help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Weinstein, L. S.
Right arrow Articles by Liu, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Weinstein, L. S.
Right arrow Articles by Liu, J.
Endocrine Reviews 22 (5): 675-705
Copyright © 2001 by The Endocrine Society

Endocrine Manifestations of Stimulatory G Protein {alpha}-Subunit Mutations and the Role of Genomic Imprinting

Lee S. Weinstein, Shuhua Yu, Dennis R. Warner and Jie Liu

Metabolic Diseases Branch (L.S.W., S.Y., J.L.), National Institute of Diabetes, Digestive, and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892; and Department of Molecular, Cellular, and Craniofacial Biology (D.R.W.), University of Louisville School of Dentistry, Louisville, Kentucky 40202

Correspondence: Address all correspondence and requests for reprints to: Dr. Lee S. Weinstein, Metabolic Diseases Branch, National Institute of Diabetes, Digestive, and Kidney Diseases, National Institutes of Health, Bethesda Maryland 20892-1752. E-mail: leew{at}amb.niddk.nih.gov


    Abstract
 Top
 Abstract
 I. Introduction
 II. Gs{alpha}-Signaling...
 III. Gs{alpha} Activating...
 IV. Gs{alpha} Loss-of-Function...
 V. Insights from Gnas...
 VI. GNAS1 Imprinting Mechanisms...
 VII. Summary
 References
 
The heterotrimeric G protein Gs couples hormone receptors (as well as other receptors) to the effector enzyme adenylyl cyclase and is therefore required for hormone-stimulated intracellular cAMP generation. Receptors activate Gs by promoting exchange of GTP for GDP on the Gs {alpha}-subunit (Gs{alpha}) while an intrinsic GTPase activity of Gs{alpha} that hydrolyzes bound GTP to GDP leads to deactivation. Mutations of specific Gs{alpha} residues (Arg201 or Gln227) that are critical for the GTPase reaction lead to constitutive activation of Gs-coupled signaling pathways, and such somatic mutations are found in endocrine tumors, fibrous dysplasia of bone, and the McCune-Albright syndrome. Conversely, heterozygous loss-of-function mutations may lead to Albright hereditary osteodystrophy (AHO), a disease characterized by short stature, obesity, brachydactyly, sc ossifications, and mental deficits. Similar mutations are also associated with progressive osseous heteroplasia. Interestingly, paternal transmission of GNAS1 mutations leads to the AHO phenotype alone (pseudopseudohypoparathyroidism), while maternal transmission leads to AHO plus resistance to several hormones (e.g., PTH, TSH) that activate Gs in their target tissues (pseudohypoparathyroidism type IA). Studies in Gs{alpha} knockout mice demonstrate that Gs{alpha} is imprinted in a tissue-specific manner, being expressed primarily from the maternal allele in some tissues (e.g., renal proximal tubule, the major site of renal PTH action), while being biallelically expressed in most other tissues. Disrupting mutations in the maternal allele lead to loss of Gs{alpha} expression in proximal tubules and therefore loss of PTH action in the kidney, while mutations in the paternal allele have little effect on Gs{alpha} expression or PTH action. Gs{alpha} has recently been shown to be also imprinted in human pituitary glands. The Gs{alpha} gene GNAS1 (as well as its murine ortholog Gnas) has at least four alternative promoters and first exons, leading to the production of alternative gene products including Gs{alpha}, XL{alpha}s (a novel Gs{alpha} isoform that is expressed only from the paternal allele), and NESP55 (a chromogranin-like protein that is expressed only from the maternal allele). A fourth alternative promoter and first exon (exon 1A) located approximately 2.5 kb upstream of the Gs{alpha} promoter is normally methylated on the maternal allele and transcriptionally active on the paternal allele. In patients with isolated renal resistance to PTH (pseudohypoparathyroidism type IB), the exon 1A promoter region has a paternal-specific imprinting pattern on both alleles (unmethylated, transcriptionally active), suggesting that this region is critical for the tissue-specific imprinting of Gs{alpha}. The GNAS1 imprinting defect in pseudohypoparathyroidism type IB is predicted to decrease Gs{alpha} expression in renal proximal tubules. Studies in Gs{alpha} knockout mice also demonstrate that this gene is critical in the regulation of lipid and glucose metabolism.

I. Introduction

II. Gs{alpha}-Signaling Mechanisms and Gene Structure

A. Gs{alpha} structure and function

B. Gs{alpha} gene (GNAS1) structure

C. Alternative GNAS1 gene products

D. Imprinting of the GNAS1 gene

III. Gs{alpha} Activating Mutations

A. Endocrine tumors

B. McCune-Albright syndrome (MAS)

C. Fibrous dysplasia of bone (FD)

IV. Gs{alpha} Loss-of-Function Mutations

A. Albright hereditary osteodystrophy (AHO)

B. Pseudohypoparathyroidism type IA (PHPIA)

C. Pseudopseudohypoparathyroidism (PPHP)

D. The role of inactivating Gs{alpha} mutations and Gs{alpha} imprinting in the pathogenesis of AHO, PHPIA, and PPHP

E. Progressive osseous heteroplasia (POH)

V. Insights from Gnas Knockout Mice

A. Tissue-specific imprinting of Gs{alpha}

B. Role of Gs{alpha} in renal function

C. Role of Gs{alpha} in energy and glucose metabolism

VI. GNAS1 Imprinting Mechanisms and Defects

A. Pseudohypoparathyroidism type IB (PHPIB)

B. Potential GNAS1 imprinting mechanisms

VII. Summary


    I. Introduction
 Top
 Abstract
 I. Introduction
 II. Gs{alpha}-Signaling...
 III. Gs{alpha} Activating...
 IV. Gs{alpha} Loss-of-Function...
 V. Insights from Gnas...
 VI. GNAS1 Imprinting Mechanisms...
 VII. Summary
 References
 
IT HAS NOW been 30 yr since Rodbell and colleagues (1, 2) provided the first evidence that guanine nucleotide is required for hormone-stimulated cAMP generation. Nine years later Gs, the heterotrimeric G protein that couples receptors to adenylyl cyclase, was purified from liver membranes (3). It is now clear that Gs is one member of a large family of G proteins that are integral components of diverse signaling pathways. Each G protein is composed of a specific {alpha}-subunit, which binds guanine nucleotide and couples to specific receptors and effectors, associated with a ß- and {gamma}-subunit. The role of G proteins in signal transduction was last summarized in Endocrine Reviews in 1992 (4).

The first example of a Gs defect leading to clinical disease was the observation that the secretory diarrhea of Vibrio cholerae infection results from posttranslational modification of the Gs {alpha}-subunit (Gs{alpha}) and abnormal regulation of cAMP in intestinal cells. As Gs{alpha} is ubiquitously expressed and is a necessary component for many signaling pathways, genetic defects of the Gs{alpha} gene (GNAS1) would be expected to lead to pleiotropic manifestations. Within the past decade it has been discovered that mutations which produce constitutively activated forms of Gs{alpha} can lead to endocrine tumors, excess hormone secretion, and skeletal and other nonendocrine abnormalities. In contrast, heterozygous inactivating Gs{alpha} mutations produce Albright hereditary osteodystrophy (AHO), a syndrome characterized by skeletal and other developmental abnormalities. Curiously, maternal inheritance of these inactivating mutations also produces hormone resistance while paternal transmission does not, and this is likely due to the fact that Gs{alpha} is imprinted (expressed only from the maternal allele) in specific tissues. Recent studies show that, in fact, GNAS1 is an extremely complex imprinted gene that produces different gene products from the maternal and paternal allele through the use of oppositely imprinted alternative promoters. Moreover, abnormal imprinting of this gene can also lead to hormone resistance. The diseases associated with Gs{alpha} defects are listed in Table 1Go.


View this table:
[in this window]
[in a new window]
 
Table 1. Human diseases associated with GNAS1 defects

 
Although the role of G protein defects in human disease has been previously reviewed (5, 6, 7, 8), our understanding of how GNAS1 defects lead to human disease has greatly increased over the past 5 yr. This review summarizes recent progress in our understanding of the molecular genetics and imprinting of GNAS1, the clinical diseases that result from GNAS1 mutations or imprinting defects, and insights that have been gained from analysis of a Gs{alpha} knockout model. Because this article focuses primarily on pathogenesis rather than on clinical diagnosis and management, we cite references that describe the clinical aspects of these disorders in greater detail.


    II. Gs{alpha}-Signaling Mechanisms and Gene Structure
 Top
 Abstract
 I. Introduction
 II. Gs{alpha}-Signaling...
 III. Gs{alpha} Activating...
 IV. Gs{alpha} Loss-of-Function...
 V. Insights from Gnas...
 VI. GNAS1 Imprinting Mechanisms...
 VII. Summary
 References
 
A. Gs{alpha} structure and function
Gs is a member of the family of heterotrimeric G proteins that transmit signals from cell surface receptors to intracellular effectors, such as ion channels or enzymes (e.g., adenylyl cyclase, PLC) which generate second messengers (e.g., cAMP, calcium, inositol triphosphate). Classical G protein-coupled receptors have seven putative helical transmembrane domains with an extracellular amino terminus and an intracellular carboxyl terminus. Each G protein is defined by its specific {alpha}-subunit (of which 20 have been identified to date), which binds guanine nucleotide and interacts with specific receptors and effectors. ß- And {gamma}-subunits form tightly but noncovalently bound dimers that are targeted to the plasma membrane through lipid modifications at the carboxyl terminus of {gamma}-subunits. G{alpha}s also undergo lipid modifications that are important for membrane targeting (9). Association with ß{gamma} is required for G{alpha}s to be activated by receptors.

Gs{alpha} is a protein found in all cell types except mature spermatozoa (10), which couples a wide variety of G protein-coupled receptors to the stimulation of adenylyl cyclase, the enzyme that catalyzes the synthesis of cAMP from ATP within cells. Many of the downstream effects of cAMP are through activation of PKA. Seven-transmembrane receptors for a wide variety of extracellular signals, including the glycoprotein and many peptide hormones, catecholamines, and other biogenic amines and neurotransmitters, all activate Gs{alpha} to raise intracellular cAMP in their target cells. However, Gs{alpha} activators may not be limited to the seven-transmembrane receptor family, as there is evidence that Gs{alpha} can also be activated by various tyrosine kinase receptors, including the epidermal growth factor (11, 12) and basic fibroblast growth factor receptors (13). Epidermal growth factor receptors appear to activate Gs{alpha} by phosphorylation of specific tyrosine residues in the {alpha}-subunit (14, 15). Nor is adenylyl cyclase the only effector activated by Gs{alpha}. Gs{alpha} has been shown to open specific Ca2+ channels in the heart (16, 17) and has recently been shown (along with Gi1{alpha}) to interact directly with and activate members of the src tyrosine kinase family (18). Gs{alpha} undergoes palmitoylation at its amino terminus, which is required for membrane targeting (19, 20, 21). While most attention has been directed at the role of Gs{alpha} as a signal transducer at the plasma membrane, Gs{alpha} is also present in intracellular membrane compartments and has been implicated as a regulator of intracellular membrane trafficking (22). A second G protein, Golf, also stimulates adenylyl cyclase, but it is produced by a separate gene and is only expressed in a small number of tissues (23, 24).

Similar to other G proteins, Gs is activated and deactivated via the GTPase cycle of its {alpha}-subunit (Fig. 1Go) (reviewed in Refs. 4 and 25). In the basal state Gs exists as an inactive Gs{alpha}{gamma} heterotrimer with GDP bound to the guanine nucleotide binding site of Gs{alpha}. Upon interaction with agonist-bound (activated) receptor, GDP is released from Gs{alpha} and replaced with GTP. Binding of GTP to Gs{alpha} switches Gs{alpha} into an active conformation, and activated GTP-bound Gs{alpha} dissociates from ß{gamma}. Upon activation, Gs{alpha} is depalmitoylated and there is evidence that some or all of the {alpha}-subunit may be released from the plasma membrane (26, 27, 28, 29). GTP-bound Gs{alpha} directly interacts with and activates its effectors. An intrinsic GTPase activity within the {alpha}-subunit hydrolyzes bound GTP to GDP, leading to reassociation with ß{gamma} to reform the inactive heterotrimer. While the GTPase activity of other G protein {alpha}-subunits is enhanced by interaction with RGS (regulator of G protein signaling) proteins, there is no evidence for the existence of an RGS protein for Gs{alpha} (30, 31). However, there is evidence to suggest that at least one isoform of adenylyl cyclase (AC5) may enhance the GTPase activity of Gs{alpha} and its ability to be activated by receptors (32). In experimental systems G protein {alpha}-subunits can also be activated by incubation with nonhydrolyzable GTP analogs (e.g., GTP{gamma}S) or aluminum fluoride (AlF4-), which binds to GDP-bound {alpha}-subunits and mimics the {gamma}-phosphate of GTP.



View larger version (18K):
[in this window]
[in a new window]
 
Figure 1. The Gs GTPase cycle. In the inactive state GDP-bound Gs{alpha} is associated with ß{gamma}-dimers in the inner leaflet of the plasma membrane. Upon binding of hormone to receptor, the activated receptor interacts with the G protein to promote the release of GDP, which is replaced by GTP. GTP binding changes the conformation of Gs{alpha}, leading to dissociation from ß{gamma}, depalmitoylation, and, at least to some extent, release from the membrane. The activated GTP-bound Gs{alpha} interacts with and activates several effectors, including adenylyl cyclase, which catalyzes the production of cAMP, Ca2+ channels, and src tyrosine kinases. The ß{gamma}-dimers can also modulate effectors, including adenylyl cyclase. Gs{alpha}, like all G protein {alpha}-subunits, has an intrinsic GTPase activity that hydrolyzes bound GTP to GDP, which deactivates the G protein and allows reformation of the heterotrimer. Covalent modification of residue Arg201 by cholera toxin or mutation of residues Arg201 or Gln227 inhibits the GTPase activity, producing a constitutively activated form of Gs{alpha} that remains in the active GTP-bound state for a long period of time.

 
X-ray crystal structures reveal that Gs{alpha} (33, 34), like all G protein {alpha}-subunits (35, 36, 37, 38, 39, 40, 41), have two domains, a ras-like GTPase domain, which includes the sites for guanine nucleotide binding and effector interaction, and a more variable helical domain (Fig. 2Go). Various Gs{alpha} isoforms with helical domains of slightly differing length are produced by alternative exon splicing (42, 43). The guanine nucleotide resides in a cleft between the two domains, and the helical domain may be important in preventing the release of GDP from the {alpha}-subunit in the inactive state (40, 44). Comparison of the structures of inactive (GDP-bound) and activated (GTP{gamma}S- or AlF4--bound) {alpha}-subunits reveals three regions within the GTPase domain (named switch 1, 2, and 3) that undergo conformational changes upon activation. The movement of switches 1 and 2 upon GTP binding is in direct response to the presence of the {gamma}-phosphate group, while switch 3 has no direct contact with bound guanine nucleotide. Upon activation, switches 2 and 3 move toward each other and form multiple interactions between acidic and basic amino acid residues that stabilize the active state (45, 46, 47, 48). Residues in switch 3 also make contacts with residues in the helical domain, and these interactions have been implicated in the maintenance of guanine nucleotide binding (40, 44) and in receptor-mediated activation (44, 49, 50, 51). Two residues in the GTPase domain that are conserved in all G{alpha}s (Arg201 and Gln227 in Gs{alpha}, see Fig. 2Go) are critical for catalyzing the hydrolysis of bound GTP, and modification or mutation of these residues produces constitutively active G proteins (37, 38, 52, 53). Crystal structures of G protein heterotrimers indicate that the G{alpha} amino terminus and switch 2 regions are important sites for interaction with the ß-subunit in the GDP-bound state (33, 39, 41).



View larger version (46K):
[in this window]
[in a new window]
 
Figure 2. Tertiary structure of the Gs{alpha} protein and location of some amino residues that are mutated in human endocrine disorders. A model of Gs{alpha} is shown ({alpha}-helices in red, ß-sheets in green) with the helical domain on the left and the ras-like domain on the right. Guanine nucleotide (purple) binds within a cleft between the two domains. Mutation of two residues important for the GTPase "turn-off" reaction [(Arg201 and Gln227), shown in blue] lead to endocrine tumors, MAS, and FD. Mutations of Arg231, Arg258, Glu259, and Arg385 (all shown in yellow) all inactivate Gs{alpha} and lead to AHO. Residues Arg231 in the switch 2 region and Glu259 in the switch 3 region stabilize the active GTP-bound state by forming mutual interactions with residues in switches 3 and 2, respectively (47 48 ). Mutation of Arg258 increases both the rate of basal GDP release (through loss of interactions with helical domain residues) and the rate of the GTPase reaction (44 51 ). Mutation of Arg385, located near the carboxyl terminus (C) uncouples Gs{alpha} from receptors (57 ). Deletion of a nearby carboxyl-terminal residue (Ile382), shown in yellow, specifically uncouples Gs{alpha} from PTH1R receptors and leads to a PHPIB-like phenotype (359 ). Mutation of Ala366 (yellow) increases the basal rate of GDP release and leads to AHO associated with testotoxicosis (345 ). The image was generated with Swiss-PDBViewer, version 3.7b2 (416 ) and rendered with POV-Ray 3.1 using the coordinates for GTP{gamma}S-bound Gs{alpha} (1AZT; Research Collaboratory for Structural Bioinformatics Protein Data Bank) (33 ).

 
Several lines of evidence implicate the carboxyl terminus as being important for interaction with receptors. Mutations of G{alpha} carboxyl-terminal residues (54, 55, 56, 57, 58), covalent modification of a specific carboxyl-terminal residue of Gi/o{alpha} by pertussis toxin (59), antibodies directed to G{alpha} carboxyl termini (60, 61, 62), and G{alpha} carboxyl-terminal peptides (63) all prevent or alter specificity of G protein-receptor coupling. Residues in the {alpha}3/ß5 loop near the carboxyl terminus may also directly interact with receptors and be required for receptor-mediated activation (58). Mutagenesis studies (64, 65), as well as crystal structures (34), have also defined the residues within the GTPase domain of Gs{alpha} that directly interact with adenylyl cyclase.

B. Gs{alpha} gene (GNAS1) structure
The single copy human Gs{alpha} gene (GNAS1) is located at 20q13.2–13.3 (66, 67, 68) while its mouse ortholog (Gnas) is located in a distal portion of chromosome 2 that is syntenic with human 20q13 (69, 70). The overall structure and regulation of this gene are well conserved between human and mouse. Originally these genes were defined by the 13 exons that encode Gs{alpha} (43) (Fig. 3Go). Two long (Gs{alpha}-1 and -2) and two short (Gs{alpha}-3 and -4) forms of Gs{alpha} result from alternative splicing of exon 3, an exon that encodes a stretch of 15 amino acids within the helical domain that are not present in other G{alpha} subunits (42, 43). Use of an alternative splice acceptor site for exon 4 leads to insertion of an extra serine residue in Gs{alpha}-2 and -4. While the ratio of long to short forms of Gs{alpha} may vary from tissue to tissue, there is little evidence to suggest that these alternative Gs{alpha} isoforms have distinct roles in signaling (71), although there may be subtle differences in their biochemical properties (72). Within the intron between exons 3 and 4 is an alternative terminal exon (3N, Fig. 3Go) with its own polyadenylation signal. Splicing to exon 3N produces transcripts that are expressed primarily in neural tissues which do not encode full-length Gs{alpha} (73). The importance of these transcripts, if any, is unknown.



View larger version (28K):
[in this window]
[in a new window]
 
Figure 3. Organization and imprinting of GNAS1. The general organization and imprinting patterns of the GNAS1 maternal (above) and paternal (below) alleles are shown (not drawn to scale) with the exons of sense transcripts depicted as boxes (coding regions, black; noncoding regions, white). Gs{alpha} exons 4–13 are shown as a single box. The five exons of the NESP55 antisense transcripts are shown as gray boxes. A20 and A21 are alternatively spliced exons that are included in a small number of transcripts generated from the XL{alpha}s promoter. Exon 3N is an alternative terminal exon that leads to the generation of a carboxyl-terminally truncated form of XL{alpha}s (XLN1a) and possibly to a truncated form of Gs{alpha}. Transcriptionally active sense and antisense promoters (arrows), as well as the splicing patterns of their respective transcripts, are shown above and below the exons, respectively. The striped arrow for the paternal Gs{alpha} promoter is to indicate that the promoter is fully active in most tissues but is presumed to be silenced in some tissues, such as renal proximal tubules. Regions that are differentially methylated (Meth) are bracketed. Depicted below the maternal and paternal alleles are the mRNA transcripts known to be expressed from each respective allele. The known or predicted protein translation products are designated to the left of their respective mRNA transcripts. The Gs{alpha} transcripts are biallelically expressed in most tissues, but are expressed primarily from the maternal allele in some tissues. Asterisks indicate that alternative splice forms of some transcripts result from the splicing in or out of exon 3. Only the largest of the paternal-specific NESP antisense transcripts is shown (gray boxes; left pointing arrow). The overall organization and imprinting of the mouse ortholog Gnas is very similar to that of GNAS1. Gnas knockout mice were generated by disrupting exon 2 (125 ).

 
The regions upstream and downstream of Gs{alpha} exon 1 are highly GC rich (43, 74) and are contained within a CpG island, which is typical for promoters of ubiquitously expressed genes such as Gs{alpha} [CpG islands are ~1- to 3,000-bp regions that are highly GC rich with a relatively high frequency of CpG dinucleotides and are usually unmethylated (75, 76)]. The GC-rich Gs{alpha} promoter region is unmethylated in both human and mouse (74, 77, 78) (see Section II.D).

It is now known that both GNAS1 and Gnas are more complex genes containing at least four alternative promoters and first exons that splice onto a common exon (exon 2). The most downstream of these exons is Gs{alpha} exon 1 (Fig. 3Go). The most upstream alternative promoter (located ~49 kb upstream of Gs{alpha} exon 1) produces transcripts encoding the chromogranin-like protein NESP55 (78, 79, 80). The entire coding sequence for NESP55 is contained within the upstream exon, and therefore Gs{alpha} exons 2–13 are within the 3'untranslated region (3'-UTR) of the NESP55 transcripts (79, 81). There is some evidence in humans for the presence of a minor splice donor site within the 5'-UTR of the NESP55 exon. In the orthologous exon in mouse (referred to as Nesp), the presence of both splice donor and acceptor sites leads to the presence of a 95-bp intron, which disrupts the 5'-UTR of Nesp. In both human and mouse, the NESP55 promoter region is methylated only on the paternal allele, and NESP55 is only transcribed from the maternal allele (78, 79, 82) (see Section II.D).

A third alternative promoter located approximately 11 kb downstream of the NESP55 exon and about 35 kb upstream of Gs{alpha} exon 1 in humans produces transcripts encoding XL{alpha}s, an isoform of Gs{alpha} with a long amino-terminal extension (77, 78, 80, 83) (Fig. 3Go). The alternative amino-terminal region of XL{alpha}s is encoded by its first exon, while the carboxyl-terminal portion of the protein, which is identical to Gs{alpha}, is encoded by exons 2–13. Just downstream of the XL{alpha}s exon are two small exons of 91 and 67 bp in length (referred to as A20 and A21, Ref. 77). While most XL{alpha}s transcripts do not contain exons A20 and A21, a small proportion of XL{alpha}s transcripts have A20 alone (84) or both A20 and A21 (77) spliced in between the upstream XL{alpha}s exon and exon 2. The presence of these exons within the spliced mRNA disrupts the coding sequence of XL{alpha}s. mRNAs that contain the A20 sequence produce a carboxyl-terminally truncated form of XL{alpha}s called XLN1b (see Fig. 3Go and Ref. 84). A role for these transcripts, if any, remains to be determined. It is not known whether exons comparable to A20 and A21 exist in the orthologous region in mouse (which is referred to as Gnasxl).

The XL{alpha}s promoter region is located within the 3'-half of a large 6-kb CpG island. In both humans and mice, XL{alpha}s is only transcribed from the paternal allele and its promoter is methylated only on the maternal allele (77, 78, 82). The region between the NESP55 and XL{alpha}s promoters is well conserved between human and mice (85). In particular, a highly conserved region located approximately 2–3 kb upstream of the XL{alpha}s exon is a promoter for antisense transcripts that traverse the NESP55 exon from the opposite direction (82, 85) (Fig. 3Go). In humans, a large variety of antisense mRNA transcripts result from alternative splicing of five exons (one upstream and four downstream of the NESP55 upstream exon) and the use of multiple splice sites (85). In the mouse, the transcripts (which have been named Nespas) originate from the same location, but the exact exon structure has not been well defined (82, 85, 86). Unlike the case in humans, the mouse Nespas mRNA product incorporates the complementary sequence spanning the upstream Nesp exon (86). Similar to XL{alpha}s, the NESP55 antisense transcripts are only expressed from the paternal allele, and their promoter is methylated on the maternal allele (82, 85, 86).

A fourth promoter that generates transcripts from the sense strand is located approximately 2.5 kb upstream of Gs{alpha} exon 1 (Fig. 3Go). The alternative first exon [which has been called A/B in humans (87) and 1' in dogs (88), and which we call 1A (74, 89)] splices onto exon 2 through the use of two alternative splice donor sites. There is no consensus AUG translational start site within exon 1A, and the resulting exon 1A transcripts are presumed to be untranslated mRNAs (74, 88). These transcripts are ubiquitously expressed in mouse, and have a tissue distribution pattern that is very similar to that of Gs{alpha} (74). The exon 1A promoter region is located within a CpG island that appears to be distinct from the Gs{alpha} exon 1 CpG island (74, 89). In both humans and mice, the exon 1A promoter is normally only methylated on the maternal allele, and exon 1A mRNAs are only transcribed from the paternal allele (74, 89). As discussed below, imprinting of this region is lost in patients with pseudohypoparathyroidism type IB (PHPIB), and this region is probably important for the tissue-specific imprinting of Gs{alpha}.

C. Alternative GNAS1 gene products
1. XL{alpha}s. XL{alpha}s is an isoform of Gs{alpha} in which the first 47 amino acids encoded by Gs{alpha} exon 1 are replaced by a large amino-terminal region encoded by XL{alpha}s exon 1 (the XL domain) to produce an acidic 78-kDa protein that has an electrophoretic mobility of 94 kDa (83, 84) (Fig. 3Go). Within the XL domain are, in sequential order, a region of Glu-Pro-Ala-Ala (EPAA) repeats, a region of Ala-Ala-Arg-Ala (AARA) repeats, a proline-rich region, and a cysteine-rich region (83, 84). The carboxyl-terminal portion of the XL domain is similar to that of the carboxyl-terminal portion of Gs{alpha} exon 1, a region required for guanine nucleotide binding and ß{gamma} association. There is also evidence for the in vivo expression of carboxyl-terminally truncated forms of XL{alpha}s resulting from the splicing of exon 3N to exon 3 (XLN1a) or the insertion of exon A20 (XLN1b) or exons A20 and A21 (see Refs. 77 and 84 and Fig. 3Go). The biological significance of these so-called XLN1 isoforms is unknown.

While RT-PCR experiments suggest that XL{alpha}s is widely distributed (82), Northern analysis, immunoblotting, and in situ hybridization experiments demonstrate that XL{alpha}s expression is limited to neural and endocrine tissues (83, 84). XL{alpha}s is expressed at very high levels in rat pituitary (particularly in the melanotrophs within the pars intermedia of the posterior lobe), and to a lesser extent in brain, adrenal, heart, and pancreatic islets (where it is expressed in only some cells and is not restricted to ß-cells). There is little or no XL{alpha}s expressed in spleen, liver, or kidney. In the rat, expression of XL{alpha}s begins during the onset of neurogenesis (90). While initial sucrose gradient studies suggested that XL{alpha}s is primarily localized to the trans-Golgi network (83), more recent sucrose gradient and immunofluorescence studies demonstrate that XL{alpha}s is mostly targeted to the plasma membrane (84). However, one study suggests that XL{alpha}s (at least when transfected into cells) may also be targeted to Golgi under certain conditions (91). It is clear that the XL domain is critical for membrane targeting (84, 91). Palmitoylation of cysteines within the cysteine-rich region is required for membrane association, and the proline-rich region is required for targeting to Golgi (91).

Biochemical studies indicate that XL{alpha}s has many properties in common with Gs{alpha} (92). Like Gs{alpha}, XL{alpha}s can be modified by cholera toxin, can interact with ß{gamma}, and is activated by the nonhydrolyzable GTP analog GTP{gamma}S. Also similar to Gs{alpha}, XL{alpha}s proteins that are activated by GTP{gamma}S or by mutating residue Gln548 (analogous to residue Gln227 in Gs{alpha} which when mutated leads to reduced GTPase activity and constitutive activation) can stimulate adenylyl cyclase activity. However, in contrast to Gs{alpha}, there is no evidence that XL{alpha}s can be activated by seven-transmembrane receptors that activate Gs{alpha} (92). Therefore, perhaps XL{alpha}s is a stimulator of adenylyl cyclase that is activated by an alternative pathway. The biological function of XL{alpha}s remains to be determined.

2. NESP55. NESP55 (neuroendocrine-specific protein of apparent molecular mass 55 kDa) is an acidic chromogranin that localizes to large dense-core granules within neuroendocrine tissues, such as the adrenal medulla, anterior and posterior pituitary, hypothalamus, and other brain regions (81, 93). Within the adrenal medulla, NESP55 is more highly expressed in epinephrine-, as opposed to norepinephrine-secreting chromaffin cells (94). Within the central nervous system, NESP55 is expressed in neural, but not glial, cells and is most prominent in the pituitary, hypothalamus, and other midbrain and brainstem regions, but absent in the neocortex, hippocampus, and cerebellum (95). Its brain distribution overlaps the distribution pattern of the norepinephrine, epinephrine, and serotonergic transmitter systems. There is also some NESP55 expression in the intestine, where it undergoes extensive proteolytic cleavage (93). The NESP55 protein is highly conserved between species and is a predicted size of approximately 27–29 kDa (96). However, it is very acidic due to addition of keratin sulfate glycosaminoglycan chains (and therefore has an apparent molecular mass of 55 kDa on immunoblots) (81, 96) and undergoes proteolytic cleavage. During midgestation in the mouse, Nesp transcripts are expressed in somites and vasculature (97). Little is known about the physiological role of NESP55, although it has been proposed that one potential proteolytic cleavage product may be an endogenous antagonist of serotonin 1b receptors (81). Loss of NESP55 expression in humans is not associated with an obvious phenotype (89).

D. Imprinting of the GNAS1 gene
Genomic imprinting is an epigenetic phenomenon affecting a small number of autosomal genes (probably 100 or less) that leads to differences in gene expression between the maternal and paternal allele (98, 99, 100, 101). Mutation or dysregulation of imprinted genes is associated with several human congenital diseases, including Beckwith-Wiedemann, Prader-Willi, and Angelman’s syndromes, and in carcinogenesis. The first clue to the existence of imprinting was the observation that both androgenetic and parthenogenetic zygotes develop abnormally, demonstrating that normal development requires both a paternal and maternal contribution to the genome (102, 103). Subsequently, specific chromosomal regions in the mouse were identified that were presumed to contain imprinted genes due to the fact that uniparental disomy (UPD) of these regions produced abnormal phenotypes (104, 105). (UPD is the inheritance of a chromosome or chromosomal region from a single parent.) Often these imprinted regions contain several closely linked imprinted genes that are coordinately regulated. One such imprinted region is in the distal portion of chromosome 2 and includes Gnas (70, 104).

In order for the parental alleles of imprinted genes to behave differently, there must be a mark placed on the gene in either the male or female gamete that is maintained in all somatic cells throughout development. This mark would have to be erased in the primordial germ cell so that the appropriate imprinting patterns could be reestablished in the mature oocyte or spermatozoa, respectively. Virtually all imprinted genes have one or more regions in which the cytosines within CpG dinucleotides are methylated on only one parental allele, and differential methylation is the most likely candidate for the imprinting mark (99, 101). Imprinting is lost in mice lacking the DNA methyltransferase DNMT1 (106, 107). In many instances, methylation within the differentially methylated regions (DMRs) is established in either the male or female gamete and maintained throughout pre- and postimplantation development, and deletion of these DMRs disrupts the imprinting of one or more genes in the imprinted region (108, 109, 110, 111, 112). As differential methylation of these regions is probably important for establishing imprinting, these regions have been referred to as methylation imprint marks or core DMRs. In contrast, other regions become differentially methylated at a later time in development, and while these regions may be important to maintain imprinted expression, they do not provide the initial signal that marks parental origin.

In many, but not all, cases, DMRs coincide with the gene promoter and lead to allele-specific expression through silencing of the methylated promoter (98, 99, 113). This occurs through the binding of methyl-CpG binding proteins, which recruit histone deacetylases, leading to a more compact chromatin that is not accessible to transcriptional factors (99, 101, 113, 114). There is also recent evidence that the DNA methyltransferases also recruit histone deacetylases (115, 116). Methylation itself may prevent the binding of transcriptional factors that are required for gene transcription. However, differential methylation of imprinted genes can lead to allele-specific expression through alternative mechanisms. For example, the H19 DMR is located between the Igf2 promoter and a set of endoderm-specific enhancers and contains several boundary elements (or insulators). On the maternal allele the DMR is unmethylated and is therefore able to bind the insulator protein CCCTC-binding factor (CTCF) and insulate the Igf2 promoter from the enhancers. As a result, Igf2 is not expressed from the maternal allele. On the paternal allele the DMR (and its boundary elements) are methylated. CTCF is unable to bind to the methylated boundary elements. Consequently, the DMR does not insulate the Igf2 promoter from its enhancers, allowing Igf2 to be expressed from the paternal allele (117, 118). A similar mechanism has been recently proposed for the DLK1-GTL2 imprinting cluster (119, 120).

GNAS1 has a complicated imprinting pattern that leads to the expression of some gene products from the maternal allele and others from the paternal allele (Fig. 3Go). The imprinting pattern of the mouse ortholog Gnas is essentially identical to that of GNAS1. The closely linked promoter regions for NESP55 and XL{alpha}s are oppositely imprinted. NESP55 and its associated transcripts are only expressed from the maternal allele, and its promoter region is methylated on the paternal allele (78, 79, 82). In contrast, XL{alpha}s and its associated transcripts are only expressed from the paternal allele, and its promoter region is methylated on the maternal allele (77, 78, 82). Based on the similar tissue distribution but opposite imprinting patterns, it is possible that these two imprinted regions are coordinately regulated. Paternal-specific methylation of the NESP55 promoter is not established until postimplantation development in mouse, and therefore this region is not the methylation imprint mark for the Gnas locus (74). More likely its imprinting is dependent on another imprinting event that occurs at an earlier stage in development.

The NESP antisense transcript is expressed only from the paternal allele, and its promoter (which is located in the same CG-rich region as the XL{alpha}s promoter) is methylated on the maternal allele (82, 85, 86). Interestingly, paternal-specific antisense transcripts have been identified in several imprinted genes (111, 121, 122, 123). One potential mechanism for the imprinting of NESP55 is that the paternal antisense transcript leads to silencing of the NESP55 promoter on the paternal allele. This is exactly analogous to the situation in the maternally expressed Igf2r gene (111). Downstream of the Igf2r promoter is a promoter for a paternal-specific antisense transcript that is methylated on the maternal allele. The maternal-specific methylation of the antisense promoter is established in the oocyte and is maintained throughout development and is therefore a methylation imprint mark, whereas methylation and silencing of the paternal Igf2r promoter is established later in development. Moreover, deletion of the differentially methylated antisense promoter region leads to loss of imprinting of Igf2r. Further studies will be required to determine when methylation of the NESP antisense promoter is established and whether or not it is required for NESP55 imprinting. One study suggested the NESP antisense expression is biallelically expressed in two tissues (82). However, one of these tissues was testis, where the transcript would be predicted to be biallelically expressed because the promoter is unmethylated in male germ cells. An in situ hybridization study of midgestational mouse embryos showed that Nesp and Nespas transcripts do not colocalize, indicating that at least during gestation Nespas probably does not play a role in the imprinting of Nesp (97).

The temporal methylation pattern of the XL{alpha}s promoter region in mice is atypical as the region appears to be partially methylated in both spermatozoa and oocytes, and this methylation is maintained in preimplantation blastocysts (at a time when the genome is undergoing global demethylation) (74). These findings suggest that allele-specific methylation is not erased in germ cells and that paternal-specific methylation is somehow reestablished at some point during postimplantation development by a novel mechanism. The role of the XL{alpha}s promoter region in establishing GNAS1 imprinting is presently undefined.

Clinical genetic studies of AHO patients (8, 124) strongly suggest that Gs{alpha} is imprinted in a tissue-specific manner, being biallelically expressed in most tissues but maternally expressed in a few tissues, such as the renal proximal tubules, and this has been confirmed in Gnas knockout mice (125). Although earlier studies did not show imprinting of Gs{alpha} in humans (77, 79, 126), Gs{alpha} has recently been shown to be imprinted in normal human pituitary glands (127). Even in tissues where Gs{alpha} is imprinted, the imprinting is not absolute in that there is some expression from the paternal allele (125, 127). The Gs{alpha} promoter is not methylated in either allele, and therefore allele-specific expression of Gs{alpha} is not due to differential methylation of its promoter (74, 77, 78). NESP55 and XL{alpha}s are expressed in tissues where Gs{alpha} is not imprinted (77, 79), while neither is expressed in tissues where Gs{alpha} is imprinted (Refs. 84, 93 and 128 and S. Yu and L. S. Weinstein, unpublished data). It is therefore unlikely that either NESP55 or XL{alpha}s or their promoters are involved in the imprinting of Gs{alpha}. Tissue-specific imprinting of Gs{alpha} and the role of Gs{alpha} imprinting in human disease are discussed in more detail in Sections IV.D, V.A, and VI.

The exon 1A promoter region is methylated on the maternal allele, and exon 1A mRNAs are only expressed from the paternal allele (74, 89). In mice the maternal-specific methylation of the exon 1A promoter is established in the oocyte and maintained throughout pre- and postimplantation development (74); therefore, this region appears to be a methylation imprint mark that may be important to establish imprinting throughout the GNAS1 locus. Like the NESP antisense transcript, the exon 1A mRNAs are untranslated mRNAs. Untranslated mRNAs are a common feature of imprinted genes, although their role in the imprinting mechanism, if any, is unknown. The close proximity of the exon 1A DMR to the Gs{alpha} promoter and the abnormal imprinting of this region in PHPIB (89), a disease likely to result from abnormal imprinting of Gs{alpha}, strongly suggests that this region is critical for the tissue-specific imprinting of Gs{alpha}. Potential mechanisms by which this region may lead to Gs{alpha} imprinting are discussed in Section VI.B.

As imprinted genes often occur in clusters, it is important to identify the genes most closely linked to GNAS1 and determine whether or not they are also imprinted. The two closest genes located downstream of GNAS1 are TH1, a gene of unknown function, and CTSZ, which encodes cathepsin Z, and initial results suggest that neither gene is imprinted (129). The imprinting status of neighboring genes upstream of GNAS1 remains to be established. In the mouse, 14 genes close to Gnas that are located within the 20-centimorgan distal chromosome 2 imprinted region were shown not to be imprinted (130, 131).


    III. Gs{alpha} Activating Mutations
 Top
 Abstract
 I. Introduction
 II. Gs{alpha}-Signaling...
 III. Gs{alpha} Activating...
 IV. Gs{alpha} Loss-of-Function...
 V. Insights from Gnas...
 VI. GNAS1 Imprinting Mechanisms...
 VII. Summary
 References
 
The first evidence that inappropriate Gs{alpha} activation can lead to pathophysiological consequences was provided by the fact that the secretory diarrhea characteristic of intestinal V. cholerae infection results from a covalent modification of Gs{alpha} that leads to constitutive activation. V. cholerae secretes an exotoxin named cholera toxin that catalyzes the ADP ribosylation of a specific Gs{alpha} residue (Arg201) that is critical for the G protein’s GTPase "turn off" mechanism (4). As a result of markedly reduced GTPase activity, Gs{alpha} remains in its active GTP-bound form for a long period of time, resulting in constitutive Gs activation (Fig. 1Go). Excess intracellular cAMP in intestinal cells alters ion transport, leading to secretory diarrhea. Subsequently, it has been established that missense mutations that encode substitutions of Arg201 or residue Gln227 also lead to decreased GTPase activity and constitutive activation (52, 53, 132) (Fig. 2Go), and such mutations have been identified in sporadic endocrine tumors, the McCune-Albright syndrome (MAS), and fibrous dysplasia of bone (FD).

A. Endocrine tumors
In many endocrine glands, cAMP stimulates both proliferation and hormone secretion, and therefore mutations that lead to increased cAMP levels, such as activating Gs{alpha} mutations, would be predicted to lead to differentiated endocrine tumors with increased rates of hormone release. The first step toward the identification of activating Gs{alpha} mutations in a human disease was the observation that a subset of human pituitary GH-secreting tumors have high basal levels of GH release and membrane-associated adenylyl cyclase activity that could not be further stimulated by GHRH, which normally stimulates adenylyl cyclase through its Gs-coupled receptor (133). These tumors, which account for 40% of GH-secreting tumors, were shown to contain heterozygous missense mutations in GNAS1 exons 8 or 9 that encode substitutions of Gs{alpha} residues Arg201 (to Cys or His) or Gln227 (to Arg or Lys), respectively (52, 134). Subsequently, a tumor containing an Arg201-to-Ser mutation was identified (135). These mutations were of somatic origin, as the same mutations were not present in peripheral blood samples from the same patients. Mutation of either Arg201 or Gln227 leads to decreased intrinsic GTPase activity (52, 53, 132), and the catalytic role of these amino acid residues in the GTPase reaction is supported by data from x-ray crystallography (37, 38) (Fig. 2Go). The mutated Gs{alpha} is therefore constitutively activated, leading to activation of adenylyl cyclase and further downstream physiological responses. In GH-secreting pituitary cells, cAMP stimulates proliferation and differentiation by inducing the tissue-specific transcription factor GHF1 (136). Consistent with these findings in human GH-secreting tumors, transgenic mice expressing cholera toxin (which covalently modifies Gs{alpha} Arg201) in their somatotrophs developed pituitary hyperplasia and excess GH secretion (137). Activating mutations are dominant acting, and such constitutively activated forms of Gs{alpha} have been designated as the gsp oncogene (52, 134).

In gsp+ GH-secreting pituitary tumors, mRNA derived from the gsp+ allele is often more abundant than mRNA derived from the wild-type allele (52, 127, 138). A recent study showed that Gs{alpha} is imprinted and expressed from the maternal allele in normal pituitary glands and that in 21 of 22 gsp+ GH-secreting tumors the gsp mutation was present on the maternal allele (127). Presumably, the gsp mutation only leads to significant expression of activated Gs{alpha} (and therefore tumorigenesis) when present on the active maternal allele (127, 139). In both gsp+ and gsp- GH-secreting tumors, there is variable loss of Gs{alpha} imprinting leading to a variable amount of Gs{alpha} expression from the normally inactive paternal allele (127). Whether this is critical for tumorigenesis or represents an epiphenomenon and whether loss of Gs{alpha} imprinting in pituitary tumors is associated with altered GNAS1 methylation remain to be determined (139).

There are few clinical differences between gsp+ and gsp- GH-secreting tumors (140, 141). Gsp+ tumors tend to be smaller than gsp- tumors and maintain their responsiveness to somatostatin analogs in terms of inhibition of GH secretion (135, 142, 143). This indicates that Gi pathways, which inhibit adenylyl cyclase, are preserved and that somatostatin analogs are useful therapeutic agents for gsp+ GH-secreting tumors. There are several mechanisms that counteract the effect of gsp mutations on intracellular cAMP levels. The levels of Gs{alpha} protein are reduced in gsp+ pituitary tumors, presumably due to more rapid turnover of activated Gs{alpha} proteins (138). Also, Gs{alpha} activation in gsp+ tumors leads to increased levels of a specific subtype of phosphodiesterase (phosphodiesterase 4), which hydrolyzes cAMP and therefore counteracts the increased rate of cAMP formation (144). Phosphodiesterase 4 expression was also elevated in one gsp+ thyroid tumor (145).

Gsp mutations are also rarely present in other pituitary tumors. Similar to GHRH in somatotrophs, CRH activates Gs in corticotrophs to stimulate ACTH release, and therefore one might predict gsp mutations to be present in corticotroph adenomas. One study identified gsp mutations in 2 of 32 corticotroph tumors (146), while another study failed to find mutations in 7 such tumors (134). It is interesting to note that corticotroph tumors are not a feature of MAS, in which the distribution of gsp mutations is widespread. Results from three studies indicate that gsp mutations are also present in about 10% of clinically nonfunctional pituitary adenomas (134, 147, 148). Studies have failed to identify gsp mutations in either thyrotroph or lactotroph tumors (134, 149). This is not surprising given the fact that their respective hypothalamic stimulating hormones do not work through Gscoupled pathways.

Constitutively activated Gs{alpha} transfected into the FRTL5 thyroid cell line leads to increased growth and thyroid hormone release (150), and transgenic mice with thyroid-specific expression of cholera toxin develop thyroid hyperplasia and hyperthyroidism (151). In thyroid cells, cAMP stimulates both cAMP-dependent transcription factors and the p38 MAPK in a PKA-dependent manner (152). Several studies have confirmed that gsp mutations are present in a small subset of autonomously functioning thyroid tumors (134, 153). More common are activating TSH receptor mutations, which activate the same Gs-coupled pathway (154, 155). The gsp mutations are also rarely found in differentiated thyroid carcinomas (153, 156, 157). One study showed that follicular neoplasms have increased levels of presumably normal Gs{alpha} (158). While gsp mutations have been identified in parathyroid and adrenocortical tumors and in pheochromocytomas (159, 160), other studies suggest that these mutations are rarely found in these or other endocrine tumors (134, 160, 161, 162).

B. McCune-Albright syndrome (MAS)
MAS was first described by McCune (163) and Albright and co-workers (164) in the 1930s and is classically defined as the concurrence of hyperpigmented (café-au-lait) skin lesions, sexual precocity, and FD (for reviews, see Refs. 6 and 165, 166, 167, 168, 169). MAS patients may also present with other endocrine or nonendocrine abnormalities. Some MAS patients present with only two features of the classic triad or one of these features and another endocrine or nonendocrine abnormality (170, 171, 172). In virtually every case, MAS occurs sporadically and does not appear in the subsequent generation.

The café-au-lait spots in MAS are single or multiple tan-brown hyperpigmented flat macules with irregular ("coast of Maine") borders (164) that become more obvious with age or with sun exposure (172, 173). Often these lesions are limited to one side of the body, which usually corresponds to the side with bone involvement and generally do not cross the midline. They are often arranged in a segmental pattern, which follows the developmental lines of Blaschko (174). Melanocytes cultured from these lesions have increased intracellular cAMP levels, increased numbers of dendrites and melanosomes, and increased levels of tyrosinase, the rate-limiting enzyme for the production of melanin (175). In MAS patients usually more than one bone is affected by FD, which is described in more detail in the next section.

In female MAS patients the first sign of sexual precocity is premature menses, sometimes occurring within the first months of life (172, 173, 176). Serum estrogen levels fluctuate, with peaks corresponding to the presence of large ovarian follicles (171, 176, 177). Usually females undergo normal development during adolescence and show normal reproductive function in adult life. The typical signs of sexual precocity in males include testicular enlargement and premature onset of secondary sex characteristics and spermatogenesis (172, 173). In one case, testicular enlargement in the absence of pubertal development was due to premature maturation of the seminiferous tubules but not of the Leydig cells (178).

Other endocrine abnormalities associated with MAS include the presence of hyperfunctional thyroid nodules, usually leading to hyperthyroidism (172, 173, 179, 180, 181), macronodular adrenal hyperplasia or adrenal adenomas leading to hypercortisolism (172, 173, 182, 183), pituitary tumors leading to acromegaly and/or hyperprolactinemia (172, 173, 184), and hypophosphatemic rickets or osteomalacia (172, 173, 185). In each case the peripheral endocrine organs (thyroid, adrenal cortex, gonads) behave as if they are being overstimulated by the pituitary even though circulating levels of their respective stimulatory pituitary hormone (e.g., TSH, ACTH, gonadotropins) are suppressed (172, 173, 179, 180, 186, 187, 188, 189). Pituitary adenomas and hyperplasia have not been associated with excess secretion of other pituitary hormones in MAS.

Hyperphosphaturic hypophosphatemic rickets or osteomalacia has been associated with MAS and polyostotic fibrous dysplasia (POFD) (172, 173, 190, 191, 192, 193, 194, 195, 196). In one study increased phosphate excretion was documented in almost 50% of patients with MAS or POFD (196). Some have postulated that hyperphosphaturia in MAS is due to a phosphaturic factor (phosphatonin) secreted from areas of fibrous dysplasia, similar to the mechanism of tumor-associated hypophosphatemic rickets (191, 194, 196), while others implicate a primary defect in the renal proximal tubule (193, 195). Increased intracellular cAMP levels in renal proximal tubules, even in the absence of hyperparathyroidism, leading to decreased phosphate reabsorption, could be one potential mechanism for hypophosphatemia in MAS patients. One study showed that MAS patients with hypophosphatemia have increased basal levels of urinary cAMP (197), while another study reported normal levels of urinary cAMP in these patients (196). In addition to increased phosphate clearance, patients with MAS or POFD have other evidence of renal tubular dysfunction, including aminoaciduria and mild proteinuria (196). These abnormalities are similar to those observed in patients with tumor-induced osteomalacia, who presumably develop renal abnormalities due to excess circulating levels of a phosphatonin [possibly fibroblast growth factor 23 (198)]. Although hyperparathyroidism has been reported to occur in patients with monoostotic fibrous dysplasia (MOFD) usually affecting the jaw (199, 200, 201), it is likely that most or all of these patients have hyperparathyroidism-jaw tumor syndrome that has been mapped to chromosome 1 (202, 203). Hyperparathyroidism is rarely present in patients with the classic MAS clinical triad (204, 205).

The abnormalities in MAS patients are generally restricted to bone, skin, and endocrine organs, and therefore there is little effect on mortality (172, 206, 207). However, some patients also develop one or more nonendocrine abnormalities that may markedly increase morbidity and mortality (208). Often these patients also have extensive POFD or hypercortisolism (206, 208). Nonendocrine organs that may be affected in MAS include the liver, heart, thymus, spleen, bone marrow, gastrointestinal tract, and brain (208, 209). Liver abnormalities associated with MAS include severe neonatal jaundice, elevated liver enzymes, and cholestatic and biliary changes on liver biopsy (208, 210). Cardiac abnormalities may include cardiomegaly associated with atypical myocyte hypertrophy, persistent tachycardia, and unexplained sudden death (208). Other rare features of MAS include thymic hyperplasia, myelofibrosis, gastrointestinal polyps, pancreatitis, breast cancer, microcephaly, and other neurological abnormalities (164, 172, 208, 211, 212, 213).

Based on the fact that the cutaneous hyperpigmentation in MAS often follows the developmental lines of Blaschko, Happle proposed that MAS is likely to be caused by a dominant somatic mutation occurring early in development or a gametic half-chromatid mutation (174). Either of these early mutational events would produce a mosaic with a widespread distribution of mutant cells. Because MAS is never inherited, it is likely that the underlying mutation is lethal if present in the germline. According to this model, the specific constellation of abnormalities within a given patient would be dependent on the distribution of mutant-bearing cells.

Although it was originally proposed that the increased growth and secretory function of diverse peripheral endocrine organs in MAS results from a hypothalamic defect (214), it is now clear that each of the peripheral endocrine glands functions autonomously in the absence of its respective trophic pituitary hormone (173, 179, 186, 187, 188, 189). These trophic hormones (e.g., gonadotropins, TSH, ACTH, GHRH) normally stimulate Gs and cAMP generation in their respective target tissues, leading to increased proliferation and hormone secretion (173, 188, 207, 215). Moreover, {alpha}-MSH normally raises cAMP levels in melanocytes by binding to the Gs-coupled melanocortin 1 receptor, leading to increased tyrosinase activity and melanin production (216, 217). A genetic lesion that leads to increased cAMP generation, even in the absence of stimulating hormone, such as a gsp mutation, could explain the endocrine and skin manifestations of MAS. This idea was supported by biochemical studies that showed cAMP production to be increased in tissues isolated from MAS patients (175, 197, 218, 219).

The gsp mutations coding for Gs{alpha} Arg201 substitutions have been identified in tissues from many MAS patients (208, 209, 220, 221). In each case, one specific mutation (either Arg201 to His or Cys) is detected in multiple tissues in variable abundance, consistent with a somatic mutation with a widespread distribution. More recently, an Arg201 to Gly mutation was identified in one MAS patient (222), and an Arg201 to Leu mutation, which was possibly germline, was identified in a patient with severe skeletal, endocrine, and developmental abnormalities (223). The gsp mutations coding for Gs{alpha} Gln227 substitutions have not been found in MAS, perhaps because these mutations are more activating than Arg201 mutations (52) and therefore may lead to a more lethal phenotype. Consistent with the mutation being dominant, the wild-type allele is always present in equal or greater abundance than the mutant allele, indicating that the affected cells are heterozygous for the gsp mutation.

The gsp mutations have been identified in affected endocrine tissues and in café-au-lait lesions from MAS patients (175, 208, 209, 221), and it is likely that these lesions are the result of increased intracellular cAMP levels. gsp Mutations have also been identified in other normal and abnormal nonendocrine tissues (208, 209, 210). It seems possible that increased activity of Gs pathways, which normally mediate the chronotropic and ionotropic responses to sympathetic stimulation, could lead to the cardiac hypertrophy and sudden death occasionally found in MAS patients (208). Some of these effects on cardiac function may result from increased PKA-dependent activation of p38 MAPK (224). The role of gsp mutations in the pathogenesis of these and other nonendocrine manifestations needs to be defined.

The clinical features present in an individual MAS patient are primarily determined by the distribution of cells that bear the gsp mutation. Patients who present with only one or two features of MAS have the same somatic mutations, but in a more limited tissue-specific distribution (presumably these mutations occur at a later time in development). Given that in some tissues Gs{alpha} is preferentially expressed from the maternal allele, MAS patients with gsp mutations in the maternal allele are perhaps more severely affected than those patients with mutations in the paternal allele. Support for this comes from the fact that in gsp+ GH-secreting tumors, the gsp mutation is almost always on the maternal allele (127). Activating mutations within the paternally expressed XL{alpha}s protein also lead to constitutive activation of adenylyl cyclase (92). Whether activated XL{alpha}s contributes to the clinical manifestations in MAS patients with paternal gsp mutations is unknown.

C. Fibrous dysplasia of bone (FD)
FD is a focal and benign fibrous bone lesion that was first described about 50 yr ago (225). Most FD patients (~70%) have a single bone lesion (MOFD), while the remainder have multiple lesions (POFD) (226). A small minority of POFD patients have other features of MAS. With rare exceptions (227, 228, 229), FD is a sporadic disease that is almost never inherited. FD lesions are typically found in the long bones of the extremities, the ribs, and craniofacial bones and are rarely found in the hands, feet, or spine (206, 226, 230). FD lesions may be asymptomatic but often lead to bone deformity, pain, pathological fractures, or cranial nerve compression (172, 206, 230). In some patients FD is associated with intra-muscular myxomas (Mazabraud’s syndrome) (193, 231, 232, 233), and gsp mutations have been identified in intramuscular myxomas both in patients with and without accompanying FD (233). The clinical, radiological, and histological features of FD are reviewed in Refs. 165 and 166 .

FD is a lesion composed mainly of fibrous tissue that originates in the medullary cavity and expands concentrically outward into the surrounding cortical bone. Most of the cells are immature mesenchymal cells with a spindle-shaped fibroblastic appearance (234). These cells express alkaline phosphate and other osteoblast-specific proteins (235). Within the fibrous tissue are spicules of immature woven bone. The histological pattern of FD varies depending on location, but all lesions have common characteristic features (236). For example, the woven bone is lined by flat cells (234) with retracted cell bodies, leading to the formation of pseudolacunar spaces (235, 236). This retraction is likely to be the result of increased intracellular cAMP, as cAMP induces retraction in cultured osteoblasts (235). In addition, in FD the collagen fibrils are perpendicular to the bone-forming surface (so-called Sharpey’s fibers) (235, 236, 237). In some FD lesions, islands of hyaline cartilage are present, and this can sometimes be a dominant feature (234, 238). Often FD lesions have osteomalacic changes, even in patients without hypophosphatemia (237).

FD originates in the inner marrow cavity and expands into surrounding cortical bone through the bone-resorbing activity of osteoclasts, which are present in greater numbers in the periphery (218, 235). Bisphosphonates have been shown to have therapeutic benefit, presumably due to their antiresorptive activity (239, 240, 241, 242). These lesions rarely expand beyond the normal boundaries of the cortical bone (243, 244) or undergo malignant degeneration (245, 246, 247, 248).

As in MAS, activating Gs{alpha} mutations have also been found in FD lesions from MAS, POFD, and MOFD patients, indicating that all forms of FD have the same underlying genetic abnormality (219, 220, 237, 249, 250, 251, 252). In most cases the mutation encodes Arg201 to Cys or Arg201 to His substitutions, although an Arg201 to Ser mutation was found in one POFD patient (252).

Recent studies suggest that FD results from abnormal proliferation and differentiation of bone marrow stromal cells (253). Under normal conditions, marrow stromal (CFU-F) cells differentiate into mature osteoblasts through a series of defined steps: a proliferative phase in which precursor cells divide and a postproliferative phase in which the cells differentiate and express osteoblast-specific gene products (e.g., alkaline phosphatase, osteopontin, and osteocalcin) that support the formation of mineralized bone (254). Cells isolated from FD lesions have an increased proliferation rate and are poorly differentiated, expressing early, but not late, markers of osteoblast differentiation (219, 235). Transplantation of these cells into immunocompromised mice produces fibrotic lesions that are similar to human FD, with abnormal woven bone ossicles and no interspersed adipocytes or hematopoietic elements (237, 255), further suggesting that the normal differentiation program is disrupted in these cells. Interestingly, these lesions are only formed when both normal and mutant cells were transplanted, suggesting that FD requires the presence of normal as well as mutant osteogenic cells.

Several lines of evidence suggest that elevated cAMP may stimulate the proliferation and inhibit the differentiation of osteogenic precursors and therefore lead to FD. Cell culture studies demonstrate that PTH and forskolin, which both stimulate cAMP production, inhibit osteoblast differentiation (256, 257, 258), while treatment of cells with a Gs{alpha} antisense oligonucleotide (which decreases Gs{alpha} expression) promotes osteoblast differentiation (259). Increased cAMP, through activation of PKA, increases the expression of c-fos and other genes (260). cAMP binds to the PKA regulatory subunits, which then dissociate from the catalytic subunits. The catalytic subunits translocate to the nucleus and phosphorylate transcription factors [e.g., cAMP-response element binding protein (CREB)]. Phosphorylated cAMP response element binding protein (CREB) binds to the promoter of c-fos, leading to increased expression of the protein Fos (261). Fos overexpression in osteogenic cells in response to increased cAMP appears to be an important common pathway in the development of FD. Fos is one component of activator protein-1 (AP-1) heterodimers, which themselves are transcriptional factors. AP-1 is present at high levels in proliferating osteoblasts, and its level declines markedly as these cells differentiate (254). One mech