Severe Obesity and Insulin Resistance due to Deletion of the Maternal G_sα Allele Is Reversed by Paternal Deletion of the G_sα Imprint Control Region

Abstract

The G protein α-subunit G_sα mediates receptor-stimulated cAMP production and is imprinted with reduced expression from the paternal allele in specific tissues. Disruption of the G_sα maternal (but not paternal) allele leads to severe obesity, hypertriglyceridemia, and insulin resistance in mice and obesity in patients with Albright hereditary osteodystrophy. Paternal deletion of a G_sα imprint control region (1A) leads to loss of tissue-specific G_sα imprinting. To determine whether the metabolic abnormalities resulting from disruption of the G_sα maternal allele could be reversed by loss of paternal G_sα imprinting, females with a heterozygous G_sα exon 1 deletion were mated to males with heterozygous deletion of the imprint control region (1A) to generate mice with maternal G_sα deletion (E1^m−), paternal 1A deletion (1A^p−), double mutants (E1^m−:1A^p−), and wild type. E1^m− mice developed obesity, glucose intolerance, insulin resistance, and hypertriglyceridemia, which were all normalized by the paternal 1A deletion in E1^m−:1A^p− mice. Obesity in E1^m− was associated with reduced energy expenditure and sympathetic nerve activity, and these were also normalized in E1^m−:1A^p− mice. 1A^p− mice had reduced body weight associated with proportional decreases in fat and lean mass as well as increased activity levels. The metabolic phenotype resulting from maternal G_sα deletion is rescued by a genetic lesion that leads to loss of tissue-specific G_sα imprinting, consistent with this phenotype being a direct consequence of G_sα imprinting in one or more specific tissues.

THE INCIDENCE OF metabolic syndrome, characterized by obesity, insulin resistance, and diabetes, has dramatically increased over the past 2 decades. Studies of monogenic obesity disorders provide important insights into genes involved in the development of the metabolic syndrome and the mechanisms by which they impact energy and glucose metabolism (1). One such disease is Albright hereditary osteodystrophy (AHO), a disorder characterized by short stature and skeletal and neurological abnormalities (2). The underlying genetic defect is heterozygous loss-of-function mutation of the ubiquitously expressed G protein α-subunit G_sα, which is required for receptor-stimulated cAMP generation. In addition to these clinical features, patients who inherit G_sα mutations from their mother also develop resistance to several hormones, such as PTH and thyroid-stimulating hormone, and obesity (referred to as pseudohypoparathyroidism type 1A). Similarly, mice with disruption of the G_sα maternal allele develop severe obesity, insulin resistance, and hypertriglyceridemia, whereas the same mutation on the paternal allele leads to only very mild obesity and insulin resistance (3).

Studies have shown that these parent-of-origin effects of G_sα mutations on hormone resistance are due to the fact that G_sα is imprinted in a tissue-specific manner, being expressed primarily from the maternal allele (suppressed on the paternal allele) in hormone-target tissues, such as the renal proximal tubules, thyroid, gonads, and pituitary somatotrophs (4,5,6,7,8). In fact, G_sα is just one product of the complex imprinted gene GNAS on human chromosome 20q13 (2,9). In addition to G_sα, this gene has alternative upstream promoters for two other gene products, the chromogranin-like protein NESP55 and the alternative G_sα isoform XLαs, which are expressed only from the maternal and paternal alleles, respectively (Fig. 1A). Transcript-specific knockouts of the mouse ortholog Gnas have helped to define the roles of each gene product in development and metabolic regulation (9). Heterozygous deletion of G_sα exon 1 (E1⁻) on the maternal allele leads to reduced postnatal survival, severe perinatal sc edema, severe obesity, insulin resistance, and hypertriglyceridemia, whereas the same deletion on the paternal allele leads to normal survival, no sc edema, and only a mild increase in adiposity and insulin resistance (3,10). Another model in which maternal G_sα was disrupted in exon 2 also led to obesity and reduced energy expenditure, although glucose metabolism was not similarly affected (11). XLαs deficiency leads to a lean and insulin-sensitive phenotype with a defect in neonatal suckling in mice (12,13) whereas NESP55 has no major role in metabolic regulation in either mice or humans (14,15). The role of XLαs in humans is less clear (9).

Figure 1.

Organization and imprinting of the Gnas locus and cross-mating of mice with different Gnas mutations. A, The maternal and paternal alleles of the Gnas locus are depicted showing four alternative first exons for NESP55 (NESP), XLαs, 1A mRNA transcripts, and G_sα (exon 1) splicing onto common exons 2–13 (shown as a single box). Regions of differential methylation (METH) are shown above and splicing patterns are shown below each allele. Active promoters are shown in white with horizontal arrows, whereas inactive promoters are shown in gray. The thin and dashed arrow for the paternal G_sα promoter indicates that this promoter is suppressed in a tissue-specific manner due to genomic imprinting. B, Female mice with paternal deletion of G_sα exon 1 (E1^p−) were mated to males with a paternal deletion of the 1A DMR (1A^p−) to generate wild-type (+/+), E1^m− (maternal deletion of G_sα exon 1), 1A^p−, and double-mutant (E1^m−:1A^p−) mice. Below each genotype is a schematic showing deletions with an X and indicating the expected effect on G_sα expression. The survival rate at weaning expressed as both total number (n = 370 total offspring) and percent of wild-type as well as the presence or absence of sc edema at birth are indicated for each group of mice.

Tissue-specific suppression of G_sα expression from the paternal allele is not associated with methylation of the G_sα promoter (16,17). Rather, tissue-specific G_sα imprinting is dependent on a differentially methylated region (1A DMR) located 2 kb upstream of the G_sα promoter (15,16) (Fig. 1A). This region is an imprinting control region for G_sα as paternal deletion of this region leads to loss of tissue-specific G_sα imprinting (18,19). This region contains an alternative first exon (exon 1A) that is expressed only from the paternal allele and that splices onto G_sα exon 2 to produce an mRNA that does not encode a fully functional protein (15,16). The role, if any, of this transcript is unknown, but it is unlikely to be involved in G_sα imprinting because it is ubiquitously expressed in a similar tissue distribution as that of G_sα transcripts (16). Rather, we speculated that the region contains one or more cis-acting negative regulatory elements for the G_sα promoter that acts in a methylation-sensitive and tissue-specific manner (18).

It is presumed that differences in the phenotypes between mice with maternal vs. paternal E1⁻ mutations result from differences in G_sα expression in tissues in which G_sα is normally imprinted, leading to much more severe G_sα deficiency in maternal E1⁻ mice (E1^m−). If this is correct, the presence of a 1A DMR deletion on the paternal allele (1A^p−) should rescue the phenotype due to loss of tissue-specific G_sα imprinting. In this study we generated E1^m−, 1A^p−, and double mutant (E1^m−:1A^p−) mice and show that the E1^m− metabolic phenotype is completely corrected by the presence of the paternal 1A DMR deletion, providing further evidence that the metabolic phenotype is a direct consequence of G_sα imprinting in one or more specific tissues. In addition, we show the 1A^p− mutation alone has effects on growth and activity levels.

Materials and Methods

Animals

To generate mice, female E1^p− and male 1A^p− mice (3,18) with deletion of G_sα exon 1 or the 1A DMR on the paternal allele, respectively, were mated. All mice were on a CD-1 genetic background and maintained on standard pellet diet (NIH-07, 5% fat by weight) and 12-h light, 12-h dark cycle. Except when noted, all experiments were performed on 12- to 14-wk-old male mice. Experiments were approved by the National Institute of Diabetes and Digestive and Kidney Diseases Animal Care and Use Committee and conducted in accord with accepted standards of humane animal care, as outlined in the ethical guidelines.

Body composition, food intake, metabolic rate, and activity measurements

Body composition was measured using the Echo MRI3-in-1 (Echo Medical Systems, Houston, TX). Food intake, metabolic rates (oxygen consumption rate by indirect calorimetry), and activity levels were determined as previously described (11).

Histology

Dissected samples were fixed overnight in 4% paraformaldehyde, embedded in paraffin, cut, and stained with hematoxylin and eosin.

Blood and urine chemistries

Blood was obtained by retroorbital bleed. Serum glucose, cholesterol, and triglyceride levels were measured by autoanalyzer at the National Institutes of Health Clinical Chemistry Laboratory. Serum insulin, leptin, and free T₄ were measured by ELISA (Millipore, Billerica, MA). TSH and IGF-I were measured by RIA (Alpco Diagnostics, Windham, NH). Urine was collected by bladder puncture and catecholamines were measured by HPLC (20) and corrected by creatinine concentration within the same samples, which were measured by Ani-Lytics, Inc. (Gaithersburg, MD).

Glucose and insulin tolerance tests

Glucose and insulin tolerance tests were performed in overnight fasted mice after ip injection of glucose (2 mg/g) or insulin (Humulin, 0.50 mIU/g), respectively. Blood glucose levels in tail vein bleeds were measured using a Glucometer Elite (Bayer, Elkhart, IN) at indicated times before and after injection.

Quantitative RT-PCR

RNA was extracted from brown adipose tissue (BAT) using TRIzol (Invitrogen, Carlsbad, CA) and treated with DNase (DNA-free; Ambion, Austin, TX) to remove DNA contamination. Reverse transcription was performed using the SuperScript first-strand synthesis system (Invitrogen). Gene expression levels were measured by quantitative RT-PCR using a real-time PCR machine (MxP3000; Stratagene, La Jolla, CA). PCRs (20 μl total volume) included cDNA, 100 nm primers, and 10 μl of SYBR Green master mix (Applied Biosystems, Foster City, CA). To get relative quantification, standard curves were simultaneously generated with serial dilutions of cDNA, and results were normalized to β-actin mRNA levels in each sample, which were determined simultaneously by the same method. Specificity of each RT-PCR product was indicated by its dissociation curve and the presence of a single band of expected size on acrylamide gel electrophoresis. Primer sequences used for each gene have been previously published (13).

Statistical analysis

Data are expressed as mean ± sem. Statistical significance between groups was determined using one- or two-factor ANOVA with Bonferroni posttest analysis or unpaired t test with differences considered significant at P < 0.05.

Results

Paternal 1A deletion reverses perinatal sc edema and lethality due to maternal G_sα disruption

Females heterozygous for a deletion of G_sα exon 1 on the paternal allele (E1^p−) were mated with males heterozygous for a deletion of the 1A DMR on the paternal allele (1A^p−) to generate four groups of offspring: wild-type (+/+) mice, mice with disruption of G_sα expression on the maternal allele (E1^m−), mice with deletion of the 1A imprinting control region on the paternal allele (1A^p−), and mice with both maternal disruption of G_sα and paternal deletion of 1A (E1^m−:1A^p−) (Fig. 1B). E1^p− as opposed to E1^m− mothers were used in the study to minimize any potential maternal gestational effects because these mice have a much less severe phenotype (3). Genetic analysis of 370 total offspring that survived to weaning showed that E1^m− had only 56% survival to weaning, compared with +/+ mice, similar to previously published results (3). 1A^p− mice had a very small reduction in survival (92% of +/+ mice), whereas the paternal 1A⁻ deletion partially rescued the lethality resulting from maternal loss of G_sα expression with E1^m−:1A^p− mice having an 80% survival rate compared with +/+ mice. All E1^m− offspring had severe sc edema at birth despite the fact that their mothers who were E1^p− never developed perinatal edema. In addition, when female E1^m−:1A^p− mice, which had a normal phenotype, were mated with wild-type males, the E1^m− offspring all had perinatal sc edema, confirming that this phenotype is due to loss of the maternal G_sα allele in the offspring, rather than a maternal effect. In contrast, the edema was absent in all E1^m−:1A^p− pups, similar to previous reports, suggesting that the edema in E1^m− mice was a consequence of G_sα imprinting (3,18,19).

Paternal 1A deletion reverses the obesity of E1^m− mice

Both male and female E1^m− mice gained significantly more weight than +/+, E1^m−:1A^p−, and 1A^p− mice over the first 20 wk of life (Fig. 2A), similar to previously published results (3). In contrast, both male and female 1A^p− mice gained less weight than +/+ mice, although this difference was significant only in males. The presence of the paternal 1A deletion reversed the increased weight gain resulting from the E1^m− mutation, with both male and female E1^m−:1A^p− mice having weight curves between those of +/+ and 1A^p− mice. These weight differences were not associated with differences in body length (Fig. 2B). Consistent with the lack of effects on length, there were also no significant differences in serum IGF-I levels between the groups (Fig. 2C). In males the body mass index [BMI; weight in grams/(nasoanal length in centimeters) squared] was significantly greater in E1^m− than +/+ mice, and this increase was reversed in E1^m−:1A^p− mice (Fig. 2D). BMI was minimally reduced in 1A^p− mice, compared with +/+ mice.

Figure 2.

Effects of genotype on growth and body composition. A, Growth curves for male (left) and female (right) mice for each genotype (n = 2–14 animals/measurement). Two-way ANOVA with Bonferroni’s posttest analysis showed that by 14 wk both male and female E1^m− mice had significantly greater growth than the other three groups, and male 1A^p− mice had significantly less growth than +/+ mice. Nasoanal body length (B), serum IGF-I levels (C), and BMI [weight in grams/(nasoanal length in cm) squared] (D) in adult male mice (n = 6–18/group). E, Fat mass expressed as percent of total body weight in male (left) and female (right) adult mice (n = 6–14/group). F, Weight (as percent body weight) of interscapular BAT (left) and parametrial (Abd; right) fat pads in adult females (n = 5–7/group). G, Histological sections of interscapular BAT obtained from adult mice and stained with hematoxylin and eosin. H, Serum leptin levels in nonfasted adult male mice (n = 7–11/group). *, P < 0.05 vs. +/+; ^, P < 0.05 vs. E1^m−; #, P < 0.05 vs. E1^m−:1A^p−.

Body composition analysis by nuclear magnetic resonance confirmed that both male and female E1^m− mice have increased fat mass, as previously reported (3), and that this increase is reversed in E1^m−:1A^p− mice (Fig. 2E). The reduced body weights observed in male 1A^p− mice (Fig. 2A) were due to similar reductions in both fat and lean mass (Fig. 2E and data not shown). Consistent with the body composition data,. female E1^m− mice had increased abdominal (parametrial) white adipose tissue and interscapular BAT pad masses (as percent body weight) compared with +/+ mice, whereas the white adipose tissue and BAT fat pad masses of E1^m−:1A^p− and 1A^p− mice were similar to +/+ mice (Fig. 2F). Histology of interscapular BAT showed E1^m− mice to have larger adipocytes with increased lipid accumulation per cell compared with the other three groups (Fig. 2G). Finally, serum leptin levels, which reflect overall adiposity, were almost 3-fold higher in E1^m− mice compared with +/+, E1^m−:1A^p−, and 1A^p− mice (Fig. 2H). Overall these findings show that the obesity in E1^m− mice is reversed by the presence of paternal 1A⁻ deletion in E1^m−:1A^p− mice and that the 1A^p− mutation has a small independent effect on body weight but a minimal effect on adiposity. Differences in adiposity were not due to differences in thyroid hormone status, as serum free T4 and TSH levels were similar in the four groups (data not shown).

Paternal 1A deletion reverses the reduced energy expenditure of E1^m− mice

We next examined food intake, energy expenditure, and activity levels in our four groups of mice to determine which of these parameters are impacting the observed differences in adiposity. We found no differences in absolute food intake per day among the four groups despite the increased body weight of E1^m− mice. When food intake was normalized to body weight E1^m− actually ate less despite their increased adiposity, and this difference was reversed by the presence of the paternal 1A⁻ deletion in E1^m−:1A^p− mice, with no independent effect of the 1A^p− mutation alone (Fig. 3, A and B).

Figure 3.

Effects of genotype on energy balance and activity levels. A, Total food intake (grams per day) in adult male mice (n = 5–9/group). B, Food intake data normalized by body weight. C, Resting and total energy expenditure rates normalized to body weights (vO₂ in milliliters per gram^0.75 per hour) at 23 and 30 C in adult male mice (n = 6–8/group). D, Total and ambulatory activity levels in male mice at 23 and 30 C (n = 6–8/group). E, Urine DHPG, NE, and EPI levels normalized to creatinine concentration (n = 7–13/group). F, Relative expression of four genes involved in thermogenesis and mitochondrial function in interscapular BAT (n = 4–6/group). *, P < 0.05 vs. +/+; ^, P < 0.05 vs. E1^m−.

Similar to previously published results (3), E1^m− had reduced resting energy expenditure (O₂ consumption) at both room (23 C) and thermoneutral (30 C) temperatures, compared with +/+ mice (Fig. 3C). This difference was reversed by the presence of the paternal 1A⁻ allele in E1^m−:1A^p− mice. In fact, resting energy expenditure at 23 C was slightly greater in 1A^p− mice compared with +/+ mice. A similar pattern was observed for total energy expenditure, except for the fact that total energy expenditure at 23 C was clearly elevated in E1^m−:1A^p− and 1A^p− mice compared with +/+ mice. This increase in total energy expenditure in these two groups of mice is likely explained by the fact that they have markedly increased ambulatory and total activity levels (Fig. 3D). Similar to our previous study (3), we observed no differences in activity levels between E1^m− and +/+ mice.

Previous studies have suggested that disruption of the maternal G_sα allele leads to reduced energy expenditure by a decrease in sympathetic nervous system activity (11). To examine this, we measured urine catecholamine levels, which were normalized to creatinine concentration to correct for differences in urine concentration and lean body mass (Fig. 3E). There were no differences in serum creatinine among the four groups (data not shown). Compared with +/+ mice, urine levels of norepinephrine (NE) and its metabolite dihydroxyphenylglycol (DHPG) were reduced in E1^m− (P = 0.07 for NE), consistent with these mice having reduced sympathetic nervous system activity as a likely contributor to their lower-than-normal energy expenditure rates. Epinephrine (EPI), which is primarily secreted from the adrenal medulla, also tended to be lower in E1^m− mice. In contrast, urine DHPG, NE, and EPI concentrations in E1^m−:1A^p− and 1A^p− mice were all similar to those in +/+ mice. This correlates with the differences in energy expenditure observed between groups and demonstrates that the presence of the paternal 1A⁻ deletion reverses both the reduced sympathetic activity and energy expenditure levels that result from disruption of G_sα on the maternal allele.

We next examined the expression of BAT genes involved in energy metabolism that are directly induced by sympathetic nerve stimulation (Fig. 3F). Peroxisome proliferator-activated receptor-γ coactivator (PGC)-1α is a cAMP-inducible gene that activates genes required for mitochondrial function and thermogenesis (21). Due to great variability in the samples, we were unable to detect differences in BAT PGC1α expression among the groups, although PGC1α expression tended to be lower in the E1^m− samples. However, expression of the thermogenic uncoupling protein (Ucp1) gene and mitochondrial transcription factor A (Tfam), a gene involved in mitochondrial function, were both reduced in BAT from E1^m− mice, consistent with their reduced energy expenditure rates and sympathetic activity. Nuclear regulatory factor 1 (Nrf1), another gene involved in mitochondrial function, tended to be lower in E1^m− mice compared with +/+ mice, and was significantly lower in E1^m− mice than in E1^m−:1A^p− mice. In contrast, expression of these three genes in BAT from E1^m−:1A^p− or 1A^p− mice was similar to that in +/+ mice, consistent with the normal levels of energy expenditure and sympathetic activity observed in these two groups of mice. Overall, our results demonstrate that the 1A^p− mutation reverses the reduced energy expenditure and sympathetic nervous system activity observed in E1^m− mice and has little or no independent effect on these parameters. Moreover, the presence of the paternal 1A⁻ deletion is associated with increased activity levels.

Paternal 1A deletion reverses abnormal glucose and lipid metabolism of E1^m− mice

We observed no differences in serum cholesterol levels among the four groups of mice (Fig. 4A). Similar to previously published results (3), serum triglyceride levels were elevated in E1^m− mice, compared with +/+ mice (Fig. 4B). Hypertriglyceridemia resulting from the E1^m− mutation was reversed by the presence of the paternal 1A⁻ deletion in E1^m−:1A^p− mice, whereas the presence of the 1A^p− mutation alone had no independent effect on triglyceride levels.

Figure 4.

Effects of genotype on glucose and lipid metabolism. Serum cholesterol (n = 6–9) (A), triglyceride (n = 6–8) (B), glucose (n = 6–9) (C), and insulin (n = 17–21/group) (D) in nonfasted adult male mice are shown. Intraperitoneal glucose tolerance tests (n = 6–10/group) (E), and insulin tolerance tests (n = 4–10/group) (F) were performed in overnight fasted adult male mice. For both glucose and insulin tolerance tests, E1^m− curves were significantly different from the other three genotypes based on two-way ANOVA. *, P < 0.05 vs. +/+; ^, P < 0.05 vs. E1^m−.

Nonfasted serum glucose levels were similar among the four groups of mice (Fig. 4C), but serum insulin levels were almost 2-fold higher in E1^m− mice compared with +/+ mice (Fig. 4D), similar to prior published results (3) and consistent with the presence of insulin resistance in these mice. The hyperinsulinemia observed in E1^m− mice was reversed by the presence of the paternal 1A⁻ deletion in E1^m−:1A^p− mice. Insulin levels tended to be even lower in 1A^p− mice, but the differences were not significant. Glucose and insulin tolerance tests were performed to examine glucose disposal and insulin sensitivity (Fig. 4, E and F). Similar to previous published results (3), E1^m− had impaired glucose tolerance, compared with +/+ mice, indicated by persistently higher glucose levels after an ip glucose load. Both E1^m−:1A^p− and 1A^p− mice had glucose curves that were similar to those of +/+ mice, indicating that the paternal 1A⁻ deletion can reverse the glucose intolerance resulting from disruption of the maternal G_sα allele and that the 1A^p− deletion has no independent effect on glucose tolerance. Likewise, E1^m− mice had almost no hypoglycemic response to ip insulin, similar to prior published results (3) and consistent with the presence of insulin resistance. In contrast, the hypoglycemic response of both E1^m−:1A^p− and 1A^p− mice was similar to that of +/+ mice, consistent with the ability of the 1A^p− mutation to reverse the insulin resistance caused by the E1^m− mutation and the lack of an independent effect of the 1A^p− mutation on insulin sensitivity.

Discussion

Observations in both AHO patients and mice show that mutations which disrupt G_sα expression from the maternal allele lead to more severe phenotypic consequences than those on the paternal allele (9). These parent-of-origin effects presumably are the result of tissue-specific G_sα imprinting with predominant expression from the maternal allele in certain tissues. In these tissues mutation of the active maternal allele leads to severe G_sα deficiency, whereas mutation of the relatively inactive paternal allele has little effect on G_sα expression. This hypothesis predicts that a perturbation that reverses G_sα imprinting would reverse the phenotypic features that are specifically associated with maternal G_sα mutations.

The initial clue that the 1A DMR is an imprint control center for G_sα was the observation that loss of 1A DMR methylation on the maternal allele leads to pseudohypoparathyroidism type 1B, an isolated form of PTH resistance (15). Based on this observation, we proposed a model in which the 1A DMR contains one or more cis-acting regulatory elements that suppress the G_sα promoter in a tissue-specific manner and that are normally active only on the paternal allele because the elements are methylated on the maternal allele. Support for this model came from studies that showed that G_sα imprinting in renal proximal tubules and neonatal BAT is reversed in mice with paternal deletion of the 1A DMR (18,19).

In this study we determined whether a paternal 1A DMR deletion that reverses G_sα imprinting could reverse the phenotype resulting from mutation of the maternal G_sα allele, with particular focus on the metabolic phenotype. In this regard we showed that paternal 1A deletion improved the postnatal survival rate of E1^m− mice. Moreover, we confirmed in a large sample size the previous observations that paternal 1A DMR deletion reverses the perinatal sc edema observed in E1^m− pups (18,19). This edema is unlikely to result from an embryonic kidney defect because it was shown in a similar model to develop on embryonic day 11.5, 5 d before embryonic kidneys begin to function (22). Considering that the edema occurs during late gestation and reverses quickly after parturition, it may be the result of placental dysfunction due to severe G_sα deficiency in the fetal portion of the placenta resulting from the combined effects of maternal G_sα mutation and imprinting of the paternal allele. We have preliminary evidence that in fact G_sα is imprinted in the placenta (Wang, J., M. Chen, L. S. Weinstein, unpublished results), similar to other imprinted genes (23).

Strikingly, paternal 1A DMR deletion also completely reversed the severe obesity, insulin resistance, and hypertriglyceridemia observed in E1^m− mice. This suggests that the metabolic syndrome in E1^m− mice results from loss of G_sα expression in one or more metabolically active tissues in which it is normally imprinted. Liver and muscle are unlikely to be sites for these maternal-specific G_sα effects because G_sα is not imprinted in these tissues (6,11,24), and liver- and muscle-specific G_sα knockout mice do not develop these metabolic features (25) (Chen, M., L. S. Weinstein, unpublished results). Whereas G_sα was initially reported to be imprinted in adipose tissue (6), subsequent reports show no evidence for G_sα imprinting in either human and mouse adipose tissue, except in mouse neonatal BAT (3,9,10,19,26,27). In any case, mice with adipose-specific G_sα deficiency do not develop the metabolic features observed in E1^m− mice (28). G_sα deficiency in β-cells also does not lead to obesity or insulin resistance (29).

A likely explanation for the severe metabolic phenotype resulting from maternal G_sα mutation is loss of G_sα expression due to G_sα imprinting in one or more regions of the central nervous system. G_sα mediates melanocortin signaling in the central nervous system, which inhibits food intake and increases sympathetic nervous system activity, energy expenditure, and insulin sensitivity (30,31,32,33). The E1^m− and similar models are characterized by obesity with reduced sympathetic activity and energy expenditure (3,11) (this study), and these are reversed by the paternal 1A DMR deletion. One potential explanation for these observations is that G_sα is normally imprinted in one or more regions that are specifically involved in regulation of sympathetic outflow. The changes in BAT histology and gene expression observed in E1^m− but not E1^m−:1A^p− mice probably result from reduced sympathetic nervous system activity in E1^m− mice.

Mice with only the paternal 1A DMR deletion (1A^p−) might be predicted to have some phenotype due to G_sα overexpression in specific tissues (Fig. 1B). In fact, 1A^p− (and to some extent E1^m−:1A^p−) mice had poor growth and increased activity levels. The growth defect was not associated with reduced length or adiposity or other associated metabolic changes, such as changes in serum leptin levels, food intake, resting energy expenditure, sympathetic activity, BAT histology, or gene expression. Rather, there was a proportional decrease in both fat and lean mass of 1A^p− mice by mechanisms that are not entirely clear. One contributing factor may be increased activity levels leading to increased total energy expenditure. Increased activity levels were also observed in mice with disruption of exon 2 on the paternal allele, which would also disrupt 1A mRNA transcripts (11) but not in mice with paternal disruption of either XLαs or G_sα exon 1 (3,13). Although 1A mRNA transcripts are considered to be noncoding, based on these observations, we cannot rule out the possibility that these transcripts play a physiological role in the regulation of locomotor activity.

In conclusion, we have demonstrated in this study that the severe metabolic features due to disruption of the maternal G_sα allele are completely reversed by a second mutation that reversed G_sα imprinting, providing further evidence that these metabolic effects result from severe tissue-specific G_sα deficiency as a result of the combined effects of maternal mutation and tissue-specific paternal imprinting. Similar mechanisms likely underlie the parent-of-origin metabolic effects observed in AHO patients. The reciprocal metabolic effects resulting from maternal vs. paternal mutation of Gnas is consistent with the parental-conflict theory for genomic imprinting, which predicts that maternally expressed genes suppress growth, whereas paternally expressed genes promote growth (34). This study also provides evidence that 1A-specific mRNA transcripts may have independent effects on growth and locomotor activity.