Molecular mapping and functional validation of GLP-1R cholesterol binding sites in pancreatic beta cells

Affiong I Oqua
Kin Chao
Liliane El Eid
Lisa Casteller
Alba Miguéns
Sebastian Barg
Ben Jones
Jorge Bernadino de la Serna author has email address
Sarah L Rouse author has email address
Alejandra Tomas author has email address

Section of Cell Biology and Functional Genomics, Division of Diabetes, Endocrinology and Metabolism, Department of Metabolism, Digestion and Reproduction, Imperial College London, London, UK
Department of Life Sciences, Imperial College London, London, UK
Department of Medical Cell Biology, University of Uppsala, Uppsala, Sweden
Section of Investigative Medicine, Division of Diabetes, Endocrinology and Metabolism, Department of Metabolism, Digestion and Reproduction, Imperial College London, London, UK
National Heart and Lung Institute, Imperial College London, London, UK

https://doi.org/10.7554/eLife.101011.1

Open access
Copyright information

Figures and data

Effect of changes in cholesterol on GLP-1R agonist responses in vivo and in primary islets.
(A) IPGTTs 6 hours post-intraperitoneal administration of vehicle (Veh) or 1 nmol/kg exendin-4 (Ex-4) in mice fed a chow vs 2% cholesterol (high chol) diet for 5 weeks; n = 8-9 female mice per diet. (B) Area under the curve (AUC) for glucose curves from (A). (C) Ex-4 over Veh glucose levels in chow vs 2% cholesterol diet fed mice. (D) Average intensity of filipin staining (to label cholesterol) in mouse islets preincubated with Veh or MβCD loaded with 20 mM cholesterol (MβCD/chol) for 1 hour; n = 5 islet preps from separate mice; representative islet images also shown; size bars, 10 μm. (E) Percentage of insulin secretion from mouse islets preincubated with Veh or MβCD/chol before stimulation with 11 mM glucose (G11) +/- 100 nM Ex-4; n = 5. (F) Ex-4-induced insulin secretion (fold over G11) in mouse islets from (E). (G) Average filipin staining in mouse islets preincubated with Veh or lipoprotein-deficient serum (LPDS) media supplemented with 10 µM simvastatin (LPDS/simv) overnight; n = 5 islet preps from separate mice; representative islet images also shown; size bars, 10 μm. (H) Percentage of insulin secretion from mouse islets preincubated with Veh or LPDS/simv before stimulation with G11 +/- 100 nM Ex-4; n = 5. (I) Ex-4-induced insulin secretion (fold over G11) in mouse islets from (H). Data is mean +/- SEM; ns, non-significant, *p<0.05, **p<0.01 by paired t-test or one-way ANOVA with Sidak’s multiple comparison test.

cgMD simulations of GLP-1R – cholesterol binding sites in model membranes.
(A) Overview of the simulation setup - GLP-1R is embedded in a model mammalian plasma membrane with the following composition: POPC (30%), DOPC (30%), POPE (8%), DOPE (7%), and cholesterol (25%) in the upper leaflet, and POPC (5%), DOPC (5%), POPE (20%), DOPE (20%), POPS (8%), DOPS (7%), PIP₂ (10%), and cholesterol (10%) in the lower leaflet. (B) Average cholesterol occupancy profile in active (left) and inactive (right) GLP-1R states shown as a heatmap (red – highest occupancy; blue – lowest occupancy), with the top 10 highest occupancy residues per state labelled. (C) Snake plot showing the top 30 highest cholesterol occupancy residues in active and inactive states, with colours indicating occupancy levels (top 10 – red; top 20 – pink; top 30 – orange). (D) Top three cholesterol binding sites in GLP-1R active (top) versus inactive (bottom) states, calculated using PyLipID. Binding sites are colour-coded as follows: site I - purple in active and cyan in inactive state; site II - orange; site III - green, with the top 3 residues with the highest residence time in each site labelled and average residence time indicated for each site. (E) GLP-1R snake plot indicating residues from top three cholesterol binding sites in both states using the same colour scheme as in (D). (F) GLP-1R snake plot indicating the 12 residues selected for screening, with the 4 residues showing a significant reduction in GLP-1R internalisation when mutated to alanine (see H) coloured in red, and the remaining residues in green. (G) Table showing the predicted structural impact of the 12 selected residues after site-directed mutagenesis to alanine in active vs inactive GLP-1R using Missense3D-TM (93). (H) Surface expression and exendin-4 (100 nM, 10 min) mediated internalisation screen of the 12 selected residues from GLP-1R-cholesterol binding sites mutated to alanine, transiently transfected in INS-1 832/3 GLP-1R KO cells; n = 4-5. Data is mean +/- SEM, *p<0.05, **p<0.01 by one way ANOVA with Dunnett’s multiple comparison test vs corresponding WT SNAP/FLAG-hGLP-1R.

GLP-1R WT vs V229A cholesterol binding propensity.
(A) Representative images of INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A sublines labelled with SNAP-Surface Alexa Fluor 647; size bars, 100 µm. (B) Surface expression of SNAP/FLAG-hGLP-1R WT vs V229A; n = 6. (C) Schematic diagram of the GLP-1R PhotoClick cholesterol binding assay. (D) SNAP/FLAG-hGLP-1R-bound cholesterol normalised to receptor levels in INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A treated with vehicle (Veh) or 100 nM exendin-4 (Ex-4) for 2 min; n = 4. (E) Representative images of INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells labelled with SNAP-Surface 488 (green) and stimulated with Veh vs Ex-4 for 2 min prior to fixation and labelling with D4H*-mCherry (red); size bars, 5 µm. (F) Quantification of co-localisation (Mander’s tM1) between SNAP/FLAG-hGLP-1R WT or V229A and D4H*-mCherry in cells from (E); n = 5. (G) Ex-4 over Veh co-localisation fold change for WT vs V229A SNAP/FLAG-hGLP-1R; n = 5. Data is mean +/- SEM, ns, non-significant, *p<0.05, **p<0.01 by paired t-test or one-way ANOVA with Sidak’s multiple comparison test.

GLP-1R WT vs V229A movement at the plasma membrane.
(A) Representative images from GLP-1R WT vs V229A RICS analysis of plasma membrane lateral diffusion in INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A cells labelled with SNAP-Surface Alexa Fluor 647 before stimulation with vehicle (Veh) or 100 nM exendin-4 (Ex-4). (B) Average RICS diffusion coefficients from GLP-1R WT vs V229A from (A); n = 4. (C) TIRF-SPT analysis of average total displacement (top) and speed (bottom) of GLP-1R WT vs V229A under Veh or Ex-4-stimulated conditions in INS-1 832/3 GLP-1R KO cells expressing hGLP-1R-mEGFP WT vs V229A; n = 4 for total displacement and n = 5 for speed. (D) Average RICS diffusion coefficients of the lipid dye Laurdan in INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells under Veh or Ex-4-stimulated conditions; n = 5. (E) Average diffusion coefficient from RICCS analysis of SNAP-Surface Alexa Fluor 647-labelled SNAP/FLAG-hGLP-1R WT vs V229A together with lipid-labelled Laurdan under Veh or Ex-4-stimulated conditions in INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells; n = 5. Data is mean +/- SEM; ns, non-significant, **p<0.05, ***p<0.001 by one-way ANOVA with Sidak’s multiple comparison test.

GLP-1R WT vs V229A oligomerisation and CCP recruitment.
(A) N&B estimation of average number of pixels for the different oligomerisation states of the GLP-1R, either as monomers-dimers, dimers-hexamers, hexamers-decamers, or higher order oligomers, calculated at different time frames after stimulation with either vehicle (Veh) or 100 nM exendin-4 (Ex-4) from INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A cells; n = 4. (B) GLP-1R WT vs V229A levels at lipid raft fractions purified from INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells under Veh or Ex-4-stimulated conditions. Results represent SNAP levels assessed by Western blotting normalised to flotillin as a marker of lipid raft enrichment; n = 5-6. (C) Left: representative TIRF images of INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A cells co-expressing clathrin light chain-GFP (CLC-GFP) labelled with SNAP-Surface Alexa Fluor 647 prior to Veh or 100 nM Ex-4 stimulation; right: quantification of association (βF/S, see Methods) of SNAP/FLAG-hGLP1-R WT or V229A with clathrin puncta; n = 29 and 127 cells for WT Veh vs Ex-4, and n = 31 and 106 cells for V229A Veh vs Ex-4, respectively; each data point represents mean of 3 cells, data collated from 3 separate experiments. Data is mean +/- SEM; ns, non-significant, *p<0.05, **p<0.01, ****p<0.0001 by unpaired t-test, one- or two-way ANOVA with Sidak’s multiple comparison test.

GLP-1R WT vs V229A trafficking profiles.
(A) Representative images of INS-1 832/3 SNAP/FLAG-hGLP-1R WT or V229A cells labelled with SNAP-Surface Alexa Fluor 647 probe under vehicle (Veh) conditions or following stimulation with 100 nM exendin-4 (Ex-4) for 10 min; size bars, 5 µm. (B) Schematic diagram of agonist-mediated SNAP/FLAG-hGLP-1R internalisation assay. (C) Percentage of internalised SNAP/FLAG-hGLP-1R WT vs V229A at the indicated time points after stimulation with 100 nM Ex-4; corresponding AUC also shown; n = 4. (D) Schematic diagram of SNAP/FLAG-hGLP-1R plasma membrane recycling assay. (E) Percentage of recycled SNAP/FLAG-hGLP-1R WT vs V229A at the indicated time points after stimulation with 100 nM Ex-4; corresponding AUC also shown; n = 3. (F) Schematic diagram of agonist-mediated SNAP/FLAG-hGLP-1R degradation assay. (G) Percentage of SNAP/FLAG-hGLP-1R WT vs V229A degradation at the indicated time points after stimulation with 100 nM Ex-4; corresponding AUC also shown; n = 4. Data is mean +/- SEM; ns, non-significant, *p<0.05, **p<0.01, ***p<0.001 by paired t-test or two-way ANOVA with Sidak’s multiple comparison test.

Signalling profiles of GLP-1R WT vs V229A.
(A) Mini-G_s recruitment dose response curves and log₁₀(Emax/EC₅₀) after stimulation with the indicated concentrations of exendin-4 (Ex-4) in INS-1 832/3 GLP-1R KO cells transiently transfected with GLP-1R-SmBiT WT or V229A and LgBiT-mini-G_s; n = 5. (B) Mini-G_q recruitment dose response curves and log₁₀(Emax/EC₅₀) after stimulation with the indicated concentrations of Ex-4 in INS-1 832/3 GLP-1R KO cells transiently transfected with GLP-1R-SmBiT WT or V229A and LgBiT-mini-G_q; n = 6. (C) β-arrestin 2 (βarr2) recruitment dose response curves and log₁₀(Emax/EC₅₀) after stimulation with the indicated concentrations of Ex-4 in INS-1 832/3 GLP-1R KO cells transiently transfected with GLP-1R-SmBiT WT or V229A and LgBiT-βarr2; n = 5. (D) Mini-G_s over βarr2 bias calculation for GLP-1R V229A vs WT. (E) GLP-1R WT vs V229A plasma membrane activation after stimulation with 100 nM Ex-4 in INS-1 832/3 GLP-1R KO cells co-transfected with Nb37-SmBiT, LgBiT-CAAX and SNAP/FLAG-hGLP-1R WT or V229A, measured by NanoBiT bystander complementation assay; AUC also shown; n = 6. (F) As in (E) but for GLP-1R WT vs V229A endosomal activation in INS-1 832/3 GLP-1R KO cells co-transfected with Nb37-SmBiT, Endofin-LgBiT and SNAP/FLAG-hGLP-1R WT or V229A; n = 6. Data is mean +/- SEM; ns, non-significant, *p<0.05, **p<0.01 by paired t-test or two-way ANOVA with Sidak’s multiple comparison test.

Functional responses of GLP-1R WT vs V229A in pancreatic beta cells and primary islets.
(A) cAMP responses of INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells transduced with the Green Up Global cAMP cADDis biosensor before stimulation with 100 nM Exendin-4 (Ex-4) followed by 100 μM isobutyl methylxanthine (IBMX) + 10 µM forskolin (FSK) for maximal response. (B) AUC and maximal response for the Ex-4 period from (A); n = 5. (C) Insulin secretion from INS-1 832/3 SNAP/FLAG-hGLP-1R WT vs V229A cells following stimulation with 11 mM glucose (G11) +/- 100 nM Ex-4; n = 5. (D) Insulin secretion Ex-4-fold increase over G11 calculated from data in (C); n = 5. (E) Representative images of GLP-1R KO mouse islets transduced with adenoviruses expressing SNAP/FLAG-hGLP-1R WT or V229A, labelled with SNAP-Surface Alexa Fluor 647 prior to stimulation with vehicle (Veh) or 100 nM Ex-4 for 5 min; size bars, 100 µm. (F) Surface expression of SNAP/FLAG-hGLP-1R WT vs V229A expressed in GLP-1R KO mouse islets from (E); n = 3. (G) Percentage of SNAP/FLAG-hGLP-1R WT vs V229A internalisation in GLP-1R KO mouse islets following stimulation with 100 nM Ex-4 for 5 min; n = 3. (H) Insulin secretion responses from GLP-1R KO islets transduced with SNAP/FLAG-hGLP-1R WT vs V229A adenoviruses following stimulation with G11 +/- 100 nM Ex-4; n = 3. (I) Insulin secretion Ex-4-fold increase over G11 calculated from data in (H); n = 3. Data is mean +/- SEM; ns, non-significant, *p<0.05, **p<0.01 by paired t-test or one-way ANOVA with Sidak’s multiple comparison test.

Schematic diagram of effects of cholesterol binding mutant GLP-1R V229A on GLP-1R function.
Thick arrows indicate increased and thin arrows decreased pathway engagement.

Overview of effects of GLP-1R V229A in inactive and active states compared with GLP-1R WT.
(↑ increased, ↓ decreased, ≈ unchanged).

Site-directed mutagenesis primers for SNAP/FLAG-hGLP-1R cholesterol binding mutant generation.

Transfected plasmid amounts for Nb37 bystander NanoBiT assays.

Sign up for email alerts