Adipocyte-selective Reduction of the Leptin Receptors Induced by Antisense RNA Leads to Increased Adiposity, Dyslipidemia, and Insulin Resistance*

Although recent evidence suggests that leptin can directly regulate a wide spectrum of peripheral functions, including fat metabolism, genetic examples are still needed to illustrate the physiological significance of direct actions of leptin in a given peripheral tissue. To this end, we used a technical knock-out approach to reduce the expression of leptin receptors specifically in white adipose tissue. The evaluation of leptin receptor reduction in adipocytes was based on real time PCR analysis of the mRNA levels, Western blot analysis of the proteins, and biochemical analysis of leptin signaling capability. Despite a normal level of leptin receptors in the hypothalamus and normal food intake, mutant mice developed increased adiposity, decreased body temperature, hyperinsulinemia, hypertriglyceridemia, impaired glucose tolerance and insulin sensitivity, as well as elevated hepatic and skeletal muscle triglyceride levels. In addition, a variety of genes involved in regulating fat and glucose metabolism were dysregulated in white adipose tissue. These include tumor necrosis factor- (cid:1) , adiponectin, leptin, fatty acid synthase, sterol regulatory element-binding protein 1, glycerol kinase, and (cid:2) 3 -adre- nergic receptor. Furthermore, the mutant mice are significantly more sensitive to high fat feeding with regard with leptin or insulin for 10 min before being harvested for analysis of leptin signaling. The analysis of PI3-kinase was carried out with a protocol described previously in which we used an anti-phospho- tyrosine (pY20) antibody (Transduction Laboratory) to precipitate the PI3-kinase activity (15). The analysis of phospho-STAT3 and phospho- MAPK was based on Western blot analysis with an anti-phospho-STAT3 (pY705) antibody and an anti-phospho-MAPK (pT202/pY204) antibody, respectively (Cell Signaling). Quantitative analysis was achieved by scanning and quantifying the bands of interest with a NIH Image 6.0 program, and all comparisons were subjected to a two-tailed Student’s t test with p (cid:3) 0.05 considered statistically significant. Re- combinant mouse leptin (AFP376C) NHPP, NFDDK, and Dr. A. F. Parlow.

Leptin deficiency in ob/ob mice and leptin receptor deficiency in db/db mice lead to hyperphagia, sterility, obesity, and diabetes (1)(2)(3)(4). The ability of leptin to regulate food intake, body weight, adiposity, and insulin sensitivity has been demonstrated extensively (for review, see Refs. 1 and 5). These vital functions of leptin have been attributed exclusively to its actions in the hypothalamus (2). Accordingly, intracerebroven-tricular injection of leptin in ob/ob mice can reduce food intake and body weight and alleviate the diabetic phenotypes (6). However, when leptin is peripherally infused or overexpressed in vivo, only a portion of the leptin-induced weight loss can be attributed to a reduction in food intake in the leptin-treated animals compared with the control animals with matched caloric intake (7)(8)(9)(10). Recent evidence at the cellular level further demonstrates that leptin can directly modulate glycerol release, insulin signaling, and gene expression in primary rodent adipocytes (11,12). At a genetic level, Kowalski et al. (13) found that transgenic expression of the long form leptin receptor, OB-Rb, in the central nervous system could only partially correct the obesity and diabetic phenotypes of db/db mice. Cohen et al. (14) demonstrated that mutant mice with brain-specific deletion of the leptin receptors developed obesity, whereas mutant mice with liver-specific deletion of leptin receptors did not exhibit any weight-related phenotype. Interestingly, the weight gain of the mutant mice with brain-specific deletion of leptin receptors was not as great as that of ob/ob and db/db mice at equivalent ages (14). These studies point to the possibility of a peripheral effect of leptin on body weight regulation. Indeed, many recent studies have suggested that at the cellular level leptin can play important physiological roles in many systems, such as adipocytes, hepatocytes, skeletal muscle, Tlymphocytes, pancreatic ␤-cells, and vascular endothelial cells (11,(15)(16)(17)(18)(19)(20)(21). In this study, we aim to create a genetic model to investigate whether leptin might play a role in regulating the peripheral functions (such as those of adipocytes) in vivo, in addition to its actions in the central nervous system.
The technical knock-out (TKO) 1 system has been deployed successfully to reduce the expression of several signaling proteins relevant to glucose and lipid homeostasis, such as G␣ i2 and G␣ q (22)(23)(24). This technology takes advantage of the tissue specificity of a rat phosphoenolpyruvate carboxykinase (PEPCK) promoter to drive and restrict the expression of short antisense sequences of a given target gene in fat and liver without affecting other nontarget genes or nontarget tissues (22,24). Because of the developmental regulatory nature of PEPCK promoter, the transgene (with the antisense sequences) is not expressed until after birth, thus eliminating concerns regarding any potential compensatory changes during embryonic development. By taking advantage of the TKO sys-tem, we can express a short stretch of antisense sequences of leptin receptor gene in fat and liver and potentially create mutant mice with selective leptin receptor deficiency in these tissues. The transgenic mice thus created are hereafter referred to as TKO-OBR mice, short for technical knock-out of OB receptor. Here, we describe our initial characterizations of such mutant mice.

EXPERIMENTAL PROCEDURES
Transgenic Mice, Genotyping, and Determination of Transgene Expression-An antisense sequence corresponding to a short stretch of sequences (39 bp) immediately 5Ј to the ATG initiation codon of the leptin receptor mRNA was inserted into the TKO vector, which was then used to generate transgenic animals (Fig. 1A). A BLAST search in the GenBank nucleotide data base indicated that this stretch of sequence showed no significant homology to any other known mouse genes. We used B6D2F1 hybrid strain for the initial production of transgenic mice. Subsequently, three independent lines were backcrossed with C57BL/6 to the F4 generations. All phenotypic changes of the TKO-OBR mice were relative to their wild-type littermates. The animals were fed standard food and water ad libitum, housed under controlled temperature at 22°C and a 12-h light-dark cycle with light from 0630 to 1830 h. All animals were handled in accordance with the guidelines established by the Institutional Animal Care and Use Committee at the University of Pittsburgh. The sequences of primers for genotyping are as follows: P1, 5Ј-CGTTTAGTGAACCGTCAGA; P2, 5Ј-TTGCCAAACCTACAGGTGGG; P3, 5Ј-CCCTTCTCATGACCTTTGGC-CGTG; P4, 5Ј-CCAGGTGTACACCTCTGAAGAAAG. For determination of transgene expression, the RNA samples (ϳ100 ng each) from different tissues were subject to a RT-PCR (with P1 and P2 as primers) followed by a Southern blot with a probe generated from the region flanked by P1 and P2. The specific primers for mouse glyceraldehyde-3-phosphate dehydrogenase gene are 5Ј-ACCACAGTCCATGCCATCAC FIG. 1. Generation of TKO-OBR mice and expression of TKO-OBR transgene. A, the antisense sequences of leptin receptor (39 bp, corresponding to the region 5Ј to the ATG initiation codon) were inserted into the first exon of a rat PEPCK gene. B, genotyping of TKO-OBR mice was carried out with a PCR screening of the tail genomic DNA samples with two sets of specific primers (P1/P2 and P3/P4) engineered into the transgene construct. Both primers are unique to the transgene construct. The appearance of both bands with expected sizes (ϳ230 bp for P1/P2, ϳ670 bp for P3/P4) indicates positive transgenic clones. In this case, the positive clones are 1, 2, 4, 5, and 7. Lane M, markers. C, tissue-specific expression of TKO-OBR gene as determined by a virtual Northern blot assay. The RNA samples (ϳ100 ng each) from different tissues were subjected to a RT-PCR (with P1 and P2 as primers) followed by a Southern blot. A semiquantitative (ϳ25 cycles) RT-PCR is also performed to determine the levels of 18 S ribosomal RNA expression, which served as a control of RNA quantity. D, an extensive RT-PCR amplification (40 cycles, with P1 and P2 as primers) detected no expression in the hypothalamic RNA samples of the transgenic mice (TKO-OBR). The three hypothalamic samples (Hyp.) were from three transgenic mice. The expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the control. The experiment was repeated twice. Lane M, markers. and 5Ј-TCCACCACCCTGTTGCTGTA. The RNA samples were extracted using a standard Trizol kit (Invitrogen).
Western Blot Assays-Protein samples were prepared by extracting tissues with an immunoprecipitation assay buffer described previously (25), separated on 8% or 10% SDS gels, and then blotted onto nitrocellulose membrane. The blots were incubated with appropriately diluted primary antibodies and horseradish peroxidase-conjugated secondary before being subjected to a chemiluminescent reaction for detection. The sources of antibodies are as follows: leptin, Peprotech, NJ; fatty acid synthase, BD Transduction Laboratory; tumor necrosis factor-␣ (TNF-␣,) sterol regulatory element-binding protein 1 (SREBP-1), ␤ 3adrenergic receptor, leptin receptor (K-20), and peroxisome proliferator-activated receptor-␥ (PPAR-␥), Santa Cruz Biotechnology, CA. A previously characterized rabbit polyclonal antibody against all variants of murine leptin receptor (15) was used to characterize the expression levels of leptin receptors except where noted.
Measurement of Food Intake and Rectal Temperature-Transgenic mice and their wild-type littermates were caged individually, and food was weighed before and at the end of each day for 2 days. The food crumbs were collected and weighed at the end of the observations to provide accurate assessment of food intake. The cumulative food intake in 48 h was divided by 2 to achieve the values of cumulative food intake in 24 h. The whole procedure was repeated three times for 6 days in a row. The temperature measurements were performed between 0930 and 1030 with a rectal microprobe and a microprobe thermometer (BAT-12) (Harvard Apparatus, MA).
Analysis of Total Body Fat Content-The analysis of total body fat content was according to a previously published protocol (14). Mice at 18 -19 weeks of age were sacrificed, and carcasses were weighed and then oven dried in a 90°C oven until the weight was constant. The carcass was then homogenized in a blender, and duplicate aliquots were extracted using a 3:1 mixture chloroform/methanol. The extracted homogenate was dried overnight and weighed to calculate fat mass and lean mass.
Intraperitoneal Glucose Tolerance Test and Insulin Sensitivity Test-After overnight fasting, the mice were injected intraperitoneally with glucose at a dose of 2 g/kg of body weight, and the blood glucose levels were monitored with Precision-Xtra strips (Medisense Products, MA) for 2 h through tail vein blood samples. For insulin sensitivity test, the mice were fasted overnight before being injected with a bolus of insulin (0.3 unit/kg of body weight), and their blood glucose levels were monitored through tail vein blood.
Measurements of Triglyceride Levels in Serum and Tissues, Serum Levels of Insulin, Leptin, TNF-␣, and Adiponectin-Hepatic and skeletal muscle triglycerides were extracted with a protocol established previously (10). The levels of triglyceride were determined with an Infinity Triglyceride kit (Sigma). The serum insulin and adiponectin levels were measured with radioimmunoassay kits from Linco Research, and the serum leptin level was determined with a radioimmunoassay kit from Alpco. The serum TNF-␣ level was determined with an enzyme-linked immunosorbent assay kit from Assay Designs, Inc. (Ann Arbor, MI). All comparisons were subjected to a two-tailed Student's t test with p Ͻ 0.05 considered statistically significant.
Determination of Leptin Signaling-Mice were fasted but given free access to water, for 8 h before intraperitoneal injection of 1 mg/kg saline or leptin. The animals were sacrificed 30 min after injection, and the white adipose tissue (WAT) and liver were rapidly removed and frozen in liquid nitrogen. For in vitro analysis, primary adipocytes were isolated from epididymal fat pad of both transgenic and wild-type mice based on a previous protocol (25). The adipocytes were subsequently treated with leptin or insulin for 10 min before being harvested for analysis of leptin signaling. The analysis of PI3-kinase was carried out with a protocol described previously in which we used an anti-phosphotyrosine (pY20) antibody (Transduction Laboratory) to precipitate the PI3-kinase activity (15). The analysis of phospho-STAT3 and phospho-MAPK was based on Western blot analysis with an anti-phospho-STAT3 (pY705) antibody and an anti-phospho-MAPK (pT202/pY204) antibody, respectively (Cell Signaling). Quantitative analysis was achieved by scanning and quantifying the bands of interest with a NIH Image 6.0 program, and all comparisons were subjected to a two-tailed Student's t test with p Ͻ 0.05 considered statistically significant. Recombinant mouse leptin (AFP376C) was obtained NHPP, NFDDK, and Dr. A. F. Parlow.

RESULTS
Genotyping of TKO-OBR mice was performed using a standard PCR screening of tail genomic DNA with two different pairs of specific primers engineered into the transgene construct (P1/P2 and P3/P4, Fig. 1A). Transgenic clones were identified by the presence of P1/P2 and P3/P4 amplicons (examples shown in Fig. 1B). Tissue specificity of the transgene at the mRNA level was determined through a sensitive reverse transcription (RT)-PCR assay (Fig. 1, C and D). The transgene proved to be expressed at high levels in the WAT and in the liver, and at much lower levels in the kidney (Fig. 1C, lanes 12, 11, and 5, respectively). Expression was not detectable either in the brain, brown fat, heart, lung, pancreas, spleen, skeletal muscle, and testis, or in the fat and liver of wild-type mice (Fig. 1C). To eliminate further the possibility that the transgene may be expressed in the hypothalamus, we also performed RT-PCR analysis on the hypothalamic samples from the transgenic mice with an extensive amplification (40 cycles). No expression of the transgene was detected in the hypothalamic RNA samples of the TKO-OBR mice (example shown in Fig. 1D). This expression pattern accurately reflects the tissue specificity of a traditional PEPCK promoter as described before (26).
The loss of leptin receptor expression in WAT was evaluated from three different aspects, which include Western analysis of OBR protein levels, real time PCR analysis of OBR mRNA levels, and examination of leptin signaling capability in WAT of the TKO-OBR mice. With a specific antibody recognizing several variants of leptin receptor (15), we analyzed OBR expression at the protein level in different tissues of the TKO-OBR mice. Although this antibody has been characterized previously (15), we further demonstrated its specificity by comparing its immunoactivity with that of another leptin receptor antibody recognizing different epitopes (K20, Santa Cruz Biotechnology, CA). The results in Fig. 2A indicate that the two different leptin receptor antibodies detected the same long and/or short forms of leptin receptors from the hypothalamic and WAT samples on Western blot assays. The sizes of the long and short forms of leptin receptors detected here are consistent with those observed in previous studies (21,(27)(28)(29). The identification of OB-Rb was confirmed by further comparing the immunoreactive bands of the wild-type brain with those of db/db hypothalamic and brain extracts because the db/db mice do not express OB-Rb but maintain the expression of short forms of OBR (4) (Fig. 2B).
The predominant forms of OBR in WAT are the short forms, but this result does not exclude the possibility of a minor presence of OB-Rb playing important physiological roles. In fact, we were able to detect a low level of OB-Rb expression in a RT-PCR assay (Supplemental Data 1). As anticipated, in all three independent lines of TKO-OBR mice, the transgene was able to reduce the expression of leptin receptors substantially in the WAT with more than 80% reduction in one line and greater than 90% reduction in the other two lines (n ϭ 5 for each genotype; p Ͻ 0.05; examples shown in Fig. 2C). The extent of OBR reduction is essentially the same between the male and female transgenics. These three lines of TKO-OBR mice had very similar phenotypes, and the results described below reflect the averages from all three transgenic lines. Although the transgene was highly expressed in the liver and modestly in the kidney of TKO-OBR mice, it failed to reduce the expression levels of the leptin receptors expressed in these tissues (Fig. 2, C and E). Because of the dynamic regulations of PEPCK gene expression, we also determined the hepatic expression level of leptin receptors under a fasting condition where the transgene was supposed to have the maximal effect because of its high expression levels. Even after an extensive fasting period (18 h), the hepatic levels of leptin receptors of the TKO-OBR mice were essentially the same as those of the wildtype littermates (data not shown). We further confirmed the

FIG. 2. Expression levels of leptin receptors in different tissues of TKO-OBR mice.
The protein expression levels of leptin receptors were determined by a Western blot assay with a specific antibody recognizing all variants of mouse leptin receptors (15). The arrows indicate detected leptin receptors with the dashed ones indicating the long form of OBR. A, Western blot analysis comparing two different antibodies against the leptin receptors in the hypothalamic (Hypoth.) and WAT extracts. Antibody Ab1 (ours) was a rabbit polyclonal antibody made against a polypeptide representing an extracellular portion of mouse leptin receptors (amino acids 634 -784). Antibody Ab2 was a commercial goat polyclonal antibody made against a peptide representing amino acids 32-51 of mouse leptin receptors (K-20). In each case, the blots were first probed with Ab1. After the Western blot reactions, the blots were stripped and reprobed with Ab2. The two different leptin receptor antibodies clearly recognize the same immunoreactive bands. B, demonstration of the absence of OB-Rb in db/db mice with our OB antibody in a Western blot assay. The long form (OB-Rb) was absent in the hypothalamic and brain samples, whereas the short forms of OBR appear to be normal in the db/db mice. Br, brain; hyp, hypothalamus. WT, wild-type. C, the predominant leptin receptors in the liver and WAT are the short forms. In all three independent lines of TKO-OBR mice, the leptin receptor levels were drastically reduced either Ͼ80% (one line) or Ͼ90% (two lines), whereas the hepatic expression levels of leptin receptors were unaffected. We used STAT3 as a control protein for loading. wt, wild-type. D, the central nervous system and hypothalamic expression of leptin receptors (both the long and the short forms) was not affected in the TKO-OBR mice. The molecular mass of the top band (ϳ150 kDa) is consistent with the predicted molecular mass of the OB-Rb. E, Western blot analysis of leptin receptor expression in multiple tissues. The predominant variants of the leptin receptor expressed in the skeletal muscle (SM; lanes 1 and 2), heart (lanes 3 and 4), lung (lanes 5 and 6), kidney (lanes 7 and 8), spleen (lanes 9 and 10), testis (lanes 11 and 12), and brown fat (lanes 13 and 14) are the short forms. The sizes of the short forms vary slightly from tissue to tissue (110 -120 kDa), most likely reflecting tissue-specific post-translational modifications. In skeletal muscle, brown fat, and to a less extent, the brain tissues, the bands with very high molecular mass (Ն200 kDa) are similar to the reported glycosylated leptin receptors (48). The lanes with odd numbers represent the tissue samples from the wild-type littermates; the lanes with even numbers represent the tissue samples from the TKO-OBR mice. reduction of adipocyte leptin receptor expression at the mRNA level. We found that the OBR mRNA level in WAT of TKO-OBR mice was reduced by 80% relative to that of wild-type littermates (n ϭ 5, p Ͻ 0.05), whereas the hepatic ORB mRNA level was essentially unchanged (n ϭ 6), further demonstrating a differential targeting of leptin receptor expression in these two tissues.
Consistent with the RNA expression pattern of the transgene, the levels of leptin receptors in the TKO-OBR mice were not affected in the brain, brown fat, heart, lung, spleen, kidney, skeletal muscle, and testis (Fig. 2, D and E). To eliminate the concerns that the transgene might alter the levels of leptin receptor in the hypothalamus, we also isolated hypothalamic tissues of TKO-OBR mice and analyzed their expression level of leptin receptor (examples shown in Fig. 2D). We found no detectable difference in the hypothalamic expression of leptin receptors between the TKO-OBR mice and their wild-type littermates (wild-type 100% versus TKO-OBR 107 Ϯ 10%, n ϭ 4 for each genotype). The results thus far indicate that TKO-OBR mice have adipocyte-selective deficiency of leptin receptors and that they maintained normal levels of leptin receptors in the central nervous system, particularly in the hypothalamus.
To provide a functional proof that the leptin actions were reduced in WAT, we examined the ability of leptin to stimulate several signaling pathways in WAT. It has been well demonstrated that leptin can activate PI3-kinase, MAPK (p42/44), and STAT3 in a variety of tissues including WAT (30 -32). Following the intraperitoneal injection of leptin (1 mg/kg of body weight), the STAT3, PI3-kinase, and MAPK were analyzed through either direct measurement of kinase activity or detection of specific phosphorylation. The injected dosage of leptin was consistent with those in previous studies (32). The representative samples are shown in Fig. 3. Compared with those of saline-injected samples, the tyrosine phosphorylation (pY705) of STAT3, the PI3-kinase activity, and the MAPK phosphorylation were all significantly elevated (by 3-, 2.5-, and 2.5-fold, respectively) in the WAT of wild-type mice (Fig. 3B). However, in parallel experiments, leptin failed to induce any detectable activation of these signaling pathways in the WAT of TKO-OBR mice (Fig. 3, A and B). In agreement with the The values from saline-treated samples were arbitrarily set at 1.0. Each set of experiments was repeated twice. n ϭ 4 for each condition in each genotype in the STAT3 experiment, n ϭ 5 for the evaluation of PI3-kinase and MAPK. The asterisk (*) indicates p Ͻ 0.05 in a two-tailed Student's t test. C, the hepatic samples were also analyzed for leptin-induced activation of PI3-kinase and MAPK (p42/44). Representative samples are shown here. Leptin activated PI3-kinase by 2.5-and 2.7-fold (p Ͻ 0.01) in the wild-type and transgenic samples, respectively. Similarly, MAPK was activated by 2.0-and 1.9-fold in the wild-type and transgenic samples, respectively (n ϭ 4 for each condition in each genotype). D, isolated primary adipocytes were treated with leptin for 10 min before being harvested for detection of leptin-stimulated phospho-MAPK and insulin-stimulated phospho-Akt. Each lane represents the adipocytes from two different animals of the same genotype, and each set of experiment was repeated twice. E, the quantitative analysis of relative activation of MAPK by leptin and c-Akt by insulin was performed with NIH Image 6.0. The symbol # indicates p Ͻ 0.01 in a two-tailed Student's t test.
observation that the hepatic expression of leptin receptors was normal (Fig. 2C), leptin still maintained its hepatic signaling in the liver of TKO-OBR mice, which was reflected in the activation of PI3-kinase and MAPK (p42/44) (Fig. 3C). To confirm further the lack of leptin signaling in the adipocytes of TKO-OBR mice, we also isolated adipocytes from TKO-OBR mice and tested the ability of leptin to stimulate MAPK in vitro. After a 10-min incubation, 5 nM leptin was able to induce MAPK (p42/44) activation by 2.7-fold in the wild-type adipocytes (p Ͻ 0.01) but failed to do so in the TKO-OBR adipocytes (Fig. 3, D and E). The TKO-OBR adipocytes still maintained the signaling response to insulin as evaluated by the phosphorylation of c-Akt at Ser-473 (Fig. 3D). However, insulin-induced activation of c-Akt was reduced significantly in the TKO-OBR adipocytes compared with that in the wild-type adipocytes (1.8fold versus 3.3-fold in the wild-types) (Fig. 3, D and E), a result consistent with the phenotype of insulin resistance (see below). These results demonstrate that the residual OBR activities in WAT are incapable of mediating leptin signaling in the WAT of TKO-OBR mice.
In agreement with the hypothalamic level of leptin receptors not being affected, we found that the TKO-OBR mice consumed the same quantity of food as the wild-type littermates (Fig. 4A). However, despite comparable caloric intake, both the male and female TKO-OBR mice gradually developed significantly higher body weight than their wild-type littermates (Fig. 4B). This difference became significant by week 8 for the males and week 10 for the females (Fig. 4B). By week 22, the body weight of both the male and the female TKO-OBR mice was on average more than 20% higher than that of wild-type littermates (37.9 Ϯ 2.2 versus 31.0 Ϯ 1.6 g, 30.1 Ϯ 2.3 versus 24.8 Ϯ 1.5 g, respectively; Fig. 4B). Virtually all of this body weight gain was from fat tissue. At 18 -19 weeks of age, the genital fat pad weight of TKO-OBR mice was significantly greater than that of wild-type littermates (Table I). A whole body composition analysis also suggests that the TKO-OBR mice had significantly more fat than their wild-type littermates (Table I). A histological analysis of the WAT revealed that the sizes of the adipocytes in the TKO-OBR mice were on average 2.3-fold of those in the wild-type littermates (p Ͻ 0.0001) (Supplemental Data 2), suggesting that the increased adiposity was due at least in part to adipocyte hypertrophy. Interestingly, the TKO-OBR mice (male or female) also have ϳfour times as much brown fat compared with their wild-type littermates (Table I). It remains to be determined histologically whether the increase in brown fat weight in TKO-OBR mice was caused by increased lipid content or a result of infiltration of white adipocytes.
The development of increased body weight and adiposity in the context of normal food intake implies that the TKO-OBR mice should have lower metabolic rate, hence energy expenditure, than their wild-type littermates. As an initial test of this concept, we measured the body (rectal) temperature of these mice because body temperature is proportional to energy expenditure. As shown in Fig. 4E, both the male and female TKO-OBR mice displayed significantly lower body temperature than their wild-type littermates. Specifically, the rectal temperature of the male transgenics was on average 1.5°C lower than the wild-type males (35.5 Ϯ 0.5 versus 37.0 Ϯ 0.2°C, p Ͻ 0.0001). Similarly, the rectal temperature of the female transgenics was about 1.4°C lower than their wild-type counterparts (35.3 Ϯ 0.5 versus 36.7 Ϯ 0.1°C, p Ͻ 0.0001). As an additional control, we also measured the rectal temperature of several male ob/ob mice (of the same age as the other male mice) in these experiments. The average body temperature of the ob/ob mice was 34.5°C, which is in complete agreement with what had been reported before (33). These results sup-ported our hypothesis that the TKO-OBR mice have lower energy expenditure than their wild-type littermates.
We explored the consequences of leptin receptor deficiency in the WAT of TKO-OBR mice in more detail, specifically with regard to insulin sensitivity, glucose, and fat metabolism, and adipocyte gene expression. Using the intraperitoneal glucose tolerance test, we found that the male TKO-OBR mice displayed impaired glucose tolerance as early as week 6 -7 (Fig.  5A) and that by week 13-14, both the male and female TKO-OBR mice became glucose-intolerant compared with the wildtype controls (Fig. 5B). Because at week 6 -7, there was no difference in body weight between the mutant and wild-type mice, the observed glucose intolerance is likely because of the reduction of leptin receptors in WAT rather than an increase in adiposity. Fasting glucose concentrations were elevated modestly in both the male and female TKO-OBR mice relative to the wild-type littermates (see the zero time point, Fig. 5B), which was associated with a 4-fold elevation of serum insulin level (Fig. 5D). The serum leptin level in the TKO-OBR mice was also four times of that in the wild-type littermates (Fig.  5D). Consistent with the glucose intolerance, the response of blood glucose concentrations to an intraperitoneal injection of insulin was also impaired (Fig. 5C). These data demonstrate that loss of leptin receptors in WAT can result in insulin resistance.
Adipocyte-specific reduction of leptin receptors also leads to marked dysregulation in fat metabolism. We found that the fasting serum triglyceride concentrations in the TKO-OBR mice were more than twice of those in wild-type littermates (93.1 Ϯ 19.0 mg versus 40.9 Ϯ 10.0 mg/ml, Fig. 6A). In addition, the triglyceride levels in the liver and skeletal muscle were almost 3-fold higher in the transgenics compared with wildtype littermates (18.0 Ϯ 3.0 mg versus 6.5 Ϯ 1.5 mg, 39.4 Ϯ 10.0 mg versus 14.1 Ϯ 4.0 mg/g of tissue, respectively; Fig. 6B). The fasting serum glycerol and free fatty acid levels were lower in the TKO-OBR mice than in the wild-type littermates (Fig. 6C), which is consistent with the reduced levels of ␤ 3 -adrenergic receptors and increased glycerol kinase expression in the WAT (Ref. 34 and Table II).
We hypothesized that leptin receptor deficiency in the WAT might cause significant changes in the expression levels of critical regulatory genes in the adipocytes. As an initial test of this concept, we employed real time PCR (Taqman) assay to analyze selectively the adipocyte expression of several genes involved in glucose and fat metabolism (Table II). The expression of TNF-␣ and glycerol kinase were sharply elevated (ϳ15and 11-fold, respectively). Similarly, the expression levels of leptin and a transcription factor, SREBP-1, were increased by more than 3-fold (Table II). On the contrary, the mRNA levels of fatty acid synthase, ␤ 3 -adrenergic receptor, and a secretory protein adiponectin (also known as ACRP30, adipoQ, and apm-1) (35,36) were reduced by 3-to ϳ6-fold in the TKO-OBR mice relative to those in the wild-type littermates (Table II). We also found that the levels of PPAR-␥ and acyl-CoA oxidase were unchanged. These results of RNA analysis were confirmed using Western blot assays. As shown in Fig. 7, the protein expression levels of leptin, TNF-␣ fatty acid synthase, SREBP-1c, ␤ 3 -adrenergic receptors, and PPAR-␥ all displayed changes similar to their corresponding mRNA levels. Consistent with these observations, the serum level of TNF-␣ of TKO-OBR mice was significantly higher, and adiponectin level significantly lower, than those of the wild-type littermates (Fig.  7B). Furthermore, lipolysis induced by a ␤ 3 -agonist, BRL37344 (Tocris, Ellisville, MO), was reduced significantly in the isolated primary adipocytes of TKO-OBR mice (Fig. 7C). Interestingly, the basal lipolysis in the TKO-OBR adipocytes was FIG. 4. Measurement of food intake, body weight, adipocyte morphology, and rectal temperature. In all cases, the filled symbols represent the TKO-OBR mice and the open symbols the wild-type mice. A, cumulative 24-h food intake was monitored for mice at 4, 10, and 20 weeks of age. There was no significant difference in food consumption between TKO-OBR mice and their wild-type littermates. n ϭ 8 for both the TKO-OBR males and the wild-type littermates; n ϭ 11 for both the TKO-OBR and the wild-type females. B, the male and female TKO-OBR mice displayed significantly greater body weight than their wild-type counterparts starting from week 8 and week 10, respectively. n ϭ 9 for both the TKO-OBR and wild-type males; n ϭ 7 for both the TKO-OBR and wild-type females. Each result is expressed as the mean Ϯ S.E. All corresponding comparisons were subjected to Student's t test. The symbols, *, **, and # denote p Ͻ 0.01, p Ͻ 0.005, and p Ͻ 0.05, respectively. C, measurement of rectal temperature. The number of animals/group (n) is indicated. All comparisons were subjected to a two-tailed Student's t test. The comparisons of the wild-type versus the transgenic, the wild-type versus ob/ob mice, and the transgenic versus the ob/ob mice all reached statistical significance. The symbols # and * indicate p Ͻ 0.0001 and Ͻ0.03, respectively, in a two-tailed Student's t test.
increased by ϳ2-fold (p Ͻ 0.05) compared with that in the wild-type (Fig. 7C), which may potentially be the result of the elevated TNF-␣ expression as reported previously (37).
Rodents fed with a high fat diet often develop increased body weight and adiposity, insulin resistance, and hyperglycemia (10,38). We challenged the TKO-OBR mice with a high fat diet starting at week 5 of age to test whether high fat feeding might further intensify the body weight growth and insulin resistance. As shown in Fig. 8, A and B, during high fat feeding, the TKO-OBR mice displayed a far more rapid body weight increase than their wild-type littermates. After 9 weeks of high fat feeding, the weight gain of male TKO-OBR mice was 25.5 Ϯ 2.1 g versus 18.0 Ϯ 1.7 g in the wild-type male littermates, representing 161% versus 106% in body weight increase, respectively (Fig. 8B). The weight of female TKO-OBR increased by 14.7 Ϯ 1.5 g versus 9.1 Ϯ 1.0 g in wild-type littermates (Fig.  8B), representing 101% versus 61% increase in body weight, respectively. Concomitant with this rapid body weight increase, the high fat-fed TKO-OBR mice displayed severe glucose intolerance (Fig. 9A). This contrasted dramatically with wild-type mice on the same high fat diet. The wild-type mice displayed a level of glucose intolerance similar to that observed in the TKO-OBR mice on standard chow diet (Figs. 9A and 5B). At the 2-h time point following glucose loading, the blood glucose level of the male TKO-OBR mice was still at 320 Ϯ 60 mg/dl, compared with 150 Ϯ 15 mg/dl in the male wild-type littermates (Fig. 9A). In the male TKO-OBR mice, fasting blood glucose levels reached 110 Ϯ 10 mg/dl, well above the 85 Ϯ 9 mg/dl level in the wild-type males (Fig. 9A), and the nonfasting blood glucose levels reached at 230 Ϯ 20 mg/dl, compared with  5. Tests of intraperitoneal glucose tolerance and insulin. For the glucose tolerance test, the mice were injected intraperitoneally with glucose at a dose of 2 g/kg of body weight, and the blood glucose levels were monitored through tail vein blood samples. The filled symbols represent the TKO-OBR mice, and the open symbols represent the wild-type littermates. A, at 6-7 weeks of age, although the females still display normal glucose tolerance, the male TKO-OBR mice have already become glucoseintolerant. B, at 13-14 weeks of age, both the male and the female TKO mice were glucose-intolerant. n ϭ 7 for both the TKO-OBR and wild-type males; n ϭ 8 for both the TKO-OBR and wild-type females. C, the response in blood glucose following an intraperitoneal injection of insulin (0.3 unit/kg of body weight). The blood glucose concentrations at different time points were expressed as a percentage of the fasting blood glucose concentrations. n ϭ 6 for each group, with 3 males and 3 females at 13-14 weeks of age. Results are expressed as mean Ϯ S.E. D, serum insulin and leptin levels of TKO-OBR mice (transgenic, Tg) and their wildtype (WT) littermates. The mice were at 19 -20 weeks of age when the blood samples were taken. n ϭ 8 for each measurement with 4 males and 4 females. The symbols *, **, and *** denote p Ͻ 0.05, p Ͻ 0.03, and p Ͻ 0.01, respectively, in a two-tailed Student's t test.
160 Ϯ 15 mg/dl (n ϭ 5 for each group) in the wild-type counterparts. Furthermore, following an intraperitoneal injection of insulin, the decline in blood glucose was severely impaired in the TKO-OBR mice (Fig. 9B). Taken together, our results demonstrate that the TKO-OBR mice, carrying adipocyte-specific reduction of leptin receptors, are significantly more sensitive to the feeding of a high fat diet both in terms of body weight growth and the development of insulin resistance than the wild-type littermates.

DISCUSSION
Although we originally intended to use the technical knockout strategy to reduce the expression of OBRs in both fat and liver, the differential targeting success only caused the diminution of OBRs in the WAT of the TKO-OBR mice. The transgene did not affect the hepatic OBR expression at both mRNA and protein levels. Similarly, the expression of leptin receptors was unaffected by the modest expression of TKO-OBR transgene in kidney (Fig. 1D). In this regard, almost identical ob-servations have been made in our recently created TKO-phosphodiesterase 3B transgenic mice where the adipocyte expression of phosphodiesterase 3B proteins was almost completely eliminated, but its hepatic protein expression was not altered (see Supplemental Data 3). These targeting failures in liver are somewhat surprising in light of the earlier success of TKO technology in reducing the hepatic expression of G␣ i2 and G␣ q (22)(23)(24). The success of antisense strategy depends on a variety of factors including the targeted genes, the unwinding/ modifying enzyme activities on the antisense-and sense-RNA duplex in a given type of cells, and the relative expression levels of the antisense sequences versus the levels of targeted mRNAs (39). Although it remains to be determined which of these factors contributed to the failure of reducing hepatic OBR expression, we suspect that one of the contributing factors in the earlier targeting success of G␣ i2 and G␣ q might have been their low mRNA levels in hepatocytes (24,40). We also estimated based on a RT-PCR analysis that the hepatic mRNA level of leptin receptor is at least 100-fold higher than those of G␣ i2 or G␣ q (data not shown). Overall, the success of the TKO strategy seems to be consistent in WAT but gene-dependent in liver. In this particular case, we were able to create the transgenic mice with adipocyte-selective deficiency of leptin receptors. It is very unlikely that the phenotypes of the TKO-OBR mice were the result of the nonspecific effects of the parental transgenic vector because previous studies of using the same strategy to target other genes (such as G␣ i2 ) produced the phenotypes quite distinct from those of the TKO-OBR mice (22). In addition, the expression of the TKO-OBR transgene alone does not affect leptin signaling unless it reduces the expression of leptin receptors (Figs. 1C and 3, A and C).
Recent studies have found that neuronal deletion of leptin receptors did not result in the body weight gain achieved by the ob/ob and db/db mice (14) and that central overexpression of OB-Rb in db/db mice only partially corrected the obesity phenotype (13). These studies led us to speculate that the peripheral effects of leptin, such as in WAT, may also help regulate body weight. The TKO-OBR mice described herein represent an ideal genetic model for our investigation of the peripheral FIG. 6. Serum and tissue triglyceride and free fatty acid levels in liver and skeletal muscle. Mice were fasted for 8 h before their serum and tissue samples were taken. A, serum triglyceride (TG) levels of TKO-OBR mice were significantly higher than those of wild-type (wt) littermates. B, the triglyceride levels in the liver and skeletal muscle samples were taken from the mice at 19 -20 weeks of age. The triglyceride levels in liver and muscle were measured with a Sigma triglyceride assay kit and are expressed as mg/g of tissue weight. C, serum levels of glycerol and free fatty acids in the TKO-OBR and wild-type mice after overnight fasting. In the triglyceride measurement, n ϭ 6 for each group and each tissue. In the glycerol and Free fatty acid measurement, n ϭ 6 and 5 for wild-type and transgenic group, respectively. The symbols * and ** denote p Ͻ 0.05 and p Ͻ 0.01 in a two-tailed Student's t test.

TABLE II Dysregulation of gene expression in white adipose tissues
of TKO-OBR mice The analysis was based on a Taqman (ABI 7700 system) quantitative RT-PCR assay. The designs of primers and probes and the linearity of the assays for each gene can be found on our web site www.pitt.edu/ Ϸazhao/. We analyzed samples from four to six for each gene, and each sample was analyzed three times at three different concentrations. The wild-type expression levels were always set at 1.0. The 18 S rRNA was used as an internal control of the RNA quantity. The results were expressed the -fold of changes over the wild-type levels.

Genes
Increase 1 or decrease (2) ϪFold of changes actions of leptin in WAT. The phenotypes of TKO-OBR mice, such as increased adiposity and increased sensitivity to high fat feeding, indicate that control of body weight by leptin requires not only its actions at the hypothalamus but also the distinct autocrine/paracrine actions on the fat tissue. Furthermore, because the majority of the leptin receptors in WAT are the short forms, it remains to be seen which of the phenotypes of the TKO-OBR mice are mediated through the short forms of OBR. The increased body weight and adiposity in TKO-OBR mice occurred despite their normal food intake, suggesting decreased energy expenditure in the TKO-OBR mice. In an initial test of this concept, the TKO-OBR mice did show lower body temperature relative to the wild-type littermates, which suggests that leptin-regulated energy expenditure likely requires its complementary actions in the hypothalamus and in the WAT. However, this concept needs to be validated further by other independent approaches such as measurement of oxygen consumption. If eventually proven, the presumed reduction of metabolic rate in TKO-OBR mice can be tentatively linked to FIG. 7. Analysis of adipocyte gene expression. A, confirmation of the results from the quantitative RT-PCR analysis with Western blot assays. The adipose tissue samples (three from the wild-type, three from the TKO-OBR mice) were homogenized in a protein extraction buffer, and the resulting protein samples were subject to a standard Western blot assay with each specific primary antibody as indicated. B, determination of serum TNF-␣ and adiponectin levels in the transgenic mice and their wild-type littermates. In the analysis of TNF-␣, n ϭ 18 (9 males and 9 females) for the wild-type (wt); n ϭ 13 for the transgenic (6 males and 7 females). In the analysis of adiponectin, n ϭ 11 for the wild-type and the transgenic (6 males and 5 females). The symbol # indicates p Ͻ 0.0001 in a two-tailed Student's t test. C, determination of lipolysis (using medium glycerol) stimulated by a ␤ 3 -agonist, BRL37344, in isolated primary adipocytes from the TKO-OBR mice and their wild-type littermates. The experiment was repeated twice, and each condition was set in triplicates. The symbols * and # indicate p Ͻ 0.05 and 0.005, respectively, in a two-tailed Student's t test. When compared with the corresponding controls, the BRL37344-stimulated lipolysis was significantly attenuated in the TKO-OBR adipocytes (at 200 nM, 2.8-fold versus 7.9-fold in the wild-types; at 1,000 nM, 3.6-fold versus 11.7-fold in the wild-types; p Ͻ 0.01). two potential mechanisms. Similar to the ob/ob mice, the TKO-OBR mice have reduced expression of ␤ 3 -adrenergic receptors in the WAT (Ref. 41, Table II, and Fig. 7A). The ␤ 3 -adrenergic receptor knock-out mice developed increased adiposity while having normal food intake (34,42), which was primarily the result of decreased energy expenditure compared with the wildtype littermates. The decreased ␤ 3 -adrenergic receptor expression in WAT will presumably lead to reduced supply of metabolic fuel (i.e. free fatty acids) to other organs, such as liver and skeletal muscle in the TKO-OBR mice. Indeed, the fasting serum levels of glycerol and free fatty acids are lower in the TKO-OBR mice than in the wild-type littermates (Fig. 6C). Alternatively, the significant increase of TNF-␣ expression in the TKO-OBR mice (Table II and Fig. 7, A and B) might have suppressed the expression of uncoupling protein-1 in the brown adipose tissue as shown before (43), which in turn could also lead to decreased thermogenesis.
Selective deficiency of leptin receptors in WAT of TKO-OBR mice led to insulin resistance and dyslipidemia. Because TNF-␣ has been well established as an antagonist of insulin actions both in vitro and in vivo (44,45), the significant increase of TNF-␣ expression in the TKO-OBR mice might be one of the FIG. 8. Comparison of body weight growth under high fat diet feeding. A, a male TKO-OBR mouse (on the right) and a male wild-type littermate (on the left) at 13 weeks of age. Both have gone through 9 weeks of high fat diet (19% lard and 1% corn oil diet). When the picture was taken, the TKO-OBR mouse weighed ϳ43 g, and the wild-type counterpart weighed ϳ32 g. B, cumulative body weight gain after 5 and 9 weeks of feeding with either a standard chow or a high fat diet. The TKO-OBR mice and the wild-type littermates were either fed with high fat diet or were continued on standard chow. The results are expressed as the mean Ϯ S.E. n ϭ 6 for each group. * denotes p Ͻ 0.01 in a two-tailed Student's t test.
main reasons for their insulin resistance. Similarly, adiponectin has been recently demonstrated to be able to stimulate fatty acid oxidation in skeletal muscle and liver (46,47). Therefore, it remains to be determined whether the reduction of adiponectin expression in the WAT of TKO-OBR mice may be one of the primary causes of elevated triglyceride levels in skeletal muscle and liver. These data also suggest to us that some of the phenotypes of the TKO-OBR mice might be secondary to the dysregulation of hormonal levels in the WAT.
Gene expression profiles in the adipocytes of ob/ob mice have been examined recently using microarray analysis (41). Because the ob/ob mice also have deficiency of leptin functions in the central nervous system, we need to distinguish between the central and the peripheral effects of leptin on adipocyte gene expression. The TKO-OBR mice can be utilized to address this issue because of their normal central nervous system expression of leptin receptors. In this regard, the expression of ␤ 3adrenergic receptor, leptin, fatty acid synthase, and SREBP-1 in the WAT of ob/ob mice all showed changes similar to those observed in the WAT of TKO-OBR mice (41), suggesting a peripheral regulation of these genes by leptin (Table II). However, other overt conditions in the TKO-OBR mice such as hyperinsulinemia and elevated TNF-␣ level may also induce changes in adipocyte gene expression. For example, previous studies with microarray assays have shown that chronic infusion of TNF-␣ into rats chronic exposure of 3T3L1 adipocytes to TNF-␣ (6 h to 5 days) could lead to dysregulation of many adipocyte genes including reduction of fatty acid synthase and adiponectin expression (37,48). Thus, further studies are required to define which of the expression changes observed in FIG. 9. The effects of high fat diet on the glucose metabolism. A, intraperitoneal glucose tolerance test. The TKO-OBR mice (filled symbols) and their wild-type littermates (open symbols), at 13-14 weeks of age had been fed a high fat diet for 9 weeks before an intraperitoneal glucose tolerance test. n ϭ 6 for each group of mice. B, response in blood glucose concentrations following an intraperitoneal injection of insulin (0.3 unit/kg of body weight). n ϭ 6 for each group with 3 males and 3 females at 13 weeks of age in each group. Open symbols represent the wild-type mice, the filled symbols the TKO-OBR mice. The symbols #, *, and ** denote p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.005, respectively, in a two-tailed Student's t test.
the TKO-OBR adipocytes are directly the result of the leptin receptor manipulation and which of the changes are secondary. Overall, our limited analysis of adipocyte gene expression suggests a peripheral regulatory effect of leptin on several adipocyte genes involved in glucose and lipid metabolism.
In summary, the TKO-OBR mice with adipocyte-specific deficiency of leptin receptors develop increased adiposity, decreased energy expenditure, insulin resistance, dyslipidemia, and dysregulation of adipocyte gene expression. Our results raise the possibility that peripheral leptin resistance (not necessarily at the receptor level) may be a major contributing factor to the insulin resistance observed in obese subjects. Based on the results seen in the TKO-OBR mice, the hypothesis that peripheral, particularly fat tissue-specific, leptin resistance may be one of the primary causes of insulin resistance and type 2 diabetes merits further investigation.