- Top of page
- Materials and Methods
- Supporting Information
Certain genes are expressed in a parent-of-origin–specific manner from a single allele, and genetic aberrations disrupting this monoallelic expression are the cause of a variety of human diseases and tumors.[1, 2] GNAS is a complex gene leading to several transcripts that show monoallelic expression.[3, 4] One of the major gene products of GNAS is the α-subunit of the stimulatory G protein (Gαs), a ubiquitously expressed signaling protein that mediates the actions of many hormones, neurotransmitters, and autocrine/paracrine factors via the generation of cAMP. More recently described products from GNAS include the maternally expressed transcript encoding NESP55 and paternally expressed transcripts A/B (also referred to as 1A), extra-large Gαs (XLαs), and GNAS antisense.[5-8] Promoters of these latter gene products are differentially methylated. In contrast, the promoter of Gαs lacks methylation, and in most tissues, Gαs expression is biallelic.[7, 9] However, despite the lack of differential methylation at its promoter, Gαs is expressed predominantly from the maternal allele in a small number of tissues including pituitary, thyroid, and renal proximal tubules.[10-13]
Mutations of GNAS cause several human diseases, and monoallelic maternal expression of Gαs contributes to the pathogenesis of most of these diseases. For example, somatic mutations leading to constitutive Gαs activity (gsp oncogene) are found in growth hormone-secreting pituitary adenomas. However, these somatotroph tumors develop only if the mutations are located on the maternal GNAS allele. Moreover, paternal Gαs expression is frequently derepressed in these tumors, regardless of the presence of activating Gαs mutations, presumably contributing to their pathogenesis. Loss-of-function mutations in or abnormal imprinting of the maternal GNAS allele lead to pseudohypoparathyroidism type-Ia (PHP-Ia) and PHP-Ib, respectively,[15, 16] which are both associated with end-organ resistance to the actions of parathyroid hormone (PTH) and some other hormones that signal via Gαs. In contrast, mutations on the paternal allele have no effect on hormone action because Gαs expression from the paternal allele is normally suppressed because of imprinting in hormone target tissues.[17, 18]
PTH binds to a Gαs-coupled receptor and acts on bone, as well as the renal proximal and distal tubules, to regulate calcium and phosphate homeostasis. In bone, PTH increases both bone formation and bone resorption. In renal proximal tubules, it induces the biosynthesis of 1,25 dihydroxy vitamin D (1,25(OH)2D) and inhibits the tubular reabsorption of phosphate, and in renal distal tubules, it enhances the tubular reabsorption of calcium. The paternal Gαs allele is silenced in the renal proximal tubule[13, 20] but not in bone or the distal part of the nephron.[9, 21] Accordingly, resistance to PTH in patients with PHP-Ia and PHP-Ib occurs primarily in the renal proximal tubule. These patients typically develop hypocalcemia and hyperphosphatemia, combined with elevated serum PTH. Responsiveness of bone and renal distal tubules to PTH, on the other hand, appears to be preserved, as patients with PHP can show evidence for increased bone turnover and hypocalciuria.
Clinical features resulting from PTH resistance in PHP-Ia and PHP-Ib are often observed after the first year of life, and it has therefore been suggested that PTH resistance in these PHP forms develops in a delayed fashion.[24-26] On the other hand, several cases of PHP-Ia with documented GNAS mutations have been reported in whom PTH levels were elevated during infancy.[27-32] Moreover, although the presence of GNAS mutations was not investigated, hypocalcemia owing to PTH resistance has been reported in a substantial number of infants and newborns.[33-41] Hence, it has remained unclear whether the renal proximal tubular PTH resistance caused by maternal loss of Gαs manifests itself during early postnatal life. This is an important question, considering that the paternal Gαs silencing has an important role in the development of this biochemical defect. If this hormonal abnormality indeed develops in a delayed fashion, it may indicate that the silencing of the paternal Gαs allele is not established at birth but instead develops postnatally, suggesting, in turn, that the mechanisms underlying this critical, yet poorly understood, epigenetic process are subject to developmental regulation. The temporal profile of allelic Gαs silencing in the renal proximal tubule, however, has also remained hitherto unknown.
We herein reviewed reported PHP-Ia cases with documented GNAS mutations and investigated mice heterozygous for Gnas disruption regarding the temporal development of PTH resistance. Our findings indicate that PTH resistance caused by the loss of maternal GNAS allele occurs in a delayed fashion. Moreover, we determined that paternal Gαs silencing is established during postnatal development of mouse renal proximal tubules, thus providing a plausible explanation for the latency of PTH resistance in PHP-Ia.
- Top of page
- Materials and Methods
- Supporting Information
In this study, we examined reported cases of PHP-Ia with documented GNAS mutations and found that elevated PTH levels with or without hypocalcemia becomes manifest mostly, but not always, after infancy. By using mouse models of this disorder, we also showed that, although the Gαs mutation is present from the time of conception, hypocalcemia and elevated PTH levels develop after early postnatal life. In addition, taking advantage of mice in which maternal or paternal Gnas exon 1 is ablated, we examined the allelic expression of Gαs in the renal proximal tubule, thus showing that both parental alleles contribute equally to Gαs expression during the early postnatal period, with increasing relative expression from the maternal allele during development.
The results of our analyses are consistent with what have been described in several case reports,[47, 48, 54, 55] indicating that the elevation of PTH precedes the manifestation of hypocalcemia. The increase in PTH may be responsible for maintaining normal serum calcium levels for a certain period of time, likely resulting from the actions of PTH in bone and renal distal tubules. It is also likely that, as evidenced in E2m−/+ and E1m−/+ mice, the degree of end-organ resistance increases with age, preventing elevated PTH levels from being able to maintain normal serum calcium. In addition, high oral calcium intake, as in breast- or formula-fed infants, could perhaps help delay or mitigate the hypocalcemia. Note that the 3-week-old E1m−/+ mice analyzed in our study were not yet weaned. In early postnatal life, a small number of PHP-Ia patients were reported to have hypocalcemia or normocalcemia with elevated PTH, unlike the data obtained from our E1m−/+ mice. This finding in humans could reflect other genetic or nongenetic factors, such as insufficient vitamin D or calcium intake during infancy, or perhaps defects related to calcium metabolism in the mother who is also a carrier of the GNAS mutation.
According to our analyses, male and female PHP-Ia patients do not differ significantly with respect to the age at which the evidence of PTH resistance is observed. Our results from E1m−/+ mice also do not indicate gender-specific differences regarding the temporal development of PTH resistance, at least based on the analysis of 3-week-old and 2-month-old mice. Using the same strain of adult E1m−/+ mice, we have previously detected no gender differences in serum calcium, PTH, phosphorus, and 1,25(OH)2D values. Likewise, no gender-specific differences were observed in serum calcium, PTH, and phosphorus levels of adult E2m−/+ mice. Thus, gender may not play a significant role in the onset and the progression of PTH resistance in PHP-Ia, but this question needs to be addressed through additional studies.
PTH-induced cAMP generation has been found to be reduced in proximal tubule-enriched renal cortices of adult E2m−/+ mice and in kidney membranes from another adult mouse model in which maternal Gnas exon 1 was ablated. Our findings in E1m−/+ and E2m−/+ mice with respect to PTH-induced changes in plasma cAMP are consistent with those previous studies, indicating resistance to PTH. Moreover, our results demonstrate a progressive increase of PTH resistance in both of these PHP-Ia mouse models. However, 3-week-old E1m−/+ mice, unlike 3-week-old E2m−/+ mice, did not show a significantly blunted plasma cAMP response to PTH. This discrepancy could reflect the differences between the background strains in which these two models are maintained (FVB and CD1, respectively) or the finding that E2m−/+ mice show retarded kidney development (LSW, unpublished data). Additionally, the ∼80% preweaning mortality in E2m−/+ mice resulted essentially in the analysis of a selected group of survivors, and this selected group may have shown poor responsiveness to PTH because of some other reasons. Note that, although there are various different PTH-responsive tissues, the PTH-induced cAMP elevation in the plasma appears to reflect cAMP generation in a tissue(s) in which Gαs is silenced from the paternal allele. This interpretation is based on our data from adult E1m−/+ and E2m−/+ mice and the finding that PTH-induced cAMP generation is markedly blunted in both urine and plasma of adult mice in which Gαs is silenced from both parental alleles owing to loss of Gnas imprinting (model of PHP-Ib). Given the data indicating that PTH-induced elevation of plasma cAMP reflects the action of this hormone in kidney, this tissue is likely to be the renal proximal tubule. Thus, the apparent absence of a significantly blunted plasma cAMP response in 3-week-old E1m−/+ mice suggests that renal proximal tubular resistance to PTH is not yet fully established at that age.
Serum 1,25(OH)2D levels have been reported to be slightly lower in 2-month-old E1m−/+ mice than wild-type littermates, although the difference was not statistically significant. This finding is consistent with the mild reduction of Cyp27b1 expression in proximal tubules of 2-month-old E1m−/+ mice. The concomitant increase in serum PTH in these animals thus indicates PTH resistance in the renal proximal tubule. Similarly, proximal tubular Cyp27b1 expression is normal despite significant elevation of serum PTH in 3-week-old E1m−/+ mice, and this inappropriately normal Cyp27b1 expression also provides evidence for PTH resistance at this age. These findings correlate well with the marked reduction of Gαs mRNA in proximal tubules of 3-week- and 2-month-old E1m−/+ mice. On the other hand, PTH reduces Cyp24a1 mRNA stability, and thus, the mild reduction of Cyp24a1 mRNA in proximal tubules of 2-month-old E1m−/+ mice could indicate that this action of PTH is not impaired. However, renal Cyp24a1 is under the control of various other factors, and the mild reduction in its expression level could be secondary to reduced serum calcium and/or 1,25(OH)2D levels. Further investigations are necessary to determine the relative roles of Cyp27b1 and Cyp24a1 in the development of hypocalcemia in PHP-Ia.
The finding that E1+/p− mice are normocalcemic correlates well with findings in patients with PPHP. However, the elevation of PTH in 3-week-old E1+/p− mice is unexpected, as patients with PPHP are described as having Albright's hereditary osteodystrophy without evidence for hormone resistance. Serum PTH has also been found to be modestly elevated in another paternal Gnas exon 1 knockout mouse model in adulthood, and like E1+/p− mice, the latter mice were normocalcemic. This discrepancy between data from humans and mice could perhaps be explained by species-specific differences. Alternatively, there might be some patients with paternal inactivating GNAS mutations who have high-normal or mildly elevated PTH levels in the absence of hypocalcemia. Given that Gαs is normally expressed from both parental alleles in bone and renal distal tubules,[9, 21] the elevation of PTH in 3-week-old E1+/p− mice may reflect a modest degree of Gαs haploinsufficiency in those tissues. In that case, however, the PTH elevation in 3-week-old E1m−/+ mice would also be owing, at least partly, to this putative Gαs haploinsufficiency because the PTH levels in E1m−/+ is only modestly, and not statistically significantly, higher than those in E1+/p− mice at this age (434 versus 325 pg/mL; p = 0.15) (Fig. 3). In fact, our finding that 3-week-old E1m−/+ mice show an apparently normal PTH-induced plasma cAMP response also correlates with this interpretation (Fig. 2E). Moreover, it has been shown that derepression of paternal Gαs allele reduces but does not normalize the elevated serum PTH levels in another mouse model of PHP-Ia, in which an inactivating missense mutation is present within maternal Gnas exon 6. It thus appears that factors other than paternal Gαs silencing may contribute to the PTH resistance that results from the loss of maternal Gαs allele. Further investigations are needed to address this possibility.
Consistent with our results, Zheng and colleagues have shown no evidence for Gαs imprinting in human fetal renal cortices. Thus, the mechanisms silencing the paternal Gαs allele in the renal proximal tubule operate only after birth, perhaps reflecting the immaturity of renal tubules at birth and during early postnatal life. In contrast, the allelic silencing of Gαs appears to occur much earlier in various other tissues, considering the findings in mice that loss of the maternal Gαs allele leads to neonatal subcutaneous edema[13, 43, 60, 61] and that derepression of the paternal Gαs allele leads to early postnatal growth retardation. Based on clinical findings in patients with PHP-Ia, allelic Gαs silencing might also be present in neonatal thyroid.[63-65] Moreover, our results, consistent with evidence obtained from mice with paternal deletion of exon A/B (termed 1A in mice), show that paternal Gαs expression is repressed in BAT already at birth. Interestingly, however, a previous study has shown that the paternal Gαs allele is expressed almost to the same extent as the maternal Gαs allele in adult BAT. Thus, the mechanisms governing this silencing event may be subject to tissue-specific temporal constraints. Of note, the postnatal decline of Gαs silencing in BAT coincides with the temporal expression of XLαs in this tissue, which is abundant at birth but reduces drastically after early postnatal development. It remains to be determined whether XLαs expression, which occurs exclusively from the paternal allele, is involved in the tissue-specific allelic silencing of Gαs. Similar to GNAS, several other genes demonstrate tissue-specific monoallelic expression. Some of those genes, such as KCNQ1, GRB10, and IGF2, are also developmentally regulated in this regard, at least in certain tissues.[68-71] It is possible that analogous mechanisms regulate the allelic silencing of these genes and GNAS via tissue-specific factors that are expressed at different stages of development.
Our results indicate that the contribution of the paternal allele is ∼35% of the total in 2-month-old mouse proximal tubules, ie, markedly more than the paternal Gαs contribution in neonatal BAT, which is ∼14% (Fig. 4A, B). This finding, consistent with previous observations made upon the analysis of renal cortices, is highly unlikely to reflect contamination from surrounding tissues because we obtained renal proximal tubules by laser-capture microdissection. By using this method in our lab, we have previously isolated mouse proximal tubules and showed that the isolated tissue was positive for various proximal tubule-specific markers (eg, aldolase, the V-ATPase E-subunit, and the Arf6-GDP/GTP exchange factor ARNO) but not distal tubule- (eg, caveolin-1 and caveolin-2) or endothelial-specific markers (eg, CD31 and ICAM1). Therefore, it appears that the paternal silencing of Gαs expression in the renal proximal tubule is incomplete. In that case, however, Gαs mRNA levels in proximal tubules of adult patients with PHP-Ib—in whom both GNAS alleles, partially or entirely, show a paternal-specific imprinting profile—would be predicted to be ∼70% of the levels in healthy individuals and higher than the levels in adult patients with PPHP, who would have ∼65% Gαs levels compared with healthy individuals. Considering that patients with PHP-Ib, but not those with PPHP, have significant hypocalcemia, it is possible that there are species-specific differences in the extent of Gαs imprinting between humans and rodents. No data are currently available from PHP-Ia patients or healthy humans regarding the paternal silencing of Gαs in the renal proximal tubule, but a more pronounced degree of silencing may indeed exist in humans, considering that the biochemical phenotype of mice heterozygous for ablation of maternal Gnas exon 1 is less severe than that observed in most patients with PHP-Ia.[13, 20]
On the other hand, we have recently generated a mouse model of PHP-Ib, in which the Gαs silencing occurs on both parental alleles resulting from specific methylation changes on the maternal Gnas allele. Adult PHP-Ib mice demonstrated hypocalcemia and elevated serum PTH. Thus, additional factors may underlie our observation that the silencing of paternal Gαs allele in the mouse renal proximal tubule is incomplete. For example, it is conceivable that the paternal Gαs allele is more dramatically silenced but only within a small portion of the proximal tubule. This would reconcile the finding that, whereas PTH-induced plasma cAMP response is almost completely blunted, substantial Gαs mRNA expression is observed in samples obtained from the entire proximal tubule (∼35% of wild type). However, an alternative explanation could also exist, as it is conceivable that the plasma cAMP response to PTH reflects mostly the action of this hormone in a different tissue in which the paternal Gαs allele is silenced. In that case, reduction of Gαs levels in that tissue may be the primary cause of the observed blunting in the cAMP response. For example, pituitary could play a role. Gαs expression is predominantly maternal in anterior pituitary, and it has been shown that PTH-induced release of prolactin observed in healthy subjects is significantly blunted in patients with PHP.
The delayed establishment of paternal Gαs silencing in the renal proximal tubule could explain the latency of PTH resistance in PHP-Ia patients. However, other possible mechanisms have yet to be ruled out. The paternally expressed Gαs variant XLαs, which can mimic Gαs actions and is expressed in mouse kidney during the early postnatal period,[45, 73, 74] remains intact because of the maternal transmission of GNAS mutations in PHP patients, making it possible that this protein mediates the early postnatal actions of PTH. Another possibility is that the proximal tubular actions of PTH during this period are mediated through another G protein or in a G-protein-independent manner, as PTHR can couple to other G proteins and signal, at least under certain conditions, through mechanisms that are G-protein independent.[75, 76]
In summary, our investigation of reported PHP-Ia cases and mice with heterozygous ablation of Gnas showed that PTH resistance resulting from loss of the maternal Gαs allele develops after early postnatal life. We also revealed that paternal silencing of Gαs in the renal proximal tubule is established after the early postnatal period, unlike in BAT in which the paternal allele is already repressed at birth. The delayed onset of paternal Gαs silencing in the renal proximal tubule provides a plausible explanation for the latency of PTH resistance in PHP-Ia patients.