Constitutive internalisation of EP2 differentially regulates G protein signalling

Abstract

The prostanoid G protein-coupled receptor (GPCR) EP2 is widely expressed and implicated in endometriosis, osteoporosis, obesity, pre-term labour and cancer. Internalisation and intracellular trafficking are critical for shaping GPCR activity, yet little is known regarding the spatial programming of EP2 signalling and whether this can be exploited pharmacologically. Using three EP2-selective ligands that favour activation of different EP2 pathways, we show that EP2 undergoes limited agonist-driven internalisation but is constitutively internalised via dynamin-dependent, β-arrestin-independent pathways. EP2 was constitutively trafficked to early and very early endosomes (VEE), which was not altered by ligand activation. APPL1, a key adaptor and regulatory protein of the VEE, did not impact EP2 agonist-mediated cAMP. Internalisation was required for ~70% of the acute butaprost- and AH13205-mediated cAMP signalling, yet PGN9856i, a Gαs-biased agonist, was less dependent on receptor internalisation for its cAMP signalling, particularly in human term pregnant myometrial cells that endogenously express EP2. Inhibition of EP2 internalisation partially reduced calcium signalling activated by butaprost or AH13205 and had no effect on PGE2 secretion. This indicates an agonist-dependent differential spatial requirement for Gαs and Gαq/11 signalling and a role for plasma membrane-initiated Gαq/11-Ca2+-mediated PGE2 secretion. These findings reveal a key role for EP2 constitutive internalisation in its signalling and potential spatial bias in mediating its downstream functions. This, in turn, could highlight important considerations for future selective targeting of EP2 signalling pathways.

Keywords: EP2; GPCR; prostaglandin; trafficking.

Figure 1

EP2 undergoes constitutive but not agonist-driven internalisation. (A) HEK 293 cells expressing FLAG-EP2 live-labelled with anti-FLAG Ab and treated with butaprost (10 μM), AH13205 (10 μM), PGN9856i (100 nM) or DMSO for 20 min ± EDTA stripping of plasma membrane signal. Cells were fixed, permeabilised and labelled with immunofluorescent secondary antibodies. Cells were imaged by confocal microscopy, and representative images shown from at least three independent experiments. Scale bar: 20 μm, inset size: 7 μm. (B) HEK 293 cells expressing FLAG-EP2 were live-labelled with anti-FLAG Ab and stimulated for 20 min with butaprost (10 μM), AH13205 (10 μM), PGN9856i (100 nM) or DMSO, and cell surface expression measured via flow cytometry. Data are shown as percentage change from basal surface expression, represented as mean ± s.e.m., n = 3. One-sample t-test: *P < 0.05. (C) HEK 293 cells expressing FLAG-EP2 live-labelled with anti-FLAG M1 antibody at 4°C with/without 1 h incubation at 37°C. Cell surface expression measured via flow cytometry. Data are shown as percentage change from surface expression in cells kept at 4°C, represented as mean ± s.e.m. n = 3. One-sample t-test: *P < 0.05.

Figure 2

Internalisation of EP2 occurs via β-arrestin independent, dynamin-dependent pathways. (A) WT HEK 293 cells and cells lacking β-arrestin 1/2 were checked via western blot for β-arrestin 1/2 expression (~50 kDa) using GAPDH as a loading control (~35 kDa). Representative image is shown. (B) Confocal microscopy images of FLAG-EP2 expressing WT HEK 293 cells and cells lacking β-arrestin 1/2. Cells were labelled live with anti-FLAG M1 antibody ± EDTA stripping of plasma membrane signal before fixation and stained with immunofluorescent secondary antibody. Representative images from three independent experiments. Scale bar 10 μM, inset 4 μM. (C) FLAG-EP2 HEK 293 cells were incubated with/without dyngo-4a (50 µM; 45 min) treatment prior to live labelling with anti-FLAG M1 antibody before fixation and stained with immunofluorescent secondary antibody. Representative image is shown; n = 10 cells per condition. Scale bar 10 μM, inset 4 μM.

Figure 3

EP2 partially colocalises with early endosomes. (A,B) HEK 293 cells stably expressing (A) FLAG-EP2 or (B) FLAG-β2AR were live-labelled with anti-FLAG Ab and treated with either butaprost (10 μM), AH13205 (10 μM), PGN9856i (100 nM), DMSO or isoproterenol (10 μM) for 20 min. Cells were stripped of plasma membrane signal using EDTA, fixed, permeabilised and labelled with anti-APPL1 primary antibody and immunofluorescent secondary antibodies. Cells were imaged on a confocal microscope. Representative images shown with the receptor in the green channel and EEA1 in the red channel. Scale bar 10 μM. Inset size 4 μM. (C) Receptor-positive endosomes were counted using ImageJ, and the data are the percentage of receptor-positive endosomes that are colocalised with EEA1-positive endosomes. Data are mean ± s.e.m., n = 3, with the mean of each experiment shown as large circles and individual cell data shown as smaller circles, coloured for each experiment. One-way ANOVA with Dunnett’s multiple comparisons test: ***P < 0.001.

Figure 4

EP2 is localised to the very early endosome but is not regulated by APPL1. (A, B) HEK 293 cells stably expressing EP2 were live-labelled with anti-FLAG Ab and treated with butaprost (10 μM), AH13205 (10 μM), PGN9856i (100 nM) or DMSO for 20 min. Cells were stripped of plasma membrane signal using EDTA, fixed, permeabilised and labelled with anti-APPL1 primary antibody and immunofluorescent secondary antibodies. Cells were imaged on a confocal microscope. (A) Representative images shown with the receptor in the green channel and EEA1 in the red channel. Scale bar 10 μM. Inset size 4 μM. (B) EP2-positive endosomes were counted using ImageJ, and the data are the percentage of EP2-positive endosomes that are colocalised with APPL1-positive endosomes. Data are mean ± s.e.m., n = 3, with the mean of each experiment shown as large circles and individual cell data shown as smaller circles, coloured for each experiment. (C) APPL1 was knocked down using SiRNA and cAMP accumulation was measured after 5 min ligand treatment. (D) Representative western blot of APPL1 knockdown for Figure 4C.

Figure 5

Internalisation via β-arrestin-independent dynamin-dependent pathways is required for full EP2 Gαs and Gαq/11 signalling. (A) HEK 293 cells expressing FLAG-EP2 incubated ± dyngo-4a (50 μM; 45 min) before an IBMX pretreatment (5 min, 500 μM) and 5 min stimulation with either butaprost (10 μM), AH13205 (10 μM) or PGN9856i (100 nM). Data shown normalised to control (ligand without dyngo-4a) and shown as mean ± s.e.m., n = 3. One-sample t-test: *P < 0.05. (B) HEK 293 cells expressing FLAG-EP2 were treated for 15 min with dyngo-4a (50 μM) then incubated with Fluo4 AM calcium indicator ± dyngo-4a for 30 min and intracellular calcium mobilisation was measured following acute stimulation with either butaprost (10 μM), AH13205 (10 μM) or PGN9856i (100 nM). Data shown as fluorescent intensity normalised to background for each cell measured (F–F0) and spots for each cell measured, colour coded to reflect inter-experimental variation. n = 3–4 independent experiments. One-way ANOVA with Šídák's multiple comparisons test: *P < 0.05, **P < 0.01. (C) cAMP accumulation was measured in primary myometrial cells as in Figure 5A. Data shown normalised to control (ligand without dyngo-4a) and shown as mean ± s.e.m., n = 3. One-way ANOVA with Tukey’s multiple comparisons test: *P < 0.05. One sample t-test: ^##P < 0.01, ^###P < 0.001. (D) Calcium mobilisation was measured in primary myometrial cells using the same protocol and analysis as Fig. 5B. n = 3 independent experiments, data shown as mean ± s.e.m. One-way ANOVA with Šídák's multiple comparisons test: *P < 0.05. (E) FLAG-EP2 expressing WT HEK 293 cells and cells lacking β-arrestin 1/2 were incubated with IBMX (5 min, 500 μM) before 5 min stimulation with either butaprost (10 μM), AH13205 (10 μM) or PGN9856i (100 nM). cAMP concentrations were normalised to protein and data shown as mean ± s.e.m., n = 3. (F) Calcium mobilisation was measured in primary myometrial cells using the same protocol and analysis as Fig. 5D. n = 3 independent experiments, data shown as mean ± s.e.m.

Figure 6

EP2-mediated PGE2 is released acutely and is not dependent on internalisation. HEK 293 cells stably expressing EP2 were pre-treated with/without dyngo-4a (45 min; 50 μM) before ligand stimulation with butaprost (10 μM), AH13205 (10 μM), PGN9856i (100 nM) or DMSO for either (A) 6 h or (B) 5 min. Media was removed after treatment and PGE2 concentration was determined by ELISA. Data are mean ± s.e.m., n = 3.