Regulation of expression of ob mRNA and protein by glucocorticoids and cAMP.

Regulation of obese gene (ob) expression in ob/ob and db/db mice and in cultured rat adipocytes was examined. It has been demonstrated that exogenous human OB protein (leptin) treatment reduces food intake and weight gain, as well as insulin, glucose, and corticosterone levels in ob/ob mice. In the present report we show that leptin treatment down-regulates endogenous adipose ob mRNA. However, treatment of isolated rat adipocytes with 100 ng/ml human or murine leptin had no direct effect on expression of endogenous ob mRNA, suggesting that leptin may be able to down-regulate its own expression by an indirect, non-autocrine mechanism. Glucocorticoids increased both ob mRNA levels and secreted leptin levels in vitro. Conversely, agents that increase intracellular cAMP, such as β-adrenergic agonists or BtcAMP itself, decreased ob mRNA expression and leptin secretion. Therefore, increased glucocorticoid levels and decreased sympathetic neural activity may contribute to the elevated ob mRNA expression observed in genetically obese, hyperglucocorticoid rodents. Furthermore, leptin might regulate its own expression through a feedback mechanism involving the hypothalamic pituitary axis.

Obesity is a major risk factor for a number of human diseases including cardiovascular disease, hypertension, and non-insulin-dependent diabetes. Although obesity in humans is apparently a polygenic disorder, numerous rodent models of obesity exist as single gene mutations. Several of these which have recently been identified include the agouti gene (1,2), the fat gene (3), and the ob (obese) gene (4). The ob locus on chromosome 6 involves two mutations in the gene for a secretory protein, which result in either premature termination (nonsense mutation at Arg-105) in the original C57BL/6J ob/ob strain or complete absence of message in the SM/Ckc-ϩ Dac ob 2J / ob 2J strain (4). The human homolog has been cloned (4,5), and there appears to be no evidence of obesity-associated mutations in human OB comparable with the ob mouse (5) nor any linkage between OB mutations and susceptibility to non-insulin-dependent diabetes (6). The ob mutation was originally thought to involve a diminished satiety signal or factor based on parabiosis experiments in both ob/ob and db/db mice (7), which suggested that ob/ob mice did not express the satiety signal while db/db mice were defective in the signal transduction pathway. Recent experiments have supported this by demonstrating that exogenous recombinant mouse Ob protein and human OB protein decrease food intake and weight gain in the ob/ob, but not in the db/db, mouse (8 -11). It has recently been suggested that the ob gene product be referred to as leptin (9).
Despite the evidence that leptin acts as a satiety factor, human obesity (5) and other forms of rodent obesity that involve different genetic loci, such as the A vy mouse (12), or non-genetic lesions in the ventramedial hypothalamus (13,14) are paradoxically associated with elevated levels of leptin expression. For this reason, we investigated the regulation of ob expression in ob/ob and db/db mice, and in 24 h cultures of mature rat adipocytes to determine what obesity associated factor(s) might regulate leptin expression.

MATERIALS AND METHODS
Animal Treatment-Five-to six-month-old obese (C57Bk/6J-ob/ob and C57Bk/Ks-db/db) mice were obtained from Jackson Laboratories, Bar Harbor, ME or Harlan, U.K. Mice were housed three to six per cage; water and Purina Formulab Chow 5008 (Purina Mills, St. Louis, MO) were available ad libitum. Animals were maintained on a normal 12 h light/12 h dark cycle. Recombinant human leptin or saline was administered subcutaneously at 300, 200, and 100 g/mouse/day for 9, 2, and 19 days, respectively, 1 h before the dark cycle as described previously (11). Blood samples for glucose, insulin, and corticosterone determinations were obtained between 8:00 and 10:00 a.m. at the end of the experimental period. Epididymal fat pads were processed as described below for Northern blot. Sprague-Dawley rats were obtained from Harlan Sprague-Dawley (Indianapolis, IN). A full presentation of food consumption and body weight change data, as well as glucose, insulin, triglyceride, and glucocorticoid measurements, has been published elsewhere (11).
Isolation and Culture of Mature Rat Adipocytes-Adipocytes were obtained from rat epididymal fat pads from 250 -300-g male rats by collagenase digestion (15). The cell suspension was filtered sequentially through 500-, 250-, and 100-m mesh and washed 6 times with Dulbecco's modified Eagle's medium/Ham's F12 (3:1, Life Technologies, Inc.) supplemented with 20 g/ml bovine serum albumin (RIA 1 grade, Sigma), 20 mM Hepes, 0.1 g/liter sodium selenite, and 4.88 mg/liter ethanolamine. Adipocytes were cultured in this medium for 24 h at approximately 5 ϫ 10 5 cells/ml in either Costar P6 trays or Corning T75 flasks. Medium was removed from beneath the adipocyte layer and stored at Ϫ20°C until used for RIA and Western blot analysis. Adipocytes were used for RNA isolation as described below. Dibutyryl cAMP, isoproterenol, dexamethasone, and hydrocortisone were obtained from Sigma. Recombinant human insulin was from Lilly.
Western Blot-10 ml of conditioned medium from 24-h cultures was treated with 100 l of 20 mg/ml deoxycholate and 1.1 ml of 20% trichloroacetic acid (16). After centrifugation, the pellets were dissolved in 200 l of Laemmli sample buffer, and 40-l samples were run on 18% polyacrylamide gel electrophoresis gels (Novex, San Diego, CA). Proteins were transferred to polyvinylidene difluoride membranes and blotted with affinity-purified antibody (1 g/ml) that had been prepared in rabbits immunized by full-length recombinant murine leptin (11). Antisera were affinity-purified on agarose columns containing immobilized murine leptin (Aminolink immobilization kit, Pierce). Western blots were detected by the Supersignal chemiluminescent system (Pierce).
Northern Blot-Total RNA was isolated from rat adipocytes or from * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Endocrine Research Division, Drop Code 0540, Lilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN 46285. mouse epididymal fat pads by the TRI REAGENT method (Molecular Research Center, Cincinnati, OH). Fat pads were removed, freeze clamped in liquid nitrogen, and stored at Ϫ80°C until RNA isolation. Ten to 15 g of total RNA were separated on 1.2% agarose gels and transferred to 0.2 m maximum strength Nytran membranes using the Turboblotter rapid downward transfer system (Schleicher & Schuell). Murine ob cDNA was prepared as described (11). 32 P-Labeled fulllength murine ob, human glyceraldehyde 3-phosphate dehydrogenase, and human ␤-actin cDNA (Clontech, Palo Alto, CA) hybridization probes were prepared by the random priming DNA labeling system (Life Technologies, Inc.), and hybridization was performed according to Sambrook et al. (17). Hybridization signals were visualized and quantitated after 3 h with a PhosphorImager SI (Molecular Dynamics, Sunnyvale, CA). 18 S RNA was quantitated by staining with ethidium bromide (0.05 g/l sample) and visualizing with a FluorImager 575 (Molecular Dynamics, Sunnyvale, CA).
Radioimmunoassay-Recombinant murine leptin was iodinated with 125 I-Bolton-Hunter reagent (Amersham Life Sciences, Arlington Heights, IL) to a specific activity of 25-40 Ci/g. Antibody was prepared in rabbits using recombinant murine leptin and Freund's complete adjuvant. RIA was performed in phosphate-buffered saline containing 1 mg/ml bovine serum albumin and 0.1% Triton X-100 (RIA buffer). Using a non-equilibrium protocol to enhance sensitivity, 100 l of conditioned medium was combined with 100 l of a 1:8000 dilution of anti-leptin antibody in RIA buffer. Standards were prepared in RIA buffer and ranged from 0.25 to 500 ng/ml. After incubation at room temperature for 4 h, approximately 15,000 -20,000 cpm of 125 I-murine leptin were added in 100 l of RIA buffer, and incubation continued overnight at 4°C. To each tube were then added 100 l of a 1:6 dilution of sheep anti-rabbit IgG (Antibodies Inc., Davis, CA), 100 l of a 1:50 dilution of normal rabbit serum (Life Technologies, Inc.), and 100 l of 10% polyethylene glycol (PEG 8000), all in RIA buffer. After a 15-min incubation at room temperature, tubes were centrifuged and decanted. Pellets were counted for 125 I and data analyzed by RIA AID™ (ICN Micromedic Systems, Huntsville, AL). In order to verify cross-reactivity of the anti-mouse leptin antibody with rat leptin, serum from obese Zucker fa/fa rats (100, 50, 25, 12.5, and 6.25 l) was diluted into varying amounts of horse serum such that the total serum volume was 100 l. Horse serum itself showed no detectable immunoreactivity with the anti-mouse antibody. The dilution curve generated from these samples was parallel over the entire range to a comparator standard curve generated from mouse leptin standards diluted in 100 l of horse serum.
In the present study, Northern analysis of total RNA from epididymal fat pads from each group of mice showed that ob mRNA levels were reduced 65% in treated ob/ob mice (ob mRNA/18 S RNA: control, 1.23 Ϯ 0.14; treated, 0.44 Ϯ 0.06, p Ͻ 0.05) but were unchanged in db/db mice (Fig. 1, A and B). Although many factors secondary to weight loss or reduced food consumption may have contributed, it is possible that the reduction in glucocorticoids and insulin or the administration of exogenous leptin itself may have acted to down-regulate endogenous ob mRNA expression. The observation that leptin downregulates its own expression in ob/ob mice, but not in db/db mice that have a presumed central defect in its signal transduction, would suggest that leptin does not act directly at the adipocyte but instead through a centrally regulated mediator.
However, these data do not rule out the possibility that a putative adipocyte leptin receptor is also defective in db/db mice.
In order to examine direct (autocrine regulation at the adipocyte) versus indirect effects of leptin on ob mRNA regulation, we treated rat adipocytes in vitro with either 25 nM dexamethasone or 100 ng/ml human or mouse leptin for 24 h. Northern analysis (Fig. 1C) shows that leptin treatment did not alter endogenous ob mRNA levels but that 25 nM dexamethasone up-regulated ob expression dramatically. Therefore, the ability of exogenous human leptin to down-regulate endogenous expression of ob mRNA in the ob/ob mouse argues for an indirect mechanism, possibly secondary to reduced glucocorticoid levels. It should be noted, however, that our measurements are of steady-state mRNA levels, and in the absence of nuclear run-on experiments we cannot distinguish transcriptional regulation from effects on mRNA stability. The recent report that treatment of rats with dexamethasone and hydrocortisone increased adipocyte ob mRNA expression levels is consistent with these in vitro results (18).
The development of a sensitive RIA allowed us to examine a number of agents to investigate their role in regulating expression of leptin. The limit of detection for this assay is approximately 0.5-1 ng/ml. Cross-reactivity between rat and mouse leptin was confirmed by showing that rat serum diluted in parallel with the mouse leptin standard (data not shown). However, the absolute level of cross-reactivity has not been determined quantitatively with recombinant rat leptin standard. The anti-mouse leptin antibody also cross-reacted with secreted rat leptin in a Western blot. These results are reasonable based on the high degree of conservation of the protein sequence between rat and mouse leptin (96% at the protein level (13)) and suggest that this assay is suitable for measuring relative, if not necessarily absolute, changes in rat leptin levels.
Because of our initial in vitro observation that dexamethasone up-regulated ob mRNA expression, we examined a number of steroids, including glucocorticoids, mineralocorticoids, estrogens, and androgens, for their ability to regulate leptin expression and secretion (data not shown). Preliminary experiments indicated that glucocorticoids such as dexamethasone, hydrocortisone, and corticosterone had a pronounced 3-4-fold stimulatory effect on leptin expression (as measured by RIA). Testosterone and aldosterone were relatively ineffective (data not shown), while 17␤-estradiol increased leptin release approximately 2-fold. Fig. 2 shows a dose response for stimulation of leptin secretion by dexamethasone, hydrocortisone, and 17␤estradiol. Both glucocorticoids reach the same maximal response (3-3.5-fold increase over basal) with the synthetic glucocorticoid being more potent (EC 50 : dexamethasone, 3.3 nM; hydrocortisone, 24.7 nM). These values are consistent with other gluccorticoid-induced functions in adipocytes, such as inhibition of preadipocyte proliferation (19). Dexamethasone and hydrocortisone increased ob mRNA expression (Fig. 3A) as well as leptin secretion (Fig. 3, B and C).
Insulin treatment increased leptin secretion only 20 -25% as measured by RIA (Fig. 3). Effects of insulin on ob mRNA expression have been variable, probably reflecting the difficulty in measuring small changes in RNA by Northern analysis. Therefore, the effect of insulin under these conditions appears modest. Incubation of adipocytes with 25 mM glucose had no effect on ob mRNA expression (data not shown), suggesting that ob expression may not be regulated directly by acute changes in glucose availability. A recent report by Saladin et al. (20) indicated that a single insulin injection or a hyperinsulinemic clamp of rats increased ob mRNA expression severalfold, independent of glucose concentration. In vitro treatment of adipocytes with insulin for 24 h also doubled ob mRNA. However, Murakami et al. (21) reported that in vitro insulin treatment of adipocytes increased ob mRNA only 10%. Insulin clearly increases leptin expression, but the degree would appear to be less than that of glucocorticoids.
Increases in intracellular cAMP result in decreased expression of ob mRNA and leptin secretion in rat adipocytes. This is demonstrated in Fig. 3, where it is shown that dibutyryl cAMP, the non-selective ␤-agonist isoproterenol, and the selective ␤ 3 agonist ICI 201,651 (22) all reduce ob mRNA and protein expression. This would suggest that when lipolysis is stimulated in adipocytes, leptin expression is reduced. This hypothesis is consistent with decreased expression of a satiety signal under conditions of starvation. A recent report has demonstrated that fasting of lean Ob/? mice (but not ob/ob mice) is associated with a reduction of ob mRNA levels, which is reversed upon refeeding (23). Pertussis toxin (50 ng/ml) and adenosine deaminase treatment (0.5 unit/ml) reduced leptin secretion (as measured by RIA) 60 and 20%, respectively, suggesting a role for G i in regulating leptin secretion. The effect of adenosine deami-nase was mimicked by the adenosine antagonist 8-phenyltheophylline (2 M) and was blocked by the adenosine agonist phenylisopropyl adenosine (10 nM) (data not shown).
Interestingly, ICI 201,651 is less effective than isoproterenol at lowering ob expression, even at the relatively high concentration of 10 M. This is similar to the effect of another ␤ 3 selective agonist BRL 37344 on cAMP production in rat adipocytes and may represent a weaker coupling of the ␤ 3 receptor to cAMP production (24). Since ␤ 3 agonists induce weight loss through increased thermogenesis in rats (25), it is interesting that they also down-regulate expression of leptin, which similarly increases the basal metabolic rate in rats (10).
We have previously demonstrated that leptin treatment of ob/ob mice reduces synthesis and release of hypothalamic NPY, and that this may represent its mechanism of action on food intake and basal metabolic rate (11). It has been suggested that glucocorticoids and insulin regulate energy balance by their central regulation of food intake through NPY and through their peripheral effects on energy storage (26). Centrally, glucocorticoids increase NPY levels and food consumption (26), while insulin inhibits NPY release (27). The data presented here suggest that leptin may represent another arm of this regulatory pathway. Increased glucocorticoids directly stimulate leptin secretion from adipose tissue, which then negatively modulates NPY release in the hypothalamus (11). Increased NPY levels would also lead to decreased sympathetic activity (28) which, based on our results, would lead to further increases in leptin secretion. This would explain the observed increased leptin expression in genetic models of rodent obesity, such as the Zucker fa/fa rat (29), since these animals are characterized by hyperglucocorticoidism, elevated hypothalamic NPY, and reduced sympathetic activity (30). Increased NPY is also associated with increased insulin secretion, which could further contribute to elevated leptin levels.
In summary, we have demonstrated that treatment of ob/ob mice with exogenous human leptin down-regulates endogenous murine ob mRNA expression, while similar treatment of obresistant db/db mice has no effect. Treatment of isolated rat adipocytes with leptin in vitro has no effect on endogenous ob mRNA levels, while glucocorticoids up-regulate expression of ob mRNA and protein secretion. Agents that increase intracellular cAMP down-regulate ob expression. These results suggest a centrally mediated mechanism of feedback regulation of ob expression involving the hypothalamic-pituitary-adrenal axis.