1154 Diabetes Volume 64, April 2015

Kenneth J. McCreath,1 Sandra Espada,1 Beatriz G. Gálvez,1 Marina Benito,2,3 Antonio de Molina,4 Pilar Sepúlveda,5 and Ana M. Cervera1,5

Targeted Disruption of the SUCNR1 Metabolic Receptor Leads to Dichotomous Effects on Obesity

Diabetes 2015;64:1154–1167 | DOI: 10.2337/db14-0346

A number of metabolites have signaling properties by G-–coupled receptors (GPCRs) constitute the major acting through G-protein–coupled receptors. Succinate, and most diverse group of membrane receptors in a Krebs cycle intermediate, increases after dysregu- eukaryotes (1). Found in the plasma membrane, these lated energy metabolism and can bind to its cognate receptors are tasked with the recognition and transmis- receptor succinate receptor 1 (Sucnr1, or GPR91) to ac- sion of messages from the external environment. An ever- tivate downstream signaling pathways. We show that increasing number of GPCRs are now being identified as Sucnr1 is highly expressed in the white adipose tissue receptors for metabolites or energy substrates (2), (WAT) compartment of mice and regulates adipose expanding the repertoire of biological targets to diseases 2/2 mass and glucose homeostasis. Sucnr1 mice were associated with the metabolic syndrome, including hyper- generated, and weight gain was monitored under basal tension and type 2 diabetes (3). Metabolites such as lac- and nutritional stress (high-fat diet [HFD]) conditions. tate (4), ketone bodies (5), and a-ketoglutarate (6), 2/2 METABOLISM On chow diet, Sucnr1 mice had increased energy known primarily to serve functions distinct to signaling, expenditure, were lean with a smaller WAT compart- have been shown to act as ligands for GPCRs. Represen- ment, and had improved glucose buffering. Lipolysis tative of this new class of receptor is Gpr91, a receptor for measurements revealed that Sucnr12/2 mice were re- the Krebs cycle intermediate succinate, now termed leased from succinate-induced inhibition of lipolysis, Sucnr1 for succinate receptor (7). The Krebs, or citric demonstrating a function of Sucnr1 in adipose tissue. acid, cycle occurs at the junction between glycolysis and Sucnr1 deletion also protected mice from obesity on oxidative phosphorylation (8) and is a key component of HFD, but only during the initial period; at later stages, aerobic metabolism by providing reducing equivalents for body weight of HFD-fed Sucnr12/2 mice was almost the electron transport chain. Interestingly, succinate has comparable with wild-type (WT) mice, but WAT content been known for many years to increase in tissues during was greater. Also, these mice became progressively hypoxia (9–11) and may act as a surrogate marker for hyperglycemic and failed to secrete insulin, although reduced oxygenation. Accordingly, Sucnr1 may be acti- pancreas architecture was similar to WT mice. These vated in times of hypoxic stress such as ischemic injury findings suggest that Sucnr1 is a sensor for dietary in liver (12) or retina (13) or during cardiomyocyte crisis energy and raise the interesting possibility that proto- (14). Moreover, dysregulated metabolism, akin to diabetes cols to modulate Sucnr1 might have therapeutic utility or the metabolic syndrome, also has the potential to in- in the setting of obesity. crease peripheral blood concentrations of succinate to

1Department of Cardiovascular Development and Repair, Fundación Centro Corresponding author: Ana M. Cervera, [email protected] or amcerveraz@ Nacional de Investigaciones Cardiovasculares Carlos III, Madrid, Spain gmail.com. 2 Advanced Imaging Unit, Department of Atherothrombosis, Imaging, and Epidemi- Received 28 February 2014 and accepted 14 October 2014. ology, Fundación Centro Nacional de Investigaciones Cardiovasculares Carlos III, This article contains Supplementary Data online at http://diabetes Madrid, Spain .diabetesjournals.org/lookup/suppl/doi:10.2337/db14-0346/-/DC1. 3CIBER de Enfermedades Respiratorias, Madrid, Spain 4Comparative Medicine Unit, Fundación Centro Nacional de Investigaciones Cardio- © 2015 by the American Diabetes Association. Readers may use this article as vasculares Carlos III, Madrid, Spain long as the work is properly cited, the use is educational and not for profit, and 5Regenerative Medicine and Heart Transplantation Unit, Instituto de Investigación the work is not altered. Sanitaria La Fe, Valencia, Spain diabetes.diabetesjournals.org McCreath and Associates 1155 levels where activation of the Sucnr1 receptor will occur with genomic DNA extracted from tail biopsies using the (15,16). Indeed, the low levels (;2–80 mmol/L) of succi- REDExtract-N-Amp Tissue PCR kit (Sigma-Aldrich, Saint nate found in blood plasma (15–18), in comparison with Louis, MO). Genotyping primers were primer 1, 59-ACTA that required for activation (in the region of 180 mmol/L TAAGCATCACTTCACCACTTCC-39; primer 2, 59-CAACGG [2]), would suggest that Sucnr1 functions more as a sensor GTTCTTCTGTTAGTCC-39; and primer 3, 59-AAGAACAA of metabolic (or oxidative) damage (19) rather than ATGGTAGAACAGTCACGG-39. A PCR product of ;500 a physiological mediator of signaling. nucleotides (nt) was obtained from WT DNA, whereas Mounting evidence points to an important role for mutant DNA produced an amplification product of Sucnr1 during periods of metabolic dysfunction, such as ;320 nt in size. These KO-first mice were designated GT diabetes-related hypertension, oxygen-induced retinopa- mice. In order to remove the critical exon of the Sucnr1 , thy, and the metabolic syndrome (19). Interestingly, together with the neomycin cassette, GT mice were crossed obesity is a key component of many metabolic-related with Sox2-Cre mice (supplied by Dr. Miguel Torres, Centro diseases (20) and can lead to diabetes, hyperlipidemia, Nacional de Investigaciones Cardiovasculares [CNIC]) to pro- and cardiovascular disease (20,21). The effect of Sucnr1 duce whole-body KO mice. Genotyping was performed deficiency on body composition and metabolism has not as before using the primer 4, 59-ACTTCCAGTTCAACAT previously been studied. We reasoned that Sucnr1 could CAGCCGCTACA-39;primer5,59-GTGATTCATCTGTATTAT be a potential modulator of metabolism since a recent TAGATGAGC-39; and primer 6, 59-CAGCACAACCATCAGA study indicated that this receptor participates in the GAAACAATGGACTC-39. A PCR product of ;650 nt indi- breakdown (lipolysis) of stored triglycerol (TG) in fat cated that recombination had occurred to produce a KO (22). To test this hypothesis, we analyzed the expression , and a product of ;300 nt identified a WT locus. All of Sucnr1 in mouse adipose tissue and observed high experiments were carried out using these KO mice. expression in the white adipocyte compartment. To gain Mice were housed at the CNIC animal facility under insight into the (patho)physiological role of Sucnr1, we pathogen-free conditions. Male mice were used for all created mice deficient for this gene using a knockout studies. Mice were given ad libitum access to food and (KO)-first approach. We show that deficiency of the water unless otherwise noted in experimental methods. Sucnr1 gene has metabolic consequences for weight gain Environmental conditions were controlled to 12/12-h and glucose homeostasis. Sucnr1 KO mice have a modest light/dark cycles. Animal studies were approved by the lean phenotype under standard diet (SD) conditions, cor- local ethics committee. All animal procedures conformed relating with a reduction in the size of white adipose to European Union Directive 86/609/EEC and recommen- tissue (WAT) depots, increased energy expenditure, and dation 2007/526/EC regarding the protection of animals an improved glucose clearance rate. Challenge with high- used for experimental and other scientific purposes, fat diet (HFD) also resulted in a significant decrease in enacted under Spanish law 1201/2005. Male Sucnr1 KO body weight in Sucnr1 KO mice, but only during initial mice and WT mice (5–6 weeks old) obtained from hetero- periods. At later periods, body weight of Sucnr1 KO mice zygous mating were fed an SD of rodent chow, containing was indistinguishable from wild-type (WT) littermates, 13.4% fat (16% energy; Diet 5K67, LabDiet, Saint Louis, but WAT content was increased. This appeared to coincide MO) or an HFD containing 34.9% fat (60.9% energy; with a hyperglycemic profile and a reduction in insulin Diet58Y1,TestDiet,SaintLouis,MO).Bodyweight secretion from the pancreas. Collectively, our results was recorded weekly. Separate cohorts of animals on point to a potentially important role for Sucnr1 in meta- SD and HFD were used for metabolic cage analysis using bolic disease related with obesity. a 16-chamber indirect calorimetry system (PhenoMaster; TSE Systems, Bad Homburg, Germany). Mice were indi- RESEARCH DESIGN AND METHODS vidually housed in the chambers for 5 days; the first Generation of Sucnr1-Mutant Mice, Feeding Regimes, 2 days were used for acclimatization, and data were ana- and Indirect Calorimetry lyzed from the last 3 days. Oxygen consumption rate, To create Sucnr1-mutant mice, we obtained two Sucnr1- carbon dioxide production rate, feeding, and total loco- targeted embryonic stem (ES) cell clones from the European motive activity were measured concurrently for each Mouse Mutagenesis Consortium (EUCOMM), designated as mouse. KO-first, conditional-readyEScells(23).Thisflexible system allows both for gene-trap (GT) reporter constructs and also MRI and Spectroscopy Analysis for conditional and null alleles to be generated by exposure Imaging was performed in the Advanced Imaging Unit at to the site-specificrecombinasesCreandflippase (www CNIC. Mice on SD or HFD were preanesthetized by .knockoutmouse.org). ES cells were injected into C57BL/6 inhalation of isoflurane/oxygen (2–4%) and monitored by blastocysts, and chimeric males were bred to C57BL/6 electrocardiogram and breathing rhythm (Model 1025; SA females to produce germ line transmission. Heterozygous Instruments, Inc., NY). Animals were positioned supine in animals were intercrossed, and genotyping of F2 progeny a customized bed with a built-in nose cone supplying inha- detected homozygous mice at the expected Mendelian ratio. latory anesthesia (1–2%) and kept at 35–37°C by a warm Genotyping was performed after weaning by PCR assay airflow. Images were acquired with an Agilent/Varian 7 1156 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015

Tesla system equipped with a DD2 console and an active Blood was collected before and 15 min after injection. shielded 205/120 gradient insert coil with 130 mT/m max- Serum nonesterified fatty acid (NEFA) was measured imum strength and a volume coil (Rapid Biomedical GmbH, using the NEFA-HR(2) kit from Wako Chemicals Rimpar, Germany) for transmit/receive. For MRI, data were (Richmond, VA). collected with a two- and three-dimensional (3D) spin-echo Tissue Collection and Blood Analysis sequence. T1 weighted two-dimensional images (repetition At the end of the study, mice were killed following an time 380 ms, echo time 13 ms, number of averages 4) were overnight fast. Blood samples were collected by cardiac acquired to visualize the fat qualitatively in high resolution. puncture, and serum was separated by centrifugation. 3D T1 weighted spin-echo acquisition was also acquired Total serum cholesterol, triglycerides, free fatty acids, to calculate the body fat–to–total body weight ratio using hepatic aspartate aminotransferase (AST), and alanine 100 ms repetition time, 8 ms echo time, 1 average. In both aminotransferase (ALT) were determined using an auto- sequences, the field of view was 90 3 45 3 45 mm3 and mated analyzer (Dimension RxL Max Integrated Chemistry the reconstruction matrix was 256 3 256 3 256. For System; Siemens, Munich, Germany) and commercially MRS, the 13Cand1H surface coil was placed over the available kits from Siemens (total cholesterol, triglycerides, epididymal fat. 13C spectra were acquired with a repeti- ALT, and AST) and Wako Chemicals (free fatty acids). tion time of 1,000 ms, 512 accumulations, and a spectral Diabetes-related biomarkers in serum were analyzed using width of 18 kHz. The 1H spectra were acquired with 1,000 the Bio-Plex Pro Mouse Diabetes standard 8-plex and ms of repetition time, 256 scans, and a spectral width of adiponectin kits (Bio-Rad Laboratories Inc., Berkeley, CA). 10 kHz. Images were processed using OsiriX (Pixmeo, Selected tissues were collected and weighed. Tissues Geneva, Switzerland). Quantification of fat composition were fixedinneutralbufferedformalinsolutionand from 3D images was estimated applying a semiautomatic processed for histological analysis or snap frozen in image thresholding segmentation plug-in included in liquid nitrogen and stored at 280°C for RNA and pro- the software and referenced to the total (body) volumes tein extraction. calculated by the same approach. 1Hand13C spectra were processed with MestReNova (Mestrelab Research, Histological Analysis Santiago de Compostela, Spain). An exponential line broad- Tissues were fixed in 10% buffered formalin (VWR In- ening (20 Hz for 13Cspectrumand3Hzfor1H spectrum) ternational, Radnor, PA), embedded in paraffin, and sec- was applied before Fourier transformation. Individual tioned at 5 mm. For morphometric analysis, sections were contributions of the signals from carbonylic, olefinic, stained with hematoxylin and eosin according to standard methylene, and methyl moieties were quantified by inte- protocols. Adipocyte size (from WAT) and islet cell area and gration after automatic baseline correction. The same pro- cell number (from pancreas) were determined with ImageJ cess was applied in 1H spectrum. The areas under each software (National Institutes of Health). Sections from six peak were used to calculate the fractional contribution animals per group were used, and at least 500 adipocytes of saturated (130 ppm), monounsaturated (130 and per animal were measured from multiple fields. Islet cell 128 ppm), and doubly (poly) unsaturated (128 ppm) fatty area was quantified from at least four sections per pancreas acids, taking as a reference the area under the carbonylic and expressed as the percentage of total pancreas area. Islet peak (24). The areas under the peak in the 1Hspectrum, cell density was assessed by counting the number of nuclei total lipids (1.3 ppm) and water peak (4.7 ppm), were used in a 1,720 mm2 areafromthesamesections.Additional to calculate the percentage of total lipids with respect to sections were subjected to heat-induced epitope retrieval water content (25). in citrate-buffer, pH 6, followed by blocking of endogenous peroxidase activity with hydrogen peroxide prior to the Ex Vivo Lipolysis Assay immunostaining. Sections were incubated overnight with Mouse epididymal adipose tissue was dissected, weighed, antibodies specific for Sucnr1 (Novus Biologicals, Littleton, divided into equal pieces, and suspended in 400 mLKrebs- CO), insulin (Cell Signaling Technology Inc., Danvers, MA), Ringer bicarbonate buffer (Sigma-Aldrich) containing 1% glucagon (Abcam plc, Cambridge, U.K.), Ki67 (Master Diag- fatty acid-free BSA (Sigma-Aldrich). The tissue samples nostica, Granada, Spain), or caspase-3 (R&D Systems Inc., were treated with either PBS or 25 nmol/L isoproterenol Minneapolis, MN). For Sucnr1 immunodetection, a tyramide or pretreated with 100 mmol/L succinate (for 10 min) or signal amplification kit (Molecular Probes; Thermo Fisher 100 ng/mL pertusis toxin for 4 h before isoproterenol Scientific, Waltham, MA) was used. For insulin/glucagon treatment. Samples were incubated on a shaker at 37°C, double staining, we used fluorescence-labeled secondary and glycerol release was measured using a Glycerol Deter- antibodies. Insulin, Ki67, and caspase-3 were detected mination kit (Cayman Chemical Company, Ann Arbor, MI). using an automatic immunostainer (Dako Autostainer Plus; Dako, Glostrup, Denmark) with a horseradish- In Vivo Lipolysis Assay peroxidase-labeled polymer anti-rabbit as secondary After an overnight fast, mice were injected intraperitoneally antibody (EnVision-HRP; Dako) and the chromogenic with succinate (1.2 mg/kg) or the ago- substrate DAB (Dako), followed by counterstaining nist (R)-N6-(2-phenylisopropyl)-adenosine (PIA; 0.15 nmol/g). with Mayer’shematoxylin. diabetes.diabetesjournals.org McCreath and Associates 1157

Western Blot Analysis culture medium supplemented with only insulin for the For examination of phospho-Akt in WAT, skeletal muscle, remaining duration of differentiation (a further 4 days). and liver samples, fasted mice were injected intraperito- For flow cytometry analysis of adipocytes, the same en- neally with saline or insulin (1.5 units/kg body mass). zyme dissociation procedure was followed for epididymal Extracts were prepared at 10 min postinjection as de- WAT digestion. Then adipocytes were recovered by cen- scribed (26). Briefly, tissues were homogenized in 500 mL trifugation (250g, 10 min, 4°C), collected in 1 mL of ice- ice-cold lysis buffer (1% Triton X-100, 150 mmol/L NaCl, cold PBS containing 0.25 mL of Draq5 (BioStatus Limited, 10 mmol/L Tris-HCl pH 7.5, 1 mmol/L EDTA, 1% Nonidet Leicestershire, U.K.) to stain the cells, and analyzed on P-40, and complete mini protease inhibitor cocktail; Roche, a LSR Fortessa flow cytometer (Becton Dickinson, San Mannheim, Germany) with a mechanical homogenizer, and Jose, CA). Data were analyzed with FlowJo 1.6 software. debris was cleared by centrifugation at 10,000g at 4°C for Triglyceride Measurement 10 min. The supernatant was collected and the quantity of Hepatic and WAT triglyceride content was measured using protein was measured with the BCA kit (Thermo Fisher tissue from mice fasted overnight as previously described Scientific). Protein extracts (40 mg of protein) were exam- (28). Briefly, total lipids were extracted from samples ined by protein immunoblot analysis by probing with anti- (50 mg) by using an 8:1 mixture of chloroform and methanol bodies to Akt and phospho-Ser473 Akt (Cell Signaling (4 h). The extracts were mixed with 1 N sulfuric acid and Technology Inc.). Immunocomplexes were detected with centrifuged (10 min). Triglyceride content was measured the enhanced chemiluminescence reagent (Amersham, GE with a kit purchased from bioMérieux Inc. (Marcy l’Etoile, Healthcare Bio-Sciences AB, Uppsala, Sweden). France). Oil Red O staining and measurement of intracel- Glucose Tolerance Test and Insulin Tolerance Test lular triglyceride levels were performed as previously de- For the intraperitoneal glucose tolerance test (IPGTT), mice scribed (27). Fixed cells or frozen liver sections (10 mm) were fasted overnight (16–18 h). Conscious mice were were stained with Oil Red O (Sigma-Aldrich) for 15–20 min then injected intraperitoneally with glucose (1.5 g/kg at room temperature. For quantification, cell monolayers body weight). Blood glucose was monitored in samples were washed extensively with water to remove unbound obtained from tail punctures before and after glucose dye, and 1 mL of isopropyl alcohol was added to the administration using a handheld glucometer (Ascencia culture dish. After 5 min, the absorbance (510 nm) of Elite; Bayer AG, Leverkusen, Germany). To measure in- the extract was measured by spectrophotometry. sulinlevelsinresponsetoglucose,bloodwascollected Total Pancreatic Insulin Content at 0 and 10 min postinjection through the submaxilar Total pancreatic insulin content was measured as de- vein, and the resulting serum was used to measure in- scribed (29). Pancreata were harvested from fed mice, sulin by ELISA (Millipore Corporation, Billerica, MA). Polytron homogenized (30 s) on ice in 5 mL cold 2 N For the intraperitoneal insulin tolerance test (IPITT), acetic acid, boiled for 5 min, transferred to ice, and centri- mice were fasted for 2 h and injected with insulin fuged at 15,000g and 4°C for 15 min. Supernatant (0.75 units/kg). extracts were then assayed for protein content and insulin Flow Cytometry Analysis, Stromal Vascular Fraction content (ELISA, Millipore Corporation). Isolation, and Preadipocyte Differentiation RNA Isolation and Gene Expression Analysis The stromal vascular fraction (SVF) was isolated from Total RNA extraction, cDNA synthesis, and quantifica- WAT by incubating the tissue for 1 h at 37°C in Krebs tion were performed by standard methods. The expres- buffer (1 g tissue/3 mL solution) containing 2 mg/mL sion of mRNA was examined by quantitative reverse collagenase A (Roche Diagnostics, Basel, Switzerland) and transcription–polymerase chain reaction (qRT-PCR) us- 20 mg/mL fatty acid–free BSA (Sigma-Aldrich). Digested ing a 7900 Fast Real Time thermocycler and Fast SYBR tissue was centrifuged at 250g, and the SVF pellet and Green assays (Applied Biosystems, Thermo Fisher Scien- adipocyte fractions were collected for RNA extraction. tific). Three technical replicates were performed per sam- For preadipocyte differentiation, the same procedure was ple, and relative expression was normalized against followed using subcutaneous WAT. Once the SVF fraction mouse Srp14. Primer sequences are given in Supplemen- was collected, SVF cells were differentiated to adipocytes ex tary Table 1. vivo using a modification of a published protocol (27). Cells were plated (50,000 cells/cm2) and maintained in Statistical Analysis culture medium (DMEM/F12, 8% FBS, 2 mmol/L L-glutamine, Calculations were performed using Biogazelle qBase+ and 50 units/mL penicillin, and 50 mg/mL streptomycin) for GraphPadPrism5.0software(SanDiego,CA)andreportedas 5 days, and differentiation was induced by incubating with mean 6 SEM. MRI and MRS data were reported as mean 6 culture medium supplemented with 5 mg/mL insulin, 1 standard deviation. Statistical comparisons were made using mg/mL dexamethasone, 25 mg/mL isobutylmethylxanthine, repeated-measures ANOVA with Bonferroni posttests for and 1 mmol/L rosiglitazone. After 2 days, cells were incu- comparisons between groups at individual time points and bated for an additional 2 days in culture medium containing Student t test for independent samples, as appropriate. Sig- insulin and rosiglitazone only and were then maintained in nificancewasestablishedasavalueofP , 0.05. 1158 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015

RESULTS after mating. All mice were on a C57BL/6 background. Sucnr1 Is Highly Expressed in WAT Mouse genotypes were evaluated by PCR using genomic Gene transcription analysis of C57BL/6 WT mouse tissues DNA from tail biopsies, with primers designed to detect revealed high Sucnr1 expression in kidney and liver, with correctly targeted events (Fig. 2B, primers 1, 2, and 3). GT/+ little or no expression in lung tissue (Fig. 1A). These Homozygous offspring from heterozygous Sucnr1 results are consistent with previous work demonstrating male and female intercrosses were viable and exhibited high Sucnr1 expression in kidney medulla (7) and also he- normal development. Whole-body KO mice, distinct patic stellate cells (12). Interestingly, significant expression from GT mice by the complete removal of exon 2, were was also detected in WAT from epididymal fat (Fig. 1A), produced following a crossing regimen with Sox2-Cre ratifying the recent grouping of this gene to an adipose mice. Removal of exon 2 and the neomycin module was fi cluster in the adult mouse (22). Analysis of fractionated con rmed by PCR using primers 4, 5, and 6, giving the fi WAT demonstrated Sucnr1 expression in the mature adi- expected ampli cation products for WT and KO (Fig. 2C). pocyte but not the stromal cell fraction, which contains Because intronic insertions of GT vectors may allow nor- preadipocytes (Fig. 1B). In contrast to WAT, Sucnr1 ex- mal mRNA splicing to some degree, we analyzed endoge- pression was minimal from brown adipose tissue (BAT) nous Sucnr1 levels in GT mice to determine whether deposits (Fig. 1C). residual transcripts were expressed. Using primers e1 and e2, which bridge exons 1–2 (Fig. 2A), PCR amplifica- Generation of Sucnr1GT/GT and Sucnr12/2 Mice tion of cDNA from different tissues of WT mice yielded To investigate the in vivo function of Sucnr1, we generated a single product with the expected size of ;190 nt Sucnr1-mutant mice by using KO-first targeted ES cell clones (Fig. 2D). Unexpectedly, amplification of cDNA from obtained from EUCOMM. The targeting vector, directed to Sucnr1GT/GT mice with the same primer pair yielded intron 1 of the Sucnr1 gene (Fig. 2A), contained a splice a larger single product of 300 nt, with no detectable WT acceptor (SA) module from the engrailed-2 (En2)gene,fol- Sucnr1 expression (Fig. 2D, left panel) suggesting that the lowed by an internal ribosome entry site, which directs SA is not skipped. Introducing an internal ribosome entry translation of a fusion reporter protein with b-galactosidase site–specific reverse PCR primer (p7), together with primer (LacZ),which,inturn,isdependentontranscriptionofthe e1, resulted in the amplification of an intermediate-sized endogenous Sucnr1 promoter. A separate transcription unit fragment (220 nt), consistent with the proper functioning contained a promoter-driven neomycin gene for antibiotic of the SA module (Fig. 2D, middle panel). PCR across exons selection. These two modules were flanked by flippase rec- 1–2inSucnr1KO/KO (Sucnr1 KO) mice failed to amplify ognition target sites, permitting flippase-mediated recombi- a reaction product, consistent with the loss of exon 2 nation. Additionally, exon 2 of the gene, which was included (Fig. 2D, right panel). To determine the origin of the un- in the targeting vector, was flanked by LoxP sites, allowing anticipated 300 nt product in Sucnr1GT/GT mice, and ex- conditional ablation of the exon following Cre-mediated re- clude the possibility of PCR artifacts, reactions were combination (30). A third LoxP site bordered the neomycin performed on mRNA from different tissues of mice, and cassette (Fig. 2A), ensuring its removal with exon 2. Thus, the products were sequenced. Results from sequencing de- depending on mating regimes, either whole-body KO mice termined precisely in all cases that the fusion transcript or conditional KOs can be produced (23). In this work, we generated by the proper joining of exon 1 to the SA se- describe only results from the GT and whole-body KO mice. quence had caused activation of a cryptic splice donor site Two targeted ES cells clones were injected into mouse located 115 nt downstream in the En2 sequence, which blastocysts, and germ line transmission was confirmed spliced to the SA of Sucnr1-exon 2 (Fig. 2E). This splicing

Figure 1—Sucnr1 tissue distribution. A: Relative expression of Sucnr1 in mouse tissues, determined by qRT-PCR. Expression levels (log10) were relative to that in lung tissue arbitrarily chosen as 1 (n = 3 mice). B: Relative expression of Sucnr1 in the SVF and adipocytes isolated from epididymal adipose tissue, examined by qRT-PCR. Adiponectin and macrophage scavenger receptor 1 served as positive controls for adipocyte and stromal fractions, respectively (n = 4 mice). C: Relative expression of Sucnr1 in WAT and BAT compartments, examined by qRT-PCR (n = 3 mice). Data represent mean 6 SEM. ADIP, adipocytes; Adp, adiponectin; H, heart; K, kidney; Li, liver; Lu, lung; Msr1, macrophage scavenger receptor 1. diabetes.diabetesjournals.org McCreath and Associates 1159

In accord with mRNA expression, strong and homog- enous b-galactosidase staining was observed throughout WATfromKOmicebutnotWTlittermates(Fig.2F,top panels), indicating robust b-galactosidase activity in this tis- sue driven by the endogenous Sucnr1 promoter. Subse- quently, immunofluorescence analysis of paraffin-embedded WAT sections revealed that Sucnr1 is present in the plasma membrane of WT adipocytes but is not detected in equiva- lent samples of KO mice (Fig. 2F, bottom panels). Collec- tively, these results identify the white adipocyte as a significant reservoir of Sucnr1 expression and suggest that it might have a functional role in adipocyte metabolism. Sucnr1 Deficiency Results in Release From Succinate- Induced Inhibition of Lipolysis Hydrolysis of fat stored as TG in adipose tissue, termed lipolysis, is tightly controlled by hormones and metabo- lites that are secreted according to nutritional status (33). Enzymatic breakdown of TG occurs through a series of steps (34) to release one molecule of glycerol and three molecules of fatty acid. Released fatty acids enter the circulation and can be taken up and oxidized by peripheral tissues. Interestingly, succinate has been shown to inhibit lipolysis in isolated adipose tissue, presumably through engagement with Sucnr1. To investigate whether lipolysis is perturbed in Sucnr1 KO mice, we dissected WAT from epididymal fat regions and measured lipolytic activity in explant tissue. Glycerol release, as a metric of lipolysis, was similar in WT and Sucnr1 KO mice, both under basal and b-adrenergic agonist (isoproterenol)-stimulated con- ditions (Fig. 3A). In accord with previous findings (22), addition of succinate (100 mmol/L) induced an inhibition of isoproterenol-stimulated glycerol release in WAT ex- Figure 2—Generation of Sucnr1-deficient mice. A: Diagram of the EUCOMM targeting vector including the LoxP-flanked exon and plants from WT mice, and pretreatment of adipocytes Sucnr1 WT locus. Primers for genotyping correct targeting events with pertussis toxin blocked the antilipolytic effect of are shown by large arrows (1–7); exon-spanning primers are shown succinate (Fig. 3B). As anticipated, inhibition of lipolysis by B C by small arrows. Black boxes represent exons. and : PCR anal- succinate was absent in equivalent WAT from Sucnr1 KO ysis of GT, KO, and WT alleles. D: Amplification of exons from mRNA produced from liver, kidney, and WAT of WT, GT, and KO mice (Fig. 3C). Moreover, intraperitoneal injection of mice. Larger product amplified with primers e1+e2 was extracted succinate (1.2 g/kg) provoked a significant decrease in for sequencing analysis. E: Schematic representation of alternative serum levels of NEFA in WT mice, whereas the decrease fi splicing around the KO- rst cassette. Correct splicing is denoted by in Sucnr1 KO mice was not significant (Fig. 3D). Further, a solid line; the alternative transcript is denoted by a broken line (not to scale). F: Representative b-galactosidase staining of intact adi- intraperitoneal injection of the adenosine A1 receptor ag- pose tissue from WT and KO mice (2003 magnification) demon- onist PIA could inhibit lipolysis both in WT and Sucnr1 strating high LacZ expression in KO mice, which was absent in WT KO mice (Fig. 3E), demonstrating that the response to mice. WAT was counterstained with nuclear fast red; immunohisto- chemical staining of WAT (2003 magnification) with an antibody to general GPCR-mediated antilipolysis was not perturbed SUCNR1 confirms its expression in WT but not KO tissue. ACT, in Sucnr1 KO mice. Taken together, these results establish actin; FRT, flippase recognition target; IR, internal ribosome entry that succinate exerts its antilipolytic actions through site; NEO, neomycin. Sucnr1 coupled to Gi-type G .

Disruption of Sucnr1 Results in a Moderately Lean resulted in a frameshift with premature stop codons within Phenotype With Reduced WAT Compartments exon 2, predicting termination of a truncated Sucnr1 trans- Given that expression of Sucnr1 in WAT alluded to an lation product. Curiously, this cryptic splice element in the involvement in energy homeostasis, we assessed weight En2 gene has been previously described in the literature gain in WT and Sucnr1 KO mice on an SD. Total body (31,32), and it seems rather puzzling that a large-scale ge- weight did not differ at weaning (13.1 6 1.1 g for male WT netic KO platform would use the En2-SA unit with a docu- vs. 13.2 6 0.6 g for male KO; n = 8), and both genotypes mented alternative splicing phenomenon. Because of this displayed similar weight gains, although a discernible incongruity, only KO mice were characterized further. trend for decreased weight gain in Sucnr1 KO mice was 1160 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015

Figure 3—Lipolysis, body weight measurement, fat distribution, and adipocyte content of Sucnr1-deficient mice A: Measurement of lipolysis in WAT explants from Sucnr1-deficient mice (KO) and WT littermates. Lipolysis was determined by measuring glycerol released from adipocytes in the absence (–) or presence (+) of 25 nmol/L isoproterenol for 4 h (n = 4 mice per group). Results are expressed as mg glycerol per liter released per mg tissue. B: WAT explants of WT mice were pretreated or not for 4 h with pertussis toxin (100 ng/mL21) prior to the addition of succinate (100 mmol/L), followed 10 min later by isoproterenol (25 nmol/L). Lipolysis was determined by glycerol release as in A. A representative experiment from four is shown. C: Lipolysis was measured in WAT explants from Sucnr1-deficient and WT littermates. Explants were incubated with isoproterenol (25 nmol/L) plus/minus succinate (100 mmol/L), and glycerol release was deter- mined after incubation for 3 h. Results show the percentage of activity with succinate (n = 4 mice per group). D: Fasted mice were injected intraperitoneally with succinate (1.2 g/kg). Serum NEFA was measured before and 15 min after injection (n = 6 mice per group). E: Fasted mice were injected intraperitoneally with PIA (0.15 nmol/g). Serum NEFA was measured before and 15 min after injection (n = 6 mice per group). F: Body weight changes in male Sucnr1-deficient mice (KO) and WT littermates fed an SD for 16 weeks (n =8–12 mice per group). G: Body weights of male Sucnr1-deficient mice (KO) and WT littermates fed an SD for 50 weeks (n =8–12 mice per group). H: WAT weight measurements after 20 weeks on diet (n =8–12 per group). I: Total fat weight taken as a percentage of body weight after 20 weeks on diet (n =8–12 per group). J: NMRI analysis of body fat content and quantification of several measurements (n = 7 mice per group). Results were expressed as proportion of fat volumes vs. total mouse volume (fat volume/total volume). K: Lipid quantification. 1H spectra of abdominal compartments from WT and Sucnr1 KO mice (n = 7 mice per group). Areas under the curve were used to calculate the portion of lipids with respect to water content. L: Representative hematoxylin and eosin–stained sections (2003 magnification) of paraffin-embedded epididymal WAT from mice after 18 weeks of diet and average adipocyte size (n = 8 mice per group). M: Flow cytometry analysis of epididymal adipocytes (n = 6 mice per group). N: Triglyceride (TG) content in epididymal fat from SD-fed mice (n = 6 mice per group). O: Relative expression of adipogenesis in epididymal WAT after 18 weeks on SD, determined by qRT-PCR analysis: Pparg, Cebpa, and Fabp4 (n = 6 mice per group). P: Representative images of Oil Red O–stained adipocytes differentiated from stromal vascular cells collected from WT and Sucnr1-deleted mice and absorbance measurement (510 nm) of Oil Red O staining (n = 4 mice per group). Q: Relative expression of adipogenesis genes in differentiated adipocytes from SD-fed mice, determined by qRT-PCR analysis; Pparg, Pref1, and Srebp1c (n =4 mice per group). Data represent mean 6 SEM. Body weight analysis was assessed by repeated-measures ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001. epi, epididymal; FSC-A, forward scatter-A; iso, isoproterenol; OD, optical density; PTX, pertusis toxin; ret, retroperitoneal; sub, subcutaneous; succ, succinate.

apparent from 10 weeks of diet (Fig. 3F), which was sig- to WT littermates (Fig. 3H). Further, compared with total nificant when measured at 50 weeks (41.6 6 1.5 g for WT body weight, cumulative fat content was significantly re- vs. 35.5 6 1.9 g for Sucnr1 KO mice; P , 0.05) (Fig. 3G). duced in Sucnr1 KO animals (Fig. 3I). Visual inspection of Analysis of body composition was performed by dissec- NMRI images of WT and Sucnr1 KO animals suggested tion of organs and fat pads, nuclear MRI (NMRI), and a greater percentage of abdominal fat mass in WT animals MRS. After 18–20 weeks on the diet, mice were killed and relative to Sucnr1 KO littermates (Fig. 3J), and quantifica- organ weights were measured. No differences between tion analysis of axial slices demonstrated a trend for reduc- genotypes were found in the weights of heart, spleen, liver, tion (Fig. 3J). In the same experimental setting, 13C-NMR and BAT at the end of the study (not shown); however, spectra of glycerides were recorded over the entire abdom- evident differences were found in the WAT depots. Accord- inal region (Supplementary Fig. 1A). Quantification of the ingly, retroperitoneal, subcutaneous, and epididymal fat spectra (24) showed that the fractional contribution of mass was significantly reduced in Sucnr1 KO mice relative saturated fatty acids to the total pool of glycerides was diabetes.diabetesjournals.org McCreath and Associates 1161 significantly increased in Sucnr1 KO animals (0.583 6 genes related to adipocyte differentiation. Although expres- 0.292 vs. 0.481 6 0.189; P , 0.05 for Sucnr1 KO vs. sion levels of the adipogenic transcription factors, peroxi- WT, respectively), whereas monounsaturated and polyun- some proliferator-activated receptor g (Pparg) and CCAAT/ saturated fatty acids were comparable between genotypes. enhancer-binding protein a (Cebpa), together with the late Finally, 1H-MRS were also acquired from the abdominal differentiation marker fatty acid binding protein 4 (Fabp4), region to assess lipid content (25). As depicted in repre- tended to be slightly lower in WAT taken from Sucnr1 KO sentative spectra of mice, the ratio of water to fat content mice, this decrease was not significant (Fig. 3O). Moreover, was altered in Sucnr1 KO mice relative to control litter- the differentiation ability of preadipocytes extracted from mates (Supplementary Fig. 1B), and area under the curve the SVF of adult mice was comparable between both measurements revealed a significant decrease in fat content groups, as measured by Oil Red O staining of neutral lipids in Sucnr1 KO mice (Fig. 3K). Collectively, these results are (Fig. 3P) and by expression of the adipocyte marker genes consistent with the fat weight analysis and strongly suggest Pparg, protein delta homolog-1 (Pref1), and SREBP-1c that Sucnr1 KO mice have a reduced WAT content. (Srebp1c), although a nonsignificant increase in Pparg In agreement with the decreased size of the different and Srebp1c expression was observed in adipocytes from anatomical WAT compartments in SD-fed Sucnr1 KO mice, Sucnr1 KO mice (Fig. 3Q). These results, taken together histological analysis of hematoxylin and eosin–stained with the findings of low levels of Sucnr1 expression in SVF, WAT established a significant decrease in adipocyte size a reduction in adipose tissue mass and adipocyte size, and relative to WT mice (Fig. 3L). Further, flow cytometry anal- decreased lipid content, collectively suggest that Sucnr1 de- ysis of disaggregated adipocytes from epididymal WAT letion does not alter adipogenesis but rather results in revealed a significant decrease in adipocyte size, measured decreased lipid accumulation and smaller adipocytes. as forward scatter (Fig. 3M). These changes were mirrored To further evaluate the metabolic changes caused by by a reduction in the concentration of adipose triglycerides Sucnr1 deletion, we measured energy expenditure by in- (Fig. 3N). As the decrease in adipose mass in Sucnr1 KO direct calorimetry. Interestingly, the rate of VO2 con- mice can result from a reduction in adipocyte size and/or sumption in the Sucnr1 KO group was consistently decreased differentiation, we examined the expression of higher than WT counterparts (Fig. 4A). The respiratory

Figure 4—Indirect calorimetry and UCP expression. Age-matched mice after 20 weeks of SD were individually placed in metabolic cages (PhenoMaster) and allowed to acclimatize for 48 h before readings were taken (n =7–8 mice per group). A: Mean whole-body oxygen 21 21 consumption rate. B: RER (VCO2/VO2). C: Mean food intake (g). D: Mean energy expenditure (kcal$h $kg ). E: Mean RER over dark and light cycles. F: Mean food intake over dark and light cycles. G: Mean energy expenditure over dark and light cycles. H: Total ambulatory activity (sum of horizontal and vertical counts) over dark and light cycles. Measurements were recorded on a cycle of dark (8:00 P.M. to 8:00 A.M.) and light (8:00 A.M. to 8:00 P.M.). I: Relative expression of UCP1, UCP2, and UCP3 in tissues (n =5–7 per group). Data represent mean 6 SEM. *P < 0.05; **P < 0.01; ***P < 0.001. D, dark; EE, energy expenditure; L, light; SM, skeletal muscle. 1162 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015

exchange ratio (RER; VO2/VCO2) of Sucnr1 KO mice was NMRI images indicated comparable abdominal fat volume also significantly higher than that of WT mice during the in both groups (Fig. 5F), and quantitative assessment of dark cycle (Fig. 4B and E), indicating that perhaps less tissue fat from axial slices revealed no differences (Fig. 5F). A fat is available for oxidation in these mice. Alternatively, more accurate quantification was carried out using the increased RER values may suggest an increase in 1H-MRS to determine lipid content. In this case, area under whole-body utilization of carbohydrates and protein the curve measurements revealed a significant increase in over lipids. Surprisingly, food intake was increased in fat content in Sucnr1 KO mice (Fig. 5G), in keeping with Sucnr1 KO mice compared with WT (Fig. 4C and F), the findings from the first cohort. and energy expenditure was greater during the dark cycle Results from indirect calorimetry at 8 weeks on HFD fi (Fig. 4D and G), with no change in locomotor activity revealed no signi cant differences in VO2 consumption or between the two genotypes (Fig. 4H). This increase in RER (Fig. 5H) in Sucnr1 KO mice compared with WT food intake together with the reduction in fat mass could control, although ingestion of the HFD reduced RER val- be due to evident increases in energy expenditure. Though ues in both groups relative to SD. Again, total food con- we were unable to find differences in BAT expansion be- sumption was modestly increased in Sucnr1 KO mice, tween WT and Sucnr1 KO mice, dysregulated energy ex- although this did not reach significance (Fig. 5H). Further- penditure often results in overexpression of uncoupling more, locomotor activity was not significantly different proteins (UCPs) to dissipate energy. However, analysis of between groups (data not shown). These results indicated UCP expression revealed comparable levels of UCP2 and that the initial slower rate of weight gain in Sucnr1 KO UCP3 in skeletal muscle and liver (Fig. 4I). Interestingly, mice under HFD was not attributable to alterations in although changes did not reach significance, there was energy expenditure, food intake, or activity. a trend for lower UCP1 expression in BAT tissue of Sucnr1 Changes in adipose mass were reflected in alterations KO mice, together with elevated levels of UCP1 and UCP2 of adipokine levels in serum of animals. Compared with expression in WAT tissue (Fig. 4I), suggesting perhaps WT littermates, the reduced quantity of adipose tissue in compensatory mechanisms. Collectively, these results sug- Sucnr1 KO mice under SD was paralleled by decreased gest that deletion of Sucnr1 results in substantial altera- circulating levels of the obesity indicator leptin (Fig. 5I). tions in energy expenditure and disposition. This finding is in keeping with the increased feeding of Sucnr1 KO animals on SD (Fig. 4C and F) and the satiety HFD Feeding Leads to Increased Fat Deposition in effect of this hormone (35). In contrast, HFD feeding Sucnr1 KO Mice and Hepatic Steatosis induced a .10-fold increase in serum leptin, which was We next challenged mice with HFD to provoke obesity. not significantly different between Sucnr1 KO and WT HFD feeding led to increased weight in both genotypes; mice (Fig. 5I). In addition, a small but significant decrease however, Sucnr1 KO mice gained weight at a slower rate in serum levels of the insulin-sensitizing hormone adipo- than WT littermates, which was significant after only nectin was observed for Sucnr1 KO mice under both di- ;3 weeks of feeding (25.7 6 0.3 vs. 28.6 6 0.5 g; P = etary regimes (Fig. 5I). However, serum levels of the 0.0006). Variance in body weight widened until week 14 adipokines resistin and plasminogen activator inhibitor- when differences became less pronounced and weight 1, although increased in HFD, were not statistically dif- curves were superimposable (Fig. 5A). Predictably, ani- ferent between genotypes (Fig. 5I). Obesity in humans mals on HFD were substantially heavier than SD-fed ani- and animal models is often associated with increased cir- mals at the end of the study (compare Fig. 5A and Fig. culating lipids and ectopic lipid accumulation in nonadi- 3F), and this correlated with an increase in the weight of pose tissues including liver (36). Consistent with this WAT and BAT compartments and also liver. Surprisingly, phenotype, both fasted serum levels of NEFAs and cho- although Sucnr1 KO mice on HFD were generally leaner lesterol were elevated in WT mice and Sucnr1 KO mice fed than WT counterparts, the weight of WAT stores at the HFD for 18 weeks, whereas TG concentrations remained end of the study was significantly increased (Fig. 5B), in the normal range (Fig. 5J). In accord with the measured whereas adipocyte size was comparable (Supplementary lipolytic rates in explants (Fig. 3A), no differences in Fig. 2). This finding was in contrast to that observed un- plasma parameters were observed between genotypes. Al- der SD, where Sucnr1 KO mice had significantly less WAT though no morphologic alterations were found in liver (Fig. 3H and I). In a second cohort of HFD-fed mice used between groups fed an SD, hepatic steatosis was induced for NMRI analysis, body weights of Sucnr1 KO mice after in both genotypes under HFD, evidenced by hematoxylin 20 weeks on diet were found to be significantly lower than and eosin staining of abundant vacuolar vesicles in liver equivalent WT mice (Fig. 5C), in keeping with the overall sections, which were confirmed as lipid droplets by Oil observations. Although no differences were found in the Red O staining (Fig. 5K). Additionally, hepatic triglyceride weight of different anatomical WAT compartments be- content in both genotypes was significantly increased un- tween groups in this cohort (Fig. 5D), the proportion of der HFD, consistent with the fatty changes to liver (Fig. 5L); fat as a percentage of total body weight was significantly however, no significant differences in liver TG content were higher in Sucnr1 KO animals, as determined by measure- found between Sucnr1 KO and WT mice under either diet. ment of combined fat depots (Fig. 5E). Inspection of Further, HFD levels of AST and ALT were found to be diabetes.diabetesjournals.org McCreath and Associates 1163

Figure 5—Analysis of mice under HFD. A: Body weight changes in male Sucnr1-deficient mice (KO) and WT littermates fed HFD for 16 weeks (n =8–12 mice per group). B: Organ weight measurements after 18 weeks on diet (n =8–12 per group). C: Body weight changes in male Sucnr1-deficient mice (KO) and WT littermates fed HFD for 20 weeks (n = 8 mice per group). D: WAT weight measurements after 20 weeks on diet (n = 8 per group). E: Total fat weight taken as a percentage of body weight after 20 weeks on diet (n = 8 per group). F: NMRI analysis of body fat content and quantification of several measurements (n = 7 mice per group). Results were expressed as proportion of fat volumes vs. total mouse volume (fat volume/total volume). G: Lipid quantification. 1H spectra of abdominal compartments from WT and Sucnr1 KO mice (n = 7 mice per group). Areas under the curve were used to calculate the portion of lipids with respect to water content. H: Indirect calorimetry of age-matched mice after 8 weeks of HFD was performed as in Fig. 3. Mean whole-body oxygen consumption, mean RER over dark and light cycles, and mean food intake over dark and light cycles. I: Serum levels of adipokines measured in fasted mice at end of study (n =8–12 mice per group). J: Serum levels of NEFAs, triglycerides, and total cholesterol from Sucnr1-deficient mice (KO) and WT littermates after 18 weeks on diets (n =8–12 mice per group). K: Representative hematoxylin and eosin–stained and Oil Red O sections of paraffin-embedded liver taken from WT and KO mice after 18 weeks on diets (1003 magnification). L: Hepatic triglyceride (TG) content after 16 h fasting (n =8–12 mice per group). M: Serum measurement of ALT and AST from Sucnr1-deficient mice (KO) and WT littermates after 18 weeks on HFD (n =8–12 mice per group). Body weight analysis was assessed by repeated-measures ANOVA. Data represent mean 6 SEM. *P < 0.05; **P < 0.01; ***P < 0.001. CHOt, total cholesterol; epi, epididymal; ret, retroperitoneal; sub, subcutaneous.

significantly increased in Sucnr1 KO mice compared with acid synthase (Fas) and stearyol CoA desaturase (Acc), WT mice, suggesting increased hepatocyte damage (Fig. 5M); were unaltered in HFD-fed mice. Similarly, expression of the origin of these differences is not known. Consistent with the fatty acid transporter CD36 was upregulated under theincreasedlipidloadinliver, expression analysis revealed HFD in both genotypes, although expression was signifi- changes in relevant genes (Supplementary Fig. 3), including cantly reduced in Sucnr1 KO mice compared with WT the lipogeneic transcription factor sterol regulatory binding littermates on SD. Additionally, a modest but significant protein 1c (Srebp1c), and its target stearyol CoA desaturase-1 increase in the expression of a second fatty acid trans- (Scd1), which were induced under HFD in both genotypes; porter, Fabp4, was found in Sucnr1 KO but not WT mice however, other genes involved in triglyceride synthesis, fatty under HFD diet, whereas expression of the fatty acid 1164 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015 transport protein 2 (Fatp2) was downregulated in Sucnr1 muscle to similar intensities in Sucnr1 KO and WT mice on KO mice under HFD. In contrast to these changes, no HFD (Fig. 6E). These results pointed to a defect in insulin differences were found between genotypes for the expres- secretion as the origin of the hyperglycemia in HFD-fed KO sion of genes related to fatty acid oxidation, although mice. To consider this, we repeated the glucose tolerance carnitine palmitoyltransferase 1a and 1b (Cpt1a and test and measured circulating insulin levels at baseline and Cpt1b) were downregulated under HFD in both geno- 10 min after a glucose load (1.5 g/kg). Whereas WT mice types. Collectively, these data indicate that although the showed a characteristic increase in insulin levels at t =10min, absence of Sucnr1 results in subtle changes to liver func- Sucnr1 KO mice failed to secrete insulin, although GLP-1 tion metrics compared with WT mice, HFD appeared to levels were similar in both genotypes (Fig. 6F). Collec- cause comparable steatosis. tively, these results imply that while Sucnr1 KO mice have significantly enhanced glucose tolerance under SD, Sucnr1 KO Mice Are Hyperglycemic Under HFD With the HFD causes a sufficient deficiency in insulin secretion Reduced Insulin Secretion to cause glucose intolerance. Although the metabolic phenotype of Sucnr1 KO mice Guided by these findings, we examined islet anatomy appeared modest, the evident differences in adiposity and and endocrine cell size in pancreatic tissue sections of weight gain prompted us to evaluate glucose and insulin HFD-fed mice (Fig. 6G). As expected, morphometric anal- homeostasis. These parameters were measured in mice ysis revealed that high-fat feeding increased islet cell size, after 16 weeks on diet. Compared with SD, HFD led to an and this increase was comparable in both genotypes increase in both fasted and fed levels of serum glucose in (Fig. 6H). Moreover, b-cell number was also similarly in- Sucnr1 KO mice and WT littermates (Fig. 6A). Serum creased in Sucnr1 KO and WT mice under HFD, compared glucose was comparable between genotypes in SD; how- with SD (Fig. 6I). Also, total (acid-extracted) pancreatic ever, despite their similar final body weights in HFD, insulin concentration was similar between genotypes on Sucnr1 KO mice had significantly higher levels of fasting HFD (Fig. 6J). Immunohistochemistry staining with an blood glucose, whereas fed glucose was unaltered (Fig. 6A, antibody to insulin showed that the morphologic features right panel). Nevertheless, Sucnr1 KO mice showed sim- of WT and KO islets were comparable (Fig. 6K, left pan- ilar levels of fasting insulin as those from WT mice in els), and detailed analysis of the islets revealed a normal HFD, which were elevated compared with those in SD cell architecture shared between genotypes, with positive (Fig. 6B). To directly assess whether glucose homeostasis immunostaining for glucagon (a-cells) in the periphery, was dysregulated in Sucnr1 KO mice, we performed glu- while the area of anti-insulin staining was observed cose tolerance tests and insulin tolerance tests. Interest- mostly in the central region (Fig. 6K, right panels). Exam- ingly, in the SD group, intraperitoneal challenge with ination of b-cell proliferation with an antibody to Ki67 1.5 g/kg glucose resulted in faster clearance in Sucnr1 indicated a similar and low (,1%) frequency of Ki67- KO mice, as glucose concentrations remained significantly positive cells in both WT and KO mice, and very few lower compared with WT mice (Fig. 6C, left panel). This is apoptotic (caspase-3 positive) b-cells were detected in ei- in keeping with the general leanness of KO animals and ther genotype (Fig. 6L). Thus our results are consistent higher RER values (Fig. 4B and E), and suggested im- with an insulin secretory defect in Sucnr1 KO mice. proved insulin sensitivity that promotes carbohydrate uti- lization. As anticipated, glucose clearance deteriorated in DISCUSSION the HFD group (Fig. 6D, left panel). However, in marked Impaired cellular metabolism is a defining hallmark of contrast to the findings under SD, Sucnr1 KO mice fed an many disease states and can affect homeostatic signaling HFD developed a clear impairment in glucose elimination processes. As the Krebs cycle is the central metabolic compared with WT mice, with manifest hyperglycemia pathway in aerobic organisms, it is particularly significant 120 min after glucose injection. Notably, this decreased that succinate, an intermediate metabolite of this path- ability to lower blood glucose was not apparent in Sucnr1 way, has an alternative occupation as a signaling molecule KO mice at 11 weeks on HFD (Supplementary Fig. 4), for Sucnr1 coupling. suggesting a progressive glucose intolerance concurring Information regarding the function of Sucnr1 is limited. with the increased WAT weight. Abnormal glucose ho- Previous studies have shown that Sucnr1 is distributed in meostasis could result from either a defect in insulin sen- organs with high metabolic demand, including the kidney sitivity of the peripheral tissues or a defect in insulin and liver (19). Signaling of Sucnr1 appears to be an early secretion. Insulin tolerance tests showed that although response to hyperglycemic conditions in the kidney and both genotypes exhibited decreased insulin sensitivity un- leads to activation of the intrarenal renin-angiotensin sys- der HFD, Sucnr1 KO mice exhibited comparable insulin tem, resulting in hypertension and potential tissue injury sensitivity to WT littermates, as demonstrated by a quick (37). Here we show that Sucnr1 is expressed in high levels decline in plasma glucose after an acute insulin injection in the adipocyte and that loss of Sucnr1 leads to dichoto- (Fig. 6C and D, right panels). Furthermore, insulin was mous effects on body weight and metabolism. Although shown to induce phosphorylation of the major insulin- not conspicuously different during early growth, after signaling protein Akt (on Ser473) in liver, WAT, and skeletal 16–20 weeks on SD (at 21–25 weeks of age) Sucnr1 KO diabetes.diabetesjournals.org McCreath and Associates 1165

Figure 6—Sucnr1 KO mice show diet-dependent alterations in glucose homeostasis. A: Plasma levels of fasting (16 h) and fed glucose in Sucnr1-deficient mice (KO) and WT littermates under SD (left) and HFD (right) (n =8–12 mice per group). B: Plasma insulin levels after 16-h fast (n =8–12 mice per group). Plasma glucose during IPGTT (left) and IPITT (right) in SD (C) and HFD (D)(n =8–12 mice per group). E: Phosphorylation of Akt (Ser473) in liver, WAT, and skeletal muscle of HFD mice. After an overnight fast, mice were injected with PBS (–) or insulin (+) (1.5 units/kg) for 10 min. F: Serum insulin and GLP-1 concentrations measured from fasted WT and Sucnr1-deficient (KO) mice on HFD, measured after intraperitoneal glucose (1.5 g/kg) injection (n = 8 mice per group). G: Representative hematoxylin and eosin–stained sections of paraffin-embedded pancreas taken from WT and KO mice after 18 weeks on diets (1003 magnification), used to determine islet area and b-cell number. H: Quantification of islet size measured as a percentage of total pancreas area (n = 6 mice per group). I: Quantification of b-cell density was assessed by counting the number of nuclei in a 1,740-mm2 islet center area (n = 6 mice per group). J: Insulin content from acid-extracted pancreata of HFD-fed mice (n = 12 mice per group). K: Representative insulin immunostaining (403 magnification) of pancreas from Sucnr1-deleted mice and WT littermates on HFD (left panels). Immunofluorescence detection (1003 magnification) of insulin (green) and glucagon (red) in islets from Sucnr1-deleted mice and WT littermates on HFD (right panels). L: Representative Ki67 and caspase-3 staining of pancreas from Sucnr1-deleted mice and WT littermates on HFD. Islets are denoted by a dotted line, immunopositive cells are marked with an arrowhead. Data represent mean 6 SEM. IPGTT/IPITT analysis was assessed by repeated-measures ANOVA. *P < 0.05; ***P < 0.001. pAkt, phospho-Akt; prot, protein; S. muscle, skeletal muscle.

mice exhibited a modestly lean phenotype compared with adiposity was not associated with ectopic triglyceride stor- WT mice, with significantly reduced levels of plasma lep- age in liver or plasma (Fig. 5) or skeletal muscle (not tin and reduced WAT mass containing smaller adipocytes shown) but correlated with a significant increase in glu- and decreased triglyceride content. This decrease in cose tolerance and elevated metabolic activity, including 1166 SUCNR1 Effects on Obesity Diabetes Volume 64, April 2015 increased energy expenditure, which might partly explain energy availability (45), enhanced energy metabolism in why Sucnr1 KO mice were lean despite hyperphagia. How- peripheral tissues, or a defect in adipose tissue storage ever, this phenotype was seen in the setting of normal capability. It remains to be determined whether one or levels of BAT content and tissue expression of UCPs, more of these mechanisms are operating in our model. which are often elevated in models of increased energy Finally, it should be noted that an underlying concept expenditure (38). Decreased weight gain in Sucnr1 KO in Sucnr1 signaling is hypoxia-induced modulation of mice was also observed in the HFD study, particularly receptor engagement through local increases in succinate during the initial 10 weeks. This finding supports our levels. This is particularly manifest in ischemic retinopa- observations on SD and strongly suggests that leanness thy, where microvascular degeneration results in an is a primary outcome of Sucnr1 deletion. Since adipogen- abnormal hypoxia-driven succinate-induced neovasculari- esis was not affected by deletion of Sucnr1 and an anti- zation through Sucnr1-dependent upregulation of proan- lipolytic effect of succinate had been previously described, giogenic factors (13,46). Interestingly, hyperglycemia can this metabolically favorable phenotype might be explained, also result in regional increases in succinate (16,46), most in part, by the release from succinate-inhibited lipolysis. likely through overactivation of the Krebs cycle (47). In- Notably, experiments both in vitro and in vivo verified deed, adipose tissue succinate has been shown to increase that succinate inhibited the breakdown of triglycerides, (twofold) in mice under HFD (48), and hypoxia has been although basal and isoproterenol-induced lipolysis was detected during expansion of adipose tissue in obesity unchanged in Sucnr1 KO animals. This phenotype, in (49,50), doubtless occurring when vascularization is in- part, is consistent with other mouse models of “unre- sufficient to maintain oxygen levels. Thus increases in strained” lipolysis and shows that release from Gi-coupled succinate, hypoxia, and/or hyperglycemia might trigger inhibition of fat mobilization results in leanness (39–41) Sucnr1-induced signaling at the level of the adipocyte. and increased energy expenditure (41). After prolonged Clearly, because Sucnr1 is expressed in organs with high high-fat feeding, Sucnr1 KO mice displayed a rather more metabolic demands, tissue-selective KOs will be required complex phenotype, presumably a result of nutritional to identify the specific cell type(s) in which Sucnr1 plays stress induced by the HFD. Under these conditions, a direct role; nevertheless, our results indicate that acti- end-of-study body weights of Sucnr1 KO mice were, in vation of Sucnr1 may have therapeutic implications in the the main, comparable with WT controls, but WAT was treatment of obesity-related disorders. significantly heavier. Furthermore, after 16 weeks on HFD, Sucnr1 KO mice had significantly greater levels of blood glucose and failed to secrete insulin in glucose tol- Acknowledgments. The authors thank Guadalupe Sabio and Tamas Rözer for helpful discussions and Roisin Brid Doohan, Ana Belén Ricote, and erance tests; however, pancreas architecture was not dif- Jesús María Ruiz-Cabello for technical assistance (all from CNIC). ferent than in WT mice. Ostensibly, this hyperglycemic Funding. This work was supported in part by a research grant from the state may have resulted from the failure to buffer serum Ministerio de Ciencia e Innovación (SAF2009-07965 to K.J.M. and SAF2010- glucose, although this inference is tempered by the result 15239 to B.G.G.). The CNIC is supported by Ministerio de Economía y Compet- that insulin sensitivity was unchanged in Sucnr1 KO mice. itividad and the Pro-CNIC Foundation. Still, compensatory mechanisms such as neuronal or Duality of Interest. No potential conflicts of interest relevant to this article endocrine modulation may account for this discrepancy were reported. between insulin secretion and resistance. Author Contributions. K.J.M. and A.M.C. conceived the project, performed Although our original hypothesis presupposed that experiments, interpreted data, and wrote the paper. S.E., B.G.G., M.B., A.d.M., and a lean phenotype would result from augmented WAT P.S. performed experiments and interpreted data. A.M.C. is the guarantor of this lipolysis, our findings of increased metabolic activity, work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. under basal conditions, in Sucnr1 KO mice points to a more systemic effect of Sucnr1 deletion. Whether these References phenotypes are triggered directly by differences in adipose 1. Flower DR, Attwood TK. Integrative bioinformatics for functional genome mass in Sucnr1 KO mice or through a more circuitous annotation: trawling for G protein-coupled receptors. Semin Cell Dev Biol 2004; route awaits further study. Nevertheless, adiponectin lev- 15:693–701 els, which are typically inversely correlated with obesity, 2. Smith NJ. Low affinity GPCRs for metabolic intermediates: challenges for were decreased in Sucnr1 KO mice and might reflect spe- pharmacologists. Front Endocrinol (Lausanne) 2012;3:1 cific changes in adipocytokine signaling (42), leading us to 3. Blad CC, Tang C, Offermanns S. G protein-coupled receptors for energy – hazard that a hyperglycemic phenotype might occur when metabolites as new therapeutic targets. Nat Rev Drug Discov 2012;11:603 619 4. Cai TQ, Ren N, Jin L, et al. Role of GPR81 in lactate-mediated reduction of fat content reaches a critical mass. adipose lipolysis. Biochem Biophys Res Commun 2008;377:987–991 Many mechanisms can account for leanness, but they 5. Taggart AK, Kero J, Gan X, et al. (D)-beta-Hydroxybutyrate inhibits adipocyte all generally result from deregulation of energy balance, as lipolysis via the nicotinic acid receptor PUMA-G. J Biol Chem 2005;280:26649– noted here. We can speculate on several potential 26652 mechanisms that may promote this phenotype in Sucnr1 6. Tokonami N, Morla L, Centeno G, et al. a-Ketoglutarate regulates acid-base KO mice (43): hypothalamic signal transduction (44), re- balance through an intrarenal paracrine mechanism. J Clin Invest 2013;123: duced intestinal absorption of foods leading to decreased 3166–3171 diabetes.diabetesjournals.org McCreath and Associates 1167

7. He W, Miao FJ, Lin DC, et al. Citric acid cycle intermediates as ligands for 29. Hinke SA, Navedo MF, Ulman A, et al. Anchored phosphatases modulate orphan G-protein-coupled receptors. Nature 2004;429:188–193 glucose homeostasis. EMBO J 2012;31:3991–4004 8. Krebs HA. Rate control of the tricarboxylic acid cycle. Adv Enzyme Regul 30. Testa G, Schaft J, van der Hoeven F, et al. A reliable lacZ expression reporter 1970;8:335–353 cassette for multipurpose, knockout-first alleles. Genesis 2004;38:151–158 9. Ashrafian H, Czibik G, Bellahcene M, et al. Fumarate is cardioprotective via 31. Adham IM, Khulan J, Held T, et al. Fas-associated factor (FAF1) is required for activation of the Nrf2 antioxidant pathway. Cell Metab 2012;15:361–371 the early cleavage-stages of mouse embryo. Mol Hum Reprod 2008;14:207–213 10. Hohl C, Oestreich R, Rösen P, Wiesner R, Grieshaber M. Evidence for 32. Galy B, Ferring D, Benesova M, Benes V, Hentze MW. Targeted mutagenesis succinate production by reduction of fumarate during hypoxia in isolated adult rat of the murine IRP1 and IRP2 genes reveals context-dependent RNA processing heart cells. Arch Biochem Biophys 1987;259:527–535 differences in vivo. RNA 2004;10:1019–1025 11. Iles RA, Barnett D, Strunin L, Strunin JM, Simpson BR, Cohen RD. The effect 33. Zechner R, Zimmermann R, Eichmann TO, et al. FAT SIGNALS—lipases and of hypoxia on succinate metabolism in man and the isolated perfused dog liver. lipolysis in lipid metabolism and signaling. Cell Metab 2012;15:279–291 Clin Sci 1972;42:35–45 34. Schweiger M, Schreiber R, Haemmerle G, et al. Adipose triglyceride lipase 12. Correa PR, Kruglov EA, Thompson M, Leite MF, Dranoff JA, Nathanson MH. and hormone-sensitive lipase are the major enzymes in adipose tissue tri- Succinate is a paracrine signal for liver damage. J Hepatol 2007;47:262–269 acylglycerol catabolism. J Biol Chem 2006;281:40236–40241 13. Sapieha P, Sirinyan M, Hamel D, et al. The succinate receptor GPR91 in 35. Frederich RC, Hamann A, Anderson S, Löllmann B, Lowell BB, Flier JS. neurons has a major role in retinal angiogenesis. Nat Med 2008;14:1067–1076 Leptin levels reflect body lipid content in mice: evidence for diet-induced re- 14. Aguiar CJ, Andrade VL, Gomes ER, et al. Succinate modulates Ca(2+) sistance to leptin action. Nat Med 1995;1:1311–1314 transient and cardiomyocyte viability through PKA-dependent pathway. Cell 36. Maruhama Y, Ohneda A, Tadaki H, Ohtsuki M, Yanbe A. Hepatic steatosis Calcium 2010;47:37–46 and the elevated plasma insulin level in patients with endogenous hyper- 15. Sadagopan N, Li W, Roberds SL, et al. Circulating succinate is elevated in triglyceridemia. Metabolism 1975;24:653–664 rodent models of hypertension and metabolic disease. Am J Hypertens 2007;20: 37. Peti-Peterdi J. High glucose and renin release: the role of succinate and 1209–1215 GPR91. Kidney Int 2010;78:1214–1217 16. Toma I, Kang JJ, Sipos A, et al. Succinate receptor GPR91 provides a direct 38. Silva JE. Thermogenic mechanisms and their hormonal regulation. Physiol link between high glucose levels and renin release in murine and rabbit kidney. Rev 2006;86:435–464 J Clin Invest 2008;118:2526–2534 39. Ahmed K, Tunaru S, Tang C, et al. An autocrine lactate loop mediates insulin- 17. Krebs HA. Chemical composition of blood plasma and serum. Annu Rev dependent inhibition of lipolysis through GPR81. Cell Metab 2010;11:311–319 Biochem 1950;19:409–430 40. Bjursell M, Admyre T, Göransson M, et al. Improved glucose control and 18. Nordmann J, Nordmann R. Organic acids in blood and urine. Adv Clin Chem reduced body fat mass in 2-deficient mice fed a high-fat 1961;4:53–120 diet. Am J Physiol Endocrinol Metab 2011;300:E211–E220 19. Ariza AC, Deen PM, Robben JH. The succinate receptor as a novel thera- 41. Jaworski K, Ahmadian M, Duncan RE, et al. AdPLA ablation increases li- peutic target for oxidative and metabolic stress-related conditions. Front Endo- polysis and prevents obesity induced by high-fat feeding or leptin deficiency. Nat crinol (Lausanne) 2012;3:22 Med 2009;15:159–168 20. Kaplan NM. The deadly quartet. Upper-body obesity, glucose intolerance, 42. Patané G, Caporarello N, Marchetti P, et al. Adiponectin increases glucose- hypertriglyceridemia, and hypertension. Arch Intern Med 1989;149:1514–1520 induced insulin secretion through the activation of lipid oxidation. Acta Diabetol 21. Wilson PW, D’Agostino RB, Sullivan L, Parise H, Kannel WB. Overweight and 2013;50:851–857 obesity as determinants of cardiovascular risk: the Framingham experience. Arch 43. Xia Z, Stanhope KL, Digitale E, et al. Acylation-stimulating protein (ASP)/ Intern Med 2002;162:1867–1872 complement C3adesArg deficiency results in increased energy expenditure in 22. Regard JB, Sato IT, Coughlin SR. Anatomical profiling of G protein-coupled mice. J Biol Chem 2004;279:4051–4057 receptor expression. Cell 2008;135:561–571 44. Nogueiras R, López M, Diéguez C. Regulation of lipid metabolism by energy 23. Skarnes WC, Rosen B, West AP, et al. A conditional knockout resource for availability: a role for the central nervous system. Obes Rev 2010;11:185–201 the genome-wide study of mouse gene function. Nature 2011;474:337–342 45. Khalifeh-Soltani A, McKleroy W, Sakuma S, et al. Mfge8 promotes obesity 24. Künnecke B, Verry P, Bénardeau A, von Kienlin M. Quantitative body by mediating the uptake of dietary fats and serum fatty acids. Nat Med 2014;20: composition analysis in awake mice and rats by magnetic resonance relaxo- 175–183 metry. Obes Res 2004;12:1604–1615 46. Hu J, Wu Q, Li T, Chen Y, Wang S. Inhibition of high glucose-induced VEGF 25. Corbin IR, Furth EE, Pickup S, Siegelman ES, Delikatny EJ. In vivo as- release in retinal ganglion cells by RNA interference targeting G protein-coupled sessment of hepatic triglycerides in murine non-alcoholic fatty liver disease using receptor 91. Exp Eye Res 2013;109:31–39 magnetic resonance spectroscopy. Biochim Biophys Acta 2009;1791:757–763 47. Brownlee M. The pathobiology of diabetic complications: a unifying 26. Petersen PS, Jin C, Madsen AN, et al. Deficiency of the GPR39 receptor is mechanism. Diabetes 2005;54:1615–1625 associated with obesity and altered adipocyte metabolism. FASEB J 2011;25: 48. Cummins TD, Holden CR, Sansbury BE, et al. Metabolic remodeling of white 3803–3814 adipose tissue in obesity. Am J Physiol Endocrinol Metab 2014;307:E262–E277 27. Zhang HH, Huang J, Düvel K, et al. Insulin stimulates adipogenesis through 49. Elias I, Franckhauser S, Ferré T, et al. Adipose tissue overexpression of the Akt-TSC2-mTORC1 pathway. PLoS ONE 2009;4:e6189 vascular endothelial growth factor protects against diet-induced obesity and 28. Sabio G, Kennedy NJ, Cavanagh-Kyros J, et al. Role of muscle c-Jun NH2- insulin resistance. Diabetes 2012;61:1801–1813 terminal kinase 1 in obesity-induced insulin resistance. Mol Cell Biol 2010;30: 50. Regazzetti C, Peraldi P, Grémeaux T, et al. Hypoxia decreases insulin sig- 106–115 naling pathways in adipocytes. Diabetes 2009;58:95–103