Parathyroid hormone (PTH), which is secreted by the parathyroid glands, is a critical regulator of serum calcium and phosphate homeostasis. PTH increases the serum calcium concentration by directly inducing calcium release from bone and stimulating calcium reabsorption in the kidney. In addition, it stimulates 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] synthesis by inducing expression of 25-hydroxyvitamin D3 1α-hydroxylase (Cyp27b1) in the kidney, and 1,25(OH)2D3 then enhances intestinal calcium absorption. PTH also decreases phosphate reabsorption in the kidney, whereas 1,25(OH)2D3 promotes intestinal phosphate absorption. Excessive secretion of PTH (hyperparathyroidism) causes hypercalcemia, whereas a defect in PTH secretion (hypoparathyroidism) owing to a PTH gene mutation or disorders in parathyroid development causes hypocalcemia. In the fetus, PTH positively regulates calcium concentration as well as skeletal mineralization.1, 2
In mice, parathyroid glands develop with the thymus from bilateral common primordia derived from the third pharyngeal pouch endoderm.3 A transcriptional network involving Hoxa3,4–6 Pax1/Pax9,7–9 Eya1,10 and Six1/Six411 regulates the third pharyngeal pouch patterning and formation of the parathyroid/thymus common primordia. Under the control of these factors, dorsal/anterior and ventral/posterior parts of the primordium are specified to become parathyroid and thymus domains, respectively, at about embryonic day 9.5 (E9.5).12–14 Each primordium begins to separate into a single parathyroid gland and a thymic lobe around E13.5. By E15.5, the parathyroid and thymus have migrated to their adult locations, near the thyroid glands and in the chest cavity, respectively.
In the presumptive parathyroid domain of the common primordium, expression of Glial cell missing 2 (Gcm2) transcription factor begins at E9.5. Loss of the gcm2 gene in mice results in parathyroid primordial cell death around E12.5 and parathyroid agenesis without affecting thymic development.12, 15, 16 Therefore, Gcm2 functions as a master regulator of parathyroid specification and controls its early organogenesis. In humans, hypomorphic and dominant-negative mutations in GCM2 lead to a decrease in PTH secretion and cause hypoparathyroidism.17–20 Because Gcm2 expression persists in developing and mature parathyroid cells, Gcm2 also may have roles in later steps of parathyroid development, including PTH gene expression. However, the genes that regulate these events in parathyroid development remain unknown.
v-maf musculoaponeurotic fibrosarcoma oncogene homologue B (MafB) is a member of the Maf family of bZip transcription factors and plays important roles in the developmental processes of various tissues, as well as in cell-type-specific gene expression. For example, disruption of the mafB gene in mice has shown that its encoded protein regulates respiratory rhythmogenesis in the brain,21 monocyte and osteoclast differentiation,22–25 podocyte differentiation in the kidney,22, 26 and maturation of pancreatic islet α and β cells.27 MafB stimulates expression of tissue-specific genes (eg, F4/80 in macrophages, glucagon in α cells, and insulin in β cells) by binding directly to Maf-recognition element (MARE)–related sequences in their promoter or enhancer regions.22, 28–32
In this study we report that MafB is expressed in developing and mature parathyroid glands and examine the roles of MafB in the regulation of parathyroid development and PTH gene expression. We found that MafB regulates later steps of parathyroid development, that is, separation from the thymus and migration toward the thyroid gland. We also found that MafB and Gcm2 interact with each other, bind directly to an evolutionally conserved region of the PTH gene promoter, and synergistically activate transcription.
Materials and Methods
The generation of mafB null mutant mice and genotyping have been described previously.22 The mafB mutant mice were backcrossed into C57BL/6J mice. To generate gcm2 null mutant mice, an entire coding region of the gcm2 gene, encompassing exons 1 through 5, was replaced by the internal ribosome-binding site (IRES) and lacZ gene followed by a neomycin-resistance gene cassette. Heterozygous mutant mice were generated from embryonic stem cells (129SVJ/RW-4) and backcrossed to the CD1 background. Detailed information for the generation and genotyping of the gcm2 null mutant mice will be provided elsewhere (SH, KI, TH, and YH, manuscript in preparation). All experiments were performed according to the guidelines for the care and use of laboratory animals of the Nara Institute of Science and Technology, National Institute for Physiological Sciences, and Tokyo Medical and Dental University.
Embryos were fixed overnight in PBS containing 4% paraformaldehyde and then processed for frozen sections. For immunostaining, frozen sections were incubated with primary antibodies and then visualized with Alexa488- or Alexa594-conjugated secondary antibodies (Invitrogen, Carlsbad, CA, USA). Nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI). The following primary antibodies were used: rabbit anti-MafB (Bethyl Laboratories, Montgomery, TX, USA), goat anti-MafB (P-20; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), rabbit anti-GFP (Invitrogen), mouse anti-GFP (Invitrogen), rat anti-CCL21 (R&D Systems, Minneapolis, MN, USA), goat anti-CaSR (F-19; Santa Cruz Biotechnology), and goat anti-PTH (N-18; Santa Cruz Biotechnology).
Serum PTH measurement
Whole blood was collected from neonatal mice (postnatal day 0, P0) within 2 hours of birth. If the serum volume obtained from an individual mouse was not enough, sera from two mice were pooled together to obtain sample volumes of 25 µL. Each intact PTH level was measured by the two-antibody method using a mouse intact PTH ELISA kit (Immutopics International, San Clemente, CA, USA).
Real-time RT-PCR analysis
Total RNA was isolated from kidneys of P0 mice using Trizol reagent (Invitrogen) and then subjected to reverse transcription using oligo-dT primer. The resulting cDNAs were analyzed using a LightCycler 480 System (Roche Applied Science, Indianapolis, IN, USA). The following primer sets were used: cyp27b1, 5'-GGG AGA CGC TTG GCA GAG CTT GAG-3' and 5'-AGT CCG GGT CAT GGG CTT GAT AGG-3'; gapdh, 5'-GGT GAA GGT CGG TGT GAA CGG ATT T-3' and 5'-TCC CGT TGA TGA CAA GCT TCC CAT T-3'.
Bone histology and micro–computed tomographic (µCT) analysis
Bone histology was analyzed by alcian blue/alizarin red staining of neonatal skeleton and von Kossa staining of neonatal tibia sections, as described previously.33, 34 CT scanning was performed using a ScanXmate-A100S Scanner (Comscantechno, Kanagawa, Japan). 3D microstructural image data were reconstructed and structural indices were calculated using TRI/3D-BON software (RATOC, Osaka, Japan). Bone morphometric analysis was performed using the full-length femurs of postnatal mice.33, 34
A series of reporter plasmids containing the human PTH gene promoter region was constructed from pPTHg108 plasmid (a gift from Dr HM Kronenberg).35 Mutations were introduced by site-directed overhang extension PCR mutagenesis. To construct an expression plasmid for Gcm2 (pHygEF2/gcm2), the entire open-reading frame of mouse gcm2 was amplified by RT-PCR and inserted into the pHygEF2 vector. Expression plasmids for hemagglutinin (HA)–tagged human MafB (pHygEF2/HA-h-mafB) and Renilla luciferase (pEF-Rluc) have been described previously.36, 37
BHK21 cells grown in 24-well plates were transfected with 0.7 µg of plasmid DNA (0.05 µg of luciferase reporter plasmid, 0.6 µg of expression plasmid, and 0.05 µg of pEF-Rluc) using 2 µL of Lipofectamine 2000 reagent (Invitrogen). Cells were harvested 24 hours after transfection. Firefly and Renilla luciferase activities were measured using a dual luciferase assay system (Promega, San Luis Obispo, CA, USA). Statistical significance was calculated by ANOVA (p < 0.05). Data represent the mean ± SE of three independent experiments.
For immunoprecipitation, a FLAG tag was fused to the amino terminus of Gcm2 (pHygEF2/FLAG-gcm2). HeLa cells grown in 6-well plates were transfected with 2.0 µg of pHygEF2/HA-h-mafB and 1.2 µg of pHygEF2/FLAG-gcm2 using 8 µL of Lipofectamine 2000 reagent (Invitrogen), and cell extracts were prepared by cell lysis in NETN buffer (150 mM NaCl, 10 mM EDTA, 10 mM Tris-HCl [pH 8.0], and 0.1% NP-40), followed by sonication and centrifugation. The extract was subjected to immunoprecipitation using anti-FLAG agarose (Sigma-Aldrich, St Louis, MO, USA) and analyzed by immunoblot with anti-FLAG (Sigma-Aldrich) or anti-HA (MBL, Nagoya, Japan) antibodies.
Electrophoretic gel mobility shift assay (EGMSA)
Nuclear extracts were prepared from BHK21 cells transfected with pHygEF2 or pHygEF2/HA-h-mafB, as described previously.38 To produce a GST fusion protein of Gcm2 in Escherichia coli, a DNA fragment of gcm2 encoding the DNA-binding domain (DBD; amino acids 1 to 188) was inserted into the pGEX-6P-1 vector (GE Healthcare, Piscataway, NJ, USA). GST and GST-Gcm2 (DBD) proteins were purified by glutathione sepharose 4B column chromatography.
The nuclear extracts and GST fusion proteins were used for EGMSA as described previously.38 An oligonucleotide probe containing ECR4-MARE was made from the oligonucleotides 5'-biotin-GAT CCA AAG CTG TGCTCAGCTCTT CTC TGT CA-3' and 5'-GAT CTG ACA GAG AAGAGCTGAGCA CAG CTT TG-3'. The ECR4-GcmBS probe was made from the oligonucleotides 5'-biotin-GAT CCA GTC TCT TGT GCCCACATA TCC CCT AA-3' and 5'-GAT CTT AGG GGA TATGTGGGC ACA AGA GAC TG-3'. The top-strand nucleotide sequences of the mutated oligonucleotides were 5'-GAT CCA AAG CTG TGC gacta TCTTCT CTG TCA-3' (mut MARE) and 5'-GAT CCA GTC TCT TGT GaaaACATAT CCC CTA A-3' (mut GcmBS).
Expression of MafB in parathyroid glands during development
Mouse embryonic sections were examined by immunostaining to determine MafB expression in the developing parathyroid. At E11.5, MafB expression was first detected in the dorsoanterior domain of the third pharyngeal pouch (Fig. 1B). MafB was expressed in the nuclei of cells that demonstrated cytoplasmic staining of chemokine CCL21, one of the earliest parathyroid differentiation markers,39 indicating that MafB was expressed in the parathyroid domain of the primordium (Fig. 1B). Coimmunostaining with the late parathyroid differentiation marker PTH40 revealed that MafB was expressed in the nuclei of parathyroid cells that were positive for cytoplasmic PTH staining and that the MafB expression persisted during embryonic development (Fig. 1C, D) and into adulthood (Fig. 1E, F). MafB expression was not detected in the third pharyngeal pouch at the earlier E10.5 stage (Fig. 1A). These results suggest that MafB is involved in parathyroid development and function.
MafB expression is not initiated in the parathyroid primordium of gcm2−/− mice
MafB expression was next examined in gcm2−/− mice. Gcm2 is the master regulator of parathyroid development, and gcm2−/− mice lack parathyroid glands owing to apoptosis of primordial cells around E12.5.16, 40 Expression of gcm2 mRNA has been shown to begin in the parathyroid primordium at E9.5,12 2 days before MafB protein expression was detected (E11.5; Fig. 2A). However, MafB expression was not detected at E11.5 in the parathyroid primordium of gcm2−/− mice (Fig. 2B). This result, together with the temporal expression patterns of Gcm2 and MafB and the phenotypes of these knockout mice (see below), suggests that MafB functions downstream of Gcm2 in the regulation of parathyroid development.
Ectopic location of parathyroids in mafB+/− and mafB−/− mice
To explore the roles played by MafB in parathyroid development, we analyzed mafB:GFP knock-in mice in which the coding sequence for mafB had been replaced by GFP, and the location of mafB expression could be monitored by GFP fluorescence.22 In wild-type mice at E18.5, the parathyroid glands are attached to the thyroid gland (shown schematically in Fig. 3E, left). In contrast, in our observations, the parathyroid glands were not located near the thyroid glands in mafB+/− or mafB−/− mice. In mafB+/− mice at E18.5 (Fig. 3A, C), bilateral clusters of GFP+ cells were found between the thyroid and thymus and attached to the carotid arteries. Because these GFP+ cells expressed the parathyroid markers PTH (Fig. 3F) and CaSR (Fig. 3G), they were concluded to represent ectopically located parathyroid cells. These bilateral ectopic parathyroids were observed in 100% of the mafB+/− E18.5 embryos and adults investigated.
In contrast, GFP+ cells were found within the anterior region of each thymic lobe at E18.5 in all mafB−/− mice (Fig. 3B, D). These cells were concluded to be of parathyroid origin because they expressed the early parathyroid differentiation marker CaSR (Fig. 3H). However, they did not express detectable levels of the late differentiation marker PTH (Fig. 3I), suggesting that these cells were immature parathyroid cells. No gross abnormalities were observed in the thymus or thyroid glands of mafB+/− or mafB−/− mice, aside from the incorporation of parathyroid-like cells into the thymus of mafB−/− mice. Later stages of mafB−/− mice could not be analyzed because they died within 24 hours of birth owing to central respiratory defects.21 Taken together, our observations indicate that MafB regulates parathyroid differentiation and localization.
Impaired PTH secretion and decreased bone mineralization in mafB−/− mice
We next measured serum PTH levels of wild-type mafB+/− and mafB−/− mice (Fig. 4A). At postnatal day 0 (P0), serum PTH levels of wild-type mice were higher (820 pg/mL) than those of embryos or adults (<100 pg/mL), probably reflecting a decline in circulating calcium levels owing to a lack of placental calcium supply after disconnection from the mother.41, 42 We found that serum PTH levels of mafB−/− mice were much lower (30 pg/mL) than those of wild-type mice, indicating that PTH secretion is impaired in mafB−/− mice. Serum PTH levels of mafB+/− mice also were lower (510 pg/mL) than those of wild-type littermates, suggesting that the ectopically localized mafB+/− parathyroid cells were not fully functional. Expression levels of renal cyp27b1 mRNA, one of the target genes activated by PTH, also were reduced greatly in mafB−/− and partially in mafB+/− mice (Fig. 4B).
Bone histology of wild-type and mafB−/− mice also was analyzed. The intact skeletons of these mice at P0 were stained with alizarin red and alucian blue to visualize mineralized bone and cartilage, respectively, and we observed that mafB−/− mice showed a grossly normal skeletal phenotype (Fig. 5A). However, closer examination of bone sections by von Kossa staining revealed that mafB−/− tibias showed less black staining, indicating that less mineral was present (Fig. 5B). Furthermore, in mafB−/− mice, tibias were shorter and their hypertrophic cartilage (HC) was enlarged (Fig. 5B).
µCT analysis of femurs revealed that bone volume per tissue volume (BV/TV), trabecular number (Tb.N), and trabecular thickness (Tb.Th) were reduced in mafB−/− mice (Fig. 5C), indicating that bone mineralization was reduced in mafB−/− mice. These results together suggest that in mafB−/− mice, PTH secretion was impaired and thus bone mineralization was reduced. However, considering that MafB is also expressed in osteoclasts and negatively regulates osteoclastgenesis,24, 25 further analyses are required to distingish cell-type-specific roles of MafB in perinatal bone mineralization.
Synergistic activation of PTH promoter by MafB and Gcm2
The observation that PTH expression was greatly reduced in mafB−/− parathyroid cells prompted us to test whether MafB directly regulates PTH gene expression. A computational analysis of the PTH genomic loci using the ECR browser (http://ecrbrowser.dcode.org/) identified four evolutionarily conserved regions (ECRs; labeled ECR1 through ECR4 from proximal to distal) within 5 kb of the 5'-flanking region of the transcription start site (Fig. 6A, above). Luciferase reporter assays then were performed using a human PTH promoter construct spanning these regions (−5341 to +52). The series of reporter constructs used in this analysis is shown schematically in Fig. 6A (below). In addition to MafB, the effect of Gcm2 was tested in this assay because parathyroid expression of Gcm2 also persists during development and into adulthood. Expression of MafB alone was found to result in a weak but significant stimulation of the reporter containing ECR1 to ECR4 (pGL4/PTH-p [−5341]; Fig. 6B, left). Gcm2 alone had no effect, but coexpression of MafB and Gcm2 synergistically stimulated reporter activity. A reporter construct containing ECR1 to ECR3 but lacking ECR4 (pGL4/PTH-p [−3879]) completely lost responsiveness to MafB and Gcm2 (Fig. 6B, right).
To further determine the role of ECR4, a reporter plasmid containing this region (−4840 to −4289) and the proximal basal promoter (−160 to +52) was constructed (pGL4/ECR4 [wt]). We found that the activity of this reporter also was stimulated weakly by MafB alone and synergistically by MafB and Gcm2 (Fig. 6C, left). Inspection of the nucleotide sequence of ECR4 revealed a highly conserved putative Maf-recognition element (MARE) and a Gcm2-binding sequence (GcmBS) located in close proximity (Fig. 6D). Mutation of the ECR4-MARE (pGL4/ECR4 [mut MARE]) resulted in a complete loss of stimulation by MafB alone and by MafB and Gcm2 (Fig. 6C, middle). This finding indicates that ECR4-MARE is indispensable for activation by MafB and Gcm2 and suggests that the synergistic effect of Gcm2 depends on MafB binding to ECR4-MARE. Mutation of the ECR4-GcmBS (pGL4/ECR4 [mut GcmBS]) did not affect activation by MafB alone but significantly reduced the synergy between MafB and Gcm2 (Fig. 6C, right). This finding indicates that ECR4-GcmBS, at least in part, mediates synergistic action by Gcm2. These results together suggest that MafB and Gcm2 cooperatively stimulate PTH gene expression through the MARE and GcmBS within the ECR4.
The domain requirements of MafB and Gcm2 for activation of the PTH promoter were examined next using a series of deletion mutants (Fig. 6E). Deletion of the transactivator (TA) domain from either MafB (ΔN) or Gcm2 (ΔC) resulted in a complete loss of synergistic action, indicating that the TA domains of both MafB and Gcm2 are required for the synergy. A Gcm2 mutant lacking the DNA-binding domain (ΔDBD) also lacked synergistic activity.
Physical interaction of MafB and Gcm2
Binding of MafB and Gcm2 to their respective putative target sequences within ECR4 was investigated next. Gel mobility shift assay using a labeled oligonucleotide probe containing the ECR4-MARE demonstrated that HA-tagged MafB protein binds to this element (Fig. 7A, left). A competition experiment using excess wild-type and mutant ECR4-MARE oligonucleotides demonstrated the specificity of MafB binding to the element (Fig. 7A, right). Similarly, GST-tagged Gcm2 protein specifically bound to the ECR4-GcmBS (Fig. 7B).
The observation that MafB and Gcm2 synergistically activate transcription by binding to their respective target sites, which are located in close proximity within ECR4, led us to explore the potential interaction between MafB and Gcm2. When coexpressed in HeLa cells, HA-tagged MafB protein was coimmunoprecipitated with FLAG-tagged Gcm2 protein (Fig. 7C), indicating that MafB physically interacts with Gcm2.
In this study, we demonstrated that MafB is expressed in developing and mature parathyroid glands. The parathyroid glands of mafB+/− mice were mislocalized between the thyroid gland and thymus, whereas those of mafB−/− mice remained attached to thymic lobes. The mafB−/− parathyroid cells did not express detectable PTH levels. Accordingly, in newborn mafB−/− mice, PTH secretion was impaired, renal cyp27b1 expression was decreased, and bone mineralization was reduced. MafB and Gcm2 also were found to interact with each other, bind to their respective binding sites within an evolutionarily conserved putative enhancer region (ECR4) of the PTH gene, and synergistically activate transcription.
Previous studies have shown that a genetic cascade involving Hoxa3,4–6 Pax1/Pax9,7–9 Eya1,10 and Six1/Six411 regulates the third pharyngeal pouch patterning and formation of the parathyroid/thymus common primordium until E9.5. Gcm2 expression then begins in the presumptive parathyroid domain of the primordium. Here we found that MafB expression begins in the parathyroid primordium at E11.5, 2 days after initiation of gcm2 mRNA expression (E9.5).12 This observation, together with our finding that MafB expression is lost in the parathyroid primordium of gcm2−/− mice, suggests that MafB acts downstream of Gcm2. Parathyroid primordial cells undergo apoptosis in gcm2−/− mice,40 whereas mafB−/− parathyroid cells survive and express early differentiation marker CaSR but lack detectable PTH expression. Thus MafB seems to regulate a part of the genetic program downstream of Gcm2 and controls later steps of parathyroid development. It remains to be determined whether Gcm2 activates mafB expression directly.
In humans, gcm2 gene mutations lead to reduced PTH secretion and hypoparathyroidism.17–20 Although Gcm2 has been reported to activate CaSR gene expression by binding directly to its promoter region,43 evidence for its regulation of PTH expression is lacking. Furthermore, little is known about the mechanisms underlying tissue-specific PTH expression in the parathyroid gland. This study demonstrated that MafB and Gcm2 bind directly to their respective binding sites within an evolutionarily conserved region (named ECR4) of the PTH promoter, physically interact with each other, and cooperatively activate transcription. We thus propose that MafB and Gcm2 are important components in the establishment of parathyroid-specific PTH expression. Furthermore, MafB and Gcm2 expression persists after parathyroid morphogenesis, suggesting that they are also required for the maintenance of parathyroid functions by regulating a parathyroid-specific gene expression program. However, a previous report has demonstrated that gcm2 expression is reduced in human parathyroid adenomas of primary hyperparathyroidism despite increased PTH production,44 suggesting that the role of Gcm2 and possibly MafB in mature parathyroid is more complex.
Our analyses of mafB+/− and mafB−/− mice demonstrated that MafB regulates multiple aspects of parathyroid development. In mafB−/− mice, parathyroid cells remained attached to thymic lobes, indicating that MafB is required for the separation of parathyroid primordium from the thymus. The parathyroid glands of mafB+/− mice were separated from thymic lobes but mislocalized between thyroid gland and thymus, indicating that MafB also regulates parathyroid migration toward the thyroid. How the processes of separation and migration are regulated under the control of MafB remains to be determined. Notably, the location of mafB+/− ectopic parathyroids was restricted to an area between the thyroid and thymus. Furthermore, the ectopic parathyroids always were attached to carotid arteries, suggesting that these arteries are a permissive area for parathyroid migration.
In summary, the results presented here demonstrate that MafB plays critical roles in the regulation of many steps in parathyroid development, at least in part in cooperation with Gcm2. Further analyses will be required to determine the detailed molecular mechanisms of underlying parathyroid developmental processes.
All the authors state that they have no conflicts of interest.
This work was supported by Grants-in-Aid for Scientific Research on Priority Areas, for Scientific Research (C), and for the Global COE Program from the Ministry of Education, Culture, Sports, Science and Technology and grants from the Mitsubishi Foundation and the Takeda Science Foundation to KK.
Authors' roles: Study design: AKK and KK. Study conduct: KK. Sample preparation: MH, TM, SH, KI, TH, YH, and ST. Data collection: AKK, MM, FS, KN, and KK. Data analysis and interpretation: AKK, IH, KN, HT, and KK. Drafting manuscript: KK. Revising manuscript content: AKK, IH, KN, HT, SH, KI, TH, YH, and SH. All authors approved final version of the manuscript. KK takes responsibility for the integrity of the data analysis.