EP3 signaling is decoupled from the regulation of glucose-stimulated insulin secretion in β-cells compensating for obesity and insulin resistance

Abstract

Of the β-cell signaling pathways altered by obesity and insulin resistance, some are adaptive while others contribute to β-cell failure. Two critical second messengers are Ca²⁺ and cAMP, which control the timing and amplitude of insulin secretion. Previous work has shown the importance of the cAMP-inhibitory Prostaglandin EP3 receptor (EP3) in mediating the β-cell dysfunction of type 2 diabetes (T2D). Here, we used three groups of C57BL/6J mice as a model of the progression from metabolic health to T2D: wildtype, normoglycemic Leptin^Ob (NGOB), and hyperglycemic Leptin^Ob (HGOB). Robust increases in β-cell cAMP and insulin secretion were observed in NGOB islets as compared to wildtype controls; an effect lost in HGOB islets, which exhibited reduced β-cell cAMP and insulin secretion despite increased glucose-dependent Ca²⁺ influx. An EP3 antagonist had no effect on β-cell cAMP or Ca²⁺ oscillations, demonstrating agonist-independent EP3 signaling. Finally, using sulprostone to hyperactivate EP3 signaling, we found EP3-dependent suppression of β-cell cAMP and Ca²⁺ duty cycle effectively reduces insulin secretion in HGOB islets, while having no impact insulin secretion on NGOB islets, despite similar and robust effects on cAMP levels and Ca²⁺ duty cycle. Finally, increased cAMP levels in NGOB islets are consistent with increased recruitment of the small G protein, Rap1GAP, to the plasma membrane, sequestering the EP3 effector, Gɑ_z, from inhibition of adenylyl cyclase. Taken together, these results suggest that rewiring of EP3 receptor-dependent cAMP signaling contributes to the progressive changes in β cell function observed in the Leptin^Ob model of diabetes.

Keywords: Diabetes; EP3 receptor; Gɑz; Rap1gap; cAMP; hyperglycemia; insulin secretion; prostaglandins; β-cell.

Graphical abstract

Figure 1.

Expression of EP3 is associated with impaired insulin secretion independent of PGE₂ production. (a) Blood glucose levels in C57Bl/6J wild-type (WT), normoglycemic Leptin^ob (NGOB) and hyperglycemic Leptin^ob (HGOB) mice. N = 13–18. (b) Islet insulin content. N = 7–10. (c) Relative Ptger3 transcript levels normalized to Actb. N = 9–15. (d) Relative Ptger3 isoform transcript levels normalized to Actb. N = 4–12. (e) Selected PGE₂ synthetic enzymes (Pt2g4a, Ptgs1, and Ptgs2) as normalized to Actb. N = 4–12. (f) 24 h PGE₂ excretion from cultured islets. N = 6–8. In all cases, N=individual mice or mouse islet preparations, and data represent the mean ± SEM. Data was analyzed by one-way ANOVA with Tukey test post-hoc (A-C, F) or multiple t-test corrected with Tukey test post-hoc (D,E). *P<.05; **P<.01; ***P<.001; and ****P<.0001. ns = not significant.

Figure 2.

Validation of β-cell-specific cAMP FRET sensor and correlation of β-cell cAMP levels with islet pathophysiology. (a) 3D projection of WT islets expressing the β-cell specific cAMP biosensor (Epac-S^H187) as recorded by two-photon microscopy. (b) Normalized fluorophore intensity (left), FRET ratio (middle) and average FRET ratio (right) from WT islets expressing cAMP biosensor treated with 100 μM 3-Isobutyl −1-methylxanthine (IBMX). N = 7 islets. Scale bar = 0.02. (c) Glucose stimulated insulin secretion from WT, NGOB and HGOB islets treated with 3.3 mM glucose or 8.3 mM glucose ±10 nM E x 4. Data is represented as both total insulin secreted (left) and fractional insulin secretion as a percent of total insulin content. N = 7–10 mice. (d) Representative Ca²⁺ recordings and duty cycle quantification of WT, NGOB and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM E x 4. Scale bar = 0.01. (e) Representative average cAMP recordings and quantification of WT, NGOB, and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM E x 4. N = 55–72 islets from three mice (NGOB and HGOB) and 90 islets from six mice (WT). Scale bar = 0.025. Error bars represent the standard error of the mean (SEM). Data was analyzed by one-way ANOVA with Tukey test post-hoc within groups and multiple T-test with Tukey test post-hoc among groups.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. For panel E, ****p < 0.0001 as compared to WT and ^####p <0.0001 as compared to NGOB.

Figure 3.

Agonist-sensitive EP3 signaling does not influence WT, NGOB, or HGOB β-cell cAMP production or function. (a) Glucose stimulated insulin secretion from WT, NGOB and HGOB islets treated with 3.3 mM glucose or 8.3 mM glucose ±10 nM DG041. Data are represented as both total insulin secreted (left) and fractional insulin secretion as a percent of total insulin content. N = 7–10 mice. (b) Representative Ca²⁺ recordings and duty cycle quantification of WT, NGOB and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM DG041. Scale bar = 0.01. (c) Representative average cAMP recordings and quantification of WT, NGOB and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM DG041. Scale bar = 0.025. In (B) and (C), N = 55–72 islets from three mice (NGOB and HGOB) and 108 islets from six mice (WT). Error bars represent the standard error of the mean (SEM). Data was analyzed by one-way ANOVA among groups or multiple T-test by treatment, both with Tukey test post-hoc.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. For panel C, ****p < 0.0001 as compared to WT and ^####p <0.0001 as compared to NGOB.

Figure 4.

The EP3 agonist sulprostone reduces Ca²⁺ duty cycle and cAMP levels but inhibits insulin secretion only in failing β-cells. (a) Glucose stimulated insulin secretion from WT, NGOB and HGOB islets treated with 3.3 mM glucose or 8.3 mM glucose ±10 nM sulprostone. Data is represented as both total insulin secreted (left) and fractional insulin secretion as a percent of total insulin content. N = 7–10 mice. (b) Representative Ca²⁺ recordings and duty cycle quantification of WT, NGOB and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM sulprostone. Scale bar = 0.01. (c) Representative average cAMP recordings and quantification of WT, NGOB and HGOB islets treated with 9 mM glucose followed by 9 mM glucose +10 nM sulprostone. Scale bar = 0.025. In (B) and (C), N = 55–82 islets from three mice (NGOB and HGOB) and 84 islets from six mice (WT). Error bars represent the standard error of the mean (SEM). Data was analyzed by one-way ANOVA among groups or multiple T-test by treatment, both with Tukey test post-hoc.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. For panel C, ****p < 0.0001 as compared to WT and ^#### p < 0.0001 as compared to NGOB.

Figure 5.

Ga_z-null islets have higher cAMP levels and E x 4 response, and EP3γ reduces cAMP production and GSIS or Ca²⁺ duty cycle in a Gɑ_z-dependent or -independent manner, respectively. (a) Representative simultaneous recordings of Ca²⁺ and cAMP in WT and Gα_z KO islets treated with 9 mM glucose followed by 9 mM glucose +10 nM exendin − 4 (E x 4). Scale bars: Ca²⁺ = 0.1 and R470/530 (cAMP) = 0.02. (b) Quantification of islet Ca²⁺ duty cycle for the data shown in (A). (c) Average temporal (left) and pre/post E x 4 (right) cAMP levels for the experiments represented in (A). (d) Islet E x 4 response, as calculated by normalizing the mean cAMP levels in 9 mM glucose + E x 4 to those in 9 mM glucose alone. In A-D, N = 67–72 islets from three mice each. (e) Representative Western blot of WT and Gɑ_z knockout (KO) C57Bl/6N islets transduced with an adenovirus encoding HA-tagged EP3γ or a GFP control adenovirus. (f) Insulin secreted in 16.7 mM glucose from cultured islets isolated from WT or Gα_z KO mice expressing EP3_γ or GFP control, with and without 10 nM sulprostone. Within each experiment, the GSIS response of a group was normalized to that of its GFP control. N = 3 independent experiments. Figure adapted from Schaid et al., 2021. (g) cAMP production from cultured islets isolated from WT or Gα_z KO mice expressing EP3_γ or GFP control, with and without 10 nM sulprostone. Within each experiment, the GSIS response of a group was normalized to that of its GFP control. Figure adapted from Schaid et al., 2021. (h) Representative recordings of Ca²⁺ oscillations (left) and duty cycle quantification (right) in islets isolated from WT or Gα_z KO mice expressing EP3_γ or GFP control, with and without 10 nM sulprostone. N = 3 independent experiments. In all panels, error bars represent the standard error of the mean (SEM). Data was analyzed by two-way ANOVA with Tukey test post hoc (A-E) or multiple unpaired t-test corrected with Holm-Sidak test post-hoc, or unpaired t-test (F). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Figure 6.

Expression and membrane localization of the Gα_z effector, Rap1GAP, is associated with higher cAMP signaling in islets from highly compensating NGOB mice. (a) Representative Western blot of Rap1GAP (total cellular and membrane-associated) and β-actin in WT, NGOB and HGOB islets. (b) Quantification of total cellular Rap1GAP in WT, NGOB and HGOB islets, as normalized to β-actin. (c) Quantification of membrane-associated Rap1GAP in WT, NGOB and HGOB islets, as normalized to β-actin. (d) Representative Western blot of phospho-PKA substrate in WT, NGOB, and HGOB islets. (e) Quantification of phospho-PKA substrate as normalized to β-actin. In (B), (C), and (E), data were normalized to relative WT expression. n = 4–10 mice/group. Error bars represent the standard error of the mean (SEM). In (B), (C), and (E), data was analyzed with one-way ANOVA with Tukey test post hoc *p < 0.05, **p < 0.01, ***p < 0.001.