A Transgenic Model of Visceral Obesity and the Metabolic Syndrome

See allHide authors and affiliations

Science  07 Dec 2001:
Vol. 294, Issue 5549, pp. 2166-2170
DOI: 10.1126/science.1066285


The adverse metabolic consequences of obesity are best predicted by the quantity of visceral fat. Excess glucocorticoids produce visceral obesity and diabetes, but circulating glucocorticoid levels are normal in typical obesity. Glucocorticoids can be produced locally from inactive 11-keto forms through the enzyme 11β hydroxysteroid dehydrogenase type 1 (11β HSD-1). We created transgenic mice overexpressing 11β HSD-1 selectively in adipose tissue to an extent similar to that found in adipose tissue from obese humans. These mice had increased adipose levels of corticosterone and developed visceral obesity that was exaggerated by a high-fat diet. The mice also exhibited pronounced insulin-resistant diabetes, hyperlipidemia, and, surprisingly, hyperphagia despite hyperleptinemia. Increased adipocyte 11β HSD-1 activity may be a common molecular etiology for visceral obesity and the metabolic syndrome.

Obesity is associated with adverse metabolic consequences such as diabetes and dyslipidemia (1). The best predictor of these morbidities is not the total body adipose mass, but the specific quantity of visceral fat (2, 3). A molecular basis for disproportionate accumulation of visceral fat has not been identified and the extent to which visceral adiposity causes or merely reflects the associated metabolic syndrome, which includes insulin resistance, glucose intolerance, and dyslipidemia, remains unclear (3).

One identified cause of visceral obesity and metabolic complications is exposure to excessive levels of glucocorticoids. Although systemic glucocorticoid excess in rare Cushing's syndrome causes visceral obesity and the metabolic syndrome (4, 5), circulating glucocorticoid levels are normal in patients with the prevalent forms of obesity (5). However, glucocorticoid action on target tissues depends on both circulating hormone levels and intracellular prereceptor metabolism (6). The enzyme 11β hydroxysteroid dehydrogenase type 1 (11β HSD-1) plays a pivotal role in determining intracellular glucocorticoid concentrations by regenerating active glucocorticoid (cortisol in humans, corticosterone in rats and mice) from inactive cortisone and 11-dehydrocorticosterone (6–8) and has been suggested to serve as a tissue-specific amplifier of glucocorticoid action (6).

Mice homozygous for a targeted deletion of the 11β HSD-1 gene are viable and developmentally normal but cannot regenerate active corticosterone from inert 11-dehydrocorticosterone in vivo, demonstrating that 11β HSD-1 is the sole 11β-reductase in the body (9). The 11β HSD-1–deficient mice show attenuated activation of glucocorticoid-sensitive hepatic gluconeogenic enzymes in response to stress or high-fat diets and have a diabetes-resistant phenotype (9). Recently, adipose tissue from obese humans has been shown to have increased 11β HSD-1 activity (7). To test the hypothesis that increased production of glucocorticoid exclusively within adipose tissue would produce visceral obesity and features of the metabolic syndrome, we created transgenic mice overexpressing 11β HSD-1 under the control of the enhancer-promoter region of the adipocyte fatty acid binding protein (aP2) gene (Web fig. 1) (10–12).

All mice studied were inbred FVB strains crosses, wild type or hemizygous for the transgene. Ribonuclease (RNase) protection assays with a rat-based cRNA probe can differentiate the transgene-derived (rat) mRNA from endogenous murine 11β HSD-1 mRNA. The transgene-derived transcript was expressed equivalently in adipose tissue from subcutaneous abdominal, epididymal, mesenteric, and interscapular brown adipose tissue (BAT) depots but was absent in brain, liver, skeletal muscle, and kidney of transgenic (Tg) mice. In line 7 male mice, in which all studies were performed unless otherwise noted, transgene-derived mRNA was increased sevenfold compared with endogenous mRNA (Web fig. 2). 11β HSD-1 activity (6, 9, 13) was 2.4-fold increased in subcutaneous abdominal fat (P < 0.03) and 2.7-fold increased in epididymal fat (P < 0.01), respectively (Fig. 1A). In adipose tissue, the fold increase in aP2/11β HSD-1 Tg mice was comparable to that seen in leptin-deficient obese ob/ob mice and in obese humans (7), demonstrating that the extent of transgenic amplification of 11β HSD-1 activity in aP2/11β HSD-1 mice is physiologically relevant.

Figure 1

Creation of transgenic mice overexpressing 11β HSD-1 in adipose tissue. The Institutional Animal Care and Use Committee (Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA) approved all studies. (A) The 11β HSD-1 activity in adipose tissue from transgenic mice (Tg) (male, 18 weeks of age, n = 6), nontransgenic littermates (non-Tg, n = 6), and age-matched male ob/ob mice (n = 3). Values were expressed as mean ± SEM. SubQ, subcutaneous abdominal fat. In adipose tissue homogenates, the reaction catalyzed by 11β HSD-1 is bidirectional, consisting of a reductase activity and a dehydrogenase activity (6). The rate is most easily measured as B ([3H]-corticosterone) to A ([3H]-11-dehydrocorticosterone) conversion. As 11β-reductase activity predominates in intact cells and tissues (6, 9), enzyme activity was assessed by the rate of B to A conversion (6, 9,13). *, P < 0.05 compared with non-Tg mice in each fat tissue. (B) Adipose tissue corticosterone (CORT) concentration [14-week-old male non-Tg (n = 9) and Tg (n = 8) mice] was determined by methanol extraction followed by radioimmunoassay (Rat CORT RIA kit, ICN Pharmaceuticals) (15, 16). Mes, mesenteric fat; Epi, epididymal fat. Values were expressed as mean ± SEM; *, P < 0.01 compared with non-Tg mice. (C) Body weight curves for male mice on low- or high-fat diet. When mice were 6 weeks of age, non-Tg and Tg mice were divided into two separately housed groups. One group was given a high-fat diet containing 45% fat (D12451, Research Diets, New Brunswick, NJ), and the second group was given a low-fat diet (D12450, Research Diets) containing 10% fat. Body weight and food intake were measured weekly. ♦, non-Tg mice fed a low-fat diet; ▴, Tg mice fed a low-fat diet; ▪, non-Tg mice fed a high-fat diet; •, Tg mice fed a high-fat diet (n = 7 in each group). Values were expressed as mean ± SEM. * (Tg mice fed a high-fat diet), P < 0.001 compared with non-Tg mice fed a high-fat diet; † (Tg mice fed a low-fat diet),P < 0.001 compared with non-Tg mice fed a low-fat diet. (D) Gross appearance of 18-week-old male non-Tg and Tg mice. Ventral and dorsal views show loss of constriction of waist and larger abdomen in Tg mice. (E) Comparison of fat depots weight between non-Tg and Tg mice. Weight of unilateral epididymal, subcutaneous abdominal, and mesenteric fat depots in 16-week-old male non-Tg and Tg mice (n = 6 in each group) was measured as in (34). The ratio (Tg compared with non-Tg) in each fat depot were expressed as mean ± SEM. (F) Fat weight in whole body (Wh.) and abdominal region (Abd.) assessed by DEXA (18) (Lunar Corporation, Madison, WI) (left). LF, low-fat diet; HF, high-fat diet. Values were expressed as mean ± SEM; *, P < 0.01 compared with non-Tg mice; †,P <0.005 compared with non-Tg mice in respective diet (18-week-old male mice, n = 8 in each group). Fat weight in Abd./Wh.(%) (right). *, P < 0.005 compared with non-Tg mice; †, P < 0.01 compared with non-Tg mice in respective diet.

To address glucocorticoid metabolism in Tg mice, we measured corticosterone concentrations in adipose tissue and in serum. In two lines (7 and 10), serum corticosterone levels, measured under nonstressed conditions (14), were similar to those in control mice [117 ± 25 and 119 ± 4 ng/ml in non-Tg mice (n = 8) compared with 96 ± 13 and 104 ± 13 ng/ml in Tg mice (n = 10), respectively. All values in the present study were expressed as mean ± SEM]. In contrast, corticosterone concentrations in adipose tissue (15,16) were significantly elevated (∼15 to 30% increase compared with non-Tg mice, P < 0.01), reflecting increased local conversion of corticosterone (Fig. 1B). To confirm that corticosterone levels in the peripheral circulation were not increased, we assessed several parameters that reflect the consequences of circulating glucocorticoids including thymic weight, bone mineral density, lean body mass weight, and linear growth; none were altered in Tg mice (Web note 1) (17).

From weaning to 9 weeks of age, body weights of non-Tg and Tg male mice fed normal (low-fat) diets were indistinguishable. However, weights diverged after 9 weeks and by 15 weeks of age, Tg mice weighed 16% more than non-Tg mice (Fig. 1C). The Tg showed increased sensitivity to weight gain on a high-fat diet. At 15 weeks of age, the body weight of Tg mice fed a high-fat diet for 9 weeks was 40.2 ± 1.2 g, which exceeded that of Tg mice fed a low-fat diet (32.5 ± 0.9 g, P < 0.001), non-Tg mice fed a high-fat diet (33.3 ± 0.7 g, P < 0.001), and non-Tg mice fed a low-fat diet (28.8 ± 0.5g, P < 0.0005). The 24 ± 1.7% weight gain in Tg mice fed a high-fat diet was greater than the 16 ± 0.6 % weight gain in non-Tg mice fed a high-fat diet (P < 0.001). Female Tg mice were not evaluated in the present study. External examination of the Tg mice suggested a prominent abdominal component to the weight gain (Fig. 1D). To evaluate the distribution of adiposity in Tg mice, we measured the weight of three different fat depots (Fig. 1E). Visceral fat is defined as fat depots located in the area of the portal circulation and is chiefly made up of omental and mesenteric fat (5). As mesenteric fat has a considerable amount of tissue and is easily isolated, the weight of mesenteric fat was measured as a representative of visceral fat (3, 5). The weight of unilateral epididymal adipose tissue did not differ between non-Tg and Tg mice [255 ± 14 mg (non-Tg) compared with 275 ± 12 (Tg)], and there was a small but significant increase in subcutaneous abdominal adipose tissue [1112 ± 75 mg (non-Tg) compared with 1432 ± 102 (Tg), P < 0.05]. In contrast, the weight of mesenteric adipose tissue was strikingly increased in Tg mice (795 ± 26 mg) compared with non-Tg mice (215 ± 14) (P < 0.001). Similar findings were observed in an additional independent line “10” [Web notes 2 and 3].

We also used dual energy x-ray absorptiometry (DEXA) (18) to measure the amount of fat in the whole body or in the abdominal region, defined as the area between the lower border of the thoracic rib cage and the upper border of the pelvic cavity (Fig. 1F). This region contains several fat depots, including visceral, retroperitoneal, and surrounding subcutaneous fat (3, 5,18). In non-Tg mice at 18 weeks of age on low-fat diets, whole body and abdominal adipose tissue weight were 4.85 ± 0.22 g and 1.35 ± 0.16 g, respectively, whereas these weights in Tg mice were 7.97 ± 0.6 g and 3.0 ± 0.14 g, respectively. Accordingly, 27.5 ± 2.1% of adipose weight was in the abdominal region in non-Tg mice, compared with 37.9 ± 1.2% in Tg mice (P < 0.005). Fat accumulation in the abdominal region of non-Tg mice on high-fat diets was comparable to that in Tg mice on low-fat diets, and the ratio was further exaggerated in Tg mice fed high-fat diets. Thus, modest overexpression of 11β HSD-1 in all adipose tissue produces disproportionate accumulation of visceral fat depots.

To examine the basis for increased energy balance in these mice, we monitored food consumption (Fig. 2A). Food intake was increased by 10.2 ± 0.7% (P < 0.05) and 17.1 ± 1.2% (P < 0.01), respectively, in lines 7 and 10, and this paralleled the increase in body weight (line 7, 10.2 ± 1.3% and line 10, 21.0 ± 2.5%). Thus, in addition to altering body fat distribution, adipose overexpression of 11β HSD-1 engenders systemic alterations that result in hyperphagia.

Figure 2

Metabolic phenotype in aP2/11β HSD-1 transgenic mice. (A) Cumulative food intake (g/week) in line 7 and line 10 mice (male, 14 weeks of age, n = 9 in each group). Values were expressed as mean ± SEM. *,P < 0.05 compared with non-Tg mice (line 7); †, P < 0.01 compared with non-Tg mice (line 10). (B) Profile of intraperitoneal glucose tolerance test (1 mg of glucose per kg of body weight in awake mice after a 12-hour fast) [line 7, 18-week-old male Tg mice (n = 8) (▪) and non-Tg (n = 10) (♦)] (left) and insulin tolerance test (0.75 U insulin per kg of body weight after a 6-hour fast) (right) (35). Values were expressed as mean ± SEM; *, P < 0.005 compared with non-Tg mice. Data were assessed by analysis of variance (ANOVA) with repeated measures analysis (Statview 4.01; Abacus Concepts). (C) Adipocyte number and size in mesenteric and subcutaneous adipose tissue from 16-week-old male non-Tg and Tg mice (n = 3 in each group) were assessed as in (35). Values were expressed as mean ± SEM; *,P < 0.01 compared with non-Tg mice; †,P < 0.05 compared with non-Tg mice. (D) Glucocorticoid receptor α isoform (GRα) (20) and LPL (21) mRNA expression in adipose tissue from 18-week-old non-Tg and Tg mice (n = 5 in each group). Specific mRNA for GRα was determined by quantitative reverse transcription (RT)-32P polymerase chain reaction (PCR) with sense (5′-TGCTATGCTTTGCTCCTGATCTG-3′) and antisense (5′-TGTCAGTTGATAAAACCGCTGCC-3′) primers. LPL mRNA was determined by Northen blot. The β-actin mRNA were determined by RNase protection assay (HybSpeed RPA kit; Ambion, Austin, TX). The results were normalized to the signal generated from β-actin mRNA. Values were expressed as mean ± SEM. GRα: *, P < 0.001 compared with non-Tg SubQ or Tg SubQ, respectively. LPL: *,P < 0.01 compared with non-Tg SubQ; †,P < 0.001 compared with non-Tg Mes. (E) Portal FFA (left) and corticosterone (right) levels in nonTg and Tg mice (20-week-old male mice, n = 8 in each group). Venous blood was taken (∼200 μl/mouse) from the proximal end of the portal vein with a 28G small syringe. Values were expressed as mean ± SEM. FFA: *, P < 0.005 compared with non-Tg systemic circulation (Circ); †,P < 0.001 compared with non-Tg portal vein (Portal). Corticosterone: *, P < 0.01 compared with non-Tg Portal.

Despite relatively modest degrees of overall obesity in Tg mice fed low-fat diets, the mice were markedly hyperglycemic [blood glucose fed ad libitum: 111 ± 6 mg/dl (non-Tg) compared with 164 ± 5 (Tg), P < 0.001] and hyperinsulinemic [230 ± 17 pg/ml (non-Tg) compared with 974 ± 254 (Tg), P < 0.03]. Intraperitoneal glucose and insulin tolerance testing in line 7 revealed pronounced glucose intolerance and insulin resistance (Fig. 2B). Similar findings were observed in a second independent line (Web note 4). Serum levels of free fatty acids (FFA), triglyceride, and leptin were also significantly increased (Table 1, top). Consistent with the notion that glucocorticoids are potent secretagogues of leptin from adipose tissue (19), leptin levels in Tg mice were disproportionately elevated as assessed by the ratio of leptin (pg/ml)/body fat (g): 412 ± 96 (non-Tg) compared with 856 ± 130 (Tg), P < 0.01, suggesting that leptin resistance developed in Tg mice.

Table 1

Metabolic parameters and adipocyte gene expression in 18-week-old male mice. Serum free fatty acids (FFA) and triglyceride were measured by NEFA-C kit and Triglyceride-E kit (Wako Chemicals, Richmond, VA). Leptin was determined by radio immunoassay (Linco Research, St. Louis, MO). Values were expressed as mean ± SEM. The mRNA for Acrp30-AdipoQ (23) and β-actin were determined by RNase protection assays (HybSpeed RPA kit, Ambion, Austin, TX). Resistin (25) and angiotensinogen (26) mRNA were determined by quantitative RT-32P PCR with the following primers. Resistin: sense, 5′-ATGAAGAACCTTTCAT TTCC-3′; antisense, 5′-CTTGCACACT GGCAGTGACA-3′; angiotensinogen: sense, 5′-CCTGAAGGCCACCATCTTCT-3′; antisense, 5′- GATCATTGCGACCTGGGCAG-3′. Mitochondrial uncoupling protein-1 (UCP-1) mRNA was determined by Northern blot (27). The results were normalized to the signal generated from β-actin mRNA and expressed as mean ± SEM. Values in Tg mice were expressed as a percentage of those in non-Tg mice. mesenteric, mesenteric adipose tissue, BAT, brown adipose tissue.

View this table:

Although there was no difference between non-Tg and Tg mice in adipocyte number in mesenteric or subcutaneous abdominal fat depots, adipocyte size in these depots was increased 3.2-fold (P < 0.006) and 1.5-fold (P < 0.05) in Tg mice compared with non-Tg mice (Fig. 2C). Consistent with previous reports (3–5), the levels of glucocorticoid receptor α isoform (GRα) (20) mRNA were threefold higher in mesenteric compared with subcutaneous adipose tissue of both non-Tg and Tg mice (P < 0.001) (Fig. 2D). Enhanced GRα expression in visceral fat may account, at least in part, for the exaggerated accumulation of fat in this depot, despite similar overexpression of 11β HSD-1 in all fat depots. Lipoprotein lipase (LPL) mRNA, which is known to be upregulated by glucocorticoids (21), was significantly increased in mesenteric adipose tissue of Tg mice (3.5-fold, P < 0.001) and less so in subcutaneous adipose tissue (2.4-fold, P < 0.01) (Fig. 2D). LPL overexpression can drive lipid accumulation in adipose depots (3, 5, 22).

We also studied the expression of adipose genes that are known or suspected to influence systemic metabolic pathways (Table 1, bottom). Adipocyte complement–related protein of 30 kD, (Acrp30)-AdipoQ (23), has been suggested to serve as an insulin-sensitizing factor (24). The mRNA level in mesenteric fat from Tg mice was markedly decreased, consistent with a role for this factor in the insulin-resistant state. The mRNA for resistin, which has been suggested to be involved in glucose homeostasis (25), was significantly decreased in Tg mice. Angiotensinogen mRNA, which is up-regulated by glucocorticoids (26), was substantially increased in Tg mice. Mitochondrial uncoupling protein-1 (UCP-1) mRNA (27) in interscapular BAT, which is down-regulated by glucocorticoids (28), was significantly decreased in Tg mice, suggesting a possible role for decreased BAT function in energy dyshomeostasis in this model. Tumor necrosis factor–alpha (TNF-α), a fat cell–derived cytokine that can cause insulin resistance (3,22), was significantly elevated in serum of Tg mice compared with non-Tg mice [39 ± 1.1 pg/ml (non-Tg) compared with 76 ± 11.2 pg/ml (Tg), P < 0.02].

In patients with the metabolic syndrome, considerable evidence suggests that increased FFA draining from visceral adipose tissue into the portal circulation contributes to hepatic insulin resistance (3, 5, 29). Portal FFA levels were increased by 3.3-fold in Tg mice (P < 0.001) (Fig. 2E). Because visceral fat produces active corticosterone through 11β HSD-1, we considered the possibility that visceral adipocytes of transgenic mice release sufficient corticosterone into the portal vein to alter the levels exposed to the liver. Indeed, portal vein corticosterone levels in Tg mice were increased 2.7-fold [129 ± 11.1 ng/ml (non-Tg) compared with 349 ± 79.5 ng/ml (Tg),P < 0.01] (Fig. 2E). Thus, visceral fat may affect hepatic metabolism by portal production of glucocorticoids as well as FFA.

Glucocorticoids regulate adipose tissue differentiation, function, and distribution, and their systemic excess produces a syndrome of central obesity with diabetes, hyperlipidemia, and hypertension, known as Cushing's syndrome (3–5). Although subtle alterations in the endocrine hypothalamic pituitary adrenal (HPA) axis have been reported in some studies of obesity (4, 5,30), these have been controversial, and no clear role for increased circulating glucocorticoids in visceral obesity has emerged. On the other hand, a role for increased local cortisol reactivation in human obesity is suggested by several findings (5–7,31). 11β HSD-1 activity is higher in human visceral compared with subcutaneous adipose tissue (31), and reactivation of cortisone to cortisol is increased selectively in adipose tissue of obese humans, while impaired in liver (7). Similar findings were reported in obese Zucker rats (8). The thiazolidinedione (TZD) class of antidiabetic agents that are ligands for peroxysome proliferator–activated receptor (PPAR) γ markedly reduce adipocyte 11β HSD-1 mRNA and activity both in vivo and in vitro (32). Because TZDs preferentially reduce visceral fat accumulation in humans (3, 5, 22,33), suppression of adipose 11β HSD-1 by TZDs could be a mechanism for this fat redistribution and may play a role in their antidiabetic effects.

Our finding that a modest increase in the activity of 11β HSD-1 in adipose tissue of mice is sufficient to cause hyperphagia with visceral obesity and its most critical metabolic complications demonstrates that glucocorticoid-dependent adipocyte pathways have an unexpectedly major impact on systemic biology. Adipose tissue of obese humans is reported to have increased activity of 11β HSD-1 of similar or greater magnitude than that observed in our transgenic mice (7). These findings strongly suggest that increased adipocyte 11β HSD-1 is a common molecular mechanism for visceral obesity and the metabolic syndrome and may be an exciting pharmaceutical target for the treatment of this prevalent disorder.

  • * To whom correspondence should be addressed. E-mail: jflier{at}


Stay Connected to Science

Navigate This Article