Genetic control of ATGL-mediated lipolysis modulates adipose triglyceride stores in leptin-deficient mice.

Dissecting the genetics of complex traits such as obesity allows the identification of causal genes for disease. Here, we show that the BALB/c mouse strain carries genetic variants that confer resistance to obesity induced by leptin-deficiency or a high-fat diet (HFD). We set out to identify the physiological and genetic bases underlying this phenotype. When compared with C57BL6/J ob/ob mice (B6), BALB/c ob/ob mice exhibited decreased food intake, increased thermogenic capacity, and improved fat catabolism, each of which can potentially modify obesity. Interestingly, analysis of F1 ob/ob (progeny of B6 ob/+ × BALB/c ob+) mice revealed that obesity resistance in BALB/c ob/ob mice principally relied upon improved fat mobilization. This was mechanistically explained by increased adipose triglyceride lipase (ATGL) content in adipocytes, along with increased lipolysis and fatty acid oxidation. We conducted a genome-wide scan and defined a quantitative trait locus (QTL) on chromosome 2. BALB/c alleles on chromosome 2 not only associated with the obesity resistance phenotype but also supported increased ATGL content in adipose tissue. In summary, our study provides evidence that leptin-independent control of adipocyte lipolysis rates directly modifies the balance of macronutrient handling and is sufficient to regulate fat mass in the absence of alterations in food intake and energy expenditure.-Marcelin, G., S-M. Liu, X. Li, G. J. Schwartz, and S. Chua.

Obesity, defi ned as the accumulation of triglycerides within adipocytes, leads to increased body fat content and exacerbates susceptibility to type 2 diabetes, cardiovascular dysfunction, and cancer (1)(2)(3). Individual susceptibility to obesity is variable and integrates both environmental Qiagen). DNA was measured by spectrophotometric method, and total DNA content from fat pads was determined by multiplying DNA concentration in each sample by the fat pad weight for each mouse ( 16,17 ).

Glucose tolerance test
Mice were fasted overnight and glucose was given intraperitoneally (1 mg/g). Glycemia was measured at the different indicated time for 2 h from tail blood.

Cold tolerance assay
BALB/c, B6, and F1 ob/ob mice were placed at 4°C, and body temperature drop was measured with a rectal probe after 75 min. Water and food were removed during the experiment.

Ex-vivo lipolysis
Fat explants freshly isolated from fed mice were cut into small pieces and incubated at 37°C under agitation in Krebs-Ringer medium containing 2% of free fatty acid BSA with or without isoproterenol (Sigma). Glycerol was measured from supernatants after 2 h of incubation. ATGL-mediated lipolysis (i.e., non-HSL lipolysis) was determined in presence of the inhibitor CAY10499 (100 M). Fat explants were preincubated for 1 h with or without CAY10499. Then the medium was changed with medium containing (or not) CAY10499, and FFA was measured from supernatants after 2 h of incubation. Six mice in each group were studied, and analysis was performed in triplicate.

Metabolites and hormone concentration determination
Blood samples for plasma or serum were always collected at the same time in the morning. Mice were either fed ad libitum (without previous fasting) or fasted overnight (15 h period of fast). Glycemia was determined using a glucometer (Abbot), serum insulin levels using ELISA (Linco Mouse Insulin kit), plasma NEFA (Wako), and serum glycerol (Cayman) using a colorimetric assay. The HOMA index uses the following formula: insulin ( U/ml) × [glucose (mmol/l) / 22.5].

Treadmill
Wild-type BALB/c and C57BL6/J males were familiarized with the calorimetric treadmill for 2-3 days preceding the actual treadmill test. The adaptation period consisted of sessions lasting 5 min at belt speed 9 m/min. The number of aversive stimulations received was recorded, and the adaptation session was repeated once after 20 min of recovery if mice received more than fi ve stimulations. For the treadmill test, the ramping protocol of the belt speed was as follows: 0 m/min for 20min, 5 m/min for 5 min (warm-up phase), 9 m/min for 5 min, 12 m/min for 5 min, 15 m/min for 5 min, 17 m/min for 2 min, 19 m/min for 2 min, and 20 m/min until the end of the test. RER was measured every 4 min during treadmill running by CLAMS calorimetry.

Immunoblot analysis
White adipose tissue was homogenized in lysis buffer containing protease and phosphatase inhibitor. Protein extracts were run on Criterion gels (Bio-Rad) and blotted onto nitrocellulose membranes. After blocking, immunoblots were incubated with primary antibodies against hormone-sensitive lipase (HSL), phospho-HSL (Ser563 and Ser660), ATGL, ␤ -actin (Cell Signaling Technology), perilipin A (Sigma) and CGI-58 (gift from Dawn Brasaemle , Rutgers University). Blots were then incubated with fl uorescent secondary antibodies, and proteins were detected using the fl uorescence-based Odyssey Infrared Imaging System (LI-COR Biosciences). Quantifi cation was performed using the Odyssey Infrared Imaging Software. BALB/c ob/ob mice and found that obesity resistance in these mice principally relied upon improved fat oxidation. This operated through enhanced adipocyte lipolysis due to increased intracellular content and activity of adipose triglyceride lipase (ATGL/PNPLA2/desnutrin). Thus, higher lipolysis rate in adipocytes sustains increased fatty acid oxidation in BALB/c ob/ob mice. In addition, a genome-wide QTL analysis uncovered a signifi cant QTL Lipq1 (for lipolytic line QTL 1) on chromosome 2 associated with body fat fraction and ATGL adipocyte content. Interestingly, Lipq1 does not alter food intake. In conclusion, our study supports that leptin-independent control of adipocyte lipolysis rates through regulating ATGL directly modifi es the balance of macronutrient handling and is suffi cient to regulate fat mass in the absence of detectable changes in food intake and energy expenditure.

Body composition and indirect calorimetry
Body composition was measured by magnetic resonance spectroscopy (MRS) using an EchoMRI (Echo Medical Systems). Fatmass percentage was calculated as the ratio between fat mass and body weight. For indirect calorimetry, mice were individually housed in metabolic chambers maintained at 20-22°C on a 12 h light, 12 h dark cycle with lights on at 0800 h. Metabolic measurements [oxygen consumption, respiratory exchange ratio (RER), locomotor activity] were obtained continuously using a CLAMS (Columbus Instruments) open-circuit indirect calorimetry system. Calorimetry data were normalized to individual lean body mass. Cumulative food intake of pelleted chow diet (#F0163, precision dustless food pellet; Bio-Serv) was measured daily and collected over fi ve days for each individual animal.

Histological analysis of adipose tissue and estimation of adipocyte number per fat pad
Fat samples were fi xed in 10% buffered formalin, embedded in paraffi n, cut into 6 m sections, and stained with hematoxylin and eosin. Adipocyte sizes were determined with ImageJ software. To assess adipocyte number, total DNA content from adipose tissue was determined. Briefl y, fat pads were resected, weighed, and immediately frozen in liquid nitrogen. About 50 mg of tissue was homogenized, and genomic DNA was extracted following the manufacturer's protocol (DNeasy Blood and Tissue Kit, 69504; accompanied by a signifi cant increase in total DNA content, which is consistent with an increased number of adipocytes in B6 ob/ob compared with BALB ob/ob ( Fig. 1D ). Similar characteristics were observed for the gonadal fat pad (data not shown). Thus, decreased adipose tissue mass in BALB/c ob/ob mice was the result of both reduced adipocyte triglyceride storage and decreased adipogenesis.
Additionally, decreased adiposity in the BALB/c strain was associated with changes in glucose homeostasis. Fasting glycemia of BALB/c ob/ob mice was lower with unchanged insulinemia compared with B6 ob/ob mice ( Table 1 ). F1 ob/ob mice displayed hyperglycemia compared with BALB/c ob/ob , and F1 females exhibited lower insulinemia compared with BALB/c ob/ob ( Table 1 ). On a HFD, fasting glucose was lowered in wild-type BALB/c relative to wild-type B6, and insulin level was decreased in BALB/c males (supplementary Fig. II-B). Glucose tolerance tests (GTT) performed on both ob/ob ( Fig. 1E ) and HFD mice (supplementary Fig. II-C) showed that glucose clearance was markedly improved in BALB/c versus B6 mice.

Genotyping
DNA was isolated from ear clips. Allelic constitution at various microsatellite markers was determined by PCR and analysis of amplicon size. The markers (from the Mouse Phenome Database and the NCBI database) used for the chromosome 2 scanning were D2Mit37 , D2Mit42 , D2Mit30 , and D2Mit456 . SNPs were also examined as described in supplementary Table I. Briefl y, we amplifi ed by PCR the region containing the single nucleotide polymorphism (SNP) and digested the amplicon with the appropriate restriction enzyme to discriminate between the BALB and B6 alleles as predicted by the database (http://insilico.ehu.es/ restriction/two_seq/).

Statistical analysis
Results are shown as average ± SEM. Comparisons between groups for indirect calorimetry were performed by two-way repeated measures ANOVA, with genotype and time as factors. One-way ANOVA and unpaired Student t -tests were performed as indicated in the fi gure legends. Signifi cance was accepted at P < 0.05. For the whole-genome scanning, Fisher's exact test was performed, and statistical signifi cance was accepted at P < 0.00125 after Bonferroni correction. Statistical analysis was performed with GraphPad Prism 5.

Protection against obesity and diabetes in BALB/c mouse strain
We studied the development of obesity in HFD-fed mice and in genetic obesity due to leptin defi ciency in C57BL6/J (B6) and BALB/c strains. Of note, the leptin-defi cient mice ( ob/ob ) were also Agrp Ϫ / Ϫ . Because insulin signaling could be a confounding factor in obesity study, we used Agrp deletion as a tool to lower glucose and insulin level (supplementary Fig. I) (18 ). Thus, all of our data were based on BALB/c ob/ob Agrp Ϫ / Ϫ mice compared with C57BL/6J ob/ob Agrp Ϫ / Ϫ mice (B6). For brevity, we skipped the Agrp We observed that both B6 males and females fed a HFD for three months exhibited signifi cantly increased body fat fraction compared with BALB/c animals (supplementary Fig. II-A). We next examined the impact of leptin deficiency in three different genetic backgrounds: BALB/c, B6, and F1 (BALB/c × B6). BALB/c ob/ob mice developed obesity but to a much lesser extent than the B6 ob/ob mice. Body composition analysis showed that three-month-old B6 ob/ob males and females had much greater fat mass (+40%) compared with BALB/c ob/ob mice ( Fig. 1A ). The F1 ob/ob mice developed obesity similar to the parental B6 strain with increased body weight and fat mass compared with BALB/c strain ( Fig. 1A ). Consequently, the BALB/c genome carries alleles recessive to B6 alleles that limit fat accumulation. Inguinal fat pad wet weight was reduced in BALB/c ob/ob (1.9 ± 0.1 g vs. 6.3 ± 0.6 g, P < 0.0001, in BALB/c and B6 ob/ob , respectively), and histological analysis of adipose tissue revealed a greater frequency of small adipocytes in subcutaneous fat depots from BALB/c ob/ob compared with B6 ob/ob ( Fig. 1B ), resulting in a signifi cant decrease in mean adipocyte cross-sectional area ( Fig. 1C ). Moreover, the increased fat mass for a given fat pad was . (E) GTT was performed by measuring blood glucose concentration after an overnight fast at the indicated times after intraperitoneal injection of glucose, and area under the curve (AUC) was calculated in BALB/c and B6 ob/ob females (n = 4). Data are expressed as average ± SEM. One-way ANOVA (A) and unpaired ttest (C-E) were performed. * P < 0.05, ** P < 0.005, *** P < 0.0005. phenotypic traits associated with obesity resistance in BALB/c ob/ob mice.
We monitored daily caloric intake and found that BALB/c and F1 ob/ob mice manifested reduced food intake compared with the B6 ob/ob mice. The BALB/c and F1 hypophagia was solely due to decreased meal frequency, as meal size was similar in both strains ( Table 1 ). As F1 and B6 ob/ob mice exhibited the same level of adiposity, we concluded that caloric intake is unlikely to be the major contributor to fat-mass deposition in our models.
Assessment of cold tolerance to probe brown adipose tissue function, a critical modulator of metabolic imbalance ( 19 ), revealed that cold-induced thermogenesis was impaired in B6 ob/ob mice compared with BALB/c and F1 ob/ob mice ( Table 1 ). As adiposity and cold tolerance were discordant in B6 ob/ob mice and F1 ob/ob mice, brown adipose tissue thermogenesis differences could not explain resistance to obesity in BALB/c ob/ob mice. Whole body energy expenditure evaluation using indirect calorimetry did not reveal any difference in oxygen consumption between strains despite decreased spontaneous activity in B6 and F1 compared with BALB/c ob/ob mice ( Table 1 ). Similarly, no difference in oxygen consumption was found between HFD BALB/c and B6 mice (supplementary Fig.  IV-A). Consequently, obesity resistance of BALB/c mice could not be ascribed to alterations in whole body energy expenditure. Interestingly, the revealing difference observed between BALB/c and both B6 and F1 ob/ob mice relates to the RER. B6 and F1 ob/ob mice exhibited higher RER values during both the light and the dark periods compared with BALB/c ob/ob mice ( Fig. 2A ). In B6 and F1 ob/ob mice, RER values were almost all greater than 1.

Increased fatty acid oxidation is associated with reduced adiposity in BALB/c mice
Given our fi ndings that F1 and B6 ob/ob mice developed similar degrees of obesity, we next compared BALB/c ob/ ob mice with both B6 and F1 ob/ob mice in search of key Data are expressed as average ± SEM and statistical signifi cance is determined for P < 0.05. a Determined after an over-night fast. b ⌬ °C is the difference between the initial temperature and the temperature measured after 75 min at 4°C. c Individual fat-free body mass measured by magnetic resonance spectroscopy was used for normalization. metabolism. Two main enzymes account for 95% of lipolytic activity in adipocytes, ATGL, and HSL ( 23 ). Interestingly, protein levels of ATGL and its coactivator CGI-58 ( 24 ) were increased in adipose tissue from BALB/c ob/ob mice compared with that from B6 and F1 ob/ob mice ( Fig. 3A , B ). Similarly, in HFD-fed mice, increased ATGL and CGI-58 protein levels were observed in BALB/c adipose tissue compared with that from B6 mice (supplementary Fig. III-B, C).
We also checked HSL expression. We found that HSL protein level was increased in adipose tissue from BALB/c ob/ob mice compared with that from B6 and F1 ob/ob mice (supplementary Fig. V-A). However, there was no difference in the amount of the activated form of HSL ( 25,26 ), phosphorylated HSL (supplementary Fig. V-B, C), or the lipolysis inhibitor perilipin A ( 27 ) (supplementary Fig.  V-D). Hence, HSL may not participate in increased adipose lipolysis observed in BALB ob/ob . Accordingly, there was no change in total HSL protein expression or HSL activation in adipose tissue from HFD-fed mice (supplementary Fig. III-B, C). We further investigated adipose lipolysis in measuring ATGL-mediated lipolysis. To that aim, we measured free fatty acid (FFA) release from adipose tissue explants in presence or not of the HSL inhibitor CAY10499 ( 28,29 ). We previously assessed the effi ciency of CAY10499 in our experimental condition as we found that induced lipolysis with a selective ␤ 3-AR agonist (CL316,243) is blunted in presence of 100 M CAY10499 (supplementary Fig. VI). Then we compared FFA release from freshly dissected adipose tissue explants This indicated that lipid metabolism is shifted toward net lipid synthesis with minor lipid oxidation in B6 and F1 ob/ob compared with BALB ob/ob ( 20 ). Moreover, lower RER values were also observed in HFD-fed BALB/c mice compared with B6 mice ( Fig. 2B ). Thus, our data reveal that the BALB/c strain is characterized by an elevation in fatty acid oxidation compared with the B6 strain, implying that increased fat catabolism confers protection against obesity.

Increased basal lipolysis in BALB/c ob/ob mice
Lower RER values ( Fig. 2A ) and reduction in adipocyte lipid droplet size in BALB/c ob/ob mice ( Fig. 1C ) suggest that increased fatty acid oxidation may be a consequence of an increase in lipolysis. Consistent with this hypothesis, higher circulating glycerol concentrations were observed in BALB/c mice compared with B6 and F1 ob/ob mice ( Fig. 2C ). Nonesterifi ed fatty acid (NEFA) concentrations were increased in BALB/c versus F1 ob/ob mice ( Fig. 2D ) while no difference was detected between BALB/c and B6 ob/ob mice ( Fig. 2D ). In addition, glycerol ( Fig. 2E ) and NEFA ( Fig. 2F ) were increased in BALB/c versus B6 HF fed mice.
Beta-3 adrenergic receptor ( ␤ 3-AR), predominantly expressed in white and brown adipose tissues, stimulates sympathetic activation of adipose tissue to promote lipolysis. By qRT-PCR, we found that ␤ 3-AR expression was signifi cantly increased in BALB/c versus B6 ob/ob white adipose tissue (supplementary Fig. IV-A). However, we observed that acute CL316,243 (a specifi c ␤ 3-AR agonist) administration raised serum glycerol concentration of BALB/c and B6 ob/ob mice to the same concentration (supplementary Fig. IV-B), suggesting similar degrees of ␤ 3-adrenergic signaling between the two strains ( 21 ). To ascertain whether the lipolysis rates were a cell-autonomous function rather than strictly regulated by the autonomic nervous system, we next quantifi ed adipose tissue lipolysis by measuring glycerol release from adipose tissue explants. In both basal and stimulated conditions, we observed that glycerol release was signifi cantly higher in explants derived from BALB/c ob/ob compared with B6 ob/ob ( Fig. 2G ). These data indicate that the increased rate of lipolysis of BALB/c adipocytes is intrinsic to adipocytes.
As fatty acid demand increases during exercise ( 22 ), we assessed the lipolytic capacity of wild-type B6 and BALB/c animals (fed a normal chow diet) during exercise involving treadmill running. No difference in RER was observed at rest. However, during exercise, whereas RER values remained stable for BALB/c mice, RER signifi cantly increased for B6 mice, indicating increasing dependence upon carbohydrates for fuel ( Fig. 2H ). We observed that exercise-induced increase in plasma NEFA was low in B6 mice, whereas BALB/c animals signifi cantly increased NEFA production by almost 3-fold ( Fig. 2I ). Thus, these data reveal an increased ability of BALB/c mice to perform lipolysis and oxidize fatty acids.

Increased amounts of ATGL in BALB/c adipocytes
We further characterized BALB/c adipocytes by analyzing expression of key enzymes involved in triglyceride  from BALB/c and B6 ob/ob inguinal fat pad. We observed that FFA production was increased in explants from BALB/c mice relative to B6 ob/ob mice when HSL activity was inhibited by CAY10499 ( Fig. 3D ). Moreover, we calculated the same difference between BALB and B6 lipolysis ( ⌬ Ϫ CAY = 431.3 ± 31.7 versus ⌬ +CAY = 397.1 ± 8.1, P = 0.31) without the HSL inhibitor (i.e., HSL + ATGL activity) or with CAY10499 (i.e., ATGL activity), respectively. Consequently, our data show that increased adipose lipolytic rate in BALB/c principally relied on increased ATGL activity. Of note, the increased ATGL activity was greater than the increase in ATGL protein content in adipose tissue. This may be due to the concomitant increase of CGI-58 expression that is known to enhance ATGL lipolytic activity.
Thus, we postulated that the increase in ATGL protein level is an intrinsic characteristic of the BALB/c strain. Indeed, ATGL protein levels were also differentially expressed in adipose tissue from four-week-old lean BALB/c and B6 mice with no difference in adiposity ( Fig. 3C ). Therefore, it is likely that the differences in ATGL protein between the BALB/c and B6 strains were not due to the development of obesity but may be the underlying cause of increased adipose lipolysis.

A genome-wide scan points to a chromosome 2 locus associated with fractional body fat
We generated obese N2 progeny (from F1 ob/+ Agrp Ϫ / Ϫ × BALB/c ob/+ Agrp Ϫ / Ϫ ). Body composition studied for each mouse showed that N2 progeny displayed fat-mass values scattered over a broad range, from low adiposity similar to BALB ob/ob , to high adiposity comparable to B6 or F1 ob/ob mice ( Fig. 4A ). Previously, we did not report any gender difference in BALB or B6 ob/ob mice. Moreover, both N2 ob/ob males and females displayed the same distribution of fat-mass values (not shown). Consequently, N2 ob/ob males and females were combined for the genomewide scan.
To map the QTL associated with obesity resistance in BALB/c strain, we only genotyped N2 ob/ob mice at the two extremes of the distribution of body fat fraction: N2 mice with high fat-mass percentage (>60%), and N2 mice with low adiposity (<50%) ( Table 2 ). On the basis of our fi ndings on BALB/c, B6, and F1 ob/ob mice, we reasoned that mice with low adiposity should be enriched with homozygous BALB/c alleles (CC genotype) at loci that control fatmass deposition, whereas mice with high adiposity should be enriched with heterozygous B6 and BALB/c alleles (BC genotype). A set of 40 microsatellite markers were used to perform the genome scan, and 49 N2 ob/ob progenies were analyzed. We found a signifi cant association with the following markers on chromosome 2: rs4136610 , D2Mit37 , and rs3667007 . No other marker exhibited a signifi cant association. Consequently, we concluded that our genome-wide analysis revealed markers on chromosome 2 strongly associated with obesity resistance of BALB/c strain, and this QTL was named Lipq1 for lipolytic line QTL 1. We further analyzed the N2 ob/ob mice according to their chromosome 2 haplotype. As expected, BALB/c alleles on chromosome 2 confer lower body weight and fat mass of leptin-defi cient animals ( Fig. 4B, C ). Moreover, Lipq1 does not alter caloric intake ( Fig. 4D ). Finally, we checked ATGL expression in adipose tissue and found that ATGL protein level is increased in N2 ob/ob mice with BALB/c alleles on chromosome 2 ( Fig. 4E ). Consequently, the genome-wide scan provides supportive evidence that a locus or a set of loci on chromosome 2 supports increased ATGL expression in adipose tissue and control of body fat fraction, independently of caloric intake.

SNP architecture of the chromosome 2 and candidate genes
C57BL6/J and BALB/c genome sequence strains are included in the Sanger sequencing database. Consequently, levels, a trait conferred by modifi er genes present on the chromosome 2 of BALB/c mice.
Triglyceride (TG) hydrolysis, or lipolysis, is a sequential process that depends on ATGL (PNPLA2, desnutrin), which initiates TG breakdown and produces diacylglycerols (DG), whereas HSL preferentially generates monoglycerides from DGs ( 30 ). Thus, ATGL is rate-limiting for the fi rst step of TG hydrolysis and may determine the maximal rate for fatty acid mobilization (31)(32)(33). In obese patients, defects in adipose tissue lipolytic activity have been reported, but it is not clear whether altered lipolysis is primary or secondary to obesity development (34)(35)(36). Here, we noticed that ATGL level was decreased in adipose tissue from B6 ob/ob mice compared with BALB/c ob/ob mice . In addition, comparison between BALB/c and B6 ob/ob mice revealed that protein levels of ATGL and its coactivator CGI-58 were elevated in the adipose tissue of BALB/c ob/ob mice, whereas active HSL levels were unchanged. Consequently, downregulation of ATGL level and activity occurred specifi cally in the obese-susceptible strain C57BL6/J, whereas the resistant strain BALB/c maintained higher levels of ATGL and its activators. In BALB/c ob/ob mice, the higher lipolysis rate was accompanied by increased whole body fatty acid oxidation. So, it is tempting to propose that even higher rates of adipocyte lipolysis could further reduce adipocyte lipid stores if fatty acid oxidation rates could accommodate the increased supply.
As lipolysis also supports fatty acid demand by exercising muscles, we further assessed the lipolytic capacity of lean BALB/c and B6 mouse using a mouse treadmill running model ( 22 ). In BALB/c mice, a fl at RER during ramped exercise indicated a high fatty acid oxidation rate suffi cient to effi ciently provide ATP to supply exercising muscles. By contrast, C57BL/6J mice fail to increase fatty acid release during exercise and were apparently unable to increase fatty acid oxidation, leading to a rapid increase in RER during exercise ramp-up. Together, these fi ndings imply that BALB/c alleles support a better lipolytic capacity in both obese and exercising mice, suggesting that increased TG hydrolysis is a principal mechanism protecting BALB/c mice from obesity.
We postulate that the increase in ATGL protein level is an intrinsic characteristic of the BALB/c strain. Indeed, ATGL protein levels were also differentially expressed in adipose tissue from young, lean BALB/c and B6 mice prior to the appearance of any differences in adiposity. Therefore, we argue that the differences in ATGL protein between the BALB/c and B6 strains are not secondary to obesity but, rather, are determined by BALB/c modifi er genes. In humans, ATGL defi ciency causes neutral lipid storage disease due to an inability to mobilize neutral lipids ( 37 ), and low body mass indexes (BMI ) have been associated with higher levels of ATGL protein level in human adipose tissue ( 38 ). Consequently, identifi cation of a regulatory mean of modulating ATGL might have important implications for obesity treatment. QTL analysis is commonly used to map chromosomal regions and genes that control obesity in mice. Having fi rst identifi ed lipolysis as the leading cellular mechanism that protects BALB/c we studied the SNP architecture of the proximal chromosome 2 spanning from 5.5 to 68 Mb (i.e., ‫ف‬ 60 Mb) and found less than 20 SNPs for gene-associated location (supplementary File I ). By contrast, the domain including the QTL (40 Mb spanning from 65 to 104 Mb) contains ‫ف‬ 500 SNPs for gene-associated location (supplementary File II). As a consequence, the low SNP occurrence in the proximal chromosome 2 indicated that BALB and B6 strain are not polymorphic at these loci. In addition to this in silico analysis, we examined SNPs defi ned as polymorphic between BALB and B6 (supplementary Table I), for which allele information is available online through the Mouse Genome Database. For the SNP genotyping, we used genomic DNA extracted from the BALB/c and B6 parental strains. Interestingly, the SNP analyzed from 24.7 Mb to 67.09 Mb were not polymorphic between BALB/c and B6 strain (supplementary Table I). Altogether, a nonpolymorphic region of 42 Mb extending from marker rs27233322 to marker rs27987286 is defi ned on chromosome 2. As QTLs are preferentially assigned to polymorphic loci between parental strains, we narrowed the critical interval to a polymorphic region of chromosome 2 spanning from 65 Mb to 104 Mb. Overall, our results support that genetic determinants on chromosome 2 are associated with expression level of ATGL and CGI-58. It is conceivable that a transcriptional or posttranslational mechanism, or both, through a direct or indirect mechanism could be responsible for increased protein expression. Thus, we established a list of the plausible candidates ( Table 3 ) to these processes by studying SNPs in conjunction with the functional information available.

DISCUSSION
Obesity is often associated with impaired response to the adipose-tissue derived hormone leptin, a central regulator of body fat content and metabolism. We proposed here to use genetics to unveil leptin-independent pathways that overcome obesity. Toward that aim, BALB/c mice, which have been shown previously to be more resistant to obesity induced by leptin defi ciency, were compared with leptin-defi cient C57BL6/J (B6) mice ( 15 ). To probe the primary function responsible for obesity resistance, we measured multiple metabolic and behavioral parameters. When compared with B6 ob/ob mice, BALB/c ob/ob mice exhibited decreased food intake, increased thermogenic capacity, and improved fat catabolism. However, analysis of F1 ob/ob (B6 ob/+ × BALB/c ob/+ ) mice revealed that obesity resistance in BALB/c ob/ob mice solely relied upon improved fat oxidation. We uncovered a leptin-independent increase in lipolysis intrinsic to adipocytes that contributed to the regulation of body fat content in BALB/c ob/ob mice. Interestingly, this process was also operative in lean mice fed a HFD, as we found similar increased fattyacid oxidation; consequently, this mechanism supported by BALB/c modifi er gene(s) was operative beyond leptin defi ciency. Finally, we showed that increased lipolysis in these mice was associated with increased ATGL protein from obesity, we wanted to fi nd the genetic loci controlling body fat content and lipolysis in these mice. We identifi ed a signifi cant QTL named Lipq1 (for lipolytic line QTL 1) on chromosome 2 associated with body fat content. Furthermore, Lipq1 supports higher ATGL protein content in adipose tissue of N2 ob/ob mice. We narrowed the region to a critical interval that encompasses a chromosome 2 portion of ‫ف‬ 30 Mb. A previous study mapped an overlapping QTL Mob that confers increased adiposity in F2 (C57BL6/J × CAST/Ei) mice inheriting chromosomal regions from B6 and lower adiposity in mice inheriting the CAST chromosomal region ( 39 , 40 ). Conversely, BALB/c alleles of Nidd5 on chromosome 2 have been shown to decrease obesity in Tsumura Suzuki obese diabetic (TSOD) mice ( 41 ). Underlying mechanisms were not described; however, analysis of lipolysis in those models would indicate whether a common locus was involved in obesity resistance. Interestingly, Lipq1 is syntenic to human chromosomal region 2q22-32 that contains loci related to Bardet-Biedl syndrome with manifestation of obesity ( 42 ), abdominal visceral fat ( 43 ), and higher BMI ( 44 ). Consequently, this observation supports that the mechanism we identifi ed in the BALB/c strain could be relevant for a better understanding of human obesity.
In summary, our fi ndings highlight a leptin-independent increase in adipose tissue lipolysis that restrains obesity and glucose intolerance. Recent advances in the understanding of the enzymology of lipolysis within adipocytes and other cell types have yielded insights into the process of triglyceride metabolism and the regulatory molecules involved ( 24,(45)(46)(47). Here, we identifi ed a QTL on mouse chromosome 2 that confers resistance to obesity through modulation of lipolysis and the adipocyte content of ATGL . We propose that identifi cation of Lipq1 will provide a novel target for the treatment of obesity and associated metabolic defects.