The parathyroid glands are the only known source of circulating parathyroid hormone (PTH), which initiates an endocrine cascade that regulates serum calcium concentration1. Glial cells missing2 (Gcm2), a mouse homologue of Drosophila Gcm, is the only transcription factor whose expression is restricted to the parathyroid glands2,3,4,5. Here we show that Gcm2-deficient mice lack parathyroid glands and exhibit a biological hypoparathyroidism, identifying Gcm2 as a master regulatory gene of parathyroid gland development. Unlike PTH receptor-deficient mice, however, Gcm2-deficient mice are viable and fertile, and have only a mildly abnormal bone phenotype. Despite their lack of parathyroid glands, Gcm2-deficient mice have PTH serum levels identical to those of wild-type mice, as do parathyroidectomized wild-type animals. Expression and ablation studies identified the thymus, where Gcm1, another Gcm homologue, is expressed, as the additional, downregulatable source of PTH. Thus, Gcm2 deletion uncovers an auxiliary mechanism for the regulation of calcium homeostasis in the absence of parathyroid glands. We propose that this backup mechanism may be a general feature of endocrine regulation.
Serum calcium is essential for many physiological functions including muscle contraction, blood coagulation, neuromuscular excitability and mineralization of bone, a tissue that contains 99% of the extracellular calcium6. Serum calcium concentration is strictly regulated by a complex endocrine pathway of which PTH is an essential component. PTH increases serum calcium concentration by favouring calcium release from bone, stimulating renal calcium reabsorption and indirectly enhancing intestinal calcium absorption by favouring 1,25(OH)2 vitamin D3 synthesis1 (Fig. 1a). No molecular determinant exclusively controlling the development of parathyroid glands has been identified. The transcription factor Gcm2, a mouse homologue of Drosophila Gcm2,3,4,5, is expressed exclusively in the parathyroid glands, and so may be a specific regulator of parathyroid gland development5. We deleted Gcm2 using embryonic stem cell technology (Fig. 1b–d) to test this proposal.
Heterozygous mice are phenotypically normal. Crosses between these heterozygotes produced homozygous mutant Gcm2m1/ Gcm2m1 mice at the predicted mendelian frequency. Thirty per cent of the Gcm2-deficient mice died shortly after birth (Fig. 2a) owing to severe hypocalcaemia (as low as 3 mg dl-1; Fig. 2b). The remaining Gcm2-deficient mice were viable, fertile and developed like their wild-type littermates. These mice had milder hypocalcaemia associated with hyperphosphataemia (Fig. 2b, c). There was also increased calcium elimination in the urine without evidence of renal failure (data not shown). These features are characteristic of hypoparathyroidism.
Eighty-five per cent of the pups generated by intercrossing Gcm2-deficient mice died within the first day of life (Fig. 2d). When we crossed Gcm2-deficient males with females heterozygous for the mutation, 78% of the Gcm2-deficient pups survived. When placed with foster mothers, 67% of the offspring of Gcm2-deficient parents survived ( Fig. 2d). These numbers are similar to those observed when intercrossing mice heterozygous for the Gcm2 mutation (Fig. 2a ). In contrast, crossing Gcm2-deficient females with heterozygous and homozygous males resulted in the death of 85% of the Gcm2-deficient pups (Fig. 2d). These data show that the lethality of the Gcm2-deficient pups born from Gcm2-deficient parents is of maternal origin, probably due to the low extracellular calcium concentration in Gcm2-deficient females and the decreased milk production in hypoparathyroidic rodents7. This was confirmed by crossing Gcm2-deficient males with Gcm2-deficient females whose serum calcium concentration had been corrected by treatment with vitamin D3, another hypercalcaemic hormone8. In this case 64% of the pups survived ( Fig. 2d). That the mutant phenotype reflects altered calcium and not PTH levels is also shown by the fact that there was no measurable PTH in the milk of either Gcm2-deficient or wild-type mice. The level of PTH related peptide (PTHrP) in both milk and serum was identical in wild-type and Gcm2-deficient mice (data not shown).
Regardless of the genotype of the parents, histological examination of Gcm2-deficient mice failed to detect any parathyroid glands ( Fig. 3a). Likewise, reverse transcriptase polymerase chain reaction (RT-PCR) did not detect any PTH-expressing cells in the thyroid and neck of these animals (Fig. 3b and data not shown). In Gcm2-deficient mice, PTH expression is not detectable at any time point analysed (see Supplementary Information). There was no evidence of increased apoptosis in the neck region (data not shown), and no other anatomical structures were affected in Gcm2-deficient mice. These data identify Gcm2 as the first specific regulator of parathyroid gland differentiation.
PTH has major effects on bone remodelling9,10; we therefore analysed the skeletons of six-month-old Gcm2-deficient mice. The number and thickness of trabeculae were increased in these mice compared with wild-type littermates (Fig. 3c). Osteoblast and osteoclast surfaces were both reduced in Gcm2-deficient mice, as has been shown in WT-TPTX animals11 (Fig. 3d, e), but the net result was a relative increase in bone volume without any cartilage abnormalities (Fig. 3c and data not shown). This low turnover—that is, a low cell number, high bone mass phenotype—could be rescued by continuous PTH infusion (Fig. 3c–e). The bone phenotype of the Gcm2-deficient mice is mildly abnormal compared with that observed in mice and humans deficient for the PTH/PTHrP receptor12,13,14, perhaps indicating that signal transduction through the PTH/PTHrP receptor was not completely abrogated in these mice, even though they lack parathyroid glands. This in turn suggested the existence of either a distinct, functionally redundant, ligand for the PTH/PTHrP receptor or an auxiliary source of PTH. To test the latter hypothesis we measured PTH in Gcm2-deficient mice using an assay that does not cross-react with PTHrP.
Serum PTH levels in Gcm2-deficient and wild-type mice were nearly identical and well above the limit of detection of the assay ( Fig. 4a). As an internal positive control we measured PTH levels in Vitamin D receptor-deficient mice which have high serum PTH15 (Fig. 4a). In the face of a marked hypocalcaemia, the level of serum PTH in Gcm2-deficient mice was too low to correct the hypoparathyroidism. Continuous PTH infusion could correct the hypocalcaemia of the Gcm2-deficient mice (Fig. 4b). There was no compensatory increase in the serum concentration of PTHrP16 or of 1,25(OH)2 vitamin D3, another hypercalcaemic hormone, in Gcm2-deficient mice (Fig. 4c, d). That the serum concentration of 1,25(OH)2 vitamin D3 was normal indicates that the PTH in the serum of the Gcm2-deficient mice was active, as PTH is the main regulator of 1,25(OH)2 vitamin D3 synthesis8. T3, T4 and calcitonin serum levels were normal in Gcm2-deficient mice (data not shown).
To test whether the auxiliary source of PTH in Gcm2-deficient mice also exists in wild-type mice we measured serum calcium and PTH for various periods of time in WT-TPTX mice kept on a regular diet. The serum calcium of these animals dropped to 5 mg dl-1 and their phosphorus level increased (Fig. 4e and data not shown), demonstrating the existence of a hypoparathyroidism. But PTH was present in the serum of every WT-TPTX animal and increased over time (Fig. 4e), showing that the absence of parathyroid glands does not lead to the absence of PTH in mice.
To locate the auxiliary source of PTH we first performed RT-PCR on multiple tissues. PTH transcripts were found in the hypothalamus, as previously described17, but also in the thymus, the other derivative of the third pharyngeal pouch (Fig. 4f). The identity of the whole length of this transcript as bona fide PTH was confirmed by DNA sequencing (data not shown). In situ hybridization revealed a cluster of PTH-expressing cells underneath the thymic capsule in wild-type and Gcm2-deficient mice (Fig. 5a and data not shown). These thymic PTH-expressing cells expressed the calcium sensing receptor (Casr) gene (Fig. 5a).
Two observations indicate that the thymus is the auxiliary source of PTH. First, Hoxa3-deficient mice, which lack parathyroid glands and thymus but have no known hypothalamic abnormalities18, also have no detectable PTH in their sera19. This indicates that the hypothalamus does not secrete PTH into the general circulation. Second, whereas thymectomy itself did not cause lethality and only 30% of Gcm2-deficient and WT-TPTX mice died, 100% of wild-type mice that were thyroparathyroidectomized and thymectomized died shortly after surgery (Figs 5b, 2a). The latter experiment provides functional evidence for thymic involvement in calcium metabolism in mice lacking parathyroid glands. To determine whether this second genetic pathway of PTH-expressing cell differentiation could involve Gcm1, the other mouse homologue of Drosophila Gcm (ref. 4), we looked for its expression in thymus and parathyroid glands. Gcm1 expression colocalized with PTH expression, indicating that Gcm1 could contribute to the differentiation of PTH-expressing cells in thymus (see Supplementary Information). We failed to detect Gcm1 expression in parathyroid glands (data not shown).
The absence of a high PTH serum level in the Gcm2-deficient and WT-TPTX animals despite hypocalcaemia suggested that the thymic PTH source already secreted its maximum amount of PTH and/or that it could not be regulated properly. Long-term treatment with vitamin D3, a hormone that increases serum calcium concentration and decreases PTH synthesis in wild-type animals20, had the same effects in Gcm2-deficient mice ( Fig. 5c), indicating that secretion of the thymic PTH could be downregulated by one physiological regulator. In contrast, treatment of Gcm2-deficient mice with sodium phosphate, which decreases serum calcium and increases PTH in wild-type mice21, failed to do so in Gcm2-deficient mice (see Supplementary Information). The absence of upregulation of PTH secretion by the thymic source in response to low serum calcium indicates that these cells, which do express Casr, secrete their maximum amount of PTH in the absence of parathyroid glands. This would be consistent with the very small size of this source compared with the size of the parathyroid glands (Fig. 5d).
From a developmental perspective this study identifies Gcm2 as the first gene, to our knowledge, specifically to control the differentiation of cells of the third pharyngeal pouch into parathyroid glands. Moreover, Gcm2 deletion uncovers the existence of an auxiliary genetic pathway, possibly involving another Gcm protein, that can achieve differentiation of PTH-producing cells in the thymus. From a physiological point of view, our results uncover a general feature of serum calcium regulation that has not been appreciated before (Fig. 5e). The existence of an auxiliary source of PTH accounts for the relatively mild biochemical and bone phenotype of the Gcm2-deficient mice. It also provides a genetic explanation for the previously suspected regulation of calcium and phosphate metabolism independently of the parathyroid glands and for the high incidence of PTH-producing tumours in thymus22. On the basis of our results we predict that a PTH-deficient mouse will have a more severe phenotype than the Gcm2-deficient mouse.
Beyond calcium regulation itself, the existence of an auxiliary source of PTH bears some resemblance to other endocrine-failure situations. In ovarian failure, for instance, the production of oestrogens by other tissues does not prevent menopause23. These two examples among others suggest that auxiliary sources of hormone are a common safety feature of endocrine regulation, implying that the phenotype of an endocrine organ failure will in most cases be less severe than the phenotype caused by the absence of the corresponding hormone.
We constructed a targeting vector containing 8 kilobases of homology to the region of the Gcm2 locus, a neo cassette for positive selection and a thymidine kinase cassette (TK) for negative selection24. The neo cassette replaces all four exons5. The vector was electroporated into AB-1 mouse embryonic stem (ES) cells and the ES clones were selected in the presence of G418 and FIAU24. DNA prepared from resistant clones was digested with XbaI and screened by Southern blot using a 0.8-kb XbaI– HindIII fragment as a probe. Correctly identified targeted clones were injected into C57BL/6 blastocysts and germline transmission was obtained as described24.
DnaseI-treated RNA isolated using Trizol (Gibco) was used for reverse transcription. PCR involved annealing at 60 °C for Gcm1, 56 °C for Gcm2 or 58 °C for Hprt and 67 °C for PTH for 35 cycles or 25 cycles for Hprt. The following primers were used: Gcm1 RT1F: 5′-GCACGAATTCAATG GAACTGGACGACTTTG-3′; Gcm1 RT1B: 5′-TAGCTGCTCAGA TCCACAGA-3′; Gcm2 F7: 5′-CGGAATTCCCTATTCTCGCTCTTAACCTTTGCC-3′; Gcm2 B7: 5′-GCTCTAGACCAGCAAGTTACAGGTCTGAGCTTC-3′; HPRT5: 5′-GTTGAGAGATC ATCTCCACC-3′; HPRT3: 5′-AGCGATGATGAACCAGGTTA-3′; PTHF-RT: 5′ATGA TGTCTGCAAGCACCATGGCT-3′; PTHB-RT: 5′-CGGTCTAGAAT ACGTCAGCATT TA-3′.
Serum calcium and inorganic phosphorus concentrations were determined colorimetrically (Sigma). PTH concentration was determined using a two-sided immunoradiometric assay specific for rat PTH (1–34) (Nichols Institute). The detection limit is 2 pg ml-1. Continous subcutaneous treatment of adult mice was done by delivery of 2.8 ng human PTH(1–34) per g weight per h (Bachem) in buffer (1mM HCl, 0.15M NaCl; 2% cystein, 10% heat inactivated BSA) for 6 days and subsequent treatment with 2.2 ng human PTH(1–34) per g weight per h for 8 days using miniosmotic alzet pumps (Alza). Controls were treated with buffer. PTHrP was analysed using a radioimmunoassay detecting the amino-terminal portion of the human protein25.
In situ hybridization and histology
In situ hybridization was carried out as described26 using digoxigenin-labelled riboprobes for rat PTH17, rat Casr27, mouse Pax9(ref. 26) and mouse Gcm1 . The Gcm1 fragment (base pairs 838–1460) was subcloned into bluescript, linearized with XbaI and transcribed with T3. Bone specimens were processed as described28
TPTX and thymectomy
Thyroparathyroidectomy and thymectomy was performed on six-week-old animals (Taconic). All animals were maintained on a regular diet ad libitum without additional calcium supplementation.
G.K. thanks R. Behringer for his generosity. We thank R. Balling, R. Civitelli, C. Johner, H. Kronenberg, B. Lanske, J. Mclaughlin, H. Peters, L.D. Quarles and S. Rebalo for reagents and advice, A. Arnold for sharing unpublished information, and R. Behringer, H. Bellen, P. Hastings, C. Silve, H. Zoghbi and members of the Karsenty laboratory for critical reading of the manuscript. This work is supported by a grant from the MOD foundation to G.K. and NIH grants to G.K. and D.A. T.G. was supported by the Deutscher Akademischer Austauschdienst (DAAD). D.J.A. is an investigator of the Howard Hughes Medical Institute.
About this article
Expression, function, and regulation of the embryonic transcription factor TBX1 in parathyroid tumors
Laboratory Investigation (2017)