Studies of the regulation and function of the G_sα gene Gnas using gene targeting technology

Abstract

The heterotrimeric G protein α-subunit G_sα is ubiquitously expressed and mediates receptor-stimulated intracellular cAMP generation. Its gene Gnas is a complex imprinted gene which uses alternative promoters and first exons to generate other gene products, including the G_sα isoform XLαs and the chromogranin-like protein NESP55, which are specifically expressed from the paternal and maternal alleles, respectively. G_sα itself is imprinted in a tissue-specific manner, being biallelically expressed in most tissues but paternally silenced in a few tissues. Gene targeting of specific Gnas transcripts demonstrates that heterozygous mutation of G_sα on the maternal (but not the paternal) allele leads to early lethality, perinatal subcutaneous edema, severe obesity, and multihormone resistance, while the paternal mutation leads to only mild obesity and insulin resistance. These parent-of-origin differences are the consequence of tissue-specific G_sα imprinting. XLαs deficiency leads to a perinatal suckling defect and a lean phenotype with increased insulin sensitivity. The opposite metabolic effects of G_sα and XLαs deficiency are associated with decreased and increased sympathetic nervous system activity, respectively. NESP55 deficiency has no metabolic consequences. Other gene targeting experiments have shown Gnas to have two independent imprinting domains controlled by two different imprinting control regions. Tissue-specific G_sα knockout models have identified important roles for G_sα signaling pathways in skeletal development, renal function, and glucose and lipid metabolism. Our present knowledge gleaned from various Gnas gene targeting models are discussed in relation to the pathogenesis of human disorders with mutation or abnormal imprinting of the human ortholog GNAS.

1. Introduction

Heterotrimeric G proteins are a group of membrane-associated complexes that are primarily involved in signaling at the plasma membrane and other intracellular sites. Each is composed of an α-subunit, which binds guanine nucleotide and mediates signals from seven transmembrane receptors to downstream effectors, as well as a βγ dimer. α, β, and γ subunits are the products of separate genes, and G proteins are generally defined by their specific α-subunit. The G_sα-subunit G_sα is a ubiquitously expressed G protein which primarily couples cell surface receptors to the stimulation of adenylyl cyclase and generation of cAMP.

Heterozygous inactivating G_sα mutations lead to Albright hereditary osteodystrophy (AHO), which is characterized by short stature, subcutaneous ossifications, brachydactyly (shortening and widening of long bones in the hands and feet), and neurobehavioral abnormalities (Spiegel & Weinstein, 2001; Spiegel & Weinstein, 2004; Weinstein, 2004). Patients who inherit AHO from their mother also develop obesity and resistance to several hormones (parathyroid hormone [PTH], thyroid stimulating hormone [TSH], gonadotropins, growth hormone) which signal through G_sα in their target tissues, a condition known as pseudohypoparathyroidism type 1a (PHP1a) (Long et al., 2006; Weinstein, 2001), while patients who inherit the same mutations on the paternal allele develop only the AHO phenotype, a condition also known as pseudopseudohypoparathyroidism (PPHP). These parent-of-origin differences are due to the fact that G_sα expression from the paternal allele is suppressed in various tissues due to genomic imprinting (Germain-Lee et al., 2002; Hayward et al., 2001; Liu et al., 2003; Mantovani et al., 2002). (Fig. 2) A small number of patients develop extensive subcutaneous ossifications which invade the deeper soft tissues (progressive osseous heteroplasia, POH) (Eddy et al., 2000; Kaplan & Shore, 2000; Shore et al., 2002; Yeh et al., 2000). A more isolated form of PTH resistance, known as pseudohypoparathyroidism type 1b (PHP1b), results from a gene imprinting defect leading tissue-specific G_sα deficiency (Bastepe et al., 2003; Liu et al., 2000a; Liu et al., 2005b; Weinstein, 2004). Somatic G_sα mutations that result in constitutive signaling may lead to endocrine tumors, fibrous dysplasia of bone, and other features of the McCune-Albright syndrome (Levine, 1999; Weinstein, 2002, 2004).

Fig. 2.

Tissue-specific G_sα imprinting and the effects of heterozygous inactivating G_sα mutations. In renal proximal tubules (upper panels) G_sα is paternally imprinted (silenced, denoted with X). Mutation (Mut) on the active maternal allele (left panel) leads to G_sα deficiency and PTH resistance while mutation on the silenced paternal allele (right panel) has little effect on G_sα expression or PTH sensitivity. This is shown in immunoblots of renal cortical membranes isolated from E2^m−/+ (m−/+) and E2^+/p− (+/p−) mice, respectively, and their wild type (WT) littermates (S. Yu et al., 1998). Studies in mice and humans shows G_sα to also be imprinted in thyroid, ovaries, pituitary somatotrophs, and neonatal BAT. In most other tissues (lower panels) G_sα is not imprinted and therefore heterozygous mutation leads to ~50% loss of G_sα expression (haploinsufficiency) whether or not the mutation is inherited maternally or paternally. This is demonstrated in immunoblots of renal inner medulla membranes from the same mice. Adapted from Weinstein et al., 2004.

Studies over the last decade have shown that the G_sα gene (GNAS in human, Gnas in mouse) is a complex imprinted gene that encodes multiple gene products through the use of alternative promoters and first exons (Weinstein et al., 2001). Gnas is unique in that is the only example in which oppositely imprinted gene products (some expressed from the maternal allele, others from the paternal allele) are expressed from the same transcriptional unit. This complexity has made it difficult to determine which gene products are critical for normal physiology and human disease, and how the various regulatory and promoter regions interact to maintain their normal epigenetic states. Various mouse models developed over the past decade in which regulatory or coding regions of Gnas were disrupted by gene targeting have been very important in understanding how the gene is regulated and the role of its gene products in normal physiology and disease. This review will summarize what has been learned from gene targeting studies of Gnas.

2. General organization of the Gnas gene

The mouse and human G_sα genes (Gnas and GNAS, respectively) are located within syntenic regions (distal chromosome 2 in mouse, 20q13.2–13.3 in humans) and have similar overall organization (Gejman et al., 1991; Levine et al., 1991; Peters et al., 1994; Rao et al., 1991). Both orthologs are also undergo genomic imprinting and have similar overall imprinting patterns. Genomic imprinting is an epigenetic phenomenon which results in partial or complete suppression of gene expression from one parental allele (Reik & Walter, 2001). The primary imprint ‘marks’ (generally assumed to be allele-specific differences in DNA methylation or histone modification) which distinguishes the two parental alleles are erased in primordial germ cells, reestablished in either male or female gametes, and maintained in somatic tissues throughout development. DNA methylation is the most well established imprint mark, as all imprinted genes have one or more regions in which the maternal and paternal allele are differentially methylated. In many cases DNA methylation is present on the silenced allele, especially when it is the promoter region which is silenced. However, DNA methylation may be present on the active allele when the mechanism of gene silencing is through a more indirect mechanism.

GNAS/Gnas generates multiple gene products through the use of four alternative promoters and first exons that splice onto a common exon (exon 2, Fig. 1). The most downstream alternative exon is G_sα exon 1, which generates transcripts encoding G_sα (Kozasa et al., 1988). The G_sα coding region spans 13 exons (12 exons in mice) and generates two long and two short forms of G_sα protein by alternative splicing of exon 3 (Bray et al., 1986; Kozasa et al., 1988). Use of an alternative terminal exon (N1) located between G_sα exons 3 and 4 generates a neural-specific splice variant of unknown function (Crawford et al., 1993). Typical of most ubiquitously expressed genes, the G_sα promoter resides within a CpG island (GC-rich region with a high frequency of CpG dinucleotides) which is not methylated (Hayward et al., 1998a; Kozasa et al., 1988; Liu et al., 2000b; Peters et al., 1999). Despite the lack of promoter methylation, G_sα is imprinted in a tissue-specific manner in both mice and humans, being expressed primarily from the maternal allele in several tissues, including pituitary, thyroid, renal proximal tubules, and gonads, but biallelically expressed in most tissues (Campbell et al., 1994; Davies & Hughes, 1993; Germain-Lee et al., 2002; Hayward et al., 2001; Liu et al., 2003; Mantovani et al., 2002; Weinstein et al., 2001; S. Yu et al., 1998).

Fig 1.

The Gnas locus and its gene products. The organization of the Gnas gene locus is shown with exons denoted as black rectangles and promoter regions Nesp, Nespas, Gnasxl, 1A, and G_sα indicated above. _Gsα exons 1, 2, 3, 4–12 (as a single box) and alternative terminal exon N1 are labeled. Only Nespas exon 1 is shown. Methylated regions (Meth), transcriptionally active promoters (horizontal arrows), and resulting mRNA transcripts from the maternal and paternal alleles are shown above and below the gene locus, respectively. Coding regions within transcripts are indicated in black. The stippled horizontal arrow at the paternal Gsα promoter indicates that the promoter is silenced on the paternal allele in a tissue-specific manner.

The most upstream alternative promoter (~45 kB upstream of G_sα exon 1 and referred to as Nesp in the mouse; Fig. 1) generates transcripts that encode the neuroendocrine-specific protein of 55 kDa (NESP55) (Hayward et al., 1998b; Kelsey et al., 1999; Peters et al., 1999). NESP55 is a chromogranin-like protein which is totally unrelated to G_sα. The entire NESP55 coding sequence is within its specific upstream exon and G_sα exons 2–13 are within the 3′ untranslated region of NESP55-specific transcripts (Hayward et al., 1998b; Ischia et al., 1997). NESP55 is only transcribed from the maternal allele and its promoter region is methylated only on the paternal allele (Hayward et al., 1998b; T. Li et al., 2000; Peters et al., 1999). Imprinting of NESP55 in mice is not established until after implantation, suggesting that this region is not a primary imprinting control region for the GNAS/Gnas locus (Liu et al., 2000b).

A third alternative promoter (~30 kb upstream of G_sα exon 1 and referred to as Gnasxl in the mouse, Fig. 1) generates transcripts encoding the G_sα isoform XLαs (Hayward et al., 1998a; Kehlenbach et al., 1994; Kelsey et al., 1999; Peters et al., 1999). XLαs has a long amino-terminal extension encoded by its specific first exon (422, 367, and 551 amino acids in mice, rats, and humans, respectively) while the remainder of the protein is identical to G_sα, being encoded by G_sα exons 2–13. High levels of a neural-specific truncated form of XLαs presumed to be nonfunctional (XLN1) are generated by alternative splicing of XLαs to the alternative terminal exon N1 (Pasolli et al., 2000). The XLαs-specific first exon also has a second open reading frame for the unrelated protein Alex, which has been reported to directly interact with the XLαs protein (Freson et al., 2003; Klemke et al., 2001). Gnasxl is oppositely imprinted to Nesp, being methylated on the maternal allele and transcriptionally active only on the paternal allele (Hayward et al., 1998a; T. Li et al., 2000; Peters et al., 1999).

Within the same differentially methylated region and just upstream of the XLαs promoter is a promoter for paternally expressed antisense transcripts that traverse the NESP55 exon from the opposite direction (Nespas) (Hayward & Bonthron, 2000; T. Li et al., 2000; Williamson et al., 2002; Wroe et al., 2000). Like Gnasxl, Nespas is methylated on the maternal allele. Within the Nespas-Gnasxl promoter region is a primary DNA methylation imprint mark which is established during oogenesis and maintained throughout development (Coombes et al., 2003). Although evidence described below suggests that Nespas is the imprint control center for Nesp, Nespas and Nesp transcripts do not always colocalize during development (Ball et al., 2001).

A fifth alternative promoter and first exon (exon 1A, also referred to as exon A/B in humans) located ~2.5 kb upstream of G_sα exon 1 (Fig. 1) is methylated on the maternal allele and generates paternal-specific transcripts that are presumably untranslated (Ishikawa et al., 1990; Liu et al., 2000a; Liu et al., 2000b). The exon 1A region is also a primary imprinting control center as its maternal-specific methylation is established in oocytes and maintained throughout development in mice (Liu et al., 2000b). In PHP1b patients who have renal PTH resistance presumably due to G_sα deficiency in renal proximal tubules, maternal-specific exon 1A methylation is absent (Bastepe et al., 2001; Jan de Beur et al., 2003; Liu et al., 2000a; Liu et al., 2005b). This suggests that exon 1A controls tissue-specific G_sα imprinting via the presence of one or more negatively regulatory cis-acting elements that are both tissue-specific and methylation-sensitive (therefore only leading to suppression of G_sα expression on the unmethylated paternal allele) (Fig. 3). Promoter competition between the exon 1A and G_sα exon 1 promoters is unlikely to be the mechanism for tissue-specific G_sα imprinting as exon 1A-specific transcripts are expressed in virtually all tissues with a similar tissue distribution pattern as G_sα (Liu et al., 2000b). Deletions identified in autosomal dominant forms of PHP1b suggest that cis-acting elements within the upstream NESP55 imprinted region and the nearby STX16 gene may be critical for the establishment and/or maintenance of exon 1A maternal-specific methylation (Bastepe et al., 2003; Bastepe et al., 2005; Linglart et al., 2005; Liu et al., 2005b).

Fig. 3.

Role of the 1A differentially methylated region in tissue-specific G_sα imprinting and the mechanism of abnormal G_sα expression in PHP1b patients and mice with 1A deletion. The model predicts that the 1A region has a cis-acting silencer element which binds a tissue-specific repressor (R). In renal proximal tubules (upper panels) the repressor is present and binds to the unmethylated silencer on the paternal allele, leading to silencing of the paternal G_sα promoter. The repressor is unable to bind to the maternal allele due to methylation of the silencer, allowing the maternal G_sα promoter to remain active. In most other tissues (lower panels) the repressor is not expressed and therefore the paternal G_sα promoter is unaffected and G_sα is biallelically expressed. Loss of maternal 1A methylation in PHP1b allows the repressor to bind to both alleles, leading to biallelic G_sα silencing, G_sα deficiency, and PTH resistance in proximal tubules (down arrow). In most other tissues the repressor is not expressed and therefore G_sα expression is unaffected (N). Maternal 1A deletion in mice (1A^m−/+) has no effect on G_sα expression. In contrast, paternal 1A deletion (1A^+/p−) results in loss of G_sα imprinting in proximal tubules due to loss of the silencer, leading to G_sα overexpression (up arrow) and increased PTH sensitivity, but has no effect on G_sα expression in other tissues. Although shown here as a silencer, the negative cis-acting element involved in tissue-specific G_sα imprinting may be an insulator. Loss of tissue-specific G_sα imprinting explains why paternal 1A deletion rescues the phenotype produced from disruption of the maternal G_sα allele. Reproduced from Liu et al., 2005a.

3. Protein products encoded by Gnas

3.1. G_sα

G_sα couples seven transmembrane receptors for a wide variety of extracellular factors, including glycoprotein and peptide hormones, catecholamines and other biogenic amines, and neurotransmitters, to adenylyl cyclase and is required for receptor-stimulated intracellular cAMP generation (Weinstein et al., 2004; Weinstein et al., 2001). G_sα is ubiquitously expressed and in almost all tissues is the G protein mediating receptor-stimulated cAMP responses. The related G protein G_olfα, which is encoded by an independent gene, also stimulates adenylyl cyclase but has a much more restricted tissue distribution (Jones & Reed, 1989).

G_sα is targeted to the inner leaflet of both plasma and intracellular membranes by palmitoylation at its amino terminus (Degtyarev et al., 1993; Linder et al., 1993; Mumby et al., 1994). It has been suggested that in addition to its role in signal transduction, it may also play a role in membrane trafficking (Bomsel & Mostov, 1992; Zheng et al., 2001). In the basal state G_sα has GDP bound to its guanine nucleotide binding pocket and combines with βγ dimers to form a heterotrimer. Activated receptors promote GDP release, allowing ambient GTP to bind to G_sα. Upon GTP binding G_sα switches to an active conformation, leading to depalmitoylation and release from βγ and the plasma membrane and direct activation of adenylyl cyclase and other effectors (Allen et al., 2005; Huang et al., 1999; Wedegaertner et al., 1996; J. Z. Yu & Rasenick, 2002). G_sα is deactivated by an intrinsic GTPase activity that hydrolyzes bound GTP to GDP. Evidence suggests that the GTPase cycle of G_sα may be influenced by an RGS protein (Zheng et al., 2001) and by adenylyl cyclase itself (Scholich et al., 1999).

Like all Gα subunits, G_sα has two structural domains, a Ras-like GTPase domain, which includes the sites for guanine nucleotide binding and effector interaction, and a helical domain which forms a cleft in which guanine nucleotides bind (Sunahara et al., 1997; Tesmer et al., 1997). Interactions between the two domains may be important to prevent constitutive guanine nucleotide release (Mixon et al., 1995; D. R. Warner et al., 1998). Alternative splicing of exon 3 leads to two long and two short forms of G_sα with helical domains of slightly differing length (Bray et al., 1986; Kozasa et al., 1988) which are all biologically active with only subtle biochemical differences between them (Jones et al., 1990; Seifert et al., 1998).

GTP binding alters the conformation of three regions (switch 1, 2 and 3 regions) within the GTPase domain (Lambright et al., 1994; Noel et al., 1993). Interactions between the switch 2 and 3 regions, as well as between switch 3 region and the helical domain, are important to maintain the active conformation (Grishina & Berlot, 1998; Iiri et al., 1997; Q. B. Li & Cerione, 1997; Marsh et al., 1998; D. R. Warner et al., 1999; D.R. Warner & Weinstein, 1999; D. R. Warner et al., 1998) Two highly conserved residues (Arg²⁰¹ and Gln²²⁷ in the long form of G_sα) are critical for catalyzing the GTPase reaction (Coleman et al., 1994; Graziano & Gilman, 1989; Landis et al., 1989; Sondek et al., 1994). Post-translational modification (ADP-ribosylation) of Arg²⁰¹ by cholera toxin or missense mutations encoding substitutions of Arg²⁰¹ or Gln²²⁷ lead to constitutive activation by disrupting the GTPase ‘turn-off’ mechanism. Such somatic mutations have been identified in sporadic endocrine tumors (pituitary, thyroid, adrenal), fibrous dysplasia of bone, and the McCune-Albright syndrome (Lyons et al., 1990; Weinstein, 2002, 2004). The amino terminus and switch 2 regions interact directly with β-subunit to form the heterotrimer (Lambright et al., 1996; Wall et al., 1995), while the carboxyl terminus is important for receptor interactions (Conklin et al., 1996; Grishina & Berlot, 2000; Mazzoni et al., 2000; Schwindinger et al., 1994; Simonds et al., 1989; Sullivan et al., 1987).

cAMP, the major second messenger resulting from G_sα signaling, mediates its effects through direct binding and stimulation of several molecules, including protein kinase A (PKA), cAMP-regulated guanine nucleotide exchange factors (cAMP-GEFs) (de Rooij et al., 1998; Kawasaki et al., 1998), and ion channels (Sudlow et al., 1993; Wainger et al., 2001). PKA is a serine/threonine kinase which activates physiological changes and gene expression through direct phosphorylation of enzymes involved in intermediary metabolism and transcription factors such as the cAMP response element binding protein (CREB) (Montminy, 1997). cAMP-GEFs are expressed at high levels in neuroendocrine cells and may lead to mitogenesis through activation of the ras family member Rap1 (Cass & Meinkoth, 1998; Mei et al., 2002; Miller et al., 1997; Vossler et al., 1997; Wang et al., 2006). There is evidence that G_sα may also directly activate effectors other than adenylyl cyclase, including cardiac Ca²⁺ channels (Mattera et al., 1989; Yatani et al., 1988) and Src tyrosine kinases (Ma et al., 2000) and may be activated by various tyrosine kinase receptors, including the epidermal growth factor (EGF) (Nair et al., 1990; Nair & Patel, 1993; Poppleton et al., 1996; H. Sun et al., 1997) and basic fibroblast growth factor receptors (Krieger-Brauer et al., 2000)

3.2. XLαs

XLαs is a G_sα isoform only produced from the Gnas paternal allele which differs from G_sα in that the first 47 amino acids of G_sα are replaced by a large amino terminal domain with multiple repeat regions (the XL domain) that is encoded by the XLαs-specific first exon (Abramowitz et al., 2004; Kehlenbach et al., 1994; Pasolli et al., 2000). The XL domain targets XLαs to the plasma membrane (Pasolli et al., 2000). Some, but not all, evidence suggests XLαs may also be localized to the Golgi (Pasolli et al., 2000; Ugur & Jones, 2000). XLαs can bind to βγ and can activate adenylyl cyclase (Klemke et al., 2000). Although one study in native cells did not show evidence for receptor signaling by XLαs (Klemke et al., 2000), a subsequent study showed that XLαs transfected into G_sα-null cells mediated stimulation of adenylyl cyclase by various receptors with the same efficacy and potency as G_sα (Bastepe et al., 2002). However whether XLαs has an important physiological function in humans is unclear, as PPHP patients with paternal GNAS mutations predicted to disrupt XLαs expression do not have a specific phenotype.

XLαs has a more restricted tissue distribution than G_sα, being primarily expressed in neuroendocrine tissues (Kehlenbach et al., 1994; Pasolli & Huttner, 2001; Pasolli et al., 2000). In rats, XLαs is present in the melanotrophs within the pars intermedia of the posterior lobe, and to a lesser extent in brain, adrenal, heart, and pancreatic islets. In mouse neonates XLαs was shown to be expressed in brown and white adipose tissue (BAT, WAT), adrenal medulla, intermediate lobe of the pituitary, brain, stomach, heart, pancreas, and kidney (Plagge et al., 2004). In situ hybridization performed in these mice showed XLαs in the brain to be localized to the hypothalamus, locus coeruleus, as well as the laterodorsal tegmental, hypoglossal, trigeminal, ambiguous, medullary reticular, raphe obscurus and facial nuclei. XLαs may play a specific role during early development, as XLαs expression in adipose tissue was shown to be lost during the first two weeks of mouse postnatal development (Xie et al., 2006). Low levels of XLαs expression have also been reported in endochondral growth plates of developing bone (Bastepe et al., 2004).

The XLαs-specific first exon also generates small amounts of an unrelated basic protein named Alex by translation from an alternative reading frame, and this protein may bind to XLαs (Abramowitz et al., 2004; Freson et al., 2003; Klemke et al., 2001). One study suggested that polymorphisms within the XLαs-specific exon may generate alternative forms of Alex and XLαs which fail to interact (Freson et al., 2003). In the central nervous system the XLαs promoter also produces large amounts of a truncated form of XLαs (XLN1) by splicing to the alternative terminal exon N1 (Pasolli et al., 2000). Although XLN1 is missing most of the elements that are critical for signaling, it may potentially act as a dominant negative regulator of G_sα signaling (Plagge et al., 2004; Xie et al., 2006)

3.3. NESP55

Rather than being a Gα subunit, NESP55 is a chromogranin-like protein located within large dense core granules of secretory cells and is presumed to be a cosecretagogue (Ischia et al., 1997). In rodents NESP55 transcripts begin to be expressed in somites and vasculature during midgestation (Ball et al., 2001), and in adults is primarily expressed in neuroendocrine tissues, including the adrenal medulla, pituitary, hypothalamus, and other midbrain and brainstem regions (Bauer et al., 1999a; Bauer et al., 1999b; Ischia et al., 1997; Lovisetti-Scamiform et al., 1999). It has been suggested that a proteolytic cleavage product of NESP55 may be an endogenous antagonist of serotonin 1b receptors (Ischia et al., 1997). However loss of NESP55 expression in humans is not associated with an obvious phenotype (Kim et al., 2000; Liu et al., 2000a), so the normal role for NESP55, if any, is presently unknown.

4. Overview of Gnas knockout mouse models

4.1. Mice with disruption of Gnas exon 2 (E2⁻)

The initial Gnas knockout model was generated by insertion of a neomycin-resistance cassette into Gnas exon 2 (S. Yu et al., 1998) (germline Gnas knockout models are summarized in Table 1). Similar to all other subsequent germline G_sα knockout lines (Cattanach et al., 2000; M. Chen et al., 2005a; Germain-Lee et al., 2005; Skinner et al., 2002), homozygous G_sα mutations were embryonically lethal with failure to develop beyond gestational day 10.5, as might be expected for total deficiency of a key signaling molecule that is ubiquitously expressed. In fact, heterozygotes were also lethal in several inbred genetic backgrounds and in the CD-1 outbred strain had only 20% survival to weaning even though they were born at the expected Mendelian ratios (S. Yu et al., 1998).

Table 1.

Germline Gnas knockout and related mouse models

Model	Gene Defect	Transcripts Disrupted	Phenotype	References
E2m−/+	Gsα exon 2 insertion (maternal)	-Gsα (maternal); -? other maternal Gnas transcripts	- Perinatal subcutaneous edema - Wide body habitus - Neurological signs - Early lethality (1–3 weeks) - Slightly reduced length - ↑ BAT mass - Obesity - ↓ Energy expenditure - ↓ Sympathetic activity - ↓ Activity levels - ↑ glucose tolerance - ↑ Insulin sensitivity - PTH resistance - Urine concentration defect in response to acute vasopressin administration	Ecelbarger et al., 1999 S. Yu et al., 2001 S. Yu et al., 2000 S. Yu et al., 1998
PatDp distal 2	Paternal duplication/maternal deletion of distal chrom 2	-Gsα (maternal) -NESP55 -? other maternal Gnas transcripts -possible increase of XLαs, XLN1 expression	- Perinatal subcutaneous edema - Slight increase in length - Wide body habitus - No survival beyond days - Hyperactivity - Neonatal microcardia	Cattanach & Kirk, 1985 Williamson et al., 1998
E1m−/+	Gsα exon 1 deletion (maternal)	-Gsα (maternal)	- Early lethality (less than E2m−/+) - Slightly reduced length - Severe obesity - Severe glucose intolerance - Severe insulin resistance - Hypertriglyceridemia - PTH resistance	M. Chen et al., 2005a Germain-Lee et al., 2005 Liu et al., 2005a
Oed	Gsα exon 6 Val159Glu point (maternal) mutation	-Gsα (maternal) -? other maternal Gnas transcripts	- Perinatal subcutaneous edema - Pre- and postnatal lethality - Slightly reduced length - Neonatal microcardia - ↑ BAT mass - Initial growth retardation - Weight gain towards weaning	Cattanach et al., 2000 Skinner et al., 2002
E2+/p−	Gsα exon 2 insertion (paternal)	-Gsα (paternal); -XLαs, XLN1	- Perinatal lethality - Slightly reduced length - Narrow body habitus - Poor suckling at birth - Perinatal hypoglycemia - Small body size - ↓ BAT mass - Reduced adiposity - ↑ Energy expenditure - ↑ Sympathetic activity - ↑ Activity levels - ↑ Glucose tolerance - Markedly ↑ Insulin sensitivity	M. Chen et al., 2004 S. Yu et al., 2001 S. Yu et al., 2000 S. Yu et al., 1998
MatDp distal 2	Maternal duplication/paternal deletion of distal chrom 2	-XLαs, XLN1 -possible increase in NESP55 expression and Gsα expression in some tissues	- Perinatal lethality (no survival) - Narrow body habitus - Poor suckling at birth - Hypoactivity - ↑ BAT mass	Cattanach & Kirk, 1985 Williamson et al., 1998
E1+/p−	Gsα exon 1 deletion (paternal)	-Gsα (paternal)	- No perinatal lethality - Slightly reduced length - Mild obesity - Mild insulin resistance - Hypertriglyceridemia	M. Chen et al., 2005a Germain-Lee et al., 2005
Sml	Gsα exon 6 Val159Glu point mutation (maternal)	-Gsα (paternal) -XLαs	- Early postnatal lethality - Hypoglycemia - ↓ BAT mass - Growth retardation - Lean adult phenotype	Cattanach et al., 2000 Skinner et al., 2002
Gnasxl+/p−	XLαs exon 1 deletion	-XLαs, XLN1	- Perinatal lethality - Narrow body habitus - Poor suckling at birth - Perinatal hypoglycemia - Small body size - ↓ BAT mass - Reduced adiposity - ↑ Energy expenditure - ↑ Sympathetic activity - ↑ Activity levels - ↑ Glucose tolerance - Markedly ↑ Insulin sensitivity	Plagge et al., 2004 Xie et al., 2006
NESP55 KO	Nesp deletion	-NESP55	- No metabolic phenotype - ↑ Reaction to new environment	Plagge et al., 2005
1A+/p−Δ1A-DMR	Deletion of 1A region (paternal)	-1A	- Loss of tissue-specific Gsα imprinting - Increased PTH sensitivity - Rescue of Oed and E1m−/+ phenotypes	Williamson et al., 2004 Liu et al., 2005a
ΔNAS-DMR (paternal)	Deletion within Nespas promoter	-NESP antisense	- Loss of Nesp imprinting - ↓ XLαs expression - Partial paternal 1A methylation - Partial loss of Gsα imprinting - Similar phenotype to Gnasxl+/p− - Partial rescue Oed phenotype - Maternal ΔNAS-DMR normal	Williamson et al., 2006

Most strikingly, heterozygotes in which the mutant allele was inherited from the mother (E2^m−/+) had a very different phenotype than those in which the mutant allele was inherited from their father (E2^+/p−) (S. Yu et al., 1998). E2^m−/+ neonates had wide, square shaped bodies, increased birth weights, and marked subcutaneous edema which resolved within 1–2 days. In addition, they had increased convex skin folds in the dorsal neck due to increased interscapular brown adipose tissue (BAT) and inset ears. Most of the mice developed neurological abnormalities, including ataxia, tremor, and imbalance, associated with delayed maturation of the cerebellar cortex with persistence of superficial granular cells, which eventually led to respiratory difficulty and premature death. Other features included thymic cortical atrophy, delayed kidney maturation, and delayed growth. With respect to the thymic cortical atrophy, thymic cortical hyperplasia has been reported in McCune-Albright syndrome due to constitutively activating G_sα mutation (Danon & Crawford, 1987; Shenker et al., 1993). E2^m−/+ mice that survived to adulthood had low levels of energy expenditure and developed obesity (S. Yu et al., 2000).

In contrast to E2^m−/+ neonates, E2^+/p− neonates did not have subcutaneous edema but had low birth weights, narrow bodies, arched backs, and concavity of the dorsal neck due to reduced BAT mass. Most failed to suckle, developed hypoglycemia, and died within hours after birth. In contrast to E2^m−/+ mice, E2^+/p− mice had markedly reduced adiposity (S. Yu et al., 2000). Both E2^m−/+ and E2^+/p− had slight reductions in body length.

E2⁻ mice have been used to develop G_sα-null cells by immortalization of cells derived from 9.5 day-old E2^m−/p− embryos (Bastepe et al., 2002). Transfection studies with these cells have proven useful in defining the role of G_sα in signaling and its regulation. Using these cells, XLαs was shown to have identical receptor-adenylyl cyclase signaling properties to G_sα (Bastepe et al., 2002). These cells were also used to determine the carboxy-terminal residues required for coupling to prostacyclin receptors (Zhang et al., 2006), to establish that G_sα is required for β2-adrenergic receptor stimulation of MAP kinase at low but not at high agonist concentrations (Y. Sun et al., 2007), and to define the role of G_sα in sensitization of adenylyl cyclase isoforms resulting from chronic D2 dopamine receptor stimulation (Vortherms et al., 2006). Embryonic stem cells have been generated from E2^m−/p− blastocysts and used to generate E2^m−/p−:wild-type chimeras to show that G_sα signaling is required for normal regulation of chondrocyte differentiation within embryonic growth plates (Bastepe et al., 2004).

4.2. PatDp and MatDp distal 2 mice

The different phenotypes observed in the maternal and paternal heterozygotes provided strong evidence for imprinting of the Gnas locus and suggested that the E2^m−/+ and E2^+/p− phenotypes result from loss of maternal and paternal-specific Gnas transcripts, respectively. Support from this model derives from the fact that mice with paternal duplication/maternal deletion of the distal chromosome 2 region which includes Gnas (PatDp distal 2) have features in common with E2^m−/+ mice, while mice with maternal duplication/paternal deletion of the same region (MatDp distal 2) develop features in common with E2^+/p− mice (Cattanach & Kirk, 1985; Peters et al., 1994; S. Yu et al., 1998). Due to the fact that PatDp and MatDp distal 2 mice have only paternally and maternally derived alleles of Gnas, respectively, they have been used to establish the imprinting of Nesp, Nespas, and Gnasxl in mice (Kelsey et al., 1999; Peters et al., 1999; Wroe et al., 2000).

Like E2^m−/+ mice, PatDp distal 2 mice were reported to have subcutaneous edema at birth, with a similar body habitus and hyperactivity, but no survival beyond several days (Cattanach & Kirk, 1985). Detailed studies showed that the edema began to develop on gestational day e11.5 (Williamson et al., 1998). As this is five days before the embryonic kidneys begin to function, the edema is not likely to be the result of abnormal renal excretion but rather may be due to a defect in placental fluid transport as the placentas of PatDp distal 2 embryos had reduced wet weight. PatDp embryos also have slightly increased bone length compared to normals (Williamson et al., 1998). Conversely, MatDp distal 2 mice had similarities to E2^+/p− mice, including similar body habitus, failure to suckle, hypoactivity, and death within hours after birth (Cattanach & Kirk, 1985). Neither PatDp or MatDp distal 2 mice survived beyond several days. The differences in survival between E2⁻ and PatDp/MatDp distal 2 mice is likely due to differences in the genetic backgrounds in which the mice were studied. It is possible that effects in the latter mice are due to overexpression of Gnas products (eg. XLαs or NESP55 overexpression in PatDp or MatDp distal 2 mice, respectively, due to the presence of two paternal or maternal Gnas alleles) or to potential effects on other unknown genes within the duplicated chromosomal region which may also be imprinted.

4.3. Mice with a point mutation in Gnas exon 6 (Oed-Sml)

Another Gnas mutation, a point mutation in G_sα exon 6 leading to a single amino acid substitution (Val¹⁵⁹ to Glu) was generated in mice by chemical mutagenesis using ethylnitrosourea (Cattanach et al., 2000; Skinner et al., 2002). Similar to E2⁻ mice, homozygotes were embryonically lethal. This mutation has been referred to as Oed-Sml because maternal transmission of the mutation leads to subcutaneous edema while paternal transmission leads to a small phenotype. Oed mice are similar to E2^m−/+ and PatDp distal 2 mice in that they have subcutaneous edema at birth, increased BAT mass, early death due to respiratory abnormalities, and initial growth retardation followed by significant weight gain. Like E2^m−/+ mice, Oed mice were shorter than normal. Unlike E2^m−/+ mice, Oed mice had reduced prenatal survival. Oed mice were also reported to have small hearts (microcardia) in the neonatal period which normalized by adulthood, and followup studies of PatDp distal 2 mice also showed evidence for neonatal microcardia (Cattanach et al., 2000). Oed mice did not display the hyperkinetic behavior observed in PatDp distal 2 mice. Sml mice had features similar to E2^+/p− and MatDp distal 2 mice, including substantial early lethality, decreased BAT mass, hypoglycemia, growth retardation, and a lean adult phenotype (Cattanach et al., 2000; Plagge et al., 2004; Skinner et al., 2002). However, the abnormal body habitus, hypokinetic behavior and abnormal suckling observed in E2^+/p− and MatDp distal 2 mice was not observed in Sml mice. Differences between Oed-Sml and the other mouse models may be due to differences in genetic background or differential effects on Gnas gene products as mentioned above.

4.4. Transcript-specific and tissue-specific Gnas knockouts

Based upon the findings in E2⁻, Oed-Sml, and PatDp and MatDp distal 2 mice, it can be concluded that loss of one or more maternal Gnas gene products leads to subcutaneous edema during gestation, increased BAT mass, obesity, and increased lethality while loss of paternal Gnas gene products leads to failure to suckle, perinatal lethality, growth retardation, and a lean phenotype in adulthood. However, because these mutations occur in exons that are not transcript-specific, it is difficult to assign specific phenotypic features to the loss of specific Gnas transcripts. To address this issue, subsequent knockout models which specifically disrupt the expression of Gnas products G_sα, XLαs, or NESP55 through disruption or deletion of their specific first exons have been generated, and are discussed in detail below (transcript-specific phenotypes summarized in Tables 1 and 2). Specific deletions of the regions presumed to be important for Gnas imprinting have also been generated.

Table 2.

Phenotype consequences resulting from loss of specific Gnas transcripts in mice

Gsα (homozygous)	Embryonic lethality
Gsα (maternal)	Perinatal subcutaneous edema
	Early postnatal lethality in a significant proportion
	Mildly reduced body length
	Severe obesity
	Severe glucose intolerance, insulin resistance
	Hypertriglyceridemia
	PTH resistance
	TSH resistance (mild, variable depending on genetic background
Gsα (paternal)	Mildly increased adiposity, normal weight
	Mild glucose intolerance, insulin resistance
	No sucking defect or perinatal lethality
XLαs	Perinatal lethality associated with sucking defect and hypoglycemia
	Severely reduced adiposity
	Markedly increased glucose tolerance and insulin sensitivity
	Increased energy expenditure and sympathetic activity levels
	Increased lipid metabolism in adipose tissue
	No increase in sensitivity of adipose tissue to sympathetic stimulation
NESP55	Increased reactivity to novel environment
	No obvious effects on development, survival, or metabolism

It is difficult to decipher the effects of G_sα deficiency in specific tissues using germline G_sα knockout models as homozygotes are embryonically lethal and G_sα is a ubiquitously expressed protein which is subject to imprinting. In order to examine the role of G_sα signaling pathways in specific tissues, floxed G_sα knockout mice have been generated in which loxP recombination sites have been placed on either side of G_sα exon 1 (E1^fl). The presence of the loxP sites themselves in E1^fl/fl mice has no effect on G_sα expression or phenotype (M. Chen et al., 2005b). Homozygous or heterozygous tissue-specific G_sα knockouts are generated by mating E1^fl/fl or E1^fl/+ mice to tissue-specific Cre recombinase transgenic lines in which Cre transgenes are driven by gene promoters with tissue-specific expression (tissue-specific G_sα knockout models are summarized in Table 3).

Table 3.

Tissue-specific G_sα knockout models

Cre-Promoter	Tissue	Phenotype	References
Collagen Ia1	Osteoblasts, osteocytes	-Hypoplasia of anterior craniofacial bones -Shortening of long bones - ↓ Primary spongiosa, trabecular bone formation - Thickened cortical bone due to reduced osteoclast activity - ↓ RANK ligand expression - ↓ Expression of late osteoblast markers	Sakamoto et al., 2005b
	?Extraskeletal tissues	Perinatal subcutaneous edema
Collagen 2a1	Chondrocytes	- Perinatal lethality with severe skeletal defects - Shortened limbs, domed skull - Shortened growth plates - Premature chondrocyte differentiation - ↑ PTHrP expression in periarticular zone -Ectopic chondrocytes in anterior aspect of proximal metaphysic of tibia	Sakamoto et al., 2005a
Albumin	Liver	- Glycogen synthesis in liver - ↑ Insulin sensitivity in liver, muscle, WAT - ↓ Expression of gluconeogenic enzymes - Hypoglycemia, hypoinsulinemia in fed state - Normal fasting glucose and insulin probably due to ↑ glycogen stores and extrahepatic gluconeogenesis - ↑ Fasting triglyceride levels due to persistent lipogenesis - ↓ Adiposity - ↑ ↑ ↑ Glucagon, GLP1 levels - Pancreatic α-cell hyperplasia - ↑ glucose-stimulated insulin secretion	M. Chen et al., 2005b
Renin	Juxtaglomerular (JGA) cells	- ↓ Renin production in JGA cells - ↓ Plasma renin, aldosterone levels - ↓ Hypotension - ↓ Glomerular filtration rate	L. Chen et al., 2007
	Renal collecting ducts	- Vasopressin resistance - Urine concentrating defect

5. Use of Gnas gene targeting in miceto study Gnas imprinting

Because PatDp and MatDp distal 2 mice carry only paternal and maternally derived Gnas alleles, respectively, they were used to first establish the allele-specific patterns of transcription activity and methylation of the Nesp, Gnasxl, and Nespas, promoters, which are each methylated on one allele and transcriptionally active on the opposite allele as described above (Kelsey et al., 1999; Peters et al., 1999; Wroe et al., 2000). By Southern analysis using restriction enzymes that were sensitive to DNA methylation it was established that the exon 1A region is methylated exclusively on the maternal allele (Liu et al., 2000a; Liu et al., 2000b). Examination of histone modifications in mice with paternal or maternal exon 1A deletions showed this region to be heterochromatic on the maternal allele and euchromatic on the paternal allele (Sakamoto et al., 2004). Exon 1A-specific transcripts were detected in E2^m−/+, but not in E2^+/p− mice, consistent with these transcripts being only transcribed from the unmethylated promoter on the paternal allele (Liu et al., 2000b), and similar paternal-specific expression of these transcripts was confirmed in humans (Liu et al., 2000a). Other studies have confirmed these imprinting patterns of the Gnas upstream promoter regions (T. Li et al., 2000).

Unlike the Nesp, Gnasxl, Nespas, and exon 1A promoters, in which one parental allele is fully silenced in all tissues due to promoter methylation, the G_sα promoter is unmethylated on both parental alleles and G_sα expression from the paternal allele is suppressed in a tissue-specific manner. Moreover, the imprinting may only be partial, with paternal-specific expression being present, but at a lower level than expression from the maternal allele. G_sα mRNA and protein expression was significantly reduced in proximal tubules of E2^m−/+ while G_sα expression was minimally affected in E2^+/p− mice, consistent with G_sα being expressed primarily from the maternal allele in renal proximal tubules (S. Yu et al., 1998) (Fig. 2). Consistent with this finding, E2^m−/+, but not E2^+/p− mice, have PTH resistance in proximal tubules (S. Yu et al., 1998). These initial findings were confirmed in a G_sα-specific knockout model in which G_sα exon 1 was deleted (Germain-Lee et al., 2005; Schwindinger et al., 1997) and PTH resistance was also observed in Oed mice (Williamson et al., 2004).

The G_sα knockout model generated by Germain-Lee and her colleagues also had G_sα imprinting in the thyroid with associated TSH resistance and elevated serum TSH levels in the maternal heterozygote (Germain-Lee et al., 2005), while TSH levels were normal in E2^m−/+ mice (S. Yu et al., 2000) and another G_sα-specific knockout model (M. Chen et al., 2005a). These differences are probably due to the fact that the model with TSH resistance was in the SvEv genetic background while the latter two were in a CD-1 background, indicating strain-specific differences in thyroid-specific G_sα imprinting or other physiological variables which regulate TSH sensitivity. G_sα expression studies in E2⁻ mice also showed evidence for G_sα imprinting within the brain (S. Yu et al., 1998). Overall these findings are consistent with what is observed in AHO patients, in whom maternal, but not paternal inheritance of G_sα mutations lead to renal PTH resistance and partial TSH resistance (Weinstein, 2004; Weinstein et al., 2001). As PTH resistance in these patients develops during the first one or two years of life and renal tubules continue to mature during this period, it is possible that G_sα imprinting is temporally regulated and is only present in mature proximal tubule cells (Weinstein, 2001).

Initial studies showing low G_sα expression in BAT and WAT of adult E2^m−/+ but not E2^+/p− suggested that G_sα is also imprinted in BAT and WAT (S. Yu et al., 2000; S. Yu et al., 1998). These findings have been refuted by studies in adult G_sα-specific knockout models in which G_sα levels were reduced to a similar extent in WAT from maternal and paternal heterozygotes (Germain-Lee et al., 2005) and with only a small difference in G_sα BAT expression between maternal and paternal heterozygotes (M. Chen et al., 2005a). One study also showed no evidence for G_sα imprinting in human WAT (Mantovani et al., 2004). We suspect that the differential G_sα expression observed in E2^m−/+ and E2^+/p− mice may result from secondary changes in G_sα expression resulting from different levels of metabolic activity in adipose tissues of these mice (S. Yu et al., 2000) rather than imprinting. G_sα expression studies in neonatal BAT of various Gnas knockout models have also suggested that G_sα is imprinted in this tissue (Plagge et al., 2004; Williamson et al., 2004; Williamson et al., 2006), but these studies must also be interpreted with caution as BAT undergoes rapid cellular changes in this period which are affected by Gnas perturbations. One study examining allele-specific expression using a polymorphism within the G_sα transcript showed strong evidence for imprinting in neonatal BAT (Williamson et al., 2006). It is likely that G_sα is imprinted in neonatal BAT, but that this imprinting is lost by the time animals reach adulthood. It is clear that temporal changes in Gnas regulation occur in adipose tissue, as XLαs is expressed in neonatal BAT and WAT but this expression is lost within 1–2 weeks (Xie et al., 2006).

Studies in E2⁻ and G_sα-specific knockout models showed that in many other tissues (liver, heart, renal inner and outer medulla, muscle, lung) disruption of the maternal or paternal G_sα allele resulted in similar (~50%) decreases in Gsα expression levels, indicating that G_sα is not imprinted in these tissues (M. Chen et al., 2005a; Ecelbarger et al., 1999; Germain-Lee et al., 2005; S. Yu et al., 2000; S. Yu et al., 1998) (Fig. 2). Although one study using PatDp and MatDp distal 2 mice suggested that G_sα was imprinted in an opposite manner in renal glomeruli with silencing of the maternal allele (Williamson et al., 1996), there was no evidence for G_sα imprinting in E2⁻ mice (S. Yu et al., 1998). Tissue-specific G_sα imprinting likely explains why PHP1a patients with maternal G_sα mutations develop resistance to some hormones (eg. PTH, TSH) but not to other hormones that also signal through G_sα (eg. adrenocorticotropic hormone, vasopressin). It is likely that in tissues where G_sα is not imprinted, G_sα haploinsufficiency is still capable of mediating hormone action without an obvious physiological effect (Weinstein et al., 2000). G_sα imprinting within the renal proximal tubule, but not within the more distal tubule segments likely explains why PHP1a patients show evidence for PTH resistance in the proximal tubule but maintain responsiveness to PTH in the distal nephron (Stone et al., 1993). Analysis of histone methylation in mice with paternal or maternal deletion of the G_sα promoter showed that tissue-specific differences in allele-specific expression were associated with corresponding tissue-specific differences in the extent of histone 3 lysine 4 methylation (a marker of transcriptional activity) within the paternal G_sα promoter (Sakamoto et al., 2004).

5.1. Mice with deletion of the exon 1A imprinting control region (1A⁻)

In mice the maternal-specific methylation of exon 1A is established during oogenesis and maintained throughout pre- and postimplantation development, features consistent with this region being an primary imprinting control center (Liu et al., 2000b). Evidence that the exon 1A region is important for G_sα imprinting comes from the observation that the maternal-specific methylation imprint in this region is lost in PHP1b patients, who have renal PTH resistance but lack the AHO phenotype and maintain normal G_sα levels in peripheral blood cells (Liu et al., 2000a).

We have proposed a model in which the exon 1A region has one or more cis-acting regulatory elements (eg. silencer elements) which suppress activity of the G_sα promoter and which are both methylation-sensitive and tissue-specific (Fig. 3) (Liu et al., 2000a; Weinstein et al., 2001). Methylation of the silencer prevents binding of a trans-acting repressor protein allowing the maternal G_sα promoter to remain active in all tissues. Suppression of the paternal G_sα allele would occur only in those tissues where the repressor is expressed (eg. renal proximal tubules). In PHPIb the maternal-specific methylation is absent, allowing the repressor to bind to both alleles leading to G_sα deficiency in renal proximal tubules and PTH resistance. In most other tissues where G_sα is normally not imprinted (eg. peripheral blood cells) G_sα expression levels will not be affected because the repressor is not present in these tissues. It is possible that the cis-element is an insulator, rather than a repressor. The imprinting mechanism is not likely to depend on the exon 1A promoter activity or its transcripts, as the tissue distribution of exon 1A expression does not correlate with the tissue distribution of G_sα imprinting (Liu et al., 2000b).

Two independent groups have generated models in which a significant proportion of the exon 1A differentially methylated region was deleted (Liu et al., 2005a; Williamson et al., 2004). In both cases imprinting of the upstream Nesp, Gnasxl, and Nespas imprinted regions was unaffected, indicating that imprinting of these regions is independent of the exon 1A imprinting control center and that the imprinting of the Gnas gene is dependent on at least two independent imprinting control centers. This is consistent with findings in PHP1b patients showing that imprinting of exon 1A may be disrupted without affecting the NESP, NESPAS, or XLαs upstream regions and that a control element within the linked STX16 gene appears to be important for imprinting of exon 1A, but not the other upstream regions (Bastepe et al., 2003; Linglart et al., 2005; Liu et al., 2000a; Liu et al., 2005b).

The G_sα imprinting model proposed above would predict that maternal deletion of exon 1A should have no effect on G_sα expression, while paternal exon 1A deletion should result in G_sα overexpression in tissues where G_sα is normally imprinted due to loss of the cis-acting regulatory elements required for silencing of the paternal G_sα promoter (Fig. 3). As predicted, Liu and colleagues showed that paternal exon 1A deletion (1A^+/p−) resulted in G_sα overexpression in renal proximal tubules (where G_sα is normally imprinted) but not in liver (where G_sα is not imprinted), while maternal deletion (1A^m−/+) had no effect on G_sα expression in either tissue (Liu et al., 2005a). Consistent with increased G_sα expression in renal proximal tubules, both paternal exon 1A deletion models showed evidence for increased PTH sensitivity with reduced serum PTH levels (Liu et al., 2005a; Williamson et al., 2004). Similarly, Williamson and colleagues showed that paternal, but not maternal deletion of exon 1A (Δ1A-DMR) resulted in G_sα overexpression in neonatal BAT, but had little effect on allele-specific G_sα expression in brain (Williamson et al., 2004). Moreover, paternal exon 1A deletion reversed the subcutaneous edema, preweaning lethality, and other gross phenotypic changes observed in neonatal G_sα-specific knockout and Oed mice, respectively (Liu et al., 2005a; Williamson et al., 2004). In these latter models mutation of the maternal G_sα allele leads to severe G_sα deficiency in tissues where G_sα is normally imprinted. The presence of the exon 1A deletion on the paternal allele reverses the imprinting, allowing G_sα to be expressed in these tissues and the phenotypes to be rescued.

5.2. Mice with deletion of the Nespas imprinting control region

A portion of the Nespas promoter region was identified as a likely primary imprinting control region based upon the fact that its maternal-specific methylation is established during oogenesis in mice and maintained through development (Coombes et al., 2003). To determine the role of this region in Gnas imprinting, a mouse model was generated in which 1.6 kilobases including the Nespas promoter and first exon were deleted (ΔNAS-DMR) (Williamson et al., 2006). Maternal transmission of ΔNAS-DMR produced no phenotype in offspring while paternal heterozygotes were born at the expected frequency but died by postnatal day 2.

Paternal ΔNAS-DMR deletion resulted in loss of paternal Nesp imprinting, with loss of paternal-specific methylation of the Nesp promoter and biallelic expression of Nesp transcripts, while the maternal ΔNAS-DMR deletion had no effect on Nesp imprinting. This provides strong evidence that Nesp imprinting, which is established during postimplantation development (Liu et al., 2000b) and therefore is not a primary imprinting event, is dependent on the Nespas imprinting control center. This is consistent with the fact that the imprinting status of the NESP55 and NESPAS promoters are virtually always concordant in PHP1b patients (Liu et al., 2005b). While antisense transcripts have been shown to be important for imprinting of other genes (Sleutels et al., 2002), it is unclear whether Nespas transcripts themselves are required for Nesp imprinting, as studies show that the tissue-specific expression of Nespas and Nesp do not correlate with each other (Ball et al., 2001; T. Li et al., 2000).

The ΔNAS-DMR model suggests that Gnasxl imprinting is not dependent upon Nespas, as the deletion on either allele did not affect Gnasxl methylation or allele-specific expression. This may well also be the case in humans, as the imprinting status of the NESPAS promoter and XLαs exon 1 are different in a significant number of PHP1b patients (Liu et al., 2005b). However XLαs expression was significantly reduced in mice with the paternal deletion, as the deleted region appears to have enhancer activity for the Gnasxl promoter (Williamson et al., 2006). The severe phenotype in paternal ΔNAS-DMR mice is likely to result from XLαs deficiency, as the phenotype is reminiscent of that observed in XLαs-specific knockout mice (see below) (Plagge et al., 2004).

The presence of the ΔNAS-DMR deletion on the maternal-allele had no effect on maternal-specific methylation of exon 1A nor any effect on expression of G_sα or exon 1A-specific transcripts, and therefore Nespas is not required to establish maternal-specific exon 1A methylation. Rather, deletion mutations identified in familial PHP1b patients suggest that elements required to establish or maintain this methylation may be located within the closely linked STX16 gene (Bastepe et al., 2003; Linglart et al., 2005) and in the vicinity of the NESP55 upstream region (Bastepe et al., 2005). In contrast, when the deletion was on the paternal allele it resulted in partial methylation of the exon 1A region on the paternal allele, associated with reduced paternal expression of exon 1A transcripts. This was associated with loss of G_sα imprinting, with G_sα overexpression in neonatal BAT but no effect on G_sα expression in liver. Moreover, the presence of the ΔNAS-DMR deletion on the paternal allele partially rescued the edematous phenotype seen in mice with the maternal G_sα Oed mutation due to relaxation of G_sα imprinting on the paternal allele. These observations provide further evidence for the important role of exon 1A for G_sα imprinting. These findings also suggest that the Nespas region may in some way be important to prevent the exon 1A region from being methylated, although how important this is in the normal imprinting process and the mechanisms involved need to be further studied. The effects on exon 1A and G_sα imprinting and expression were not due to reduced XLαs expression, as there were no similar changes in XLαs knockout mice (Plagge et al., 2004).

6. Transcript-specific Gnas knockout mouse models

6.1. G_sα-specific knockout models

In order to determine the effects of deleting G_sα specifically without affecting the expression of the other Gnas transcripts, two groups have generated models in which the G_sα-specific exon 1 was deleted (M. Chen et al., 2005a; Germain-Lee et al., 2005). In both models homozygotes were embryonically lethal, confirming that total G_sα deficiency is sufficient to impair gestational development.

In one model (E1⁻) mice were studied in the CD-1 genetic background so they could be directly compared to E2⁻ mice (M. Chen et al., 2005a). Similar to E2^m−/+ mice, E1^m−/+ mice were born with subcutaneous edema, confirming that this manifestation is secondary to loss of G_sα expression from the maternal allele (see Table 2 for summary of phenotypic manifestations due to loss of specific Gnas gene products). As this manifestation is absent in E1^+/p− mice, the edema seen in several models with disruption of the maternal G_sα allele presumably results from severe G_sα deficiency in one or more tissues where G_sα normally undergoes imprinting. Support from this comes from the fact that the edema phenotype is rescued by loss of paternal G_sα imprinting resulting from deletion or methylation of the exon 1A imprinting control region on the paternal allele (Liu et al., 2005a; Williamson et al., 2004; Williamson et al., 2006). G_sα imprinting in placenta may account for this phenotype, as several genes have been shown to be imprinted in placenta and abnormal placental function may lead to abnormal water balance.

However in several respects, E1^m−/+ and E2^m−/+ are not alike, as E1^m−/+ mice lack the neurological phenotype present in E2^m−/+ mice and have better survival. Moreover adult E1^m−/+ mice have more severe obesity, hypertriglyceridemia, and insulin resistance (M. Chen et al., 2005a), whereas E2^m−/+ mice have reduced triglyceride levels and increased insulin sensitivity (S. Yu et al., 2001; S. Yu et al., 2000). These differences may result from disruption of other maternal-specific Gnas products in E2^m−/+ mice. Disruption of NESP55 is unlikely to account for these differences, as its expression is not disrupted in E2^m−/+ mice and NESP55 deficiency does not appear to produce a major phenotype in mouse (Plagge et al., 2005) or humans (Kim et al., 2000; Liu et al., 2000a). Whether the differences observed between E2^m−/+ and E1^m−/+ mice result from disruption of other potential maternal Gnas transcripts (Holmes et al., 2003) or from genetic drift in the outbred CD-1 background over time is unknown.

Comparison of E1^+/p− and E2^+/p− mice shows very dramatic differences, as E1^+/p− do not exhibit the suckling defect and perinatal lethality observed in E2^+/p− mice (M. Chen et al., 2005a). In addition, adult E1^+/p− mice have mildly increased fat mass, insulin resistance, and hypertriglyceridemia, whereas adult E2^+/p− mice are very lean with increased sensitivity and hypotriglyceridemia (M. Chen et al., 2004; S. Yu et al., 2001; S. Yu et al., 2000). Studies in XLαs-specific knockout mice (Plagge et al., 2004; Xie et al., 2006) confirm that the severe neonatal and adult phenotype in E2^+/p− mice results from disruption of XLαs expression from the paternal allele (see below). The more severe phenotype observed in E1^m−/+ compared to E1^+/p− mice (subcutaneous edema, increased lethality, more severe metabolic phenotype) presumably results from the effects of tissue-specific G_sα imprinting, as reversal of paternal G_sα imprinting has been shown to rescue some of the features of the E1^m−/+ phenotype (Liu et al., 2005a; Williamson et al., 2004; Williamson et al., 2006).

Another model with deletion of G_sα exon 1 was studied in a different genetic background (SvEv or SvEv/CD-1) (Germain-Lee et al., 2005). Studies in these mice confirmed tissue-specific G_sα imprinting, and the presence of PTH and TSH resistance in the maternal heterozygote. Both maternal and paternal heterozygotes had greater preweaning lethality than their counterparts studied in the CD-1 genetic background (M. Chen et al., 2005a). Similar to E1⁻ mice, the maternal heterozygotes weighed more than their paternal counterparts and many of the early features seen in E2^m−/+ and E2^+/p− mice were absent. There were other differences besides lethality between the two G_sα-specific knockout models. Germain-Lee do not report any perinatal subcutaneous edema in the maternal heterozygotes. There was also a tendency for the paternal heterozygotes (particularly the males) to weigh less than their normal littermates in the Germain-Lee study, which may be partly accounted for by reduced body length and differences in genetic background. The Germain-Lee study also provided evidence that maternal heterozygotes (both male and female) have reduced fertility, perhaps due to gonadotropin resistance as seen in PHP1a patients (Mantovani et al., 2002; Namnoum et al., 1998), and that female maternal heterozygotes show poor mothering behavior. Their study showed no obvious skeletal phenotype or changes in bone remodeling based upon histomorphometric studies. These mice were also used to show that loss of G_sα signaling leads to reduced ethanol consumption and increased sensitivity to the sedative effects of ethanol (Wand et al., 2001).

6.2. XLαs-specific knockout model

XLαs-specific knockout mice were generated in which 60 base pairs within XLαs exon 1 were deleted (Gnasxl⁻) (Plagge et al., 2004). This deletion had no effect on Gnas imprinting and only resulted in XLαs deficiency with an associated phenotype when present on the paternal allele as XLαs is only paternally expressed. Overall Gnasxl^+/p− mice had a phenotype very similar to that of E2^+/p−, MatDp distal 2, and to some extent Sml mice, with failure to suckle associated with hypoglycemia and neonatal lethality. Like the other models, these mice had markedly reduced lipid accumulation in BAT and WAT and maintained a lean phenotype through adulthood (Plagge et al., 2004; Xie et al., 2006). It was proposed that XLαs deficiency may result in reduced suckling by a direct effect on orofacial muscle activity, as XLαs is expressed in the facial, hypoglossal, and trigeminal nuclei, which provide motor innervation to the orofacial muscles (Plagge et al., 2004). It was also suggested that loss of XLαs in the locus coeruleus and the cholinergic laterodorsal tegmental nucleus may underlie the observed hypoactivity. Based upon the findings in Gnasxl^+/p− mice and the lack of similar findings in E1^+/p− mice, it is clear that the severe phenotype in E2^+/p− and MatDp distal 2 is due to XLαs deficiency, which appears to have a dominant effect.

Studies of adult Gnasxl^+/p− mice in a CD-1 genetic background (Xie et al., 2006) showed them to have a metabolic phenotype very similar to that of E2^+/p− mice (M. Chen et al., 2004; S. Yu et al., 2001; S. Yu et al., 2000), including reduced adiposity, increased energy expenditure, hypolipidemia, and greater than normal glucose tolerance and insulin sensitivity. Gene expression studies in BAT, WAT, muscle, and liver suggested that the phenotype results from increased lipid oxidation in adipose tissue, particularly in BAT. Several lines of evidence showed that these changes were due to BAT overstimulation due to increased sympathetic nervous system activity, rather than cell-autonomous effects on adipocytes (Xie et al., 2006). First, similar to E2^+/p− mice (S. Yu et al., 2000), Gnasxl^+/p− mice had very elevated levels of urinary norepinephrine, consistent with increased sympathetic nervous system activity. Second, Gnasxl^+/p− mice did not show evidence for increased responsiveness to a β3 adrenergic agonist which specifically activates adipose tissue. Finally, XLαs is not expressed in adipose tissue by two weeks of age (Xie et al., 2006), making it unlikely that disruption of XLαs expression would result in a cell-autonomous change in adipocytes.

It therefore appears most likely that XLαs is involved in regulation of sympathetic output from the central nervous system. Whether this occurs by acting at sites in the central nervous system that are distinct from G_sα or whether it or its alternative splice form XLN1 directly inhibit G_sα signaling remains to be determined. A role for XLN1 may explain why the Sml phenotype is not as severe, as the Sml exon 6 mutation alters the coding sequence of G_sα and XLαs, but not that of XLN1 (Cattanach et al., 2000; Plagge et al., 2004; Skinner et al., 2002). Although paternal deletion of 20q13 including GNAS has been associated with feeding difficulties, poor growth, and abnormal adipose tissue distribution in humans (Genevieve et al., 2005), it remains unclear if XLαs has an important role in humans as PPHP patients with paternal GNAS mutations that disrupt XLαs do not produce a similar phenotype.

6.3. NESP55-specific knockout model

Mice with specific deficiency of NESP55 were generated by deleting 26 bp at the NESP55 start codon (Plagge et al., 2005). NESP55 knockout mice were reported to have a grossly normal phenotype with no lethality, obvious changes in body habitus, or decrease in fertility. Therefore, NESP55 deficiency does not contribute to the severe phenotypes observed in E2^m−/+, PatDp distal 2, and Oed mice. This is not unexpected, as the E2⁻ and Oed mutations do not disrupt the NESP55 coding regions and NESP55 expression is unaffected in E2^m−/+ mice (T. Xie, unpublished results). Moreover, loss of NESP55 expression due to biallelic methylation of the NESP55 promoter in PHP1b patients (Liu et al., 2000a) or due to deletion within the coding region (Kim et al., 2000) is not associated with an obvious phenotype. Behavioral studies in NESP55 knockout mice showed these mice to have increased reactivity to a novel environment that was not due to general changes in locomoter activity or anxiety, and was not associated with gross differences in monoamine concentrations within the pons or prefrontal cortex.

7. Role of Gnas in development and physiology based upon mouse knockout models

7.1. Glucose and energy metabolism

Studies in E2⁻ mice showed them to have opposite changes in energy metabolism depending on parental inheritance (S. Yu et al., 2000). E2^m−/+ mice developed obesity with increased lipid accumulation in BAT and WAT, reduced energy expenditure at ambient temperature and activity levels. These changes are most likely due to reduced sympathetic activity, as these mice had normal energy expenditure at the thermoneutral temperature 30° when sympathetic activity is minimized, normal metabolic and biochemical responsiveness to a β3-adrenergic agonist, and reduced urine excretion of norepinephrine and its metabolites (S. Yu et al., 2000). In contrast, E2^+/p− mice had a severely lean phenotype with increased levels of activity and energy expenditure, presumably due to increased sympathetic activity as they had increased urine excretion of norepinephrine (S. Yu et al., 2000). Leptin levels varied in proportion to adiposity in both groups of mice. Despite the opposite effects on energy metabolism, both groups of mice had hypolipidemia and improved glucose tolerance and insulin sensitivity, although the effects were greater in E2^+/p− mice (M. Chen et al., 2004; S. Yu et al., 2001; S. Yu et al., 2000). Isolated skeletal muscles from both E2^m−/+ and E2^+/p− mice showed similar increases in insulin-stimulated glucose uptake without any changes in the expression of the glucose transporter GLUT4 (S. Yu et al., 2001). E2^+/p− have increased insulin sensitivity in liver, muscle, and adipose tissue (M. Chen et al., 2004; S. Yu et al., 2001).

Studies in transcript-specific Gnas knockouts have begun to decipher the role of individual gene products in metabolic regulation. E1⁻ (G_sα-specific knockout) mice develop increased fat mass, glucose intolerance, insulin resistance, and hypertriglyceridemia, but the effects are much greater in E1^m−/+ than in E1^+/p− mice (M. Chen et al., 2005a). Another G_sα-specific knockout model also showed greater obesity in mice with maternal inheritance of the mutation (Germain-Lee et al., 2005). These findings mimic the pattern of inheritance in humans, as obesity is a more prominent feature of PHP1a (who have maternal G_sα mutations) as compared to PPHP patients (who have paternal G_sα mutations) (Long et al., 2006). Therefore the obesity phenotype resulting from heterozygous G_sα mutation appears to be an imprinted phenotype, similar to multihormone resistance. The central nervous system appears to be a good candidate site for this effect, as E1^m−/+ mice were shown to have altered balance between food intake and energy expenditure rates, which are both regulated in the central nervous system, and G_sα is known to mediate pathways involved in energy balance (eg. melanocortin signaling through MC4 receptors).

E1^+/p− mice had mild obesity and insulin resistance, in contrast to the lean phenotype with increased insulin sensitivity observed in E2^+/p− mice, thus implicating that the E2^+/p− metabolic phenotype was due to loss of an alternative paternal-specific Gnas transcript. As discussed above, Gnasxl^+/p− mice had a metabolic phenotype that was very similar to that of E2^+/p− mice (Plagge et al., 2004; Xie et al., 2006), confirming that XLαs deficiency is the underlying cause of the E2^+/p− phenotype. Studies in E2^+/p− and Gnasxl^+/p− outlined above implicate XLαs (or XLN1) as a negative regulator of sympathetic nervous system activity in the central nervous system, either by directly counteracting G_sα function or by acting in brain regions distinct from those of G_sα (Xie et al., 2006). Studies in NESP55 knockout mice suggests that NESP55 does not play an important role in metabolic regulation (Plagge et al., 2005).

Tissue-specific knockouts are also useful to examine the role of G_sα signaling in specific tissues on metabolic regulation. Mice with liver-specific G_sα deficiency due to homozygous G_sα deletion (LGsKO) were generated by mating E^fl/fl mice with albumin-promoter-Cre transgenic mice (M. Chen et al., 2005b). Consistent with G_sα mediating the effects of glucagon, which counterregulates insulin in the liver, these mice had increased hepatic glycogen synthesis and had reduced expression of hepatic genes involved in gluconeogenesis, such as phosphoenolpyruvate carboxykinase and glucose-6-phosphatase. Although LGsKO mice were hypoglycemic and hypoinsulinemic at baseline, they maintained normal glucose and insulin levels after prolonged fasting, probably due to prolonged breakdown of increased hepatic glycogen stores and increased extrahepatic gluconeogenesis. LGsKO mice did not lower their triglyceride levels upon fasting due to inappropriate persistence of lipogenesis. Hepatic glucagon resistance and/or hypoglycemia led to markedly elevated circulating glucagon and glucagon-like peptide 1 (GLP-1) levels and pancreatic α-cell hyperplasia. LGsKO mice also had increased levels of pancreatic insulin secretion, possibly due to increased GLP-1 levels, and increased insulin sensitivity in skeletal muscle and WAT. These mice also had reduced adiposity, although the mechanism for this is unclear.

7.2. Skeletal development

Heterozygous G_sα mutations in AHO patients are associated with several skeletal abnormalities, including short stature, ectopic ossifications, and brachydactyly, in which specific long bones in the hands and feet are shortened due to premature chondrocyte differentiation and fusion of the growth plates, consistent with a defect in endochondral ossification. Similar skeletal abnormalities have not been reported in the analogous genetic mouse models (E1⁻ or E2⁻) (M. Chen et al., 2005a; Germain-Lee et al., 2005; S. Yu et al., 1998).

Endochondral ossification is a process of ordered chondrocyte differentiation and osseous replacement at growth plates leading to longitudinal bone growth (Kronenberg, 2003). The epiphyseal growth plate consists of three zones (from the end toward the metaphysis): the periarticular zone, the proliferative zone, and the hypertrophic zone. The periarticular zone consists of immature chondrocytes which proliferate and supply cells which form ordered columns of proliferating chondrocytes in the proliferative zone. At the other end of the columns cells undergo hypertrophic differentiation and these cells form the scaffold for ossification which occurs at the end of the growth plate towards the diaphysis (center) of the bone. Periarticular chondrocytes secrete PTH-related peptide (PTHrP) which works in a paracrine manner to inhibit chondrocyte differentiation and expression of Indian hedgehog (Ihh). Once cells reach a certain distance from the periarticular zone where PTHrP levels are below a critical level, cells begin to differentiate (prehypertrophic chondrocytes) and secrete Ihh. Ihh positively feeds back on the periarticular zone to stimulate PTHrP secretion and has independent effects on chondrocyte differentiation.

PTHrP binds to the PTH/PTHrP receptor (PPR) (also referred to as PTH1R), which stimulates both G_s-cAMP and G_q/11-phospholipase C pathways (Bringhurst et al., 1993; Schwindinger et al., 1998). Evidence suggests that these two signaling pathways may have opposite effects on cellular proliferation and differentiation (G_s-cAMP stimulating proliferation and inhibiting differentiation, G_q/11 having opposite effects) (Guo et al., 2002; Guo et al., 2001). The G_s-cAMP action appears to be dominant as deficiency of either PTHrP or PPR results in severe chondrodysplasia due to premature hypertrophic differentiation of growth plate chondrocytes in both mice and humans (Jobert et al., 1998; Karaplis et al., 1994; Lanske et al., 1996; Vortkamp et al., 1996).

Two related models using targeted disruption of Gnas have provided direct evidence that G_sα plays an important role in regulating chondrocyte differentiation. In one model G_sα-null embryonic stem cells were generated from E2^m−/p− blastocysts and injected into normal blastocysts to generate E2^m−/p−:wild-type chimeras (Bastepe et al., 2004). Examination of growth plates showed mutant chondrocytes to undergo premature hypertrophic differentiation, which can be rescued by transfection of a G_sα-expressing construct. Similar studies using E2^m−/+ or E2^+/p− cells showed only a partial effect, which was somewhat greater using E2^+/p− cells. Expression studies showed no differences in G_sα expression between E2^m−/+ and E2^+/p− chondrocytes, and it was suggested that E2^+/p− mice may show a greater effect due to disruption of XLαs, which is expressed at low levels in chondrocytes. While the columns of proliferative chondrocytes were elongated in PPR^−/−:wild-type chimeras (Chung et al., 1998), this was not observed in E2^m−/p−:wild-type chimeras. One potential explanation for this difference that was proposed is that PTHrP stimulation of G_q-phospholipase C is only preserved in the latter model.

In the second model, mice with chondrocyte-specific G_sα deficiency were generated by mating E1^fl:fl mice with transgenic mice in which the Cre recombinase gene was driven by the collagen 2a1 gene promoter (Sakamoto et al., 2005a). Similar to PTHrP and PPR knockout mouse models (Karaplis et al., 1994; Kobayashi et al., 2002; Lanske et al., 1996), these mice died soon after birth with a severe skeletal abnormality including shortened limbs and domed skull and had severe shortening of the growth plated due to premature chondrocyte differentiation. While PPR and Ihh expression in prehypertrophic chondrocytes were unaffected, PTHrP expression in periarticular chondrocytes was increased presumably because Ihh-secreting prehypertrophic chondrocytes were in closer proximity to these cells. These findings confirm the critical role for G_sα in mediating the effects of PTHrP on chondrocyte differentiation in the growth plate. These mice also developed ectopic chondrocyte differentiation under the perichondrium in the anterior surface of proximal metaphysis of the tibia near the end of gestation. This effect is unlikely to be due to disruption of PTHrP signaling, as PTHrP is not expressed within this area and this phenotype was not observed in mice with chondrocyte-specific PPR deficiency (Kobayashi et al., 2002). Consistent with the lack of a skeletal phenotype in E1⁻ or E2⁻ heterozygotes, mice with heterozygous deletion of G_sα exon 1 in chondrocytes did not demonstrate an obvious phenotype.

The role of G_sα signaling in osteoblasts and osteocytes was examined by mating E1^fl:fl mice with mice with a collagen Ia1-Cre recombinase transgene (BGsKO) (Sakamoto et al., 2005b). All BGsKO mice died soon after birth and had subcutaneous edema similar to other maternal G_sα knockout models. It is presumed that this effect was due to extraskeletal Cre expression leading to G_sα deficiency in other tissues. BGsKO had severe hypoplasia of the anterior craniofacial bones, often associated with cleft palate, suggesting an important role for G_sα in craniofacial skeletal development, along with premature endochondral ossification of Meckel’s cartilage, a feature also observed in PTHrP knockout mice (Ishii-Suzuki et al., 1999).

BGsKO mice showed no effects on early skeletal development, including formation of the initial cartilaginous template or bone collar, nor were there effects on the growth plates. However, mice had shortening of long bones associated with reduced formation of primary spongiosa, which is the precursor for trabecular bone. In contrast, the cortical bone was thickened due to reduced osteoclast resorption at the endosteal surface resulting from reduced expression of the osteoclast-stimulating factor receptor activator of nuclear factor-κB (RANKL). These differential effects on cortical and trabecular bone with low bone remodeling are similar to those observed in PTH and PPR knockout mice (Lanske et al., 1999; Miao et al., 2002) and opposite to those seen in mice with osteoblast-specific expression of a constitutively activated PPR (Calvi et al., 2001), indicating that G_sα mediates the differential effects of PTH on cortical and trabecular bone. Osteoblasts in BGsKO mice had normal levels of alkaline phosphate, an early marker of osteoblast differentiation, but reduced levels of later differentiation markers, suggesting that G_sα pathways are important for the later stages of osteoblast differentiation in vivo.

7.3. Renal function

G_sα-cAMP pathways mediate different receptor signaling pathways and physiological responses along the nephron, and the effects of heterozygous G_sα deficiency in mice is further complicated by the fact that G_sα imprinting is variable in different segments of the nephron (Ecelbarger et al., 1999; Weinstein et al., 2000; S. Yu et al., 1998). In the proximal tubule G_sα mediates the action of PTH to reduce phosphate reabsorption, leading to increased renal phosphate wasting, and to increase conversion of 25 hydroxyvitamin D to 1, 25 dihydroxyvitamin D, the active form which stimulates calcium absorption from the gut and calcium release from bone. PHP1a and PHP1b patients develop hypocalcemia, hyperphosphatemia, and low circulating 1, 25 dihydroxyvitamin D levels due to G_sα deficiency and loss of PTH signaling in renal proximal tubules. Studies in E2⁻ mice showed G_sα to be imprinted in renal proximal tubules and E2^m−/+, but not E2^+/p− mice, develop renal PTH resistance due to loss of the active maternal G_sα allele (S. Yu et al., 1998). This pattern was mimicked in G_sα-specific knockout mice (Germain-Lee et al., 2005). These results explain why PHP1a patients with maternal G_sα mutations develop PTH resistance while PPHP patients with paternal G_sα mutations do not. Moreover, paternal deletion of 1A in 1A^+/p− and paternal Δ1A-DMR mice leads to loss of G_sα paternal imprinting, G_sα overexpression in renal proximal tubules, and increased PTH sensitivity (Liu et al., 2005a; Williamson et al., 2004). Overall these findings show that G_sα is a rate-limiting signaling component, consistent with results observed in the kidney cell line HEK293 when G_sα levels were altered (Yang et al., 1997).

In contrast to what is observed in proximal tubules, there is no evidence for G_sα imprinting in more distal portions of the nephron, including the loop of Henle, distal convoluted tubules, and collecting ducts (Ecelbarger et al., 1999; S. Yu et al., 1998). Expression of the cAMP-responsive Na-K-2Cl cotransporter gene in the outer medulla (primarily thick ascending limb) was similarly reduced by ~50% in both E2^m−/+ and E2^+/p− mice, consistent with reduced levels of hormone-stimulated cAMP generation in this renal segment (Ecelbarger et al., 1999). This was accompanied by low levels of adenylyl cyclase type VI, suggesting a potential feed-forward mechanism for regulation of this enzyme. Reduced expression of the cotransporter would be predicted to lead to reduced countercurrent multiplication by the loop of Henle and reduced medullary solute concentrations. As predicted, E2^m−/+ mice had reduced urine concentration after a single bolus of vasopressin (Ecelbarger et al., 1999). More chronic vasopressin stimulation resulting from 48 hours of water deprivation resulted in correction of the urinary concentrating defect and normalization of Na-K-2Cl cotransporter gene expression (Ecelbarger et al., 1999; S. Yu et al., 1998), indicating that chronic hormone stimulation can raise cAMP levels sufficiently to elicit a maximal physiological response. In the inner medulla (primarily collecting ducts) hormone-stimulated cAMP, adenylyl cyclase type VI, and aquaporin 2 levels were all unaffected in E2^m−/+ and E2^+/p− mice, suggesting that G_sα is not a rate-limiting signaling component in this segment of the nephron (Ecelbarger et al., 1999).

Mice in which G_sα was deleted in juxtaglomerular cells were generated by mating E1^fl/fl mice with mice in which Cre recombinase was under the control of the endogenous renin promoter (L. Chen et al., 2007). These mice had markedly reduced basal and stimulated plasma renin levels, along with reduced blood pressure, aldosterone levels, and glomerular filtration rate. Studies in juxtaglomerular cells confirmed that renin production was markedly reduced, confirming that renin production is dependent on one or more receptor pathways which are mediated by G_sα. Unexpectedly, these mice also had Cre expression and G_sα deficiency in the renal medulla, which led to vasopressin resistance and a urinary concentrating defect.

8. Conclusion

Gnas is a unique gene in that it has multiple promoters that are oppositely imprinted, resulting in maternally, paternally, and biallelically expressed gene products. This provides an explanation for the unique observation that disruption of the maternal or paternal alleles of Gnas leads to distinct, and in many respects opposite, phenotypes. One prominent theory for why genomic imprinting occurs is the parent-offspring conflict theory, which hypothesizes that for multiparous species the paternal genome promotes increased allocation of resources to the offspring in order to confer survival advantage, while the maternal genome promotes the opposite effect to conserve maternal resources for future litters (Haig, 2004; Moore & Haig, 1991). This theory predicts that disruption of paternally expressed imprinted genes inhibits fetal and early postnatal growth, while disruption of maternally expressed imprinted genes promotes growth. Studies in Gnas knockout mouse models fits these predictions, as E2^m−/+ and PatDp distal 2 mice have increased body mass, while E2^+/p− and MatDp distal 2 mice have reduced body mass.

Studies in Gnas knockout mouse models have confirmed the tissue-specific nature of G_sα imprinting, which provides a molecular mechanism for the different parent-of-origin effects in PHP1a and PPHP, respectively, and have confirmed the important role of the 1A region in tissue-specific G_sα imprinting, providing a molecular model for the pathogenesis of PHP1b. Future tissue-specific G_sα knockout models will define the tissue where disruption of the maternal G_sα allele leads to obesity in PHP1a patients and mice with maternal G_sα mutations in the germline.

Abbreviations

G_sα

stimulatory G protein α-subunit

AHO

Albright hereditary osteodystrophy

XLαs

extra-large isoform of G_sα

NESP55

neuroendocrine-specific protein of 55 kDa

E2⁻

mice with disruption of Gnas exon 2

PatDp and MatDp distal 2

paternal or maternal duplication of distal chromosome 2, respectively

Oed-Sml

maternal or paternal Gnas exon 6 mutation, respectively

E1⁻

mice with deletion of G_sα exon 1

1A⁻

Δ1A-DMR, mouse models with deletion of exon 1A differentially methylated region

PTH

parathyroid hormone

TSH

thyroid stimulating hormone

cAMP

cyclic adenosine monophosphate

PKA

cAMP-dependent protein kinase

PHP1a

PHP1b, pseudohypoparathyroidism types 1a and 1b

PPHP

pseudopseudohypoparathyroidism

POH

progressive osseous heteroplasia

XLN1

neural-specific truncated form of XLαs

Alex

alternative protein transcribed from XLαs exon 1

E1^fl/fl

mice with loxP recombination sites surrounding G_sα exon 1

BAT

WAT, brown and white adipose tissue

ΔNAS-DMR

mice with deletion of Nespas promoter region

PTHrP

PTH-related peptide

Ihh

Indian hedgehog

PPR

PTH/PTHrP receptor, BGsKO, osteoblast/osteocyte-specific G_sα knockout

LGsKO

liver-specific G_sα knockout

GLP-1

glucagon-like peptide 1