Glucocorticoid Resistance Remodels Liver Lipids and Prompts Lipogenesis, Eicosanoid, and Inflammatory Pathways

Abstract

We have previously demonstrated that the glucocorticoid receptor β (GRβ) isoform induces hepatic steatosis in mice fed a normal chow diet. The GRβ isoform inhibits the glucocorticoid-binding isoform GRα, reducing responsiveness and inducing glucocorticoid resistance. We hypothesized that GRβ regulates lipids that cause metabolic dysfunction. To determine the effect of GRβ on hepatic lipid classes and molecular species, we overexpressed GRβ (GRβ-Ad) and vector (Vec-Ad) using adenovirus delivery, as we previously described. We fed the mice a normal chow diet for 5 days and harvested the livers. We utilized liquid chromatography-mass spectrometry (LC-MS) analyses of the livers to determine the lipid species driven by GRβ. The most significant changes in the lipidome were monoacylglycerides and cholesterol esters. There was also increased gene expression in the GRβ-Ad mice for lipogenesis, eicosanoid synthesis, and inflammatory pathways. These indicate that GRβ-induced glucocorticoid resistance may drive hepatic fat accumulation, providing new therapeutic advantages.

Graphical Abstract.

High levels of hepatic GRβ cause glucocorticoid resistance, which increases monoacylglycerol and cholesterol ester lipid species. The elevation of GRβ activates genes regulating lipogenesis, eicosanoid metabolism, and inflammation.

INTRODUCTION

In the past two decades, an obesity epidemic has emerged, and adults currently display the largest weights known in history. The obesity rate is 42.4%, and combined with people who are overweight, comprise nearly 70% of the population [1]. Obesity is mostly caused by overeating and diets high in fat that, if continued, can lead to metabolic-associated comorbidities such as non-alcoholic fatty liver disease (NAFLD) [2, 3] that can instigate type II diabetes and cardiovascular diseases [4–6]. NAFLD is a problem worldwide and is estimated to have a prevalence of 32% in adults [7]. Chronic high-fat feeding has been shown to raise circulating levels of glucocorticoids and induce NAFLD [8, 9]. People who have been chronically administered glucocorticoids have higher chances of developing NAFLD [10–13]. Similarly, people with Cushing’s syndrome, who have significantly elevated levels of the endogenous glucocorticoid cortisol, have a 33% increased risk of developing NAFLD [14].

The glucocorticoid receptor (GR) regulates lipid and glucose homeostasis in the liver [15]. However, the specific functions of the GR isoforms and their regulation of NAFLD have yet to be established. The GR has two isoforms (GRα and GRβ) that arise from the GR gene (NR3C1) due to the alternative splicing of exon 9 [15, 16]. The GRα isoform binds to glucocorticoids, while the GRβ has a truncated C-terminus that does not allow it to bind to them [17–20]. However, GRβ can still dimerize with GRα on gene promoters, which reduces glucocorticoid responsiveness. Hence, increasing GRβ levels induce glucocorticoid resistance [13, 15, 20–23]. It has been previously shown that chronic glucocorticoid administration can lead to NAFLD [11]. GRβ levels significantly increase in high-fat-fed mice [24] and are elevated by chronic glucocorticoid therapy [20, 24], suggesting that these are possibly glucocorticoid-resistant states. Glucocorticoid resistance has been implicated in numerous pathologies, including asthma, depression, obesity, cardiovascular disease, and NAFLD [13]. Glucocorticoid resistance occurs for several reasons, including genetic or acquired [13, 25–28]. Still, it always results in reduced glucocorticoid responses, possibly due to increased GRβ levels [29, 30].

Glucocorticoids have been shown to increase several genes that encode enzymes that mediate de novo lipogenesis and triacylglycerol (TAG) synthesis. Specifically, the luciferase activity of acetyl-CoA carboxylase 1 and 2 (ACACA and ACACB) promoters was activated by adding dexamethasone, a potent GRα agonist [31]. Reporter gene activity analysis using the fatty acid synthase gene (FASN) promoter has also been shown to be increased with dexamethasone treatment [32]. In 3T3-L1 adipocytes, Yu et al. identified GR response elements (GREs) in the promoters of genes involved in triglyceride synthesis [33], suggesting a direct role from one of the GR isoforms in regulating lipid accumulation. Hepatocyte-specific knockout (KO) of the GR gene (Nr3c1) in male C57BL/6 mice displayed TAG accumulation, hepatic steatosis, and an increase of sphingomyelins and ceramides but no change to serum TAGs [34]. They had reduced lipid accumulation after shRNA intervention, possibly due to increased hairy enhancer of split 1 (Hes1) activity [35]. This model also represents a glucocorticoid-resistant state, indicating that this may be a mechanism in NALFD development. However, these disruptions were in the entire GR gene, so how the GR isoforms regulate hepatic de novo lipogenesis and lipid accumulation has yet to be determined.

In the present study, we hypothesized that GRβ regulates a pool of hepatic fatty acid species in the liver that might lead to liver dysfunction. We found that overexpression of GRβ regulated specific fatty acids. We previously demonstrated that GRβ regulates hepatic steatosis by decreasing the β-oxidation pathway that burns fat through a reduction in the peroxisome proliferator-activated receptor α (PPARα) nuclear receptor that controls it [24]. GRβ has been implicated in driving de novo lipogenesis and increased hepatic inflammation by increasing FASN protein [36]. This study is the first to determine that GRβ specifically regulates lipid content in the liver.

RESULTS

GRβ regulates the hepatic lipidome.

To determine how GRβ signaling affects lipid species in the liver, we used 8-week-old mice injected with either GRβ adenovirus (GRβ-Ad) or adenovirus vector (Vec-Ad) control (Supplemental Figure 1), as we have previously described in [24]. The mice were fed a normal chow diet for 5 days, and as we showed before using the same GRβ-Ad, the livers accumulated fat (data not shown). As we performed in [34–36], the livers were harvested and extracted for lipidomic analysis. After normalization and quality control, 915 lipid species were analyzed in both groups. Of these lipid species, 508 were significantly changed due to GRβ overexpression (GRβ-Ad) compared to the Vec-Ad controls. From the 508 significantly (p < 0.05) changed lipid species, 248 were downregulated, and 260 were upregulated in the GRβ-Ad livers compared to the Vec-Ad. A PCA plot was used to analyze lipid clustering, which indicated two distinct lipid clusters between the Vec-Ad and GRβ-Ad (Figure 1A). In Figure 1B, the heatmap shows the diverse effects of GRβ overexpression on the significantly upregulated and downregulated individual lipid species. Analysis of the top 10 changed lipid species showed a strong effect of GRβ on the lipid clustering (Figure 1C). A lipid class enrichment plot analysis identified cholesterol esters (CEs) and monoacylglycerides (MAGs) as the most significantly increased lipid species (Figure 1D). We generated a volcano plot (Figure 1E) to visualize the overall changes in individual lipid species in the GRβ-Ad compared to Vec-Ad. We found that some of the highest changed individual lipids were MAGs, as indicated by the light blue dots (Figure 1E). This can be further visualized in the log 2-fold change waterfall plot in Figure 1F, which shows the top 150 upregulated lipid species. The lipidomic data indicate that GRβ increases specific lipid classes, which we determine further below for each subtype and pathways mediating these differences.

Figure 1. Hepatic lipid species clusters in the livers of Vec-Ad and GRβ-Ad.

(A) PCA plot for clustering the Vec-Ad and GRβ-Ad liver lipid species. (B) Heat map indicating the upregulation and downregulation of lipid species. (C) Variable PCA plot for the top lipid species indicated by the arrows with their contribution to clustering. (D) Enrichment plot for the most significantly changed lipid classes. (E) Volcano plot of the most upregulated and downregulated lipid species. (F) The waterfall plot for the top 150 increased lipids is represented by log2 (fold change) for the statistically significant lipid species. The color of the bubbles denotes the −log10 (p-value). The darker shades of red indicate greater significance. [n =5 per group].

Lipidome network and GRβ regulation of lipid class.

In Figure 2A, we generated a network analysis for the lipid classes from the livers of the GRβ-Ad and Vec-Ad mice. The shaded green arrows represent positive Z scores for the active lipid classes and purple arrows for those suppressed. The network analysis identified the highest positive Z scores for sphingomyelin (SM) (Z=2.792), MAGs (Z=4.304), and lysophosphatidylinsitol (LPI) (Z=3.305) due to increased GRβ expression. The three lowest positive Z scores due to increased GRβ activity were phosphatidylcholine (PC) (Z=1.699), phosphatidylserine (PS) (Z=1.989), and lipopolysaccharides (LPS) (Z=2.209). No shaded purple arrows indicated the lack of negative Z scores affected by GRβ expression for these lipid classes. Since it was indicated in our analysis in Figure 1D that CEs were one of the most significantly increased lipid species, we analyzed the individual CE classes. We found no significant differences in the GRβ-Ad compared to the Vec-Ad mice for all CE classes. However, a trend showed an increase in the overall levels of all CEs for each class in the GRβ-Ad mice compared to the Vec-Ad (Figure 2B). The network analysis also indicated an increase in the SM lipids in the GRβ-Ad compared to the Vec-Ad mice. Further investigation into the individual SM lipid class indicated no significant differences between the GRβ-Ad and Vec-Ad mice (Figure 2C). To quantitate the hepatic lipidome effects of GRβ, we next analyzed other lipid classes in ceramide synthesis and their metabolite pathways.

Figure 2. Network analysis of lipid species analysis in Vec-Ad and GRβ-Ad mice.

(A) Network analysis of lipid classes, where node shape indicates lipid class and node color represents if the lipid class was active/suppressed (green) or unchanged (white). The purple arrow represents a negative Z-score, while the green line represents a positive Z-score. A shaded arrow indicates that the z-score is involved in the active/suppressed status. Sub-lipid species levels from two lipid classes: (B) cholesterol esters and (C) sphingomyelin. [n =5 per group; *, p < 0.05].

GRβ mediates ceramide synthesis.

Investigating the ceramide and dihydroceramides showed a reduction in several ceramide species in the GRβ-Ad mice compared to the Vec-Ad mice. Figure 3A shows six significantly decreased long-chain ceramides (CERs) [22:1, 20:1, 20:0, 18:0, 16:0, and 14:0]. Figure 3B shows three significantly decreased dihydroceramides (DCERs) [24:1, 22:0, and 18:1]. There are modifications to the ceramides, such as the hexosylceramides (HCER), which belong to the group of cerebrosides within the sphingolipids. We observed one significantly decreased HCER [20:0] in Figure 3C. Lactosylceramides (LCER) are another group of modified ceramides belonging to the diglycosylceramides within the sphingolipids. We found no significant differences in lactosylceramide classes (Figure 3D).

Figure 3. Ceramide lipid species analysis in Vec-Ad and GRβ-Ad mice.

Sub-lipid species levels from the ceramide lipid class: (A) ceramides, (B) dihydroceramides, (C) hexosylceramides, and (D) lactosylceramides. [n =5 per group; *, p < 0.05].

We investigated the different ceramide metabolism gene pathways to dissect further the changes GRβ overexpression made (Figure 4). We found that there was significantly increased (p<0.05) expression in UDP-glucose ceramide glucosyltransferase (Ugcg) (p<0.01) and sphingomyelin synthase 1 (Sgms1) mRNAs in the GRβ-Ad mice compared to the Vec-Ad mice livers. We also found a significant decrease (p<0.05) in sphingomyelin phosphodiesterase 2 (Smpd2) mRNA. No significant changes in other ceramide gene pathways were measured. These indicate that GRβ controls hepatic lipid content by controlling gene expression. Next, we wanted to determine how GRβ impacts hepatic triglyceride content.

Figure 4. Gene analysis of ceramide synthesis pathways.

Real-time PCR measurement of mRNAs of genes in the ceramide synthesis pathway. [n =6–7 per group; *, p < 0.05].

GRβ controls hepatic triglyceride synthesis.

GRβ-Ad mice displayed increased lipid species in Figure 1D for MAGs and CEs, of which the CEs were found not significantly increased per each class. Further analysis of the individual MAG lipid species showed high overall levels in the GRβ-Ad mice (Figure 5A). Still, there were no significant differences between the GRβ-Ad and Vec-Ad mice. Although there was an increase in the overall MAGs for each class, it was an upward trend with no significance. Analysis of diacylglycerols (DAGs) in Figure 5B, showed a significantly decreased DAG 18:1/18:2 in the GRβ-Ad mice compared to the Vec-Ad mice. Then, we analyzed lyso-phosphatidyl lipid classes (Figure 5C). We observed a significantly decreased level in lyso-phosphatidyl-ethanolamine [LPE 22:4], lyso-phosphatidylglycerol [LPG 22:4, 20:4, and 20:3], and lyso-phosphatidylinositol [LPI 22:4] but no significant differences to any lyso-phosphatidylcholine (LPC) or lyso-phosphatidylserine (LPS). Because of the overall increase in the lipid classes, we next analyzed the effects of GRβ on gene pathways that regulate lipogenesis, eicosanoids, and inflammatory pathways that they later activate.

Figure 5. Triglyceride analysis in Vec-Ad and GRβ-Ad mice.

The species levels of (A) monoacylglycerol, (B) diacylglycerol, (C) lyso-phosphatidylcholine, lyso-phosphatidyl-ethanolamine, lyso-phosphatidylglycerol, lyso-phosphatidylinositol, and lyso-phosphatidylserine from Vec-Ad and GRβ treated mice. [n =5 per group; *, p < 0.05].

GRβ regulation of gene pathways.

We had previously demonstrated that GRβ-Ad mice had an increase in fatty acid synthase [36] protein expression, which regulates de novo lipogenesis of lipids [24]. We assessed other genes in the lipogenesis pathway to determine whether GRβ overexpression induced significant changes. There was a significantly increased (p<0.05) level in MLX interacting protein-like (Mlxipl) and stearoyl-CoA 9-desaturase (Scd1) mRNA (Figure 6A). This was accompanied by raised levels of peroxisome proliferator-activated receptor γ2 (PPARγ2) but not significantly higher (p=0.1702) and significantly decreased (p<0.01) in PPARγ in the GRβ-Ad mice when compared to the Vec-Ad. The PPARγ2 isoform has previously been shown to be regulated by GRβ [37]. The PPARγ2 isoform mediates gene pathways for lipid and glucose uptake, which could be one mechanism by which GRβ regulates hepatic fat content.

Figure 6. Gene analysis of lipid regulatory pathways in Vec-Ad and GRβ-Ad mice.

Real-time PCR quantification of genes involved in (A) lipogenesis, fatty acyl-CoA regulation, (B) fatty acid elongation, (C) PPARα targets, (D) eicosanoid metabolism, and (E) inflammatory pathways. [n =6–7 per group; *, p < 0.05; **, p < 0.01].

Hepatic fat content is regulated by de novo lipogenesis or by controlling the accumulation and catabolism of triglycerides. There were no significant differences in the enzymes that regulate fatty acyl-CoA, such as Acly or Dgat2 mRNA levels (Figure 6A). Because of the increase in some long-chain fatty acids (shown in Supplemental Figure S2B), we quantified mRNAs for genes that are elongation factors. The fatty acid elongase gene Elovl fatty acid elongase 3 (Elovl3) mRNA was significantly increased (p<0.05) (Figure 6B). However, no significant differences in the levels of the other elongation genes [Elovl1, Elovl2, Elovl4, Elovl5, Elovl6, and Elovl7]. The Elovl family of genes has been shown to be regulated by PPARα. We have previously demonstrated that GRβ suppresses PPARα transcriptional activity in the liver [24]. To determine genes regulated by PPARα, we measured Cyp4a10, G0s2, Abcd3, and Bdh1, well-known PPARα-target genes [38–41]. All four genes were lower with higher GRβ levels, but none were significantly decreased.

Since the levels of ceramides were reduced and the overall MAG levels were increased in the GRβ-Ad mice, we hypothesized that the MAGs can be used as precursors to generate eicosanoids, enabling arachidonic acid epoxygenase activity that drives inflammation. The Cyp2j6 and Cyp2c55 mRNA levels significantly increased (p<0.01) in the GRβ-Ad mice compared to the Vec-Ad controls. There were no significant differences in Cyp4a11 or Cyp4f14 mRNA (Figure 6D), which are cytochrome P450 enzymes that regulate monooxygenases that catalyze reactions involved in drug metabolism and synthesis of cholesterol, steroids, and other lipids.

Since there were differences in genes that regulate eicosanoid metabolism, and GRβ levels are increased with inflammation [30], we measured prostaglandin E synthase 2 (Ptges2) that produces inflammatory intermediates that stimulate the immune system. Ptges2 mRNA was significantly increased (p<0.05) in the GRβ-Ad mice when compared to the Vec-Ad (Figure 6E). Another marker associated with hepatic inflammation is tumor growth factor beta (TGFβ), which stimulates liver fibrosis [42]. Tgfb1 mRNA was significantly increased (p<0.05) in the GRβ-Ad mice compared to the Vec-Ad. Transcriptional and immune response regulator (Tcim) is another gene that controls transcriptional and immune responses, and the GRβ-Ad mice had Tcim mRNA levels that were significantly higher (p<0.05) than the Vec-Ad controls (Figure 6E). Analysis of the mRNA levels of the proinflammatory C-X-C motif chemokine ligand 9 (Cxcl9) showed significantly raised (p<0.05) mRNA levels in the GRβ-Ad compared to the Vec-Ad control mice. Thus, GRβ signaling increases inflammatory genes in the liver after 5 days. The conclusion from the data above is that glucocorticoid resistance by increasing GRβ levels changed the levels of genes that regulate eicosanoid synthesis and inflammation, inducing liver dysfunction (Graphical Abstract).

Discussion

The present study found that raising hepatic GRβ alters the liver lipidome, triggering the expression of inflammatory genes. The GRβ-induction of the Ptges2 gene may be a player in regulating the inflammatory gene response, as it encodes for the prostaglandin E synthase (PGES) protein that produces proinflammatory cyclooxygenases [43]. Another pro-inflammatory cytokine that was increased in the GRβ-Ad livers was TGFβ mRNA, which is well documented to induce liver fibrosis via stimulation of hepatic stellate cells (HSCs) [42, 44] One of the most notable changes was an upregulation in the levels of MAGs and CEs in the GRβ-Ad livers. NAFLD has been associated with changes in the lipid composition of the liver in humans [45]. An increase in MAGs has been positively associated with adverse outcomes of cardiovascular disease and type 2 diabetes [46]. However, very little is known about these lipids and how they impact NAFLD. The progression of NAFLD to non-alcoholic steatohepatitis [47], which includes hepatic inflammation and fibrosis, typically has elevated levels of CEs in the liver [48].

GRβ regulation of lipid synthesis might control the overall levels of the CEs, as the production of lipids is needed as it is fused with cholesterol to form them. We have shown that GRβ overexpression increased hepatic lipogenesis FASN protein level and lipid accumulation [24]. Here, we show an increased Mlxipl mRNA, which encodes for the carbohydrate response element binding protein (ChREBP) transcription factor known to induce glucose intolerance, lipogenesis, fatty liver, and metabolic dysfunction [49]. This is the first study to show that GRβ-induces ChREBP, which might be related to increased lipids and higher blood glucose levels in mice with increased GRβ in the liver [24]. Hwang et al. showed synergism between ChREBP and GR in increasing the fructose transporter GLUT5 in Caco-2BBE cells when the cells were treated with dexamethasone [50]. This study demonstrated that silencing ChREBP using a siRNA significantly decreased responsiveness to dexamethasone [50]. They validated these findings in the ChREBP KO mice [50]. GRβ has been shown to suppress GLUT1 mRNA expression [24], which transports glucose, galactose, mannose, glucosamine, and ascorbic acid within cells. Not surprisingly, the overexpression of GRβ in the livers of mice increased blood glucose levels [24]. However, whether GRβ regulates hepatic glucose uptake or export is yet to be established. GRβ may regulate glucose transport into the liver by increasing ChREBP and raising blood glucose levels. It was also shown to regulate hepatic gluconeogenesis enzyme mRNA expression (G6Pase and PEPCK), which was described as contributing to higher blood glucose levels [24]. More work is needed to determine the function of GRβ on hepatic glucose production, glucose uptake and export. Studies of GRβ in the ChREBP LKO mice by overexpression will be of value to determine if this co-signaling molecule is the culprit for GRβ increasing blood glucose levels.

We also showed increased levels of Cyp2j6 and Cyp2c55 mRNA in the GRβ-Ad mice compared to Vec-Ad, genes encoding for two enzymes that regulate arachidonic acid epoxygenase activity. The CYP2J6 enzyme has been shown to produce metabolites that suppress the PPAR signaling pathway [51]. In the gonadal adipose tissues of mice fed a high-fat diet, there was a decrease in CYP-derived fatty acid epoxides and a reduction in CYP2J6 [52]. Cyp2c55 metabolizes arachidonic acid to epoxyeicosatrienoic acids and was shown to be abundant in the liver [53]. The increase in Cyp2j6 and Cyp2c55 mRNA in the GRβ-Ad mice compared to Vec-Ad could be another mechanism that GRβ inhibits fat-burning β-oxidation and reduces PPARα expression and transcriptional activity, as we showed in [24].

The PPARα-target genes Abcd3, G0s2, and Cyp4a10 were lower in GRβ-Ad mice but not significantly compared to Vec-Ad controls. PPARα regulates select family members (Elovl2 and Elovl5) of the ELOVL class of genes that elongate fatty acids [54]. The PPARα regulated ELOVL genes were not significantly changed, but Elovl3 was significantly higher in the GRβ-Ad mice. Interestingly, Elovl3 levels were significantly raised when mice were treated with dexamethasone [55]. Male and female mice with ELOVL3 deletion (Elovl3^-/-) fed a high-fat diet were resistant to diet-induced obesity and exhibited a decrease in serum adiponectin, PPARγ expression, and lipogenesis enzymes FAS and DGAT2 [56]. PPARγ is an essential mediator of adiposity [57], and regulation of its levels or transcriptional activity reduces lipid accumulation [58–63].

Wu et al. demonstrated that PPARγ2 is regulated by GRβ binding to specific regions on the Pparg gene promoter that regulates alternative splicing of this isoform [37]. Others have shown that PPARγ increases Elovl3 [64]. GRβ-Ad mouse livers may have increased Elovl3 mRNA as PPARγ2 mRNA was raised ~2-fold but not significantly. The Elovl3^−/− mice have significantly higher levels of eicosanoids, but other triglycerides longer than 20 carbons were undetectable [65]. Qin et al. showed that Elovl3 levels are significantly higher in obese mice livers compared to lean [66]. However, they found that mice with a liver-specific knockout of Elovl3 still developed hepatic steatosis. The regulation of GRβ to control Elovl3 likely contributes to increased MAG and CE levels found in the livers, but more studies are needed to determine the significance.

Glucocorticoids drive lipid metabolism in the liver, and chronically high levels have been associated with fatty liver development [67, 68]. Mice with decreased circulating cortisol had lower body weights and attenuation of fatty liver [8]. Glucocorticoids have been shown to have positive and negative effects on inhibiting mitochondrial acyl-CoA dehydrogenases and PPARα transcriptional activity [24, 69]. This might be due to inflammation, which has been shown to increase the GRβ levels [30] and cause glucocorticoid resistance [13]. GRβ inhibition of PPARα resulted in reduced β-oxidation fat-burning [24], which caused fatty acid accumulation. These imply that chronic glucocorticoid therapy might induce glucocorticoid resistance by raising the GRβ levels as part of an intracellular negative feedback loop (discussed further in [13]).

Lipidomic analysis of human skin with topical glucocorticoid therapy showed decreased triacylglycerols and esterified ceramides [70]. Administration of high levels of corticosterone in rats showed decreased PC and PE lipid levels and increased LPC, CER, PA, and PG in the brain [71]. These indicate that glucocorticoids and GRβ-induced glucocorticoid resistance affect the specific lipids in a tissue-specific manner. In the present study, we also show a decrease in some ceramide lipid classes. Alternatively, treatment with the selective glucocorticoid receptor modulator Cort118335 increased very low-density lipoprotein (VLDL) production and lowered hepatic lipid accumulation [72]. The GRβ-Ad mice had significantly increased Sgms1 expression, which encodes for the sphingomyelin synthase 1 protein (known as SMS1, TMEM23, MOB) involved in converting PC to DAG. An increase in SMS1 has previously been implicated in NAFLD-hepatocellular carcinoma (HCC) patients, and the findings suggest a possible contribution to HCC [73, 74]. More studies are needed on the involvement of the lipid species in regulating hepatic function and how they may impact HCC.

Further analysis of the GR isoforms in different cell types is also needed. Hepatocytes regulate de novo lipogenesis and fat-burning β-oxidation [4], and there is another cell type called hepatic stellate cells (HSCs) that also store lipids and vitamin A as retinoic acid [75]. The HSCs are activated by TGFβ (and other hormones described further in [75]), and in this work here, we showed that GRβ increased its mRNA expression, which might indicate that GRβ induces liver fibrosis. However, GRβ also increases PPARγ expression, which has been shown to inhibit HSC activation [76, 77]. Once TGFβ activates the HSCs [44], they proliferate, expel their lipids, and produce and shed collagen, which causes liver fibrosis [42]. A recent finding by Bates et al. showed that FOXS1 is a transcription factor essential for activating HSCs and can serve as a liver fibrosis biomarker [44]. Whether the GR isoforms are expressed in HSCs and if they control lipid accumulation or activation is unknown. Studies to determine the function of the isoforms and whether GRβ regulates PPARγ to inhibit liver fibrosis is a possibility that needs further investigation.

In conclusion, we provided evidence that GRβ contributes to changes in lipid classes and alterations in the types of lipids that accumulate to cause liver dysfunction. Increasing GRβ levels caused significant alterations in the overall MAGs and CEs hepatic lipid pools, but this was not observed for the specific sub-classes of these lipids. A longer duration of high GRβ levels would likely increase several sub-classes, but this has yet to be performed. Studies using GRβ knockout mice [78] in the future might reveal how this isoform regulates these pathologies. A GRβ-specific inhibitor could be tested for cell studies to validate GRβ’s function in humans and whether its signaling mechanism requires PPARγ [79, 80]. Future studies are also needed to understand the long-term effects of GRβ and glucocorticoid resistance and how they increase MAGs and CEs. The work here indicates that GRβ could be a therapeutic target for controlling hepatic lipids, but more work is needed to understand its mechanisms.

Materials and methods

Animals

This study’s experimental procedures and protocols conform to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The Institutional Animal Care and Use Committee at the University of Toledo approved them. Eight-week-old male C57BL/6J mice were purchased from Jackson Labs (Bar Harbor, ME, USA). Mice were individually caged in specific pathogen-free enclosures with a 12-h light/dark cycle at 22 to 24 °C and randomly assigned to either empty Vec-Ad or GRβ-Ad. All mice were fed a normal diet with full access to tap water. Mice were injected with adenovirus, as described in [24]. The mice were fed a normal chow diet ad libitum for 5 days, as we previously reported [24]. After 5 days, the livers were harvested for lipidomic and gene analysis. The GRβ levels were significantly increased, but not the GRα isoform (Supplemental Figure 1).

Quantitative Real-Time PCR

Gene expression was measured via real-time PCR, as previously described by [42]. Total RNA was extracted from the vec-Ad and GRβ-Ad infected mouse livers. In short, the tissue was lysed using a Qiagen Tissue Lyser LT (Qiagen, Hilden, Germany) and extracted using a 5-Prime PerfectPure RNA Tissue Kit (Fisher Scientific Company, LLC). RNA was quantitated using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA), and cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA, USA). PCR amplification of the cDNA was performed by quantitative real-time PCR using TrueAmp SYBR Green qPCR SuperMix (Alkali Scientific, Fort Lauderdale, FL, USA) for gene-specific primers (Table 1) and as previously described [24]. Samples were measured using a Quantstudio6 Pro (Thermo Fisher Scientific). The thermocycling protocol consisted of 5 min at 95 °C, 40 cycles of 15 s at 95 °C, 30 s at 60 °C, and an extension at 72 °C depending on the product size, finishing with a melting curve ranging from 60 to 95 °C to allow distinction of specific products. Normalization was performed in separate reactions with primers to 36B4.

Table 1.

List of all real-time PCR primers.

Gene	Forward Primer	Reverse Primer
Asah1	CGGAAGGAAAGATGCCCAGT	CAGACTTCTTGCCTCCCAGG
Asah2	CTCTTCGGGACCAGCTCTTG	TTCGTCCCGACATGGTTGTT
Cers1	GGTTCTGGTTCCGCCTCTAC	CCATCAGGAGCAACAGCAGA
Cers3	CTTTCGAGGGTGGGTCTCTG	AGCCAGTATCTCTCCGACCA
Cers5	GTGGAACCCAATGACACCCT	CGTTGGAGGCTTGTCCTGAT
Degs1	AGACCGGATGGCTCCTCAAA	AGCCAGAGTCATGGAGTGGT
Sgms1	TATGGGTGGACACTTGGGCT	AGCCTGTGTGGTCTATGGTG
Sgms2	GTGAGGGCCTGGGTATCTTC	ATGCTCTCTTCGGTGCCTTT
Smpd1	GTGGGACTCCTTTGGATGGG	CCCAAAGAACCGTGGAGTCA
Smpd2	ACTTCAGAAGCGGGATGATAGG	ACGTAGGCATTGAGCACCAG
Gba1	TGGAGAGAAGTGTGCTGGTG	GGTAAGGTCACGGGGTCAAG
Galc	ACCCGCACAATGGCTAACA	ACAACAATAAGGGCACCGCA
Ugcg	GTGTGACGGGGATGTCTTGT	GAAAACCTCCAACCTCGGTC
Ugt8a	CCGAAGGACGCGCTATGAAG	AGGCCGATGCTAGTGTCTTG
Mlxipl	CCAGAGTCCCCGCAGGAT	GACCTGGGAGGAGCCAATGT
PPARγ2	AAACTCTGGGAGATTCTCCTGTTG	GAAGTGCTCATAGGCAGTGCA
Scd1	GTACCGCTGGCACATCAACT	AACTCAGAAGCCCAAAGCTCA
Acly	ATGCCCCAAGATTCAGTCCC	GGTATGTCGGCTGAAGAGGG
Dgat2	ACTGGAACACGCCCAAGAAA	GTAGTCTCGGAAGTAGCGCC
Elovl1	GGGCAGGAGTCTCAAAGAGC	CAGCCTCCATCCTGGCTAAG
Elovl2	GACAGCGCATCGCGGC	GCCATATCGAGAGCAGGTACG
Elovl3	CTGAGGGTGTGTGGAAGGAC	TTCCTTTCAGAGGCTTGGGTC
Elovl4	ACAATGAGCCGAAGCAGTCA	CTCTCCTTTTGGCTTCCCGT
Elovl5	CAGCTTGCTTCTGTTCCCG	AGTGACGCATCGAAATGTTCC
Elovl6	TGAACAGGGAGGAAGGGCTA	GTGTGAGGTCGAACAGGGAG
Elovl7	GGCTGAGACCCACAGATTGAG	CTGTCAGTCCAAGCCTTCTTTC
Cyp4a10	AGCCACAAGGGCAGTGTTCAGG	CCAAGCGGCCATTGGAAGAAAG
G0s2	GCCACCGAATCCAGAACTGA	TTGATTGCTCGCACAGCCTA
Abcd3	CGGGAAGCCAAAGTCCACG	CGGGGGAAACTGGGAGGATA
Bdh1	ACATCAAGCCCGGAGAGATT	TGCTTCCTTCTGGTGGCTCT
Cyp2j6	GCAAAGGGGAGTGGACAGAA	GGCTAACAAGGAGCCGGTAG
Cyp22c55	ATCCCCAACAATGTGCCCC	GTCATCATGCAGCACAGAAGTC
Cyp4a11	ACAAGGACCAGCTTTGAGGG	GATATGGGCAGACAGGAAGGG
Cyp4f14	AGCTCTGGTTATTCCCCTCAC	GGGTGAAGAGGTACAGGAGTG
Ptges2	CAGGTGGTGGAGGTGAATCC	CTGCCCTGAAACCAGGTAGG
Tgfb1	ACTGGAGTTGTACGGCAGTG	GGGGCTGATCCCGTTGATTT
Tcim	TTCAGCAGGCGTTGAGAACT	CAGTGTTCTGGTCAGTGCCT
Cxcl9	CTGGGCAGAAGTTCCGTCTT	ACCAGCAGCACAAAAACCAC
GRα	AAAGAGCTAGGAAAAGCCATTGTC	TCAGCTAACATCTCTGGGAATTCA
GRβ	AAAGAGCTAGGAAAAGCCATTGTC	CTGTCTTTGGGCTTTTGAGATAGG

Lipidomics Data Acquisition

Lipidomics was done as previously described in [81, 82]. Acidified organic solvents were used to extract lipids, as described in [83–85]. Stable isotope-labeled lipid-class specific internal standards, SPLASH^®LIPIDOMIX^® Mass Spec (Avanti Polar Lipids, Alabaster, AL, USA), were added at the start of extraction. Reconstituted dried lipid residues were analyzed. A Shimadzu Nexera UHPLC system coupled with an AB Sciex 6500+ Q-Trap linear ion trap/triple quadrupole mass spectrometer instrument was used. Employing method adaptations from the literature, HILIC chromatography with a Phenomenex Luna Silica Column and guard column of the same material lipids were separated and detected in multiple reaction monitoring modes previously described in [86]. Method optimization used a pooled liver lipid sample to exclude lipid species that were not reliably detected. Lipids were analyzed in experimental samples using the final optimized method, and data for three technical replicates was collected. Data were analyzed using AB Sciex MultiQuant software (Framingham, MA, USA) for peak finding and integration. The raw peak areas were normalized for recovery of the appropriate internal standards. Lipid species with coefficients of variation greater than 20% in the technical replicates were excluded from the final report.

Lipidomics Analysis

The LipidSig differential analysis tool used the web-based platform [51] to analyze the lipidomics data. The lipidomics data were normalized to the sample weight, and duplicate values were removed before analysis (n = 5 for Vec-ad and n = 5 for GRβ-Ad). The normalized data and sample annotations were uploaded to LipidSig for analysis. Lipid species missing from more than half of the samples were excluded from the analysis. A t-test method using the p-value was used to identify lipid species significantly different between samples. For global visualization of altered lipid species, hierarchical clustering was performed with a Pearson distance measure and a complete clustering method. Only significantly changed lipid species were analyzed using hierarchical clustering. Differential lipid species were shown using a relative signal intensity range and visualized via a heatmap. The BioPAN tool provided by Lipid Maps was used to create the network analysis through the web-based platform [87].

A second analysis was performed using LipidSig and LipidR (R package version 2.16.0) [88]. Data was normalized to body weight before quality control and analysis. Principle component analysis was used to visualize sample grouping. Total lipid intensity and lipid intensity by class were used as quality control measures. No outliers were removed from the analysis. Univariate analysis of lipid species comparing GRβ-Ad to Vec-Ad was conducted using limma (Version 3.58.1). An adjusted p-value of <0.05 was used to determine significance.

Statistical Analysis

Real-time PCR data was graphed and analyzed with Prism 10 (GraphPad Software, San Diego, CA, USA) using an analysis of variance and Tukey’s post-hoc test compared to the group means. Results are expressed as mean ± S.E.M. A two-tailed, unpaired t-test was used to determine statistical significance when comparing two groups. p-values of 0.05 or smaller were considered statistically significant.

HIGHLIGHTS.

• Non-alcoholic fatty liver disease may be a state of glucocorticoid resistance.

• GRβ-induced glucocorticoid resistance regulates liver lipid species generation.

• Hepatic glucocorticoid resistance activates genes for eicosanoid synthesis.

• Raising the hepatic GRβ level increases inflammatory gene expression.

Data Availability

Raw data is available on GitHub (*information to be added).