The Alternative Stimulatory G Protein α-Subunit XLαs Is a Critical Regulator of Energy and Glucose Metabolism and Sympathetic Nerve Activity in Adult Mice*

The complex imprinted Gnas locus encodes several gene products including Gsα, the ubiquitously expressed G protein α-subunit required for receptor-stimulated cAMP generation, and the neuroendocrine-specific Gsα isoform XLαs. XLαs is only expressed from the paternal allele, whereas Gsα is biallelically expressed in most tissues. XLαs knock-out mice (Gnasxlm+/p–) have poor suckling and perinatal lethality, implicating XLαs as critical for postnatal feeding. We have now examined the metabolic phenotype of adult Gnasxlm+/p– mice. Gnasxlm+/p– mice had reduced fat mass and lipid accumulation in adipose tissue, with increased food intake and metabolic rates. Gene expression profiling was consistent with increased lipid metabolism in adipose tissue. These changes likely result from increased sympathetic nervous system activity rather than adipose cell-autonomous effects, as we found that XLαs is not normally expressed in adult adipose tissue, and Gnasxlm+/p– mice had increased urinary norepinephrine levels but not increased metabolic responsiveness to a β3-adrenergic agonist. Gnasxlm+/p– mice were hypolipidemic and had increased glucose tolerance and insulin sensitivity. The similar metabolic profile observed in some prior paternal Gnas knock-out models results from XLαs deficiency (or deficiency of the related alternative truncated protein XLN1). XLαs (or XLN1) is a negative regulator of sympathetic nervous system activity in mice.

The incidence of both obesity and diabetes has markedly increased over the past two decades. A basic understanding of the regulation of energy and glucose metabolism is required for the identification of potential new therapeutic targets. One gene shown to be important for metabolic regulation is GNAS on chromosome 20q13 in humans (Gnas on chromosome 2 in mice) (1,2). GNAS/Gnas encodes several protein products, including the ubiquitously expressed G protein ␣-subunit G s ␣ 2 that couples receptors to cAMP generation, the alternative G s ␣ isoform XL␣s, and the 55-kDa neuroendrocrine-specific secretory protein (NESP55). Heterozygous mutations resulting in partial G s ␣ deficiency lead to obesity in mice (3) and in patients with Albright hereditary osteodystrophy (1,2).
GNAS/Gnas is a complex imprinted gene whose major gene products are generated by alternative promoters and first exons that splice onto a common exon (exon 2) (1, 2) (Fig. 1). NESP55 and XL␣s are expressed from oppositely imprinted Nesp and Gnasxl promoters; NESP55 is only expressed from the maternal allele, whereas XL␣s is only expressed from the paternal allele. In contrast, the G s ␣ promoter is imprinted in a tissue-specific manner, with G s ␣ being biallelically expressed in most tissues but primarily expressed from the maternal allele in a small number of tissues.
XL␣s is structurally identical to G s ␣ except for a long aminoterminal extension encoded within its specific first exon (4,5). In contrast to G s ␣, which is ubiquitously expressed at high levels in virtually all tissues, XL␣s has a more restricted tissue distribution with expression primarily limited to neuroendocrine tissues (4,6). XL␣s has also been shown to be expressed in adipose tissue in 4-day-old mice (7). In vitro biochemical studies have shown that XL␣s is capable of stimulating adenylyl cyclase (8,9) and mediating receptor-stimulated cAMP production (9), although its role in vivo is less clear. The XL␣s promoter also generates a neural-specific truncated form of XL␣s (XLN1) produced by splicing of exon 3 onto an alternative terminal exon (6). Whether the resultant truncated protein has any biological function or whether it can act as a dominant negative inhibitor of G s ␣ or XL␣s signaling is unknown. NESP55 does not appear to be an important metabolic regulator in either humans (10) or mice (11).
We originally generated a Gnas knock-out in which exon 2, an exon common to all transcripts, was disrupted (12). Het-erozygotes with mutation of the paternal allele (E2 mϩ/pϪ ), leading to complete XL␣s deficiency and partial G s ␣ deficiency, failed to suckle, and most died soon after birth with hypoglycemia. 3 Those that survived had markedly reduced adiposity, increased metabolic rate, and activity levels and greater than normal glucose tolerance and insulin sensitivity (13)(14)(15). E2 mϩ/pϪ mice had increased urinary norepinephrine excretion, suggesting that these metabolic effects may be the consequence of increased sympathetic nervous system activity (14). In contrast, E1 mϩ/pϪ mice with paternal deletion of G s ␣ exon 1 leading only to partial G s ␣ deficiency had normal suckling and survival and developed obesity, glucose intolerance, and insulin resistance (3). Comparison of E2 mϩ/pϪ and E1 mϩ/pϪ mice indirectly implicates XL␣s as an important regulator of postnatal feeding as well as energy and glucose metabolism in older animals and suggests that XL␣s and G s ␣ may have opposite effects on metabolic regulation in the whole animal.
We have recently generated mice with deficiency of only XL␣s (Gnasxl mϩ/pϪ ) and showed that these mice have poor suckling and perinatal survival (7), similar to that observed in E2 mϩ/pϪ mice (12) and other mouse models with disruption or loss of the paternal Gnas allele (16 -18), implicating XL␣s as critical for postnatal feeding adaptation in mice. We have now examined the metabolic phenotype of adult Gnasxl mϩ/pϪ mice and show that these mice have markedly reduced adiposity, increased metabolic rate with increased lipid metabolism in adipose tissue, and increased glucose tolerance and insulin sensitivity. We provide evidence that these metabolic changes are due to increased sympathetic nervous system activity rather than adipocyte cell-autonomous effects. XL␣s (or possibly XLN1) appears to be an important negative regulator of sympathetic nervous system activity in mice. We also show that XL␣s expression in adipose tissue is developmentally regulated.

EXPERIMENTAL PROCEDURES
Animals-Gnasxl mϩ/pϪ mice (7) were repeatedly mated with CD1 wild-type mice (Charles River, Wilmington, MA) for Ͼ5 generations and were maintained on a standard pellet diet (NIH-07, 5% fat by weight) and 12 h:12 h light/dark cycle. Except when noted, all experiments were performed on 12-14week old male mutant mice and wild-type littermates. Experiments were approved by the NIDDK Animal Care and Use Committee.
Body Composition, Food Intake, Metabolic Rate, and Activity Measurements-Body composition was measured using the Minispec mq10 NMR analyzer (Bruker Optics Inc., Woodlands, TX) that was calibrated with corn oil for fat mass and rat muscle for lean mass. Food intake, metabolic rates (oxygen consumption rate by indirect calorimetry), and activity levels were determined as previously described (14). The effect of the ␤3-specific adrenergic agonist CL316243 (19) on metabolic rate was measured as follows. At 9 a.m. mice were placed into calorimetry chambers that were prewarmed to 30°C. At 1 p.m. CL316243 at the indicated dose (from a 1 mg/ml stock in saline) or a saline vehicle control was injected intraperitoneally, and after a 1-h delay data were collected for 3 h. Food and water were available at all times.
Microscopic Analysis-For standard histology, samples were fixed overnight in 4% paraformaldehyde, embedded in paraffin, and cut and stained with hematoxylin and eosin. For mitochondrial staining, 10-M cryostat sections were stained for succinate dehydrogenase by incubation in 130 mM Tris-HCl buffer (pH 7.4) containing 0.2 mM nitro blue tetrazolium and 60 mM sodium succinate (Sigma) at 37°C for 30 min. The slides were then rinsed with deionized water, dehydrated, and washed with xylene before coverslip mounting.
Hyperinsulinemic-Euglycemic Clamp Studies-Hyperinsulinemic-euglycemic clamps studies and in vivo glucose flux analysis were performed as previous described (15). Insulin levels on samples during the clamp studies were measured by radioimmunoassay kit (Linco Research, St. Charles, MO).
Blood, Tissue, and Urine Chemistries-Blood was obtained by retroorbital bleed. Serum glucose, cholesterol, and triglyceride levels were measured with an auto-analyzer by the NIH Clinical Chemistry Laboratory. Serum insulin, leptin, glucagon, and adiponectin were measured by Linco Research assay services. For C-peptide/insulin ratios, both were measured by radioimmunoassay (Linco). Serum resistin was measured using an enzyme-linked immunosorbent assay kit (Linco). Serum corticosterone was measured by Ani Lytics, Inc., Gaithersburg, MD. Serum free fatty acids (FFA) were measured using a kit from Roche Diagnostics. Triglyceride and glycogen content of hindlimb muscles were measured as previously described (15). Urine was collected by bladder puncture, and catecholamines were measured by high performance liquid chromatography (20), and corrected by creatinine concentration within the same samples, which were measured by Ani-Lytics.
Triglyceride Clearance Test-Clearance of triglycerides (12.5 l of peanut oil/g of body weight delivered by gavage) from the circulation was measured in mice after a 4-h fast. Blood was taken before gavage and hourly for 6 h after gavage, and plasma triglycerides were measured (kit #TR22421, Thermo DMA, Waltham, MA). Glucose and Insulin Tolerance Tests-Glucose and insulin tolerance tests were performed in overnight-fasted mice after intraperitoneal injection of glucose (2 mg/g) or insulin (Humulin, 0.50 mIU/g). Blood glucose levels in tail vein bleeds were measured using a Glucometer Elite (Bayer, Elkhart, IN) immediately before and at the indicated times after injection.
Portal Vein Insulin Injection and Immunoblot Analysis-Overnight-fasted mice were anesthetized with avertin (0.25 FIGURE 1. Expression patterns of encoded proteins from the GNAS/Gnas locus. The maternal (above) and paternal (below) alleles of Gnas are shown with active (white) and inactive (gray) promoters and first exons encoding NESP55 (Nesp), XL␣s (Gnasxl), and G s ␣ (exon 1), respectively, splicing to the common exon 2 (not drawn to scale). Exons downstream of exon 2 and promoters and exons for noncoding transcripts are not shown. mg/g of body weight intraperitoneal). Insulin (100 l of 15 g/ml, Sigma) was injected via the portal vein. At 2 or 4 min after injection, interscapular brown adipose tissue (BAT), epididymal white adipose tissue (WAT), and hind leg muscles were dissected and immediately frozen. Tissues were then homogenized, and protein concentrations of protein extracts were measured as previously described (15). Immunoprecipitations were performed on tissue extracts (1 mg of protein) using an anti-Akt1 Ab (Upstate Biotechnology, Lake Placid, NY) as per manufacturer's instructions followed by immunoblot analysis using an anti-Akt1 phospho-Ser 473 Ab (Upstate). Ab binding was determined by ECL (ECL kit, Amersham Biosciences). The same blots were stripped and rehybridized with the anti-Akt1 Ab.
Quantitative Reverse Transcription-PCR-RNA was extracted from frozen tissues using TRIzol (Invitrogen) or an Mini RNeasy kit (Qiagen, Germantown, MD) 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). PCR reactions (20 l total volume) included cDNA, 100 nM primers, and 10 l of SYBR Green MasterMix (Applied Biosystems, Foster City, CA). To get relative quantification, standard curves were simultaneously generated with serial dilutions of cDNA, and results were normalized to ␤-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. Sequences for gene-specific primers are listed in the supplementary table.
Semiquantitative RT-PCR-XL␣s mRNA expression was examined in BAT and WAT by semiquantitative RT-PCR. After first-strand synthesis, duplex PCR was performed (32 cycles, 60°C annealing temperature) using a pair of XL␣s-specific primers (1 M each) and ␤-actin-specific primers (100 nM each) (primer sequences listed in the supplemental table). Products were separated on agarose gels, and gels were stained with ethidium bromide.
Statistical Analysis-Data are expressed as the mean Ϯ S.E. Statistical significance between groups was determined using unpaired Student's t test (two-tailed) with differences considered significant at p Ͻ 0.05.

Gnasxl mϩ/pϪ Mice Have Reduced Survival and Adiposity-
In this study we examined the survival and adult metabolic phenotype of Gnasxl mϩ/pϪ mice, which were placed in a CD1 genetic background to compare these mice to other Gnas knock-out models that we have studied on the same genetic background. In most other inbred genetic backgrounds E2 mϩ/pϪ mice have no survival beyond several days. 3 In the CD1 background most Gnasxl mϩ/pϪ mice failed to suckle and died soon after birth, as was previously described in both these mice (7) and E2 mϩ/pϪ mice (12). By 3 weeks of age, Gnasxl mϩ/pϪ mice had a survival rate that was ϳ27% that expected by Mendelian inheritance (8 of 59 total offspring), and the effect appeared to be similar in males and females. This survival rate is similar to the survival rate that we previously reported for E2 mϩ/pϪ mice in the CD1 background (23%) (12).
Growth curves ( Fig. 2A) demonstrate that both male and female Gnasxl mϩ/pϪ mice grew very slowly through the first 30 days relative to their wild-type littermates, and the resulting weight differences were maintained during adulthood. Adult Gnasxl mϩ/pϪ mice weighed ϳ45% less than wild-type littermates and had a small but significant reduction in nasoanal body length (Fig. 2B). Body composition analysis by NMR showed that Gnasxl mϩ/pϪ mice had a markedly reduced fat mass and increased lean mass relative to total body weight ( Fig.  2C). Consistent with this, the relative weights of retroperitoneal and epididymal WAT pads and interscapular BAT pads were greatly reduced in Gnasxl mϩ/pϪ mice, whereas the relative weights of most other organs were unchanged (Fig. 2D). The relative weight of testes and lungs were slightly, although significantly increased in Gnasxl mϩ/pϪ mice as compared with controls. Histology of both BAT and WAT from Gnasxl mϩ/pϪ mice showed reduced lipid accumulation and size of adipocytes (Fig. 2E). BAT from Gnasxl mϩ/pϪ mice had markedly increased succinate dehydrogenase staining compared with controls ( Fig. 2F), indicating that the mutant mice had greater mitochondrial content in BAT. Overall, our findings show that, similar to E2 mϩ/pϪ mice (12,14), Gnasxl mϩ/pϪ mice have reduced preweaning survival, poor growth, and maintain a very lean phenotype in adulthood.
Hypermetabolism in Gnasxl mϩ/pϪ Mice Is Associated with Increased Norepinephrine Levels-The markedly reduced adiposity of Gnasxl mϩ/pϪ mice indicates that they may have either reduced food intake or increased energy expenditure rates relative to controls. Both male and female Gnasxl mϩ/pϪ mice had significantly increased food intake when corrected for body mass (Fig. 3A), indicating that reduced adiposity is not due to hypophagia. Hyperphagia in Gnasxl mϩ/pϪ mice may be a physiologic response to low circulating leptin levels (Table 1), which is an expected consequence of their severely lean phenotype. Both resting and total energy expenditure (O 2 consumption) rates were very significantly increased at both ambient temperature (24°C) and thermoneutral temperature (30°C) (Fig. 3B). There were no differences in the respiratory exchange ratio (ratio of CO 2 produced to O 2 consumed) (Fig. 3C) or activity levels (Fig. 3D) between wild-type and mutant mice. The metabolic changes are similar to those previously observed in E2 mϩ/pϪ mice, although we found E2 mϩ/pϪ mice to be hyperactive (14), whereas Gnasxl mϩ/pϪ mice were not. Sex and age differences between the mice examined in this and the prior E2 mϩ/pϪ mouse study may account for these discrepant results.
We next examined whether the hypermetabolic state could be explained by increased metabolic responsiveness to sympathetic stimulation by measuring the resting metabolic rate after administration of various doses of the ␤3-adrenergic agonist CL316243 at thermoneutral temperature (30°C). Resting metabolic rate at thermoneutrality reflects the metabolic rate in the absence of adaptive thermogenesis with minimal endogenous FIGURE 3. Studies of energy balance, activity, ␤3-adrenergic responsiveness, and urine catecholamine excretion. A, food intake in male (left) and female ( right) wild-type (closed bars) and Gnasxl mϩ/pϪ (open bars) mice (n ϭ 5-11/group). Shown are resting and total metabolic rate (B), respiratory exchange ratio (RER) (C ), and ambulatory and total activity levels (D) measured over 24 h at the indicated temperatures (n ϭ 6/group). Also shown are resting metabolic rate (E ) and RER (F) at thermoneutral temperature (30°C) after the indicated doses of CL316243 administered to male mice (n ϭ 3/group). G, Urine norepinephrine (NE ) and epinephrine (Epi) levels expressed as ng/mg creatinine in male and female mice (n ϭ 7-9/group). p Ͻ 0.05 (*) or p Ͻ 0.01 (**) versus wild type (W T). stimulation of adipose tissue by the sympathetic nervous system (21,22). Because ␤3 adrenergic receptors are specifically expressed in adipose tissue and CL316246 has been shown to specifically activate adipose tissue (23), increases in metabolic rate after CL316246 reflect the metabolic responsiveness of adipose tissue to catecholamine stimulation by ␤3-adrenergic receptors, a major receptor stimulated by sympathetic nerves in adipose tissue. Our results show that adipose tissue in Gnasxl mϩ/pϪ mice was not more sensitive to ␤3-adrenergic stimulation (Fig. 3E). In fact Gnasxl mϩ/pϪ mice tended to have an increased base-line activity (p ϭ 0.079) and a somewhat reduced response to higher doses of the ␤3 agonist, suggesting that they may be subtly resistant to ␤3-adrenergic stimulation. There were no significant differences in the respiratory exchange ratio in response to the ␤3 agonist (Fig. 3F). These results are similar to what was previously observed before and after maximal ␤3 agonist stimulation in E2 mϩ/pϪ mice (14).
We also assessed the ability of the ␤3 agonist to acutely raise serum FFA levels, which is a measure of the acute lipolytic response in adipose tissue. In the fed state Gnasxl mϩ/pϪ mice tended to have elevated serum FFA levels compared with controls (p ϭ 0.098) even though they had a reduced adipose mass (Fig. 4A). The increase in FFA levels measured 20 min after injection of CL316243 (100 g/kg) was significantly lower in Gnasxl mϩ/pϪ mice than in controls. The ratio of unstimulated/ stimulated FFA levels was significantly higher in Gnasxl mϩ/pϪ mice (0.33 Ϯ 0.05 in Gnasxl mϩ/pϪ versus 0.21 Ϯ 0.02 in control mice). These results are similar to the slightly reduced metabolic responsiveness to CL316243 observed in the mutant mice. In the fasted state (when insulin levels are low) Gnasxl mϩ/pϪ mice had 2-fold higher serum FFA levels than controls (Fig. 4B), indicative of a very high lipolytic rate in mutant mice. After administration of insulin (0.75 mIU/g of body weight), FFA levels fell to similar levels in Gnasxl mϩ/pϪ and control mice, indicating that the ability of insulin to suppress lipolysis is maintained in Gnasxl mϩ/pϪ mice.
To assess overall sympathetic nerve activity, we measured urinary norepinephrine concentrations normalized to creatinine to correct for differences in urinary concentration and in lean mass. Similar to E2 mϩ/pϪ mice (14), both male and female Gnasxl mϩ/pϪ mice had significantly increased urine norepinephrine (NE) levels (Fig. 3G). Urine epinephrine levels, which primarily reflect adrenomedullary activity, were also increased, although the difference was only significant in males. There was no significant difference in urine creatinine concentration between the two groups (data not shown). Overall, our results suggest that increased metabolic rate in adult Gnasxl mϩ/pϪ mice is primarily the result of increased sympathetic nervous system activity (and possibly adrenomedullary secretion) rather than an inherent difference in sensitivity to metabolic stimulation.
Gnasxl mϩ/pϪ Mice Have Improved Glucose Tolerance and Insulin Sensitivity-Random serum glucose, insulin, cholesterol, and triglyceride levels were all significantly lower in Gnasxl mϩ/pϪ as compared with control mice (Table 1). Similar to E2 mϩ/pϪ mice (15), Gnasxl mϩ/pϪ mice had an increased clearance rate of an oral triglyceride load and very low muscle triglyceride levels (Figs. 4, C and D), consistent with an increased rate of overall lipid metabolism. Glucose and insulin tolerance tests showed that Gnasxl mϩ/pϪ mice had increased glucose tolerance and insulin sensitivity compared with controls (Fig. 5, A and B). Improved insulin sensitivity is not explained by altered serum levels of adipocytokines, as circulating levels of leptin and the insulin-sensitizing factor adiponectin were reduced relative to controls, whereas resistin levels were unaffected in Gnasxl mϩ/pϪ mice (Table 1). Consistent with the severely lean phenotype, serum leptin levels were very low in Gnasxl mϩ/pϪ mice (Table 1). In contrast to young (100 g/kg body weight; n ϭ 6/group). B, mice were fasted for 9 h, and serum FFAs were measured before or 20 min after intraperitoneal insulin administration (0.75 mIU/g body weight; n ϭ 5-6/group). WT, wild type. C, serum triglyceride levels measured in wild-type (f, solid lines) and Gnasxl mϩ/pϪ (Ⅺ, dashed lines) mice before and hourly after lipid (peanut oil) administration by oral gavage (n ϭ 6/group). The areas under the curve are 972 Ϯ 135 for wild-type and 648 Ϯ 114 for Gnasxl mϩ/pϪ mice ( p Ͻ 0.05). D, triglyceride content of hindlimb muscles from non-fasting wild-type and mutant mice (n ϭ 7/group). Studies in panels A-C were performed in 6-month-old female mice. p Ͻ 0.05 (*) or p Ͻ 0.01 (**) versus wild type.  JULY 14, 2006 • VOLUME 281 • NUMBER 28

JOURNAL OF BIOLOGICAL CHEMISTRY 18993
Gnasxl mϩ/pϪ pups (7) and similar to adult E2 mϩ/pϪ mice (15), serum glucagon levels were unaffected in adult Gnasxl mϩ/pϪ mice (Table 1). Corticosterone levels were also unaffected in Gnasxl mϩ/pϪ mice (Table 1). Low tissue triglyceride levels are generally associated with increased insulin sensitivity and may contribute to the increased insulin sensitivity observed in Gnasxl mϩ/pϪ mice. Euglycemic-hyperinsulinemic clamp studies were performed to further characterize insulin action in specific tissues. Before insulin infusion, both basal glucose and insulin were lower in Gnasxl mϩ/pϪ mice (Fig. 5C). During the clamp glucose levels were maintained at similar levels in Gnasxl mϩ/pϪ mice and wild-type controls (Fig. 5C). Although receiving similar infusion rates corrected for body size, Gnasxl mϩ/pϪ mice had significantly lower insulin levels during the clamp (Fig. 5C), suggesting that they may have increased insulin clearance. The serum C-peptide/insulin ratio measured in non-fasting mice was 3-fold higher in Gnasxl mϩ/pϪ than in control mice (Fig.  5G), which is also consistent with these mice having an increased insulin clearance rate. Despite the lower insulin levels attained, Gnasxl mϩ/pϪ mice had significantly higher rates of glucose infusion and whole body glucose uptake (Fig. 5C), consistent with increased insulin sensitivity. Increased whole body glucose utilization is most accounted for by increased whole body glycolysis, although this difference did not reach statistical significance. There were no differences in whole body glycogen synthesis.
Glucose uptake was significantly increased in muscle and WAT and increased 2-fold in BAT (Fig. 5D). Consistent with these findings, phosphorylation of the downstream kinase Akt1 in response to acute insulin administration was greater in WAT, BAT, and skeletal muscle of mutant mice (Fig. 5F). Gene expression studies showed the insulin receptor, insulin receptor substrate 1, and Akt2 to be overexpressed in WAT and the insulin-sensitive glucose transporter glucose transporter 4 to be overexpressed in skeletal muscle of Gnasxl mϩ/pϪ mice (Fig.  6). Despite the increased muscle glucose uptake observed during the clamp study, Gnasxl mϩ/pϪ mice tended to have reduced muscle glycogen levels in the fed state (Fig. 5E, p ϭ 0.058). Potential causes of reduced muscle glycogen levels include hypoinsulinemia (Table 1, Fig. 5C), increased glucose utilization via glycolysis and/or increased glycogenolysis.
Basal (pre-clamp) endogenous glucose production rate, which primarily reflects glucose output from the liver, tended to be higher in Gnasxl mϩ/pϪ mice before the clamp, although this difference was not significant (Fig. 5C). Suppression of endogenous glucose production rate during the clamp was much greater in Gnasxl mϩ/pϪ mice despite the lower insulin levels attained in these mice. Comparison of Gnasxl mϩ/pϪ mice during the clamp with wild-type mice before the clamp, where insulin levels were similar in the two groups, showed that Gnasxl mϩ/pϪ mice had a much lower endogenous glucose production at similar circulating insulin levels (Fig. 5C). Overall our results show that, similar to E2 mϩ/pϪ mice (13,15), Gnasxl mϩ/pϪ mice have increased glucose tolerance and insulin sensitivity in muscle, adipose tissue, and liver.
Gene Expression Studies in Tissues Involved in Energy and Glucose Metabolism-We examined gene expression patterns in WAT, BAT, liver, and muscle from non-fasted mice by realtime quantitative RT-PCR (Fig. 6). Consistent with their reduced adiposity, Gnasxl mϩ/pϪ mice had reduced leptin and increased adiponectin expression in both BAT and WAT. Resistin was also overexpressed in WAT of Gnasxl mϩ/pϪ mice. These changes in leptin, adiponectin, and resistin expression mimic those observed in E2 mϩ/pϪ mice (14,15). The discrepancies between gene expression in adipose tissue and serum levels of resistin and adiponectin are most likely due to the fact that Gnasxl mϩ/pϪ have markedly reduced adipose tissue mass, and similar discrepancies between gene expression and serum protein levels have been previously observed in other knockout mouse models (15,24). The adipocytokine retinol binding protein 4, which has been reported to reduce insulin sensitivity (25), tended to be expressed at higher levels in WAT of Gnasxl mϩ/pϪ mice, although the difference was not significant.
Peroxisome proliferator activated receptor (PPAR) ␥-coactivator 1␣ is a cAMP-inducible gene that activates genes required for thermogenesis, mitochondrial function, and lipid oxidation (26). This gene as well as the downstream thermogenic uncoupling protein 1 gene and the mitochondriagenic genes nuclear respiratory factor 1 and mitochondrial transcription factor A were all significantly up-regulated in WAT and BAT of Gnasxl mϩ/pϪ mice, consistent with increased sympathetically mediated cAMP formation and mitochondrial biogenesis in adipose tissue (see Fig. 2F). Although uncoupling protein 1 was shown to be induced in both BAT and WAT, WAT had Ͻ1000fold lower uncoupling protein 1 mRNAs levels than BAT. The lipid oxidation gene acyl-CoA oxidase was also significantly upregulated in WAT. Lipoprotein lipase and hormone-sensitive lipase, which catalyze triglyceride uptake and hydrolysis, respectively, as well as other adipogenic and signaling genes (PPAR␥ coactivator 1␤, PPAR␥, sterol regulatory elementbinding protein 1c (SREBP1c), cAMP-response element-binding protein (CREB), and thyroid hormone receptor ␣) were also up-regulated in WAT of Gnasxl mϩ/pϪ mice. Several of these genes (PPAR␥, cAMP-response element-binding protein (CREB), lipoprotein lipase) were up-regulated in BAT of Gnasxl mϩ/pϪ mice as well (Fig. 5B). Increased lipoprotein lipase expression is consistent with the increased triglyceride clearance FIGURE 5. Studies of glucose metabolism and insulin sensitivity. Blood glucose in Gnasxl mϩ/pϪ mice (Ⅺ, dashed lines) and wild-type (WT) littermates (f, solid lines) measured at the indicated time points before and after glucose (A) or insulin administration (B) (n ϭ 4 -5 per group; results for insulin tolerance test are expressed as % of base line). Genotype had a significant effect by two-factor analysis of variance for both studies. C, basal (preclamp) and clamp endogenous glucose production (EGP) rate, glucose infusion rate, and rates of whole body glucose uptake, glycolysis, and glycogen synthesis in wild-type (closed bars) and Gnasxl mϩ/pϪ mice (open bars) during hyperinsulinemic-euglycemic clamp studies (n ϭ 7/group). Basal and clamp glucose (top panel ) and insulin levels (bottom panel) are shown to the right. D, glucose uptake rates in gastrocnemius muscle, WAT, and BAT, respectively, during the clamp study. E, glycogen content of hindlimb muscles obtained from non-fasting wild-type (open bar) and mutant (closed bar) mice (n ϭ 5-7/group). F, results of immunoblotting using anti-Akt1 phospho-Ser 473 (top of each panel) followed by anti-Akt Ab (bottom of each panel) in BAT, WAT, and muscle tissue extracts derived from wild-type (W T) and Gnasxl mϩ/pϪ (XL) mice which were sacrificed at the indicated times after insulin administration. G, serum C-peptide/insulin ratios from non-fasted 8-month-old male mice (n ϭ 5-6/group). p Ͻ 0.05 (*) or p Ͻ 0.01 (**) versus wild type. JULY 14, 2006 • VOLUME 281 • NUMBER 28 observed in Gnasxl mϩ/pϪ mice (Fig. 4C). Gnasxl mϩ/pϪ mice also have a 4-fold higher expression of the forkhead transcription factor FOXC2 in WAT. Adipose-specific FOXC2 overexpression leads to increased ␤-adrenergic/G s ␣/cAMP signaling and a metabolic phenotype similar to that present in Gnasxl mϩ/pϪ mice (27,28). Expression of the ␤2and ␤3-adrenergic receptor genes were increased (only significant for ␤3 receptors), whereas there were no changes in ␤1or ␣2a-adrenergic receptors. Overall these changes in gene expression are consistent with increased cAMP-driven metabolic activation and energy dissipation (lipolysis and lipid oxidation) in adipose tissue of adult Gnasxl mϩ/pϪ mice. In contrast, there were no significant changes in genes involved in energy metabolism and thermogenesis in skeletal muscle (Fig. 6C), suggesting that the increased metabolic rate of Gnasxl mϩ/pϪ is primarily due to increased energy dissipation in adipose tissue.

Role of XL␣s in Metabolic Regulation
The major changes in liver (Fig. 6D) include reduced expression of genes involved in lipogenesis (PPAR␥, sterol regulatory element-binding protein 1c (SREBP1c), ATP citrate lyase, and fatty acid synthase) and increased expression of the lipid oxidation gene for carnitine palmitoyltransferase 1a. These changes as well as the increased lipid oxidation in adipose tissue probably lead to the low serum and tissue triglyceride levels and increased triglyceride clearance in Gnasxl mϩ/pϪ and E2 mϩ/pϪ mice (14,15). The glucose-metabolizing gene glucokinase was down-regulated, whereas the gluconeogenic gene for phosphoenolpyruvate carboxykinase (PEPCK) tended to be somewhat up-regulated, although the difference for PEPCK was not statistically significant. Insulin stimulates lipogenesis and glucose metabolism and inhibits gluconeogenesis, whereas catecholamines lead to opposite effects on hepatic glucose metabolism (29). The gene expression changes in livers of Gnasxl mϩ/pϪ mice most likely result from these mice having very low insulin levels and increased catecholamine levels and are consistent with the tendency for these mice to have higher basal endogenous glucose production rates, which were observed during the clamp study.
XL␣s Expression in Adipose Tissue Is Lost during Postnatal Development-To determine whether XL␣s deficiency in adipose tissue could contribute to the adult Gnasxl mϩ/pϪ metabolic phenotype, we examined XL␣s mRNA expression in adipose tissue of wild-type mice at various ages by duplex RT-PCR. As we showed previously (7), XL␣s mRNA was expressed in both BAT and WAT during the first week postpartum (Fig. 7). However XL␣s expression in both BAT and WAT decreased over the first 2 weeks after birth, leading to barely detectable expression at postpartum days 15 and 20 and total absence in adult mice. Integrity and amount of RNA was confirmed by successful coamplication of ␤-actin mRNA in all samples (Fig.  7, A and B, lower major band). XL␣s expression is developmentally regulated in adipose tissue, being present only during the early postnatal period. It is, therefore, unlikely that genetic loss of XL␣s should lead to cell-autonomous changes in adipocytes in adult mice.

DISCUSSION
In this study we show that XL␣s deficiency leads to reduced adiposity and increased metabolic rate and insulin sensitivity. These changes as well as reduced survival, poor suckling, and reduced body weight of Gnasxl mϩ/pϪ pups (7) are also observed in E2 mϩ/pϪ mice (12). These results confirm that the E2 mϩ/pϪ phenotype results from XL␣s deficiency, which has a dominant effect over the partial G s ␣ deficiency in these mice. This explains why E1 mϩ/pϪ mice, which have loss of G s ␣ expression from the paternal allele but no disruption of XL␣s expression, lack the Gnasxl mϩ/pϪ or E2 mϩ/pϪ phenotype and in fact develop obesity and insulin resistance (3).
Fat stores reflect the balance between energy intake and energy expenditure. Reduced adiposity of Gnasxl mϩ/pϪ mice results from increased energy expenditure. Both Gnasxl mϩ/pϪ and E2 mϩ/pϪ mice (14,15) have increased energy expenditure associated with increased triglyceride clearance, consistent with an increase in lipid oxidation. In Gnasxl mϩ/pϪ mice the most striking changes in morphology and gene expression are in WAT, with a gene profile pattern consistent with increased metabolic activity. Multiple genes involved in lipolysis (hormone-sensitive lipase) and fatty acid oxidation and thermogenesis (PPAR␥ coactivator-1, uncoupling protein 1, acyl-CoA oxidase, thyroid hormone receptor ␣, nuclear respiratory factor 1, and mitochondrial transcription factor A), most of which are known to be stimulated by ␤-adrenergic/G s ␣/cAMP-signaling pathways (26,30), are up-regulated in WAT, and to a lesser extent in BAT, of Gnasxl mϩ/pϪ mice. Consistent with increased expression of genes involved in oxidative metabolism and mitochondrial biogenesis, we showed that BAT from Gnasxl mϩ/pϪ mice has increased mitochondrial content. In contrast, genes regulating energy metabolism were unaffected in skeletal muscle or liver, except for an increase in carnitine palmitoyltransferase 1a in the latter. It, therefore, appears that Gnasxl mϩ/pϪ mice are lean primarily as a result of increased lipid mobilization and oxidation in adipose tissue. A similar metabolic profile was observed in mice with adipose-specific overexpression of the forkhead transcription factor FOXC2 (27,28) and mice in which the C/EBP␣ gene was replaced with the C/EBP␤ gene (30). Both of these models have increased ␤-adrenergic/G s ␣/cAMP signaling in adipocytes, and increased G s ␣ expression in cultured white adipocytes was shown to promote mitochondrial biogenesis and lipid oxidation (30). Interestingly, FOXC2 is overexpressed in WAT of Gnasxl mϩ/pϪ mice, suggesting that this gene may have a role in the induction of lipid metabolism in adipose tissue by sympathetic nervous system stimulation. The sympathetic nervous system stimulates lipolysis in WAT (31, 32) and lipolysis, lipid oxidation, and thermogenesis in BAT (26) via ␤-adrenergic/G s ␣/cAMP-signaling pathways. The metabolic and molecular changes in adipose tissue of Gnasxl mϩ/pϪ mice could result from a cell-autonomous effect leading to increased G s ␣/cAMP signaling in adipocytes or from increased sympathetic nervous system activity. Our results for adult mutants are most consistent with altered sympathetic nerve activity rather than adipocyte cell-autonomous differences as the primary mechanism for the changes observed in Gnasxl mϩ/pϪ mice. Similar to E2 mϩ/pϪ mice (14), Gnasxl mϩ/pϪ mice had increased urine norepinephrine levels, indicative of increased sympathetic nervous system activity, and increased urine epinephrine levels in males, indicative of increased adrenal medullary activity. Adult Gnasxl mϩ/pϪ mice had increased energy expenditure rates and serum FFA levels in the basal state and reduced responses in both of these parameters after administration of an adipose-specific ␤3-adrenergic agonist. These results suggest that these mice are not intrinsically more sensitive to catecholamine stimulation (at least not through ␤3 adrenergic pathways) but, rather, have increased sympathetic nervous system activity. We cannot rule out the possibility that increased ␤-adrenergic receptor gene expression, which we documented in BAT from Gnasxl mϩ/pϪ mice, may lead to increased sensitivity of adipose tissue to catecholaminergic stimulation and increased metabolic activation. However, this is not likely to be the primary defect as we show that XL␣s is not expressed in the adipose tissue of adult wild-type mice, and therefore, genetic loss of XL␣s should produce no intrinsic changes to adipocytes. We cannot rule out the possibility that XL␣s deficiency in adipocytes during the perinatal period leads to altered adipose function later in development.
Hyperleptinemia, which results from increased leptin expression in adipose tissue, acts in the central nervous system to inhibit food intake and stimulate sympathetic nervous system activity, leading to increased energy expenditure. Sympathetic stimulation of ␤-adrenergic/G s ␣/cAMP signaling in adipocytes inhibits leptin expression (33). Hypoleptinemia likely contributes to the hyperphagia in Gnasxl mϩ/pϪ mice. However, the increased sympathetic nervous system activity in Gnasxl mϩ/pϪ mice is opposite to what is expected in the setting of hypoleptinemia, suggesting that these mice have a primary defect in sympathetic nervous system regulation. Chronically increased sympathetic stimulation of adipose tissue contributes to reduced leptin expression and increased adiponectin and resistin expression, similar to what we have observed in E2 mϩ/pϪ mice (14,15,34,35).
XL␣s is expressed within the distribution of the sympathetic trunk in the central nervous system (7,36), suggesting that it could play a primary role as a negative regulator of sympathetic nervous system activity. G s ␣ deficiency may lead to opposite metabolic effects by acting at sites that are distinct from those expressing XL␣s. For example, G s ␣ presumably mediates the effects of melanocortins to stimulate sympathetic nervous system activity within the hypothalamus (31,32,(37)(38)(39). Another possibility is that the alternative truncated Gnasxl product XLN1 is a dominant-negative inhibitor of G s ␣ signaling. XL␣s has also been shown to be expressed in the adrenal gland (6), and therefore, XL␣s deficiency could perhaps additionally lead to dysregulation of catecholamine secretion from the adrenal medulla.
Similar to E2 mϩ/pϪ mice (13), Gnasxl mϩ/pϪ mice have increased glucose tolerance and low insulin levels. Although XL␣s has been shown to be expressed in pancreatic islets (6), hypoinsulinemia is not likely due to impaired insulin secretion from ␤-cells, as insulin secretion is intact in E2 mϩ/pϪ mice (15). Rather, like E2 mϩ/pϪ mice (13,15), Gnasxl mϩ/pϪ mice have increased insulin sensitivity in BAT, WAT, muscle, and liver. The low insulin levels attained in Gnasxl mϩ/pϪ during the clamp study and as well as their elevated serum C-peptide/ insulin ratios suggest that increased insulin clearance is a likely contributing factor to their hypoinsulinemia. The insulin receptor gene is overexpressed in WAT and to a lesser extent in liver of Gnasxl mϩ/pϪ mice, which may result in increased insulin clearance through receptor binding and internalization.
Increased insulin sensitivity of Gnasxl mϩ/pϪ mice cannot be explained by changes in circulating levels of the adipocytokines leptin, adiponectin, or resistin. More likely, increased insulin sensitivity is secondary to reduced tissue triglyceride content resulting from chronically increased lipid oxidation and energy expenditure, which we have now documented in Gnasxl mϩ/pϪ mice and had documented previously in E2 mϩ/pϪ mice (15). Reduced expression of hepatic lipogenic genes, most likely secondary to chronic hypoinsulinemia, also contributes to hypolipidemia in Gnasxl mϩ/pϪ mice. Sympathetic nerve stimulation also stimulates glucose uptake in BAT and muscle by insulinindependent mechanisms (40 -42). Hypoinsulinemia and elevated catecholamine levels (29) may lead to increased basal glucose production as well as the up-regulation of phosphoenolpyruvate carboxykinase and down-regulation of glucokinase in livers of Gnasxl mϩ/pϪ mice.
We have now shown that XL␣s is only expressed in adipose tissue during the early postnatal period, which suggests that this G s ␣ isoform may play a specific role during development. The parental conflict hypothesis of genomic imprinting predicts that paternally expressed imprinted genes promote the delivery of resources to the offspring during the gestational and preweaning period, and loss of these genes leads to reduced fetal and early postnatal growth (43,44). Another paternally expressed imprinted gene Igf2, which encodes a fetal growth factor, is also only expressed at high levels during gestation and early development (45,46). The parental conflict theory predicts that the evolutionary drive for expression of these imprinted genes may be lost after weaning. Small paternal 20q13 deletions including the GNAS locus have been associated with a global dysmorphic and developmental syndrome including pre-and post-natal growth retardation, abnormal adipose tissue distribution, and feeding difficulties (47), suggesting that XL␣s deficiency may lead to similar defects in humans. However, the role of XL␣s in humans remains unclear, as Albright hereditary osteodystrophy patients with paternal GNAS mutations predicted to disrupt XL␣s expression do not develop a specific syndrome similar to 20q13 deletion or reminiscent of that observed in Gnasxl mϩ/pϪ mice.