A Polygenic Model of the Metabolic Syndrome With Reduced Circulating and Intra-Adipose Glucocorticoid Action

Fat and Lean mouse lines were selected from a three-way cross of two inbred lines (JU and CBA) and one outbred line (CFLP), as described in detail previously (22). Briefly, selection for the first 20 generations was on the ratio of gonadal fat pad weight to body weight of 10-week-old males and subsequently on dry matter content of males at 14 weeks of age. Inbred lines were initiated from a single family of each of the lines after 47 generations of divergent selection and maintained by full sib mating (22). In the current studies, we analyzed age-matched male Fat and Lean mice from inbred lines of generations 35–42 (n = 8–10 per group, n = 5 for corticosterone addition studies) at ∼6–9 months of age.

Mice were fed standard chow (Rat and Mouse no. 3; Special Diet Services, Witham, Essex, U.K.) from weaning. Animals were housed in standard cages with controlled lighting (12-h light, lights on at 7:00 a.m.). For experiments involving unstressed plasma corticosterone, ACTH, and glucose sampling, animals were acclimated to single housing and tail nicks obtained within 1 min of disturbing the cage. For high-fat diet experiments, mice were switched to a high-fat diet (58% kcal as fat, D12331; Research Diets, New Brunswick, NJ) or a calorie-matched control diet (11% kcal as fat, calories substituted with sucrose, D12328; Research Diets) for a further 18 weeks.

Insulin and leptin were measured using enzyme-linked immunosorbent assay kits (Crystalchem, Downers Grove, IL), plasma corticosterone was measured using an in-house radioimmunoassay (19). Plasma glucose was measured with an Infinity glucose hexokinase kit (Thermo Electron, Melbourne, Australia), nonesterified fatty acids (NEFAs) with a NEFA C kit (Wako Chemicals, Nuess, Germany), and triglycerides with an L-Type triglyceride kit (Wako). Insulin, glucose, NEFA, and triglycerides were measured at ∼1:00 p.m. after a 6-h fast. Measurements of plasma renin and angiotensinogen concentrations involved radioimmunoassay of angiotensin I generated from incubations of renin plus angiotensinogen. Briefly, for renin concentration, plasma was diluted fivefold with Tris buffer and then incubated with excess substrate (plasma from a binephrectomized rat collected 48 h after surgery). Angiotensinogen concentration was measured in plasma diluted 10-fold with Tris buffer then incubated with excess renin solution (cytosol extracted from the submaxillary glands of male 129 mice). ACTH was measured with an enzyme-linked immunosorbent assay (Biomerica, Newport Beach, CA). Fecal steroids were measured by a radioimmunoassay method modified from Pihl and Hau (25). Briefly, 24-h fecal samples were collected into preweighed tubes and stored at −20°C. Samples were homogenized in four volumes of water. Then, ∼1 g of homogenate was weighed and extracted twice with five volumes of dichloromethane. The dichloromethane fractions were pooled and washed once with one volume of NaOH (0.1 mol/l) and twice with one volume of water. The dichloromethane extract was dried down under a stream of nitrogen and reconstituted in assay buffer.

Systolic blood pressure was measured in conscious warmed mice by tail cuff plethysmography as previously described (26), using a tail cuff designed for mice (Harvard Apparatus, Edenbridge, Kent, U.K.).

Liver (0.02 mg/ml protein) and adipose (0.2 mg/ml protein) samples were homogenized as previously described (19) and incubated with 10 nmol/l [1,2,6,7-³H]corticosterone (Amersham Pharmacia Biotech, Amersham, U.K.) and an excess (400 μmol/l) of the 11β-HSD-1–specific cofactor NADP. (Under in vitro conditions in homogenized tissues, 11β-HSD-1 is bidirectional, but it is more stable with assay of dehydrogenation.) This assay was in the linear range of protein concentration and product formation. Briefly, after a 30-min (liver), 20-min (epididymal and subcutaneous), or 1-h (mesenteric) incubation, steroids were extracted with ethyl acetate, separated by thin-layer chromatography, identified by migration in comparison to unlabeled standards, exposed to phosphorimager film (FLA2000; Fujifilm, London), and analyzed by quantitative phosphorimager software (Aida; Raytek Scientific, Sheffield, U.K.) (19).

Tissues were dissected and rapidly frozen under dry ice. Tissues were homogenized in Trizol (Gibco, Paisley, U.K.). Total RNA was purified with a binding matrix (RNaid Plus kit, BIO 101; Anachem, Luton, Bedfordshire, U.K.) and eluted in diethylpyrocarbonate-treated water containing 400 units/ml RNAsin (Promega, Southampton, U.K.) and 10 mmol/l dithiothreitol. RNA (5–20 μg) was resolved on a 1% MOPS formaldehyde gel and blotted according to standard Northern blot procedures in 20 × SSC onto Hybond N⁺ membranes (Amersham, Little Chalfont, U.K.). Probes were labeled with ^32Pd-CTP using a random primed labeling kit (Amersham), purified through NICK columns (Pharmacia-Amersham, Little Chalfont, U.K.), and hybridized overnight in high-SDS (6%) phosphate buffer (0.2 mol/l NaH₂PO₄, 0.6 mol/l Na₂HPO₄, 5 mmol/l EDTA) containing 0.5 mg/ml denatured salmon testes DNA (Sigma) at 65°C. Blots were washed at 65°C to a maximum stringency of 0.5 × SSC and 0.1% SDS and quantified with phosphorimager analysis (19). Probes were as described (9,19,20). All liver gene expression levels were corrected to U1 RNA as an internal standard and are presented as arbitrary units. Adipose gene expression was similarly corrected to cyclophilin B.

Livers were sectioned (30 μm) and stained with oil red O (Sigma) to identify neutral lipids, cholesterol, and fatty acids (red color). Nuclei were counter-stained (blue) with hematoxylin (Sigma). Rinses were as follows: water, 60% isopropanol, 0.1% wt/vol oil red O in isopropanol for 10 min, 60% isopropanol, three water rinses, Mayers hematoxylin for 90 s, water, 1% ammonium, and water (19). Sections from livers in each experimental group were processed simultaneously. Finally, slides were covered in aqueous mount under a coverslip and viewed with a light microscope at equal light intensity. Images shown are at 1.25× magnification. Liver triglycerides were extracted by homogenization in 1 ml isopropanol/100 mg liver and then shaken for 45 min. Samples were spun at 3,000g for 10 min and supernatants assayed using triglyceride reagent (Thermotrace).

Expression of glucocorticoid receptor and 11β-HSD-1 were measured by autoradiography with ³⁵S-dUTP probes in 10-μm tissue sections (17). Briefly, 4 × 10⁶ cpm of each specific probe was hybridized overnight at 50–55°C, followed by an RNase step at 37°C for 1 h, a series of washes at 60–70°C, and exposure to film for autoradiographic quantification or to emulsion for silver grain analysis (17).

Significant differences between Fat and Lean lines were tested with one- and two-way ANOVA for diet/treatment and line interactions. Post hoc tests for group differences were determined with Tukey tests.

Chow-fed male Fat and Lean mice were killed at 8:00 a.m. (the diurnal nadir for corticosterone in mice) for analysis of gene expression and enzyme activities. Fat mice exhibited substantially higher adipose depot–to–body weight ratios than Lean mice (Table 1), as previously described (22–24). Notably, Fat mice had a proportionately higher subcutaneous and epididymal fat mass relative to the metabolically disadvantageous (1,4) visceral mass, compared with Lean mice (Table 1). Fat mice had elevated fasting plasma glucose, insulin, free fatty acid, and triglyceride levels, consistent with insulin resistance and dyslipidemia (Table 1). Fat mice also exhibited fatty liver with markedly increased oil red O staining in liver sections (Fig. 1A) and significantly higher hepatic triglyceride levels (Fig. 1B). Fat mice had elevated blood pressure compared with Lean mice (Fig. 2A). This was associated with markedly increased (threefold) renin and maintained plasma angiotensinogen levels (Fig. 2B and C).

Many rodent models of obesity are characterized by hypercorticosteronemia, including monogenic Lep^ob mice and Zucker fa/fa rats (5,16). However, unstressed plasma corticosterone levels in Fat mice were significantly lower than in Lean mice (Table 2). We found that 24-h fecal corticosterone, an indicator of integrated daily corticosterone exposure, was also lower in Fat mice (Table 2). Consistent with reduced circulating plasma glucocorticoid levels, adrenal weight was reduced, and thymus weight, where glucocorticoids induce involution, was increased (Table 2). A higher thymus-to-adrenal ratio further indicated lower corticosterone exposure, independent of potentially confounding effects of organ/body weight correction in the Fat and Lean lines. Circulating ACTH was threefold lower in Fat mice (Table 2), in line with their lower circulating corticosterone levels, and this suggested either lower hypothalamo-pituitary-adrenal axis (HPA) drive and/or increased central HPA feedback. However, glucocorticoid receptor and 11β-HSD-1 mRNA levels, determined by in situ hybridization in well-established central HPA feedback sites, were similar in Fat and Lean mice (Table 2).

Lower indexes of glucocorticoid exposure but unaltered glucocorticoid receptor and 11β-HSD-1 in hypothalamic feedback sites suggested that Fat mice had increased relative glucocorticoid sensitivity. To test this, we implanted corticosterone (releasing ∼100 μg/day) or vehicle pellets for 3 weeks and analyzed metabolic responses in Fat and Lean mice. Similar corticosterone levels were achieved by this method in the two strains in corticosterone-implanted animals (Table 3). There was no effect of corticosterone on total body weight gain over the 3-week treatment (Table 3). However, two-way ANOVA indicated that corticosterone increased subcutaneous and mesenteric, but not epididymal, fat mass (P < 0.05) (Table 3) in Fat and Lean mice. Because epididymal fat did not respond to corticosterone in either strain, we used this as a reference to look at potential within-strain effects of corticosterone on fat accumulation in the visceral and peripheral depots. On analyzing subcutaneous-to-epididymal and mesenteric-to-epididymal ratios in treated versus untreated mice of each strain, a strong interaction between strain and corticosterone treatment was found (P < 0.008), indicating that subcutaneous fat is hypersensitive to corticosterone in the Fat but not the Lean mice (P < 0.02). Consistent with this redistribution of fat and the acknowledged role of subcutaneous fat as the major depot of leptin expression (27), Fat mice showed an exaggerated increase in the glucocorticoid-inducible (28,29) plasma leptin levels, relative to the increased fat mass, in response to exogenous corticosterone. Furthermore, corticosterone administration caused a profound increase in circulating insulin levels (Table 3) in Fat but not Lean mice, indicative of exacerbated glucocorticoid-mediated peripheral insulin resistance (30).

Since obesity is associated with an altered profile of peripheral glucocorticoid metabolism, we assessed indicators of glucocorticoid action in peripheral tissues. In contrast to monogenic rodent obesity (7,16), adipose 11β-HSD-1 activity and mRNA levels were markedly reduced in Fat mice (Fig. 3A and B). However, in Fat mice, visceral adipose (mesenteric) 11β-HSD-1 levels were relatively higher compared with their other adipose depots (Fig. 3A), similar to the depot-selective increase found in Lep^ob mice and Zucker fa/fa rats (7,16,31). Therefore, some aspects of 11β-HSD-1 dysregulation with obesity are preserved in this model, albeit from a lower baseline than Lean mice. Adipose glucocorticoid receptor mRNA levels were also markedly reduced (Fig. 3C) in Fat mice. Consistent with a relative deficiency of glucocorticoid signaling in adipose tissue (7,20), glucocorticoid-regulated transcript levels encoding angiotensinogen and lipoprotein lipase were markedly lower in the adipose tissue of Fat mice (Fig. 3D and E). Expression of mRNA encoding adipocyte protein 2 (FABP4), a protein involved in lipid accumulation, was higher in Fat mice (Fig. 3F), as expected, suggesting that lipid accumulation is largely independent of glucocorticoid action in Fat mice.

In contrast to adipose tissue, hepatic 11β-HSD-1 mRNA and activity levels were elevated in Fat compared with Lean mice (Fig. 4A and B). Liver glucocorticoid receptor mRNA levels were also higher in Fat mice and correlated positively with 11β-HSD-1 levels (Fig. 4C). Consistent with elevated hepatic 11β-HSD-1 (9) and thus increased intra-hepatic glucocorticoid action, angiotensinogen mRNA levels were higher in liver of Fat mice than that of Lean mice (Fig. 4D). LDL receptor mRNA levels were elevated in Fat mice (177 ± 19.7 vs. 100 ± 15.3% for Fat vs. Lean mice, respectively; P < 0.05), again consistent with hepatic gene expression changes in transgenic mice overexpressing 11β-HSD-1 selectively in liver (9) and also with a recent study describing altered cholesterol metabolism deriving from a chromosome 15 quantitative trait loci in Fat mice (32). We also measured mRNAs encoding enzymes of hepatic glucose production (PEPCK), fatty acid synthesis (fatty acid synthase), fat oxidation (carnitine palmitoyl transferase-1), and lipid uptake (hepatic lipase), none of which differed between Fat and Lean mice (data not shown).

Chronic high-fat feeding accelerates the onset of obesity and metabolic disease in many mouse strains. We fed cohorts of Fat and Lean mice a high-fat or isocaloric control diet for 18 weeks to determine whether their divergent patterns of glucocorticoid metabolism might impact on fat accretion/distribution and parameters of metabolic disease (20). Weight gain was similar in Lean mice on both diets. Fat mice gained substantially more weight than Lean mice on control diet (2.5-fold more) and exhibited an exaggerated weight gain on high-fat diet (3.5-fold more), despite consuming 28% less control diet (P < 0.001) and 36% less high-fat diet (P < 0.001) than Lean mice (Table 4). Weight gain in Fat mice on the high-fat diet was associated with a 34% (body weight adjusted) increase in subcutaneous fat mass but a 25% reduction in mesenteric adipose fat mass with high-fat feeding (Table 4), a pattern of preferential fat deposition similar to 11β-HSD-1–deficient mice (20). Intriguingly, Lean mice exhibited a significant and consistent loss of fat mass (decreases of 34–39%) in all adipose depots with high-fat feeding (Table 4).

High-fat feeding did not worsen the already elevated plasma insulin and liver triglyceride levels in Fat mice (Fig. 5A and B), consistent with the notion that peripheral-type fat accumulation might be relatively protective (20). Intriguingly, Lean mice showed a beneficial and adaptive reduction of these indexes of metabolic disease (Fig. 5A and B), indicating a “superlean” response to high-fat feeding. High-fat feeding in the Fat mice increased blood pressure, whereas Lean mice resisted high-fat diet–induced hypertension (Fig. 5C). Finally, both genotypes exhibited diet-induced downregulation of mesenteric adipose tissue 11β-HSD-1 activity (Fig. 5D). However, adipose 11β-HSD-1 activity was downregulated to a greater extent, relative to the control-fed “set point,” with high-fat feeding in mesenteric (visceral) adipose tissue (Fig. 5D) of Lean compared with Fat mice. This is consistent with previous studies where a metabolic disease–resistant strain of mice exhibited a greater magnitude of adipose 11β-HSD-1 downregulation than a metabolic disease–susceptible strain in response to high-fat feeding (31).

Recent evidence suggests that increased glucocorticoid action within adipose tissue explains the similarities between Cushingoid and idiopathic obesity in the absence of high plasma glucocorticoids (7,12–17). This notion is supported in rodents, where some rare single-gene obesity mutations cause defects in glucocorticoid metabolism similar to those found in human obesity (7,16). However, idiopathic human metabolic syndrome results from multiple gene-environment interactions, each believed to be of relatively small effect compared with the extremely rare monogenic defects that produce profound and morbid obesity (2). We therefore tested the glucocorticoid-obesity hypothesis in a unique model that more closely reflects the polygenic make up of human obesity. Furthermore, although obesity often associates with insulin resistance and metabolic disease (33), this relationship is not inevitable (34,35). We demonstrated that the obesity of Fat mice was indeed associated with fasting hyperglycemia, insulin resistance (hyperinsulinemia, hypertriglyceridemia, and elevated free fatty acid levels), fatty liver, and hypertension. Thus, Fat mice have a full metabolic syndrome that is determined by its underlying genes per se and is not secondary to hyperphagia.

Having established that Fat mice model some important aspects of human metabolic syndrome, our primary aim was to test whether the changes in glucocorticoid biology hypothesized to underlie metabolic syndrome in humans and monogenic obesity in rodents would be recapitulated in this arguably more relevant polygenic model. Circulating basal corticosterone levels and indexes of glucocorticoid exposure were lower in Fat mice, in agreement with some studies in human obesity (12), but in contrast with rodents carrying monogenic defects in leptin or its receptor (5,16). Though unusual, obesity can occur with glucocorticoid deficiency in mice. One example is proopiomelanocortin (POMC)-null mice (36), where centrally driven, melanocortin-deficient hyperphagia models a rare monogenic obesity syndrome in humans (2). The POMC gene (a precursor for ACTH) maps within the chromosome 12 obesity quantitative trait loci of Fat mice (24) and was therefore an attractive candidate for mediating reduced ACTH, low corticosterone, and obesity (36). However, Fat and Lean mice had similar arcuate nucleus POMC expression levels (V.D., unpublished observations) and relative hypophagia, indicating that central POMC as well as primary leptin (23) and neuropeptide Y (37) defects were not contributory. Our data also suggested that altered central glucocorticoid receptor (38) and 11β-HSD-1 (20) were not responsible for the divergent food intake profiles of Fat and Lean mice.

Although obesity and metabolic disease in Fat mice was not caused by plasma glucocorticoid excess, we considered the possibility that differential tissue-specific sensitivity to glucocorticoids (7,12–16,31) might contribute to the phenotype. Accordingly, we measured expression of two key determinants of peripheral tissue glucocorticoid action, adipose and liver glucocorticoid receptor and 11β-HSD-1, as well as the physiological response to exogenous glucocorticoids. In striking contrast with Lep^ob mice (7,31) and Zucker fa/fa rats (16), Fat mice had lower adipose but increased hepatic levels of 11β-HSD-1 and glucocorticoid receptor. This pattern is not consistent with glucocorticoid-mediated obesity. However, Fat mice did show induction of alternate adipose lipid accumulation pathways, such as adipocyte protein 2, that likely compensate, with other mechanisms (32), for low lipoprotein lipase levels. Fat mice also had a pronounced response to exogenous corticosterone, with markedly increased circulating levels of the glucocorticoid-inducible (28,29) adipokine leptin and exacerbation of peripheral insulin resistance (30). Exogenous corticosterone increased fat mass in both strains of mice. However, in subcutaneous fat, this increase was greater in Fat compared with Lean mice. Consistent with the notion of glucocorticoid hypersensitivity, Fat mice showed a disproportionate increase in the glucocorticoid-regulated plasma leptin levels relative to the gain in fat mass with exogenous corticosterone, whereas leptin levels remained unchanged in Lean mice.

Although it is abundantly clear that aberrantly elevated adipose 11β-HSD-1 can drive a complete metabolic syndrome in mice (7,17) and potentially contributes to it in humans (13–15), there is a precedent for obesity with low intra-adipose glucocorticoid reactivation. Thus, comparable adiposity develops in C57BL/6J and A/J mice on high-fat feeding, despite dynamic downregulation of adipose 11β-HSD-1 (31). In addition, 11β-HSD-1^−/− mice develop obesity—albeit attenuated—on high-fat feeding (20). However, in these cases, relative or complete 11β-HSD-1 deficiency is associated with a marked protection from the detrimental metabolic consequences of the high-fat feeding (20,31). We observed that Fat mice preferentially distribute their fat mass into subcutaneous fat depots and away from visceral fat depots with high-fat feeding, a distribution associated with lowered risk of metabolic disease (1,4) and consistent with the effects of adipose 11β-HSD-1 deficiency (20). Furthermore, high-fat diet–fed Fat mice showed an unexpectedly mild response to the high-fat diet, with, for example, no worsening of basal insulinemia, glycemia, or circulating triglycerides. Fat mice were originally selected for elevated epididymal fat mass (22) and have 4-fold higher mass of peripheral fat but only 2.5-fold more visceral fat than Lean mice. Epididymal fat is akin to peripheral (e.g., subcutaneous) depots, where fat accumulation may even be metabolically protective relative to visceral fat (39). Indeed, in human Prader-Willi syndrome, peripheral obesity associates with an improved metabolic profile compared with subjects of comparable total fat mass but pronounced visceral distribution (40). Therefore, initial selection for increased “peripheral-type” fat mass (22), and possibly an associated selection of low adipose 11β-HSD-1, has afforded some degree of metabolic protection when the Fat mice are challenged with a high-fat diet (20). Overall, Fat mice and other strains (31) suggest that adipose 11β-HSD-1, while not strongly determining total adiposity, may influence the metabolic consequences of increased adipose tissue mass through altering fat distribution and insulin sensitivity. This view is supported by recent human studies showing that adipose 11β-HSD-1 correlates more strongly with insulin resistance than with adiposity per se (15,41). It is also striking that Lean mice appear to lose weight on high-fat feeding. Although a number of obesity-resistant strains of mice exist (31), they tend to maintain rather than lose fat mass with high-fat feeding. Further studies will be needed to elucidate putative mechanisms of fat loss, such as increased thermogenic capacity, or increased physical activity in Lean mice.

In contrast to adipose tissue, a role for elevated liver glucocorticoid action is indicated in the detrimental metabolic phenotype of Fat mice. Transgenic mice overexpressing 11β-HSD-1 selectively in the liver exhibit insulin resistance and hepatic fat accumulation as well as elevated liver angiotensinogen and LDL receptor mRNA levels, similar to our findings in Fat mice (9). Elevated liver 11β-HSD-1 is also found coassociated with insulin resistance and metabolic disease in human myotonic dystrophy (18), suggesting that insulin resistance, secondary to obesity in Fat mice, might be a key factor in elevating hepatic 11β-HSD-1. However, because hepatic overexpression of 11β-HSD-1 does not create obesity even with high-fat feeding (9), this feature of Fat mice is very likely a consequence rather than a cause of the increased adiposity, although a more direct genetic contribution to liver-derived dyslipidemia is possible (32). The reasons for elevated liver glucocorticoid receptor levels in Fat versus Lean mice are unclear because tissue glucocorticoid receptor levels are not affected by, for example, chronically increased intra-adipose glucocorticoid reamplification (7). These data have suggested that 11β-HSD-1 is a more crucial determinant of tissue glucocorticoid action than circulating corticosterone or tissue glucocorticoid receptor levels (7). Because glucocorticoids positively regulate 11β-HSD-1 levels in liver cells (42), an 11β-HSD-1–mediated feed-forward loop on 11β-HSD-1 expression may explain the correlation of glucocorticoid receptor and 11β-HSD-1 in liver of Fat but not Lean mice.

Hypertension in the Fat mice appears to be driven by an activated renin-angiotensin system, with elevated plasma renin levels, consistent with a compensatory response of blood pressure regulation to low ACTH levels. It is intriguing that Fat mice share the feature of elevated renin levels with the fat-selective 11β-HSD-1 overexpressor, despite having opposite intra-adipose glucocorticoid action profiles. One possible common metabolic link between the adipose and liver 11β-HSD-1 transgenic overexpression models and Fat mice is hyperinsulinemia, a known modulator of hypertension through multiple mechanisms (43). However, because insulin levels remain unchanged while blood pressure increases with high-fat feeding in Fat mice, additional processes are implicated, and further studies are required to unravel the mechanism of hypertension in Fat mice.

In summary, we describe a polygenic model of the metabolic syndrome that exhibits a novel profile of reduced HPA activation and adipose tissue glucocorticoid deficiency but selective liver glucocorticoid amplification. This suggests a similar endocrine subset may exist within the “continuum” of human metabolic syndromes. Tentative evidence for low adipose 11β-HSD-1 and obesity in humans has been reported (44). However, further detailed studies are necessary to identify such, presumably atypical, human populations. Although the therapeutic inhibition of 11β-HSD-1 is still supported by the current work, the emphasis for tissue-selective drug targeting in these putative subjects is switched to the liver, where the enzyme is most highly expressed and responds to such a therapy (21,45). The further identification of genes contributing to obesity in Fat mice will illuminate novel interactions between fat accumulation, metabolic syndrome, and glucocorticoid metabolism.

This work was funded by a Cardiovascular Research Initiative intermediate fellowship from the Wellcome Trust (to N.M.M.) and a program grant from the Medical Research Council (to C.J.K.). J.R.S. is funded by a Wellcome Trust program grant, and M.W. is a British Heart Foundation PhD student. L.B. appreciates funding from SEERAD.

We thank Moira Stewart and Linda Schoen for excellent technical assistance with the Fat and Lean line maintenance. We also thank members of the Endocrinology Unit for their useful comments.