17β-Estradiol suppresses visceral adipogenesis and activates brown adipose tissue-specific gene expression

Animals, treatment and tissue processing

Seven-week-old female C57BL/6J mice (Scanbur BK, Sweden) were housed at 22°C–23°C in a 12 h-12 h light-dark cycle with free access to water and food ad libitum. Mice were acclimatized for 1 week and divided randomly for maintenance on either a low-fat diet (LFD; 4.5 g fat/100 g; Lactamin, Sweden) or a high-fat diet (HFD; 34.9 g fat/100 g; D12492, Research Diets, USA) until 12 months of age [9]. At 11 months of age, both groups were divided for subcutaneous injection for 30 days with either 50 μg/kg/day E2 (Sigma-Aldrich, Sweden; LFD+E2 and HFD+E2 mice) prepared in vehicle or with the same volume of vehicle (sesame oil/ethanol, 9:1; LFD, HFD mice). Body weight and food consumption were measured at the start of treatment and thereafter at 48-h intervals. Tissues were collected from overnight-fasted, euthanized mice, snap-frozen and stored at –80°C. All procedures were approved by the local Ethical Committee.

Intraperitoneal glucose tolerance test (IPGTT)

Mice were fasted overnight, and blood glucose levels were assessed by a glucometer (MediSence, Abbott Scandinavia, Solna, Sweden) before, and at 30, 60 and 120 min following an intraperitoneal injection of glucose (2 g/kg body weight).

Plasma insulin levels

Insulin levels in plasma were determined by using an AlphaLISA kit (Perkin-Elmer, Sweden), according to the manufacturer’s protocol.

Magnetic resonance imaging (MRI)

For in vivo MRI, LFD, HFD and HFD+E2 mice were anesthetized with 1.5%–2.0% isoflurane in O₂ while positioned on a heated pad over which a current of air heated to 37°C was passed to further maintain the body temperature. Fat distribution measurements were performed on a Bruker 4.7-T field strength magnet and a 40-cm horizontal-bore diameter (Bruker Biospec Avance 47/40; Bruker, Karlsruhe, Germany), equipped with a commercially available circular resonator (Bruker) with an inner diameter of 24 mm for RF pulse application and signal detection using a rapid acquisition with relaxation enhancement (RARE) imaging sequence. MRI was performed to acquire a total of 29–34 contiguous 1.5 mm thick transverse slices, corresponding to the thoracic to pelvic regions. Output image series were collected in Paravision 3.01 (Bruker) and exported to ImageJ (1,47v, Wayne Rasband, NIH, USA) for calculation by cumulative pixel analyses of total body volume and volumes of the visceral and subcutaneous fat depots. Comparable ranges of consecutive slices were analyzed for all individuals. A signal threshold was used after a Gauss filter of maximum likelihood, and a class-select interaction was applied to exclude all non-fat tissues in each slice. Results are expressed as a percentage of body volume.

Histology and immunohistochemistry (IHC)

For histology, three sections of 20 μm thickness were cut from adipose tissue from three mice per group by cryostat. The sections were stained with hematoxylin and eosin, and three images per section were collected using a digital camera (10× and 40× magnifications, Leica DM 3000 LED, LAS Version 4.3.0). Adipocyte size was quantified in 80–170 cells per section by ImageJ (1,47v, Wayne Rasband, NIH, USA).

For IHC, sections of 20 μm thickness were prepared from acetone-fixed adipose tissue and blocked with 5% rabbit serum in antibody diluent (Cell Signaling Technology) for 1 h at room temperature. After blocking, the sections were incubated with anti-uncoupling protein 1 (UCP1) antibody (Abcam, UK; 250× dilution in antibody diluent) overnight at 4°C. After three washes with Tris-buffered saline (TBS), the sections were incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG polyclonal antibody (Abcam; 1000× dilution in antibody diluent). Immunostaining was enhanced by the use of diaminobenzidine. After counterstaining with hematoxylin, the sections were treated with ethanol and xylene prior to mounting. All of these steps included negative controls, which consisted of similarly treated sections except that the primary antibody was omitted. Three images per section were collected using a digital camera (10× and 40× magnification, Leica DM 3000 LED, LAS Version 4.3.0).

Western blotting

Snap-frozen adipose tissue was homogenized in radioimmunoprecipitation assay (RIPA) buffer and centrifuged for 30 min at 12,000 g and 4°C. The supernatants were collected, and the protein concentrations were assayed according to a standard assay protocol (DC protein assay, Bio-Rad, USA). Protein samples were run on 10% ECL gels (Amersham, GE Healthcare, UK) and transferred to polyvinylidene difluoride (PVDF) membranes. The membranes were blocked for 1 h in 5% fat-free dry milk (Bio-Rad Laboratories, Sweden) dissolved in TBS, and the membrane was washed once with wash buffer [TBS/0.1% (v/v) Tween-20]. Primary antibody (Table 1) was added in a 10 mL volume of TBS buffer, and the incubation was continued overnight at 4°C. The membranes were then washed and incubated with secondary antibody in 10 mL of TBS for 1 h at room temperature, according to the manufacturer’s recommendations (Table 1). After three further washes with wash buffer, bands were visualized using an ECL detection kit (SuperSignal West Femto, Thermo Scientific Pierce, Nordic Biolabs Sweden) and documented using a LAS 3000 system (Fuji Film Co., Tokyo, Japan). Band intensities were quantified employing ImageJ.

RNA preparation and real-time RT-PCR

Total RNA was prepared using RNeasy Lipid kits and treated with DNase I (Qiagen, Sweden), and reverse-transcribed using TaqMan^® Reverse Transcription Reagents and random hexamer primers (Applied Biosystems, Sweden). All target mRNAs were assayed in triplicate 25 μL reactions which included the respective Assay-on-Demand (Table 2) and 1× Master Mix (both from Applied Biosystems, Sweden) using a 7300 Real-Time PCR system. Expression levels were normalized to 18S (Applied Biosystems, Sweden). Fold changes in target gene expression were determined by the Δ-Δ C_T method.

DNA preparation and methylation analysis

Genomic DNA was extracted from adipose tissue using a PureLink^® Genomic DNA Mini Kit (Thermo Fisher Scientific, Sweden), according to the manufacturer’s protocol for mammalian tissues. Analysis of methylation was carried out using an OneStep qMethyl Kit (Zymo Research, Irvine, CA, USA). Aliquots of 280 ng of each sample of DNA were placed into test and reference reactions containing 28 μL of 2×Test/Reference Reaction Premix in a final volume of 140 μL. The Test Reaction Premix contains methylation sensitive restriction enzymes, while the Reference Reaction Premix does not. All the samples were incubated for 5 h at 37°C, after which 7 μL aliquots of each test and reference reaction were employed as template for duplicate 25 μL real-time PCR reactions containing 2×SYBR Green Master Mix (Applied Biosystems, Sweden) and 50 nM of each primer of interest (Thermo Fisher Scientific, Sweden; Table 3). The primers employed were designed using the web-based primer design tool Primer3. Primer pairs were designed to amplify the mouse homologues of regions which include CpG13S in the human SCD1 promoter [21], CpG48-50 in the human FASN promoter [21], methylation site #6 of the human CPT1b promoter [22] and regions amplified by the ADRB3 and DIO2 primers employed by Kurylowicz et al. [23]. The specificity of amplification of each primer pair was verified by initial melting curve analyses. Amplifications were carried out using a 7300 Real-Time PCR system. As positive and negative controls, aliquots of methylated and non-methylated genomic DNA (Zymo Research, Irvine, CA, USA) were subjected to parallel methylation-PCR analyses. The methylation level of the region of interest of each target gene was expressed as a percentage as determined using the following equation: methylation (%)=100×2^−ΔCt. The ΔCt is the average Ct value of the test reaction minus the average Ct value of the reference reaction.

Statistical analyses

Data were analyzed using the Prism Graph Pad Software (CA, USA). All values are presented as mean±SEM. Significances were calculated using analysis of variance(ANOVA), followed by the Bonferroni post hoc analysis with multiple comparisons. The values of p<0.05 were considered significant.

Glucose tolerance, plasma insulin levels, body weight and food intake

HFD mice exhibited increased fasting glucose levels and impaired glucose tolerance (Figure 1A), in parallel with increased plasma levels of insulin (Figure 1B). E2 had no effects on glucose tolerance and insulin levels in LFD+E2 mice but suppressed both glucose and insulin levels in HFD+E2 mice to levels comparable with those in LFD mice (Figure 1A and B).

Figure 1:

Intraperitoneal glucose tolerance test (IPGTT), plasma insulin levels, body weight and food intake in LFD and HFD mice treated with E2 or vehicle.

(A) Levels of blood glucose were determined before and after administration of a glucose load (2 g/kg body wt ip) at the indicated time points. †††, p-value<0.001 for HFD vs. LFD and ***, p-value<0.001 for HFD+E2 vs. HFD. (B) Plasma insulin levels. ***, p-value<0.001 for HFD vs. LFD and **, p-value<0.01 for HFD+E2 vs. HFD. (C) and (D) Body weight and food intake, respectively, were measured before and at 48-h intervals during treatment of mice that were housed in pairs. Body weight in HFD mice is significantly higher than that of LFD mice at all time points. †††, p-value<0.001 for HFD vs. LFD while *, p-value<0.05, **, p-value<0.01 and ***, p-value<0.001 for HFD+E2 vs. HFD mice; n=6–8 mice in (A) and (C), and 3–4 mice in (B) and (D).

Compared with LFD mice, maintenance of mice on an HFD for 10 months led to significant increases in body weight (Figure 1C). E2 treatment resulted in reductions in the body weight of HFD+E2 mice from day 10 of treatment, compared to HFD mice, and by the end of E2 treatment, body weights in HFD+E2 mice were 74% of those of mice in the HFD group. In LFD+E2 mice, E2 treatment did not affect either body weight (Figure 1C) or food intake (Figure 1D). However, body weight changes in HFD+E2 mice were accompanied by decreases in food intake from 4 days of treatment; although these reductions were not significantly different at all time points (Figure 1D).

Body composition

As the effects of E2 on body weight may reflect alterations in adiposity, the volumes of the subcutaneous and visceral fat depots were quantified by MRI. LFD+E2 mice were not included in this study, as body weight was unchanged in this group (Figure 1C). E2 treatment led to a non-significant reduction in subcutaneous adipose volume (Figure 2A, B). However, the volume of the visceral adipose depot was significantly increased in HFD mice, and in HFD+E2 mice, the volume of this depot decreased to 67% of that of the HFD group (Figure 2A, C).

Figure 2:

Magnetic resonance imaging (MRI) and visceral adipocyte morphology in LFD mice, and HFD mice treated with E2 or vehicle.

(A) Representative axial MRI images of the abdominal region in LFD mice, and in the vehicle- and E2-treated HFD mice. Arrows indicate visceral (VF) and subcutaneous (Sc) fat depots. (B) and (C) Volumes of the subcutaneous and visceral adipose depots, respectively, presented as the percentage of total body volume. (D) Hematoxylin and eosin-stained sections were prepared from dissected visceral adipose of vehicle- or E2-treated LFD and HFD mice; representative images from each group are shown at 10× and 40×. (E) Adipocyte area was calculated from three hematoxylin and eosin-stained sections per group. ***, p-value<0.001; n=6 mice in (B) and (C), and n=3 mice in (D) and (E).

Visceral adipocyte morphology

As the reductions in the visceral adipose volume in HFD+E2 mice correlated with body weight reduction, we studied the effects of E2 in this depot. Changes in adipose tissue volume may be due to alteration in adipocyte size, and morphological analyses showed that visceral adipocyte size in HFD mice was 2.2 times greater than that of LFD mice (Figure 2D, E). In HFD+E2 mice, adipocyte size was reduced 2-fold by E2 treatment (Figure 2D, E), suggesting that E2 treatment in HFD+E2 mice reduced body weight and central adiposity in association with a hypotrophic effect on the visceral adipocyte population.

Gene expression and protein levels in visceral adipose tissue

As adipocyte hypertrophy is associated with an increase in the ratio of lipid deposition to lipolysis, the expression of genes involved in lipogenesis, fatty acid uptake and lipolysis was studied in the visceral adipose. Transcript levels for the lipogenic genes scd1 and fatty acid synthase (fas) were increased significantly and non-significantly, respectively, in HFD mice, and E2 administration in HFD+E2 mice reduced expression levels of both genes (Figure 3A). Conversely, E2 treatment in the HFD+E2 group increased expression of the lipolytic gene adipose triglyceride lipase (pnpla2). Thus reduced visceral adipocyte hypertrophy in HFD+E2 mice was associated with reciprocal alterations in the expression of genes involved in fatty acid synthesis and triglyceride mobilization. In the LFD+E2 group, E2 treatment had no significant effect on any of these genes.

Figure 3:

Expression of lipogenic genes, transcription factors, adipokines and adipokine receptors in visceral adipose from LFD and HFD mice treated with E2 or vehicle.

Total RNA was prepared from visceral adipose tissue dissected from vehicle- and E2-treated LFD and HFD mice and relative gene expression levels of lipogenic genes, transcription factors, and adipokines and adipokine receptors were assessed by real-time PCR (A, B and D, respectively). (C) Nr2c2/TR4 protein levels were assessed by Western blot. *, p-value<0.05, **, p-value<0.01 and ***, p-value<0.001; n=6 mice in (A), (B) and (D), and n=3 mice in (C).

To understand how E2 altered lipogenic gene expression, the expression of transcription factors involved in its control was investigated. mRNA levels of neither srebp1, which regulates the expression of scd1, fas and acetyl-CoA carboxylase (acc1), nor of lxrα, which controls srebp1 expression, were changed between the HFD and HFD+E2 groups (Figure 3B). Similarly, the expression levels of proliferator-activated receptor alpha (pparα), a regulator of fatty acid oxidation, and of pparγ, which activates genes involved in lipid uptake and adipogenesis, were unchanged in these groups. However, transcript levels of the nuclear receptor nr2c2/tr4 were reduced in HFD+E2 mice (Figure 3B). Consistent with this, Nr2c2/TR4 protein levels were significantly reduced in the HFD+E2 group, compared with the HFD group, while there were no differences in Nr2c2/TR4 protein levels between the LFD and LFD+E2 groups (Figure 3C).

Studies of adipokine expression levels showed that leptin and adiponectin were significantly upregulated and non-significantly reduced, respectively, in HFD mice (Figure 3D). LFD+E2 mice exhibited increased leptin expression levels. Both leptin and resistin mRNAs were reduced in HFD+E2 mice, whilst adiponectin showed a tendency to increase (Figure 3D). Leptin receptor expression was reduced in HFD mice but increased in HFD+E2 mice.

Although the expression of stat3, which was identified previously as a direct target gene of E2, was unchanged in LFD+E2 mice, its expression was increased in HFD+E2 mice (Figure 4A). In addition to its role in adipokine signaling, Stat3 is also involved in regulation of the differentiation of brown adipose tissue [19], and so we examined the expression of the mitochondrial ucp-1. Expression of ucp-1 was increased in HFD mice but was increased further in HFD+E2 mice (Figure 4A). Consistent with these observations, Western blotting showed that UCP-1 protein levels were significantly increased in HFD+E2 mice, compared with HFD mice (Figure 4B). Additionally, immunohistochemical staining for UCP-1 demonstrated an increased level of UCP-1 protein expression in adipose sections from HFD+E2 mice (Figure 4C). Furthermore, the expression of a second thermogenic gene, type II iodothyronine deiodinase (dio-2), was increased in HFD+E2 mice (Figure 4D). Despite this, the levels of mRNAs for peroxisome proliferator-activated receptor gamma coactivator 1-alpha (pgc-1α), a known driver of ucp-1 expression, and of cell death-inducing DNA fragmentation factor (cidea) and cytochrome c oxidase subunit VIIIb (cox-8b), markers of brown adipocytes, were not altered. However, expression levels of adrβ3, which regulates ucp-1 expression during the thermogenic response [24], were increased in HFD+E2 mice (Figure 4D). Expression levels of carnitine palmitoyltransferase 1b (cpt-1b), a brown adipose-specific gene involved in fatty acid utilization, were also increased in HFD+E2 mice.

Figure 4:

Expression of genes involved in the browning of visceral adipose in LFD and HFD mice treated with E2 or vehicle.

Total RNA was prepared from visceral adipose tissue dissected from vehicle- and E2-treated LFD and HFD mice, and real-time PCR was used to assess the relative expression levels of Stat 3 and UCP-1 (A) and brown adipose-specific genes (D). UCP-1 protein levels were assessed by Western blot (B) Immunohistochemical staining of visceral adipose sections for UCP-1 at 10× and 40× (C). *, p-value<0.05, **, p-value<0.01 and ***, p-value<0.001; n=6 mice in (A) and (D), and n=3 mice in (B) and (C).

Methylation status of SCD1, FAS, CPT-1b, DIO-2 and ADRβ3

To assess whether E2-induced expression changes were associated with epigenetic alterations in those genes, we assessed changes in methylation status at specific sites in several genes which were regulated by E2 in HFD mice, but not in LFD mice. There were no significant changes in the extent of methylation between the different groups with regard to the lipogenic genes scd1 (Figure 5A) and fas (Figure 5B), nor for the brown adipose-specific gene cpt-1b (Figure 5C). However, HFD feeding resulted in increases in the extent of methylation of adrβ3, compared to LFD, whilst E2 decreased methylation levels significantly in HFD+E2 mice to levels comparable to those of the LFD and LFD+E2 groups (Figure 5D). Finally, E2 treatment in the HFD+E2 group significantly suppressed methylation levels in the dio-2 gene, compared to HFD mice (Figure 5E).

Figure 5:

Methylation analyses of lipogenic and brown adipose-specific genes subject to differential regulation in visceral adipose tissue extracted from LFD and HFD mice treated with E2 or vehicle.

Analyses of methylation were carried out using an OneStep qMethyl Kit followed by real-time PCR, and the extent of methylation at each target sequence in the test reaction was expressed as a percentage of that target in the reference reaction; SCD1 (A), FAS (B), CPT-1b (C), ARβ3 (D) and DIO-2 (E), ***, p-value<0.001; n=5–6 mice.