Dexamethasone-Mediated Changes in Adipose Triacylglycerol Metabolism Are Exaggerated, Not Diminished, in the Absence of a Functional GR Dimerization Domain

Abstract

The glucocorticoid (GC) receptor (GR) has multiple effector mechanisms, including dimerization-mediated transactivation of target genes via DNA binding and transcriptional repression mediated by protein-protein interactions. Much attention has been focused on developing selective GR modulators that would dissociate adverse effects from therapeutic anti-inflammatory effects. The GR^dim/dim mouse has a mutation in the dimerization domain of GR and has been shown to have attenuated transactivation with intact repression. To understand the role of GR dimerization-dependent targets in multiple tissues, we measured metabolic fluxes through several disease-relevant GC target pathways using heavy water labeling and mass spectrometry in wild-type and GR^dim/dim mice administered the potent GC dexamethasone (DEX). Absolute triglyceride synthesis was increased in both wild-type and GR^dim/dim mice by DEX in the inguinal and epididymal fat depots. GR^dim/dim mice showed an exaggerated response to DEX in both depots. De novo lipogenesis was also greatly increased in both depots in response to DEX in GR^dim/dim, but not wild-type mice. In contrast, the inhibitory effect of DEX on bone and skin collagen synthesis rates was greater in wild-type compared with GR^dim/dim mice. Wild-type mice were more sensitive to DEX-dependent decreases in insulin sensitivity than GR^dim/dim mice. Wild-type and GR^dim/dim mice were equally sensitive to DEX-dependent decreases in muscle protein synthesis. Chronic elevation of GCs in GR^dim/dim mice results in severe runting and lethality. In conclusion, some metabolic effects of GC treatment are exaggerated in adipose tissue of GR^dim/dim mice, suggesting that selective GR modulators based on dissociating GR transactivation from repression should be evaluated carefully.

Glucocorticoids (GCs) are potent therapies for a wide range of inflammatory disorders. However, their use is limited by adverse metabolic effects including osteoporosis, weight gain, insulin resistance, altered body fat distribution, hypertension, muscle atrophy, and skin atrophy, a constellation known as Cushing's syndrome (1, 2). GCs mediate these responses via binding the GC receptor (GR) (3, 4). The challenge in designing modern GCs as medicines is not in increasing their potency or specificity for GR but rather in designing compounds that evoke therapeutically desirable actions of GR while limiting adverse actions of GR, thus increasing therapeutic index (5–7).

GR is a member of the nuclear hormone receptor transcription factor superfamily and contains several modular domains, including a DNA-binding domain, ligand-binding domain, and activation domains 1 and 2 (AF1 and AF2) (4, 8). When a ligand binds to and activates GR, GR translocates to the nucleus. Once in the nucleus, activated GR can homodimerize and bind to a sequence of DNA, a classical GC response element (GRE), located in the proximal promoter, enhancer, or other regions of a gene (4). In addition, various transcriptional cofactors are recruited, and transcription of the target gene is either induced or repressed (4, 6). GR can modulate transcription of target genes by other mechanisms, such as binding to DNA as a monomer on a GRE half-site (4, 6), binding to DNA and to another transcription factor that is also binding the DNA at a composite element (4, 6), or binding via protein-protein interactions exclusively to another transcription factor (tethering) (4, 6). Many of the well-characterized anti-inflammatory mechanisms of GR, for example, involve binding with proinflammatory transcription factors, such as nuclear factor-κB and activator protein-1, via tethering or at a composite element, whereby activated GR binds to and inhibits their normal proinflammatory actions (6, 9–14).

Historically, GCs have been thought of as either pure agonists or pure antagonists (2). Recently, however, the concept of selective GR agonists or selective GC receptor modulators (SGRMs) has emerged (5–7, 15). The variable actions of known SGRMs on GR-dependent pathways are thought to be due to their ability to induce variations in cofactor recruitment or variable affinity of activated GR for different GRE sequences (7, 15). Because many of the potent anti-inflammatory actions of GR have been attributed to inhibitory events involving tethering of GR with proinflammatory transcription factors (16), a model has emerged depicting the anti-inflammatory actions of GCs as being mediated by GR repression, which are dependent largely on GR tethering with another transcription factor, and the adverse metabolic effects as being mediated by transactivation, which are dependent largely on GR-GR homodimerization and DNA binding at a classic GRE. Based on this model, much attention has been focused on the possibility of developing SGRMs that would retain anti-inflammatory properties with reduced metabolic adverse effects (7, 15, 17–20). Site-directed mutagenesis of the GR dimerization domain (within the DNA-binding domain) revealed mutants that abrogated GR-GR homodimerization-dependent transactivation of GRE-reporter constructs while leaving tethering-based repression intact (21). Homologous recombination has been used to knock-in this point mutation in mice (22). In contrast to GR^−/− mice that die at birth, GR^dim/dim mice are viable (22). In addition, many of the GR-dependent anti-inflammatory pathways of GR^dim/dim mice have been previously shown to be intact in these mice (23). GR^dim/dim mice have served not only as a genetic model for a novel class of potential SGRMs (7, 15, 17–20) but also to tease out mechanisms by which GR mediates its effects on target genes and metabolic pathways, in particular in pathways associated with the therapeutically undesirable effects of GR activation (22, 24–26).

In this study, we sought to determine the requirements for an intact GR dimerization domain in mediating the metabolic effects of GCs in multiple target pathways of GCs. We directly assessed the roles of GR dimerization-dependent targets in mediating several therapeutically undesirable metabolic effects of GCs by use of stable isotope labeling with gas chromatographic-mass spectrometric (GC-MS) methods. Kinetic measurements included bone and skin collagen synthesis, muscle protein synthesis, adipose tissue lipid dynamics, and insulin sensitivity in wild-type (WT) and GR^dim/dim mice administered vehicle or dexamethasone (DEX). Our goal was not only to understand molecular requirements for GR effects on GC-dependent pathways but also to further explore the GR^dim/dim mutation as a potential genetic model for SGRMs.

Materials and Methods

Animal husbandry and breeding

GR^dim/+ (GR-A458T backcrossed onto a C57BL/6 background) mice were a gift of Günther Schütz (German Cancer Research Center, Heidelberg, Germany). GR^dim/+ mice were rederived at University of California, San Francisco (UCSF), by embryo transfer. GR^dim/dim mice and GR^+/+ (WT) mice were housed in groups of no more than 5 under temperature-controlled conditions with a 12-hour light, 12-hour dark cycle. Mice were fed PicoLab Rodent Diet 20 (5053; PMI Nutrition International, St Louis, Missouri; 13% of calories from fat) ad libitum. All studies received prior approval from the Animal Care and Use Committee at UCSF and were performed in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals. GR^dim/dim mice did not survive to weaning at the expected Mendelian frequency. Intercrosses of GR^dim/+ mice resulted in approximately 5% GR^dim/dim mice with a ratio of GR^+/+ to GR^dim/+ that suggested dropout of GR^dim/dim mice. In addition, many dead pups were found at birth that were genotyped as GR^dim/dim. We therefore used GR^dim/dim male × GR^dim/+ female crosses to generate GR^dim/dim mice. Here, too, decreased viability of GR^dim/dim mice was seen with only ∼10% of progeny surviving to weaning from these crosses being GR^dim/dim. WT littermate controls were bred in parallel from our colony.

Dual-energy x-ray absorptiometry

Body composition analysis was performed using dual-energy x-ray absorptiometry (DEXA) (GE Lunar Piximus, Madison, Wisconsin). Mice were anesthetized with isoflurane during DEXA scanning. The head region was excluded from analysis using the region of interest feature of Piximus software.

Heavy water (²H₂O) labeling protocol

WT and GR^dim/dim mice (male, 2 or 4 months old) were labeled with ²H₂O to body water enrichments of approximately 5% for 7 days as described elsewhere (27).

DEX treatment protocol for ²H₂O-labeled mice

Mice were injected ip with vehicle or 10 mg/kg DEX every other day for 1 week. Specifically, DEX injections were performed on days 1, 3, and 5 of ²H₂O labeling.

Measurement of body ²H₂O, hydroxyproline, alanine, triglyceride-glycerol, and triglyceride-palmitate enrichments by GC-MS in heavy water labeling studies

GC-MS was performed using a 5890 gas chromatograph attached to either a 5971 or 5973 mass spectrometer (Hewlett Packard, Palo Alto, California; Agilent Technologies, Santa Clara, California). Water was distilled from plasma, and ²H enrichment was measured using GC-MS after conversion to tetrabromoethane, as described in detail elsewhere (28–30). l-4-(O-tert-butyldimethylsilyl)-hydroxyproline pentafluorobenzyl ester was prepared from bone and skin and analyzed by GC-MS as previously described (28, 31). Pentafluorobenzyl-N,N-di(pentafluorobenzyl)-alanine was prepared from skeletal muscle (quadriceps) and analyzed by GC-MS as previously described (28, 32). Triglyceride (TG)-glycerol and fatty acids (FAs) were isolated from adipose tissue and liver, TG-glycerol was derivatized to glycerol-triacetate and separated FA-methyl esters were analyzed by GC-MS as described previously (28, 30, 33–36). All compounds were resolved on a DB-225 gas chromatograph column (Agilent Technologies). Mass spectrometry was performed in negative chemical ionization mode for derivatized hydroxyproline and alanine, positive chemical ionization mode for glycerol-triacetate, and electron impact ionization mode for FA-methyl esters with helium as the carrier gas for all analytes. Mass-to-charge ratios corresponding to unlabeled, singly labeled, and doubly labeled (M₀, M₁, and M₂) mass isotopomers of each derivatized compound were analyzed by selected ion monitoring.

Calculation of fractional synthesis rates of tissue collagen, muscle proteins, TG-glycerol, and TG-palmitate

Fractional synthesis (f), or fraction newly synthesized during the 7-day labeling period was calculated for collagen (hydroxyproline), total protein (alanine), TG-palmitate, and total TG (TG-glycerol). For each analyte, the ratio of the measured excess M₁ isotopomer (EM₁) to the asymptotic value of EM₁ (EM_1,max) was calculated as previously described (28, 30–32, 36–38) using the following formula: f = EM_1(sample)/EM_1,max. EM_1(sample) is the enrichment of M₁ isotopomers in excess of that which occurs in nature, calculated for each analyte using the abundances of the M₀, M₁, and M₂ mass isotopomers measured in the sample and in an unenriched standard: EM_1(sample) = [M₁/(M₀ + M₁ + M₂)]_sample − [M₁/(M₀ + M₁ + M₂)]_standard. EM_1,max is calculated for each analyte in each animal based on the measured body ²H₂O enrichment, as described previously (28, 30–32, 36–38), and represents the calculated EM₁ of the fully turned-over analyte at the body ²H₂O enrichment measured in the animal.

Calculation of absolute TG-glycerol synthesis rates

Absolute TG synthesis rates in inguinal and epididymal fat depots were calculated as previously described (28, 30, 39). Briefly, absolute synthesis rates of adipose TG were calculated from fractional TG synthesis (f_TG) multiplied by adipose TG mass multiplied by the fraction of TG in adipose tissue by weight: absolute synthesis (grams per week) = f_TG × adipose TG mass (grams) × (0.9).

Calculation of absolute rates of de novo lipogenesis

Absolute rate of de novo lipogenesis (DNL) in inguinal and epididymal fat depots was calculated as previously described (28, 40) from fractional TG-palmitate synthesis (f_DNL), adipose TG mass, and the fraction of palmitate relative to other FAs in TG using the following formula: absolute DNL (grams per week) = f_DNL × adipose TG mass (grams) × fraction TG-palmitate. The fraction of TG-palmitate present in adipose TG was taken to be 20% (41, 42).

Ex vivo lipolysis

WT and GR^dim/dim mice were treated with vehicle or 10 mg/kg DEX every day for 4 days. Mice were killed, inguinal and epididymal adipose tissue was collected, and lipolysis was measured as previously described (43). We also measured lipolysis in adipose tissue explants incubated in 1μM isoproterenol as a control.

[²H]Glucose disposal test

The [²H]glucose disposal test (²H-GDT), a modified oral glucose tolerance test, was used to measure insulin-mediated glucose utilization (insulin resistance) and pancreatic β-cell compensation, as previously described (28, 44). At age 4 months, mice were subjected to the GDT to assess baseline (vehicle-treated) values. All mice were allowed to recover for 1 month, treated with DEX as above, and then again subjected to the GDT. Blood was taken at regular time points for blood glucose and plasma insulin concentrations as well as for analysis by isotope ratio infrared spectrometry to quantify ²H₂O and H₂¹⁸O enrichments. Insulin sensitivity index (SI) was calculated as the percentage of glucose load metabolized to H₂O divided by the product of insulin area under the curve (AUC) and glucose AUC, thus, whole-body glycolytic utilization of a glucose load per unit of ambient insulin exposure per unit of ambient glucose level.

Gene expression

A separate cohort of WT and GR^dim/dim mice were treated with either saline or DEX 10 mg/kg at 7:00 am. At 1:00 pm (6 hours later), mice were killed and adipose tissue was collected and frozen on liquid nitrogen. RNA was isolated using Trizol, cDNA was prepared using iScript (Bio-Rad, Hercules, California), and genomic DNA removed using Turbo deoxyribonuclease (Ambion, Austin, Texas). Quantitative PCR was performed on an ABI 7900 (Applied Biosystems, Foster City, California) and mRNA levels calculated using the ΔΔC_t method with normalization to the housekeeping gene cyclophilin. Intron-flanking gene-specific primer sequences were obtained at www.mouseprimerdepot.org.

Statistical analyses

Statistical analyses were performed using GraphPad Prism version (GraphPad Software Inc, La Jolla, California). In heavy water labeling, ex vivo lipolysis, and ²H-GDT studies, 2-way ANOVA was performed to assess statistically significant main effects of DEX treatment and genotype, as well as interaction effect of the 2 main variables of genotype and DEX treatment, on parameters measured. Bonferroni posttests were performed to assess statistically significant differences between vehicle-treated and DEX-treated groups for each genotype. P < .05 was considered to be significant. For DEXA analyses, t tests were used.

Results

GR^dim/dim mice exposed to DEX suppress collagen synthesis in skin and bone

We measured collagen synthesis rates in bone (femurs) and skin in vehicle- and DEX-treated WT and GR^dim/dim mice. DEX treatment did not result in significantly lower rates of collagen synthesis in either WT or GR^dim/dim mice compared with their respective vehicle controls (Figure 1A).

Figure 1.

Fractional synthesis rates (f) of collagen (hydroxyproline) in bone in 4-month-old and 2-month-old cohorts and in skin in a 4-month-old cohort of WT and GR^dim/dim (DIM) mice; n = 13 for each vehicle-treated 4-month-old group, n = 14 for each DEX-treated 4-month-old group, n = 6 for the vehicle-treated WT 2-month-old group, and n = 5 for the other 3 2-month-old groups. Data are shown as mean ± SD. *P < .05; **P < .01; ***P < .001 between vehicle- and DEX-treated groups within genotypes by 2-way ANOVA followed by Bonferroni posttests. A, Bone collagen fractional synthesis in 4-month-old mice. Interaction effect was not significant (P > .05, 2-way ANOVA). B, Bone collagen fractional synthesis in 2-month-old mice. WT mice were significantly more sensitive to DEX (interaction P < .01, 2-way ANOVA). C, Skin collagen fractional synthesis in 4-month-old mice. WT mice were significantly more sensitive to DEX (interaction P < .01, 2-way ANOVA).

Because the trend for reduced bone collagen synthesis rates was not statistically significant in 4-month-old mice, we measured bone collagen synthesis rates in a separate cohort of 2-month-old WT and GR^dim/dim mice, an age at which we have previously observed significant effects (28). In the femurs of 2-month-old mice, DEX treatment resulted in significantly lower rates of collagen synthesis in both WT (P < .001) and GR^dim/dim (P < .05) mice relative to their respective vehicle controls (Figure 1B). The inhibitory effect of DEX treatment on bone collagen synthesis rates was significantly greater (P < .01) in WT mice compared with GR^dim/dim mice.

In the dorsal skin of 4-month-old mice, DEX treatment resulted in significantly lower (P < .001) rates of collagen synthesis in both WT and GR^dim/dim mice compared with their respective vehicle controls (Figure 1C). The inhibitory effect of DEX on fractional skin collagen synthesis rates, specifically the absolute difference between vehicle- and DEX-treated values, was greater (P < .01) in WT mice compared with GR^dim/dim mice. However, within both WT and GR^dim/dim groups, DEX-treated mice had approximately 90% lower rates of skin collagen synthesis relative to their respective vehicle controls.

In summary, 1 week of DEX treatment significantly inhibited rates of bone collagen synthesis in 2-month-old, but not 4-month-old, WT and GR^dim/dim mice. DEX also significantly inhibited rates of skin collagen synthesis in both WT and GR^dim/dim mice. GR^dim/dim mice exhibited less inhibition of collagen synthesis than WT in response to DEX in both bone and skin.

GR^dim/dim mice are not protected from GC-induced suppression of skeletal muscle protein synthesis

Because GCs are known to decrease lean tissue mass in part by inhibiting muscle protein synthesis (28, 45–47), we measured rates of total protein synthesis in quadriceps muscle of vehicle and DEX-treated GR^dim/dim and WT mice. DEX treatment resulted in significantly lower (P < .001) rates of muscle protein synthesis in both WT and GR^dim/dim mice compared with their respective vehicle controls (Figure 2). Synthesis rates were 17% lower in WT and 26% lower in GR^dim/dim mice. Differences in sensitivity to DEX between genotypes were not significant (P > .05).

Figure 2.

Fractional synthesis rates (f) of muscle protein (alanine) in skeletal muscle (quadriceps) of WT and GR^dim/dim (DIM) mice; n = 13 for each vehicle-treated 4-month-old group, n = 14 for each DEX-treated 4-month-old group, n = 6 for the vehicle-treated WT 2-month-old group, and n = 5 for the other 3 2-month-old groups. Data are shown as mean ± SD. *P < .05; **P < .01; ***P < .001 between vehicle- and DEX-treated groups within genotypes by 2-way ANOVA followed by Bonferroni posttests. Interaction effect was not significant (P > .05, 2-way ANOVA).

Body weight and body composition

We examined body composition of 2-month-old and 4-month-old WT and GR^dim/dim mice. Two-month-old GR^dim/dim mice did not differ from WT mice with respect to total mass or lean mass, but had higher fat mass (Table 1; 3.4 ± 0.7 vs 2.7 ± 0.2, P < .05) and percent fat mass than WT mice (Table 1; 15.7 ± 3.0 vs 13.3 ± 1.0, P < .05). Four-month-old GR^dim/dim mice had the same pattern, with no difference in total mass compared with WT mice, but with increased fat mass and percent fat mass (Table 1; 5.8 ± 1.4 vs 4.3 ± 1.1 and 20.2 ± 3.3 vs 15.7 ± 3.8, P < .05, respectively).

Table 1.

Body Composition in 2-Month-Old and 4-Month-Old WT and GR^dim/dim Mice^a

	n	Mass	Lean	Fat	% Fat
WT, 2 months Pre	11	21.6 ± 1.2	17.7 ± 0.9	2.7 ± 0.2	13.3 ± 1.0
WT, 2 months Veh	6	24.6 ± 1.9	19.3 ± 2.2	2.7 ± 0.3	12.1 ± 1.6
WT, 2 months Dex	5	21.1 ± 1.1	16.1 ± 1.2	2.8 ± 0.4	15.1 ± 1.8
DIM, 2 months Pre	10	22.1 ± 1.8	18.1 ± 1.6	3.4 ± 0.7b	15.7 ± 3.0b
DIM, 2 months Veh	5	24.3 ± 1.8	19.1 ± 2.0	3.2 ± 0.3	15.0 ± 1.3
DIM, 2 months Dex	5	22.7 ± 1.5	17.4 ± 1.3	3.6 ± 0.5c	16.5 ± 1.8c
WT, 2 months	12	29.2 ± 2.0	23.1 ± 1.9	4.3 ± 1.1	15.7 ± 3.8
DIM, 2 months	13	30 ± 2.6	22.8 ± 1.5	5.8 ± 1.4b	20.2 ± 3.3b

Abbreviations: Dex, after 1 week of treatment with DEX; DIM, GR^dim/dim; Pre, before treatment; Veh, after 1 week of treatment with vehicle.

^aData are shown as mean ± SD.

^bP < .05 from WT.

^cP < .05 from vehicle by t test.

Adipose tissue lipid dynamics in vivo

Fat pad mass

In the inguinal depot DEX treatment resulted in significantly larger (P < .01) fat pads in GR^dim/dim mice, but not in WT mice, relative to their respective vehicle controls (Table 2). In the epididymal depot, DEX treatment did not significantly affect fat pad mass in either WT or GR^dim/dim mice relative to their respective vehicle controls (Table 2). Differences in sensitivity to DEX between the 2 genotypes were not significant for fat pad mass (P > .05) in either depot.

Table 2.

Fat Pad Mass and Fractional Synthesis in 4-Month-Old WT and GR^dim/dim (DIM) Mice Treated with Vehicle or DEX and Labeled With Heavy Water^a

Group	n	Inguinal Depot			Epididymal Depot
		Fat Pad Mass (mg)	Fraction New TG/wk (f)	Fraction New Palmitate/wk (f)	Fat Pad Mass (mg)	Fraction New TG/wk (f)	Fraction New Palmitate/wk (f)
WT, Veh	13	248 ± 78	0.35 ± 0.14	0.21 ± 0.12	421 ± 167	0.19 ± 0.06	0.08 ± 0.05
WT, DEX	14	372 ± 221	0.40 ± 0.16	0.24 ± 0.13	638 ± 400	0.25 ± 0.07b	0.12 ± 0.07
DIM, Veh	13	353 ± 116	0.29 ± 0.16	0.16 ± 0.14	728 ± 253	0.15 ± 0.09	0.07 ± 0.10
DIM, DEX	14	534 ± 95c	0.40 ± 0.09	0.33 ± 0.10c	964 ± 311	0.29 ± 0.05d	0.24 ± 0.07d

Abbreviations: DIM, GR^dim/dim; Pre, before treatment; Veh, vehicle.

^aData are shown as mean ± sd. P values are compared with vehicle of same genotype by 2-way ANOVA followed by Bonferroni posttests.

^bP < .05.

^cP < .01.

^dP < .001.

Adipose tissue TG fractional synthesis rates

In the inguinal depot, DEX treatment did not significantly affect synthesis rates of TG in either WT or GR^dim/dim mice relative to their respective vehicle controls (Table 2). In the epididymal depot, DEX treatment resulted in significantly higher rates of TG synthesis in WT (P < .05) and GR^dim/dim (P < .001) mice relative to their respective vehicle controls (Table 2). Differences in sensitivity to DEX between genotypes were not significant (P > .05) in either depot.

Adipose tissue absolute TG synthesis rates

By combining fat pad mass with fractional TG synthesis rates, we calculated absolute TG synthesized over the week-long labeling period in inguinal and epididymal fat depots (Figure 3, A and B). In both the inguinal and epididymal depots, DEX treatment resulted in significantly more TG synthesized in both WT (P < .05) and GR^dim/dim (P < .001) mice compared with their respective vehicle controls. Furthermore, in both the inguinal and epididymal depots, GR^dim/dim mice were significantly more sensitive to DEX-dependent increases in absolute TG synthesis compared with WT mice (P < .001 and P < .01 in inguinal and epididymal depots, respectively).

Figure 3.

Absolute TG synthesis and DNL in inguinal and epididymal fat pads of 4-month-old WT and GR^dim/dim (DIM) mice; n = 13 for each vehicle-treated group, n = 14 for each DEX-treated group. Data are shown as mean ± SD. *P < .05; ***P < .001 between vehicle- and DEX-treated groups within genotypes by 2-way ANOVA followed by Bonferroni posttests. A, Absolute TG synthesis in the inguinal depot. GR^dim/dim mice were significantly more sensitive to DEX (interaction P < .001, 2-way ANOVA). B, Absolute TG synthesis in the epididymal depot. GR^dim/dim mice were significantly more sensitive to DEX (interaction P < .01, 2-way ANOVA). C, Absolute DNL in the inguinal depot. GR^dim/dim mice were significantly more sensitive to DEX (interaction P < .001, 2-way ANOVA). D, Absolute DNL in the epididymal depot. GR^dim/dim mice were significantly more sensitive to DEX (interaction P < .001, 2-way ANOVA).

Fractional rates of DNL

DEX treatment resulted in significantly higher fractional rates of palmitate DNL in both depots of GR^dim/dim mice (P < .01 and P < .001 in the inguinal and epididymal depot, respectively), but not in WT mice, compared with their respective vehicle controls (Table 2). Although significant differences between genotypes in sensitivity to DEX were not found in the inguinal depot, in the epididymal depot, GR^dim/dim mice were found to be significantly more sensitive (P < .01) to DEX-dependent increases in rates of DNL compared with WT.

Absolute DNL

By combining fat pad mass with fractional rates of DNL, we calculated absolute palmitate synthesized by DNL over the week-long labeling period (Figure 3, C and D). In both the inguinal and epididymal fat depots, DEX treatment resulted in significantly more (P < .001) absolute DNL in GR^dim/dim mice, but not in WT mice, compared with their respective vehicle controls. Furthermore, in both fat depots, GR^dim/dim mice were significantly more (P < .001) sensitive to DEX-dependent increases in absolute DNL compared with WT.

Summary of adipose tissue lipid dynamics in vivo

In summary, GR^dim/dim mice have slightly larger inguinal and epididymal fat pads than WT mice. In the absence of DEX, GR^dim/dim and WT mice synthesize similar amounts of TG and synthesize similar amounts of FA via DNL in both fat depots. Although both GR^dim/dim and WT mice synthesize more TG in response to DEX treatment, only GR^dim/dim mice synthesize more FA via DNL. Finally, GR^dim/dim mice are much more sensitive to a DEX-dependent increase in TG synthesis and DNL in both inguinal and epididymal adipose depots.

Ex vivo lipolysis

We measured glycerol release as a surrogate for lipolysis in inguinal and epididymal adipose tissue explants (Figure 4). Rates of lipolysis were significantly lower (P < .05) by 40% in inguinal adipose tissue from GR^dim/dim mice treated with DEX. GR^dim/dim mice were also significantly more (P < .05) sensitive to DEX-dependent decreases in rates of lipolysis compared with WT in the inguinal depot. No other significant differences in rates of lipolysis or sensitivity to DEX were found in adipose tissue from either depot.

Figure 4.

Ex vivo rates of lipolysis in inguinal (A) and epididymal (B) fat depots of WT and GR^dim/dim (DIM) mice. Rates of lipolysis were measured in dissected fat as mass (nanograms) of glycerol released into the surrounding media per milligram of adipose tissue per minute; n = 4 per group. *P < .05 between vehicle- and DEX-treated groups within genotypes by 2-way ANOVA followed by Bonferroni posttests. GR^dim/dim mice were significantly more sensitive to DEX in the inguinal depot only (interaction P < .05, 2-way ANOVA).

Adipose gene expression

To determine the molecular mechanisms for the altered response to DEX in GR^dim/dim mice, we analyzed gene expression from WT and GR^dim/dim mice that had been acutely treated with DEX. Most genes encoding enzymes in the triacylglycerol synthesis pathway were not responsive to DEX in both WT and GR^dim/dim mice (Agpat2, Dgat1, and Dgat2, data not shown). There was a nonstatistically significant trend for induction of Gpam in WT adipose, encoding a glycerol-3-phosphate O-acyltransferase enzyme, which is rate limiting in TG synthesis (Figure 5). This trend was not observed in GR^dim/dim adipose (Figure 5). We did observe a DEX-dependent increase in Lpin3 gene expression in WT and not GR^dim/dim mice (Figure 5, WT 4.71 ± 2.02 and GR^dim/dim 2.47 ± 2.33; all values are log₂). Lipocalin 2 (Lcn2), a gene involved in lipid metabolism, but not directly in the TG synthesis pathway, was upregulated by DEX in both WT (5.63 ± 1.91) and GR^dim/dim adipose tissue (2.48 ± 1.55). Pdk4, encoding pyruvate dehydrogenase kinase, another gene impacting lipid metabolism, was also induced in WT (5.45 ± 2.01) but not GR^dim/dim (−0.45 ± 1.33). We were also able to observe gene induction of some classic GR targets in WT, but not GR^dim/dim adipose including Fkbp5 (WT 6.58 ± 2.18, GR^dim/dim 2.37 ± 2.02) and Agt (WT 3.39 ± 0.86 and GR^dim/dim 0.71 ± 1.06). We did not observe DEX-mediated induction of Hsl and Lpl, 2 genes involved in adipose triacylglycerol metabolism previously shown to be regulated by GCs.

Figure 5.

GR dimerization-dependent gene expression in adipose tissue. WT (n = 10) and GR^dim/dim (DIM, n = 8) mice were injected with 10 mg/kg DEX (n = 5 for WT and n = 4 for DIM) or vehicle (n = 5 for WT and n = 4 for DIM) at 7:00 am. Mice were killed 6 hours later, and gene expression was determined. *P < .05 for DEX different from vehicle.

Insulin sensitivity

GCs are widely known to decrease insulin sensitivity. Accordingly, we measured peripheral tissue insulin sensitivity using the ²H-GDT. DEX treatment did not significantly affect blood glucose AUC in WT (15 ± 3.8 × 10³ mg · min/dL for vehicle vs 15 ± 2.7 × 10³ mg · min/dL for DEX) or GR^dim/dim mice (13 ± 2.7 × 10³ mg · min/dL for vehicle vs 12 ± 2.5 × 10³ mg · min/dL for DEX) compared with their respective vehicle controls (Figure 6A). Average blood glucose AUC of all GR^dim/dim mice was significantly lower (12 ± 2.5 × 10³ mg · in/dL for GR^dim/dim vs 15 ± 3.2 × 10³ mg · min/dL for WT, P < .05, main effect of genotype, 2-way ANOVA) compared with all WT mice.

Figure 6.

GDT, a modified oral glucose tolerance test. Blood glucose concentrations, plasma insulin concentrations, percentage of orally administered glucose load absorbed by peripheral tissues and metabolized to water (% glc load metabolized), and SI, calculated percent glucose load metabolized divided by the product of plasma insulin AUC and blood glucose AUC; n = 7 per group. Data are shown as mean ± SD. **P < .01; ***P < .001 between vehicle- and DEX-treated groups within genotypes by 2-way ANOVA followed by Bonferroni posttests. A, Blood glucose concentrations throughout the test. GR^dim/dim groups had on average significantly lower blood glucose than WT groups (main effect of genotype P < .05, 2-way ANOVA). Interaction effect was not significant (P > .05, 2-way ANOVA). B, Plasma insulin concentrations throughout the test. Interaction effect was not significant (P > .05, 2-way ANOVA). C, Percent glucose load metabolized to water. Interaction effect was not significant (P > .05, 2-way ANOVA). D, SI. WT mice were significantly more sensitive to DEX (interaction P < .05, 2-way ANOVA).

DEX treatment resulted in significantly higher insulin levels in both GR^dim/dim (65 ± 17 ng · min/mL for vehicle vs 180 ± 50 ng · min/mL for DEX, P < .001) and WT (32 ± 8.4 ng · min/mL for vehicle vs 150 ± 77 ng · min/mL for DEX, P < .01) mice compared with their respective vehicle controls (Figure 6B). GR^dim/dim mice were not significantly more sensitive to DEX-dependent increases in insulin AUC compared with WT (P > .05).

Percent glucose load metabolized to water was used as a metric of uptake and metabolism of an oral glucose load to water by peripheral tissues. No significant differences between groups were found (Figure 6C).

By combining glucose AUC, insulin AUC, and percent glucose load metabolized to water, we calculated SI (Figure 6D). Treatment with DEX resulted in significantly lower (P < .001) SI in WT mice, but not in GR^dim/dim mice, compared with their respective vehicle controls. Furthermore, WT mice were significantly more (P < .05) responsive to DEX-dependent decreases in SI compared with GR^dim/dim mice.

In summary, GR^dim/dim mice had lower blood glucose levels than WT, and DEX increased plasma insulin levels similarly in both GR^dim/dim and WT mice. Absorption of blood glucose and metabolism to water by peripheral tissues was similar in both GR^dim/dim and WT mice and was unaffected by DEX administration. DEX treatment resulted in decreased SI in WT mice, but SI was not significantly decreased by DEX in GR^dim/dim mice. Lastly, WT mice were more responsive to DEX-dependent decreases in insulin sensitivity than were GR^dim/dim mice.

Chronically elevated GCs in GR^dim/dim mice results in runting and lethality

Given the profound abnormalities in the response of GR^dim/dim mice to acute GCs, we wanted to know what their response to chronically elevated GCs would be. To answer this question, we bred GR^dim/dim mice to CRH-Tg⁺, a transgenic line with a constitutively active CRH transgene resulting in chronically elevated ACTH and corticosterone (48). CRH-Tg⁺ mice have increased fat mass, decreased lean mass, and decreased bone content (Harris, C., unpublished observations). Of 148 pups born from crosses of CRH-Tg⁺, GR^dim/+ males to GR^dim/dim females, 15 were GR^dim/dim. Of these, only 3 were CRH-Tg⁺ (expected number is 7.5, P < .01 by χ² test). The surviving GR^dim/dim, CRH-Tg⁺ mice were severely runted, and all 3 mice generated died before 6 weeks of age, weighing ∼5 g. We were therefore unable to further study metabolic parameters in these mice.

Discussion

An intact GR dimerization domain is not required for suppression of collagen synthesis in skin or bone or suppression of protein synthesis in skeletal muscle

Previous studies have shown that the effects of GCs on bone as assessed by calcein-guided bone formation rates, histomorphometry, and collagen mRNA expression did not require an intact GR dimerization domain (25). However, our study is the first study to examine directly the effects of GCs on skin and bone collagen synthesis in vivo in GR^dim/dim mice. We were unable to detect significant DEX-dependent decreases in bone collagen synthesis rates in 4-month-old WT or GR^dim/dim mice (Figure 1A). We believe that this is due to the fact that the rate of bone collagen turnover decreases as mice get older, so 1 week of treatment and labeling was likely not sufficient to observe reductions in bone collagen synthesis.

Previously, we had observed significant differences in 2-month-old mice treated with prednisolone for 1 week (28). We repeated the experiment here in a cohort of 2-month-old WT and GR^dim/dim mice and observed significant DEX-dependent decreases in bone collagen synthesis rates in both genotypes (Figure 1B). Interestingly, we found that the difference between vehicle-treated and DEX-treated groups was significantly greater in WT mice compared with GR^dim/dim mice. Although the effect was greater in WT mice, vehicle-treated (baseline) bone collagen synthesis rates appeared to be slightly lower in GR^dim/dim mice. As expected, baseline bone collagen synthesis rates were lower in 4-month-old mice than in 2-month-old mice, and the 4-month-old GR^dim/dim mice did not have lower baseline rates of bone collagen synthesis compared with WT.

It is not clear from these studies whether the 2-month-old GR^dim/dim mice are less sensitive to DEX-dependent decreases in bone collagen synthesis rates or whether they have a slower baseline rate of bone turnover at this age. But because WT and GR^dim/dim mice have similar bone mass and bone mineral density (Harris, C., unpublished observations), we hypothesize that there is no consequential difference in baseline bone collagen synthesis rates up to 4 months of age. Furthermore, we conclude that key pathways responsible for GC-dependent decreases in bone collagen synthesis rates in WT mice are mostly intact in GR^dim/dim mice.

We observed DEX-dependent decreases in skin collagen synthesis rates in WT and GR^dim/dim mice (Figure 1C). There was a significant difference in sensitivity to DEX between genotypes, and baseline skin collagen synthesis rates appeared lower by approximately half in GR^dim/dim mice compared with WT. Because no differences in skin thickness were observed (data not shown), we assume that turnover of skin collagen is also lower in GR^dim/dim mice. Despite differences in baseline skin collagen synthesis rates, there was approximately a 90% reduction in rates of skin collagen synthesis in both genotypes treated with DEX. Thus, we conclude that key pathways responsible for GC-dependent decreases in skin collagen synthesis rates in WT mice are mostly intact in GR^dim/dim mice.

In both WT and GR^dim/dim mice, we saw modest, but significant, suppression of protein synthesis rates in skeletal muscle (Figure 2). This modest decrease is in line with our previous observations in 2-month-old WT mice (28), with the observations of other groups studying GR^dim/dim mice (26), and with GCs causing muscle wasting predominantly by inducing protein degradation rather than suppressing synthesis (7, 16, 49–52). Thus, we conclude that key pathways responsible for GC-induced decreases in muscle protein synthesis in WT mice are intact in GR^dim/dim mice; however, we cannot rule out the possibility that pathways responsible for muscle protein degradation are not.

GC-mediated changes in adipose lipid dynamics are exaggerated in the absence of a GR dimerization domain

We found that GC-induced increases in both TG synthesis and DNL were greatly exaggerated in GR^dim/dim mice compared with WT mice (Figure 3). Based on body compositions and fat pad masses (Tables 1 and 2), we conclude that GR^dim/dim mice accrue fat faster than WT in response to DEX. We also observed that the difference in the amount of TG synthesized is roughly the same as the difference in fat pad masses between vehicle-treated and DEX-treated WT and GR^dim/dim animals. Thus, TG accrued was roughly similar to the amount of TG synthesized in response to DEX in these depots, and DEX had much more of an effect on TG synthesis compared with lipolysis. Furthermore, when we directly measured lipolysis in dissected adipose tissue from WT and GR^dim/dim mice treated with vehicle or DEX for 4 days, we found no significant change in rates of lipolysis in adipose tissue from either depot in WT mice and a decrease and no change in rates of lipolysis in adipose tissue from the inguinal and epididymal depot, respectively, of GR^dim/dim mice (Figure 4). In summary, we conclude, first, that DEX increases TG synthesis to a much greater extent in GR^dim/dim mice than in WT mice, and second, net increases in synthesis are proportionally much greater than net changes in lipolysis. The increased TG synthesis in adipose of GR^dim/dim mice could not be explained by increased activation of genes encoding TG synthesis in adipose tissue as demonstrated by focused gene expression studies; however, whole-genome experiments may uncover the molecular explanation for the increased TG synthesis we observed.

Even more striking were genotypic differences in DEX-dependent changes in DNL. DEX did not significantly increase DNL in either depot of WT mice, but dramatically increased DNL in both depots of GR^dim/dim mice (Figure 3). One limitation of our in vivo studies is that because of the prolonged time required for labeling adipose TG, we cannot determine the site of action of the effects of GCs on DNL, such that the increased rate of appearance could be due to increased adipose DNL or increased hepatic DNL with subsequent very-low-density lipoprotein export to adipose tissue. In summary, DEX-treated GR^dim/dim mice, but not WT mice, retain severalfold more FAs derived from DNL in TG in both inguinal and epididymal adipose depots compared with vehicle-treated GR^dim/dim mice. These physiological changes could not be easily explained by observed changes in gene expression in the adipose tissue; however, a whole-genome analysis would be needed to rule this out. Furthermore, loss of GR dimerization in other relevant organs such as liver could indirectly affect lipid homeostasis in adipose tissue through delivery of substrates.

In the absence of DEX, GR^dim/dim mice are less insulin sensitive than WT mice

The decrease in SI in response to DEX was statistically significant in WT mice, but not in GR^dim/dim mice (Figure 5D). However, both WT and GR^dim/dim mice had approximately the same SI when treated with DEX. DEX treatment mainly affected insulin levels (Figure 5B) and didn't affect glucose levels (Figure 5A) or the percentage of the oral glucose load metabolized to water (Figure 5C). GR^dim/dim mice had significantly lower glucose levels compared with WT, independent of DEX treatment (Figure 5A). Because baseline insulin levels were slightly higher in GR^dim/dim mice relative to WT (Figure 5B), vehicle-treated SI was lower in GR^dim/dim mice than in WT. Thus, DEX treatment resulted in increases in insulin levels in both genotypes, without changes in glucose levels, and similar SI values in both genotypes. Although one interpretation could be that GR^dim/dim mice are protected from GC-induced decreases in SI, the results are also consistent with the conclusion that GR^dim/dim mice are less insulin sensitive than WT mice at baseline, and both genotypes are rendered equally insulin resistant when treated with DEX.

It has been previously demonstrated that physiological concentrations of corticosterone inhibit glucose-stimulated insulin secretion in isolated islets from mice (53). Based on this previous work and our observations that plasma insulin levels are slightly higher and blood glucose levels are slightly lower at baseline in GR^dim/dim mice compared with WT mice, we speculate that GC-dependent inhibition of glucose-stimulated insulin secretion may be dependent on GR homodimerization. When DEX is administered, however, peripheral insulin resistance dominates, the pancreas secretes approximately equal amounts of insulin to compensate, and SI is similar in both genotypes.

Chronically elevated GCs in GR^dim/dim mice is not consistent with life

We observed significantly decreased viability of GR^dim/dim mice. This decrease in viability did not appear to result in genetic selection because we have not seen any change in the percentage of GR^dim/dim mice surviving over greater than 15 generations. The severe runting and lethality of CRH-Tg⁺, GR^dim/dim mice suggests a gain-of-function property of the GR^dim/dim mutation, given that this phenotype is not seen in either GR^dim/dim mice with normal corticosterone levels or CRH-Tg⁺ (with WT GR). This may be due to the mixed gain and loss of function of the GR^dim/dim mutation. The mutation was originally described as being pure loss of function for transactivation of reporter constructs (22). However, careful analysis of several target genes has revealed potential gain-of-function properties of the GR^dim/dim mutation, because a subset of genes were hyperactivated in U2OS cells overexpressing GR-dim vs WT GR (54). This does not appear to be an overexpression artifact, because the same phenomenon has been observed with endogenous levels of GR (24). Microarray analysis of liver from WT and GR^dim/dim mice revealed that in addition to genes that were activated by GCs in WT but not GR^dim/dim mice (GR dimerization-dependent), there was a subset of genes that were induced equally by DEX in both genotypes (GR dimerization-independent) and also a set of genes that were activated in the livers of GR^dim/dim mice but not WT mice by DEX (24) (Harris, C., unpublished observations). We propose the term rogue targets for genes that are up-regulated by DEX in GR^dim/dim but not in GR^+/+. These genes cannot be definitely labeled as GR primary targets without first performing chromatin precipitation experiments, but the early time points used favor detection of primary GR targets. Although the phenotype of GR^dim/dim mice may be a result of imbalance between GR-dependent and -independent genes, the runting of CRH-Tg⁺, GR^dim/dim mice is likely due to rogue gene activation, because it did not occur in GR^dim/dim mice with normal corticosterone levels. In our focused adipose gene expression studies, we were able to observe dimerization-dependent (eg, Agt and Pdk4) and dimerization-independent GR targets (Lcn2, Figure 6) but not rogue genes. Because rogue genes are a small percentage of genes, whole-genome expression studies will be needed to identify them. We suspect that rogue gene activation occurs in a tissue-specific manner but do not know which tissue or rogue genes are responsible for the runting of CRH-Tg⁺, GR^dim/dim mice. Whether the mixed function of the GR^dim/dim allele applies to repression as well as transactivation is an open question.

Implications for SGRMs

Much attention has been focused on the search for SGRMs that would dissociate the therapeutic anti-inflammatory effects from the adverse metabolic effects. The GR^dim/dim mouse has the promise of being a useful model for dissociating GC action at the level of GR instead of the ligand. The fact that the GR^dim/dim mutation is not devoid of transactivating some target genes and can also mediate gain-of-function gene expression is instructional in the search for SGRMs. In addition to having the beneficial properties of not activating certain subsets of GR targets that mediate adverse effects, it is possible that SGRMs might activate genes not normally activated by nondissociated GCs, such as the physiological ligand cortisol (corticosterone in mice). Therefore, it will be important to assess the tissue-specific effects of all potential SGRMs on global transcription and GC-dependent metabolic pathways.