The Endocrine Disrupting Chemical Tolylfluanid Alters Adipocyte Metabolism via Glucocorticoid Receptor Activation

Abstract

Glucocorticoid signaling plays a critical role in regulating energy metabolism. Emerging data implicate environmental endocrine-disrupting chemicals as contributors to the obesity and diabetes epidemics. Previous studies have shown that the phenylsulfamide fungicide tolylfluanid (TF) augments glucocorticoid receptor (GR)-dependent luciferase expression in 3T3-L1 preadipocytes while modulating insulin action in primary murine and human adipocytes. Studies were performed to interrogate glucocorticoid signaling in primary adipocytes exposed to TF. TF mimicked the gene transcription profile of the murine glucocorticoid corticosterone (Cort). Cellular fractionation assays demonstrated that TF treatment promoted the activating serine phosphorylation of GR, augmenting its cytoplasmic-to-nuclear translocation as well as its enrichment at glucocorticoid response elements on the glucocorticoid-induced leucine zipper gene promoter. After acute treatment, Cort or TF promoted insulin receptor substrate-1 (IRS-1) gene and protein expression. Either treatment also enriched GR binding at an identified glucocorticoid response element in the IRS-1 gene. TF or Cort each increased insulin-stimulated lipogenesis, an effect resulting from increased lipogenic gene expression and enhanced insulin-stimulated dephosphorylation of acetyl-coenzyme A carboxylase. The augmentation of insulin-stimulated lipogenesis was mediated through a specific enhancement of Akt phosphorylation at T308. These findings support modulation of IRS-1 levels as a mechanism for glucocorticoid-mediated changes in insulin action in primary adipocytes. Albeit with less affinity than Cort, in silico analysis suggests that TF can interact with the ligand binding pocket of GR. Collectively, these studies identify TF as a structurally unique environmental glucocorticoid. Glucocorticoid signaling may thus represent a novel pathway by which environmental toxicants promote the development of metabolic diseases.

The glucocorticoid receptor (GR) is a ubiquitously expressed nuclear transcription factor that controls widespread physiological processes, including metabolism, growth and development, and immune system activation. Interestingly, many of these functions are cell-, tissue-, and even context-dependent. In the adipocyte, the GR is necessary for adipocyte differentiation, plays a role in adipose insulin sensitivity, and functions as a regulator of lipid metabolism (1, 2). Excess glucocorticoids have long been known to exert adverse metabolic effects, as evidenced by the obesity, insulin resistance, and diabetes seen in patients with Cushing's syndrome (3). Based on its diverse functions, the GR is an important regulator of the metabolic state in adipocytes, and dysregulation of glucocorticoid action can lead to severe metabolic disruptions.

It is well established that obesity and diabetes rates have increased exponentially in the United States over the last several decades (4, 5). Increased consumption of a calorically-dense, Western diet and decreased physical activity are clearly the major contributors to the pathogenesis of metabolic diseases. However, these factors alone fail to fully account for the magnitude and rapidity of these epidemics. It has recently been proposed that environmental pollutants acting as endocrine-disrupting chemicals (EDCs) may disturb hormonal axes regulating fat mass and energy utilization, thereby promoting the accretion of body fat and disrupting energy homeostasis (6–8). Indeed, the increase in diabetes rates closely parallels the rise in synthetic chemical production in the United States, and an increasing body of epidemiological studies has linked EDC exposure with the development of diabetes (9). Although provocative, many of these studies fail to provide insight into the cellular and molecular mechanisms by which these environmental pollutants exert their effects. Because glucocorticoids play a central role in energy homeostasis, disruption of glucocorticoid signaling is a potential mechanism of environmentally-mediated metabolic disruption.

The GR belongs to the superfamily of ligand-activated nuclear hormone receptors, the classical members of which bind hydrophobic ligands and control gene expression through their functions as transcription factors. Studies have shown that a structurally diverse array of hydrophobic EDCs can mimic or inhibit the action of certain nuclear receptor agonists and thereby disrupt endocrine signaling (7, 10). Most studies to date have focused on disruptions in sex steroid and thyroid hormone signaling (11–13). However, there has been increasing recognition that environmental toxicants may disrupt energy regulation and contribute to the pathogenesis of metabolic diseases (14). In fact, recent studies have shown the ability of certain EDCs to decrease systemic insulin sensitivity (15), increase adipogenesis (7, 10, 16), and modulate lipid metabolism (17). In many cases, however, mechanistic insights into the causes of EDC-mediated metabolic disturbances are lacking.

The adipocyte is an important model for studying endocrine disruption due to its central role in systemic energy metabolism and its ability to bioaccumulate hydrophobic EDCs. Impairments in adipocyte function have been shown to have far-reaching repercussions on metabolic health and have been implicated in the development of type 2 diabetes (18). Several members of the superfamily of ligand-activated nuclear hormone receptors are vital to the processes that govern normal adipocyte function, including adipogenesis, insulin sensitivity, and lipid metabolism. The hydrophobicity of the lipid droplet also permits the bioaccumulation of lipophilic molecules, including potential EDCs. This can lead to high local concentrations in the fat pad while also prolonging exposure through slow leaching from the lipid droplet after acute exposures. Because of this local enrichment, understanding how environmental agents modulate the activity of ligand-activated nuclear receptors in the adipocyte is critical for determining their potential to contribute to the development of metabolic diseases.

Tolylfluanid (TF) is a phenylsulfamide used as a fungicide on fruit crops as well as a booster biocide in marine paints (19). Although tissue levels have not been reported, human exposure is likely to occur through contaminated food and water as well as occupationally among those using the compound. TF has been detected on fruit crops across Europe, where it is often one of the most commonly detected pesticides (20–23). Furthermore, TF has been shown to be resistant to removal by washing (20). Marine waters are contaminated with phenylsulfamides in the low nanomolar range through agricultural run-off and leaching from antifouling paints (24, 25). Importantly, use of phenylsulfamides in the antifouling industry is predicted to rise, because they are increasingly used as substitutes for banned organotin compounds (25). Finally, significant exposure can occur occupationally to agricultural and shipyard workers applying TF (26, 27). Although the bioaccumulation of TF has not been determined, it is highly hydrophobic (19), strongly suggesting the capacity to concentrate in lipid-rich tissues such as adipose. Previous studies have demonstrated the ability of TF to activate GR-dependent luciferase expression in 3T3-L1 preadipocytes while also promoting their differentiation into fully mature adipocytes (10). Additionally, TF has been shown to displace radiolabeled glucocorticoid from the GR in a cell-free assay (28), suggesting that TF has the capacity to modulate glucocorticoid signaling. Because glucocorticoids induce adipocytic insulin resistance and because binding to the GR could lead to context-dependent agonist activity, the current study investigated the ability of TF to stimulate GR signaling in primary murine adipocytes.

Materials and Methods

Chemicals and reagents

TF was purchased from Fluka (St Louis, Missouri). Type I collagenase was purchased from Worthington (Lakewood, New jersey). BSA, anti-β-actin antibody, and all inorganic chemicals were purchased from Sigma (St Louis, Missouri). Fetal bovine serum and phenol red-free DMEM/High Modified was obtained from HyClone (Logan, Utah). PureProteome protein A magnetic beads, the Magna ChIP G kit, and antiphosphotyrosine antibody were purchased from Millipore (Billerica, Massachusetts). E.N.Z.A. Total RNA kit II was obtained from Omega Bio-Tek (Norcross, Georgia). The qScript cDNA Synthesis kit and PerfeCTa SYBR Green FastMix were purchased from Quanta BioSciences (Gaithersburg, Maryland). Anti-pS473-Akt, anti-pT308-Akt, anti-Akt, anti-insulin receptor substrate-1 (IRS-1), anti-pS79-acetyl-coenzyme A (CoA) carboxylase (ACC), anti-ACC, anti-TATA box binding protein (TBP), anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and anti-phospho-GR (human S211, murine S220) monoclonal antibodies were purchased from Cell Signaling Technology (Danvers, Massachusetts). Anti-GR antibody was purchased from Santa Cruz Biotechnology, Inc (Santa Cruz, California). Radiolabeled ¹⁴C-glucose was purchased from American Radiolabeled Chemicals (St Louis, Missouri).

Quantitative real-time PCR

RNA was isolated from cultured primary adipose tissue using the E.Z.N.A. Total RNA kit II (Omega). The purity and concentration of the isolated RNA was assessed using a Nanodrop 2000; 260/280 ratios were approximately 2.0. The Quanta Biosciences qScript kit was used to synthesize cDNA. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using SYBR Green on a Bio-Rad MyiQ RT-PCR detection system (Bio-Rad, Hercules, California). Primers were from Integrated DNA Technologies (Coralville, Iowa) and were taken from the literature or databases as cited or designed using Primer3 (http://frodo.wi.mit.edu/) (Table 1). Primer specificity was assessed by melt curve analysis. Gene expression levels were evaluated by the ΔΔ threshold cycle (Ct) method after confirmation that amplification efficiency was between 90% and 110% for all primer pairs; 18S rRNA was used as a reference gene to control for total mRNA recovery.

Table 1.

Quantitative RT-PCR Primers

Gene	Forward primer	Reverse primer	Reference
IRS-1	GCCAGAGGATCGTCAATAGC	GAGGAAGACGTGAGGTCCTG	34
GILZ	TGACTGCAACGCCAAAGC	CTGATACATTTCGGTGTTCATGGTT	53
IRS-1 GRE	TGGCTCTCTACACCCGAGAC	ACCCGTGTCATAGCTCAAGTC
GILZ GRE	GAGCCCTTGAGAAACCAGTG	AGCTCTGGCAGAAAACGAAG
ACC	TACTGCCATCCCATGTGC	GCTTCCAGGAGCAGTCGT	54
SCD-1	TTCTTGCGATACACTCTGGTGC	CGGGATTGAATGTTCTTGTCGT	55
HK2	CTCCGGATGGGACAGAAC	TCGGCAATGTGGTCAAAC	54
FAS	GGCTCTATGGATTACCCAAGC	CCAGTGTTCGTTCCTCGGA
18S RNA	CGGCTACCACATCCAAGGA	GCTGGAATTACCGCGGCT	56
TNFα	CATCTTCTCAAAATTCGAGTGACAA	TGGGAGTAGACAAGGTACAACCC	57
Lipe	CACACCTACTACACAAATCC	GGCATAGTAGGCCATAGCA	37
Lipin-1	CGCCAAAGAATAACCTGGAA	TGAAGACTCGCTGTGAATGG	37
GyK	AGCCTCTCTATAATGCCGTGG	TGCACTGAAATACGTGCTAAGT
AdipoQ	GTTCTACTGCAACATTCCGG	TACACCTGGAGCCAGACTTG

AdipoQ, adiponectin; GyK, glycerol kinase; Lipe, hormone-sensitive lipase.

Adipose organ culture

Primary adipose tissue was obtained from male C57BL/6 mice (∼8 wk of age) killed humanely according to Institutional Animal Care and Use Committee-approved protocols. Perigonadal fat pads were harvested by sterile dissection with careful attention to removal of the testes and epididymis. The dissected fat was placed in phenol red-free DMEM/High Modified supplemented with 10% fetal bovine serum, penicillin-streptomycin, and L-glutamine; and the fat was coarsely minced into a pipettable slurry followed by washing 3 times with the same media. The adipose tissue was then equally apportioned into either vehicle-supplemented (absolute ethanol), 1nM to 1μM TF-supplemented, or 100pM to 10nM corticosterone (Cort)-supplemented DMEM and incubated for the indicated time at 37°C in 5% CO₂; the total ethanol concentration was less than or equal to 0.1%. Primary adipocytes were then isolated from fat pads by collagenase digestion and flotation centrifugation as previously described (29).

Insulin signaling assay

Perigonadal fat pads from 8-week-old C57BL/6 mice were treated with vehicle, 1nM Cort, or 100nM TF and cultured for 4 hours as outlined above. The adipose was then treated with 5nM insulin in culture for 30 minutes. After treatment, the adipose samples were put on ice, the insulin-containing medium was removed, and the samples were washed with ice-cold Krebs-Ringer bicarbonate HEPES buffer. Samples for immunoblotting were prepared by the addition of an equal volume of homogenization buffer (29) and sonication. Samples were then centrifuged at 10 000g at 4°C for 10 minutes, and the infranatant (between the pelleted cell debris and floating lipid layer) was removed. Laemmli 4× buffer (167mM Tris, 8mM EDTA, 27% glycerol, 1.3% β-mercaptoethanol, 416mM sodium dodecyl sulfate, and 0.3mM bromophenol blue) was added, and the samples were heated at 95°C for 5 minutes.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

All samples were resolved on 7% or 10% sodium dodecyl sulfate polyacrylamide gels and transferred to polyvinylidene fluoride after preconditioning the membranes with methanol. Western blottings were probed as described (30). Blots were then incubated with horseradish peroxidase-conjugated goat antirabbit or goat antimouse IgG (Bio-Rad) and developed using Amersham ECL Advance (GE Healthcare, Princeton, New Jersey). Densitometry was performed for immunoblots using ImageJ version 1.44 (National Institutes of Health, Bethesda, Maryland). Insulin effects were determined by calculating the ratio of the total areas for the bands corresponding to pS473-Akt or pT308-Akt to total Akt. ACC dephosphorylation was determined by calculating the ratio of the total areas for the bands corresponding to pS79-ACC to total ACC. GR phosphorylation was determined by calculating the ratio of the total areas for the bands corresponding to pS220-GR to total GR.

Immunoprecipitation (IP)

Protein lysates from cultured primary adipocytes were prepared as outlined above, and the infranatant layer removed after centrifugation was used for the IP. The PureProteome protein A magnetic beads were used to IP IRS-1 per the manufacturer's instructions. Briefly, antibodies were conjugated to magnetic beads by nutation at room temperature for 10 minutes followed by incubation of the protein lysate with the bead-antibody complex for an additional 10 minutes with nutation at room temperature. The supernatant was removed via magnetic separation and stored for immunoblotting. The protein-bead complex was then washed multiple times with PBS, and the isolated proteins were eluted via the addition of Laemmli buffer and heating at 95°C for 5 minutes. The IP samples underwent immunoblotting for IRS-1, phosphotyrosine, and Akt. Additionally, cultured adipose tissue underwent cellular fractionation as described below, and the isolated nuclear fraction was used to IP GR. The same IP protocol was used after the coupling of the protein A beads to the GR antibody. The IP samples underwent immunoblotting for GR and pS220-GR, and the nuclear lysate was then used to immunoblot TBP, GAPDH, and β-actin.

Cellular fractionation

Isolated adipocytes were prepared as described above and then suspended in nuclear lysis buffer (250mM sucrose, 20mM HEPES, 10mM KCl, 1.5mM MgCl₂, 1mM EDTA, 1mM EGTA, and the protease inhibitors benzamidine and aprotinin [added immediately before use]). The cells were then lightly homogenized with a handheld homogenizer and then centrifuged at 1000g for 10 minutes at 4°C to pellet the nuclei. The fat cake and supernatant were removed and then centrifuged at 10 000g for 10 minutes at 4°C to isolate the cytoplasmic fraction. The nuclear pellet was then resuspended in homogenization buffer and harshly sonicated followed by centrifugation of the nuclear lysate at 10 000g for 2 minutes to remove any lipid debris. The samples were then subjected to IP as outlined above, followed by addition of Laemmli buffer and heating for 5 minutes at 95°C. The proteins were separated by SDS-PAGE as described above and were immunoblotted for GR, pS220-GR, TBP, GAPDH, and β-actin.

Chromatin IP (ChIP)

ChIP of the GR was performed using the Magna ChIP protocol (Millipore). Briefly, cultured adipocytes were collagenase digested and then formaldehyde fixed. After quenching with glycine and washing, the cells were suspended in a cell lysis buffer and centrifuged at 1000g for 10 minutes at 4°C. The fat cake and cytoplasmic fraction were removed, and the nuclear pellet was washed and then resuspended in a nuclear lysis buffer. The cells were then harshly sonicated to shear the chromatin and centrifuged at 12 000g for 10 minutes at 4°C. The suspension was aliquoted and incubated overnight at 4°C with protein G magnetic beads and the appropriate antibody: nonspecific mouse IgG or anti-GR. The chromatin-complexed beads were washed, diluted with an elution buffer, and incubated with proteinase K for 2 hours at 62°C followed by an additional 10 minutes at 95°C. The DNA was then separated from the beads by magnetic separation followed by purification using column chromatography. GR enrichment at the IRS-1 and glucocorticoid-induced leucine zipper (GILZ) glucocorticoid response elements (GREs) was assessed via qRT-PCR and reported as percent input; nonspecific DNA binding was assessed using the GILZ and IRS-1 qRT-PCR primers, which amplify DNA regions outside of the known GREs (Table 1).

Lipogenesis assay

Isolated adipocytes were prepared as described above, and 20–30 μL of packed cells were apportioned into 7-mL scintillation vials. Krebs-Ringer bicarbonate HEPES buffer containing 1% BSA, 5mM glucose, and 1 μCi of [¹⁴C]-glucose ± 5nM insulin was added to the vials, and the cells were incubated at 37°C under gentle agitation for 30 minutes. The addition of ice-cold PBS to a final volume of 1 mL was used to stop the reaction. Lipids were then extracted overnight with the addition of 4 mL of Betafluor (National Diagnostics, Atlanta, Georgia). Incorporation of radiolabeled glucose into lipid was estimated by liquid scintillation counting of the organic phase of the Betafluor extract.

In silico modeling

Binding interactions were determined for both TF and Cort with Autodock Vina (31) using the published structure of the ligand binding domain of the murine GR (32) with the assistance of Dr David A. Ostrov, PhD (University of Florida College of Medicine, Gainesville, Florida). Figures were generated with PyMOL (Schrödinger, Cambridge, Massachusetts).

Statistical analyses

Significance was determined using 2-tailed Student's t test for comparisons between 2 conditions and by ANOVA for comparisons of more than 2 conditions. To test for differences in insulin-stimulated responses among the treatment groups, 2-way ANOVA was performed. The Holm-Sidak post hoc test was used to correct for multiple comparisons. Significance represents differences relative to vehicle treatment except where indicated by a horizontal bar. The relationship between TF- and Cort-induced gene expression was determined by linear regression by the least squares method. All statistical analyzes were performed using GraphPad Prism version 6 (GraphPad, La Jolla, California).

Results

TF- and Cort-induced GR activation

Glucocorticoids are known to affect insulin sensitivity and various metabolic processes in multiple tissues, including adipose. Therefore, the effect of TF on GR activation was investigated using expression of the GR target gene GILZ as a physiological readout. The GILZ promoter contains multiple GREs, and GILZ gene expression is a sensitive measure of glucocorticoid activity (33). Perigonadal fat pads from 8-week-old male C57BL/6 mice were sterilely removed, coarsely minced, washed, and then cultured with vehicle (ethanol) or 100nM TF for the indicated times. This dose of TF has previously been shown to have robust effects in primary adipose tissue without evidence of toxicity (34). At each time point, expression of GILZ was measured by qRT-PCR. After a 1- to 6-hour exposure, TF induced a significant 48%–81% increase in GILZ expression (Figure 1A). To compare the effects of TF with that of the endogenous murine glucocorticoid Cort, the dose-response relationship of Cort on GILZ expression was analyzed at a range of doses from 100pM to 10nM. A Cort dose of 1nM increased GILZ expression by 79% after a 4-hour treatment, a comparable response with that induced by 100nM or 1μM TF at 4 hours (Figure 1, B and C). Next, the time dependence of Cort effects on GILZ expression was determined at 1nM. Cort increased GILZ expression after a 1- to 2-hour exposure (218%–237% induction) (Figure 1D), a pattern of expression similar to that observed with TF treatment. To ascertain whether TF and Cort similarly altered transcription of other genes, expression of 5 additional genes with varying responsiveness to GR activation was analyzed. At equal levels of potency for GILZ induction, 4 hours of treatment with Cort (1nM) and TF (100nM) resulted in synchronous transcriptional effects (Figure 1E).

Figure 1.

Effect of TF and Cort on GILZ gene expression. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), Cort (100pM to 10nM), or TF (1nM to 1μM) for the indicated times. After TF treatment, RNA was isolated, and expression of GILZ was determined. The time dependence of TF-induced GILZ expression was determined (A). The concentration of Cort that induced the same gene expression as 100nM TF was determined (B). The effect of various concentrations of TF on GILZ expression is shown (C). The time dependence of Cort on GILZ expression is also shown (D). Additional genes were analyzed for expression after a 4-hour treatment with Cort (1nM, black bars) or TF (100nM, gray bars) relative to vehicle (dashed line) (E). These genes included GILZ, tumor necrosis factor-α (TNFα), hormone-sensitive lipase (Lipe), Lipin-1, glycerol kinase (GyK), and adiponectin (AdipoQ). Data are presented as mean ± SEM of the fold of GILZ relative to 18S rRNA with data normalized to vehicle at each time point (n = 3). Statistical significance was determined by 1-way ANOVA with Holm-Sidak post hoc testing to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01, ***P < .001, ****P < .0001, n.s. (not significant).

Previously, TF has been shown to modulate insulin signal transduction through a specific modulation of the insulin signaling intermediate IRS-1, an effect that was partially mediated by a reduction in IRS-1 gene transcription (34). To ascertain whether these effects were mediated by activation of GR signaling, the temporal effects of Cort and TF were examined on IRS-1 gene expression. Acute TF treatment increased IRS-1 expression after a 1- to 2-hour treatment (85% augmentation) (Figure 2A). Cort exposure resulted in a similar pattern of IRS-1 expression with an acute increase after a 2-hour exposure (155% induction) (Figure 2B). At later time points (10–12 h), Cort decreased IRS-1 levels, although the effect did not reach statistical significance. This is similar to previous studies showing that treatment with the synthetic glucocorticoid dexamethasone for 24 hours decreased IRS-1 transcript levels in 3T3-L1 murine adipocytes (35). As previously reported, TF treatment of at least 24 hours results in a reduction in IRS-1 gene expression (Figure 2C) (34). Thus, over the complete 48-hour range of exposure time, both the endogenous and putative environmental glucocorticoid exert a similar biphasic influence on a critical insulin signaling intermediate, thereby demonstrating the potential capacity for glucocorticoids to differentially modulate insulin action after acute and chronic exposures.

Figure 2.

Effect of TF and Cort on IRS-1 gene expression. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for the indicated times. After TF and Cort treatment, RNA was isolated, and expression of IRS-1 was determined (A and B, respectively). (C) Full time course of TF effects on IRS-1 expression. Data are presented as mean ± SEM of the fold of IRS-1 relative to 18S rRNA with data normalized to vehicle at each time point (n = 3). Statistical significance was determined by 1-way ANOVA with Holm-Sidak post hoc testing to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01, ***P < .001, ****P < .0001.

TF induction of GR nuclear translocation

Like other nuclear receptors, GR binds its ligand in the cytoplasm and then translocates into the nucleus, where it binds regulatory elements and alters gene expression. This process results from ligand-induced conformational changes in the protein, including those mediated by posttranslational modifications, such as phosphorylation. One such event is the ligand-dependent activating phosphorylation of serine 211 (S220 in mice), which promotes translocation into the nucleus (36). To assess the ability of TF to trigger GR activity, the effect of TF on GR phosphorylation and nuclear translocation was investigated after a 2-hour exposure to either vehicle, 100nM TF, or 1nM Cort; these concentrations were chosen because they resulted in equal induction of GILZ expression (Figure 1B). Analyses were conducted by immunoblotting after IP of the GR in each subcellular fraction with the fractional purity confirmed by the presence of GAPDH (cytoplasmic marker) or TBP (nuclear marker) (Figure 3A). Cort and TF decreased cytoplasmic GR compared with vehicle by 47% and 40%, respectively, whereas nuclear GR was simultaneously increased by 207% and 131%, respectively (Figure 3B, left and middle panels, respectively). Compared with vehicle, treatment with Cort and TF increased the presence of pS220-GR in the nucleus by 87% and 119%, respectively (Figure 3B, right panel). The role of S220 phosphorylation in promoting GR nuclear translocation is evident by the vast majority of pS220-GR residing in the nuclear fraction. These data show the ability of TF to promote the activating phosphorylation and nuclear translocation of GR, mimicking the action of the endogenous murine glucocorticoid Cort.

Figure 3.

Effect of Cort and TF on GR phosphorylation and nuclear translocation. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for 2 hours before cellular fractionation. Isolated adipocytes were suspended in a cell lysis buffer, briefly homogenized, and then centrifuged at 1000g to extract the nuclei. After treatment and cellular fractionation, the GR was immunoprecipitated from each fraction. GR, pS220-GR, GAPDH, TBP, and β-actin were immunoblotted in the cytoplasmic and nuclear lysate fractions. GAPDH is used as a cytoplasmic marker, and TBP serves as a nuclear marker; β-actin is used as the total protein loading control. A representative immunoblot is shown (A). Normalized densitometry for cytoplasmic GR (cytoplasmic GR/cytoplasmic β-actin), nuclear GR (nuclear GR/nuclear β-actin), and pS220-GR (pS220-GR/[nuclear GR/nuclear β-actin]) is shown in B (n = 4). Data are presented as mean ± SEM normalized to vehicle-treated adipocytes. Statistical significance was determined by 1-way ANOVA with Holm-Sidak post hoc testing to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01.

TF- and Cort-mediated recruitment of GR to GREs

Upon agonist binding and nuclear translocation, GR binds to GREs in the promoters of glucocorticoid-responsive genes and facilitates the assembly of the transcriptional complex. The capacity of TF to promote GR binding to GREs in GILZ and IRS-1 was investigated using ChIP. The IRS-1 exonic GRE was previously identified as a putative glucocorticoid binding region in a GR ChIP-sequence study using 3T3-L1 adipocytes (37). Using this GR ChIP-sequence study to identify the putative binding sequence and verifying the putative transcription factor binding site using the prediction program PROMO (38), primers were created to analyze GR recruitment to the IRS-1 GRE using qRT-PCR. At concentrations that promoted equal GILZ induction, GR enrichment at the GILZ GRE was enhanced 2.1-fold with TF and 2.7-fold by Cort after 2 hours of treatment (Figure 4A). GR enrichment at the IRS-1 GRE was enhanced 2.0-fold with TF treatment and 3.5-fold with Cort treatment (Figure 4B). GR binding to these GREs was specific, because there was no significant binding of the GR to DNA regions outside of the response elements in GILZ or IRS-1 at baseline or with exposure to Cort or TF (data not shown). Similarly, there was no effect of Cort or TF on nonspecific IgG binding to the GILZ and IRS-1 GREs (Figure 4, A and B, respectively). These data show the ability of TF to act as a GR agonist and promote GR binding to GRE sequences on the glucocorticoid-responsive genes GILZ and IRS-1. Furthermore, binding of the GR to this putative GRE on the IRS-1 gene may mediate the transcriptional effects of both Cort and TF on this important insulin signaling intermediate.

Figure 4.

Cort- and TF-induced binding of GR to GREs. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for 2 hours followed by collagenase digestion. The adipocytes were fixed, washed, and then a cell lysis buffer was added. The adipocytes were centrifuged at 1000g after undergoing brief homogenization. The nuclear/chromatin pellet was harshly sonicated, and then the chromatin extract was washed and isolated. The samples were immunoprecipitated with anti-GR antibody overnight. The IP lysate was reverse cross-linked, the DNA was purified, and the extent of GR binding to DNA was determined by qRT-PCR using primers specific for the GREs of GILZ (A) and IRS-1 (B). Data are presented as mean ± SEM of the percent of input normalized to vehicle treatment (n = 4). Statistical significance was determined by 2-way ANOVA with Holm-Sidak post hoc testing to correct for multiple comparisons. Statistical significance is shown as: *P < .05, ****P < .0001.

Adipocytic insulin sensitivity after acute exposures to TF

The ability of glucocorticoids to acutely increase yet chronically decrease IRS-1 transcript levels has been previously described, but the physiological significance of this acute increase is not known (39–42). Studies were therefore undertaken to investigate the effects of both TF- and Cort-mediated increases in IRS-1 transcription on IRS-1 protein levels. Fat pads were treated with either TF or Cort at doses that gave equal GR activity (100nM and 1nM, respectively) for 2 and 4 hours. The samples were treated with 5nM insulin for 10 minutes, and IRS-1 protein levels were quantified by IP followed by immunoblotting. Importantly, the acute increase in IRS-1 transcription was translated to IRS-1 protein levels, with elevations demonstrated with Cort treatment at 2 and 4 hours and with TF treatment at 4 hours (Figure 5A). However, Cort and TF did not alter the fraction of IRS-1 that was tyrosine phosphorylated upon insulin stimulation at either time point (Figure 5B). Taken together, IRS-1 transcription and protein levels were elevated at acute time points after exposure to both TF and Cort.

Figure 5.

Effect of Cort and TF on IRS-1 protein and phosphorylation. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for 2 or 4 hours. IRS-1 was immunoprecipitated as described, and a whole cell lysate was prepared. IRS-1, phosphotyrosine IRS-1, and Akt protein levels were assessed by immunoblotting. IRS-1 protein levels (A) and IRS-1 phosphotyrosine levels after a 10-minute stimulation with 5nM insulin (B) in response to 2- and 4-hour TF and Cort exposure are shown normalized to Akt. Data are presented as mean ± SEM normalized to vehicle-treated adipocytes (n = 3). Statistical significance was determined by 1-way ANOVA with Holm-Sidak post hoc testing to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01.

TF augmentation of insulin-stimulated lipogenesis

One of the central functions of adipocytes in regulating global energy metabolism is the uptake and storage of lipid. Dysregulation of adipocyte lipid homeostasis can result in global insulin resistance and development of the metabolic syndrome (18). Glucocorticoid exposure can promote lipogenesis and the redistribution of adipose depots as is seen clinically in patients with Cushing's syndrome (3). For these reasons, the effect of TF on de novo lipogenesis was investigated. At concentrations that gave approximately equal induction of GILZ expression, Cort and TF both increased insulin-stimulated lipogenesis by a comparable 27% and 35%, respectively, whereas basal lipogenesis was unaffected with either treatment (Figure 6A). To investigate the mechanism underlying this increase in lipid synthesis in the adipocyte, the expression of key genes regulating glucose delivery and fatty acid metabolism was assessed after a 30-minute insulin stimulation. Expression of ACC, stearoyl-CoA desaturase (SCD), hexokinase 2 (HK2), and fatty acid synthase (FAS) was examined under the same culture conditions as those to measure insulin-stimulated lipogenesis. Cort and TF treatment similarly increased expression of ACC, SCD-1, and HK2 expression in primary adipocytes, whereas FAS expression was unchanged with either treatment (Figure 6B). These data show that TF mimics the lipogenic action of Cort by similarly up-regulating key genes regulating lipogenesis. Moreover, when gene expression is analyzed across the full panel of genes examined in the present study, TF and Cort effects are strongly correlated with a near linear relationship, suggesting they function similarly in adipose tissue (Figure 7).

Figure 6.

Effect of Cort and TF on insulin-stimulated lipogenesis. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for 4 hours. After collagenase digestion, the isolated adipocytes were incubated with ¹⁴C-glucose ± 5nM insulin for 30 minutes. The lipid was extracted overnight with Betafluor, and the amount of labeled glucose incorporated into lipid was assessed by liquid scintillation counting of the organic phase (A). A subset of the samples was used for RNA isolation for assessment of the expression of ACC, SCD-1, HK2, and FAS by qRT-PCR (B). Lipid synthesis data are presented as mean ± SEM normalized to vehicle-treated controls. Gene expression data are expressed as mean ± SEM of the fold of the respective gene relative to 18S rRNA with data normalized to vehicle (n = 3). Statistical significance for lipogenesis data were determined by 2-way ANOVA and for gene expression by 1-way ANOVA; Holm-Sidak post hoc testing was used to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01, not significant (n.s.).

Figure 7.

Correlation between TF- and Cort-induced gene expression. Gene expression changes induced by 1nM Cort were plotted against those induced by 100nM TF from data in Figures 1, 2, and 6 for the following genes: ACC, HK2, SCD, IRS-1, GILZ, lipin-1, glycerol kinase (GyK), hormone sensitive lipase (Lipe), FAS, TNFα, and adiponectin (AdipoQ). The relationship between the treatments was examined by linear regression analysis using the least squares method. Dashed lines represent 95% confidence intervals. Error bars were omitted for data clarity.

TF- and Cort-mediated augmentation of ACC dephosphorylation and Akt-selective phosphorylation

Although TF and Cort similarly augmented insulin-stimulated lipogenesis and coordinately increased expression of key lipogenic genes, lipogenesis is also tightly regulated by insulin at the protein level (43). For these reasons, we analyzed the posttranslational regulation of one of these proteins and the effect of insulin as well as TF and Cort on its activity. ACC catalyzes the conversion of acetyl-CoA into malonyl-CoA and is the rate-limiting enzyme in fatty acid synthesis. AMP-regulated kinase phosphorylates and inactivates ACC on S79, whereas insulin promotes ACC dephosphorylation and activation through stimulation of protein phosphatase 2 activity (44). Compared with vehicle, a 3.5-hour treatment with Cort or TF followed by a 30-minute treatment with 5nM insulin augmented insulin-stimulated dephosphorylation of ACC by 48% and 36%, respectively (Figure 8, A and B, left panel). To analyze this apparent insulin sensitization, the effect of Cort and TF on insulin-stimulated Akt phosphorylation was investigated. Interestingly, neither treatment affected S473 phosphorylation (Figure 8, A and B, middle panel). However, relative to vehicle treatment, T308 phosphorylation was significantly increased by 72% and 96% with Cort and TF treatment, respectively (Figure 8, A and B, right panel). These results show that the endogenous glucocorticoid, Cort, and environmental glucocorticoid, TF, selectively augment insulin-stimulated Akt phosphorylation and promote insulin-stimulated ACC dephosphorylation, leading to an increase in lipogenesis.

Figure 8.

Effect of Cort and TF on ACC and Akt phosphorylation. Perigonadal fat pads from C57BL/6 mice were cultured and treated with vehicle (Veh, ethanol), 1nM Cort, or 100nM TF for 4 hours. A subset of the samples was then treated with 5nM insulin for 30 minutes. Whole cell lysates were prepared and immunoblotted for ACC, pS79-ACC, Akt, pS473-Akt, and pT308-Akt. (A) Representative immunoblot. The ratio of insulin-stimulated to unstimulated ACC phosphorylation after Cort and TF treatment is shown in B (left). Insulin-stimulated and unstimulated Akt phosphorylation is also shown in B (pS473-Akt/Akt [middle] and pT308-Akt/Akt [right]). Data are presented as mean ± SEM normalized to vehicle-treated adipocytes (n = 4). Statistical significance for Akt phosphorylation was determined by 2-way ANOVA and for insulin suppression of ACC phosphorylation by 1-way ANOVA; Holm-Sidak post hoc testing was used to correct for multiple comparisons. Statistical significance is shown as: *P < .05, **P < .01, ***P < .001, not significant (n.s.).

In silico modeling of TF and Cort binding to the GR

To better understand the capacity for TF to stimulate glucocorticoid signaling, TF and Cort interactions with the ligand binding pocket of the murine GR (32) were analyzed using Autodock Vina (31). Based on this analysis, TF is able to intercalate into the ligand binding pocket and bind GR (Figure 9). The affinity of the receptor for TF was, however, less than that found for Cort (ΔG = −6.3 kcal/mol vs ΔG = −9.8 kcal/mol, respectively), and the model did not predict TF interactions with the 5 polar contacts identified between Cort and the GR. These findings are consistent with the physiological data suggesting that TF functions as an environmental glucocorticoid but with less potency than the endogenous GR agonist.

Figure 9.

Chemical structures of TF and Cort with in silico modeling of binding to the ligand binding domain of murine GR. TF (green) and Cort (magenta) were modeled into the murine GR ligand binding domain (cyan), PDB code 3MNP, using Autodock Vina. Black dashes represent polar contacts.

Discussion

Many provocative epidemiological studies in the field of endocrine disruption have identified intriguing links between environmental pollutants and metabolic diseases (9). However, in most cases, the precise mechanisms underlying these effects remain unknown. The study described here is among the first to demonstrate a molecular mechanism by which a novel metabolic disruptor can alter metabolic signaling, namely through the activation of the GR and changes in expression of key regulators of lipid metabolism. These results support previous work demonstrating the capacity of TF to function as a novel glucocorticoid agonist in 3T3-L1 preadipocytes (10), while also raising intriguing questions regarding the hormonal signaling pathways that may be disrupted by environmental toxicants.

Glucocorticoid signaling results from GR phosphorylation and nuclear translocation, binding of response elements, and modulation of gene expression. In the present studies, TF mimicked the action of the endogenous glucocorticoid Cort in each of these actions. Furthermore, glucocorticoid signal transduction is also regulated by a variety of factors influencing GR activity, including associations with protein complexes, posttranslational modifications, and conformational changes. Both TF and Cort were found to coordinately promote an important GR modification, namely the ligand-dependent phosphorylation of S220. Mutational studies have demonstrated that cyclin A/cyclin-dependent kinase 2 phosphorylates S211 (equivalent of murine S220) on the GR (36). Interestingly, the relative phosphorylation state of the GR at S203, S211, and S226 (corresponding to mouse S212, S220, and S234, respectively) has been shown to differentially regulate gene expression of glucocorticoid-responsive genes (36). This may explain why TF and Cort had no effect on FAS gene expression despite the demonstration of GREs in its promoter (37). Because structurally diverse EDCs are increasingly shown to adversely affect adipocyte function, the possibility that these compounds may modulate pathways regulating GR sensitivity and responsiveness will need to be investigated.

One of the most intriguing aspects of the present studies is the structural dissimilarity between TF and Cort despite their similar biological effects (Figure 9). This raises intriguing questions about the biophysical properties of TF that allow it to activate the GR. Cort binds in the ligand binding pocket of the GR, and studies have shown that TF can dislodge tritiated dexamethasone from the GR (28), suggesting that TF may bind similarly. In silico modeling of the GR ligand binding pocket revealed that TF has the capacity to bind the receptor (Figure 9). However, the affinity was less than that found for Cort, and TF was not predicted to engage in the 5 polar contacts between Cort and the GR. Aside from the ligand binding pocket, alternate modes of activation, such as allosteric regulation or modulation of cofactor association, could mediate TF-dependent GR activation as well. Likewise, TF could function as a glucocorticoid through modulation of the upstream kinases regulating GR phosphorylation, eg, cyclin-depdendent kinase 2. Further studies will be required to determine whether TF has the capacity to modulate GR activity indirectly through such mechanisms. Importantly, a lack of structural similarity to known nuclear hormone agonists and failures of in silico analyses to predict interactions has plagued the EDC field. Thus, directed studies such as those described here are necessary for identifying other putative environmental glucocorticoids.

The adipocyte is vital to systemic energy metabolism through its role in storing and mobilizing lipid. Dysregulation of lipid homeostasis can result in a variety of pathophysiological outcomes that have been shown to play key roles in the development of obesity and diabetes (18). Published work has shown that the synthetic glucocorticoid dexamethasone augments insulin-stimulated lipogenesis, an effect likely mediated by increased expression of important lipogenic genes, such as ACC and SCD-1 (45, 46). Finally, evidence suggests that glucocorticoids can also directly modulate the activity of lipogenic enzymes (45). In the present study, both TF and Cort promoted insulin-stimulated lipogenesis. Critical genes regulating lipid synthesis were induced, including ACC and SCD-1, and the insulin-stimulated activating dephosphorylation of the ACC protein was enhanced by both TF and Cort treatment. This result would be hypothesized to promote excessive adipocytic lipid accumulation and resultant cellular hypertrophy, potentially resulting in an alteration in metabolic responsiveness, disruption of adipokine secretion, and expansion of adipose mass.

The interactions between glucocorticoids and insulin are complex. Evidence suggests that glucocorticoid effects on lipid homeostasis are nutrient dependent, specifically increasing lipogenesis in the fed state while augmenting lipolysis in the fasted state (47, 48). The present ex vivo model likely mimics an endogenous fed state due to the presence of high glucose concentrations in culture (25mM) with simultaneous exposure to insulin (5nM) during assessments of lipogenesis. Interestingly, TF and Cort treatment only affected insulin-stimulated lipogenesis, with basal rates of lipogenesis remaining unchanged. Canonically, the insulin/IRS-1/phosphatidylinositol 3-kinase pathway leads to an increase in phosphatidylinositol 3-phosphate and promotes Akt recruitment to the cell membrane and activation of phosphatidylinositol-dependent kinase 1, the enzyme responsible for T308 phosphorylation of Akt. Therefore, the TF- and Cort-mediated increase in IRS-1 levels would predictably increase T308 phosphorylation. In contrast, S473 phosphorylation of Akt is thought to be mammalian target of rapamycin complex 2 dependent. However, the activating mechanism of mammalian target of rapamycin complex 2 remains unclear. In the present study, S473 phosphorylation was unchanged by treatment with the endogenous or environmental glucocorticoid. There are a few studies that have previously characterized the physiological effects of differential Akt phosphorylation, yet the effect of glucocorticoids on this process is currently unknown (49–51). Although substrate specificity has been shown to be altered with other enzymes (36), whether the differential phosphorylation states of Akt alter its target preference has not been examined. Based on the studies reported here, insulin-stimulated T308 phosphorylation of Akt appears sufficient for promoting adipocyte lipogenesis through the induction of both gene transcription as well as the activating dephosphorylation of the important lipogenic mediator, ACC.

The observed increase in IRS-1 levels reflects a significant induction of IRS-1 gene expression by TF and Cort treatment. By assessing GR binding to GRE sequences, the present studies showed that both TF and Cort not only promote GR enrichment at a GRE for GILZ but binding of the GR to a putative GRE in the IRS-1 gene as well. Previous studies have only identified the exonic IRS-1 GRE as a potential glucocorticoid binding region in the 3T3-L1 cell line (37). The present study demonstrated that TF and Cort can enrich GR at the IRS-1 GRE in primary adipocytes, providing evidence of a mechanism by which glucocorticoids can modulate IRS-1 gene expression. Acute TF exposure also led to an increase in IRS-1 levels. These studies complement previous work demonstrating a specific down-regulation in adipocytic IRS-1 gene and protein expression after more chronic exposure to TF (24–48 h) that resulted in a consequential increase in adipocytic insulin resistance (34). This biphasic response of IRS-1 expression was also observed with Cort, which acutely increased but chronically reduced IRS-1 mRNA levels. This more chronic effect of Cort on IRS-1 is similar to that induced by dexamethasone in murine 3T3-L1 adipocytes (35) and mimics the clinical presentation of Cushing's syndrome with its characteristic insulin resistance (3). Thus, these data demonstrate a time dependence of glucocorticoid action on the insulin signaling cascade. In the adipocyte, acute glucocorticoid exposure leads to GR activation with a resultant increase in insulin sensitivity, whereas more chronic exposure leads to prolonged activation of a GR-mediated transcriptional cascade that induces insulin resistance. Similar temporal effects have been observed in other model systems (39–42), but the mechanisms by which glucocorticoids create such divergent effects remain unresolved. Importantly, this time-dependent transition in the adipocyte state from insulin-sensitive to insulin-resistant is analogous to other nonmonotonic effects mediated by EDCs (52) and emphasizes the importance of carefully interrogating both the concentration- and time-dependent effects of these chemicals on cellular physiology.

One important limitation of the present study is that it is focused solely on effects in adipocytes. To explore the consequences of TF-mediated GR activation on whole-body energy homeostasis, future work will require in vivo models that account for potential effects on the multitude of tissues that regulate metabolism. The current study is also restricted to mechanistic investigations at 1 concentration of TF. Although in the nanomolar range, the relationship between this concentration and potential levels in humans remains unknown. There is limited human exposure data for TF, and like many EDCs, tissue levels in human populations have not been determined. Despite this lack of data, there are numerous reports that provide strong evidence that humans are likely exposed (20–22). Based on these and previous data, specific investigations examining the extent of exposure, tissue bioaccumulation, and consequential effects on global energy metabolism of this novel environmental glucocorticoid are warranted.