Human MC4R variants affect endocytosis, trafficking and dimerization revealing multiple cellular mechanisms involved in weight regulation

Abstract

The Melanocortin-4 Receptor (MC4R) plays a pivotal role in energy homeostasis. We used human MC4R mutations associated with an increased or decreased risk of obesity to dissect mechanisms that regulate MC4R function. Most obesity-associated mutations impair trafficking to the plasma membrane (PM), whereas obesity-protecting mutations either accelerate recycling to the PM or decrease internalization, resulting in enhanced signaling. MC4R mutations that do not affect canonical Gα_s protein-mediated signaling, previously considered to be non-pathogenic, nonetheless disrupt agonist-induced internalization, β-arrestin recruitment, and/or coupling to Gα_s, establishing their causal role in severe obesity. Structural mapping reveals ligand-accessible sites by which MC4R couples to effectors and residues involved in the homodimerization of MC4R, which is disrupted by multiple obesity-associated mutations. Human genetic studies reveal that endocytosis, intracellular trafficking, and homodimerization regulate MC4R function to a level that is physiologically relevant, supporting the development of chaperones, agonists, and allosteric modulators of MC4R for weight loss therapy.

Keywords: GPCRs; Gα(s); MC4R; MSH; melanocortin; obesity; therapy; weight loss; β-arrestin.

Graphical abstract

Figure 1

MC4R mutants impair PM localization and agonist-dependent endocytosis (A) Schematic of human MC4R. Amino acids affected by obesity-associated and obesity-protecting MC4R mutants are indicated in black (n = 50). Sequence-based generic GPCR residue numbering for Family A GPCRs (Ballesteros-Weinstein) is included. ECL, extracellular loop; ICL, intracellular loop. (B) Agonist-induced MC4R internalization is quantified using ebBRET-based sensors expressed at either the plasma membrane (PM; rGFP-CAAX sensor), early endosomes (EEs; rGFP-FYVE sensor), or in late endosomes (LEs; drGFP-Rab7 sensor). Renilla luciferase II (RlucII) serves as the ebBRET donor and is coupled to MC4R. (C) Quantification of receptor levels at the PM by basal ebBRET signal. Twenty-two mutants show PM expression of ≥85% of WT and were taken forward for functional assays. (D and E) Agonist-induced MC4R internalization as quantified using the PM sensor (D) and the EE sensor (E), both normalized to basal ebBRET. (F) Basal ebBRET as a proxy of receptor level in EEs. (G) Agonist-induced internalization quantified using LE sensor for the gain-of-function (GoF) mutants V103I and I251L, normalized to basal ebBRET quantified using the PM sensor. Data in (C)–(F) are expressed as log_e (mutant/WT) and plotted as mean ± standard error from 3–6 independent experiments. Data in (D) and (E) represent mean ± standard error from the sum curves and therefore do not include data points from individual experiments. Data in (G) are expressed as % WT MC4R and plotted as mean ± standard error from 3 independent experiments. Mutants are classified as GoF (orange), LoF (blue), or WT-like (gray) based on statistically significant differences between WT and mutant (unpaired t test with Welch’s correction; p < 0.05, or extra sum-of-squares F test for D and E; p < 0.05). See also Figures S1–S3 and S6.

Figure 2

β-arrestin-2 drives MC4R internalization, and obesity-associated MC4R mutants impair β-arrestin recruitment (A–D) Agonist-dependent endocytosis of MC4R in ARRB1/2 knockout and parental HEK293SL control cells transfected with a β-arrestin-2 expression vector, as measured using the PM sensor (A) or the EE sensor (B); basal ebBRET with the PM sensor (C) and the EE sensor (D) in this experiment. (E and F) Agonist-dependent endocytosis quantified in HEK293SL cells transfected with siRNAs targeting no mRNA (siControl) and β-arrestin-1 (siARRB1) (E) or β-arrestin-2 (siARRB2) (F). (G) Basal ebBRET in this experiment using siControl and siARRB1 or siARRB2. (H and I) Agonist-dependent β-arrestin-1 (H) and β-arrestin-2 (I) recruitment for MC4R mutants that do not impair receptor number at the PM. Internalization expressed as percentage of the maximal ΔebBRET obtained with WT receptor (see STAR Methods) and plotted as mean ± standard error from 3 independent experiments. β-arrestin recruitment data are expressed as log_e (mutant/WT) and plotted as mean ± standard error from 3–6 independent experiments. In (H) and (I), mutants are classified as GoF (orange), LoF (blue), or WT-like (gray) based on statistically significant differences between WT and mutant (unpaired t test with Welch’s correction; p < 0.05). See also Figure S4.

Figure 3

MC4R-derived MAPK activation is dependent on receptor internalization and modulated by β-arrestins (A and B) HEK293 cells transiently expressing MC4R WT were used to assess the effect of dynasore on maximal efficacy of NDP-αMSH in cAMP production (A) and ERK1/2 phosphorylation (B). (C and D) NDP-αMSH-dependent ERK1/2 phosphorylation in HEK293 cells treated with the β-arrestin/AP2 inhibitor barbadin (C) and siControl, siARRB1, or siARRB2 (D). (E) Effect of siControl, siARRB1, or siARRB2 on maximal efficacy of NDP-αMSH-induced cAMP production. Representative immunoblots shown were probed for total ERK1/2, phosphorylated ERK1/2 (p-ERK1/2), Gα_s, β-arrestin-1/2, and vinculin as an additional loading control. Data are expressed as percentage of maximal control response (% WT) and plotted as mean ± standard error from 3–14 independent experiments. Statistical significance was determined by two-way ANOVA and Dunnet’s multiple comparison test (*p < 0.05, ^∗∗∗p < 0.001, compared to NDP-αMSH-stimulated MC4R WT [control]). See also Figure S4.

Figure 4

Gα_s signaling is essential for MC4R-dependent ERK1/2 phosphorylation, and obesity-associated mutants impair the interaction between Gα_s and MC4R (A and B) HEK293 cells transiently expressing WT MC4R were used to assess the effect of siControl or Gα_s (siGNAS) on maximal efficacy of NDP-αMSH in cAMP production (A) and ERK1/2 phosphorylation (B). (C) NDP-αMSH-dependent ERK1/2 phosphorylation in HEK293 cells treated with H89 (protein kinase A inhibitor). (D) Agonist-dependent Gα_s:MC4R interaction for mutants that do not impair PM expression. (E) NDP-αMSH-mediated ERK1/2 phosphorylation for MC4R mutants. (A–C) Data are expressed as percentage of maximal control response (% WT) and plotted as mean ± standard error from 3–14 independent experiments. Statistical significance was determined by two-way ANOVA and Dunnet’s multiple comparison test (^∗∗∗p < 0.001, compared to NDP-αMSH-stimulated MC4R WT (control)). Representative immunoblots shown were probed for total ERK1/2, phosphorylated ERK1/2 (p-ERK1/2), β-arrestin-1/2, Gα_s, and vinculin as an additional loading control. (D and E) Data are expressed as log_e (mutant/WT) and plotted as mean ± standard error from 3–9 independent experiments. (F) GoF and LoF MC4R mutants affecting coupling to Gα_s, β-arrestin-1, and β-arrestin-2 are shown mapped into a cartoon representation of the crystallized MC4R receptor structure. The crystallized ligand SHU9119 is shown in dark gray (PDB: 6W25) (Yu et al., 2020). In (D)–(F), mutants were classified as GoF (orange), LoF (blue), or WT-like (gray) based on statistically significant differences between WT and mutant (unpaired t test with Welch’s correction; p < 0.05). See also Figures S4–S6.

Figure 5

Structural mapping of MC4R mutants suggests impaired dimerization (A) Receptor mutants affecting at least one coupling partner (blue) and those that do not affect coupling (gray). Two of the latter map to one of the proposed MC4R homodimeric interfaces, being within 4Å of the opposite MC4R partner (shown in gray in a van der Waals representation). S191T maps to a TM-4/TM-5 interface (left inset); T53I maps to a TM-1 /TM-7 interface (right inset). (B) BRET saturation curve from HEK293SL cells co-transfected with a constant amount of MC4R RlucII donor construct and increasing amounts of the MC4R Venus acceptor construct. Soluble (s) Venus acceptor construct was used as a negative control. (C) MC4R WT:MC4R mutant receptor dimerization as measured in NanoBiT protein:protein interaction assay (see STAR Methods). Data are expressed as log_e (mutant/WT) and plotted as mean ± standard error from 4–7 independent experiments. Mutants were classified as LoF (blue) or WT-like (gray) based on statistically significant differences between WT and mutant (unpaired t test with Welch’s correction; p < 0.05). (D) Structural mapping of variants affecting dimerization (blue sticks) and variants with preserved dimerization (gray sticks). Membrane-facing residues are highlighted in a surface representation. The left panel shows details from a top (extracellular) view of the receptor, whereas the right panel shows a bottom (intracellular) view. See also Figure S4.

Figure 6

Obesity-associated and obesity-protecting mutations in MC4R affect multiple molecular mechanisms (A) GoF mutations increase, whereas LoF MC4R mutations decrease receptor numbers at the cell surface. Molecular chaperones (heat shock proteins and MRAP2), chemical chaperones (e.g., 4-phenylbutyrate), and pharmacological chaperones (antagonists and agonists) have the potential to stabilize and target intracellularly retained MC4R mutants to the PM (chaperones represented as purple diamond). (B) MC4Rs homodimerize, and obesity-associated mutations can disrupt this process. (C) Production of the second messenger cAMP is dependent on Gα_s protein and modulated by β-arrestins; MC4R mutations can differentially affect coupling to these signaling transducers. G protein-induced cAMP production is essential for MAPK activation (ERK1/2), which leads to changes in gene expression by activation of cAMP response element-binding protein (CREB-TF). (D) Agonist-dependent endocytosis of MC4R is mainly driven by β-arrestins and is critical for ERK1/2 activation. Several mutations in MC4R negatively affect the sequestration of receptors from the PM, the translocation to EEs, or both. GoF mutations cycle more avidly by rapid recycling pathways (EE to PM), rather than being directed to late compartments (slow recycling) or degradative pathways (lysosomes) and as such are more available at the surface for subsequent rounds of agonist-induced signaling. Semi-transparent (faded) symbols for G protein subunits and β-arrestins indicate potential involvement of these intracellular effectors in megaplexes (Thomsen et al., 2016), although this has not been formally evaluated in this study.