Chronic Glucocorticoid Exposure Induced an S1PR2-RORγ Axis to Enhance Hepatic Gluconeogenesis in Male Mice Free

It is well established that chronic and excess glucocorticoid exposure, through multiple mechanisms, including prolonged stress, long-term glucocorticoid use for treating inflammatory and autoimmune disease, depression, chronic stress, or Cushing syndrome, cause hyperglycemia and insulin resistance (1–3). Glucocorticoids act through their intracellular glucocorticoid receptor (GR), promoting transcription of genes encoding enzymes in hepatic gluconeogenesis, such as phosphoenolpyruvate carboxykinase (Pck1) and the catalytic subunit of glucose 6 phosphatase (G6pc1) (1,4,5). Studies have shown that glucocorticoids induce additional mechanisms to enhance hepatic glucose production. We previously showed that prolonged glucocorticoid treatment increases ceramide production, a process that requires angiopoietin-like 4 (Angptl4), a glucocorticoid-induced gene encoding a secreted protein that inhibits lipoprotein lipase activity and promotes adipose tissue lipolysis (6–8). Ceramide-initiated signaling in the liver activates protein phosphatase 2A (PP2A) and protein kinase Cζ (PKCζ) to suppress Akt, a key component in insulin signaling, causing insulin resistance and elevating hepatic gluconeogenesis (9).

In addition to ceramides, hepatic sphingosine-1-phosphate (S1P) levels are also augmented by glucocorticoid treatment (9). Ceramides can be converted to sphingosine, which can be further metabolized to S1P. S1P is a signaling molecule whose role in glucocorticoid-mediated metabolic regulation has not been reported. S1P can exert its actions intracellularly and extracellularly (10,11). Here, we explored the potential role of extracellular S1P in chronic glucocorticoid exposure–induced glucose homeostasis disorder. Extracellular S1P binds to its cognate membrane G-protein–coupled receptors S1PR1–5 to exert its actions (12). S1PR1–3 are highly expressed in the liver (13). We found that hepatic knockdown of S1PR1, 2, and 3 in male mice all modulated glucocorticoid-regulated glucose homeostasis. Specifically, reducing hepatic S1PR2 expression blunted glucocorticoid-induced glucose intolerance and hepatic gluconeogenesis. RNA sequencing (RNA-seq) found that glucocorticoid suppression of RAR-related orphan receptor γ (Rorc) (called RORγ henceforward) gene expression was enhanced in glucocorticoid-treated hepatic S1PR2 knockdown mouse liver. Because RORγ has been previously shown to positively regulate hepatic gluconeogenic gene transcription (14), we further analyzed its role in chronic glucocorticoid exposure–enhanced gluconeogenesis.

Andras Nagy (Samuel Lunenfeld Research Institute, Mount Sinai Hospital) and Jeff Gordon (Washington University) provided Angptl4^+/+ and Angptl4^−/− mixed C57BL/6J:129 SV background male mice, which were cohoused as previously described (15). All experiments were approved by the University of California, Berkeley (AUP-2014-07-6617). This study was performed entirely in male mice.

Adeno-associated virus serotype 8 (AAV8) expressing shRNA scramble, shRNA targeting S1PR1, S1PR2, S1PR3, RORγ, and green fluorescent protein (GFP), and mouse RORγ cDNA were purchased from Vector Biolabs (Malvern, PA) and were injected in the tail vein (3 × 10¹¹ genome copies/mouse). The shRNA sequences are provided in Supplementary File 1.

Dexamethasone (Dex) water 0.84 mg/kg body weight (PHR 1768; Sigma) was previously described (15). AAV mice were switched to Dex-supplemented water 2 days after injection for 4 weeks.

Mice were fasted for 16 h for intraperitoneal glucose tolerance tests (IPGTTs) (1 mg/kg d-glucose, 50-99-7; Sigma) and pyruvate tolerance tests (PTTs) (2 mg/kg sodium pyruvate, P2256; Sigma) and fasted 2 h for insulin tolerance tests (1 unit/kg of insulin, 5 mL I0516; Sigma). A Contour Next glucometer was used to measure blood glucose.

Fasted blood was collected in EDTA tubes (16.44.100; Sarstedt). CO₂ euthanasia was performed at 3–5 h into the light cycle (10:00 a.m.–12:00 p.m.). Tissues were snap frozen in liquid nitrogen and stored at −80°C.

Plasma insulin, liver phosphorylated Akt at serine 473 (pAkt) and Akt levels, and liver and plasma S1P levels were measured using ELISA. The detailed protocols are provided in Supplementary File 1.

RNA-seq was performed by BGI Americas (Cambridge, MA). The protocol of the analysis of RNA-seq is provided in Supplementary File 1.

Tissues were prepared for SDS-PAGE as previously described (15). Antibody information is provided in Supplementary File 1. Images were analyzed using ImageJ software.

Gene expression was performed as previously described (15), and Rpl19 was used for internal normalization. Primers are listed in Table 1.

Liver chromatin immunoprecipitation (ChIP) was performed as previously described (15) with the following modifications: 10 strokes with dounce homogenizer, sonication 60% amplitude (5 min total), and 10-s bursts (40-s rests). Four micrograms of the following antibodies were used: purified rabbit IgG (whole molecule) control (A01008; GeneScript), GR (IA-1 [a polyclonal rabbit antibody against human GR amino acids 84–112 QPDLSKAVSLSMGLYMGETETKVMGNDLG]; a gift from Dr. Miles Pufall, University of Iowa), and RORγ (14-6988-82; Invitrogen). Forty microliters of 25% protein A/G sepharose beads (Santa Cruz Biotechnology) were used (15).

H4IIE cells were cultured in DMEM with 10% FBS and serum starved overnight before treatment. For 4 h, 100 nmol/L Dex (D1756; Sigma), 100 nmol/L Ly3214996 (501932152; MedChemExpress), and/or 2.5 μmol/L XY018 were added. During the last 2 h, 0.5 μmol/L CYM-5520 was added (5418/10; R&D Systems). For transfection, H4IIE cells were plated at 0.5 × 10⁶ cells/well and cultured in DMEM containing 5% FBS. The detailed protocol is provided in Supplementary File 1.

GraphPad Prism 9.0 software was used for the data analysis. Data are presented as mean with SD and SEM using Student t test, one-way ANOVA, and two-way ANOVA with Fisher least significant difference post hoc and Benjamini and Hochberg false discovery rate correction tests as indicated.

All data generated or analyzed during this study are included in the published article and its online supplementary files. RNA-seq data are deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession no. GSE150626).

Wild-type (WT) male mice infected with AAV8 expressing scramble shRNA (AAV8-shScr) were treated with or without Dex (a synthetic glucocorticoid) for 4 weeks. Dex treatment resulted in glucose intolerance (Fig.1A). As previously reported, hepatic, but not plasma, S1P levels were elevated by Dex treatment (9) (Fig.1B). To learn whether extracellular S1P is involved in glucocorticoid-regulated glucose homeostasis, WT mice were infected with AAV8 expressing shRNA targeting S1PR1 (AAV8-shS1PR1), AAV8-shS1PR2, or AAV8-shS1PR3. These mice, along with AAV8-shScr mice, were treated with Dex for 4 weeks. The expression of S1PR1, 2, and 3 was reduced in the liver but not in gastrocnemius muscle, inguinal white adipose tissue, or brown adipose tissue (Fig.1C and Supplementary Fig. 1A). Hepatic S1PR1 knockdown had no effect, whereas hepatic S1PR3 knockdown further exacerbated Dex-induced glucose intolerance (Fig.1D). In contrast, hepatic S1PR2 knockdown blunted the Dex response (Fig.1D). Similarly, hepatic S1PR1 knockdown had no effect on Dex-induced fasting hyperglycemia, while hepatic S1PR3 knockdown worsened and hepatic S1PR2 knockdown improved Dex-induced fasting hyperglycemia (Fig.1E). Compared with AAV8-shScr mice, fasting plasma insulin levels were lower in AAV8-shS1PR2 mice and higher in AAV8-shS1PR1 mice, but fasting insulin levels were not different between AAV8-shS1PR3 and AAV8-shScr mice (Fig.1F). Overall, these results demonstrate that reducing hepatic S1PR1, 2, and 3 expression all affected glucose homeostasis in the presence of Dex. Importantly, reducing their hepatic expression did not affect plasma ALT levels (Supplementary Fig. 1B). Thus, hepatic S1PR1, 2, and 3 knockdowns did not cause liver injury. Of note, we found that without Dex treatment, AAV8-shS1PR2 mice were also more glucose tolerant than AAV8-shScr mice (Fig.1G), though there were no differences in fasting plasma insulin levels (Fig.1H). Next, we specifically investigated the mechanism underlying the effects of altered hepatic S1PR2 signaling in the context of chronic glucocorticoid exposure–induced glucose disorder.

Because hepatic S1P induction by Dex depends on the presence of ANGPTL4 in male mice (9), we tested whether the effects of S1PR2 knockdown on glucose tolerance also requires ANGPTL4. As previously reported, Dex-induced glucose intolerance was compromised in Angptl4^−/− mice (9) (Supplementary Fig. 2). In contrast to the observations in WT mice, hepatic S1PR2 knockdown did not affect glucose tolerance in Dex-treated Angptl4^−/− mice (Supplementary Fig. 2), and a trend toward improved glucose tolerance was observed in Angptl4^−/− mice not treated with Dex (Supplementary Fig. 2). These results demonstrate that while ANGPTL4 is required for the effects of S1PR2 signaling in the context of chronic Dex treatment–induced glucose intolerance, it perhaps is not required for the effects of S1PR2 signaling with respect to the regulation of basal glucose tolerance.

We performed an insulin tolerance test on AAV8-shScr and AAV8-shS1PR2 mice treated with or without Dex. Without Dex, these mice had similar insulin tolerance (Fig.2A). Dex treatment induced insulin intolerance, but, interestingly, hepatic S1PR2 knockdown did not affect Dex-induced insulin intolerance (Fig.2A). This observation is consistent with analyses of the ratio of pAkt and total Akt (pAkt/Akt), an indicator of insulin signaling activity (16) in liver, gastrocnemius muscle, and epididymal white adipose tissue in AAV8-shScr and AAV8-shS1PR2 mice treated with Dex and injected with insulin or PBS. pAkt/Akt levels were similar in all three tissues examined between AAV8-shScr and AAV8-shS1PR2 mice (Fig.2B). Thus, reducing hepatic S1PR2 signaling did not affect Dex-modulated insulin signaling in these tissues.

We next performed a PTT to monitor the hepatic gluconeogenic capacity of AAV8-shScr and AAV8-shS1PR2 mice treated with or without Dex. Dex treatment increased fasting glucose levels and gluconeogenesis in AAV8-shScr mice (Fig.2C). Hepatic S1PR2 knockdown reduced gluconeogenic capacity in both control and Dex-treated mice (Fig.2C). Gene expression analysis revealed that hepatic S1PR2 knockdown reduced the expression of gluconeogenic genes Pck1 and G6pc1 (Fig.2D). Notably, without Dex, Pck1 and G6pc1 expressions were similar between these mice (Fig.2D). These indicate that S1PR2 signaling is involved in Dex-stimulated and basal gluconeogenesis. However, the mechanisms underlying S1PR2’s role in these two processes are different.

One potential mechanism to explain the lower induction of gluconeogenic genes by Dex in hepatic S1PR2 knockdown mouse liver would be through reduced Dex-induced recruitment of GR to the GREs of gluconeogenic genes. We performed GR ChIP to examine this model. Without Dex, GR occupancy on the GREs was minimal in AAV8-shScr and AAV8-shS1PR2 mouse liver (Fig.2E). Dex treatment significantly increased GR recruitment to the GREs of Pck1 (4) and G6pc1 (17) genes in AAV8-shScr mouse liver (Fig.2E). However, GR recruitment was significantly lower in Dex-treated AAV-shS1PR2 mice than Dex-treated AAV8-shScr mice (Fig.2E).

To probe potential mechanisms for the effects of S1PR2 knockdown, we performed RNA-seq to identify differentially expressed genes in Dex-treated AAV8-shScr versus Dex-treated AAV8-shS1PR2 mouse liver. Using a twofold difference and adjusted P value of 0.05 as the cutoff, we found that 238 genes had lower expression, and 182 genes were higher in Dex-treated AAV8-shS1PR2 versus Dex-treated AAV8-shScr mouse liver (Fig.3A). A list of these genes is presented in Supplementary File 2. Gluconeogenic genes were among the list of downregulated genes in AAV8-shS1PR2 liver, including G6pc1, fructose bisphosphatase 1 (Fbp1), and tyrosine aminotransferase (Tat). Pck1 was not scored based on the cutoff standard, but conventional real-time PCR analysis clearly showed a lower Pck1 expression in the liver of Dex-treated AAV8-shS1PR2 mice (Fig.2D). Moreover, RNA-seq also found that S1PR2 knockdown reduced the expression of Rorc, which encodes a nuclear factor that can potentiate Pck1 and G6pc1 transcription (14). Gene ontology analysis found that genes encoding the epoxygenase p450 pathway, lipid, steroid, acetyl-CoA, and glycolytic metabolic processes were highly represented among S1PR2 knockdown–reduced genes (Fig.3B). In contrast, genes involved in immune response, antiviral defense, estrogen responses, lipid transport, and lipoprotein metabolism were highly represented among S1PR2 knockdown–induced genes (Fig.3B). Thus, S1PR2 signaling participates in many facets of glucocorticoid actions. The role of RORγ in glucocorticoid-regulated hepatic gluconeogenesis was of particular interest since this had not been previously reported. Immunoblotting showed that without Dex, S1PR2 knockdown did not affect RORγ expression (Fig.3C). While Dex treatment reduced RORγ expression in the liver AAV8-shScr mice, Dex suppression of RORγ expression was even more pronounced in the liver of AAV8-shS1PR2 mice (Fig.3C). Similarly, nuclear levels of RORγ were suppressed in Dex-treated AAV8-shScr and further suppressed in AAV8-shS1PR2 mice (Supplementary Fig. 3A). RORα has also been shown to activate the Pck1 gene (18,19). However, protein levels of RORα were not affected by hepatic S1PR2 knockdown (Supplementary Fig. 3B).

To explore whether RORγ is involved in Dex-regulated gluconeogenic gene expression, H4IIE rat hepatoma cells were treated with 1) DMSO (vehicle), 2) Dex, 3) CYM-5520 (an agonist of S1PR2) (20), 4) Dex plus CYM-5520, 5) XY018 (an RORγ antagonist) (21), 6) Dex plus XY108, 7) CYM-5520 plus XY018, and 8) Dex plus CYM-5520 plus XY108 for 4 h, then Pck1 and G6pc1 expression were analyzed. Pck1 and G6pc1 expression were induced by Dex but not by CYM-5520 (Fig.4A). However, CYM-5520 enhanced Dex-induced Pck1 and G6pc1 expression (Fig.4A). XY018 did not affect basal Pck1 and G6pc1 expression but suppressed both Dex and Dex plus CYM-5520–induced Pck1 and G6pc1 expression (Fig.4A). Thus, RORγ activity was required for maximal Dex and S1PR2 signaling response on gluconeogenic gene expression.

We next tested RORγ’s role in Dex-induced glucose intolerance in vivo. WT male mice were treated with or without Dex for 4 weeks. For the final 2 weeks, one-half of each group was treated with PBS or XY018. Dex-treated mice with PBS injections were more glucose intolerant than mice that only received PBS injections (Fig.4B). XY018 treatment alleviated Dex-induced glucose intolerance but did not affect basal glucose tolerance (Fig.4B). PTTs showed that Dex treatment augmented glucose production in WT mice, whereas XY018 alone had no effect (Fig.4C). Dex-treated mice that also received XY018 had a lower gluconeogenic capacity than mice treated with Dex alone (Fig.4C). Not surprisingly, XY018 reduced Pck1 and G6pc1 expression in Dex-treated WT mice (Fig.4D). XY018 did not affect Pck1 and G6pc1 gene expression in mice not treated with Dex (Fig.4D). These results were consistent with the results observed in H4IIE cells.

We performed liver ChIP in these mice to monitor the GR and RORγ occupancy at the RORγ response elements (ROREs) (18) and GREs (4) of Pck1 and G6pc1 (Fig.4E). Without treatment, GR was modestly recruited to the GREs (Fig.4E). This occupancy was likely due to endogenous corticosterone. RORγ, however, was not found at the ROREs (Fig.4E). Dex treatment increased both GR and RORγ recruitment to the GRE-RORE regions of these two genes (Fig.4E). XY018 did not affect basal GR and RORγ occupancy but attenuated Dex-induced GR and RORγ recruitment (Fig.4E).

WT mice were infected with AAV8-shScr or shRNA targeting Rorc (AAV8-shRORγ) and treated with Dex. Immunoblotting confirmed a significantly reduced RORγ expression in the liver, but not brown adipose tissue or gastrocnemius muscle, in AAV8-shRORγ mice (Fig.5A). Dex-treated AAV8-shRORγ mice showed a trend toward better glucose tolerance than AAV8-shScr mice (Fig.5B). Moreover, PTT showed that hepatic gluconeogenesis was significantly lower in Dex-treated AAV8-shRORγ mice (Fig.5C). Fasting plasma insulin levels were also lower in AAV8-shRORγ mice (Fig.5D). Interestingly, while Pck1 expression was reduced in livers of AAV8-shRORγ mice, there were no differences in G6pc1 expression (Fig.5E). Nonetheless, the results from hepatic RORγ knockdown mice generally agreed with the observation from the RORγ antagonist studies in Fig. 4. Notably, hepatic RORγ knockdown did not affect plasma ALT levels (Supplementary Fig. 4B).

To further test the functional interaction between RORγ and GR, a rat Pck1 promoter luciferase reporter was transfected into H4IIE cells along with expression plasmids of human GR and human RORγ (hRORγ). Without hRORγ overexpression, Dex treatment induced the reporter activity (Fig.5F). hRORγ overexpression did not affect the reporter activity. However, transfecting 300 ng of hRORγ expression plasmid significantly potentiated the Dex response (Fig.5F). We tested whether an RORγ agonist can further potentiate this response. In separate experiments, H4IIE cells were transfected with a rat Pck1 promoter reporter, expression plasmids of human GR, and 300 ng hRORγ. Twenty-four hours after transfection, cells were treated with DMSO, Dex, LYC-55716 (an RORγ agonist) (22), and Dex plus LYC-55716. Interestingly, 100 nmol/L LYC-55716 treatment alone showed a trend toward increased basal reporter activity. Both 50 nmol/L and 100 nmol/L LYC-55716 significantly enhanced the Dex response (Fig.5G). Overall, these results suggest that RORγ enhances GR activation of rat Pck1 promoter.

Next, we tested whether RORγ overexpression in hepatic S1PR2 knockdown mice restored Dex-induced glucose intolerance and gluconeogenesis. Male WT mice infected with AAV8-shS1PR2 were treated with or without Dex. One week later, they were infected with AAV8 expressing GFP (AAV8-shS1PR2-GFP) or mouse RORγ (AAV8-shS1PR2-RORγ). Dex treatment was continued for 3 more weeks, and IPGTT and PTT were performed. Western blotting validated liver specificity of RORγ overexpression (Fig.6A). Dex treatment caused glucose intolerance and increased gluconeogenesis in AAV8-shS1PR2-GFP mice (Fig.6B and C). Overexpression of RORγ further enhanced these phenotypes (Fig.6B and C). Hepatic Pck1 and G6pc1 expression was also higher in Dex-treated AAV8-shS1PR2-RORγ mice (Fig.6D). Without Dex, RORγ overexpression did not increase glucose intolerance or gluconeogenesis of AAV8-shS1PR2 mice (Fig.6B and C). Thus, RORγ is not involved in the regulation of basal glucose tolerance and gluconeogenesis regulated by S1PR2 signaling.

The results of ChIP assays showed that the Dex effect on GR recruitment to the Pck1 and the G6pc1 GREs in AAV8-shS1PR2-GFP mice was minimal (Fig.6E), similar to the observation in AAV8-shS1PR2 mice (Fig.2E). Overexpression of RORγ enhanced GR recruitment (Fig.6E). RORγ was not present in the Pck1 and G6pc1 ROREs in AAV8-shS1PR2-GFP mice (Fig.6E). Surprisingly, without Dex, RORγ was also not present in the Pck1 and G6pc1 ROREs in AAV8-shS1PR2-RORγ mouse liver (Fig.6E). RORγ was only recruited to the ROREs in AAV8-shS1PR2-RORγ mouse liver upon Dex treatment (Fig.6E).

Overall, these results demonstrate that upon chronic Dex treatment, RORγ is recruited to the ROREs of Pck1 and G6pc1 genes and is required for maximal GR recruitment to the GREs. This augments the expression of hepatic gluconeogenic genes, which enhances gluconeogenesis and glucose intolerance (Fig. 7).

In this study, we showed that chronic Dex treatment increased liver S1P levels in male mice and that reducing the expression of hepatic S1PR1, 2, and 3 all modulated the effect of Dex treatment on glucose metabolism, though with distinct effects. Hepatic S1PR3 knockdown exacerbated Dex-induced glucose intolerance and fasting hyperglycemia but had no effect on fasting hyperinsulinemia. In contrast, hepatic S1PR1 knockdown did not affect glucose tolerance and fasting glucose levels but increased fasting insulin levels. We speculate that S1PR3 signaling negatively regulates chronic glucocorticoid-induced hepatic gluconeogenesis, whereas S1PR1 signaling attenuates chronic glucocorticoid-induced insulin resistance. These models will be examined in our future studies. We also found that hepatic S1PR2 knockdown attenuated Dex-induced glucose intolerance. However, hepatic S1PR2 knockdown had no effect on insulin tolerance, and Dex effect on insulin stimulated Akt activity in the metabolic tissues tested. PTTs showed that S1PR2 knockdown reduced Dex-induced gluconeogenesis. Therefore, prolonged Dex treatment induced hepatic S1PR2 signaling to enhance gluconeogenesis without affecting insulin signaling. This report focuses on elucidating the role of hepatic S1PR2 in chronic glucocorticoid-induced glucose disorders.

RNA-seq identified Rorc as a gene whose expression was reduced in the liver by hepatic S1PR2 knockdown in Dex-treated mice. RORγ was previously shown to potentiate transcription of gluconeogenic genes (14), but its role in glucocorticoid’s effects has not been explored. We found that RORγ overexpression in hepatic S1PR2 knockdown mouse liver restored Dex-induced glucose intolerance and increased gluconeogenesis and gluconeogenic gene expression. In contrast, RORγ antagonist XY018 treatment attenuated Dex-induced glucose intolerance and reduced gluconeogenic gene expression and hepatic gluconeogenesis. Similarly, Dex-induced hepatic gluconeogenesis was lower in hepatic RORγ knockdown mice, and they were trending toward having better glucose tolerance under Dex treatment. These results demonstrated RORγ’s participation in chronic Dex-induced hepatic gluconeogenesis. RORγ overexpression in mouse liver did not affect Pck1 and G6pc1 basal expression or basal glucose and pyruvate tolerance. Similarly, XY018 did not affect basal glucose tolerance or hepatic gluconeogenesis. The studies of Pck1 and G6pc1 mRNA in H4IIE cells were consistent with these in vivo results. A previous report showed that liver-specific Rorc knockout mice had lower gluconeogenic activity and gluconeogenic gene expression during the daytime (14). Our results did not exactly contradict these results. Perhaps the dose and/or duration of XY018 we administered was sufficient to improve Dex-induced glucose intolerance but not enough to affect basal glucose tolerance in control mice. In fact, in rat Pck1 promoter reporter assay, we found that RORγ agonist LYC-55176 was trending toward increasing the reporter gene activity. Interestingly, reporter gene studies showed that even without adding LYC-55716, overexpressing RORγ was able to potentiate Dex response, and LYC-55176 further enhanced this effect. This suggests that endogenous ligands of RORγ are enough to potentiate the Dex response to a certain degree and that more agonists further enhanced the effect.

In any case, because hepatic S1PR2 knockdown improved basal glucose tolerance and reduced basal gluconeogenesis, we propose that S1PR2 signaling exerts at least two mechanisms that enhance chronic glucocorticoid exposure–induced hepatic gluconeogenesis. The first mechanism is through suppressing glucocorticoid-reduced RORγ expression. The other is to regulate basal gluconeogenesis, for which the mechanisms remain unclear. Although S1PR2 signaling did not regulate basal gluconeogenic gene expression, other mechanisms could modulate the rate of gluconeogenesis (23). Notably, the regulation of basal glucose tolerance by S1PR2 signaling was likely independent of the presence of ANGPTL4. In contrast, ANGPTL4 was required for S1PR2 signaling to enhance glucocorticoid-induced glucose intolerance. This is not surprising because ANGPTL4 was required for Dex-induced hepatic S1P production (9). A previous study showed that basal glucose tolerance was not different between WT and S1pr2 null mice fed a chow diet (24). Here, we specifically reduced S1PR2 in the liver of male mice. Perhaps S1PR2 signaling in different tissues has distinct roles on whole-body glucose homeostasis. In S1pr2-null mice, S1PR2 signaling in certain tissues somehow offsets its role in liver gluconeogenesis.

The recruitment of GR to Pck1 and G6pc1 GREs was decreased in hepatic S1PR2 knockdown mouse liver. When RORγ was overexpressed in the liver of these mice, it restored GR occupancy. In contrast, XY018 attenuated Dex-induced GR occupancy. Thus, mechanistically, RORγ assists GR recruitment to the GREs of gluconeogenic genes. Previous studies of glucocorticoid activation of rat Pck1 gene found that HNF4 and FoxA1/2 (also known as HNF3α/β) bind upstream of GREs (25,26) and enhance GR binding to the GREs (27). The RORE in the mouse Pck1 gene overlaps with another accessory element required for maximal glucocorticoid response located downstream of GREs that has been previously shown to bind to chicken ovalbumin upstream promoter transcription factor (COUP-TF) (28). The fact that XY018 reduced GR and RORγ recruitment to the GRE-ROREs of gluconeogenic genes suggests that RORγ’s transcriptional activation activity is required for its enhancement of Dex response. This is reminiscent of the requirement of HNF4 and FoxA’s transactivation domains to assist GR with activating rat Pck1 gene transcription (29). RORγ likely participates in Dex-stimulated gluconeogenic gene transcription through recruiting transcriptional coactivators into a GR-containing transcriptional complex.

Overall, the novel GR-SIPR2-RORγ axis identified here provides valuable insights to improve glucocorticoid therapy. However, two important questions remain to be explored. First, glucocorticoids are potent anti-inflammatory and immunomodulatory agents, but their applications are limited by their adverse effects, including hyperglycemia. This study suggests that including the antagonists (ideally liver specific) of S1PR2 and/or RORγ in glucocorticoid therapy to dampen hyperglycemia could provide improved flexibility of glucocorticoid use. Notably, there are active studies ongoing for designing antagonists of both S1PR2 (10,30) and RORγ (31,32) for other diseases. Obviously, this approach requires that S1PR2 and/or RORγ antagonists do not affect the anti-inflammatory response of glucocorticoids. This notion should be confirmed. Second, how hepatic S1PR2 signaling antagonizes glucocorticoid-suppressed RORγ expression is unclear. This mechanism should be explored in future studies.

Acknowledgments. The authors thank Dr. Donald Scott (Icahn School of Medicine at Mount Sinai) for his comments. Model design was performed using BioRender.

Funding. This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grants DK113019, DK124866, and DK131194. J.-C.W. is a member of the UCSF Liver Center (NIDDK grant DK026743). Work in the laboratory of R.M.O. is supported by NIDDK grant DK132259.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.A.L. and M.C. analyzed the data. R.A.L., M.C., A.T., Y.R.L., D.L., N.Y., S.T., and M.Z. executed the experiments. R.A.L., M.C., R.M.O., and J.-C.W. wrote the manuscript. R.A.L., M.C., and J.-C.W designed and supervised the experiments. R.A.L and J.-C.W. planned and conceptualized the study. J.-C.W. is the guarantor of this work and, as such, had full access to all data in the study and takes responsibility for the integrity of the data and accuracy of data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Endocrine Society Conference, Chicago, IL, 15–18 June 2023.