Identification of PTGR2 inhibitors as a new therapeutic strategy for diabetes and obesity

Abstract

Peroxisome proliferator-activated receptor γ (PPARγ) is a master transcriptional regulator of systemic insulin sensitivity and energy balance. The anti-diabetic drug thiazolidinediones (TZDs) are potent synthetic PPARγ ligands with undesirable side effects, including obesity, fluid retention, and osteoporosis. 15-keto prostaglandin E2 (15-keto-PGE2) is an endogenous PPARγ ligand metabolized by prostaglandin reductase 2 (PTGR2). Here, we confirmed that 15-keto-PGE2 binds to and activates PPARγ via covalent binding. In patients with type 2 diabetes and obese mice, serum 15-keto-PGE2 levels were decreased. Administration of 15-keto-PGE2 improves glucose homeostasis and prevented diet-induced obesity in mice. Either genetic inhibition of PTGR2 or PTGR2 inhibitor BPRPT0245 protected mice from diet-induced obesity, insulin resistance, and hepatic steatosis without causing fluid retention and osteoporosis. In conclusion, inhibition of PTGR2 is a new therapeutic approach to treat diabetes and obesity through increasing endogenous PPARγ ligands while avoiding side effects including increased adiposity, fluid retention, and osteoporosis.

Keywords: 15-keto-PGE2; Diabetes; Obesity; PPARγ; PTGR2.

Figure 1. 15-keto-PGE2 activates murine PPARγ through binding to cysteine 313 residue.

(A) Metabolism of 15-keto-PGE2. (B) Activation of murine PPARγ (mPPARγ) measured by Gal-PPARγ/UAS-LUC reporter assay in HEK293T cells (n = 3 per group, 3 biological replicates with 1 technical replicate each). Cells were transfected with Gal4-PPARγ, UAS-LUC, and TK-Rluc (Renilla luciferase), and treated with pioglitazone (**P = 0.0021, ****P < 0.0001, ***P = 0.0002) or 15-keto-PGE2 (**P = 0.0015, *P = 0.0137, **P = 0.0014). RT-qPCR of (C) Glut4 (****P < 0.0001, **P = 0.0005, ****P < 0.0001) and other mPPARγ-downstream genes including (D) Irs2 (****P < 0.0001, *P = 0.0475, ****P < 0.0001), (E) Sorbs1 (****P < 0.0001, **P = 0.0023), (F) Cd36 (***P = 0.0002, ***P = 0.0006), (G) Acs (****P < 0.0001, **P = 0.0029), (H) Cepba (**P = 0.0016, *P = 0.0274, **P = 0.0025), and (I) Adipoq (**P = 0.0023, ****P < 0.0001, ****P < 0.0001) in differentiated 3T3-L1 adipocytes treated with 15-keto-PGE2 (n = 3 per group, 3 biological replicates with 2 technical replicate each). (J) Effect of 15-keto-PGE2 on insulin-stimulated glucose uptake in differentiated 3T3-L1 adipocytes (****P < 0.0001, ***P = 0.0002, ****P < 0.0001; n = 4 per group, 4 biological replicates with 1 technical replicate each). (K) HEK293T cells transfected by mPPARγ and treated with 15-keto-PGE2. Covalent binding of 15-keto-PGE2 to mPPARγ detected by liquid-chromatography tandem mass spectrometry (LC-MS/MS). (L) Reciprocal co-immunoprecipitation of mPPARγ and cysteine-15-keto-PGE2. Myc-DDK-mPPARγ and Myc-DDK-mPPARγ C313A were expressed in HEK293T cells, and immunoprecipitation (IP) conducted using either anti-DDK (anti-Flag) or anti-15-keto-PGE2-cysteine-BSA antibody, followed by immunoblotting with anti-15-keto-PGE2-cysteine-BSA and anti-DDK antibody. (M) PPRE reporter activity after addition of 15-keto-PGE2 to HEK293T cells transfected with wild-type and C313A mutant mPPARγ (**P = 0.0037, **P = 0.0077, ****P < 0.0001, ****P < 0.0001; n = 3 per group, 3 biological replicates with 1 technical replicate each). (N) Native mass spectrometry spectrum showed the binding of 15-keto-PGE2 to wild-type and mPPARγ mutants (C313A and H351A). The spectrum of unbound free-form proteins was shown in the left panel. The spectrum of bound form after the addition of 15-keto-PGE2 was shown in the right panel (n = 4 per group, 4 independent experiments with 1 technical replicate each) and (O) histogram (****P < 0.0001). (P) 15-keto-PGE2 enhanced insulin-stimulated glucose uptake in PPARγ-null 3T3-L1 clones (#1296; n = 4 per group, 4 biological replicates with 1 technical replicate each) rescued with wild-type mPPARγ (****P < 0.0001, **P = 0.0036) but not in those rescued with mutant mPPARγ (C313A) (****P < 0.0001). (Q) Diagram showing the motifs of mPPARγ and 15-keto-PGE2 binding site. Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by one-way analyses of variance (ANOVA) with Tukey’s post hoc test in (B–J, P) and two-sample independent t-test in (M, O). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns means no statistical difference. Source data are available online for this figure.

Figure 2. 15-keto-PGE2 prevents diet-induced obesity and improves glucose homeostasis in mice.

(A) Serum 15-keto-PGE2 concentration of non-diabetic human subjects and patients with type 2 diabetic patients (****P < 0.0001; n = 24:24 patients). Correlation of serum 15-keto-PGE2 with (B) homeostasis model assessment of insulin resistance (HOMA-IR) and (C) fasting glucose with serum 15-keto-PGE2 levels in 50 non-diabetic human subjects. (D) Serum 15-keto-PGE2 concentration (****P < 0.0001; n = 5:10 mice) and (E) 15-keto-PGE2 content in inguinal fat (****P < 0.0001; n = 15:7 mice), and (F) perigonadal fat (*P = 0.0124; n = 14:14 mice) of C57BL6/J mice on high-fat high-sucrose diet (HFHSD) or chow. (G) Body weight (*P = 0.0292, *P = 0.0145, *P = 0.0227, *P = 0.0144, **P = 0.0063, **P = 0.0074, **P = 0.0028, **P = 0.0026, **P = 0.0014, ***P = 0.0005; n = 15:15 mice), (H) glucose levels during intraperitoneal glucose tolerance test (ipGTT) (*P = 0.0383, *P = 0.0389, **P = 0.0037, *P = 0.0151, *P = 0.0126, *P = 0.0127; n = 15:15 mice), and (I) insulin tolerance test (ITT) (*P = 0.0448, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, *P = 0.0434, ****P < 0.0001; n = 15:15 mice) of HFHSD-fed C57BL6/J mice treated with 15-keto-PGE2 or vehicles. (J) Weights of inguinal fat (*P = 0.0418), perigonadal fat (*P = 0.0253), mesenteric fat (**P = 0.0044), liver, and brown adipose tissue (BAT) of HFHSD-fed C57BL6/J mice treated with 15-keto-PGE2 or vehicles (n = 15:15 mice). (K) Body composition of HFHSD-fed C57BL6/J mice treated with 15-keto-PGE2 or vehicles (*P = 0.0284; n = 15:15 mice) (L) Phospho-Akt levels in perigonadal fat, inguinal fat, muscle, and brown adipose tissue after intraperitoneal insulin injection. (M) H&E stain of perigonadal fat (scale bar=100 μm) and (N) average adipocyte size (**P = 0.0099; n = 15:15 mice). (O) Energy expenditure measured by indirect calorimetry (****P < 0.0001; n = 15:15 mice), (P) food intake (n = 15:15 mice), and (Q) physical activity (n = 15:15 mice) of HFHSD-fed C57BL6/J mice treated with 15-keto-PGE2 or vehicles. (R) Relative Ucp1 expression in inguinal fat (**P = 0.0040), perigonadal fat (*P = 0.0422), and brown adipose tissue (*P = 0.0338) of HFHSD-fed C57BL6/J mice treated with 15-keto-PGE2 or vehicles (n = 13:14 mice with 2 technical replicates each). (S) Body surface temperatures of interscapular area (****P < 0.0001, ***P = 0.0004, **P = 0.0025), inguinal area (***P = 0.0006, *P = 0.0113), and rectal (***P = 0.0001, **P = 0.0019, **P = 0.0050, ****P < 0.0001, *P = 0.0286) temperatures of 15-keto-PGE2-treated mice and vehicle control mice after HFHSD feeding (n = 15:15 mice). (T) Body surface temperatures of interscapular area (*P = 0.0250, *P = 0.0192, *P = 0.0154, *P = 0.0145, *P = 0.0139), inguinal area (*P = 0.0101, **P = 0.0075), and rectal (*P = 0.0233, *P = 0.0376) temperatures of 15-keto-PGE2-treated mice and vehicle control mice during cold test (n = 15:15 mice). Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by two-sample independent t-test in (A, D–K, N–T) and Spearman’s correlation analyses in (B, C). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure 3. Ptgr2 knockout mice were protected from diet-induced obesity, insulin resistance, glucose intolerance, and fatty liver without fluid retention and reduced bone density.

(A) Body weight (*P = 0.0418, **P = 0.0010, ***P = 0.0004, ***P = 0.0003, ***P = 0.0006, ***P = 0.005; n = 20:21 mice), (B) fasting glucose (***P = 0.0004, ****P < 0.0001, ****P < 0.0001, ***P = 0.0001, ****P < 0.0001, ****P < 0.0001; n = 20:21 mice), and (C) glycemic level during intraperitoneal glucose tolerance test (ipGTT) (****P < 0.0001, *P = 0.0331, **P = 0.0081, ***P = 0.0005; n = 20:21 mice) and (D) insulin tolerance test (ITT) (****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ***P = 0.0004, ****P < 0.0001, ****P < 0.0001; n = 19:20 mice) of Ptgr2 ^−/− and Ptgr2 ^+/+ mice on high-fat high-sucrose diet (HFHSD). (E) ¹⁸F-FDG PET scan of Ptgr2^−/− and Ptgr2^+/+ mice after injection of glucose and insulin. (F) ¹⁸F-FDG uptake in perigonadal fat (*P = 0.0285), inguinal fat (*P = 0.0373), skeletal muscle, and brown adipose tissue (BAT) after injection of glucose and insulin (n = 14:12 mice). (G) Phospho-Akt levels after intraperitoneal insulin injection in perigonadal fat, inguinal fat, liver, brown adipose tissue, and skeletal muscle. (H) Tissue weights of perigonadal fat (**P = 0.0051), inguinal fat, mesenteric fat, liver (*P = 0.0326), and BAT (*P = 0.0347; n = 11:7 mice). (I) Body composition (****P < 0.0001; n = 14:17 mice), (J) bone mineral density of femur (14:17 mice), (K) micro computed tomography (CT) image of femur of C57BL6/J mice (n = 14:17 mice), and (L) average adipocyte size (*P = 0.0484, n = 14:16 mice with at least 100 cells measured for each mouse). (M) F4/80 immunohistochemical stain of perigonadal fat (scale bar = 50 μm) and (N) percentage of F4/80-positive cells in perigonadal fat (****P < 0.0001; n = 43:56 technical replicates, from 14 and 17 mice with at least 3 images measured for each mouse). (O) H&E stain of liver (scale bar=50 μm) and (P) hepatic triglyceride contents of (**P = 0.0061; n = 6:8 mice) of Ptgr2^−/− mice and Ptgr2^+/+ on HFHSD. Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by two-sample independent t-test in (A–D, F, H, I, J, L, N, O, P). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure 4. Ptgr2 knockout mice displayed increased energy expenditure and thermogenesis.

(A) Energy expenditure (the white bar indicates light-up time, the black bar indicates light-off time) (*P = 0.0316, *P = 0.0477, *P = 0.0230, *P = 0.0258, *P = 0.0314, *P = 0.0180, *P = 0.0468, *P = 0.0252, *P = 0.0438, *P = 0.0478, *P = 0.0177, *P = 0.0254, ***P = 0.0004, **P = 0.0042, *P = 0.0361, **P = 0.0024, **P = 0.0035, *P = 0.0303, *P = 0.0292, *P = 0.0405; n = 12:12 mice) (B) Immunoblots showing Ucp1 level in different fat pads. RT-qPCR of genes involved in browning (n = 6:8 mice with 2 technical replicates each) of (C) perigonadal fat (*P = 0.0213, *P = 0.0100, **P = 0.0040, *P = 0.0406), (D) inguinal fat (*P = 0.0213, *P = 0.0213, *P = 0.0213, *P = 0.0100, **P = 0.0023), and (E) brown adipose tissue (BAT) (*P = 0.0293, *P = 0.0200, *P = 0.0293, *P = 0.0200). For the diet-induced thermogenesis test, 24-week-old mice were fasted overnight. Body surface including (F) interscapular area (**P = 0.0034, ***P = 0.0003, ***P = 0.0004, ***P = 0.0004, **P = 0.0029, **P = 0.0056), (G) inguinal area (*P = 0.0448, *P = 0.0334, *P = 0.0216), and (H) core rectal (*P = 0.0375, *P = 0.0481, *P = 0.0158) temperatures at 0, 30, 60, 90, 120, 150, 180, and 240 min after HFHSD refeeding (n = 14:16 mice). Body surface temperatures in (I) interscapular area, (J) inguinal area (*P = 0.0181, *P = 0.0287), and (K) rectal (*P = 0.0411, *P = 0.0466) temperatures during cold tolerance test of Ptgr2^−/− mice and Ptgr2^+/+ mice on HFHSD (n = 10:10 mice). Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by two-sample independent t-test in (A, C–K). *P < 0.05, **P < 0.01, ***P < 0.001. Source data are available online for this figure.

Figure 5. PTGR2 inhibitor BPRPT0245 protected mice from diet-induced obesity, insulin resistance, glucose intolerance, and fatty liver without fluid retention and reduced bone density.

(A) Structure of BPRPT0245. (B) Half-maximal inhibitory concentration (IC₅₀) of BPRPT0245. (C) Effect of BPRPT0245 on 15-keto-PGE2-dependent PPARγ transcriptional activity in PTGR2-transfected HEK293T cells (****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, **P = 0.0014; n = 5 per group, 5 biological replicates with 1 technical replicate each). (D) Intracellular 15-keto-PGE2 content of differentiated 3T3-L1 adipocytes treated with BPRPT0245 (****P < 0.0001, ****P < 0.0001, ****P < 0.0001, **P = 0.0013, **P = 0.0016; n = 3–6 per group, 3–6 biological replicates with 1 technical replicate each) and (E) insulin-stimulated glucose uptake of differentiated 3T3-L1 adipocytes treated with BPRPT0245 (**P = 0.0034, *P = 0.0105; n = 3–4 per group, 3–4 biological replicates with 1 technical replicate each). (F) Molecular docking of 15-keto-PGE2 (green), NADPH (orange brown), and BPRPT0245 (cyan) of PTGR2, Tyr259 (red) of PTGR2. (G) Molecular docking showing that PTGR2 inhibitor BPRPT0245 (cyan) interferes the interaction between 15-keto-PGE2 and NADPH through a strong biding network, Pi-Alkyl (CH) interaction, and Pi-Pi interaction with PTGR2 and NADPH. (H, I) Lineweaver–Burk plots showing the competitive inhibitory action of BPRPT0245 (50 nM) for the interaction between 15-keto-PGE2 and PTGR2 (n = 5, 5 independent experiments with 2 technical replicates) values of Vmax and Km for 15-keto-PGE2 with or without the addition of BPRPT0245 (50 nM) (n = 5, 5 independent experiments with 2 technical replicates). (J) Body weight (*P = 0.0480, *P = 0.0466; n = 15:15 mice), (K) fasting glucose (**P = 0.0054, *P = 0.0141, **P = 0.0070, *P = 0.0414, *P = 0.0251, *P = 0.0427; n = 15:15 mice), (L) glycemic level during intraperitoneal glucose tolerance test (ipGTT) (**P = 0.0014, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ****P < 0.0001, ***P = 0.0004, ****P < 0.0001; n = 15:15 mice), (M) insulin tolerance test (ITT) (*P = 0.0163, ***P = 0.0004, *P = 0.0165, *P = 0.0236; n = 15:15 mice), (N) tissue weight of perigonadal fat (**P = 0.0039), inguinal fat (***P = 0.0003), mesenteric fat (**P = 0.0022), liver (**P = 0.0050), brown adipose tissue (BAT) (n = 10:10 mice), (O) body composition (n = 10:10 mice), (P) F4/80 immunohistochemical stain of gonadal fat (scale bar= 100 μm), (Q) percentage of F4/80-positive in perigonadal fat (****P < 0.0001; n = 15:15 mice, with at least 3 images measured for each mouse), (R) adipocyte size (**P = 0.0030; n = 10:10 mice, with at least 100 cells measured for each mouse), (S) representative micro computed tomography (CT) image of femur of C57BL6/J mice, (T) bone mineral density (n = 10:10 mice), (U) representative H&E stain of liver section (scale bar = 100 μm), and (V) hepatic triglycerides content (**P = 0.0014; n = 10:10 mice) of high-fat high-sucrose-fed C67BL6/J mice treated with BPRPR0245 (100 mg/kg/day) and vehicles. Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by one-way analyses of variance (ANOVA) with Tukey’s post hoc test in (C–E) and two-sample independent t-test in (J–M, O, Q, R, T, V). Statistical significance was calculated by two-sample independent t-test in (A, C–K). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure EV1. Relative intracellular 15-keto-PGE2 levels in cultured 3T3-L1 cells treated with exogenous 15-keto-PGE2 of different concentrations (*P = 0.0109; n = 6 per group, 6 biological replicates with 1 technical replicate).

Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by one-way analyses of variance (ANOVA) with Tukey’s post hoc test. *P < 0.05.

Figure EV2. 15-keto-PGE2 enhanced insulin-stimulated glucose uptake in PPARγ2-null adipocytes rescued with wild-type PPARγ2 but not with mutant PPARγ2 (C313A).

Immunoblots showing PPARγ expression in PPARγ-null 3T3-L1 clones #1296 and #2328 using the CRISPR techniques and then overexpress (A) mutant PPARγ2 (C313A) or (B) wild-type PPARγ2. 15-keto-PGE2 enhanced insulin-stimulated glucose uptake in clone #2328 rescue (n = 4 per cell clone, 4 biological replicates with 1 technical replicate) with (C) wild-type (****P < 0.0001, **P = 0.0036) or (D) mutant PPARγ2 (C313A) (***P = 0.0001, *P = 0.0111, *P = 0.0363). Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by one-way analyses of variance (ANOVA) with Tukey’s post hoc test and two-sample independent t-test (C, D). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure EV3. Relative serum 15-keto-PGE2 concentration and 15-keto-PGE2 content in perigonadal fat were higher in Ptgr2^−/− mice compared to Ptgr2^+/+ controls.

(A) Relative serum 15-keto-PGE2 level (*P = 0.0352) and (B) relative 15-keto-PGE2 content (**P = 0.0013) in perigonadal fat (n = 8:18 mice) of Ptgr2^−/− and Ptgr2^+/+ mice on high-fat high-sucrose diet (HFHSD). Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by two-sample independent t-test in (A, B). *P < 0.05, **P < 0.01.

Figure EV4. Relative serum 15-keto-PGE2 concentration and 15-keto-PGE2 content in perigonadal fat were higher in mice treated with BPRPT0245 compared to those receiving the vehicle.

(A) Relative 15-keto-PGE2 contents in perigonadal fat (n = 6:6 mice) and (B) inguinal fat (*P = 0.0117; n = 6:6 mice) after oral gavage of BPRPT0245 (100 mg/kg/day) for 4 days. Samples are harvested 2 h after oral gavage of the latest dose. Data information: Data are presented as mean and standard error (S.E.M.). Statistical significance was calculated by two-sample independent t-test (A, B). *P < 0.05.

Figure EV5. Comparison between 2D [¹H,¹⁵N]-TROSY-HSQC NMR (nuclear magnetic resonance) spectra of apo-form and 15-keto-PGE2 bound PPARγ LBD (ligand binding domain).

(A) Comparison between 2D [¹H,¹⁵N]-TROSY-HSQC NMR spectra of apo-form and 15-keto-PGE2 bound PPARγ LBD (ligand binding domain). Cyanide color indicates missing peak. magenta color indicates chemical shift with Δδ > 0.05. (B) NMR missing peak (cyanide color) and chemical shift (magenta color) mapped onto PPARγ LBD structure.