouse mutants (Lin et al., 1993; Gage et al., 1995; Sornson et al., 1996) and the Ghr (growth hormone receptor-deficient embryos) and Cga (αGsu subunit mutant) knockout mice (Kendall et al., 1995; Lupu et al., 2001). Since Hesx1 has been implicated in hypopituitarism in mice and humans (Dattani et al., 1998; Kelberman and Dattani, 2007) and Six3 is reportedly expressed in the pituitary gland (Oliver et al., 1995), we hypothesised that hypopituitarism could be responsible for the growth impairment and perhaps the lethality in the Six3^+/− ;Hesx1^Cre/+ animals. To test this hypothesis we performed morphological analyses on the testes and thyroid glands, two pituitary target tissues (Kendall et al., 1995), in Six3^+/−;Hesx1^Cre/+compound mice compared with Six3^+/−, Hesx1^Cre/+ and wild-type littermates at 5 weeks postnatal. Histological analysis of Six3^+/−;Hesx1^Cre/+compound mice revealed the presence of small seminiferus tubules, which were hypocellular, and the absence of elongated spermatids (Fig. 1C,D; Supplementary Fig. S1). This phenotype has been observed in mouse mutants manifesting deficient secretion of the pituitary hormones FSH and LH (Kendall et al., 1995; Kumar et al., 1997; Nasonkin et al., 2004). Moreover, the thyroid gland showed a dramatic reduction in the number of follicles on histological sections (Fig. 1E-H). It has been shown that adequate pituitary input through the secretion of TSH is essential for normal thyroid function (Kendall et al., 1995; Nasonkin et al., 2004). Therefore, abnormal testicular and thyroid morphology may suggest pituitary dysfunction. Since none of these defects were observed in Six3^+/−, Hesx1^Cre/+ or wild-type littermates, we conclude that there is a genetic interaction between the two transcription factors during normal pituitary development and that hypopituitarism is the most likely condition underlying the growth retardation in Six3^+/−;Hesx1^Cre/+ double heterozygous mice.

Abnormal morphogenesis, but normal specification and terminal differentiation of the anterior pituitary in Six3^+/−;Hesx1^Cre/+ embryos

Hesx1^−/− mutants show severe pituitary defects, which are evident during embryogenesis (Dattani et al., 1998; Dasen et al., 2001). We analysed the morphology and gene expression pattern of the developing anterior pituitary in Six3^+/−;Hesx1^Cre/+ compound embryos compared with unaffected littermates. A total of 31 Six3^+/− ;Hesx1^Cre/+ compound embryos from 10.5-17.5 dpc were analysed.

In situ hybridisation at 13.5 dpc with Lhx3, a marker of the anterior pituitary (Nasonkin et al., 2004), revealed two distinct phenotypes. Some embryos showed a dysmorphic anterior pituitary with a single lumen (6/9 embryos) (Fig. 2D), whilst in others, the pouch was bifurcated, had failed to separate from the oral ectoderm and exhibited multiple lumens (3/9 embryos) (Fig. 2E). This latter phenotype may be the consequence of a developmental delay in Rathke’s pouch formation, since there was no apparent variability in the morphology of the pituitary gland at 15.5 dpc (6/6 embryos) (Fig. 3B). Morphological abnormalities were not generally observed in wild-type, Six3^+/− or Hesx1^Cre/+ embryos (Fig. 2A-C), although one Six3^+/− embryo (n=17) showed a slight abnormality in pituitary shape (data not shown).

Figure 2.

Abnormal morphology, but normal gene expression in Rathke’s pouch in Six3^+/−;Hesx1^Cre/+ embryos. In situ hybridisation on sagittal sections through the pituitary gland of 13.5 dpc embryos. Anterior is to the left. (A-E) Lhx3 is expressed in all genotypes, but note the abnormal morphology of Rathke’s pouch in Six3^+/− ;Hesx1^Cre/+ embryos (D,E) compared with single mutants (B,C) or wild-type littermates (A). The phenotype shown in D,I,N,S,X was observed in 6/9 Six3^+/− ;Hesx1^Cre/+ embryos, while the phenotype displayed in E,J,O,T,Y was seen in the remaining three embryos. (F-O) The expression patterns of Prop1 (anterior pituitary) and Six3 (anterior pituitary and ventral diencephalon) are comparable among all genotypes. (P-T) Pomc1 expression in the ventral diencephalon (arrow in P) and anterior pituitary (arrowhead in P) is reduced (S) or absent (T) in Six3^+/−;Hesx1^Cre/+ embryos in comparison with control embryos (P-R). (U-Y) Likewise, expression of Pit1 in the ventral region of the anterior pituitary is either down-regulated (X) or not detected (Y) in double mutants compared with Six3^+/− (V), Hesx1^Cre/+ (W) or wild-type littermates (U).

Figure 3.

Normal anterior pituitary specification in Six3^+/−;Hesx1^Cre/+ embryos at 15.5 dpc. In situ hybridisation with Lhx3 (A,B), Pomc1 (C,D), Cga (encoding αGSU) (E,F) and Pit1 (G,H) in wild-type (+/+) and Six3^+/−;Hesx1^Cre/+ embryos. The expression domains are expanded and the anterior pituitary appears hyperplastic in the double heterozygous mutants.

Expression of several diagnostic pituitary and/or ventral diencephalic markers, including Prop1, Six3, Pomc1, Pit1, and Cga (Kendall et al., 1995; Cohen et al., 1996; Dasen et al., 2001; Lamolet et al., 2004) was essentially normal in Six3^+/−;Hesx1^Cre/+ compound embryos at 13.5-15.5 dpc, although the expression domains were enlarged and the anterior pituitary was bifurcated in these mutants (Fig. 2F-Y and 3C-H). The only relevant difference was a transient delay in the onset of Pomc1 and Pit1 expression in the hypothalamus (Pomc1; Fig. 2S,T) and anterior pituitary (Pomc1 and Pit1; Fig. 2S,T,X,Y) in Six3^+/−;Hesx1^Cre/+ compound embryos compared with other genotypes. A similar delay in Pomc1 expression has been reported in mouse embryos harbouring inactivating Hesx1 mutations (Dasen et al., 2001). NeuroD1 expression, which precedes that of Pomc1, was normal in double mutants when compared with wild-type littermates (Fig. 7I,J) (Lamolet et al., 2004).

Figure 7.

Up-regulation of the β-catenin/LEF1-PITX2 pathway in Six3^+/−;Hesx1^Cre/+ pituitary glands. (A-J) In situ hybridisation on sagittal sections at the level of the pituitary gland of 12.5 dpc (A-F) and 13.5 dpc (G-J) wild-type and Six3^+/− ;Hesx1^Cre/+embryos. (A,B) Sox2 expression in the ventral diencephalon and infundibulum (arrows in A), and pituitary gland (arrowhead, A) is similar between genotypes. (C,D) Wnt4 is expressed in the anterior pituitary gland of Six3^+/− ;Hesx1^Cre/+ (D) and wild-type (C) embryos. (E,F) Wnt5a is expressed in the ventral diencephalon and infundibulum of the double mutant (F) and wild-type (E) embryos. (G,H) Axin2 is expressed in a high-ventral to low-dorsal gradient within the wild-type anterior pituitary (G). (H) Axin2 expression is increased in the ventral anterior pituitary of the Six3^+/−;Hesx1^Cre/+ embryo and appears to reach more dorsal levels than in the wild-type littermate. (I,J) NeuroD expression is detected in the ventral side of the anterior pituitary in both genotypes. (K,L) Pituitary glands of 13.5 dpc wild-type (K) and Six3^+/−;Hesx1^Cre/+ (L) embryos stained with an antibody against Cyclin D2. Note the increase of Cyclin D2-positive cells in the mutant pituitary gland (L). (M) Quantitative Real-Time PCR analysis of Hesx1, Lef1, Pitx2, Cyclin D2 and Axin2 expression in pituitary glands dissected from wild-type (black bars), Hesx1^Cre/+ (white bars), Six3^+/− (grey bars) and Six3^+/−;Hesx1^Cre/+ (striped bars) embryos at 12.5 dpc. Note an up-regulation of Lef1 (p<0.005, one-way ANOVA) and Pitx2 (p<0.05, one-way ANOVA followed by Dunn’s method) in Six3^+/−;Hesx1^Cre/+ compound embryos when compared with other genotypes. CyclinD2 is increased in double mutants (p<0.05, one-way ANOVA) in comparison with wild-type and Six3^+/−, but not compared with Hesx1^Cre/+ embryos. Axin2 is up-regulated in Six3^+/−;Hesx1^Cre/+ double heterozygotes compared with wild-type and Hesx1^Cre/+ embryos (p<0.05, one-way ANOVA). Axin2 expression appears elevated in Six3^+/− embryos compared with wild-type and Hesx1^Cre/+, but the differences do not reach statistical significance.

Analysis of embryos at 17.5 dpc showed that hormone-producing cells were present in the anterior pituitary of the Six3^+/−;Hesx1^Cre/+ compound embryos, as evidenced by the expression of the terminal differentiation markers Gh, Cga, Pomc1, Tshb, Prl, Lhb and Sf1 (Fig.4). The only significant difference was the apparent reduction in Lhb-expressing cell numbers, possibly reflecting a developmental delay in the terminal differentiation of the gonadotrophs. Pre-gonadotroph specification was normal in compound embryos when compared with wild-type littermates, evidenced by correct expression of Sf1, a marker of pre-gonadotrophs. However, we observed two morphologically distinct phenotypes: (i) Six Six3^+/−;Hesx1^Cre/+ compound embryos showed pituitary glands normally located between the hypothalamus and the basisphenoid bone (Fig. 4B,E,H,K,N,Q,T). Morphologically, these were smaller along the medio-lateral axis but expanded along the rostrocaudal axis. In two of the latter, pituitary tissue could be seen invading the nasopharyngeal cavity (Supplementary Fig. S2); (ii) Three additional embryos appeared to have an absent pituitary gland, as it could not be morphologically recognised in its normal position in paraffin sections. However, in situ hybridisation revealed the presence of a large mass of anterior pituitary tissue ectopically located in the nasopharyngeal cavity (Fig. 4C,F,I,L,O,R). Taken together, these results suggest that there is no major spatial or temporal disruption of the specification or differentiation programs of the anterior pituitary in the Six3^+/−;Hesx1^Cre/+ compound embryos, but morphogenesis was severely impaired, occasionally resulting in misplacement of the pituitary into the nasopharyngeal cavity. Ectopic pharyngeal pituitary has also been reported in human embryos in association with craniofacial defects (Kjaer et al., 2000; Osman et al., 2006)

Figure 4.

Normal terminal differentiation of hormone-producing cells in Six3^+/− ;Hesx1^Cre/+ anterior pituitary glands. In situ hybridisation on transverse sections through the pituitary glands of wild-type (A,D,G,J,M,P,S) and Six3^+/−;Hesx1^Cre/+ compound embryos showing either mild (B,E,H,K,N,Q,T) or severe (C,F,I,L,O,R,U) defects. In the mild phenotype, the pituitary gland is positioned normally and all cell types are detectable. However, in severe cases, the pituitary gland is ectopically misplaced in the nasopharyngeal cavity (np in C). In general, all cell types are identifiable in Six3^+/−;Hesx1^Cre/+ mutants, although Lhb expression was barely detectable in O. This is likely to be as a consequence of delayed differentiation of the gonadotrophs, as Sf1 expression was normal in the anterior pituitary (S-U). The apparent increase of signal in the hypothalamic area in U is due to the axial level of the section.

Finally, in situ hybridisation analysis of Six3^+/−;Hesx1^Cre/+ compound mice at postnatal week four revealed the presence of all six cell types, although the number of expressing cells appeared lower compared with single heterozygous or wild-type animals (Fig. 5).

Figure 5.

Hormone-producing cells are present in the postnatal pituitary gland of Six3^+/−;Hesx1^Cre/+ animals. (A-B) Photographs of pituitary glands isolated from wild-type (A) or Six3^+/−;Hesx1^Cre/+ (B) mice at four weeks of age. (C-N) In situ hybridisation on transverse sections through the pituitary glands shown in A,B. Note the specific signal in Six3^+/−;Hesx1^Cre/+ sections with all the probes analysed, suggesting that hormone-producing cells are still present at this stage.

Increased cellular proliferation in Six3^+/−;Hesx1^Cre/+ compound embryos

It has been proposed that increased cellular proliferation and recruitment of additional oral ectoderm into Rathke’s pouch may account for the observed enlargement and bifurcation of the anterior pituitary in Hesx1^−/− mutants (Dasen et al., 2001). Alternatively or in addition, decreased cell death might also contribute to the observed phenotype. We decided to analyse the contribution of these mechanisms to the pituitary phenotype in Six3^+/−;Hesx1^Cre/+ compound embryos.

The initial induction of Rathke’s pouch was assessed by in situ hybridisation with Fgf8 and Lhx3 at 10.5 dpc. Fgf8 was expressed in the ventral diencephalon in Six3^+/−;Hesx1^Cre/+double heterozygotes and no major differences were observed in these embryos compared with Six3^+/−, Hesx1^Cre/+ and wild-type littermates (Supplementary Fig. S3C,D). Likewise, no apparent differences were observed in Lhx3 expression in the developing Rathke’s pouch (Supplementary Fig. S3A,B). The expression pattern of Bmp4 was also comparable between wild-type and Six3^+/− ;Hesx1^Cre/+ mutants at 11.5 dpc (Supplementary Fig. S3E,F). From this analysis, we conclude that the initial induction of Rathke’s pouch occurs normally in Six3^+/− ;Hesx1^Cre/+ compound embryos and that it is unlikely that the anterior pituitary hyperplasia in these mutants is primarily caused by the recruitment of additional oral ectoderm into Rathke’s pouch at 10.5 dpc.

Cell death was analysed by TUNEL staining on sections of the pituitary at 12.5 dpc. The number of apoptotic cells within Rathke’s pouch and the developing anterior pituitary was almost negligible in Six3^+/−;Hesx1^Cre/+double heterozygotes, single mutants and wild-type embryos (data not shown) (Elsworth et al., 2008). Therefore, decreased cell death is unlikely to be an important contributor to the enlargement and bifurcation of the developing pituitary gland in double heterozygous embryos. Cell death was not affected in Hesx1-deficient embryos compared with wild-type littermates (data not shown; Dasen et al., 2001).

Cellular proliferation was assessed at 11.5 and 12.5 dpc, when the pituitary phenotype is first detectable. At 11.5 dpc, more proliferating cells were seen in the developing Rathke’s pouch of the Six3^+/−;Hesx1^Cre/+ compound embryos compared with other genotypes, but differences were not statistically significant (data not shown). In contrast, at 12.5 dpc, it was evident that the anterior pituitary of the Six3^+/− ;Hesx1^Cre/+ mutants was hyperplastic (Fig. 6K). Likewise, analysis of Hesx1^Cre/Cre pituitaries at this stage revealed an increased mitotic index in three mutants, correlating with mild forebrain abnormalities, as previously suggested (n=5; Fig. 6K) (Dasen et al., 2001). The remaining two mutants did not show an increase in proliferation (data not shown). These mutants had significant forebrain and craniofacial defects, and an apparent developmental delay in Rathke’s pouch formation, i.e. Rathke’s pouch remained connected to the oral cavity and resembled that of an 11.5 dpc embryo (Supplementary Fig. S4). It is important to emphasise that there is no general developmental delay of Hesx1-deficient embryos, but it is only restricted to anterior forebrain structures, including the telencephalon, eyes, frontonasal mass and pituitary gland. From these data, we conclude that the Six3^+/− ;Hesx1^Cre/+ mutants phenocopy the mild end of the spectrum of forebrain and pituitary defects observed in Hesx1 null embryos.

Figure 6.

Increased cellular proliferation in the anterior pituitary of Six3^+/−;Hesx1^Cre/+ and Hesx1^Cre/Cre embryos. (A-J) Fluorescent images of sections of 12.5 dpc embryos at the level of the pituitary gland (anterior to the left) stained with either anti-phospho-histone H3 antibody (A-E) or DAPI (F-J). (K) Mitotic index of luminal cells in the anterior pituitary of different genotypes. Six3^+/−;Hesx1^Cre/+ and Hesx1^Cre/Cre have a significantly higher mitotic index (p<0.001 one-way ANOVA) compared with wild-type, Six3^+/− or Hesx1^Cre/+ embryos. Note the slight increase in the mitotic index in Six3^+/− embryos, which failed to reach statistical significance compared with wild-type.

Cell death and proliferation analyses were also performed at 17.5 dpc and at 30-33 days of age (postnatal week 4.5). Cell death, as measured by TUNEL staining, was comparable between Six3^+/−;Hesx1^Cre/+ mutants and wild-type littermates (data not shown). However, cellular proliferation in the pituitary gland was significantly reduced in compound mutants (Supplementary Fig. S4). At 17.5 dpc, the pituitary gland of Six3^+/−;Hesx1^Cre/+ mutants and wild-type littermates were similar in size (although the mutant pituitary was generally smaller). However, by 4.5 weeks postnatally, the wild-type pituitary was much larger than that of compound mutants (Supplementary Fig. S4). In fact, the latter appears similar in size to that of 17.5 dpc embryos.

Taken together, two main conclusions can be drawn from the marker and proliferation analyses. First, the Hesx1/Six3 interaction appears to be relevant for the normal control of proliferation in periluminal progenitors of Rathke’s pouch at 12.5 dpc. This is consistent with previous results showing the involvement of these two genes in the growth of the pituitary gland (Hesx1) and retina (Six3) and with the normal pattern of expression of these two genes during pituitary development (Dasen et al., 2001; Dyer, 2003). Second, the increased proliferation at early stages leads to an abnormal morphogenesis of the pituitary gland in compound embryos without affecting lineage specification or final differentiation at 17.5 dpc. However, there is a knock-on effect in compound embryos that causes a reduction in the proliferation rate of differentiated progenitors at 17.5 dpc and postnatally, possibly leading to hypopituitarism.

Increased proliferation might be mediated by up-regulation of the Wnt/β-catenin pathway through LEF1-PITX2

Having established that increased proliferation is the primary cellular defect responsible for the pituitary phenotype of the Six3^+/−;Hesx1^Cre/+ compound embryos, we next sought a possible molecular mechanism. Hesx1 and Six3 have been proposed to antagonise the ectopic activation of the Wnt/β-catenin signalling pathway within the anterior forebrain (Lagutin et al., 2003; Andoniadou et al., 2007). Compelling evidence indicates that the Wnt/β-catenin signalling pathway controls proliferation during normal development in a variety of cell types, including the early pituitary (van Noort and Clevers, 2002; Zhu and Rosenfeld, 2004; Arce et al, 2006; Hoppler and Kavanagh, 2007). For instance, the β-catenin/LEF1/PITX2 pathway has been shown to stimulate the expression of growth-regulating genes, including Cyclin D1 (Ccnd1), Cyclin D2 (Ccnd2) and c-Myc (Kioussi et al., 2002; Baek et al., 2003; Briata et al., 2003). We hypothesised that over-activation of the Wnt/β-catenin signalling pathway might contribute, at least in part, to the pituitary phenotype in the Six3^+/− ;Hesx1^Cre/+ compound embryos.

Firstly, we performed in situ hybridisation with several markers that have been shown to act upon the Wnt/β-catenin signalling pathway. SOX2 and SOX3, two inhibitory β-catenin partners that are required for normal pituitary development (Fig. 7A-B; data not shown) (Zorn et al., 1999; Rizzoti et al., 2004; Mansukhani et al., 2005; Kelberman et al., 2006), were expressed normally in the ventral diencephalon (Sox2 and Sox3) and developing Rathke’s pouch (Sox2) in Six3^+/−;Hesx1^Cre/+ compound embryos at 12.5 dpc (n=3). Likewise, expression of Wnt4 (ventral diencephalon and pituitary) and Wnt5a (ventral diencephalon), was unaffected in Six3^+/−;Hesx1^Cre/+ compound embryos compared with single heterozygous and wild-type embryos (Fig. 7C-F; n=3) (Treier et al., 1998; Cha et al., 2004). We also analysed the expression of known Wnt/β-catenin targets such as Lef1, Pitx2 and Axin2. Axin2 is a known Wnt/β-catenin signalling target, which is ectopically expressed within the anterior forebrain in both Hesx1^−/− (Jho et al., 2002) and Six3^−/− embryos (G. Oliver, personal communication). Axin2 is normally expressed in a high ventral to low dorsal gradient in the anterior pituitary at 12.5 dpc (Olson et al., 2006) (Fig. 7G). In Six3^+/−;Hesx1^Cre/+ compound embryos this characteristic gradient of expression was less evident and Axin2 transcripts were detected in more dorsal regions of the developing anterior pituitary (Fig. 7H; n=4). Moreover, the hybridisation signal appeared stronger in the anterior pituitary of the compound embryos when compared with wild-type littermates. An apparent up-regulation of Pitx2 expression in the developing pituitary of Six3^+/−;Hesx1^Cre/+ compound embryos compared with wild-type littermates was also observed (Supplementary Fig. S3; n=3). However, in situ hybridisation with Lef1 antisense riboprobes detected comparable low levels of expression in the developing pituitary of compound embryos and wild-type littermates at 12.5 dpc (data not shown). In conclusion, the up-regulation of Axin2 and Pitx2 expression in the mutant Rathke’s pouch and developing anterior pituitary suggests that over-activation of Wnt/β-catenin signalling might contribute to the pituitary hyperplasia observed in the double heterozygous mutants.

Next, quantitative real-time PCR was used to analyse the expression levels of some components of the Wnt/β-catenin-LEF1-PITX2 pathway. Lef1, Pitx2 and Cyclin D2 expression levels were significantly higher in the pituitary gland of Six3^+/;Hesx1^Cre/+ compound embryos compared with control pituitaries from other genotypes (Fig. 7M). Immunostaining with a specific antibody against cyclin D2 revealed the presence of higher numbers of cycling cells in the pituitary gland of the double heterozygotes compared with wild-type littermates (Fig. 7K,L). Axin2 expression was elevated in Six3^+/−;Hesx1^Cre/+ and surprisingly, in Six3^+/− embryos, suggesting that Six3 haploinsuficiency alone may cause up-regulation of Axin2, which may contribute to the mild pituitary phenotype observed sporadically in Six3^+/− embryos. This analysis also revealed that Lef1 and Cyclin D2 expression was higher in Hesx1^Cre/Cre embryos compared with wild-type littermates, but Pitx2 and Axin2 expression was unaffected in these mutants (Supplementary Fig. S5). However, Cyclin D1 and c-Myc expression levels remained unchanged in compound heterozygotes compared with single mutants and wild-type embryos (data not shown). Two conclusions can be drawn from these results: (i) Although the cellular consequences of lacking either the two Hesx1 alleles in Hesx1^Cre/Cre embryos, or only one of each in Six3^+/−;Hesx1^Cre/+ mutants are comparable (cellular hyperproliferation), there are molecular differences in the transcription levels of particular Wnt/β-catenin components associated with specific genotypes; (ii) The increased proliferation observed in Six3^+/−;Hesx1^Cre/+ and Hesx1^Cre/Cre embryos may be partly due to the up-regulation of the Wnt/β-catenin/LEF1 pathway.

Molecular nature of the Hesx1/Six3 interaction

The possibility that Hesx1 is regulating Six3 expression appears to be unlikely. Firstly, Six3 expression is normal in the pituitary of Hesx1-deficient embryos (Supplementary Fig. S3). Unfortunately, Six3^−/− embryos fail to form any pituitary tissue, which precludes the analysis of Hesx1 expression. However, levels of Hesx1 mRNA in Six3^+/− and Six3^+/+ (wild-type) pituitaries are comparable by qRT-PCR (Fig. 7M). This does not support a regulatory role for Six3 on the Hesx1 locus. Moreover, within the anterior forebrain, Hesx1 is expressed in Six3^−/− mutants and Six3 is expressed in Hesx1^−/− embryos. Taken together, the data suggest that it is unlikely that Six3 and Hesx1 regulate each other.

We next explored whether the mechanism underlying the genetic interaction between Hesx1 and Six3 might be a direct protein-protein interaction. Firstly, this possibility was analysed in a mammalian two-hybrid system. Co-transfection of CHO and 293T cells with plasmids expressing either full-length HESX1 protein or a fusion of SIX3 to the activation domain of the herpes simplex virus protein VP-16 (SIX3-VP16) failed to reveal any significant activation over the basal levels of a luciferase reporter containing five copies of a consensus paired-like binding site (P3) (Supplementary Fig. 6A). In contrast, co-expression of HESX1 and TLE1-VP16 led to a significant increase in luciferase activity. TLE1 has been previously shown to interact with HESX1 (Dasen et al., 2001). Similar results were obtained when cells were transfected with p-Gal4BS-luciferase reporter plasmid, which harbours GAL4 binding sites, and plasmids expressing GAL-4-HESX1 and SIX3-VP16 fusion proteins (data not shown). Secondly, four attempts to co-immunoprecipitate tagged HESX1 and SIX3 in mammalian cells were unsuccessful, even though both proteins were expressed at high levels and could be immunoprecipitated with specific anti-tag antibodies (Supplementary Fig. S6B). Taken together with the fact that a HESX1/SIX3 interaction has not been reported in several screening studies aiming to identify HESX1 and SIX3 protein partners (Zhu et al., 2002; Lopez-Rios et al., 2003; Sajedi et al., 2007), we conclude that it is unlikely that HESX1 and SIX3 interact at the protein level.

SIX3 is expressed in the pituitary gland of the human embryo

It has been previously reported that SIX3 is expressed in the eye of the developing human embryo (Granadino et al., 1999), but it is unknown whether SIX3 may also be expressed in the human pituitary gland. To investigate this, we performed in situ hybridisation on coronal paraffin sections through the developing anterior pituitary and prospective hypothalamus of human embryos at Carnegie Stage (CS) 17 and 20, and Fetal Stage (F) 1 (Fig. 8A). SIX3 expression was detected in Rathke’s pouch and overlying ventral diencephalon at CS 17 (Fig. 8A) and in the anterior and posterior lobes of the pituitary at CS 20 (Fig. 8B). At F1, SIX3 expression was observed in the anterior pituitary, and the ventricular zones of the hypothalamus and telencephalic vesicles (Fig. 8C). This analysis indicates that SIX3 expression occurs during normal human pituitary gland embryogenesis, suggesting that its function in this organ may be conserved between mice and humans.

Figure 8.

SIX3 is expressed in the developing pituitary of human embryos. In situ hybridisation on coronal sections of human embryos at Carnegie Stage (CS) 17 and 20, and at Fetal Stage 1 specified on the left side of the panel. (A) SIX3 is expressed in the ventral diencephalon (arrows) and developing anterior pituitary (arrowhead) at CS17. (B) Note the expression in the infundibulum (arrows) and anterior pituitary (arrowheads) at CS20. (C) At Fetal Stage 1 (F1), SIX3 transcripts are detected in the pituitary gland (arrowhead) and in the ventricular zone of the hypothalamus (hy, inset) as well as in the telencephalic vesicles (tv, inset). bs: basisphenoid cartilage.

Discussion

In this manuscript we show that haploinsufficiency of both Six3 and Hesx1 leads to severe pituitary dysmorphogenesis during development, which results in hypopituitarism with dramatic postnatal growth retardation of Six3^+/−;Hesx1^Cre/+ compound mice. This research provides the first evidence indicating an essential role for the homeobox gene Six3 during normal pituitary formation.

Hesx1 and Six3 control cell proliferation of anterior pituitary progenitors during development

Our data indicate that the primary cellular defect underlying the anterior pituitary phenotype in Six3^+/−;Hesx1^Cre/+ compound embryos is an increased proliferation of anterior pituitary progenitors between 11.5 and 12.5 dpc. This defect is also observed in a proportion of Hesx1-deficient (Hesx1^Cre/Cre) embryos, suggesting that a similar molecular mechanism might underlie the pituitary phenotype in both mutants.

Normal pituitary growth is dependent on several signalling pathways, including the FGF, TGFβ and Wnt/β-catenin pathways (Treier et al., 1998; Ericson et al. 1998). Different lines of evidence suggest that down-regulation of the Wnt/β-catenin signalling pathway in anterior pituitary precursors during early stages of pituitary development is required for normal pituitary formation. Firstly, conditional removal of normal β-catenin function with a Pitx1-Cre driver has no phenotypic consequences from 9.5 to 11.5 dpc, suggesting that β-catenin does not play an essential role during early pituitary development (Olson et al., 2006). Secondly, nuclear β-catenin is not detected in the developing pituitary gland from 10.5 to 13.5 dpc (Brinkmeier et al., 2007). Thirdly, activation of Wnt/β-catenin signalling through LEF1 has been proposed to induce Pitx2 expression, which in turn, controls the transcription of critical cell cycle regulators that promote cell-type-specific proliferation (Kioussi et al., 2002; Baek et al., 2003; Briata et al., 2003). Finally, while mutations in Wnt4 are associated with pituitary hypoplasia (Treier et al., 1998), embryos carrying a mutation in Tcf4, which usually acts as a Wnt/β-catenin repressor (van Noort and Clevers, 2002; Arce et al., 2006; Hoppler and Kavanagh, 2007), exhibit elevated cell proliferation (Brinkmeier et al., 2007). In the forebrain, both Hesx1 and Six3 function to antagonise the activation of Wnt/β-catenin signalling within the anterior forebrain precursors to prevent a posterior cell fate transformation. Our data indicate that expression of Lef1, Pitx2 and Cyclin D2 is elevated in both Six3^+/−;Hesx1^Cre/+ and Hesx1^Cre/Cre embryos and Axin2 (another Wnt/β-catenin target) is expressed ectopically and at high levels in Six3^+/−;Hesx1^Cre/+ compound heterozygotes compared with wild-type littermates. Therefore, it is feasible that enhanced or ectopic Wnt/β-catenin signalling might be a major contributor to the pituitary phenotype observed in the Six3^+/−;Hesx1^Cre/+ compound and Hesx1^Cre/Cre embryos. Although we cannot exclude a role for the FGF and TGFβ signalling pathways, we have shown that Fgf8 and Bmp4 expression is not grossly affected in either Six3^+/−;Hesx1^Cre/+ or Hesx1-deficient embryos compared with single mutants or wild-type littermates at 10.5-11.5 dpc. It is worth noting that the phenotypic consequences of enhanced Wnt/β-catenin signalling within the anterior forebrain (cell fate transformation) and anterior pituitary progenitors (increased proliferation) are different, supporting the idea that the specific response to a de-regulation of Wnt/β-catenin signalling is highly context dependent (van Noort and Clevers, 2002; Arce et al., 2006; Hoppler and Kavanagh, 2007).

The molecular basis of the genetic interaction between Hesx1 and Six3 is not fully understood. We present data suggesting that a HESX1/SIX3 protein-protein interaction is unlikely. Whether these two transcription factors act in parallel pathways during pituitary development or alternatively, they directly regulate transcription of a subset of common genes, acting synergistically to regulate Wnt/β-catenin signalling in the pituitary, is an important question that merits further molecular studies.

Haploinsufficiency of Hesx1 and Six3 leads to hypopituitarism in mice

A puzzling aspect of the phenotype in the Six3^+/−;Hesx1^Cre/+ compound heterozygotes is that the pituitary gland is hyperplastic during development but it becomes hypoplastic postnatally. Moreover, normal pituitary function is impaired as judged by the arrested growth of the compound mice and the histological defects observed in pituitary target organs such as the thyroid and the testes. However, all hormone-producing cell types are present in the pituitary gland of Six3^+/−;Hesx1^Cre/+ compound mice. This observation can be partly explained when considering that a proportion of anterior pituitary tissue escapes from its normal location (between the hypothalamus and basisphenoid bone) into the nasopharyngeal space. Although this tissue contains hormone-producing cells, it is likely that its function may be impaired as there is no contact with the hypothalamus, which controls expression and secretion of anterior pituitary hormones. Moreover, this ectopic tissue might be lost once the affected pups are born, either by degeneration or by physical damage, as we have found that it is often absent when the embryos are processed for histological analysis. This is consistent with our observations that there is a significant reduction of pituitary tissue in Six3^+/−;Hesx1^Cre/+ compound embryos at 17.5 dpc.

However, a significant amount of anterior tissue remains in its normal location in the compound mutants at 17.5 dpc. Our data indicate that this remaining tissue fails to proliferate postnatally and the pituitary gland of Six3^+/−;Hesx1^Cre/+ compound mice is very small when compared with normal littermates. This suggests that the hypothalamic-pituitary axis may be disrupted in these mutants, as it has been shown that hypothalamic input through the hypophysiotrophic factors is essential for normal postnatal expansion of differentiated anterior hormone-producing cells (Billestrup, et al., 1986; Korbonits, etal., 2004; Ward et al., 2005, 2006; Zhu et al., 2007). Analysis of the Six3^+/−;Hesx1^Cre/+ compound mice (this study) and other mouse models carrying inactivating Hesx1 mutations initially identified in humans suffering from hypopituitarism and septo-optic dysplasia (E.Sajedi and J.P. Martinez-Barbera, unpublished), suggests a hypothalamic contribution to the pituitary phenotype in these mutant mice.

This is exemplified by the fact that although developmentally, the pituitary defects in Six3^+/−;Hesx1^Cre/+ embryos are very similar to those described in embryos carrying these inactivating Hesx1 mutations (Dasen et al., 2001; E.Sajedi and J.P. Martinez-Barbera, unpublished), the phenotypic consequences in newborn mice are rather different. The great majority of Hesx1^−/− mice, which carry a null allele, die soon after birth (Andoniadou et al., 2007). In contrast, newborns harbouring a Hesx1 hypomorphic allele, in which there is a substitution of the conserved isoleucine 26 by threonine, are generally viable and show no growth defects (E. Sajedi and J.P. Martinez-Barbera, unpublished). Finally, Six3^+/−;Hesx1^Cre/+ mice exhibit dwarfism and die by the 5^th-6^th week of age. It is important to emphasise that the main phenotypic difference between these genotypes is the presence of severe forebrain defects in Hesx1^−/− embryos, whereas both Hesx1^I26T/I26T and Six3^+/−;Hesx1^Cre/+ mutants exhibit only mild forebrain defects (data not shown). Since the pituitary defects are very similar between the three genotypes, it is conceivable that there may be a neural contribution to the growth retardation in Six3^+/−;Hesx1^Cre/+ compound mice. In fact, Hesx1 is expressed in the prospective hypothalamic region between 8.5 and 9.5 dpc, and Six3 expression remains high in this area during development until 16.5 dpc, making it feasible that a gene dosage reduction of these two transcription factors could affect normal diencephalic development, including the hypothalamus. Alternatively, or in addition, defects in the portal blood system or impaired connection between hypothalamus and anterior pituitary might contribute to the postnatal hypopituitarism (Ward et al., 2006). Although further research is needed to clarify the reasons underlying the postnatal phenotype of the Six3^+/−;Hesx1^Cre/+ compound mice, it is interesting that conditional inactivation of Six3 either within the brain or the pituitary gland leads to postnatal growth retardation and premature death (A. Lavado and G. Oliver, personal communication), implying that Six3 is required not only within the pituitary gland but also in the forebrain for normal pituitary function.

It has been shown that dominant mutations in HESX1 associated with human cases of hypopituitarism show a marked variablility in both the severity and penetrance of the phenotype (Dattani et al., 1998; Kelberman and Dattani, 2007). Often, this variability is observed between individuals carrying the same heterozygous mutation within the same family, suggesting that factors other than the HESX1 genotype might be involved. Our research has revealed that while single heterozygous individuals are normal, double heterozygous Six3^+/−;Hesx1^Cre/+ animals exhibit hypopituitarism. This raises the possibility that some of the HESX1 heterozygous human patients might also be heterozygous for SIX3. The fact that SIX3 expression in the developing pituitary gland is conserved in human embryos supports this idea. Future research will evaluate the role of SIX3 in the aetiology of these human conditions.