Central Leptin Regulates the UCP1 and obGenes in Brown and White Adipose Tissue via Different β-Adrenoceptor Subtypes*

The three known subtypes of β-adrenoreceptors (β1-AR, β2-AR, and β3-AR) are differentially expressed in brown and white adipose tissue and mediate peripheral responses to central modulation of sympathetic outflow by leptin. To assess the relative roles of the β-AR subtypes in mediating leptin's effects on adipocyte gene expression, mice with a targeted disruption of the β3-adrenoreceptor gene (β3-AR KO) were treated with vehicle or the β1/β2-AR selective antagonist, propranolol (20 μg/g body weight/day) prior to intracerebroventricular (ICV) injections of leptin (0.1 μg/g body weight/day). Leptin produced a 3-fold increase in UCP1 mRNA in brown adipose tissue of wild type (FVB/NJ) and β3-AR KO mice. The response was unaltered by propranolol in wild type mice, but was completely blocked by this antagonist in β3-AR KO mice. In contrast, ICV leptin had no effect on leptin mRNA in either epididymal or retroperitoneal white adipose tissue (WAT) from β3-AR KOs. Moreover, propranolol did not block the ability of exogenous leptin to reduce leptin mRNA in either WAT depot site of wild type mice. These results demonstrate that the β3-AR is required for leptin-mediated regulation ofob mRNA expression in WAT, but is interchangeable with the β1/β2-ARs in mediating leptin's effect on UCP1 mRNA expression in brown adipose tissue.

In mice the absence of leptin (ob/ob) or its functional receptor (db/db) produces a complex metabolic syndrome characterized by hyperphagia, endocrine abnormalities, and morbid obesity (1). Deposition of excess body fat occurs even when food intake is controlled, suggesting that an important function of leptin is to regulate energy balance through modulation of metabolic efficiency. This view is supported by studies in ob/ob mice showing that leptin-injected animals lose more weight than pair-fed vehicle-injected littermates (2,3). Of particular interest is the observation that leptin-induced weight loss occurs specifically in adipose tissue with little effect in other tissues (3,4). The loss of adipose tissue is associated with an increase in fat oxidation, and the associated shift in fuel selection can be measured as a decrease in the respiratory quotient during leptin repletion (5,6). Thus, adipose tissue is an important target of leptin action and the primary effect is a shift from fat storage to fat mobilization and oxidation.
This leptin-mediated shift in adipocyte function involves a coordinated change in gene expression. Two mechanisms have been postulated and include both centrally mediated effects and direct effects through functional leptin receptors (Ob-Rb) on the adipocyte (7)(8)(9). It should be noted, however, that although supraphysiologic levels of leptin are capable of producing significant direct effects on adipose tissue (10,11), increments of plasma leptin in the physiological range are thought to act primarily through receptors in the hypothalamus (10). Occupancy of hypothalamic leptin receptors promotes activation of the sympathetic nervous system (12)(13)(14)(15), and recent studies using surgical (16), chemical (17), and transgenic approaches (18) have shown that norepinephrine is required for leptin effects on gene expression in both brown and white adipose tissue (19 -22). Thus, several lines of evidence support an emerging consensus that norepinephrine represents the peripheral signal linking hypothalamic leptin receptors to leptindependent changes in adipocyte gene expression.
Occupancy of each of the three known ␤-adrenoreceptor (␤-AR) 1 subtypes leads to activation of adenylyl cyclase in adipose tissue (23), but the combination of unequal expression and differing affinities for endogenous agonists has made it challenging to assess the relative contributions of each receptor subtype in various physiological states. Recent studies also demonstrate that ␤-adrenoreceptor subtypes may be differentially coupled to various functions within the adipocyte (24 -26). Therefore, we have attempted to identify the ␤-adrenoreceptor subtype(s) that mediate the effects of leptin on gene expression in various adipose tissue depot sites. Using intracerebroventricular injections of leptin in mice lacking ␤ 3 -adrenoreceptors, we show that different complements of ␤-adrenoreceptor subtypes are required to transduce leptin's effects on gene expression in white versus brown adipose tissue.

EXPERIMENTAL PROCEDURES
Experimental Animal Protocol-Mice in each of the four experiments described below were housed in pairs in solid-bottom cages with continuous access to chow (Purina Mouse Chow, Ralston Purina, St. Louis, MO) and water. Room temperature was maintained at 22-23°C, and the lights were on a 12-h light/dark cycle. The animals were acclimated for 1-2 weeks prior to each study, and injected thereafter with various agents according to protocols specified under each experiment. All in-jections were given 2 h following the start of the light cycle, and all mice were sacrificed 2 or 4 h after the last injection in the series. Thereafter, interscapular BAT, as well as epididymal, retroperitoneal, and inguinal WAT depots were carefully removed, weighed, and used for preparation of total RNA or isolation of adipocytes.
Intracerebroventricular Injections-Mice were anesthetized by inhalation of isoflurane and a guarded, blank 27-gauge 0.5-inch needle was used to create a guide injection site 0.7 mm posterior to bregma and 1.0 mm lateral to midline at a depth of 4.0 mm (27). In the experiments proper, a 10.0-l Hamilton 1700 series gastight syringe (Hamilton, Reno, NV) was used to inject artificial cerebrospinal fluid (aCSF), murine leptin, or rat neuropeptide Y in a volume of 2-5 l. Correct positioning of the guide injection site was confirmed prior to the start of each experiment by monitoring feeding behavior following injection of neuropeptide Y (0.075 g/g body weight). Mice failing to respond to neuropeptide Y were removed from the experiment. aCSF, consisting of 70 mM NaCl, 6 mM KCl, 0.7 mM CaCl 2 , 0.85 mM MgCl 2 , 0.75 mM Na 2 HPO 4 , 0.10 mM NaH 2 PO 4 , and 0.1% untreated bovine serum albumin, was injected in a volume of 5 l. Thereafter, mice were monitored to ensure full recovery.
Experiment 1-Seven-week-old male C57BL/6J mice were obtained from Jackson Laboratories (Bar Harbor, ME) and acclimated for 2 weeks prior to the study. Thereafter, mice were injected ICV with leptin (0.1 g/g body weight/day) for 1, 2, or 3 days, and representative mice were sacrificed 2 and 4 h following the respective injections on each day. Separate groups of mice received intraperitoneal injections of CL-316,243 (1 g/g body weight/day) for 1 or 2 days, and were sacrificed 2 h after injection. Control mice were injected ICV with aCSF. Interscapular BAT, as well as epididymal, retroperitoneal, and inguinal WAT depots were removed, weighed, and used to prepare total RNA as described previously (19).
Experiment 2-Seven-month-old male FVB/NJ (WT) mice and agematched FVB/NJ male mice with a targeted disruption of the ␤ 3adrenoreceptor (␤ 3 -AR KO) gene (28) were acclimated as described above and equal numbers of each phenotype were randomly assigned to one of four treatment groups. The mice in group 1 received ICV injections of aCSF for 3 days, and food was provided ad libitum. Mice in group 2 received ICV injections of mouse leptin (0.1 g/g body weight/ day), while group 3 received ICV leptin (0.1 g/g body weight/day) in combination with intraperitoneal injections of propranolol (20 g/g body weight/day). Food was provided ad libitum, and intake was measured. Mice in group 4 received intraperitoneal injections of propranolol (20 g/g body weight/day) for 3 days and were pair-fed to the mean intake of mice in groups 2 and 3. Two h after the final injection, the mice were sacrificed and tissues processed as described above.
Experiment 3-Seven-month-old male FVB/NJ mice and agematched ␤ 3 -AR KO male mice were acclimated as described above. Thereafter, the WT and ␤ 3 -AR KO mice were sacrificed, and the adipose tissue depot sites were carefully removed and used for cell isolation and membrane preparation as described below.
Experiment 4 -Male C57BL/6J (ob/ob) mice and their lean littermates (ϩ/?) were obtained from Jackson Laboratories at 6 weeks of age and randomly assigned to one of two treatment groups. The mice were housed individually at 23°C and equilibrated for 4 days before beginning the experiment. On the morning of the 5th day and for 2 mornings thereafter, half the mice in each phenotype received intraperitoneal injections of recombinant mouse leptin (20 g/g body weight/day), while the remaining mice in each phenotype received vehicle injections. Within each phenotype, the mice receiving vehicle were pair-fed with the mice receiving leptin. Three h following the final injections on day 7, the mice were weighed, sacrificed, and adipose tissue processed as described previously.
Experiment 5-Seven-month-old male FVB/NJ mice and agematched ␤ 3 -AR KO male mice were acclimated as described above. Thereafter, half of the WT and ␤ 3 -AR KO mice were moved to 4°C environmental chambers while the remaining mice in each group were maintained at 23°C. After 4 h, all mice were sacrificed, and the adipose tissue depot sites were carefully removed and used for isolation of total RNA.
Ribonuclease Protection Assay-RNA probes complementary to UCP1 mRNA and leptin mRNA were prepared, labeled, and used as described previously to assay the respective mRNA species (18,19). Using the same method, a ␤ 3 -AR probe was produced by reverse transcription-polymerase chain reaction (5Ј to 3Ј; forward, accccagtgcagccaacacca; reverse, cgcaaccagtttcgcccaagg), purified, and cloned into the pGEM-3Z riboprobe vector (Promega, Madison, WI). The identity of the cloned fragment was confirmed by sequencing, and it corresponded to nucleotides 630 -890, producing a protected fragment of 261 base pairs. ␤ 1 -and ␤ 2 -AR probes corresponding to nucleotides 534 -971 and 789 -971 of the rat ␤ 1 -and ␤ 2 -AR cDNAs, respectively, were obtained from Dr. James Granneman (29). The concentration of UCP1, leptin, and ␤ 1 -, ␤ 2 -, and ␤ 3 -AR mRNA in each sample was determined by comparison to known amounts of sense strand fragments for each gene that were hybridized simultaneously in the ribonuclease protection assay. After autoradiography and densitometry, standard curves were constructed for each gene and used to estimate mRNA concentrations in unknown samples as described previously (18).
Isolation of Adipocytes and Preparation of Plasma Membranes-White adipocytes were prepared from pooled epididymal, retroperitoneal, and inguinal fat pads of 8-month-old WT FVB/NJ and ␤ 3 -AR KO mice by collagenase treatment according to Rodbell (30) with slight modification (31). The cells were washed and resuspended in Krebs-Henseleit-HEPES buffer (pH 7.4), then broken in a Dounce homogenizer in 10 mM TES (pH 7.0) containing 0.25 M sucrose. Crude membranes were collected following centrifugation at 48,000 ϫ g for 20 min and stored at Ϫ80°C. Interscapular brown adipose tissue was carefully dissected free of surrounding tissue, finely minced with scissors, and washed in Krebs-Henseleit-HEPES buffer (pH 7.4). Following centrifugation at 1000 ϫ g, the infranatant was aspirated and the fat cells resuspended in 10 mM TES (pH 7.0) containing 0.25 M sucrose. The brown fat cells were then broken using an overhead stirrer and filtered through nylon mesh, and crude membranes were collected following an initial low speed spin to remove unbroken cells and nuclei. The crude membranes were resuspended and stored at Ϫ80°C.
Competition Binding Assay for ␤ 1 -and ␤ 2 -AR-Radioligand binding assays were conducted with adipocyte membranes using our modification (32) of previously described methods (32)(33)(34). Briefly, 20 g of white and 40 g of brown adipocyte membranes were incubated at 37°C for 1 h with 30 pM 125 I-cyanopindolol (ICYP) in 25 mM HEPES buffer (pH 7.4) containing 12.5 mM MgCl 2 , and increasing concentrations of the ␤ 1 -AR-specific antagonist, CGP-20712A (RBI, Natick, MA). Bound ICYP was collected on filters in a Skatron cell harvester (Molecular Devices, Sunnyvale, CA), and the components of ␤ 1 -AR and ␤ 2 -AR binding of ICYP were resolved by fitting a two-site competition curve to the data by least squares (Prizm; Graphpad Software, San Diego, CA) and analyzed as described previously (32).
Adenylyl Cyclase Assay-Adenylyl cyclase (AC) activity was determined in adipocyte membranes using combinations of receptor subtype selective agonists and antagonists to assess the functional activity of each receptor subtype (31,32,35). The ␤ 3 -AR antagonist, SR 59230A (36) was obtained from Dr. Luciano Manara (Sanofi Midy, Milan, Italy) and solubilized in ethylene glycol at 200 g/ml. The antagonist was used at a final concentration of 10 M as reported previously (36). Other agonists and antagonists were from commercial sources.
Methods of Analysis-One-way analysis of variance was used to compare group means for UCP1 mRNA, leptin mRNA, ␤ 1 -AR mRNA, ␤ 2 -AR mRNA, ␤ 3 -AR mRNA, adenylyl cyclase activity, and ICYP binding in the respective experiments for each variable. The level of protection against type I errors was set at 5%, and P values for specific treatment comparisons are presented under "Results." Materials-All reagents, except where noted, were obtained from Sigma and were of the highest reagent grade. T1-RNase and Trizol LS reagent were from Life Technologies, Inc. Oligonucleotide primers and DNA sequencing were generated at the DNA Core Facility at the Medical University of South Carolina. [␣-32 P]CTP and [ 125 I]iodocyanopindolol were purchased from PerkinElmer Life Sciences. 1-Chloro-2,2,2-trifluoroethyl difluoromethyl ether (isoflurane) was from Ohmeda PPD (Liberty Corner, NJ). CL-316,243 was a gift from Wyeth Ayerst Research (Princeton, NJ). Recombinant methionyl mouse leptin was kindly provided by Amgen (Thousand Oaks, CA). The ␤ 3 -adrenoreceptor antagonist SR59230A was kindly provided by Dr. Luciana Manara (Sanofi, Italy). Breeding pairs of the ␤ 3 -AR KO mice (28)were kindly provided by Dr. Brad Lowell (Beth Israel Hospital, Boston, MA).

Time Course of Central Leptin Effects on BAT UCP1 in C57BL/6J
Mice-To determine the time course for leptin's effects on UCP1 expression in BAT and distinguish between central versus peripheral effects of the peptide, mice were given ICV injections of leptin or artificial CSF and sacrificed 2 and 4 h later on successive days. This protocol complemented our normal treatment regimen of 3 days and revealed that leptin increased UCP1 mRNA 2 and 4 h after the initial injection on day 1 (Fig. 1). The ϳ2-fold increase was maximal 4 h after leptin injection and comparable to the increase produced by the ␤ 3 -adrenoreceptor selective agonist, CL316,243 (data not shown). Similar results were observed on day 2, but, because a somewhat larger increase in UCP1 mRNA was seen on day 3, the 3-day injection protocol was adopted for subsequent studies (see Fig. 2). These results confirm that the effect of leptin was both rapid and centrally mediated, and they are consistent with our previous work showing that leptin regulates gene expression in adipose tissue through a central site of action (18).
Effect of Leptin on BAT UCP1 in ␤ 3 -KO Mice-The purpose of experiment 2 was to determine whether the absence of the ␤ 3 -AR in both BAT and WAT would compromise the ability of centrally administered leptin to regulate gene expression in each depot site. A secondary objective was to determine whether the ␤ 1 /␤ 2 -ARs could substitute for the missing ␤ 3 -AR or whether ␣-adrenoreceptors might also be involved in mediating the response. WT age-matched FVB/NJ mice served as positive controls and responded to ICV-injected leptin with a 4 -5-fold increase in UCP1 mRNA in BAT (p Ͻ 0.01, Fig. 2A). Intraperitoneal injection of propranolol to block ␤ 1 /␤ 2 -ARs reduced basal expression of UCP1 mRNA in WT mice, but did not block leptin's ability to induce UCP1 mRNA expression by 4-fold (p Ͻ 0.01, Fig. 2A). These results suggest that the ␤ 1 /␤ 2 -ARs and ␤ 3 -ARs are interchangeable with respect to their ability to mediate central effects of leptin on UCP1 expression. This conclusion is supported by results from the ␤ 3 -AR KO mice, which show that central leptin produced a 3-4-fold increase in UCP1 mRNA (p Ͻ 0.01, Fig. 2B), and the finding that propranolol completely blocked the ability of leptin to increase UCP1 mRNA in BAT from these animals (p Ͻ 0.01, Fig. 2B). These findings also make the case that ␣-adrenoreceptors are not involved in the response and that ␤-adrenoreceptors are the primary adrenergic receptors mediating the effects of leptin on UCP1 mRNA.
Effect of ICV Leptin on Leptin mRNA in WAT-To test the hypothesis that centrally administered leptin regulates its own expression in WAT, we measured leptin mRNA in epididymal and retroperitoneal depot sites from WT and ␤ 3 -AR KO mice treated with leptin. The two sites were chosen as being representative of WAT depots that respond differently to adrenergic stimulation (19). In WT mice, ICV leptin produced a highly significant reduction (p Ͻ 0.01) of leptin mRNA in epididymal WAT from 0.021 Ϯ 0.003 fmol of leptin mRNA/g of RNA to 0.005 Ϯ 0.002 fmol/g of RNA (Fig. 3A). A similar reduction in leptin mRNA (p Ͻ 0.001) was noted in the retroperitoneal WAT depot of these mice (Fig. 4A). Figs. 3A and 4A show that propranolol produced a modest increase in basal leptin mRNA in both depot sites of WT mice (p Ͻ 0.05), but did not impair leptin's ability to decrease expression of its own message in either site (Figs. 3A and 4A). In contrast, ICV leptin failed to decrease leptin mRNA in epididymal or retroperitoneal WAT from ␤ 3 -KO mice (Figs. 3B and 4B). Propranolol alone produced a slight increase in leptin mRNA in both depot sites of ␤ 3 -AR KO mice (p Ͻ 0.05), and treating these mice with propranolol and leptin produced no additional change in leptin mRNA (Figs. 3B and 4B). The failure of ICV leptin to down-regulate leptin mRNA in WAT from ␤ 3 -KO mice coupled with the inability of propranolol to block leptin's ability to down-regulate leptin mRNA in WT mice indicates that the ␤ 3 -adrenoreceptor is both necessary and sufficient for mediation of this response.
Additional data that support the importance of the ␤ 3 -AR in FIG. 1. Ribonuclease protection assay of UCP1 mRNA from BAT of C57BL/6J mice. 0.5 g of total RNA from BAT was hybridized with an antisense probe for UCP1 mRNA (nucleotides 7-300) and 18 S rRNA (nucleotides 715-794). Mice were injected ICV with either aCSF (5 l) or leptin (Lep, 2 g) for 1 day and sacrificed either 2 or 4 h following injection as described under "Experimental Procedures." The relative abundance of UCP1 mRNA was quantitated by comparing the densitometric intensities of protected fragments from each treatment group to known amounts of sense strand transcripts that were hybridized simultaneously. Individual RNA samples from each animal were analyzed to calculate group means, and representative samples are presented in the figure (aCSF, 1.71 Ϯ 0.04 fmol of UCP1 mRNA/g of RNA, n ϭ 3; leptin at 2 h, 2.32 Ϯ 0.16 fmol of UCP1 mRNA/g of RNA, n ϭ 6; leptin at 4 h, 2.94 Ϯ 0.29 fmol of UCP1 mRNA/g of RNA, n ϭ 6).

FIG. 2. Ribonuclease protection assay of UCP1 mRNA from BAT of FVB/NJ (A) and ␤ 3 -AR KO mice (B)
. 0.5 g of total RNA from BAT was hybridized with an antisense probe for UCP1 mRNA (nucleotides 7-300) and 18 S rRNA (nucleotides 715-794). Mice were injected ICV with either aCSF (5 l) or leptin (2 g), in the presence and absence of intraperitoneally injected propranolol (20 g/g body weight/day) for 3 days and sacrificed 2 h following the final injection as described under "Experimental Procedures." The relative abundance of UCP1 mRNA was quantitated by comparing the densitometric intensities of protected fragments from each treatment group to known amounts of sense strand transcripts that were hybridized simultaneously. Individual regulating leptin expression can be found by comparing leptin mRNA and serum leptin in WT and ␤ 3 -AR KO mice. If our hypothesis is correct, the absence of the ␤ 3 -AR should lead to up-regulation of leptin mRNA. The data from epididymal WAT support this conclusion by showing that leptin mRNA is ϳ3fold higher (p Ͻ 0.01) in the ␤ 3 -AR KOs compared with the WT mice (WT (0.021 Ϯ 0.003 fmol of leptin mRNA/g of RNA) versus ␤ 3 -AR KO (0.072 Ϯ 0.005 fmol/g of RNA)). The difference was even more striking in the retroperitoneal depot, where leptin mRNA was 5-fold higher in ␤ 3 -AR KO compared with WT mice (WT (0.012 Ϯ 0.004 fmol/g of RNA) versus ␤ 3 -AR KO (0.070 Ϯ 0.009 fmol/g of RNA)). Circulating leptin levels were also higher (p Ͻ 0.05) in ␤ 3 -AR KOs (8.7 Ϯ 0.4 ng/ml) compared with WT mice (5.6 Ϯ 0.4 ng/ml), although the magnitude of the difference was less than that noted with leptin mRNA. Taken together, the results support the conclusion that the absence of the ␤ 3 -AR leads to up-regulation of leptin expression.
To evaluate whether the observed changes in mRNA levels are reflected by comparable changes in ␤ 1 -AR and ␤ 2 -AR binding capacity, we used a competition radioligand binding approach with ICYP in the presence of the selective ␤ 3 -AR agonist, CL316,243, to eliminate low affinity binding of ICYP to ␤ 3 -ARs (37). Then, using crude membrane preparations from brown and white adipocytes of each phenotype, the highly selective ␤ 1 -AR antagonist, CGP-20712A, which displays ϳ1000-fold selectivity for ␤ 1 -AR over ␤ 2 -AR (38), was used to resolve ICYP binding into the components contributed by the ␤ 1 -and ␤ 2 -AR subtypes (Fig. 6, A and B). With ICYP at 30 pM (Fig. 6A), total ICYP bound by both ␤ 1 -and ␤ 2 -ARs in white adipocyte membranes was similar between WT FVB/NJ (21.3 Ϯ 0.5 fmol/mg of protein) and ␤ 3 -AR KO mice (19.7 Ϯ 0.3 fmol/mg of protein). The high affinity binding component of the curves, resolved by curve fitting and defined as the ␤ 1 -AR, accounted for 26% (5.6 Ϯ 0.6 fmol/mg) of the total binding sites in membranes from WT FVB/NJ mice. In ␤ 3 -AR KO mice, the high affinity component accounted for 19% (3.8 Ϯ 0.5 fmol/mg) of total ICYP binding (Fig. 6A). The second component of the curves represents ICYP bound by the ␤ 2 -AR and represents the remaining binding sites (WT FVB/NJ, 74%; ␤ 3 -AR KO, 81%). Although the total ICYP bound to white adipocyte membranes did not differ between the two groups, comparison of the respective binding curves using an F test indicated that the proportion of binding sites contributed by the ␤ 1 -and ␤ 2 -AR subtypes in each group was significantly different (p Ͻ 0.01). Thus, the modest decrease in ␤ 1 -AR binding sites and increase in ␤ 2 -AR binding sites noted in the ␤ 3 -AR KOs (Fig. 6A) are consistent with the observed changes in ␤ 1 -and ␤ 2 -AR mRNA levels in this group (Fig. 5) compared with the WT mice.
In the case of BAT, total ICYP binding did not differ between WT FVB/NJ (18.6 Ϯ 0.3 fmol/mg) and ␤ 3 -AR KO mice (17.5 Ϯ 0.4 fmol/mg, Fig. 6B). The high affinity ␤ 1 -AR component comprised 27% (5.0 Ϯ 0.6 fmol/mg) of total binding sites in WT FVB/NJ membranes and 28% (4.9 Ϯ 0.4 fmol/mg) of total binding sites in the ␤ 3 -AR KO group (Fig. 6B). The remaining ␤ 2 -AR binding sites comprised 73% and 72% of total ICYP binding in BAT membranes from WT and ␤ 3 -AR KO mice, respectively. In contrast to WAT, where subtle alterations were noted, we found no evidence of a change in the proportion of ␤ 1or ␤ 2 -AR binding sites in BAT between the groups (Fig. 6B). In contrast to WAT, where binding and mRNA results were consistent, the similarity in binding capacity was not reflective of the observed group differences in BAT ␤ 1 -and ␤ 2 -AR mRNA levels (Fig. 5).

Functionality of ␤ 1 -and ␤ 2 -ARs in Adipocyte Membranes of FVB/NJ and ␤ 3 -KO Mice-
The third component of studies in experiment 3 was to assess the functional coupling of ␤ 1 -and ␤ 2 -ARs to AC in adipocyte membranes from the two groups. The goal of these studies was to complement the binding studies by testing the hypothesis that expressed ␤-adrenoreceptor subtypes are equivalently coupled to AC in each group. Clarification of this point would rule out altered expression or effector coupling of ␤ 1 -or ␤ 2 -ARs as the basis for the differential requirements of receptor subtypes in regulating gene expression between BAT and WAT that was shown in experiment 2. To that end, combinations of nonselective and selective receptor agonists and antagonists were used in a systematic attempt to evaluate the relative contributions of the ␤ 1 /␤ 2 -ARs and ␤ 3 -ARs in activating AC in adipocyte membranes from each group. As shown in Table I, basal AC activity in white and brown adipocyte membranes did not differ between WT and ␤ 3 -AR KO mice. As judged by forskolin stimulation, total AC activity in BAT and WAT was also unaffected by the absence of the ␤ 3 -AR (Table I). The selective ␤ 3 -AR antagonist, SR59230A, and the selective ␤ 1 /␤ 2 -AR antagonist, propranolol, were used in combination with isoproterenol to show that the ␤ 1 /␤ 2 -ARs accounted for between 10% and 30% of the total AC activation elicited by isoproterenol in WAT and BAT membranes from WT mice, respectively (Table I). A similar assessment of receptor subtype contribution was reached with epinephrine, which was used at 1 M to fully activate the ␤ 1 -and ␤ 2 -ARs without significantly activating the ␤ 3 -AR (Table I). Using both strategies, AC activation attributable to the ␤ 1 /␤ 2 -ARs in membranes from WT mice was similar to the stimulation of AC activity in ␤ 3 -AR KO mice that is solely attributable to the ␤ 1 /␤ 2 -ARs (Table I). Taken together, these data establish the presence and functionality of ␤ 1 -and ␤ 2 -ARs on brown and white adipocytes in ␤ 3 -AR KO mice at levels that are comparable to those in WT mice.
Effect of Leptin on Leptin mRNA in ob/ob Mice-A practical test of our conclusion that the ␤ 3 -AR is required for leptin-dependent regulation of leptin mRNA in WAT can be made with the ob/ob mouse. This follows from our previous demonstration that expression of the ␤ 3 -AR is severely compromised in both brown and white adipocytes from ob/ob mice (35). Ob/ob mice are also particularly well suited for this purpose because of the absence of functional leptin and the resulting sensitivity to leptin replacement. Results from experiment 4 show that treat-ment of ob/ob mice for 3 days with leptin failed to reduce leptin mRNA in either retroperitoneal or epididymal WAT (Fig. 7). Similar treatment of lean mice produced a 6 -7-fold reduction in leptin mRNA in both sites ( Fig. 7 and Table II). The reduction in ␤ 3 -AR mRNA in WAT from the ob/ob mice used here was on the order of 50 -100-fold (Table II). These findings are consistent with and support our conclusion that the ␤ 3 -AR is required for leptin-dependent regulation of leptin mRNA in WAT. To test whether the requirement for the ␤ 3 -AR is specific to WAT or specific to the gene being regulated, we examined the effect of leptin in BAT from the same mice. Expression of the ␤ 3 -AR is also compromised in BAT from ob/ob mice (Table  II), so if the effect is dependent on the ␤ 3 -AR, regardless of site, we would predict that leptin mRNA would not be reduced by leptin in BAT. Quite the contrary, BAT from the same ob/ob mice showed a significant reduction in leptin mRNA following leptin treatment (vehicle, 0.020 Ϯ 0.001 fmol/g of RNA; leptin, 0.007 Ϯ 0.001 fmol/g of RNA). Collectively, these results indicate that regulation of leptin expression requires the ␤ 3 -AR in WAT but is capable of using the ␤ 1 /␤ 2 -AR in BAT.
Effect of Cold Exposure on Adipocyte Gene Expression in ␤ 3 -AR KO Mice-A more general test of our hypothesis that the sympathetic nervous system utilizes different complements of ␤-adrenoreceptors to mediate the effects of norepinephrine in white versus brown adipose tissue can be made by exposing ␤ 3 -AR KO mice to cold. In experiment 5, the ␤ 3 -AR KOs adapted to cold exposure readily and increased UCP1 mRNA to the same extent as their WT controls (data not shown). This finding agrees with observations made by the original developers of this transgenic mouse line (28), but of particular interest is our finding that cold exposure does not reduce leptin mRNA in WAT of ␤ 3 -AR KO mice (room temperature, 0.056 Ϯ 0.009 fmol/g RNA; 4°C, 0.094 Ϯ 0.015 fmol/g RNA). In contrast, 4 h of cold exposure in WT mice reduced leptin mRNA to the detection limit of the assay (data not shown). Taken together, these studies show that the ␤ 3 -AR is required to transduce the effects of changes in sympathetic outflow on leptin expression in WAT. As noted with ICV leptin, the ␤ 1 /␤ 2 -AR is capable of substituting for the ␤ 3 -AR in mediating sympathetic effects of leptin on gene expression in BAT. DISCUSSION In the present study, we have used the ICV route of administration to study centrally-mediated effects of leptin on gene TABLE I Relative contributions of ␤-adrenoceptor subtypes to activation of adenylyl cyclase in brown and white adipocyte membranes from control and ␤ 3 -adrenoceptor knockout mice BAT and WAT were isolated from control and ␤ 3 -adrenoceptor knockout mice as described under "Experimental Procedures." Crude adipocyte membranes were obtained from each genotype and tissue type and assayed for adenylyl cyclase activity under conditions designed to assess the relative contribution of each receptor subtype to cyclase activation. The means and their standard errors were from four replicate assays on two separate membrane preparations. The ␤ 3 -adrenoceptor antagonist SR59230A (36) was used at 10 M to selectively block ␤ 3 -adrenoceptor activation. Propranolol was used at 100 nM to selectively block ␤ 1 -and ␤ 2 -adrenoceptor activation, while isoproterenol was used at 100 M to fully activate ␤ 1 -, ␤ 2 -, and ␤ 3 -adrenoceptors. Epinephrine was used at 1 M to fully activate the ␤ 1 -and ␤ 2 -adrenoceptor subtypes without activating the ␤ 3 -adrenoceptor. Forskolin was used at 100 M to assess the total adenylyl cyclase catalytic activity in adipocyte membranes from each tissue type and genotype. expression in adipose tissue. The technique was used with mice lacking ␤ 3 -adrenoreceptors (␤ 3 -AR KO) to evaluate the involvement of this and other adrenoreceptor subtypes in mediating peripheral effects of leptin. This experimental approach is based on an emerging consensus that leptin regulates adipocyte function through central modulation of the sympathetic nervous system. The initial evidence came from a study showing that leptin increased norepinephrine turnover in BAT (12), and was followed by studies mapping the neural pathways within the sympathetic nervous system activated by leptin (13)(14)(15). Subsequent studies have shown that surgical (16), chemical (17,39), or genetic (18) sympathectomy blocked leptin's effects on gene expression in both BAT and WAT. The latter studies established a requirement for the long form of the leptin receptor (Ob-Rb), but did not distinguish between central versus peripheral sites of action for leptin. The requirement of norepinephrine (18) suggested that central leptin receptors were activated, but the involvement of peripheral leptin receptors that communicate with the brain by afferent autonomic nerves (40,41) could not be excluded. Results from the present study argue that peripheral leptin receptors are not required for leptin's effects on adipocyte gene expression, and make the case that central leptin receptors mediate the observed effects on UCP1 mRNA in BAT and leptin mRNA in WAT. Additionally, based on the comparable magnitude of the effects produced by central versus peripheral administration of leptin in our studies, it seems reasonable to assume that peripherally injected leptin is primarily working through central leptin receptors. Taken together, the evidence is compelling that leptin regulates adipocyte gene expression through central modulation of sympathetic nervous system outflow. The premise that norepinephrine is the peripheral mediator of leptin action is the basis for using the present experimental approach to deduce which adrenoreceptor(s) mediate the respective responses in brown and white adipose tissue.
Several recent studies have suggested the existence of a putative ␤ 4 -AR based on interpretation of pharmacological profiles from adipocytes expressing the ␤ 3 -AR at various levels (42)(43)(44). The existence of a fourth ␤-adrenoreceptor could complicate interpretation of the present experiments, which assume the existence of only three ␤-adrenoreceptors. The evidence for a ␤ 4 -AR centers around the observation that CGP12177A, a partial agonist of the ␤ 3 -AR and antagonist of ␤ 1 /␤ 2 -ARs, activates lipolysis in adipocytes containing little or no ␤ 3 -AR (42,43). Complementing this result is the observation that ␤ 1 -AR-selective or nonselective ␤-AR antagonists block this response while the ␤ 3 -AR-selective antagonist, SR59230, does not (43). It should be noted, however, that the documented heterogeneity in the relationship between binding affinity and coupling efficiency with ligands for ␤-adrenoreceptor subtypes complicates interpretation of pharmacological approaches in adipoyctes (45)(46)(47)(48)(49). Such seems possible with CGP12177 after the careful studies of Konkar et al. (50) have systematically documented an unexpected activation of the ␤ 1 -AR by this ligand. Thus it is unnecessary to invoke the existence of a ␤ 4 -AR to explain the results obtained with CGP12177 if its binding to the ␤ 1 -AR results in partial activation. In the present study, conclusions regarding requirements for specific ␤-receptors in WAT and BAT are not predicated on the existence of the putative ␤ 4 -AR because of the adequacy of a model containing ␤ 1 -, ␤ 2 -, and ␤ 3 -ARs in explaining the observed results.
The most significant finding from the present study is the differential requirement for the ␤ 3 -AR in mediating leptin's effects in brown versus white adipose tissue. The concept of differential coupling of ␤-adrenoreceptor subtypes to various responses within the same cell is not new. The subject has received significant attention in BAT, where densely innervated brown adipocytes are stimulated by norepinephrine in response to environmental and physiological stimuli (12,51). Using isolated brown adipocytes from both Syrian hamsters (52) and rats (24,53), Zhao et al. addressed the role of ␤ 3 -ARs versus ␤ 1 -ARs in mediating catecholamine-stimulated oxygen consumption. Based on analysis of Schild plots constructed from propranolol-mediated inhibition of norepinephrine doseresponse curves, the authors concluded that the thermogenic response was coupled solely to the ␤ 3 -AR (24,53). This conclusion is at odds with results from experiments with ␤ 3 -AR knockouts where thermogenic responses to norepinephrine, isoproterenol, and cold exposure were normal (28). Thus, the absence of the ␤ 3 -AR did not compromise the ability of BAT to respond to catecholamines. A role for the ␤ 1 -AR is also supported by the work of Atgie et al. (25), who used isolated brown adipocytes to show that a specific ␤ 1 -AR antagonist effectively blocked the effects of low concentrations of norepinephrine (25-100 nM) on respiration. The authors concluded that, within the physiological range of norepinephrine concentrations, the ␤ 1 -AR makes a significant contribution to activating respiration. Chaudhry and Granneman (26) investigated the possibility that ␤ 1 -ARs and ␤ 3 -ARs serve different signaling functions in brown adipocytes. Supported by data showing the predicted compartmentalization of phosphorylated proteins (cAMP-responsive element-binding protein and perilipin), the authors concluded that norepinephrine induced UCP1 expression preferentially through the ␤ 1 -AR and lipolysis primarily through the ␤ 3 -AR (26). Our results extend these findings to the intact animal and show for the first time that centrally mediated effects of leptin on UCP1 expression in brown adipose tissue can be transduced interchangeably by ␤ 1 /␤ 2 -or ␤ 3 -ARs. Our results demonstrate that this conclusion is not an artifact of altered expression or function of the ␤ 1 /␤ 2 -ARs in BAT from ␤ 3 -AR KO mice. The present findings are also consistent with the original report with these mice showing comparable coldinduced increases in UCP1 mRNA in BAT from wild type and ␤ 3 -AR KO mice (28). A study by Revelli et al. (54) using C57BL/6J mice with the ␤ 3 -AR gene deleted by homologous recombination found a positive correlation between UCP1 and ␤ 1 -AR mRNA, indicating a role for the ␤ 1 -AR in the absence of the ␤ 3 -AR. It should also be noted that, despite significant reductions in ␤ 3 -AR expression in BAT from ob/ob mice (35), FIG. 7. Ribonuclease protection assay of leptin mRNA from epididymal and retroperitoneal WAT of lean and ob/ob mice. 2.0 g of total RNA from epididymal WAT (EWAT) and retroperitoneal WAT (RPWAT) of lean and ob/ob mice was hybridized with an antisense probe for leptin mRNA (nucleotides 1-355) and 18 S rRNA (nucleotides 715-794). The relative abundance of leptin mRNA was quantitated by comparing the densitometric intensities of protected fragments from each treatment group to known amounts of sense strand transcripts that were hybridized simultaneously. Individual RNA samples from each animal were analyzed to calculate group means presented in Table II. they respond to leptin repletion with robust increases in BAT UCP1 expression (19). Taken together, the evidence is internally consistent and supports our conclusion of interchangeability of receptor subtypes in this tissue.
In contrast to BAT, where the ␤ 3 -AR is sufficient but not necessary to mediate leptin effects on gene expression, the present work makes the case that the ␤ 3 -AR is required for inhibitory regulation of leptin mRNA in WAT. Four lines of evidence support this conclusion. First, norepinephrine is required for inhibitory regulation of leptin mRNA by leptin (18). Second, centrally administered leptin fails to lower leptin in WAT from ␤ 3 -AR KO mice while effectively producing the response in wild type mice even when ␤ 1 /␤ 2 -ARs are blocked. Third, cold exposure produces intense sympathetic outflow that results in the complete disappearance of leptin mRNA from WAT of wild type mice, but fails to lower leptin mRNA in WAT from ␤ 3 -AR KO mice. Fourth, the relative absence of the ␤ 3 -AR in WAT from ob/ob mice (35) compromises the ability of exogenous leptin or CL316,243 to lower leptin mRNA in WAT from these mice. Our previous work made the case that the ␤ 3 -AR is also required for norepinephrine-mediated inhibition of leptin release from the adipocyte (55). Together, these findings support the conclusion that the ␤ 3 -AR plays a central role in regulating both the expression and release of leptin from the adipocyte. In the sense that adipose tissue is a primary target for leptin's effects, it can be inferred that the ␤ 3 -AR also plays a pivotal role in modulating the biological effects of leptin.
An important question arising from the present work is whether the requirement for the ␤ 3 -AR is specific to WAT or specific to the gene being regulated. One way to address this question would be to determine whether leptin is capable of inhibiting leptin mRNA in BAT or inducing UCP1 expression in WAT. We have examined this question in ob/ob mice, which have deficits in ␤ 3 -AR expression in both BAT and WAT (35), and found that exogenous leptin decreases leptin mRNA in BAT (present study) and increases UCP1 expression in retroperitoneal WAT (4244}. We have obtained comparable results in ␤ 3 -AR KO mice, where cold exposure decreases leptin mRNA in BAT but not in WAT. Similarly, Deng et al. (56) reported that stimulation of each of the coexisting ␤-ARs in mouse brown adipocytes decreased leptin mRNA. Evans et al. (39) also found that cold exposure decreased leptin mRNA in BAT, and this effect was only prevented by complete ␤-AR blockade (propranolol ϩ SR 59230A). Collectively, these results support the conclusion that the requirement for the ␤ 3 -AR resides with the tissue (WAT) and not with the gene (leptin). Our findings also suggest that cAMP generated by the ␤ 3 -AR in WAT may be targeted to a compartment that is necessary for inhibitory regulation of leptin expression. If so, this would be consistent with the observations of Chaudhry and Granneman (26), who reported differential targeting of phosphorylated proteins to subcellular compartments produced by ␤ 1 -versus ␤ 3 -AR activation in brown adipocytes (26). It will be interesting to determine whether the morphological differences between brown and white adipocytes are responsible for the differential coupling of adrenoreceptors to gene expression in the two cell types, and whether the differences described here extend to other genes regulated by leptin in the two types of adipocytes. II ␤ 3 -Adrenoceptor and leptin mRNA from lean and ob/ob mice treated with leptin Lean and ob/ob mice received intraperitoneal injections of vehicle or murine leptin (20 g/g body weight/day) on each of three successive days as described under "Experimental Procedures." 2.0 g and 10.0 g of total RNA from WAT and BAT, respectively, was hybridized with an anti-sense probe for leptin mRNA (nucleotides 1-355) or ␤ 3 -AR mRNA (nucleotides 630 -890). A probe for 18 S rRNA (nucleotides 715-794) was included to correct for differences in total RNA loaded. The relative abundance of mRNA was quantitated by comparing the densitometric intensities of protected fragments from each treatment group to known amounts of sense strand transcripts that were hybridized simultaneously. Individual RNA samples from each animal were analyzed to calculate group means, which are presented in the table. The data were subjected to analysis of variance. Mean values are expressed as femtomoles of protected mRNA/g of total RNA Ϯ S.E. EWAT, epididymal WAT; RPAT, retroperitoneal WAT; IBAT, interscapular BAT.