Induction of Galphaq-specific antisense RNA in vivo causes increased body mass and hyperadiposity.

Transgenic BDF-1 mice harboring an inducible, tissue-specific transgene for RNA antisense to Gαq provide a model in which to study a loss-of-function mutant of Gαq in vivo Gαq deficiency induced in liver and white adipose tissue at birth produced increased body mass and hyperadiposity within 5 weeks of birth that persisted throughout adult life. Gαq-deficient adipocytes display reduced lipolytic responses, shown to reflect a newly discovered, α1-adrenergic regulation of lipolysis. This α1-adrenergic response via phosphoinositide hydrolysis and activation of protein kinase C is lacking in the Gαq loss-of-function mutants in vivo and provides a basis for the increased fat accumulation.

Heterotrimeric G-proteins (G-proteins) 1 mediate transmembrane signaling from a populous group of cell-surface receptors to a lesser group of effector molecules that includes adenylyl cyclase, phospholipase C, and various ion channels (1). G-proteins have been shown to regulate complex biological processes, including cellular differentiation (2,3), neonatal development (4 -6), and oncogenesis (7). The expression of G␣ o , for example, is highly localized to the growth cones of developing neurites (2). Suppression of G␣ o expression provokes the collapse of developing growth cones (8), whereas expression of constitutively active mutants of G␣ o promote increased expression of neurites (9). In adipogenesis of 3T3 L1 embryonic fibroblasts, G␣ s acts as a suppressor (2). Inducers of differentiation stimulate a sharp decline in G␣ s levels, and constitutive expression of G␣ s blocks induction of differentiation (2). G␣ i2 has been shown to regulate the progression of embryonic stem cells to primitive endoderm (4), acting via phospholipase C (PLC) and protein kinase C to suppress progression (10). The morphogen retinoic acid induces primitive endoderm by stimulating a sharp decline in G␣ i2 (4). Mimicking the decline with oligodeoxynucleotides antisense to G␣ i2 provokes progression in the absence of retinoic acid (4,10). Study of G-proteins in vivo is a formidable task. The role of G␣ i2 in vivo has been studied through inducible, tissue-specific ablation by antisense RNA (5,6) and gene inactivation by homologous recombination (11). Deficiency in G␣ i2 leads to a runted phenotype (5,6,11), insulin resistance (12), and for the transgenic mice with the inactivated G␣ i2 gene, ulcerative colitis and adenocarcinoma of the colon (11).
Little is known about the role of G-proteins of the G q family in vivo. Two highly homologous members of the G q subfamily of G-proteins, G␣ q and G␣ 11 , can stimulate PLC and are insensitive to pertussis toxin (13)(14)(15)(16)(17). G␣ 11 have been shown to mediate growth in fibroblasts in response to bradykinin and thrombin (18), hypertrophy in cultured neonatal ventricular myocytes (19), and transformation in NIH 3T3 cells (20). In the current work, we employ conditional, tissue-specific expression of RNA antisense to G␣ q in transgenic mice to explore the role of this G-protein in vivo and more specifically in white adipocytes made deficient of G␣ q . 3 ), Gene Screen Plus, and anti-G␣ i3 antibodies (EC2) were purchased from Dupont NEN. All other reagents were purchased from Sigma or standard suppliers (5).

Reagents and Supplies-[ 3 H]cyclic AMP, [ 32 P]dCTP, [␥-32 P]ATP, [ 3 H]inositol 1,4,5-trisphosphate (IP
Mice-The B6D2F1 (BDF1) strain of mice was purchased from Taconic Farms Inc. and handled in accordance with the guidelines established by the Institutional Animal Care and Use Committee at the State University of New York at Stony Brook.
Experimental Design of the Antisense RNA Strategy-The pPCK-ASG␣ q expression vector was constructed as described below using standard techniques. In order to insert the antisense sequence at the BglII site within the first exon of the phosphoenolpyruvate carboxykinase (PEPCK) gene, the 7.0-kb PEPCK gene was subcloned as a 1.0-kb EcoRI/HindIII and a 5.8-kb HindIII/BamHI fragment into the vector pGEM7Zf(ϩ) (Promega). The vector harboring the 1.0-kb gene fragment was digested with BglII, and the restriction ends were made flush using the Klenow fragment. The 39-bp antisense sequence was obtained within a 235-bp NheI/SstI fragment excised from the vector pLNC-ASG␣ i2 (4). The restriction ends of this fragment were filled-in, ligated with the BglII-digested gene fragment, and used to transform XL-1 Blue strain of Escherichia coli (Stratagene) under selection with ampicillin. Plasmids with the 1.2-kb fragment and an insert oriented to produce antisense RNA were identified by direct DNA sequencing. The 1.2-kb fragment containing the antisense sequence was digested with SacII and ClaI (sites present in the 235-bp sequence harboring the antisense sequence), and a 59-bp oligomer containing the sense sequence to G␣ q (Ϫ33 to ϩ3) flanked by restriction enzyme sites for BclI (5Ј) and SalI (3Ј) was ligated via force cloning into the SacII and ClaI sites. Insertion of the oligomer was confirmed by restriction digest analysis with BclI and SalI. The presence of unique restriction sites within the 235-bp fragment facilitates the removal and insertion of different antisense sequences in a cassette-like fashion. The 1.2-kb fragment containing the antisense sequence was excised and ligated into the plasmid harboring the 5.8-kb gene fragment to produce the 7.0-kb pPCK-ASG␣ q construct. In addition, the synthesis of primers complimentary to the flanking ends of the 235-bp insert allows for discrimination between the pPCK-ASG␣ q RNA and the endogenous PEPCK RNA in subsequent reverse transcription-PCR amplification reactions (see below).
In choosing the antisense RNA target sequence, we had to consider the large degree of nucleotide identity among the G-protein ␣-subunits within their respective open reading frame regions. This prompted us to look for unique target sequences within the 5Ј-and 3Ј-untranslated regions of the G␣ q mRNA. The 36 nucleotides immediately upstream of and including the translation initiation codon were chosen to serve as the antisense target sequence. This 39-base pair sequence (5Ј-CGCGC-CGGCGGGGCTGCAGCGAGGCACTTCGGAAGAATG-3Ј) did not show any significant homology with sequences present in the GenBank TM data base, including G␣ 11 (33% homology) and G␣ 14 (27% homology). Cell Culture and Transfection-FTO-2B cells were cultured in a 5% CO 2 , 95% O 2 chamber and maintained in Ham's F-12/Dulbecco's modified Eagle's medium (1:1) supplemented with 10% fetal bovine serum. Cells were cotransfected with a plasmid that would allow for neomycin selection of positive transfectant clones. The control vector or pPCK-ASG␣ q construct was added in 5-fold excess relative to the plasmid containing the selectable marker. Transfection was carried out using the Lipofectin reagent (Life Technologies, Inc.) according to the manufacturer's protocol.
Detection of Antisense RNA Expression-Total RNA was extracted as described previously (4). One microgram of total RNA was reverse transcribed using a pPCK-ASGa q -specific downstream primer and then PCR-amplified in the presence of both the upstream and downstream primer set according to the manufacturer's protocol (Perkin-Elmer). The sequences for the upstream and downstream primers were dCGTTTAGTGAACCGTCAGA and dAGGTGGGGTCTTTCATTCCC, respectively.
Production of Transgenic Mice-Transgenic lines of mice were produced at the Transgenic Mouse Facility at SUNY Stony Brook using standard techniques (5,6). Briefly the pPCK-ASG␣ q construct was excised free of vector sequences and purified prior to microinjection into single-cell preimplantation embryos. Microinjected embryos were then transferred to pseudopregnant females. Offspring carrying the transgene were identified by PCR amplification and subsequent Southern analysis using a pPCK-ASG␣ q -specific probe uniformly labeled with [ 32 P]dCTP (5). Five separate founder lines were identified by Southern analysis and bred over 10 generations (5).
White Adipocyte Isolation-White adipocytes were isolated from epididymal and parametrial fat pads by collagenase digestion, as described previously (5). Briefly, 0.5-1.0 g of adipose tissue was excised from male and female mice, weighed, and added to an equal volume of Krebs-Ringer phosphate buffer (KRP) containing 3% bovine serum albumin (KRP/BSA), prewarmed to 37°C. The tissue was digested for 1 h using collagenase (1 mg/ml) at 37°C in an orbital, shaking water bath. The isolated adipocytes were washed twice with the KRP/BSA buffer and then resuspended to a final volume to achieve 62.5 mg of wet weight of packed adipocytes/ml in the same buffer. The KRP/BSA buffer was supplemented with adenosine deaminase at a concentration of 0.5 unit/ml.
Cyclic AMP Accumulation and Lipolysis-Briefly, collagenase-digested white fat cells from epididymal and epoophoronal pads were incubated at 37°C in KRP buffer supplemented with 3% bovine serum albumin and adenosine deaminase (0.5 unit/ml) for 30 min in the absence or presence of the drugs indicated. For lipolysis determinations, the assays were terminated with 0.65 N HClO 4 . Samples were deproteinized and neutralized with KOH/KCl/imidazole (2.6/0.52/0.52 M, respectively) and the glycerol content determined by measuring the reduction of NAD ϩ to NADH in a coupled assay. NADH production was assayed using a microplate fluorometer set to an excitation wavelength of 360 nm and an emission wavelength detection of 460 nm. The mass of glycerol per sample was extrapolated from a standard curve of stock glycerol. Cyclic AMP accumulation was measured using a competitive binding assay. Briefly, 80 l of fat cells (ϳ5 mg/tube) were treated with various agents for 6 min at 37°C. The reaction was stopped by the addition of HCl (0.1 N final) and boiling for 1 min. The samples were neutralized with NaOH and assayed for cyclic AMP content. Cyclic AMP accumulation was measured in adipocytes stimulated with either phenylephrine, epinephrine, norepinephrine, or isoproterenol The data are expressed as the mean values in picomoles of cyclic AMP (ϮS.E.) per million cells from three independent trials, each performed in triplicate. IP 3 and 1,2-sn-Diacyglycerol (DAG) Accumulation-Cells were incubated with the indicated agents and the IP 3 measured as described previously (21). DAG was determined using the DAG kinase assay (10). The data are expressed as the mean values in nanomoles (ϮS.E.) per million cells from three independent trials, each performed in triplicate.
G-protein Immunoblot Analysis-Membrane fractions were prepared from rat hepatoma FTO-2B cells, as described previously (5). Cell membranes were prepared from adipose tissue of transgenic mice and their control littermates (5). Aliquots of cell membrane were subjected to SDS-polyacrylamide (10% acrylamide) gel electrophoresis, and the separated proteins were transferred to nitrocellulose blots. The blots of the membrane proteins were probed with anti-peptide antibodies specific for G␣ q (antibodies E973 and E976), G␣ i2 (antibody CM112), G␣ s (antibody CM129), G␤ 2 (antibody CM162), G␣ 11 (antibody E976), or G␣ i3 (antibody EC2), and the immune complexes were made visible by staining with a calf alkaline phosphatase-conjugated, goat anti-rabbit IgG second antibody (4,5). The "CM" antibodies were prepared by our laboratory (4,5).

RESULTS AND DISCUSSION
Defining the role of a specific G-protein subunit, like G␣ q , in vivo is a formidable task. We adopted the strategy of conditional, antisense RNA to ablate G␣ q in vivo in a tissue-specific manner, creating loss-of-function mutants in adipose and liver, prominent sites of G␣ q expression. The degree of nucleotide identity among the G-protein ␣-subunits within the open reading frames dictated selection of the 5Ј-untranslated region immediately upstream and including the ATG initiator codon (Ϫ33 to ϩ3) as the antisense RNA sequence targeting Ga q (Fig.  1). This region is unique with respect to sequences with the GenBank TM data base and does not share significant homology with other G-protein ␣-subunits (Fig. 1A), including other members of the G q family, G␣ 11 (33% homology with respect to G␣ q ) and G␣ 14 (27% homology with respect to G␣ q ). A doublestranded oligodeoxynucleotide fragment antisense to G␣ q was inserted into BclI and SalI sites of the pPCK-AS vector (Fig.  1B), an inducible expression vector driven by the promoter of the PEPCK gene (5,6). Screening of FTO-2B hepatoma cells stably transfected with pPCK-ASG␣ q was performed from day 0 to day 12 following induction of the promoter with the chlorophenylthio analogue of cyclic AMP (CPT-cyclic AMP, 25 M). Immunoblots reveal that levels of G␣ i2 , G␣ s , and G␤ 2 were unaffected by induction of pPCK-ASG␣ q , while staining with an antibody to G␣ q displayed a 54% loss by day 6, and 86% loss of G␣ q by day 9 following induction ( Fig. 1C). By day 12 of the induction with CPT-cyclic AMP, G␣ q was not detectable in the immunoblots of FTO-2B cell membranes. Induction of the pPCK-SG␣ q vector, harboring the sense as compared to antisense sequence for G␣ q , resulted in a null phenotype, i.e. G␣ q expression was normal. G␣ q activates PLC-␤ in the liver (12)(13)(14)(15)(16) and suppression of G␣ q in FTO-2B hepatoma cells reduced basal PLC activity from 2.7 Ϯ 0.6 to 1.0 Ϯ 0.3 (p Յ 0.05 for difference) and abolished PLC stimulation in response to either 10 nM angiotensin II (0.9 Ϯ 0.3) or 10 M norepinephrine (0.95 Ϯ 0.2), as determined by mass assay of intracellular IP 3 accumulation at 30 s following hormonal stimulation (n ϭ 5, pmol of IP 3 accumulation/g of cellular protein).
The pPCK-ASG␣ q construct was excised as a 7.0-kb EcoRI-BamHI fragment, microinjected into single cell, preimplantation embryos, and the microinjected embryos were transferred into pseudopregnant recipients. BDF1 mice harboring the transgene were identified by PCR of tail DNA. Five independent founders were identified from two rounds of microinjection and implantation. Five separate founder lines have been propagated for more than 10 generations. Immunoblots of crude membranes from fat, liver, brain, and lung subjected to SDS-PAGE and stained with a G␣ q -specific antiserum reveal the near absence of G␣ q in the tissues, fat and liver, targeted by the transgene (Fig. 1D). Immunoblots of brain and lung, tissues not targeted by the PEPCK vector, displayed normal levels of G␣ q (Fig. 1E). Expression of G␣ i2 , G␤ 2 , and G␣ 11 (not shown) were not significantly altered in the transgenic mice.
Necropsy and histology of the transgenic mice were performed. Prominent was the increase in body weight observed in the mice harboring the pPCK-ASG␣ q transgene ( Fig. 2A). By 5 weeks of age, the transgenic mice were Ͼ135% of the body weight of their control littermates, for both male and female mice alike (Fig. 2B). Progeny of the five founder lines harboring the transgene all display the increased body weight (not shown). The fat mass at 4 weeks after birth increased by 50% in the transgenic mice (Fig. 2C). At 8 weeks of age, the white epididymal and epoophoronal fat mass of the transgenic mice was 1.75-fold greater than that of the control mice. Segregated by gender for males, white fat mass (mg) was 215 Ϯ 5 and 333 Ϯ 8 (n ϭ 5, p Յ 0.05) for 12-week-old control and transgenic mice, respectively. For females, white fat mass was 160 Ϯ 10 mg and 303 Ϯ 8 mg (n ϭ 6, p Յ 0.05) for 12-week-old control and transgenic mice, respectively. By 24 weeks, the transgenic mice displayed a 1.4-fold increase in fat mass, and the percentage of whole body weight as fat mass was 2.3 Ϯ 0.2 as compared to 1.1 Ϯ 0.3 (n ϭ 6) for control mice (Fig. 2C). Total body protein and nasal-anal length were unaffected by the presence of the transgene over this same range in age (not shown). Equally notable was the dramatic increase in adiposity, i.e. fat cell number, that occurred in the transgenic mice lacking G␣ q expression in adipose tissue (Fig. 2D).
To assess the effects of G␣ q deficiency on cell signaling, we investigated the adipocytes isolated from transgenic mice and their control littermates. PLC-␤ signaling by loss-of-function G␣ q -deficient white adipocytes was virtually abolished, i.e. IP 3 and DAG accumulation in response to norepinephrine, vasopressin, phenylephrine, or bradykinin (all hormones that activate PLC) was markedly attenuated (Fig. 3, A and B, respectively). Suppression of G␣ q in adipocytes of the pPCK-ASG␣ q mice and the stably transfected hepatoma cells abolished PLC activation by a variety of hormones ( Figs. 1 and 3). This loss of signaling in G␣ q deficiency occurs, although expression of the G␣ 11 subunit was found to be normal (not shown). Both G␣ q and G␣ 11 are expressed in a number of tissues (12)(13)(14), including fat and liver. The observations from the present study suggest that G␣ q and G␣ 11 may not be redundant with respect to PLC activation in vivo.
Since PLC activation and accumulation of either IP 3 or DAG have not been implicated in controlling lipolysis, the pharmacology of the lipolytic response observed in the G␣ q -deficient adipocytes came as a great surprise (Fig. 3C). The lipolytic response to a mixed ␣and ␤-adrenergic agonist norepinephelectrophoresis, transferred to nitrocellulose blots, and probed with rabbit polyclonal antisera specific for the G-protein subunits indicated (5,6). Immunecomplexes were made visible with goat anti-rabbit IgG coupled to calf alkaline phosphatase and colorimetric development (panel C). Crude membranes were prepared from epididymal and epoophoronal white fat and liver (panel D) and brain and lung (panel E) tissues obtained from 24-week-old control (C) and transgenic (T) mice. Samples (15 g of protein/lane) were subjected to SDS-polyacrylamide gel electrophoresis on a mini-gel apparatus and transferred to nitrocellulose for immunoblot analysis of various G-protein subunits, as described earlier (5,6). For immunoblotting, the sample loading was limited to 15 g/lane, within the range established for linearity between sample loading and quantification of immunostaining (not shown). Quantification of the blots revealed no significant change in the Gprotein subunits tested between control and transgenic mouse tissues, with the exception of the loss of G␣ q in liver and fat tissues. Expression of G␣ s was normal in liver and fat, although reduced (Ͻ15%) occasionally in fat, but not liver, of some transgenic mice (not shown). Scanning densitometry values for immunoblots of fat tissue from control and transgenic mice, respectively, were as follows: G␣ q , 0.26, 0.01; G␣ s , 0.92, 0.88; G␤, 0.31, 0.33; and G␣ i2 , 0.72, 0.75 arbitrary OD units. Scanning densitometry values for immunoblots of liver tissue from control and transgenic mice, respectively, were as follows: G␣ q , 0.51; 0.02; G␣ s , 0.68, 0.64; G␤, 0.27, 0.25; and G␣ i2 , 0.22, 0.25 arbitrary OD units. Scanning densitometry of immunoblots from brain and lung revealed no significant differences in the values obtained with tissues from transgenic as compared to control mice (not shown). The antibodies employed for staining of immunoblots for specific G-proteins subunits were as follows: E973 for G␣ q ; CM112 for G␣ i2 ; CM129 for G␣ s ; E976 for G␣ 11 ; and CM162 for G␤ 2 .
FIG. 1. G␣ q expression is suppressed in hepatoma cell stably transfected with pPCK-ASG␣ q and in mice harboring the pPCK-ASG␣ q transgene. Comparison of the 5Ј-untranslated region from Ga q (nucleotides Ϫ33 to ϩ3) with G␣ 14 and G␣ 11 , additional members of the G q family (panel A). The pPCK-ASG␣ q construct for inducible expression of RNA antisense to G␣ q (panel B). The 36-nucleotide sequence upstream of and including the translation initiation codon was inserted into the first exon of the rat phosphoenolpyruvate carboxykinase gene (PCK) to provide a 2.8-kb hybrid pPCK-ASG␣ q antisense RNA, driven by a promoter which is silent in utero and activated at birth. Crude membranes (0.2 mg of protein/SDS-polyacrylamide gel electrophoresis lane) were prepared from rat hepatoma FTO-2B cells that were stably transfected with the pPCK-ASG␣ q construct and induced with CPTcyclic AMP for 0, 6, 9, and 12 days, subjected to SDS-polyacrylamide gel rine was blunted in the G␣ q -deficient adipocytes. Lipolysis in response to the ␤-adrenergic agonist isoprenaline was impaired, whereas the response to the ␣ 1 -adrenergic agonist phenylephrine was abolished in the loss-of-function mutant cells. These results were unexpected, since neither a direct role of G␣ q in activating adenylyl cyclase nor the existence of a prominent ␣ 1 -adrenergic stimulation of lipolysis have been reported. Analysis of cyclic AMP accumulation provided direct proof linking loss of G␣ q to impaired lipolysis in response to ␤-adrenergic stimulation (acting via adenylyl cyclase) as well as to ␣ 1 -adrenergic stimulation (acting via PLC). Forskolin (10 M)-stimulated cyclic AMP accumulation, in contrast, was actually elevated in the G␣ q -deficient as compared to control adipocytes (125 Ϯ 15 and 160 Ϯ 5 pmol/10 6 cells, respectively). Forskolin (10 M)-stimulated lipolysis was equivalent in transgenic and control mice (14.1 Ϯ 2.9 and 13.9 Ϯ 0.8 mol of glycerol release/10 6 cells, respectively), as were the abundance of ␤-adrenergic receptors (140 Ϯ 4 and 133 Ϯ 9 fmol/mg of protein, respectively) in crude adipocyte membranes and the amounts of cyclic AMP phosphodiesterase activity (1.33 Ϯ 0.02 and 1.32 Ϯ 0.09 pmol/min/mg of protein, respectively) in extracts of whole fat pads.
In the G␣ q -deficient cells, the impaired lipolytic response stimulated by norepinephrine was sensitive to the strict ␤-adrenergic antagonist propranolol, reflecting a residual ␤-adrenergic, cyclic AMP-mediated response (Fig. 4A). Adipocytes from control mice display sensitivity to both propranolol and the ␣ 1 -adrenergic antagonist prazosin. The former reflects the ␤-adrenergic response acting via cyclic AMP, while the latter reflects this newly discovered ␣ 1 -adrenergic response first detected through its loss in the G␣ q -deficient cells. Vasopressin (1 M), which activates PLC, also stimulated lipolysis in adipocytes from control mice 1.8-fold over basal. The ␣ 1 -adrenergic stimulation of lipolysis was abolished by prazosin, but not by FIG. 2. Suppression of G␣ q expression by RNA antisense to G␣ q causes increased body weight and fat mass. The pPCK-ASG␣ q transgenic mice have increased body weight (panels A and B), increased white fat mass (panel C), and increased adiposity (panel D). Five founders and the progeny of each of the founder lines for 10 generations were maintained and propagated by independent outbreeding to BDF1 control mice. The five founders and all members of their lineages at each generation display this phenotype (see Table I). In all cases, the data displayed were obtained from members of at least three of the five founder lines. Transgenic mice and their littermate controls were analyzed at ages spanning 3-24 weeks. The data are expressed as the mean values Ϯ S.E. from at least six animals for each age group, transgenic and control alike. Fecundity and litter size were no different in the transgenic as compared to control mice. Mice shown in panel A were 12-week-old males, left-hand panel, transgenic mouse and right-hand panel, control mouse. Panel B, the body weight of pPCK-ASG␣ q transgenic mice and their control littermates, segregated by sex. Panel C, the fat pad mass of pPCK-ASG␣ q transgenic mice and their control littermates, pooled from male and female mice at the ages indicated for the sake of simplicity. Pair-feeding of the animals did not diminish the obese character of the transgenic mice. Nose-to-anus dimensions were not altered in the transgenic as compared to control mice. Adiposity was measured by determining the total white fat cell number from collagenase-digested, isolated epididymal and epoophoronal fat pads of single transgenic mice and paired, littermate controls, by cell counting using a hemocytometer. Throughout this work, statistical analysis was performed using the Student's t test. An asterisk denotes statistical significance with p Յ 0.05 for the difference between the mean values for transgenic (G␣ q -deficient) as compared to control mice. propranolol (Fig. 4B). The loss-of-function G␣ q -deficient cells, in contrast, have essentially lost the lipolytic response to phenylephrine stimulation.
Although G␣ q is not known to regulate adenylyl cyclase directly, the loss-of-function G␣ q mutants displayed impaired ␤-adrenergic stimulation of cyclic AMP accumulation and lipol- were measured in cells from mice 18 -24 weeks of age. Intracellular IP 3 accumulation was measured by the mass assay employing the rabbit cerebellar IP 3 -binding protein. DAG was assayed using a DAG kinase assay followed by thin-layer chromatographic separation of radiolabeled phosphate generated by the reaction. The DAG mass was calculated from a standard curve using authentic DAG. Lipolysis (panel C) and cyclic AMP accumulation (panel D) were measured in cells isolated from transgenic and control mice. The cells were challenged with varying concentrations of the mixed ␣and ␤-adrenergic agonist norepinephrine (NOR), the ␤-adrenergic agonist isoprenaline (ISO), or the ␣-adrenergic agonist phenylephrine (PHEN) for 15 min (cyclic AMP accumulation) or 60 min (lipolysis via glycerol release). For cyclic AMP determinations, the assays were terminated at 15 min, and the accumulation of intracellular cyclic AMP measured using a competitive binding assay with bovine adrenal cyclic AMP binding protein. The data presented are mean values Ϯ S.E. from at least three separate experiments, each performed on separate occasions. An asterisk denotes statistical significance with p Յ 0.05 for the difference between the mean values for transgenic (G␣ q -deficient) as compared to control mice. ysis. G␣ q , acting via PLC to promote IP 3 and DAG accumulation, may augment the cyclic AMP response indirectly, perhaps via effects on calcium-or protein kinase C-sensitive forms of adenylyl cyclase (22)(23)(24)(25). We tested the role of protein kinase C using bis-indolylmaleimide and calphostin C, selective inhibitors of protein kinase C (Fig. 4C). Both calphostin C (100 nM) and bis-indolylmaleimide (1 M) abolished the ␣ 1 -adrenergic stimulation of lipolysis, whereas the protein kinase A inhibitor KT-5720 (1 M) was without effect. At 100 nM, bis-indolylmaleimide effectively blocked phenylephrine (10 M)-stimulated lipolysis in adipocytes from control mice; glycerol release (mol/10 6 cells), in response to this ␣ 1 -adrenergic agonist, declined from 4.7 to 1.2 in the absence versus presence of this protein kinase C inhibitor. The K I for nonselective inhibition of protein kinase A by bis-indolylmaleimide is Ͼ2 M (26). Since the protein kinase A inhibitor KT-5720 itself was without effect, nonselective effects of protein kinase C inhibitors, if they indeed occurred at these lower concentrations, would be irrelevant. FIG. 5. Adipocytes from pPCK-ASG␣ q transgenic mice display loss of function with respect to activation of protein kinase C following challenge with norepinephrine. White adipocytes were isolated from epididymal and epoophoronal fat of transgenic mice and control littermates by collagenase digestion. The activity of protein kinase C was measured in cells from mice 18 -24 weeks of age. The cells were challenged for 5 min without (Basal) and with 10 M of the mixed ␣and ␤-adrenergic agonist norepinephrine (NOR), in the absence or presence of 0.1 mM bis-indolylmaleimide (NOR ϩ BIS), a potent protein kinase inhibitor. Protein kinase C activity was measured in DEAEcellulose-purified cell homogenates, as described elsewhere (30). The data presented are mean values Ϯ S.E. from at least three separate experiments, each performed on separate occasions. An asterisk denotes statistical significance with p Յ 0.05 for the difference between the mean values for transgenic (G␣ q -deficient) as compared to control mice.
for study of the lipolytic response to adrenergic agonists. The lipolytic response was measured as described in the legend to Fig. 3 The results are mean values Ϯ S.E. from three to five separate experiments for each. An asterisk denotes statistical significance with p Յ 0.05 for the differences from mean basal values obtained with adipocytes isolated from both transgenic (G␣ q -deficient) and control mice.
FIG. 4. The pharmacology of the adrenergic lipolytic response reveals the existence of an ␣ 1 -adrenergic stimulatory pathway, absent in the G␣ q -deficient loss-of-function mutants. White adipocytes were isolated from transgenic mice and their control littermates Measurement of protein kinase C activity in DEAE-cellulosepurified homogenates of cells challenged with and without norepinephrine was performed using adipocytes from the control and transgenic mice (Fig. 5). In adipocytes from control mice, norepinephrine stimulates protein kinase C activity, an action blocked by the addition of bis-indolylmaleimide. Suppression of G␣ q in adipocytes of the pPCK-ASG␣ q mice results in a frank reduction in protein kinase C activity in the basal state and a loss of norepinephrine-induced activation of protein kinase C. Total protein kinase activities for adipocytes from control and transgenic mice are equivalent, 310 Ϯ 20 and 315 Ϯ 18 pmol/min/million cells, respectively. Thus, ␣ 1 -adrenergic control of lipolysis is shown to be mediated via protein kinase C, a pathway revealed by its absence in the G␣ q -deficient state.
The absence of G␣ q resulted in increased fat accumulation and hyperadiposity, observed within 5 weeks of age and sustained through adult life. Obesity has been reported in transgenic mice after genetic ablation of brown adipose tissue (27), supporting the role of this specialized tissue in preventing obesity (28). The pPCK-ASG␣ q transgene was not expressed in brown adipose tissue (not shown). Expression of G␣ q , the uncoupling protein UCP, and the mRNAs for both were equivalent in brown adipose tissue from transgenic and control mice (not shown), suggesting no involvement of brown adipose tissue in enhanced fat accumulation by the pPCK-ASG␣ q -expressing mice.
The absence of G␣ q abolished an important stimulatory control of lipolysis, apparently predisposing the mice to accumulation of fat. Recently, G-proteins have been shown to play prominent roles in differentiation (2, 3) and neonatal growth (4 -6). For progression of F9 teratocarcinoma cells to primitive endoderm (4) and for development of nerve growth cones (3), G-proteins appear to be acting directly or indirectly via protein kinase C (10,29). In the present study we demonstrate the key role of PLC and protein kinase C in adipocyte signaling in the mature cells. The basis for the hyperadiposity in the cells deficient in G␣ q , however, remains to be established, but may reflect a critical role of G␣ q in controlling adipogenic conversion in vivo.