Endocrine Unit (M.B., H.J.), Department of Medicine, Massachusetts General Hospital; and Harvard Medical School, and Pediatric Nephrology (H.J.), MassGeneral Hospital for Children Boston, Massachusetts 02114
Address all correspondence and requests for reprints to: Harald Jüppner, M.D., Massachusetts General Hospital, Endocrine Unit, Wellman 5, Boston Massachusetts 02114. E-mail: jueppner{at}helix.mgh.harvard.edu.
In 1942, Fuller Albright and his colleagues described a group of patients who presented with certain developmental and skeletal defects, now collectively termed Albrights hereditary osteodystrophy (AHO), as well as biochemical abnormalities that suggested end-organ failure to respond to the actions of PTH (1). Several variants of this disorder, which was dubbed pseudohypoparathyroidism (PHP), have since been defined, and the different forms have been classified based on the absence or presence of AHO, with or without hormone resistance.
The most frequently encountered variants of PHP include PHP type-Ia (PHP-Ia), pseudo-pseudohypoparathyroidism (PPHP) and PHP type-Ib (PHP-Ib). Patients with PHP-Ia have features of AHO, which typically include obesity, short stature, brachydactyly, ectopic ossification, and mental retardation, and present with hypocalcemia and hyperphosphatemia despite elevated serum PTH levels (2, 3). However, hormone resistance is usually not limited to PTH, in that affected individuals frequently show evidence also for resistance to TSH and gonadotropins (2, 3). Resistance toward other hormones, such as GHRH (4, 5, 6) and calcitonin (7, 8), has also been reported for some PHP-Ia patients, but clinical manifestations and the frequencies of these abnormalities have thus far remained poorly defined. Patients with PPHP have the typical features of AHO but do not show evidence for resistance to PTH or other hormones. In contrast, patients with PHP-Ib present with signs and symptoms of PTH-resistance but lack features of AHO, and hormone resistance seems to be confined to the renal actions of PTH. Only recently, mild TSH-resistance has been observed in some PHP-Ib patients (9, 10, 11), which raises the possibility that additional endocrine systems may also be affected in this variant of PHP.
The molecular defects responsible for these three main forms of PHP appear to reside in GNAS, the gene encoding the -subunit of the stimulatory GTP binding protein (Gs
) and other recently identified transcripts. Patients with PHP-Ia have heterozygous mutations in one of the 13 exons encoding Gs
(12), leading to reduced Gs
protein level and cellular activity, as shown in erythrocytes and other easily accessible cells (2, 3). These mutations explain that all hormones whose actions are impaired in PHP-Ia act via receptors that couple to Gs
. Interestingly, Gs
mutations are also present in patients with PPHP, who do not present with hormone resistance although they have typical AHO features. Furthermore, PHP-Ia and PPHP are frequently found in the same kindred, and the development of one disorder or the other depends on the gender of the parent transmitting the genetic defect (13, 14). Inheritance of the molecular defect from a male affected by either PHP-Ia or PPHP leads to PPHP, i.e. AHO without hormone resistance, while inheritance of the same Gs
mutation from a female affected by either disorder leads to PHP-Ia, i.e. both AHO and hormone resistance. Thus, the identified Gs
mutations are necessary but not sufficient for the phenotypic expression of these two disorders. The genetic mutation leading to an autosomal dominant form of PHP-Ib appears to be an approximately 3-kb deletion more than approximately 200-kb upstream of GNAS (15). Similar to PHP-Ia, this form of PHP-Ib develops only if the genetic defect is transmitted from a female carrier. It is postulated that the approximately 3-kb deletion leads, through yet unknown mechanisms, to a loss-of-methylation defect in GNAS, thereby impairing transcription of Gs
mRNA in a tissue-specific manner.
Gs is a ubiquitously expressed signaling protein, which mediates cellular actions of numerous hormones, paracrine/autocrine factors, and neurotransmitters (16). In PHP-Ia, however, only a limited number of hormones that rely on Gs
signaling appear to be affected. For example, cellular responses to isoproterenol, glucagon, and vasopressin, all of which mediate their actions through Gs
-coupled receptors, are seemingly intact (17, 18, 19). This selective nature of hormone resistance in PHP-Ia, and the parent-of-origin specific mode of inheritance in PHP-Ia/PPHP kindreds suggest that Gs
mRNA is transcribed in certain hormone-responsive cells only from one parental allele. Consequently, heterozygous Gs
mutations can lead to hormone resistance only in those tissues or cells in which monoallelic Gs
expression occurs. In other tissues, the approximately 50% decrease in the Gs
level is expected to cause a reduction in cAMP formation, but the cAMP levels generated can sufficiently maintain all physiological responses, i.e. there is currently no evidence for haploinsufficiency. However, earlier studies using RNA preparations from whole organs, but not from single cells, could only demonstrate biallelic expression of Gs
(20, 21, 22, 23), and it has therefore remained unclear whether monoallelic Gs
expression plays a role in hormone responsiveness in PHP-Ia (or PHP-Ib).
Detailed investigations of GNAS have recently provided new insights into Gs expression and the mechanisms underlying hormone resistance. We now know that, besides Gs
, GNAS gives rise to several other transcriptional units derived from the sense or antisense strand. Transcripts encoding XL
s, a large variant of Gs
, show exclusive paternal expression and are derived through the use of an alternative promoter and a unique first exon, which splices onto exons 213 (21, 24, 25). The encoded XL
s protein has a distinct amino-terminal domain but is identical to Gs
over the carboxyl-terminal 348 residues. Another protein product of GNAS is NESP55, a 55-kDa neuroendocrine secretory protein. Transcripts encoding NESP55, which show exclusive maternal expression (22, 24), are derived from a different promoter and use a separate first exon, which also splices onto exons 213. The region comprised by these latter exons, however, is located in the 3'-untranslated portion of the mRNA (22, 26). Therefore, in contrast to XL
s, NESP55 does not share any amino acid sequence identity with Gs
. GNAS gives rise to at least one additional transcript derived from the sense strand. The A/B transcript uses exon A/B [also referred to as exon 1A or 1' (Refs.27, 28, 29)] as the first exon, which also splices onto exons 213. In addition, the opposite strand of GNAS gives rise to an antisense transcript, which consists of distinct exons that do not overlap with any of the described sense exons (30, 31). Both A/B and antisense transcripts show expression from the paternal allele only and are thought to be nontranslated gene products (28, 30, 31). Similar to other imprinted loci, the promoter regions of these GNAS transcripts show allele-specific methylation, and in each case, the nonmethylated promoter drives the expression (22, 24, 28, 30, 31).
Transcripts encoding Gs are comprised of exons 113 (12). Unlike all the other promoter regions in GNAS, the promoter region giving rise to Gs
transcripts does not exhibit allele-specific methylation, and in most tissues, expression takes place from both parental alleles (20, 21, 22, 23, 32). Nevertheless, as-yet-unknown mechanisms appear to silence paternal Gs
expression in certain cells, because maternal- specific expression has recently been demonstrated in some tissues. Generation and analysis of mice heterozygous for disruption of maternal or paternal Gnas exon 2 revealed that, whereas expression of Gs
was biallelic in renal medulla, it was predominantly maternal in the cells of renal proximal tubules and in white and brown adipose tissue (33). In the pituitary gland, thyroid, and gonads, Gs
expression was also shown to be largely maternal (34, 35, 36). Thus, predominantly monoallelic Gs
expression appears to take place in a tissue- or cell-specific manner.
Previous studies to determine hormone responsiveness of the pituitary in PHP-Ia yielded conflicting results, particularly regarding GH synthesis and release (4, 5, 6, 37). Two independent studies, published in this issue of JCEM, have now re-addressed this question through careful investigations of multiple individuals with this disorder. In nine patients with PHP-Ia, seven of whom had documented Gs mutations, Mantovani et al. (38) evaluated GH secretory responses of somatotrophs. The mean GH peak in response to arginine/GHRH infusion in their patient cohort was significantly lower than in the normal population. Six of the nine subjects demonstrated significantly impaired GH responses to arginine/GHRH, and two of the remaining three individuals showed partially impaired GH responses. As an indication of GH deficiency, low-normal serum IGF-I levels accompanied these findings in seven patients. Independently, Germain-Lee et al. (39) investigated 13 PHP-Ia subjects, 10 of whom had documented Gs
mutations. In nine subjects, the pituitary GH responses to arginine/L-dopa and arginine/GHRH infusion were reduced, and serum concentrations of IGF-I were low. Hence, these two studies provide strong evidence for the conclusion that GH deficiency is a common finding in PHP-Ia, and that impaired responsiveness of somatotrophs to GHRH, or other central regulators of GH synthesis and release, is part of multiple hormone resistance in this disorder.
The study by Mantovani et al. (38) has also evaluated corticotroph responsiveness in three PHP-Ia patients. In response to corticotropin-releasing factor, these patients exhibited appropriate increases in serum ACTH and subsequently cortisol levels, indicating that the responsiveness of corticotrophs to corticotropin-releasing factor, which, like PTH, signals through a class B Gs-coupled receptor, is intact. These results corroborate previous findings in PHP-Ia (37, 40) and, taken together with the frequently impaired GH secretion, suggest that the presence or the extent of Gs
imprinting within the pituitary gland is cell-type specific. However, additional investigations through the use of animal models or single-cell analyses, are required to determine whether cell type-specific Gs
imprinting is actually responsible for these observations.
The studies by Mantovani et al. (38) and by Germain-Lee et al. (39) raise important questions regarding evaluation and treatment of PHP-Ia patients. Should testing of patients with PHP-Ia include evaluation of GH synthesis and release, and if GH deficiency is documented, should the patient be treated with recombinant GH? Based on their findings, Germain-Lee et al. (39) suggest that GH deficiency may contribute to obesity and short stature, suggesting that PHP-Ia patients with documented GH deficiency might benefit from such treatment. However, short stature and obesity are also seen in patients with PPHP, who typically have heterozygous paternally inherited Gs mutations. If the GH deficiency documented in PHP-Ia is only the result of maternal Gs
mutations (combined with paternal silencing of Gs
transcription in somatotrophs), PPHP patients would be predicted to have no such deficiency. It is thus possible that short stature and obesity in AHO may not be directly related to a lack of GH, but rather reflect intrinsic abnormalities in bone growth and adipocyte metabolism, respectively. Controlled trials with recombinant GH treatment in patients affected by PHP-Ia (with and without documented GH deficiency) and those affected by PPHP will be necessary to clarify these points.
Clinical and laboratory consequences of GHRH resistance in PHP-Ia appear to be subject to individual variation. Interestingly, the two patients with normal IGF-I levels studied by Mantovani et al. (38) were siblings, and furthermore, five PHP-Ia siblings were previously demonstrated to have normal GH responses (37). It is therefore possible that resistance toward GHRH or other GH secretagogues does not fully develop in every genetic background. There are several possibilities to explain these differences among PHP-Ia patients. For example, the paternally derived extra-large variant of Gs, XL
s, which shows abundant expression in the pituitary (25, 41) and can act as a Gs-like signaling protein (42), may compensate for the lack of Gs
in some patients depending on the presence of certain genetic modifiers. Alternatively, the differences in GH deficiency may reflect individual variation in the proportion of paternally derived Gs
transcripts in somatotrophs. In fact, analysis of a small number of whole pituitary glands has previously revealed significant variation among individual samples with respect to the degree of paternal Gs
expression (34, 35), which was also observed among different thyroid samples and gonads (11, 35, 36). It is therefore possible that the degree of paternal Gs
silencing is much more variable in the pituitary than in renal proximal tubules, explaining why GH deficiency in PHP-Ia is encountered less frequently than PTH-resistance.
As a potential explanation for the tissue-specific imprinting of Gs expression, it has been postulated that a repressor protein reduces the generation of Gs
transcripts from the paternal allele in a tissue- or cell-specific manner (43). Differences in the pituitary expression levels of this putative repressor among individuals may be responsible for the varying degrees of Gs
expression from the paternal GNAS allele, thus influencing the development of GH deficiency in patients with PHP-Ia.
An important clue regarding the mechanism of tissue-specific Gs imprinting has been obtained from recent findings in patients with PHP-Ib. These patients, who typically have renal PTH resistance [and sometimes mild TSH resistance (11)], but lack coding Gs
mutations, exhibit a loss-of-methylation defect at GNAS exon A/B, which leads to biallelic expression of A/B transcripts (10, 44). These epigenetic changes, which appear to be caused by an approximately 3-kb deletion upstream of exon A/B (15) and make the maternal allele look like the paternal allele, are proposed to be responsible for the maternal silencing of Gs
expression in the renal proximal tubule (9, 10, 44). Thus, differential methylation at exon A/B, located immediately upstream of the Gs
promoter, seems critical for proper expression of Gs
in certain tissues. In fact, it is possible that the normally nonmethylated exon A/B region on the paternal GNAS allele allows for binding of the putative repressor, resulting in a reduction in Gs
expression.
The study of the clinical and molecular aspects of the different forms of PHP has provided the research community with many remarkable insights into the intricate regulation of Gs expression. Yet, there are still a lot of unknowns regarding the molecular and genetic mechanisms underlying each of the different PHP forms. As in the case of pituitary responsiveness in PHP-Ia, information from molecular and cellular studies is likely to stimulate additional clinical investigations to clarify previously unappreciated aspects of hormone resistance in different endocrine/paracrine systems. Such a combination of clinical, molecular, and cellular studies will likely provide further optimization of the clinical management of patients afflicted by PHP-Ia and other forms of PHP.
Footnotes
Abbreviations: AHO, Albrights hereditary osteodystrophy; Gs, GTP binding protein; PHP, pseudohypoparathyroidism; PHP-Ia, PHP type-Ia; PHP-Ib, PHP type-Ib; PPHP, pseudo-pseudohypoparathyroidism.
Received July 22, 2003.
Accepted July 22, 2003.
References