Gcm2 is required for the differentiation and survival of parathyroid precursor cells in the parathyroid/thymus primordia

Abstract

The parathyroid glands develop with the thymus from bilateral common primordia that develop from the 3^rd pharyngeal pouch endoderm in mouse embryos at about E11, each of which separates into one parathyroid gland and one thymus lobe by E13.5. Gcm2, a mouse ortholog of the Drosophila Glial Cells Missing gene, is expressed in the parathyroid-specific domains in the 3^rd pouches from E9.5. The null mutation of Gcm2 causes aparathyroidism in the fetal and adult mouse, and has been proposed to be a master regulator for parathyroid development. In order to study how Gcm2 functions in parathyroid development, we investigated the mechanism that causes the loss of parathyroids in Gcm2 null mutants. Analysis of the 3^rd pouch-derived primordium in Gcm2^−/− mutants showed the parathyroid-specific domain was present before E12.5, but underwent programmed cell death between E12-12.5. RNA and protein localization studies for parathyroid hormone (Pth) in wild-type embryos showed that the presumptive parathyroid domain in the parathyroid/thymus primordia started to transcribe Pth mRNA and produce PTH protein from E11.5, before the separation of parathyroid and thymus domains. However in Gcm2^−/− mutants, the parathyroid-specific domain in the common primordium did not express Pth and could not maintain the expression of two other parathyroid marker genes, CasR and CCL21, although expression of these two genes was initiated. Marker gene analysis placed Gcm2 downstream of the known transcription and signaling pathways for parathyroid/thymus organogenesis. These results suggest that Gcm2 is not required for pouch patterning or to establish the parathyroid domain, but is required for differentiation and subsequent survival of parathyroid cells.

Introduction

Mammals are equipped with an efficient system to regulate ionized calcium and phosphorus homeostasis in the extracellular environment that is composed of parathyroid glands, bone, kidney, and intestine. In this system, the parathyroid glands are the most important endocrine regulator to maintain the calcium homeostasis in the circulation (Ramasamy, 2006). The primary function of the parathyroids is to produce and release an 84-amino acid hormone called parathyroid hormone (PTH), which directly targets receptors on osteoblasts to regulate bone resorption and on distal tubule epithelial cells in the kidney to increase renal calcium reabsorption (Houillier et al., 2003). PTH also indirectly stimulates intestinal calcium absorption by increasing 1,25(OH)₂D₃ production in the kidney (Ramasamy, 2006). The requirement of PTH in the regulation of calcium homeostasis was found not only postnatally, but also at fetal stages (Kovacs et al., 2001a; Kovacs et al., 2001b; Miao et al., 2002). PTH is also essential for fetal bone formation (Miao et al., 2002). The production and secretion of PTH in the parathyroid glands is controlled by the membrane-bound calcium-sensing receptor (CasR), which regulates PTH secretion by sensing the changes of extracellular ionized calcium concentration (Chang and Shoback, 2004; Chen and Goodman, 2004).

Serum calcium plays many physiological functions including neuromuscular excitability, muscle contraction, blood coagulation and bone mineralization (Ramasamy, 2006). Due to the importance of PTH in the calcium homeostasis, PTH deficiency (hypoparathyroidism) caused by the failure of or disorders in parathyroid development causes disease in humans (Thakker, 2001). Hypoparathyroidism can be caused by the mutation of the genes that are required for normal parathyroid physiological functions, including Pth (Ahn et al., 1986; Goswami et al., 2004) and CasR (Suzuki et al., 2005; Thakker, 2004). It also can result from the mutation of genes that function in parathyroid development, like Gata3(Van Esch et al., 2000), Sox3 (Bowl et al., 2005), and Gcm2 (Ding et al., 2001; Thomee et al., 2005). The study of parathyroid organogenesis can therefore help us to understand the mechanisms of human hypoparathyroidism.

In mouse, the parathyroids are bilateral organs that develop with the thymus from two common parathyroid/thymus primordia originating from the 3^rd pharyngeal pouch endoderm. Beginning at E8.0, the pharyngeal endoderm develops four bilateral pouches that give rise to several organs, including the thymus and parathyroids (Graham, 2003; Graham and Smith, 2001). The 3^rd pharyngeal pouches are formed at E9.5-10 days, and are patterned into dorsal/anterior parathyroid and ventral/posterior thymus domains (Gordon et al., 2001; Moore-Scott and Manley, 2005; Patel et al., 2006). The 3^rd pouch endoderm proliferates to form bilateral parathyroid/thymus common primordia at E11-11.5. Each primordium separates into one parathyroid gland and one thymus lobe at E12.5-13.5, which then migrate to their eventual adult locations by about E14.5 (Blackburn and Manley, 2004; Manley, 2000; Manley and Blackburn, 2003). In the adult mouse, the parathyroids are located near or embedded within the thyroid gland, and the thymus is situated in the anterior chest cavity. Thus, the early stages of parathyroid organogenesis are closely linked with thymus organogenesis.

The molecular mechanisms that regulate pouch patterning and early parathyroid/thymus organogenesis are beginning to be identified. The Hoxa3, Pax1/9, Eya1, and Six1/4 transcriptional regulators have been implicated as a pathway/network regulating early organogenesis of both organs, since mice that lack these genes have normal initial pouch formation, but then fail to form or have hypoplastic parathyroids and thymus. The Hoxa3 null mutation causes the most severe defects in parathyroid/thymus organogenesis, as the Hoxa3^−/− mutants fail to initiate the formation of the parathyroid/thymus primordia (Chisaka and Capecchi, 1991; Kameda et al., 2004; Manley and Capecchi, 1995; Manley and Capecchi, 1998; Su and Manley, 2002). A Pax1/9-Eya1-Six1/4 network has been identified to act downstream of Hoxa3 during patterning and early organogenesis of both the thymus and parathyroids (Dietrich and Gruss, 1995; Manley and Capecchi, 1995; Neubuser et al., 1995; Peters et al., 1998; Su et al., 2001; Su and Manley, 2000; Wallin et al., 1996; Xu et al., 2002; Zou et al., 2006).

The mechanism by which the parathyroid- and thymus-specific domains in the 3^rd pouch and subsequent primordia are specified is beginning to be understood. Gcm2 and Foxn1 are organ-specific transcription factors that are localized to the parathyroid- or thymus-specific domains of the common primordia before their separation (Gordon et al., 2001). Foxn1 expression beings at E11.25 in a domain that is complementary to Gcm2 expression in the parathyroid/thymus primordia (Gordon et al., 2001). The Foxn1 null mutation, nude, causes failure of thymic epithelial cell differentiation, but does not affect the initiation of thymus organogenesis (Blackburn et al., 1996; Nehls et al., 1996). The Gcm2 null mutation has been reported to cause complete and specific failure of parathyroid development (Gunther et al., 2000). Gcm2 expression begins at E9.5 in the dorsal-anterior pharyngeal endoderm of the 3^rd pouch and is maintained in the presumptive parathyroid domain at later stages (Gordon et al., 2001). The early expression pattern and apparent failure of parathyroid organogenesis suggests that Gcm2 may specify the parathyroid domain in the 3^rd pharyngeal pouch prior to primordium formation, and be required for initial organogenesis.

Gcm2 is member of the Glial Cells Missing (Gcm) transcription factor family, which have a conserved Gcm DNA binding domain (Cohen et al., 2003). The first Gcm gene was found in Drosophila, which was shown to function to as a binary switch between neuronal and glial cells determination in Drosophila central nervous system (Hosoya et al., 1995; Jones et al., 1995). In mammals, there are two Gcm orthologs: Gcm1 and Gcm2 (Kim et al., 1998). However, neither gene is required in the nervous system in mice. Gcm1 is expressed at the placenta and is required for labyrinth formation (Schreiber et al., 2000), while Gcm2 expression is restricted to the parathyroid gland (Gordon et al., 2001; Gunther et al., 2000; Kim et al., 1998). The role of Gcm as a binary switch specifying glial cell fate in Drosophila nervous system development and the complementary expression domains of Foxn1 and Gcm2 in the common primordium suggests that parathyroid organogenesis may fail in Gcm2^−/− mutants because the parathyroid domain is transformed to a thymus fate. This possibility is supported by previous studies in our lab of the Sonic hedgehog (Shh) mutant phenotype. In the Shh null mutant, Gcm2 is never expressed, and no parathyroid domain forms. In contrast, there is an expanded thymus domain in the 3^rd pouch, marked by expanded Bmp4 and subsequently Foxn1 positive domains (Moore-Scott and Manley, 2005; Patel et al., 2006). These results are consistent with a model in which in the absence of Shh, and therefore of Gcm2, the parathyroid domain may be transformed to a thymus fate.

In the current study, we determined the role of Gcm2 in parathyroid organogenesis by studying the mechanism of aparathyroidism in Gcm2^−/− mutants. In contrast to previous reports, we showed that the parathyroid-specific domain was present and morphologically normal until E12 in Gcm2^−/− embryos. However, parathyroid-specific markers were either not expressed or not maintained at E11.5, and the parathyroid domain underwent coordinated programmed cell death at E12 and was totally lost by E12.5. Consistent with these and previous results, marker gene analysis showed normal expression of the Hoxa3-Pax1/9-Eya1 transcription factor and the Shh-Bmp4 signaling networks in Gcm2^−/− mutants, indicating that these pathways act upstream of Gcm2. We further found that Tbx1 expression, which is also restricted to the parathyroid-specific domain in the 3^rd pouch and/or the parathyroid/thymus common primordia, was not affected by the Gcm2 null mutation. This raises the possibility that Tbx1 may function to specify the parathyroid-specific domain downstream of Shh and upstream of Gcm2. Our data indicate that in spite of its early expression at E9.5, Gcm2 is not required for early patterning of the dorso-anterior parathyroid domain or initiation of parathyroid organogenesis. Also, Gcm2 does not act as a binary cell fate switch between parathyroid and thymus fates in the 3^rd pouch, but is instead required for the differentiation and survival of parathyroid precursor cells after initial organ domain formation.

Materials and methods

Mice

The generation of the Gcm2 null mutant and genotyping have been described (Gunther et al., 2000). Gcm2 mutant mice used for experiments were originated on 129/SvEv-C57BL/6J and had been backcrossed to C57BL/6J mice for more than 4 generations.

Pax9^lacZ (Peters et al., 1998), Bmp4^lacZ (Lawson et al., 1999), and Noggin^lacZ (McMahon et al., 1998) alleles were each crossed with Gcm2^+/− to obtain double heterozygous F1 mice, which were then crossed with Gcm2^+/− to produce Gcm2 heterozygotes and homozygous mutants carrying the marker alleles for analysis.

Embryos were collected with the day of the vaginal plug designated as E0.5. We also used somite number, eye pigment, and the morphology of parathyroid/thymus primordium to stage the embryos. All experiments were carried out with the approval of the UGA institutional animal care committee.

TUNEL Assay

The TUNEL assay was performed as described (Su et al., 2001). Staged embryos were fixed in 4% paraformaldehyde for 2 hours and processed for paraffin embedding. Sections were cut at 8μm. The TUNEL assay was performed on the paraffin-embedded tissue sections following the manufacturer’s guidelines (Roche Diagnostics).

Immunohistochemistry

PTH immunohistochemistry was performed using an anti-PTH antibody as described (Wurdak et al., 2005). Staged embryos were fixed in 4% paraformaldehyde overnight and processed for paraffin embedding. Sections were cut at 8-10μm and stained with a goat anti-PTH antibody using the Vectastain method (VectorLab).

Section and whole mount in situ hybridization

Paraffin section in situ hybridization was performed as described (Moore-Scott and Manley, 2005). Staged embryos were fixed in 4% paraformaldehyde overnight and processed for paraffin embedding. 8-10μm sections were hybridized with digoxigenin labeled RNA probes at 0.5 μg/ml. Alkaline phosphatase-conjugated antidigoxigenin Fab fragments were used at 1:5000. BM-purple (Roche) was used as a chromagen to localize hybridized probe. Probes for Gcm2 and Foxn1 (Gordon et al., 2001), Tbx1 (Chapman et al., 1996), CasR (Bowl et al., 2005), Shh (Echelard et al., 1993), and Ptc1 (Goodrich et al., 1996) have been described. CCL21 probe was a gift from Yousuke Takahama. Pth probe was cloned using the primers: 5-CTGCAGTCCAGTTCATCAGC-3 and 5-AAGCTTGAAAAGGTAGCAGCA-3.

For the Foxn1 section in situ hybridization on the alternative sections with the sections used for TUNEL experiment, the paraffin sections prepared for TUNEL experiment were refix in 4% paraformaldehyde for 20 minutes after rehydrate, then preformed section in situ hybridization like normal procedure.

Whole-mount in situ hybridization was performed as described (Carpenter et al., 1993; Manley and Capecchi, 1995). Probes for Hoxa3 (Manley and Capecchi, 1995), Pax1(Manley and Capecchi, 1995), and Eya1(Xu et al., 1997) have been described. Gcm2, Tbx1, Gata3, CasR, and CCL21 probes were the same as the above used for section in situ hybridization.

X-gal staining

Whole-mount X-gal staining was performed to the staged embryos as described (Patel et al., 2006). After lacZ staining, embryos were embedded in paraffin and 8 μm sections cut and counterstained with nuclear fast red.

Results

Initial parathyroid domain formation is normal in Gcm2 null mutants

Previous studies had concluded that the parathyroids were absent in Gcm2^−/− embryos as early as E11.5 (Gunther et al., 2000). To investigate how parathyroid organogenesis fails in Gcm2^−/− mutants, we performed a detailed morphological and cell fate analysis of the 3^rd pouch-derived primordium in control and Gcm2 mutant embryos from E10.5-E13.5. Gcm2 expression is normally restricted to a small dorso-anterior domain in the third pouch and in the subsequent shared organ primordium, with Foxn1 expressed in the remainder of the primordium after E11.5 (Gordon et al., 2001) (Fig. 1A-C). As Gcm2 mRNA is not present in the mutant embryos, we used Foxn1 gene expression as a marker for thymus cell fate to assess the formation of the parathyroid domain in Gcm2 mutants. The Foxn1-expressing thymus domain was normal at all time points assayed, and the presumptive parathyroid domain was clearly identifiable as a Foxn1-negative domain at the dorsal and anterior aspect of the common parathyroid/thymus primordium at E11.25, E11.5 and E12 in both wild-type and mutant embryos (Figs. 1C-F, and data not shown). At E12.5-13, we saw presumptive parathyroids consisting of Foxn1-negative cells separating from the thymus domain in the wild-type embryos (Fig. 1G, I). However, at these stages the parathyroid domain was absent in the Gcm2^−/−embryos (Figs. 1H, J). Thus, the parathyroid domain did form in Gcm2^−/−embryos and initially appeared morphologically normal, but was lost by E12.5. Furthermore, the parathyroid domain does not appear to be transformed to a thymus-specific fate, since in Gcm2^−/−embryos the Foxn1-positive thymus domain did not expand to the whole primordium at E11.25-E12 (Figs. 1D, F).

Fig. 1. Cell fate analysis of the 3^rd pouch-derived parathyroid/thymus common primordia in wild-type and Gcm2^−/− mutant embryos .

Section in situ hybridization of Gcm2 on sections of wild-type embryos at E10.5 (A) and E11.5 (B) embryos is shown for comparison. Foxn1 in situ hybridization was performed on the sections from wild-type (C, E, G and I) and Gcm2^−/− (D, F, H and J) embryos at E11.5 (C, D), E12 (E, F), E12.5 (G, H), and E13 (I, J) stages. Sections were cut in the sagittal plane. In all figures, anterior is up, and dorsal is to the right. Ages of embryos are indicated in the upper right corner of each panel. The thymus domain in (A, B), and the parathyroid domains in (C-J) are outlined. pt, parathyroid; th, thymus.

Loss of the parathyroid domain in Gcm2^−/− mutants by programmed cell death

To determine whether the presumptive parathyroid domain underwent programmed cell death in Gcm2^−/− mutants, we used the TUNEL assay at E11.5-12.5. At E11.5, there was no difference between Gcm2 null mutant and WT control embryos (Figs. 2A, and B). As we have previously reported, at this stage apoptosis is normally present at in the region of endoderm where the primordium is separating from the pharyngeal pouch (Gordon et al., 2004). This apoptosis is very transient, and at E12 few or no TUNEL-positive cells were seen in the wild-type primordium (Fig. 2E). In contrast, there was a concentration of apoptotic cells in the presumptive parathyroid-specific domain at E12 in the Gcm2 null mutants (Fig. 2F). We confirmed that the apoptotic domain at E12 corresponded to the presumptive parathyroid domain by in situ hybridization for Foxn1 on alternate sections (Fig. 2C, D). At E12.5, apoptosis could not be detected in the Gcm2^−/− mutants (data not shown), consistent with the absence of the parathyroid domain at this time point (Fig. 1H). These results indicated that loss of the presumptive parathyroid domain was via coordinated apoptosis between E12 and E12.5.

Fig. 2. TUNEL analysis of cell death in the parathyroid/thymus primordia in wild-type and Gcm2^−/− mutant embryos.

TUNEL was performed on a complete sagittal section series prepared from wild-type controls (A, E) and Gcm2^−/− embryos (B, F) at E11.5 (A and B) and E12 (E and F) stages. Anterior is up, and dorsal is to the right. In (A-B), the common parathyroid/thymus primordium is outlined in white, and the 3^rd pouch from which the primordium is undergoing separation by apoptosis is outlined in yellow (p3). At E12, in situ hybridization for Foxn1 to indicate the thymus domain (C, D) was performed on alternate sections with the sections used for TUNEL (E, F) to confirm that the location of the apoptotic cells are in the parathyroid domain in Gcm2 null mutant. The dorso-anterior parathyroid domain at E12 is indicated by a dashed line in C-F.

Parathyroid differentiation initiated, but was subsequently blocked in Gcm2^−/− mutants

To determine the earliest time point when parathyroid differentiation markers initiate and whether they were ever expressed in the Gcm2^−/− mutants prior to loss of the parathyroid domain at E12.5, we assayed the expression of calcium sensor receptor (CasR), the chemokine CCL21, and parathyroid hormone (Pth), in wild-type and Gcm2^−/− embryos. This analysis was performed to determine whether parathyroid specification and differentiation was initiated in Foxn1 negative domain in the Gcm2 mutants at E10.5-11.5.

The two earliest differentiation markers for the parathyroid domain other than Gcm2 are the calcium sensing receptor (CaSR) and the chemokine CCL21. CasR is functionally required for parathyroid cells to respond to modulating calcium concentrations. Our previous study showed that CCL21 protein is produced by the parathyroid-specific domain at E11.5, and is absent from the parathyroid domain in Gcm2^−/− mutants at E11.5 (Liu et al., 2006). This result indicated that Gcm2 regulates parathyroid-specific expression of CCL21, and providing functional evidence that the parathyroids play a role in the attraction of hematopoietic-derived T-lymphoid progenitor cells to the developing parathyroid/thymus primordium. Both CCL21 and CaSR mRNA are expressed in the dorso-anterior Gcm2 positive parathyroid domain of the 3^rd pouch at E10.5 in wild type embryos (Fig. 3A, C, E ) (Bowl et al., 2005). CasR is also expressed in a similar domain of the other three pharyngeal pouches at E10.5 (Fig. 3E), but by E11, CasR expression was specifically maintained in the dorsal parathyroid domain of the 3^rd pouch but down regulated in the other three pouches (Fig. 3G). In Gcm2^−/− mutants, both CCL21 and CasR expression in the pharyngeal region was initiated normally, although CCL21 was reduced compared to controls (Fig. 3D, F). At E11, CaSR was not maintained in pouch 3 (Fig. 3H). Thus, expression of both of these early parathyroid markers was initiated in the Gcm2 mutants, indicating that the initial patterning of the parathyroid domain in the 3^rd pouch is normal in the Gcm2^−/− mutants

Fig. 3. The expression of CasR and CCL21 in wild-type and Gcm2^−/− mutant embryos.

Whole mount in situ hybridization for Gcm2 at E10.5 (A) and E11 (B) is shown for comparison. CCL21 (C, D) and CasR (E-H) expression is shown in wild-type (C, E, G) and Gcm2^−/− (D, F, H) embryos at E10.5 (C-F). CasR expression is also shown at E11 (G, H). In all panels, dorsal is to the right, anterior is up. CasR expression was initiated at the dorsal sides of all four pouches (p1-p4) in the wild-type E10.5 embryo (E), and this expression was not affected by Gcm2 null mutation (F). The initiation of CCL21 expression at the dorsal side of 3^rd pouch (C) partially required Gcm2 function (D). CasR expression was maintained only in the parathyroid domain at E11 in the wild-type (G), and this expression required Gcm2 function (H). a1, first arch; a2, second arch.

At E11.5 in control embryos, both CasR and CCL21 were present specifically in the parathyroid domain of the common primordium (Fig. 4A, B, D). In contrast, both of these markers were absent from the presumptive parathyroid domain in the common primordium in E11.5 Gcm2^−/− mutants (Fig. 4C, E). These results show that while the initial expression of parathyroid-specific markers does not absolutely require Gcm2, Gcm2 is required for wild type initial levels of CCL21 and for maintenance of both CCL21 and CasR expression in the parathyroid domain of the shared primordium.

Fig. 4. Maintenance of CCL21 and CasR expression in the parathyroid domain at E11.5 requires Gcm2.

Sections were cut at sagittal plane. In all figures, anterior is up, and dorsal is to the right. The common parathyroid/thymus primordium is outlined and the parathyroid domain is indicated with an arrow in (A, B, D). in situ hybridization for Gcm2 at E11.5 (A) is shown for comparison. CCL21 (B) and CasR (D) are restricted to the parathyroid domain at E11.5 in wild-type embryos. Neither is expressed in Gcm2 mutants (C, E). a1, first arch; a2, second arch.

Perhaps the most characteristic parathyroid marker is parathyroid hormone. The major function for the mature parathyroid glands is to produce and secrete PTH in the circulation to regulate extracellular calcium concentration. Pth mRNA has been detected in the presumptive parathyroid domain as early as E11.5 (Gunther et al., 2000). To determine whether this represented the earliest time point of Pth expression, we checked Pth mRNA expression and protein localization in the wild-type embryos beginning at E9.5, when Gcm2 is first expressed in the 3^rd pouch. At E9.5 and E10.5, Pth mRNA was not detected in the 3^rd pouch (Fig. 5A and data not shown). The initial expression of Pth mRNA was seen at E11.5 in the parathyroid-specific domain (the anterior and dorsal part) of the parathyroid/thymus primordium (Fig. 5B), consistent with previous data (Gunther et al., 2000). Pth mRNA was then maintained in the parathyroid at all subsequent stages examined (Figs. 5C-F). The timing of initial Pth expression characterizes Pth as a late differentiation marker for parathyroid cells, compared to CCL21 and CasR. Immunohistochemistry using a PTH antibody also showed that PTH protein was produced in the parathyroid-specific domain as early as E11.5, as soon as Pth RNA was detected, and was maintained in the parathyroid at all subsequent stages examined (Figs. 5G-I). These Pth RNA and protein studies suggest that the parathyroid precursor cells in the parathyroid-specific domain can produce PTH at E11.5, prior to separation from the thymus in wild-type embryos. However, Pth mRNA and protein were never present in the parathyroid domain in Gcm2^−/− mutant embryos (Figs. 5J-N), even though the parathyroid domain is morphologically normal before E12.5 in Gcm2^−/− mutant embryos (Figs. 1D, F). This result confirms that Pth expression requires Gcm2 as in the previous report (Gunther et al., 2000).

Fig. 5. The expression of Pth in wild-type and Gcm2^−/− mutant embryos.

Section in situ hybridization of Pth was performed on the sections prepared from the wild-type embryos at E10.5 (A), E11.5 (B), E12 (C), E12.5 (D), E13.5 (E), and E18.5 (F). Immunohistochemistry of PTH was performed on the sections prepared from the wild-type embryos at E11.5 (G), E12.5 (H), and E18.5 (I). The initiation of Pth mRNA and protein expression was not present in the Gcm2^−/− mutants at E11.5 (J and M), and Pth mRNA and protein expression was also not present in late stages at E12 (K and N), and E12.5 (L). Sections were cut at sagittal plane. In all figures, anterior is up, and dorsal is to the right. Arrows in B-C indicate the Pth positive parathyroid domain. The common parathyroid/thymus primordium is outlined in (B, C, G, H, J-N). p3, third pouch; a1, first arch; a2, second arch; h, heart; th, thymus; tr, thyroid; pt, parathyroid.

Taken together, these marker studies present a time course of parathyroid differentiation, initiating with Gcm2 at E9.5, followed by CCL21 and CasR at E10.5 and Pth at E11.5. The earliest markers of the parathyroid domain, CaSR and CCL21, are both initiated at the right time and place in Gcm2 mutants, and are entirely or partially independent of Gcm2 at this stage. However, subsequent differentiation, as indicated by initiating Pth expression and maintenance of the early markers, failed in the Gcm2 mutants.

Tbx1 expression in the dorsal pouch is normal in Gcm2 mutants

Our previous studies and others have shown that the transcription factor Tbx1 is also expressed in the dorsal and anterior 3^rd pouch endoderm in wild-type embryos at E10.5, strikingly similar to the Gcm2 expression domain (Manley et al., 2004; Vitelli et al., 2002; Zhang et al., 2005) (Fig. 6A, C). Tbx1 has also been identified as a downstream target of Shh signaling in the pharyngeal endoderm (Garg et al., 2001; Yamagishi et al., 2003), as has Gcm2 (Moore-Scott and Manley, 2005). Gcm2 expression at E10.5 and parathyroid organogenesis are also absent in Tbx1^−/− mutants, although in this case the loss is secondary to the absence of the 3^rd and 4^th pouches (Ivins et al., 2005; Jerome and Papaioannou, 2001; Vitelli et al., 2002). Consistent with this early similarity between Tbx1 and Gcm2 expression, at E11.5 Tbx1 expression was restricted to the parathyroid-specific domain (Fig. 6E). There was no change in Tbx1 expression in Gcm2^−/− mutant embryos at E10.5 or E11. 5 (Fig. 6B, D, F). The restricted expression pattern of Tbx1 in the 3^rd pouch endoderm and in the parathyroid-specific domain of the common primordium in the Gcm2 mutants places Tbx1 upstream of Gcm2 in the parathyroid domain.

Fig. 6. Expression of Tbx1 in wild-type and Gcm2^−/− mutant embryos.

Whole mount (A, B) or paraffin section (C-F) in situ hybridization for Tbx1 at E10.5 (A-D) and E11.5 (E, F). In the wild-type, Tbx1 expression was present at the dorsal side of 3^rd pouch at E10.5 (A, C) and at the parathyroid-specific domain in the parathyroid/thymus primordium at E11.5 (E). This expression is not affected by Gcm2 null mutation (B, D, F). Sections were cut in the saggital plane. Arrows in (E, F) indicate the parathyroid domain. p3, third pouch; a1, first arch; a2, second arch.

Gcm2 is downstream of known transcription factor and signaling networks in parathyroid/thymus organogenesis

A number of transcription factor and signaling pathways have been identified as playing a role in the patterning of the third pouch into parathyroid and thymus domains and/or in the initiation of primordia formation. Mutants for Hoxa3, Pax1, Pax9, Eya1, and Shh all have absent or reduced Gcm2 expression at E10.5-11.5 (Moore-Scott and Manley, 2005; Su et al., 2001; Xu et al., 2002). These transcription factor and signaling pathways should function upstream to Gcm2, based on their expression patterns and mutant phenotypes.

To confirm that Gcm2 is downstream of the Hoxa3-Pax1/9-Eya1 pathway, we studied these genes’ expression in Gcm2^−/− mutants using whole mount in situ hybridization or LacZ reporter transgenic mice. Hoxa3 (Figs. 7A and B), Pax1 and Pax9 (Figs. 7C-F), and Eya1 expression (Figs. 7G and H) were all the same in both wild-type and Gcm2^−/− mutant embryos. Combined with previous data, these results show that Gcm2 expression is down stream of this transcriptional network.

Fig. 7. Expression of the Hoxa3-Pax1/9-Eya1 pathway in wild-type and Gcm2^−/− mutant embryos.

Whole mount in situ hybridization (A, B, C, D, G, and H) or LacZ staining (E and F) was performed on wild-type (A, C, E and G) and Gcm2^−/− (B, D, F, and H) embryos at E10.5 for Hoxa3 (A and B), Pax1 (C and D), Pax9 (E and F), and Eya1 (G and H). The expression of these genes in the 3^rd pouch was normal in Gcm2 mutants. The 3^rd pouch is indicated in each panel (p3).

Our previous data have implicated the Shh and Bmp4 signaling pathways in the patterning of the organ-specific domains in the 3^rd pouch, with Shh required for parathyroid domain formation and Gcm2 expression, and opposing Bmp4 in the presumptive thymus domain (Moore-Scott and Manley, 2005; Patel et al., 2006). The Bmp antagonist Noggin is also expressed in the dorsal anterior 3^rd pouch overlapping with Gcm2 at E10.5-11.5 (Patel et al., 2006). In situ hybridization for Shh and Ptc in wild-type and Gcm2^−/− mutant embryos did not show any difference (Figs. 8A-D), consistent with this signaling pathway functioning upstream Gcm2. We used Bmp4^lacZ and Noggin^lacZ transgenic mouse strains to perform a detailed study of Bmp4 and noggin expression from E10.5-11.5 in the wild-type and Gcm2^−/− mutant embryos. Both Bmp4 and noggin expression were unchanged in wild-type and Gcm2^−/− mutant embryos at all stages tested (Fig. 9A-L), suggesting that this pathway is also not affected by Gcm2 mutation, and confirming co-localization of Bmp4 with presumptive thymus cells and normal patterning of the pouch and primordium in the Gcm2^−/− mutants.

Fig. 8. Expression of Shh and Ptc1 in the 3^rd pouch of wild-type and Gcm2^−/− mutant embryos.

Section in situ hybridization for Shh (A and B) and Patched1 (C and D) was performed on transverse paraffin sections from wild-type (A and C) and Gcm2^−/− (B and D) embryos at E10.5. The expression of these genes in the 3^rd pouch was normal in Gcm2 mutants. Dorsal is up. p3, third pouch.

Fig. 9. Expression of Bmp4 and Noggin in the 3^rd pouch of wild-type and Gcm2^−/− mutant embryos.

Whole mount LacZ staining for Bmp4^LacZ (A- F) and Noggin^LacZ (G-L) was performed on wild-type (A, C, E, G, I, and K) and Gcm2^−/− (B, D, F, H, J, and L) at E10.5 (A, B, G and H), E11 (C, D, I and J), and E11.5 (E, F, K and L). In panels (A-F), arrows indicate Bmp4 expression restricted to the ventral/posterior thymus domain in the 3^rd pouch (p3) and common primordium. In (E and F), the Bmp4-negative dorsal domain is indicated with a white arrow. In (G-L), arrows indicate dorsally restricted noggin domain. Arrowheads in (K and L) indicate the ventral thymus domain.

Discussion

Our data suggest that Gcm2 is required for the differentiation of parathyroid precursor cells in the parathyroid-specific domain, but is not required for initial patterning or initial expression of differentiation markers of the parathyroid domain in the 3^rd pharyngeal pouches and common parathyroid/thymus primordia. Our data also show that Gcm2 acts downstream of the known transcription and signaling pathways that function in parathyroid/thymus organogenesis. In this gene network, Gcm2 acts as a parathyroid-specific regulator gene for parathyroid differentiation, such that in the absence of Gcm2, the parathyroid precursor cells form and express initial differentiation markers, but cannot complete differentiation, and subsequently undergo apoptosis.

Based on its expression pattern, the initial description of the Gcm2 mutant phenotype in mouse, the Shh mutant phenotype, and the role of the Drosophila Gcm gene in cell fate specification, Gcm2 has been proposed by us and others to act as the master regulator gene that establishes the initial specification of the parathyroid domain (Balling and Erben, 2000; Berg, 2002; Manley and Blackburn, 2004). This role for Gcm2 predicted that aparathyroidsim in the Gcm2 mutants would be due to failure to specify the parathyroid domain in the third pouch. This failure might then result in either early apoptosis of the cells that would normally form the parathyroid domain in the third pouch, or transformation of the parathyroid domain to a thymus fate.

Our results did not support either of these predicted roles for Gcm2. The cell fate analysis of the 3^rd pouch-derived parathyroid/thymus primordium showed that the parathyroid domain formed and was morphologically normal before E12.5 in Gcm2^−/− mutants. Gcm2 also did not act as a binary switch between parathyroid and thymus fates, since the presumptive parathyroid domain did not express the thymus-specific marker Foxn1 in Gcm2^−/− mutants, and did express the early parathyroid differentiation markers CaSR and CCL21. The Foxn1 expressing domains were also a similar size in Gcm2^−/− mutants and wild-type controls, and Bmp4 expression in the presumptive thymus domain was also normal in Gcm2^−/− embryos. These results further showed that Gcm2 does not normally act to suppress thymus fate or thymus-specific gene expression in the parathyroid domain in the third pouch.

Our results show Gcm2 is required for the differentiation of the parathyroid domain after it is formed. Among three different parathyroid cell marker genes we test in Gcm2^−/− mutants, we found three different regulation models. Gcm2 is required for initial Pth expression at E11.5. On the other hand, the initial expression of CasR and CCL21 genes is at least somewhat independent of Gcm2, but do require Gcm2 to maintain their expression. For CasR, this may not be surprising, as it is initially expressed in all four pouches at E10.5 – in this case, Gcm2 seems to take the role of maintaining expression only in the parathyroid domain. However, CCL21 expression is initially restricted to the 3^rd pouch, but only partially depends on Gcm2 for full initial expression levels. The expression analysis of these three parathyroid differentiation markers reveals a surprising complexity in the regulation of parathyroid differentiation, and indicates that at least one other transcription factor is required to establish correct initial expression of these factors. The maintenance of all of these genes at later stages may also require Gcm2, since Gcm2 expression is maintained in adult parathyroid cells, although that remains to be experimentally determined.

Gcm2 not only regulates the differentiation procedure of parathyroid precursor cells, it is also required for their survival. This phenotype may reflect a direct role for Gcm2 in promoting cell survival. Our previous analysis of Hoxa3^+/−Pax1^−/− compound mutants showed that loss of Gcm2 expression after E11.5 resulted in progressive loss of parathyroids (Su et al., 2001). Interestingly, there was a delay of up to two days before this phenotype was evident, similar to the time delay between initial expression of Gcm2 at E9.5 and cell death at E12 in the Gcm2 mutants. Alternatively, in the absence of normal differentiation parathyroid cells either fail to acquire responsiveness to survival signals in the environment, or activate a default apoptotic fate.

The restricted expression of several genes in the presumptive parathyroid domain even in the absence of Gcm2 strongly suggests that this domain is specified and initiates parathyroid differentiation in these mutants. If Gcm2 does not specify the parathyroid domain, what does? From its expression pattern and its position upstream of Gcm2, Tbx1 is a likely candidate to specify the parathyroid domain in the 3^rd pouch. At E10.5, Tbx1 expression is restricted to the dorsal and anterior part in the 3^rd pouch endoderm that will become the parathyroid domain (Manley et al., 2004; Vitelli et al., 2002; Zhang et al., 2005), and remains restricted to the parathyroid domain in the parathyroid/thymus primordia (Fig. 6E). This expression pattern is consistent with Tbx1 functioning in early parathyroid organogenesis. The Shh pathway is also required to establish the parathyroid domain, as Gcm2 expression is lost and the thymus domain extended into the pharynx in Shh mutants (Moore-Scott and Manley, 2005). Tbx1 is also thought to be a down stream target of the Shh signaling pathway (Garg et al., 2001; Yamagishi et al., 2003). Taken together, these data support a model in which specification of the parathyroid domain is regulated through a Shh-Tbx1-Gcm2 pathway.

In addition to the Shh pathway, a Hoxa3-Pax1/9-Eya1 pathway is also required for 3^rd pouch patterning and initiation of parathyroid/thymus primordium formation, and may directly regulate Gcm2 expression. Gcm2 expression is absent or down regulated in Hoxa3^−/− and Eya1^−/− embryos at E10.5 (our unpublished data)(Xu et al., 2002). Consistent with this data, the expression of the Hoxa3-Pax1/9-Eya1 pathway in Gcm2^−/− embryos is normal at E10.5. How the Shh-Tbx1 and Hoxa3-Pax1/9-Eya1 pathways converge to regulate Gcm2 and parathyroid cell fate is a key remaining question in understanding parathyroid organogenesis.

The current results indicate that in spite of their different timing for initial expression, both Gcm2 and Foxn1 play analogous roles in the development of the parathyroid and thymus domains in the common primordium – both genes are not required for initial specification of the organ domains or initial primordium formation, but are required for subsequent tissue-specific differentiation events. However, while Foxn1 initial expression is at E11.25, similar to the appearance of the first identifiable phenotype, Gcm2 expression in the 3^rd pouch endoderm is initiated from E9.5, well before the parathyroid/thymus primordium is formed. The decreased initial expression of CCL21 does indicate some role for Gcm2 at least at E10.5, even though it is not required for domain specification. The earlier expression could also be due to differences in the timing of expression of upstream regulation of the two genes. The location of parathyroids within the endoderm of the pharyngeal pouches is evolutionarily conserved, and in teleost fishes is required for gill bud formation, which precedes thymus organogenesis in these animals (Graham et al., 2005; Okabe and Graham, 2004; Schorpp et al., 2002; Willett et al., 1997). The molecular pathways that specify the parathyroid domain may turn on earlier than those of the thymus due to this evolutionary legacy. A consequence of this earlier expression may also be to carve out the parathyroid domain within the third pouch and protect it from a thymus fate.