Alternative Gnas gene products have opposite effects on glucose and lipid metabolism

Abstract

Gnas is an imprinted gene with multiple gene products resulting from alternative splicing of different first exons onto a common exon 2. These products include stimulatory G protein α-subunit (G_sα), the G protein required for receptor-stimulated cAMP production; extralarge G_sα (XLαs), a paternally expressed G_sα isoform; and neuroendocrine-specific protein (NESP55), a maternally expressed chromogranin-like protein. G_sα undergoes tissue-specific imprinting, being expressed primarily from the maternal allele in certain tissues. Heterozygous mutation of exon 2 on the maternal (E2^m-/+) or paternal (E2^+/p-) allele results in opposite effects on energy metabolism. E2^m-/+ mice are obese and hypometabolic, whereas E2^+/p- mice are lean and hypermetabolic. We now studied the effects of G_sα deficiency without disrupting other Gnas gene products by deleting G_sα exon 1 (E1). E1^+/p- mice lacked the E2^+/p- phenotype and developed obesity and insulin resistance. The lean, hypermetabolic, and insulin-sensitive E2^+/p- phenotype appears to result from XLαs deficiency, whereas loss of paternal-specific G_sα expression in E1^+/p- mice leads to an opposite metabolic phenotype. Thus, alternative Gnas gene products have opposing effects on glucose and lipid metabolism. Like E2^m-/+ mice, E1^m-/+ mice had s.c. edema at birth, presumably due to loss of maternal G_sα expression. However, E1^m-/+ mice differed from E2^m-/+ mice in other respects, raising the possibility for the presence of other maternal-specific gene products. E1^m-/+ mice had more severe obesity and insulin resistance and lower metabolic rate relative to E1^+/p- mice. Differences between E1^m-/+ and E1^+/p- mice presumably result from differential effects on G_sα expression in tissues where G_sα is normally imprinted.

The ubiquitously expressed stimulatory G protein α-subunit (G_sα) couples receptors to adenylyl cyclase and is required for the intracellular cAMP response to hormones and other extracellular signals (1). Heterozygous G_sα inactivating mutations lead to Albright hereditary osteodystrophy (AHO), a syndrome characterized by obesity and skeletal and neurobehavioral defects (1). Patients who inherit AHO maternally also develop resistance to parathyroid hormone (PTH), thyroid-stimulating hormone (TSH), and gonadotropins, a condition known as pseudohypoparathyroidism type 1A. In contrast, patients who inherit AHO paternally do not develop multihormone resistance (referred to as pseudopseudohypoparathyroidism) due to the fact that G_sα is paternally imprinted in a tissue-specific manner, being primarily expressed from the maternal allele in certain hormone-responsive tissues (2-5).

G_sα is encoded by a complex imprinted gene (GNAS at 20q13 in human, Gnas on mouse chromosome 2) that produces multiple gene products via the use of multiple alternative promoters and first exons that splice onto a common set of downstream exons (Fig. 1A) (1). Promoters for 55-kDa neuroendocrine-specific protein (NESP55), a chromogranin-like protein, and extralarge G_sα (XLαs), a neuroendocrine-specific G_sα isoform with a long N-terminal extension encoded by its alternative first exon, are located 47 and 35 kb upstream of the G_sα promoter, respectively (6). NESP55 is unrelated to G_sα, and its entire coding region is within its first exon; G_sα exons 2-13 are within the 3′ untranslated region of NESP55 transcripts (7). XLαs is capable of mediating receptor-stimulated cAMP production in transfected cells (8), and XLαs knockout mice (which also lack a neural-specific isoform XLN1 and an alternative translation product named ALEX) have a severe lean and insulin-sensitive phenotype with early lethality (9). NESP55 and XLαs are oppositely imprinted: NESP55 is expressed from the maternal allele, and its promoter is DNA-methylated on the paternal allele, whereas XLαs is paternally expressed, and its promoter region is methylated on the maternal allele (6, 10). Exon 1A, another alternative first exon located just upstream of the G_sα promoter, is methylated on the maternal allele and generates paternal-specific untranslated transcripts (11, 12). Loss of exon 1A imprinting leads to pseudohypoparathyroidism type 1B, a syndrome of isolated parathyroid hormone resistance (12).

Fig. 1.

Generation of E1^- mice. (A) The maternal (Mat) and paternal (Pat) alleles of Gnas are depicted with alternative first exons for NESP55 (NESP), XLαs, 1A, and G_sα (1) splicing to a common set of exons (exons 2-12, not shown). Differentially methylated regions (Meth) and transcriptionally active promoters (arrows) are indicated. The dashed arrow indicates that transcription from the paternal G_sα promoter is suppressed in some tissues. Antisense transcripts from the XLαs promoter are not shown. The diagram is not drawn to scale. (B) The upstream portion of wild-type Gnas including exons 1A, 1, 2, and 3 is shown at the top. The positions of the 5′ and 3′ probes used for Southern analysis are shown above. The scale above is in kb, with position 0 being the G_sα translational start site. A loxP site (triangle) was inserted within an NruI site (N) at +419, and upstream and downstream BamH1 (B) fragments were cloned into pLoxpneo to generate the targeting construct. The E1^neo-fl and E1^- alleles are shown below. S, SacI; Bg, BglII; H, HpaI; Neo, neomycin resistance gene; TK, thymidine kinase cassette. (C) Southern analysis of genomic DNA from ES cells (two right lanes) or mice (two left lanes) after SacI digestion and hybridization with the 5′ probe. Genotypes are indicated above each lane. (D)G_sα mRNA levels (expressed as percent of wild-type) in liver (Left) and BAT (Right) from E1^+/p- and E1^m-/+ mice (n = 4-5 per group).

We previously generated mice with mutation of Gnas exon 2 (E2), an exon common to all Gnas transcripts, which can potentially disrupt multiple Gnas gene products. Homozygotes were embryonically lethal, whereas heterozygotes with mutation on the maternal (E2^m-/+) and paternal (E2^+/p-) allele had distinct phenotypes (referred to as m-/+ and +/p- in ref. 2). E2^m-/+ mice had s.c. edema at birth, and most died between 1 and 3 weeks of age, after developing neurological signs. In contrast, E2^+/p- mice were small at birth, and most failed to suckle and died soon after birth. These features were also present in mice with paternal and maternal uniparental disomy of distal chromosome 2 where Gnas resides, suggesting that the phenotypes result from loss of maternal- and paternal-specific Gnas gene products, respectively (13, 14). Similar phenotypes were also observed in mice with a heterozygous missense mutation in exon 6 (15, 16). Surviving E2^m-/+ mice became obese, hypometabolic, and hypoactive, whereas E2^+/p- survivors had reduced adipocity, with increased metabolic rate, activity levels, glucose tolerance, insulin sensitivity, and lipid clearance (17-19). The opposite metabolic phenotypes in E2^m-/+ and E2^+/p- mice were associated with decreased and increased urinary norepinephrine excretion, respectively, suggesting they might be due to opposite changes in sympathetic activity (17).

We have now generated mice with deletion of G_sα exon 1 (E1), which should specifically disrupt expression of G_sα but not alternative Gnas gene products. Homozygotes were embryonically lethal. In contrast to XLαs knockout and E2^+/p- mice, heterozygotes with paternal E1 mutation (E1^+/p-) had normal survival and developed obesity and insulin resistance. The E2^+/p- phenotype appears to result from XLαs deficiency. Alternative Gnas gene products have opposing effects on glucose and lipid metabolism. The E1^m-/+ and E2^m-/+ phenotypes differed in several respects, raising the possibility of other maternal-specific Gnas gene products. E1^m-/+ mice also differed from E1^+/p- mice, presumably because of differences in G_sα expression in tissues where G_sα is normally imprinted.

Materials and Methods

Construction of Targeting Vector. A genomic clone including the Gnas upstream region was isolated by screening a 129/SvEv genomic library with a 1.4-kb probe containing E1. A 10.8-kb NotI fragment was subcloned into pBluescript SKII (Promega) (Fig. 1B). Complementary DNA linkers containing a loxP site (forward sequence, 5′-GAAGCTTATAACTTCGTATAGCATACATTATACGAAGTTATGAGCTCCATATCG-3′) were ligated into an NruI site at position +419 (all positions are relative to G_sα translational start site). A 3.1-kb BamH1 fragment (-4,728/-1,601) was ligated into a HpaI site located upstream of the loxP-neomycin resistance (Neo)-loxP cassette of pLoxpneo (20), and a 5.6-kb BamH1 fragment (-1,601/+3,962) was ligated into a BamH1 site located downstream of loxP-Neo-loxP.

Generation of G_sα Knockout Mice. TC-1 ES cells (21) were electroporated with the linearized targeting vector and selected with G418 and 1-(2′-deoxy-2′-f luoro-β-d-arabinofuranosyl)-5-iodouracil. Doubly resistant clones were screened for the E1^neo-fl allele by Southern analysis by using 5′ and 3′ probes, which were located outside the targeting construct (Fig. 1 B and C). These cells were injected into C57BL/6 blastocysts implanted into pseudopregnant foster mothers. Three chimeric mice (≈70-80% derived from ES cells, as judged by coat color) were bred with Black Swiss mice, and two had germ-line transmission. F₁ mice were bred with EIIa promoter-cre transgenic mice (22), and mice with recombination at the most upstream and downstream loxP sites were selected, resulting in mice with deletion of E1 (E1^- allele, Fig. 1B, referred to as Δ^-1601/+419 in ref. 23). E1^- mice were mated with wild-type CD-1 mice (Charles River Laboratories) for several generations to place them in the same genetic background as previously studied E2^- mice (2, 17-19). Subsequent genotyping was performed by PCR of mouse tail DNA by using the common upstream primer 5′-GAGAGCGAGAGGAAGACAGC-3′ and downstream primers 5′-TCGGGCCTCTGGCGGAGCT T-3′ and 5′-AGCCCTACTCTGTCGCAGTC-3′, which generated 330- and 250-bp bands from the wild-type (E1⁺) and E1^- alleles, respectively. Primers correspond to GenBank accession no. AF152375, bases 1293-1312, 1619-1600, and 3507-3488, respectively. Animals were maintained on a 12:12-h light/dark cycle and standard pellet diet (NIH-07, 5% fat by weight). Except when noted, all experiments were performed on 12- to 14-week-old male mutant mice and wild-type littermates. Experiments were approved by the National Institute of Diabetes and Digestive and Kidney Diseases Animal Care and Use Committee.

Northern Analysis. Total tissue RNA was isolated by using TRIzol reagent (Invitrogen). For Northern analysis, membranes were hybridized with a ³²P-labeled E1-specific cDNA probe, which was generated by PCR by using primers 5′-GGCAACAGTAAGACCGAGGACCAG-3′ and 5′-ACCCAGCAGCAGCAGGCG-3′ (GenBank accession no. AF152375, bases 3011-3034 and 3139-3122, respectively). G_sα signals were quantified by using a Fuji BAS1500 phosphorimager and normalized to 18S RNA, which was quantified by ethidium bromide by using alphaease fc software (Alpha Innotech, San Leandro, CA).

Body Composition, Food Intake, Metabolic Activity, and Activity Measurements. Body composition was measured by using the Minispec mq10 NMR analyzer (Bruker Optics, The Woodlands, TX). Food intake, metabolic rates (oxygen consumption rate by indirect calorimetry), and activity levels were determined as described (17).

Blood and Tissue Analysis. Blood was obtained by retroorbital bleed. Random serum glucose and triglyceride levels were measured with an autoanalyzer by the National Institutes of Health Clinical Chemistry Laboratory. Serum insulin, leptin, glucagon, and adiponectin, and free T4 were measured by radioimmunoassays (Linco Research, St. Charles, MO), and free fatty acids were measured by using a free fatty acid kit (Roche Diagnostics). TSH was measured by using a rat TSH assay kit (Amersham Pharmacia). Liver triglycerides were measured by solvent extraction followed by a radiometric assay for glycerol (24). For histology, tissues were fixed in 10% neutral formalin and paraffin-embedded, and sections were stained with hematoxylin/eosin. Urine was collected as described (17), and norepinephrine was measured by HPLC (25).

Glucose and Insulin Tolerance Tests. i.p. glucose and insulin tolerance tests were performed in overnight-fasted male mice after injection of glucose (2 mg/g) or insulin (Humulin, 0.75 milli-units/g). Blood glucose values were obtained from tail-vein bleeds by using a Glucometer Elite (Bayer, Elkhart, IN).

Statistical Analysis. Data are expressed as mean ± SEM. Statistical significance among groups was determined by using a paired or unpaired Student t test or multifactor ANOVA, with differences considered significant at P < 0.05.

Results

Generation of Mice with a E1 Deletion (E1^-). To study mice with specific disruption of G_sα, we generated mice in deletion of E1 (Fig. 1B). Correct targeting and subsequent cre-mediated recombination were confirmed by Southern analysis (Fig. 1 B and C and data not shown), and loss of G_sα expression was confirmed by Northern analysis of liver and brown adipose tissue (BAT) RNA from E1^+/p- and E1^m-/+ mice, respectively (Fig. 1D).

As with E2^- mice (2), matings between E1^- heterozygotes failed to produce homozygous E1^-/- offspring, confirming that total G_sα deficiency is embryonically lethal. E1^m-/+ mice had s.c. edema at birth, which resolved within 1-2 days, a feature also present in E2^m-/+ mice (2). E1^m-/+ mice had a 3-week expected survival rate of 51% (27 E1^m-/+ vs. 53 E1^+/+), a higher survival rate than that observed in E2^m-/+ mice (2). Moreover, most E2^m-/+ mice die between 1 and 2 weeks after birth (2), whereas the survival loss of E1^m-/+ mice occurred within 3 days after birth.

In contrast to both E2^+/p- (2) and XLαs knockout (9) mice, E1^+/p- mice had normal survival (103 E1^+/p- vs. 113 E1^+/+ mice at 3 weeks) and no obvious preweaning phenotype. Moreover, the phenotype observed in adult E2^+/p- (2) was absent in adult E1^+/p- mice (see below). The similarity between E2^+/p- and XLαs knockout phenotypes and the absence of this phenotype in E1^+/p- mice confirm that the E2^+/p- phenotype is a direct result of XLαs, rather than G_sα, deficiency.

E1^+/p- and E2^+/p- Mice Have Opposite Metabolic Phenotypes. We next examined whether there were differences in the metabolic phenotype of adult E1^+/p- and E2^+/p- mice. In contrast to E2^+/p- mice (17), E1^+/p- mice had normal body weight (Fig. 2A) and increased fat mass based on NMR analysis and increased BAT and white adipose tissue (WAT) pad weights (Fig. 2 B and C). Interscapular fat pads of E1^+/p- mice showed greater lipid accumulation in both WAT and BAT (Fig. 2D). E1^+/p- mice had normal liver, heart, and spleen weights and a slight decrease in kidney weights (Fig. 2B). Consistent with the fact that E1^+/p- mice have normal body weight, there were no differences in food intake or resting and total metabolic rates between E1^+/p- mice and wild-type littermates (Fig. 3 A and B). There were also no differences in total or ambulatory activity levels (Fig. 3D) or in urinary norepinephrine excretion (E1^+/p- 389 ± 31 vs. wild-type 335 ± 31 nmol/nmol creatinine; n = 5 per group). These results are in contrast to those in E2^+/p- mice, in which metabolic rate, activity, and urinary norepinephrine were all increased (17). E1^+/p- mice had a small decrease in the respiratory exchange quotient at 22°C (Fig. 3C).

Fig. 2.

Body and organ weights and body composition. (A) Body weights of E1^+/p- (Left) and E1^m-/+ (Right) mice and their respective wild-type littermates (mutants, filled bars; wild type, open bars) (n = 6 per group). (B) Organ weights in E1^+/p- mice and wild-type littermates expressed as percent body weight (n = 6-9 per group). BAT, interscapular BAT; WAT, epidydimal WAT. (C) Total, fat, and lean mass of E1^+/p- (Left) and E1^m-/+ (Right) mice and their respective wild-type littermates (n = 6 per group). (D) Representative sections (original magnification, ×100) of interscapular fat pads from E1^+/p- (Left), wild-type (+/+, Center), and E1^m-/+ (Right) mice. In each section, BAT is above and WAT is below. ^*, P < 0.05 vs. wild type.

Fig. 3.

Food intake, metabolic rate, and activity. (A) Food intake in E1^+/p-(Left) and E1^m-/+ (Right) mice (n = 4-6 per group). (B) Resting and total metabolic rate (oxygen consumption) over 24 h at the indicated temperatures in E1^+/p- (Upper) and E1^m-/+ (Lower) mice. (C) Respiratory exchange ratio (RER) measured over 24 h at the indicated temperatures in E1^+/p- (Left) and E1^m-/+ (Right) mice (n = 6 per group). (D) Total and ambulatory (Amb) activity levels at ambient temperature in E1^+/p- (Left) and E1^m-/+ (Right) mice (n = 5-6 per (group). Mutants and their respective wild-type littermates are shown as filled and open bars, respectively. ^*, P < 0.05 vs. wild type.

Random-fed E1^+/p- had normal serum glucose and elevated serum insulin levels (Table 1). Glucose and insulin tolerance tests showed E1^+/p- mice to be both glucose-intolerant and insulin-resistant (Fig. 4). This finding contrasts with E2^+/p- mice, which have increased glucose tolerance and insulin sensitivity (18, 19). Random glucagon levels were unaffected in E1^+/p- mice (E1^+/p- 154 ± 46 vs. wild-type 99 ± 16 pg/ml; n = 5 per group). E1^+/p- mice also showed changes in lipid metabolism that were opposite those found in E2^+/p- mice. Serum leptin levels were very low in E2^+/p- mice (17) but were significantly elevated in E1^+/p- mice (Table 1). Because leptin levels correlate with adiposity, increased leptin levels are consistent with the increased fat mass of E1^+/p- mice. E1^+/p- mice tended to have increased serum (Table 1) and liver triglyceride levels (E1^+/p- 13.0 ± 2.2 vs. wild-type 9.2 ± 2.1 μmol/g; n = 8 per group) and had very reduced triglyceride clearance after an intragastric lipid bolus (Fig. 5, which is published as supporting information on the PNAS web site). Overall triglyceride metabolism in E1^+/p- mice showed changes opposite those in E2^+/p- mice, which have reduced serum and liver triglyceride levels and increased triglyceride clearance (17, 19). Serum free fatty acid and adiponectin levels were unaffected in E1^+/p- mice (Table 1).

Table 1. Serum chemistries in E1^+/p- and E1^m-/+ mice and wild-type littermates.

	E1+/p- controls	E1+/p-	E1m-/+ controls	E1m-/+
Glucose, mg/dl	180 ± 9	165 ± 5	156 ± 18	209 ± 21†
Insulin, ng/ml	2.1 ± 0.3	5.7 ± 1.9*	3.2 ± 1.6	44.9 ± 20.7†
Free fatty acids, nM	0.410 ± 0.034	0.412 ± 0.032	0.468 ± 0.066	0.675 ± 0.079*†
Triglycerides, mg/dl	173 ± 19	215 ± 31	122 ± 12	200 ± 25*
Leptin, ng/ml	10.6 ± 4.3	16.4 ± 2.2*	14.4 ± 4.7	39.2 ± 6.0*†
Adiponectin, mg/ml	6.30 ± 0.58	6.87 ± 0.53	6.89 ± 1.57	6.97 ± 0.65
Free T4, ng/dl	0.65 ± 0.09	0.62 ± 0.09	0.71 ± 0.16	0.84 ± 0.06
Thyrotropin, ng/ml	1.51 ± 0.04	1.36 ± 0.03	1.40 ± 0.09	1.42 ± 0.08

n = 5-22 for E1^+/p- mice and controls; n = 5-8 for E1^m-/+ mice and controls. ^*, P < 0.05 vs. +/+. †, P < 0.05 vs. E1^+/p-.

Fig. 4.

Glucose and insulin tolerance tests. Blood glucose in mutants (▴) and wild-type littermates (▪) was measured at the indicated time points after glucose (A) or insulin (B) administration (n = 5-6 per group; results for insulin tolerance expressed as percent of baseline). Results for E1^+/p- and E1^m-/+ mice are shown in Left and Right, respectively. The genotype had a significant effect by two-factor ANOVA in all studies.

E1^m-/+ Mice Have a More Severe Metabolic Phenotype than E1^+/p- Mice. Adult E1^m-/+ mice were 20% heavier than either E1^+/p- or wild-type mice (Fig. 2 A), and virtually all of this increase was due to a >2-fold increase in fat mass (Fig. 2C) and increased lipid accumulation in BAT and WAT (Fig. 2D). Consistent with more severe obesity, E1^m-/+ mice had serum leptin levels that were ≈3-fold higher than in wild-type mice and >2-fold higher than in E1^+/p- mice (Table 1). Food intake was unaffected in E1^m-/+ mice (Fig. 3A). However, unlike E1^+/p- mice, E1^m-/+ mice had a significantly lower resting metabolic rate at 23°C and 30°C and total metabolic rate at 30°C (Fig. 3B). Activity levels also tended to be lower in E1^m-/+ mice (Fig. 3D). Glucose and lipid metabolism were also more severely affected in E1^m-/+ mice. Serum glucose, insulin, and free fatty acid levels were significantly higher in E1^m-/+ than in E1^+/p- mice, and triglyceride levels were higher in E1^m-/+ mice than in their wild-type littermates (Table 1). E1^m-/+ mice were glucose-intolerant and had much more severe insulin resistance than E1^+/p- mice, with virtually no hypoglycemic response to insulin (Fig. 4). Serum adiponectin levels were unaffected in E1^m-/+ mice (Table 1). Metabolic differences between E1^m-/+ and E1^+/p- mice are not due to hypothyroidism, because there are no differences in serum TSH and free thyroxine (T4) levels between either mutant group and wild-type mice (Table 1).

Discussion

Because Gnas has multiple gene products, gene knockout models that specifically disrupt single gene products are required to understand the biological role of each one. In this study, we generated mice with specific loss of G_sα through deletion of its specific first exon. Homozygotes are embryonically lethal, confirming that loss of G_sα cannot be compensated by other similar genes, such as the olfactory G protein (26). Like E2^m-/+ mice (2) and Oed mice with a maternal G_sα missense mutation, which presumably disrupts G_sα function (15, 16), E1^m-/+ mice develop s.c. edema at birth, confirming that this manifestation results from loss of maternal G_sα expression (see Table 2 for a summary of Gnas knockout models). That edema is absent in E1^+/p- mice suggests that this manifestation results from severe G_sα deficiency in one or more tissues where G_sα is normally paternally imprinted. Several imprinted genes have been shown to affect placental function. Additional studies are necessary to determine whether G_sα imprinting in placenta could account for the s.c. edema during late gestation observed in several maternal G_sα knockout models.

Table 2.

Summary of Gnas knockout mouse lines

Mouse line (source)	Genetic deficiencies	Phenotypes
E2m-/+ (2, 17, 18)	Maternal Gsα	s.c. edema at birth
	Possibly other maternal products (not NESP55)	Severe preweaning lethality Obesity; increased insulin sensitivity
E2+/p- (2, 17-19)	Paternal Gsα; XLαs	Perinatal lethality
		Very lean, greatly increased insulin sensitivity
Oed (maternal mutation) (15, 16)	Maternal Gsα	s.c. edema at birth
	Possibly other maternal products	Severe preweaning lethality
		Increased BAT mass
Sml (paternal mutation) (15, 16)	Paternal Gsα; XLαs	Similar to E2+/p-
XLαs knockout (9)	XLαs, XLN1, ALEX	Similar to E2+/p-
NESP55 knockout (27)	NESP55	Normal viability and metabolism
E1m-/+ (this study)	Maternal Gsα	s.c. edema at birth
		Moderate preweaning lethality (50%)
		Severe obesity, insulin resistance
E1+/p- (this study)	Paternal Gsα	Obesity, insulin resistance

However, E1^m-/+ and E2^m-/+ phenotypes differ in many respects. E1^m-/+ mice have better survival and more pronounced obesity compared with E2^m-/+ mice and lack the neurological defects often observed in 1- to 3-week-old E2^m-/+ mice (2). Moreover, whereas E2^m-/+ mice have increased insulin sensitivity (18), E1^m-/+ are severely insulin-resistant. The more severe obesity of E1^m-/+ mice likely contributes to their insulin resistance. However, G_sα deficiency is also an important factor, because E1^+/p- mice, which are only mildly obese, are also insulin-resistant. Disruption of other maternal-specific Gnas gene products in E2^m-/+, but not E1^m-/+, mice is the most likely explanation for the phenotypic differences between these two mutant lines. However, the only other known maternal-specific Gnas gene product, NESP55, is not disrupted in E2^m-/+ mice (T.X., data not shown), and NESP55 knockout mice do not have altered viability or metabolism (27). Moreover, loss of NESP55 expression in some pseudohypoparathyroidism type IB patients does not lead to an obvious phenotype (12). Our attempts to identify other maternal Gnas products by 5′ RACE have been unsuccessful. Another partial maternal-specific mRNA transcript has been identified (28), although it is unknown whether this transcript extends into exon 2 or generates a protein product. Although both E1^m-/+ and E2^m-/+ mice were studied in a CD-1 background, we cannot exclude the possibility that genetic background differences contributed to phenotypic differences observed between these two lines.

The absence of early lethality or lean phenotype in E1^+/p- mice confirms that these manifestations in E2^+/p- mice are not caused by G_sα deficiency. XLαs knockout mice show a phenotype very similar to E2^+/p- mice (9), indicating that the effects in the latter result from complete loss of paternally expressed XLαs. Moreover, adult XLαs knockout mice have the same hypermetabolic insulin-sensitive phenotype as adult E2^+/p- mice (T.X., L.S.W., A. Plagge, and G. Kelsey, unpublished results). Therefore, the E2^+/p- phenotype at all stages of life is determined by XLαs deficiency, and XLαs deficiency appears to be dominant over paternal-specific G_sα deficiency. The opposite metabolic phenotypes of E2^+/p- and XLαs knockout mice vs. E1^+/p- and E1^m-/+ mice demonstrate that G_sα and XLαs have opposing effects on glucose and lipid metabolism. Although it has been suggested that G_sα and XLαs may directly counteract each other biochemically (9), this has not been established, and studies examining the role of XLαs in receptor-mediated signaling have provided conflicting results (8, 29). We have proposed that XLαs may normally inhibit sympathetic activity in the central nervous system, and that XLαs deficiency therefore leads to increased sympathetic activity (19). Whatever the mechanism, the role of XLαs in metabolic regulation appears to be species-specific, because pseudopseudohypoparathyroidism patients with paternal GNAS mutations that disrupt XLαs expression do not develop a similar phenotype (1).

The presence of obesity in both E1^m-/+ and E1^+/p- mice provides further evidence that the obesity in AHO patients results from G_sα deficiency. This hypothesis is further supported by the fact that patients with E1 mutations, which disrupt only G_sα expression, have the same phenotype as patients who inherit mutations in downstream exons common to all GNAS transcripts. Because G_sα is ubiquitously expressed and mediates important signals for metabolic regulation in almost all tissues, it is unclear whether the obesity in heterozygous E1^- mice and AHO patients results from one specific metabolic defect or is the cumulative result of subtle metabolic defects in multiple tissues. It is unlikely that obesity in E1^- heterozygotes is solely due to the resistance of fat cells to sympathetic- or hormone-mediated metabolic activation, because E2^m-/+ mice have a normal metabolic response to a β₃ adrenergic agonist, which primarily targets adipose tissue (17), and preliminary observations show that mice heterozygous for an adipose tissue-specific G_sα knockout do not develop significant obesity (M.C. and L.S.W., unpublished results). It is likely that obesity in E1^- mice contributes to their insulin resistance, although direct effects of partial G_sα deficiency on insulin sensitivity cannot be ruled out.

In many respects (s.c. edema at birth, early lethality, more severe obesity, and insulin resistance), E1^m-/+ mice have a more severe phenotype than E1^+/p- mice. These differences are presumably the consequence of differential expression of G_sα in tissues where G_sα is normally preferentially expressed from the maternal allele, leading to tissue-specific G_sα deficiency in E1^m-/+, but not E1^+/p-, mice. This hypothesis is supported by the observation that many features of the E1^m-/+ phenotype are reversed by paternal deletion of the exon 1A region, a region required for tissue-specific paternal G_sα imprinting (30, 31). Although it is possible that maternal effects during gestation could contribute to the E1^m-/+ phenotype, the phenotype of E1^m-/+ offspring is not affected by whether the mutant mother is E1^m-/+ or E1^+/p-.

The greater severity of the metabolic phenotype in E1^m-/+ as compared with E1^+/p- mice likely results from more severe G_sα deficiency in one or more tissues involved in metabolic regulation. Given that metabolic rates are reduced in E1^m-/+, but not E1^+/p-, mice, one hypothetical mechanism is reduced hypothalamic melanocortin-G_sα signaling in E1^m-/+ mice due to the absence of G_sα resulting from maternal mutation and paternal imprinting. However, it is presently unknown whether G_sα is imprinted within the hypothalamus. Although pseudohypoparathyroidism type 1A patients develop TSH resistance, thyroid function is normal in E1^m-/+ mice, and therefore hypothyroidism is not a cause of their reduced metabolic rates. Additional studies in this and tissue-specific G_sα knockout models will provide a greater understanding of the role of the Gnas locus in development and metabolic regulation.