Hepatocyte cholesterol content modulates glucagon receptor signalling

Abstract

Objective: To determine whether glucagon receptor (GCGR) actions are modulated by cellular cholesterol levels.

Methods: We determined the effects of experimental cholesterol depletion and loading on glucagon-mediated cAMP production, ligand internalisation and glucose production in human hepatoma cells, mouse and human hepatocytes. GCGR interactions with lipid bilayers were explored using coarse-grained molecular dynamic simulations. Glucagon responsiveness was measured in mice fed a high cholesterol diet with or without simvastatin to modulate hepatocyte cholesterol content.

Results: GCGR cAMP signalling was reduced by higher cholesterol levels across different cellular models. Ex vivo glucagon-induced glucose output from mouse hepatocytes was enhanced by simvastatin treatment. Mice fed a high cholesterol diet had increased hepatic cholesterol and a blunted hyperglycaemic response to glucagon, both of which were partially reversed by simvastatin. Simulations identified likely membrane-exposed cholesterol binding sites on the GCGR, including a site where cholesterol is a putative negative allosteric modulator.

Conclusions: Our results indicate that cellular cholesterol content influences glucagon sensitivity and indicate a potential molecular basis for this phenomenon. This could be relevant to the pathogenesis of non-alcoholic fatty liver disease, which is associated with both hepatic cholesterol accumulation and glucagon resistance.

Keywords: Cell membrane; Cholesterol; Glucagon; Glucagon receptor; Non-alcoholic fatty liver disease; Type 2 diabetes mellitus.

Figure 1

Acute changes in cellular cholesterol levels influence glucagon-stimulated cAMP production in a concentration-dependent manner (A) cAMP concentration-response curves in Huh7-GCGR cells pre-treated with cholesterol-deplete MβCD at indicated concentrations or cholesterol-saturated MβCD (Chol), then stimulated with glucagon (GCG), n = 4. (B) cAMP responses over a wider glucagon concentration range with or without pertussis toxin pre-treatment (PTX; 10 ng/ml) to block Gα_i-mediated cAMP inhibition and reveal the Gα_s- and Gα_i-specific responses, n = 6. The effect of pre-treatment with MβCD or cholesterol is shown. (C) Balance between Gα_s and Gα_i-mediated cAMP effects from (B); all inter-group statistical comparisons non-significant by one-way repeated measures ANOVA. (D) cAMP dynamics measured in Huh7-GCGR cells transduced with cADDis sensor (pictured; scale bar = 80 μm) pre-treated with MβCD or cholesterol and stimulated with 100 nM glucagon. Representative images from cells at baseline, 5 min after glucagon stimulation, and 5 min after addition of 100 μM IBMX and 10 μM forskolin (FSK). AUC comparison by one-way repeated measures ANOVA with Tukey's test, n = 8. (E) As for (D) but 50 pM glucagon, n = 9. (F) As for (D) but using membrane-targeted cADDis cAMP sensors as indicated, n = 7. (G) PKA activation in Huh7-GCGR cells expressing AKAR4-NES and pre-treated with MβCD or cholesterol. Representative FRET images of same cells at baseline, 5 min after glucagon stimulation, and 5 min after addition of 100 μM IBMX and 10 μM forskolin (FSK); scale bar 40 μm, n = 5. (H) cAMP accumulation in primary cadaveric human hepatocytes, 10 min stimulation with 100 μM IBMX, n = 4. The two panels are from the same experiments and separated for clarity. (I) Quantification of signalling potency and maximum responses from (H) in relationship to cholesterol modulating treatments, with one-way repeated measures ANOVA and linear test for trend. (J) cAMP accumulation in primary mouse hepatocytes, 10 min stimulation with 100 μM IBMX, n = 5. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are shown as mean ± SEM, with individual experimental replicates where possible.

Figure 2

Simvastatin treatment enhances glucagon-stimulated signalling and glucose output. (A) The effect on glucagon-stimulated cAMP production (10 min) in Huh7-GCGR cells pre-treated with simvastatin (Simva) or serum-free medium (SFM) overnight, with concurrent or subsequent treatment with or without mevalonate (mev; 50 μM) or cholesterol-saturated MβCD (chol; 50 μg/ml). Results are normalized to forskolin (FSK; 10 μM), n = 5. (B) Association between a combined measure of cAMP efficacy and potency (log-transformed E_max/EC₅₀) and cellular cholesterol for each of the treatments shown in Figure 1A and Figure 2A, both normalized to vehicle control, with linear regression line ± 95% confidence intervals shown. (C) Steady state cAMP concentrations in Huh7-GCGR cells co-treated with glucagon and simvastatin/vehicle for 16 h, with normalisation to the acute 10 μM FSK response taken at the end of the incubation, n = 5. (D) Glucose production in primary mouse hepatocytes pre-treated with simvastatin or vehicle overnight, then stimulated with glucagon for 24 h, expressed as fold change over no-glucagon control stimulation, n = 7, with paired t-test. Data are shown as mean ± SEM, with individual experimental replicates in (D).

Figure 3

Predicted GCGR-cholesterol interactions. The top four ranked binding sites for cholesterol from coarse-grained MD simulations of the glucagon receptor (GCGR) in inactive (A) and active (B) conformations in plasma membrane-like bilayers containing 25% cholesterol. Each conformation was simulated for 10 × 10 μs. Distinct binding sites are coloured yellow (site-1), red (site-2), lilac (site-3) and blue (site-4). Residues comprising each site are shown as spheres scaled by per residue cholesterol residence times. The residence time for cholesterol binding to each site is indicated. Binding sites and associated residence times were calculated using PyLipID [43]. (C) Snapshot from atomistic simulations of the top ranked cholesterol binding pose at site-4, as identified by PyLipID from the coarse-grained simulations. GCGR is shown in surface representation and cholesterol is shown as sticks. F387 is coloured blue and encloses a pocket which shields the cholesterol hydroxyl group from the membrane (see inset for coordinating residues of TM7). (D) Minimum distance between the site-4 cholesterol and L395 (a key residue in site-4) across atomistic simulations. Simulations were initiated from the top ranked cholesterol pose whereby the hydroxy group was located towards the center of the bilayer (3 × 1 μs, blue) or with the cholesterol reversed by 180° such that the hydroxyl group was in proximity to the lipid phosphate groups (3 × 500 ns, grey). (E) Binding saturation curves for cholesterol binding to each site from equilibrium MD simulations (5 × 5 μs at each % free cholesterol). Site % occupancy was calculated using PyLipID and plotted against the free cholesterol % (see methods) in binary bilayers composed of POPC and cholesterol. (F) BRET signal (535/460) indicating interaction between nanoluciferase-tagged mini-G_s and GCGR in the plasma membrane in Huh7-GCGR cells expressing KRAS-venus, 30 min after stimulation with vehicle or 100 nM glucagon, n = 7, compared by two-way repeated measures ANOVA with Sidak's test. (G) As for (F) but using mini-G_i. ∗p < 0.05, ∗∗∗p < 0.001. Data are shown as mean ± SEM, with individual experimental replicates in Figure 2F, G.

Figure 4

Increase in hepatic cholesterol in mice decreases responsiveness to glucagon. (A) body weight, n = 9–12, (B) hepatic triglyceride, n = 9–12, (C) hepatic cholesterol, n = 9–12, (D) fasting glucose, n = 9–12, (E) fasting alanine, n = 6, and (F) glucagon:alanine index, n = 6, in mice fed different diets for 7–12 days; statistical comparisons are by one-way ANOVA with Tukey's test. (G) glucagon/pyruvate challenge test, n = 12, with blood glucose compared using two-way repeated measures ANOVA with Sidak's test. (H) the association between change in AUC during glucagon challenge test and hepatic cholesterol in mice fed different diets as indicated by colour code; the linear regression line ± 95% confidence intervals is shown. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are shown as mean ± SEM, with individual experimental replicates where possible.