Exposure of MC4R to agonist in the endoplasmicreticulum stabilizes an active conformation of thereceptor that does not desensitizeSusana Granell1, Brent M. Molden1, and Giulia Baldini2

Department of Biochemistry and Molecular Biology, University of Arkansas for Medical Sciences, Little Rock, AR 72205

Edited by Melanie H. Cobb, University of Texas Southwestern Medical Center, Dallas, TX, and approved October 29, 2013 (received for reviewNovember 16, 2012)

Melanocortin-4 receptor (MC4R) is a G protein-coupled receptorexpressed in neurons of the hypothalamus where it regulates foodintake. MC4R responds to an agonist, α-melanocyte–stimulatinghormone (α-MSH) and to an antagonist/inverse agonist, agouti-related peptide (AgRP), which are released by upstream neurons.Binding to α-MSH leads to stimulation of receptor activity andsuppression of food intake, whereas AgRP has opposite effects.MC4R cycles constantly between the plasma membrane and endo-somes and undergoes agonist-mediated desensitization by beingrouted to lysosomes. MC4R desensitization and increased AgRPexpression are thought to decrease the effectiveness of MC4Ragonists as an antiobesity treatment. In this study, α-MSH, insteadof being delivered extracellularly, is targeted to the endoplasmicreticulum (ER) of neuronal cells and cultured hypothalamic neu-rons. We find that the ER-targeted agonist associates with MC4Rat this location, is transported to the cell surface, induces constantcAMP and AMP kinase signaling at maximal amplitude, abolishesdesensitization of the receptor, and promotes both cell-surfaceexpression and constant signaling by an obesity-linked MC4R var-iant, I316S, that otherwise is retained in the ER. Formation of theMC4R/agonist complex in the ER stabilizes the receptor in an activeconformation that at the cell surface is insensitive to antagonismby AgRP and at the endosomes is refractory to routing to thelysosomes. The data indicate that targeting agonists to the ER canstabilize an active conformation of a G protein-coupled receptorthat does not become desensitized, suggesting a target for therapy.

Melanocortin-4 receptor (MC4R) is a G protein-coupledreceptor (GPCR) expressed in the brain where it controls

food intake and energy expenditure (1–3). MC4R-knockout miceare obese (4), as are humans with naturally occurring mutationsof the receptor (5–8), thus indicating the central importance ofMC4R in energy homeostasis. Although MC4R is expressedubiquitously in the central nervous system (9, 10), regulation offood intake is restored in MC4R-knockout mice when MC4R isexpressed in the paraventricular nuclei of the hypothalamus andin the amygdala (11). Multiple anorexigenic hormones from theperiphery, including leptin from the adipose tissue, insulin fromβ-cells of the pancreas, and glucagon-like peptide-1 and chole-cystokinin from the gut, are received by pro-opiomelanocortin(POMC) neurons localized in the arcuate nucleus of the hypo-thalamus. POMC neurons, projecting to the paraventricularnucleus of the hypothalamus, release the anorexigenic hormoneα-melanocyte–stimulating hormone (α-MSH), which binds toMC4R expressed by downstream melanocortin neurons. On theother hand, other neurons in the arcuate nucleus that respond toorexigenic hormones from the periphery increase the productionof agouti-related protein (AgRP), which is the natural antago-nist/inverse agonist of MC4R (12–18). AgRP signaling andAgRP neurons (19–21) appear to be essential to promote feed-ing. Exposure of MC4R to α-MSH promotes receptor couplingto the heterotrimeric stimulatory G protein with downstream ac-tivation of adenylate cyclase and increased intracellular levels of

cAMP. These effects are antagonized by AgRP, which also acts asan inverse agonist by inhibiting constitutive signaling of MC4R.Following agonist stimulation, MC4R is desensitized by

a population of receptors disappearing from the cell surface viaa mechanism that includes retention of the receptor in the in-tracellular compartment and its traffic to lysosomes (22–25).Administration of the potent MC4R agonist melanotan II and ofMSH/adrenocorticotropin (ACTH) 4–10 to lean and obese micecauses a decrease in both food intake and body weight (26, 27),thus suggesting that treatment with MC4R agonist could preventor reverse obesity. However, in mice, prolonged agonist treat-ment causes tachyphylaxis by a mechanism that may includea compensatory up-regulation of AgRP mRNA levels and de-sensitization (26, 27). Moreover, overweight humans appear tobe resistant to treatment with the MC4R agonists (28). Theseobservations suggest that agonism of MC4R by extracellularlydelivered ligands is an ineffective target to treat obesity becausethe receptor is desensitized and/or because of compensatory up-regulation of AgRP. Intracellular delivery of GPCR agonists andantagonists has been used to rescue conformationally defectivereceptors along the biosynthetic route and to promote theirdelivery to the cell surface (29). In this respect, it has been foundthat many MC4R variants linked to obesity have the tendency tomisfold and be retained in the endoplasmic reticulum (ER) (6,30–32). These mutant receptors can be rescued by pharmaco-perones, which are MC4R antagonists that can permeate the

Significance

Melanocortin-4 receptor (MC4R) is a cell-surface hormone re-ceptor in the brain that is central to the control of appetite.Upstream pathways sensing energy balance in the organismlead to the secretion of α-melanocyte–stimulating hormone(α-MSH), which stimulates MC4R activity and leads to de-creased appetite. Previously, it was found that MC4R becomesresistant to treatment with extracellular agonists by a mecha-nism that may involve desensitization of the receptor, arguingagainst their usefulness in treating obesity. In this study, weshow that exposing MC4R to α-MSH in the endoplasmic re-ticulum leads to constant signaling of MC4R without de-sensitization. Our results indicate that the cellular localizationwhere agonist binding initially takes place affects the confor-mation and the desensitization properties of MC4R, suggestinga target for therapy.

Author contributions: S.G., B.M.M., and G.B. designed research; S.G. and B.M.M. per-formed research; S.G. and B.M.M. contributed new reagents/analytic tools; S.G., B.M.M.,and G.B. analyzed data; and S.G., B.M.M., and G.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1S.G. and B.M.M. contributed equally to this work.2To whom correspondence should be addressed. E-mail: GBaldini@uams.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219808110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1219808110 PNAS | Published online November 18, 2013 | E4733–E4742

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

membrane, as well as by chemical chaperones (31–35). Here wereasoned that delivery of MC4R agonists, rather than antago-nists, to the ER could reduce the binding of MC4R to AgRP andpromote the traffic of MC4R variants retained in the ER bychanging the conformation of the receptor at this location. Wefind that binding MC4R to α-MSH in the ER abolishes the re-ceptor response to AgRP and promotes the folding of an obesity-linked variant retained in the ER. In addition, binding MC4R toα-MSH in the ER induces a persistent activity of the receptor ata constantly maximal amplitude because desensitization androuting of the receptor to the lysosomes does not occur. The dataindicate that desensitization of a GPCR can be modulated notonly by the type of agonist to which the receptor is exposed, as inthe case of opioid receptors (36–39), but also by the cell locali-zation where the agonist/GPCR interaction initially takes place.The data presented here suggest that exposure of MC4R toα-MSH in the ER induces a conformation of the receptor dif-ferent from that stabilized by extracellular α-MSH, so that thatthe receptor is both active and resistant to desensitization, in-dicating a potential target to promote GPCR signaling.

Resultsα-MSH and γ+α-MSH Targeted to the Secretory Pathway Bind toExogenous, Tagged MC4R. Posttranslational processing of POMCgenerates α-, β-, and γ- (γ-MSH) melanocyte–stimulating hor-mones and ACTH. These hormones are agonists of MC4R andof other melanocortin receptors (40, 41). To target α-MSH to theER, we generated a construct in which the signal sequence ofPOMC directing the hormone precursor to the ER was retainedand was followed directly by the sequence of α-MSH with a Myc-tag at the C terminus (ER-α-MSH). The tetrapeptide motif His6-Phe7-Arg8-Trp9 of α-MSH is essential for stimulation of MC4R(40, 42–44), and mutation of His6 to Trp6 within this tetrapeptidereduces potency toward MC4R (40). We generated constructswith α-MSH lacking the tetrapeptide motif (ER–α-MSH–delta)and with His6 mutated to Trp6 (ER–α-MSH–Trp) (Fig. 1A).Corresponding constructs were generated in which WT andmutated α-MSH are preceded by the N-terminal peptide ofPOMC, which includes the weak MC4R agonist γ-MSH in ad-dition to α-MSH (ER–γ+α-MSH) (Fig. S1A). We previouslygenerated Neuro2A (N2A) cells, which, in addition to endoge-nous MC4R, express exogenous HA-MC4R-GFP with the HAtag attached to the N terminus and exposed to the extracellularmedium and GFP attached the C terminus and exposed to thecytosol (N2AMC4R cells) (24). The tags do not change the abilityof MC4R to bind to α-MSH or to respond to α-MSH stimulationby increasing intracellular cAMP levels (24, 31). Immunofluo-rescence microscopy of fixed and permeabilized N2AMC4R cellstransiently expressing ER–α-MSH, ER–α-MSH–Trp, and ER–

α-MSH–delta indicated that all these proteins were expressedat similar levels (Fig. S1B). However, ER–α-MSH and ER–

α-MSH–Trp accumulated more clearly in a perinuclear com-partment than ER–α-MSH–delta, indicating a different cellulardistribution (Fig. S1B, arrows). To determine whether ER–

α-MSH, ER–α-MSH–Trp, and ER–α-MSH–delta reached theplasma membrane, N2AMC4R cells were incubated at 4 °C withantibodies against the Myc tag of the ER–α-MSH proteins andantibodies against the HA tag of HA-MC4R-GFP. ER–α-MSHand ER–α-MSH–Trp colocalized with HA-MC4R-GFP at theplasma membrane (Fig. 1B, second and third rows), but ER–

α-MSH–delta did not (Fig. 1B, fourth row). Thus, ER–α-MSHand ER–α-MSH–Trp, but not ER–α-MSH–delta, bind to HA-MC4R-GFP at some point along the biosynthetic pathway to betransported to the cell surface. We have shown that MC4R isconstitutively endocytosed and recycled back to the plasmamembrane (24, 25). Consistent with that process, when liveN2AMC4R cells were incubated at 37 °C with anti-HA antibodies,the antibodies were internalized into a perinuclear endosomal

compartment where HA-MC4R-GFP also localizes (Fig. 1C, firstrow). Also, the anti-Myc antibodies against the Myc tag of theER–α-MSH and ER–α-MSH–Trp internalized to the same peri-nuclear compartment (Fig. 1C, second and third rows). However,when ER–α-MSH–delta was expressed, the anti-Myc antibodydid not appear in the perinuclear endosomes (Fig. 1C, fourth

Fig. 1. Intracellular α-MSH (ER–α-MSH) expressed in N2AMC4R cells binds totagged MC4R along the biosynthetic pathway and is transported to the cellsurface and to endosomes. (A) Schematic diagram of ER–α-MSH constructsused in this work. ER–α-MSH corresponds to the signal peptide of POMC (SP,amino acids 1–26) followed by the fragment of POMC that encodes forα-MSH (amino acids 138–150). (B and C) N2AMC4R cells stably expressing HA-MC4R-GFP were transfected with empty vector (EV) or ER-α-MSH (WT, Trp, ordelta). (B) To determine whether ER–α-MSH is expressed at the cell surfacetogether with HA-MC4R-GFP, cells were incubated with anti-HA antibodiesat 4 °C. Total HA-MC4R-GFP (GFP, green fluorescence), HA-MC4R-GFP at thecell surface (anti-HA, Cy3, red fluorescence), and ER–α-MSH at the cell sur-face (anti-Myc, Cy5, blue fluorescence) are shown. (C) To determine whetherER–α-MSH internalizes together with HA-MC4R-GFP, cells were incubatedwith anti-HA antibodies at 37 °C. Total HA-MC4R-GFP (GFP, green fluores-cence), HA-MC4R-GFP at the cell surface and endosomes (anti-HA, Cy3, redfluorescence), and ER–α-MSH at the cell surface and endosomes (anti-Myc,Cy5, blue fluorescence) are shown. Immunofluorescence and confocal im-aging was done as described in SI Methods, and experiments were per-formed three times. (Scale bar, 5 μm.)

E4734 | www.pnas.org/cgi/doi/10.1073/pnas.1219808110 Granell et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

row). These data indicate that ER–α-MSH and ER–α–MSH–Trpremain bound to HA-MC4R-GFP at the cell surface and throughone or more cycles of receptor endocytosis.

Expression of Intracellular α-MSH in Neuronal Cells ExpressingTagged HA-MC4R-GFP Induces Constant cAMP Signaling. BecauseER–α-MSH appears to reach the plasma membrane/endosomesbound to HA-MC4R-GFP, we reasoned that the intracellularα-MSH might promote signaling of the receptor in the absenceof exposure to extracellular α-MSH. MC4R couples to G proteinwith activation of adenylate cyclase and thereby increases theproduction of cAMP. To test this possibility, we used N2A cellstransiently cotransfected with HA-MC4R-GFP and either emptyvector or ER–α-MSH. When the N2A cells transiently expressingHA-MC4R-GFP were exposed for 15 min to α-MSH delivered tothe medium, the cAMP level increased by ∼20-fold (Fig. 2, lanes1 and 2). Conversely, when ER–α-MSH was coexpressed withHA-MC4R-GFP, cAMP already was elevated to the level in-duced by acute (15-min) exposure to extracellular α-MSH (Fig.2, lanes 2 and 3). The addition of extracellular α-MSH to thesecells did not further increase the level of cAMP being generated(Fig. 2, lanes 3 and 4). The experiment suggests that all the HA-MC4R-GFP appearing at the cell surface is in a complex withER–α-MSH and signals constantly. Coexpression of HA-MC4R-GFP and ER–α-MSH–delta did not induce any increase ofcAMP in the absence of extracellular α-MSH (Fig. 2, lanes 7 and8). This result is consistent with the expected inability of α-MSH–

delta to act as an agonist (40) and with the observation that ER–

α-MSH–delta does not appear to bind to HA-MC4R-GFP (Fig.1). Coexpression of HA-MC4R-GFP and ER–α-MSH–Trpincreases intracellular cAMP by ∼10-fold, lower than the levelinduced by the expression of ER–α-MSH (Fig. 2, lanes 3 and 5);the addition of the extracellular hormone increased signalingfurther to the level generated by the receptor acutely exposed toextracellular α-MSH (Fig. 2, lanes 6 and 2). This result is con-sistent with the observation that mutation of His6 to Trp6 withinα-MSH reduces potency toward MC4R (40). These experiments

indicate that the expression of intracellular α-MSH leads tocAMP signaling that is constant and that, when the tetrapeptidemotif His6-Phe7-Arg8-Trp9 of α-MSH is intact, has the sameamplitude as the signaling induced by extracellular α-MSH.When ER–γ+α-MSH (WT, Trp, or delta) was expressed in the

N2AMC4R cells, the cAMP levels were equivalent to those in-duced by ER–α-MSH (Fig. S2A, lanes 5–7 and 2–4). This resultsuggests that, only α-MSH, and not γ-MSH in the same protein,functions to induce signaling of MC4R. It is possible that theconstant signaling observed with the expression of ER–α-MSHoccurs because the hormone is secreted in the medium andinduces cAMP signaling by binding to MC4R expressed by thesame or another cell. However, when the medium from N2Acells expressing ER–γ+α-MSH was transferred to another pop-ulation of N2AMC4R cells, it did not induce an increase of in-tracellular cAMP (Fig. S2B). Thus, the increase in cAMPdepends not on the agonist being released in the medium butrather on the binding of the intracellular agonist to MC4R alongthe biosynthetic pathway.

Binding of Intracellular α-MSH to Tagged MC4R Occurs in the ER. Thedata in Fig. 1 indicate that ER–α-MSH binds to MC4R along thesecretory pathway. However, the cellular compartment in whichthis interaction takes place is unclear. Western blot analysisshowed that HA-MC4R-GFP expressed without ER–α-MSHmigrated at a different, higher molecular weight (MW) than HA-MC4R-GFP coexpressed with ER–α-MSH, suggesting divergencein posttranslational processing (Fig. 3A). In mammalian cells,N-glycosylation is initiated in the ER and requires the presenceand availability of the acceptor sequence Asn-X-Ser/Thr (45, 46).HA-MC4R-GFP expressed either with or without ER–α-MSHand then treated with peptide-N-glycosidase F (PNGase F) mi-grated as a band with a lower MW than that of the receptor notexposed to glycosidase treatment (Fig. 3B). The shift in MW ofMC4R upon PNGase F treatment indicates that the receptor isN-glycosylated. The deglycosylated HA-MC4R-GFP constructmigrated equally, whether the receptor was coexpressed with orwithout ER–α-MSH. These observations suggest that coex-pression of MC4R with ER–α-MSH changes the glycosylationpattern of the receptor. MC4R possesses four putative N-gly-cosylation sites, three at the N terminus (Asn3, Asn17, and Asn26)and one at the first extracellular loop (Asn108) (2). HA-MC4R-GFP receptors with mutations at each of these sites migratedfaster than the WT receptor on SDS/PAGE (Fig. 3C, lanes 1 and2–9, respectively). This result indicates that MC4R is N-glyco-sylated at each of these sites. The abundance of all HA-MC4R-GFP N-glycosylation mutants at the cell surface was similar tothat the WT receptor (Fig. S3A). Thus, reduced N-glycosylationof MC4R does not appear to affect the ability of the receptor tomature. HA-MC4R-GFP with mutated Asn3, Asn26, and Asn108

underwent a further lower shift in MW when coexpressed withER–α-MSH (Fig. 3C, lanes 3 and 4, 7 and 8, and 9 and 10),indicating that ER–α-MSH inhibits glycosylation at a site dif-ferent from that blocked by the mutation. Conversely, HA-MC4R-GFP with the Asn17 mutation had the same MW whenexpressed without or with ER–α-MSH (Fig. 3C, lanes 5 and 6).Moreover, HA-MC4R-GFP with the Asn17 mutation had anapparent MW similar to that of the WT receptor coexpressedwith ER–α-MSH (Fig. 3C, lanes 2 and 5 and 6, respectively).This result suggests that mutation at Asn17 and coexpression withER–α-MSH have the same effect on N-glycosylation of MC4R.Together, these experiments indicate that ER–α-MSH binds toMC4R early during synthesis of MC4R in the ER, beforeN-glycosylation at Asn17 takes place. MC4R has intrinsic constitu-tive activity (17), which can be modulated by mutations at theN-terminal domain of MC4R (47). To determine if ER–α-MSH–

mediated inhibition of glycosylation at Asn17 changed the consti-tutive activity of the receptor, we measured both the α-MSH–

Fig. 2. Coexpression of intracellular α-MSH and tagged MC4R inducesconstant cAMP signaling at maximal amplitude. N2A cells were transientlycotransfected with HA-MC4R-GFP and with either empty vector (EV) or ER-α-MSH (WT, Trp, or delta). cAMP was measured in the absence (open bars) orin the presence of exogenous 100 nM α-MSH (gray bars). The graph showsa representative experiment in which samples were run in replicates ofeight. The experiment was performed three times. Data are expressed asmean ± SD. **P < 0.01, ***P < 0.001; one-way ANOVA; ns, nonsignificant,

Granell et al. PNAS | Published online November 18, 2013 | E4735

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

stimulated and constitutive activity of WT HA-MC4R-GFP andHA-MC4R-GFP N17Q. We found that the constitutive andα-MSH–stimulated activity by the WT and N17Q receptors wereequivalent (Fig. S3 B and C). This result indicates that theincrease in cAMP resulting from coexpression of MC4R withER–α-MSH is not caused by enhanced constitutive activityof the receptor but rather by signaling after binding to theintracellular ligand.

Binding of Intracellular α-MSH to Obesity-Linked MC4R Variant I316SChanges the Conformation of the Receptor to Promote Cell SurfaceExpression and Constant Signaling. The data presented in Fig. 3indicate that binding of ER–α-MSH to MC4R occurs in the ER.In this respect, it has been found that delivery of pharmaco-perones, which are small lipophilic agonists and antagonists ofGPCR that can pass through biological membranes, can changethe conformation of these receptors by forming interactionsalong the biosynthetic pathway so that they can progress to thecell surface to signal (29, 48, 49). Pharmacoperones that areeither inverse agonists or antagonists of MC4R have been usedto promote progression along the secretory pathway of obesity-linked variants of MC4R that otherwise are retained in thebiosynthetic pathway (32–34). We reasoned that if interactionwith ER–α-MSH changed the conformation of MC4R I316S inthe ER, then it might rescue the mutated receptor to the plasmamembrane/endosomes. To determine if such rescue occurred,N2A cells transiently expressing either WT HA-MC4R-GFP orHA-MC4R-GFP I316S were incubated in the presence of anti-HA antibody for 1 h at 37 °C. Although in cells expressing WTHA-MC4R-GFP the staining by the anti-HA antibody virtuallycoincided with the fluorescence of GFP (Fig. 4A, first row, EV),in cells expressing HA-MC4R-GFP I316S the GFP fluorescencehad a predominantly reticular distribution that differed from thestaining by the anti-HA antibody (Fig. 4B, first row, EV). Thisresult is consistent with previous findings that MC4R I316S hastendency to be retained in the ER (8, 30, 31, 35, 50). The ratio ofred fluorescence (Cy3, recycling receptor) to green fluorescence(GFP, total receptor) is a measure of the fraction of total HA-MC4R-GFP that has exited the biosynthetic pathway to reachthe cell surface and recycle (31). The coexpression of ER–

α-MSH and ER–α-MSH–Trp increased (by ∼100%) the fractionof mutated (Fig. 4C, lanes 5–7) but not the fraction of WT HA-MC4R-GFP (Fig. 4C, lanes 1–3) able to reach plasma mem-brane/endosomes (Fig. 4C), whereas the coexpression of ER–

α-MSH–delta did not have any effect (Fig. 4C, lanes 5 and 8).Thus, ER–α-MSH and ER–α-MSH–Trp appear to promote theexit of HA-MC4R-GFP I316S from the ER to the cell surface/endosomes. The data indicate that binding of the intracellularagonist to HA-MC4R-GFP I316S changes the conformation ofthe receptor in the ER so that it no longer is retained by thequality-control machinery.Next, we determined that the HA-MC4R-GFP I316S coex-

pressed with ER–α-MSH signals constantly, as does the WTreceptor coexpressed with ER–α-MSH (Fig. 4D, lanes 1 and 3).

However, unlike the signal generated by WT HA-MC4R-GFP,the signal generated by HA-MC4R-GFP I316S in a complex withER–α-MSH was more elevated than that generated by the re-ceptor variant acutely exposed to extracellular α-MSH (Fig. 4D,lanes 2 and 3), most likely because ER–α-MSH concomitantlyincreases the pool of receptors exiting the biosynthetic pathwayto reach the plasma membrane/endosomes. In conclusion, thedata in Fig. 4 indicate that interaction with ER–α-MSH and ER–

Fig. 3. Binding of ER–α-MSH to HA-MC4R-GFP occursin the ER and changes the posttranslational processingof the receptor. (A) N2A cells were cotransfected withWT HA-MC4R-GFP and with or without ER-α-MSH. Celllysates were immunoprecipitated and analyzed byWestern blot as described in SI Methods. (B) N2Acells were cotransfected, and cell lysates wereimmunoprecipitated as in A. Immunoprecipitateswere incubated with 0 or 20 units of PNGase F asdescribed in SI Methods. Immunoprecipitates wereanalyzed as in A. (C) Asparagine residues of potential N-glycosylation sites were mutated to glutamine as described in SI Methods. N2A cells werecotransfected with WT or mutated HA-MC4R-GFP and with either empty vector (EV) or ER-α-MSH. Cell lysates were immunoprecipitated and analyzed as in A.The images in A–C show a representative gel from three independent experiments.

Fig. 4. Binding of intracellular α-MSH to HA-MC4R-GFP I316S induces boththe exit of the mutated receptor from the ER and constant cAMP signaling.(A and B) N2A cells were cotransfected with either empty vector (EV) or ER-α-MSH (WT, Trp, or delta) and with either WT HA-MC4R-GFP (A) or I316S HA-MC4R-GFP (B). Immunofluorescence staining, confocal imaging, and quan-tification of the data were done as described in SI Methods (C) The fractionsof total cell receptor at the cell surface and endosomes from the experi-ments shown in A and B were measured as described in SI Methods [WT HA-MC4R-GFP cotransfected with EV, number of cells (n) = 22; WT ER–α-MSH, n =91; ER–α-MSH–Trp, n = 60; ER–α-MSH–delta, n = 53. I316S; HA-MC4R-GFPcotransfected with EV, n = 38; ER–α-MSH-WT, n = 127; ER–α-MSH–Trp, n =136; ER–α-MSH–delta, n = 95). Quantification was done from three in-dependent experiments. Each experiment was done three times. (D) N2Acells were cotransfected as in B. cAMP was measured in the absence (lightbars) or in the presence (dark bars) of exogenous 100 nM α-MSH. The graphshows a representative experiment in which samples were run in replicatesof eight. Each experiment was done three times. Data are expressed as mean ±SD. **P < 0.001, ***P < 0.0001; one-way ANOVA; ns, nonsignificant.

E4736 | www.pnas.org/cgi/doi/10.1073/pnas.1219808110 Granell et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

α-MSH–Trp induces changes in conformation of an obesity-linked variant of MC4R in the ER. Such changes appear tounderlie the ability of the receptor variant to exit the biosyntheticpathway and to signal constantly.

Expression of Intracellular α-MSH Induces cAMP Signaling at MaximalAmplitude in Immortalized GT1-7 Hypothalamic Cells. Expression ofintracellular α-MSH in neuronal cells with exogenous MC4Rinduces constant cAMP signaling at the amplitude induced byacute exposure to extracellular α-MSH. This constant activityover a prolonged period is unexpected, because it has beenreported that prolonged exposure to extracellular α-MSH indu-ces desensitization of MC4R, which is detectable by the disap-pearance of a population of MC4R from the cell surface and,functionally, by decreased amplitude of cAMP signaling upona rechallenge with the hormone (22, 24, 25). As is consistent withthese reports, pretreatment of N2A cells transiently expressingHA-MC4R-GFP with 1 μM α-MSH for 4 h blunted the responseto acute challenge with α-MSH by ∼70% (Fig. 5A, lanes 2 and 4).Conversely, the level of cAMP constantly generated by HA-MC4R-GFP coexpressed with ER–α-MSH was similar to thelevel induced by the acute, 15-min exposure to extracellularα-MSH in N2A cells that were not pretreated with the hormone(Fig. 5A, lanes 2 and 5), as is consistent with data in Fig. 2, ratherthan to the level observed in response to the rechallenge withα-MSH after the prolonged pretreatment with the hormone (Fig.5B, lanes 4 and 5). These data indicate that MC4R in a complexwith α-MSH initiated in the ER signals constantly at maximalamplitude, unlike MC4R in a complex with extracellular α-MSH,which appears to become desensitized over time. However, it ispossible that ability of ER–α-MSH to induce maximal signalingby HA-MC4R-GFP is an artifact caused by the overexpression ofMC4R. To test this possibility, ER–α-MSH was expressed inN2A cells and GT1-7 immortalized hypothalamic neurons (Fig. 5

B and C), both of which express endogenous MC4R (25, 51–53).The GT1-7 cells were transduced with lentiviral particles toachieve ER–α-MSH expression in most of the cells (Fig. S4). Inboth N2A and GT1-7 cells, acute stimulation with extracellularα-MSH was induced by 60-min incubation with the hormone,which led to a 50% increase in intracellular cAMP (Fig. 5 B andC, lanes 1 and 2). The magnitude of this signal is similar to thatfound previously in the GT1-7 cells (52). Conversely, when N2Aand GT1-7 cells were pretreated with extracellular α-MSH for4 h and then were rechallenged with the hormone for 1 h, theamount of cAMP did not increase (Fig. 5 B and C, lanes 1 and 3).This result indicates that cells expressing endogenous MC4Rundergo profound desensitization upon prolonged incubationwith the agonist. However, when N2A and GT1-7 cells expressedER–α-MSH, the abundance of cAMP was increased by ∼50%, tothe same extent as in cells acutely exposed to extracellularα-MSH (Fig. 5 B and C, lanes 2 and 4). These experiments in-dicate that when endogenous MC4R binds to α-MSH in the ER,the receptor signals constantly and at maximal amplitude, asobserved with overexpressed HA-MC4R-GFP, and suggest thatthe receptor does not become desensitized.In vivo experiments indicate that MC4R signaling leads to

inhibition of AMP kinase (AMPK) activity in the hypothalamus,which is necessary to reduce food intake (54). Consistent withthese data are reports that exposure of immortalized hypotha-lamic neurons to α-MSH induces dephosphorylation of AMPKat Thr172, resulting in a loss of AMPK activity (55). When N2Acells were exposed for 15 min to α-MSH added to the medium,phosphorylation of AMPKα at Thr172 was reduced by ∼40%compared with cells that were not treated with the agonist (Fig.5D, lanes 1 and 2). This result indicates that the α-MSH signalingto decrease AMPK activity is reproduced in the N2A cells.However, when N2A cells were treated with α-MSH for 4 h, the

Fig. 5. Expression of intracellular α-MSH inducesconstant cAMP and AMPK signaling in neuronalN2A cells and immortalized GT1-7 hypothalamiccells expressing endogenous MC4R. (A) N2Acells were cotransfected with HA-MC4R-GFP andeither EV or ER–α-MSH. Cells were incubatedin the absence or presence of 1 μM α-MSH for4 h (pretreatment). Cells were washed and stim-ulated by incubating the cells with 100 nMα-MSH for 15 min in the presence of 3-isobutyl-1-methylxanthine (IBMX). The upper panel showsa schematic diagram of the experiment shown inthe lower panel. (B) N2A cells were transfectedwith empty vector (EV) or ER–α-MSH. Cells wereincubated in the absence or presence of 1 μMα-MSH for 4 h (pretreatment). Cells were stimu-lated by incubating the cells with 100 nM α-MSHfor 60 min in the presence of IBMX. The upperpanel is as in A. (C ) GT1-7 cells were either mock-transduced or transduced with ER–α-MSH–plenti6lentiviral particles at a final concentration of106 IU/mL in the complete culture medium, andcells were incubated at 37 °C for 72 h. The ex-periment was done as in B. (D) N2A cells weretransfected with either empty vector (EV) orER–α-MSH and were incubated at 37 °C in thepresence or absence of 1 μM α-MSH for 4 h(pretreatment). Cells were washed and treatedwith or without 100 nM α-MSH for 15 min(stimulation). Expression of phospho-AMPK andtotal AMPK were measured by Western blot asdescribed in SI Methods. (Left) A schematic dia-gram of the experiment. (Center) A representa-tive Western blot from three independent experiments. (Right) The quantification of three independent experiments. Data are expressed as mean ±SD. *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA; ns, nonsignificant.

Granell et al. PNAS | Published online November 18, 2013 | E4737

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

level of phosphorylated AMPK did not change, whether or notcells were washed to be acutely reexposed to extracellularα-MSH (Fig. 5D, lanes 3 and 4). This result again indicatesprofound desensitization of MC4R. Conversely, when ER–

α-MSH was expressed in N2A cells, the level of AMPK phos-phorylation at Thr172 was reduced to that seen in cells acutelyexposed to the hormone (Fig. 5D, lanes 2 and 5). These dataindicate that endogenous MC4R in a complex with of ER–

α-MSH signals through the AMPK pathway at maximal ampli-tude, suggesting that MC4R does not become desensitized.

MC4R Exposed to Intracellular α-MSH Does Not Become Desensitized.Formation of the MC4R/intracellular α-MSH complex in the ERappears to decrease the propensity of the receptor to becomedesensitized. However, constant signaling by MC4R/ER–α-MSHinstead might result from the continuous delivery of newly syn-thesized complexes to the cell surface/endosomes and/or fromsignaling along the biosynthetic pathway. To test these possibil-ities, cells were treated for 2 h with cycloheximide to blockprotein synthesis and with 1 μM α-MSH to compare the effectsof prolonged exposure to extracellular α-MSH on HA-MC4R-GFP in a complex with ER–α-MSH and on unoccupied HA-MC4R-GFP. After 2-h incubation with extracellular α-MSH, morethan 80% of the ER–α-MSH immunoreactivity disappeared fromthe plasma membrane of N2A cells cotransfected with ER–α-MSHand HA-MC4R-GFP, indicating that exposure to the extracellularα-MSH displaced most of the intracellular agonist from theMC4R binding site (Fig. 6 B and C). Upon prolonged exposure toextracellular α-MSH, cells without coexpressed ER–α-MSH had∼40% less HA-MC4R-GFP at the plasma membrane, and cellswith coexpressed ER–α-MSH had ∼20% less HA-MC4R-GFP(Fig. 6D, lanes 1 and 2 and lanes 3 and 4). This result indicatesthat the expression of intracellular α-MSH blunts the loss of re-ceptor abundance at the plasma membrane caused by prolongedexposure to extracellular α-MSH. Functionally, when rechallengedwith the hormone, cAMP levels were reduced by ∼40% in cellsincubated with extracellular α-MSH without coexpressed ER–α-MSH and were unchanged in cells incubated with extracellularα-MSH with coexpressed ER–α-MSH (Fig. 6 E and F). The dataindicate that when the MC4R/intracellular α-MSH complex ini-tially forms in the ER, the propensity of the receptor to becomedesensitized is decreased. In addition, blocking protein synthesisby cycloheximide seems not to disrupt the ability of HA-MC4R-

GFP in a complex with intracellular α-MSH to generate cAMP atthe amplitude generated by the receptor when acutely exposed toextracellular α-MSH (Fig. 6F, lanes 1 and 3). This result suggeststhat continuous delivery of newly synthesized MC4R/ER–α-MSHcomplexes does not contribute to the constant signaling of thereceptor. Moreover it appears that the MC4R/intracellularα-MSH complex signals after exiting the biosynthetic pathway.These observations are consistent with the finding that MC4R ina complex with ER–α-MSH signals at same level as MC4R acutelyexposed to extracellular α-MSH.

MC4R Reaching the Cell Surface in a Complex with Intracellular α-MSHIs Stabilized in an Active Conformation Different from That of theReceptor in a Complex with Extracellular α-MSH. The data shown inthis paper indicate that, when MC4R binds to intracellularα-MSH, the receptor is stabilized into an active conformation thatsignals constantly and does not become desensitized. Theseobservations open the possibility that MC4R can have differentfunctional properties at the cell surface, depending on whether thereceptor initially binds α-MSH in the ER or at the plasma mem-brane. Because MC4R cycles constantly between the plasmamembrane and the endosomes, dissecting possible differences inthe function of the receptor residing at the cell surface requires anassay with temporal resolution. In this respect, we have found thatexpression of ER–α-MSH does not change the fraction of recy-cling HA-MC4R-GFP localized at the cell surface (Fig. S5B).Expression of ER–α-MSH also does not change the rate of in-ternalization of HA-MC4R-GFP (t1/2 ∼10 min) compared withthat of the unoccupied receptor (Fig. S5A). Thus, the distributionand dynamics of HA-MC4R-GFP associated with intracellularα-MSH along the recycling compartments are similar to those ofthe unoccupied receptor. We measured cAMP generation by us-ing a real-time FRET-based approach with the genetically enco-ded sensor m-Turquoise-TEpacVV-YFP (TEpacVV), which iscomposed of the cAMP-responsive protein Epac sandwichedbetween two fluorescent tags, mTurquoise (donor) and YFP(acceptor) (56). When N2AMC4R cells were transfected withTEpacVV and incubated with forskolin, which directly activatesadenylate cyclase, the fluorescence of mTurquoise increased andthat of YFP decreased (Fig. 7A), thus increasing the ratio ofmTurquoise fluorescence to YFP fluorescence (the FRET ratio)(Fig. 7B), as is consistent with a previous report (56). These dataindicate that TEpacVV can detect changes in intracellular cAMP

Fig. 6. HA-MC4R-GFP in a complex with intracellularα-MSH does not become desensitized. (A) Schematicdiagram of the experiments shown in B–D. N2A cellswere cotransfected with WT HA-MC4R-GFP and ei-ther EV or ER–α-MSH. CHX, cycloheximide. (B) Stain-ing of ER–α-MSH at the cell surface (anti-Myc, Cy3,red fluorescence) and HA-MC4R-GFP at the cell sur-face (anti-HA, Cy5, blue fluorescence) was done asdescribed in SI Methods. (Scale bar, 5 μm.) The figureshows a representative sample from two indepen-dent experiments. (C) The graph shows the quantifi-cation of the experiments in B [without α-MSH,number of cells (n) = 30; with 1 μMα-MSH, n = 35]. (D)Cells were incubated in the absence or presence of 1μM α-MSH for 2 h. Cell surface HA-MC4R-GFP wasmeasured as described in SI Methods. The graphshows a representative experiment in which sampleswere run in replicates of four. Each experimentwas done three times. (E ) Schematic diagram ofthe experiments shown in F. (F ) Cells were in-cubated in the absence or presence of 1 μMα-MSH for 2 h. Intracellular cAMP was measured in the presence of 100 nM α-MSH. The graph shows a representative experiment in which sampleswere run in replicates of eight. Each experiment was done three times. Data are expressed as mean ± SD. Statistical significance was assessed witha Student t test (C ) or one-way ANOVA (D and F ), ***P < 0.001; ns, nonsignificant.

E4738 | www.pnas.org/cgi/doi/10.1073/pnas.1219808110 Granell et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

concentration over time in N2AMC4R cells. When N2AMC4R cellsexpressing TEpacVV were exposed to extracellular α-MSH, theFRET ratio increased, indicating that binding of the agonist toHA-MC4R-GFP induces an increase of intracellular cAMP. Thisincrease occurred within ∼2.5 min, and the ratio remained ele-vated over the next 15 min. The addition of α-MSH at 100 nMappeared to be sufficient to bind to the entire population of cellreceptors, because subsequent additions of the hormone at higherconcentrations did not increase further the amount of intracellularcAMP being generated (Fig. S6). When AgRP was added to themedium of N2AMC4R cells previously exposed to extracellularα-MSH, the FRET ratio dropped (Fig. 7E). This decrease indi-cates that AgRP, by displacing extracellular α-MSH bound to HA-

MC4R-GFP, leads to a decrease in the intracellular concentrationof cAMP, consistent with the role of AgRP as an antagonist (12–16). The AgRP-dependent decrease in cAMP level also occurredwithin ∼2.5 min, an interval during which the population of HA-MC4R-GFP receptors at the cell surface changes by less than15%. [This value is derived from modeling, as described inSI Methods, the internalization data of HA-MC4R-GFP, which wepreviously found to be unchanged by binding to extracellular α-MSH(24).] Conversely, in the N2AMC4R cells cotransfected with ER–α-MSH, the TEpacVV FRET ratio already was elevated to the levelseen in N2AMC4R cells acutely stimulated with α-MSH and, as isconsistent with data in Fig. 2, did not change with the additionof extracellular α-MSH to the medium (Fig. 7D). Importantly,

Fig. 7. At the cell surface, the sensitivity of MC4Rin a complex with intracellular α-MSH to antago-nism by AgRP is different from that of MC4R ina complex with extracellular α-MSH. (A and B)N2AMC4R were transfected with mTurquoise-TE-PACVV-YFP. Cells were pretreated with 100 mMIBMX for 10 min and subsequently were treatedwith 1 mM forskolin at the indicated time. Thefluorescence of mTurquoise was excited at 458 nm,and the intensities of fluorescence emitted at 480–495 nm (mTurquoise) and at 535–565 nm (YFP) wererecorded over time (A), along with the ratio of theintensity of mTurquoise fluorescence to the in-tensity of fluorescence YFP (B). (C–F ) N2AMC4R cellswere transfected with mTurquoise-TEPACVV-YFPand with either empty vector (EV) (C and E) or ER–α-MSH (D and F). Cells were treated with 100 nMα-MSH and with or without 600 nM AgRP as in-dicated. Fluorescent ratios were calculated as in B.Graphs shown are of representative cells for eachcondition. (G) Schematic diagram of the experi-ments shown in H. (H) N2AMC4R cells transfectedwith mTurquoise-TEPACVV-YFP and ER–α-MSH wereincubated in the presence of 1 μM α-MSH for 2 h.Cells were washed once in DMEM and then wereincubated in DMEM with 100 nM α-MSH through-out the experiment. Cells were treated with 600 nMAgRP as indicated in the figure. Fluorescent in-tensities were recorded, and fluorescent ratios werecalculated as described above. For each condition,at least 15 cells were quantified in at least threeseparate experiments.

Granell et al. PNAS | Published online November 18, 2013 | E4739

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

the addition of AgRP did not decrease the cAMP level in cellsexpressing ER–α-MSH (Fig. 7F). Thus, it appears that AgRPdoes not displace α-MSH from MC4R at the cell surface whenthe complex initially was formed in the ER but does so when thecomplex initially was formed at the cell surface. However, thepossibility still existed that AgRP is unable to displace the in-tracellular α-MSH from HA-MC4R-GFP because of changes inthe α-MSH protein itself, such as the Myc epitope at the C ter-minus. To exclude this possibility, ER–α-MSH at the cell surfacewas replaced by extracellular α-MSH by pretreating cells with1 μM α-MSH, as in Fig. 6B, and then exposing them to 100 nMα-MSH for the duration of the assay (Fig. 7G). Importantly, evenwhen ER–α-MSH bound to HA-MC4R-GFP was replaced byextracellular α-MSH, the addition of AgRP did not decrease thecAMP level in cells expressing ER–α-MSH (Fig. 7H). The datasuggest that MC4R reaching the cell surface in a complex withintracellular α-MSH is stabilized in an active conformation that isdifferent from that of the receptor in a complex with extracellularα-MSH. These data also indicate that, when MC4R folds in thepresence of α-MSH in the ER, there is a persistent change inconformation of the receptor, and this change makes it resistant toantagonism by AgRP, whether or not the intracellular α-MSH hasbeen replaced by the extracellular hormone.

MC4R in a Complex with Intracellular α-MSH, Unlike MC4R in a Complexwith Extracellular α-MSH, Recycles Efficiently to the Plasma Membrane.MC4R desensitization occurs because a population of constantlyrecycling receptors, when bound to extracellular α-MSH, areretained along the endosomal pathway and do not reappear at theplasma membrane by being routed to lysosomes (22–25, 57). Wehave shown here that upon coexpression of ER–α-MSH thefraction of recycling HA-MC4R-GFP residing at the cell surface isthe same as that of the ligand-free receptor (Fig. S5B). Thereforeit is possible that when α-MSH bound to MC4R in the ER, thereceptor, although stabilized in an active conformation, is notretained in the endosomal/lysosomal compartment and thereforerecycles to the plasma membrane as efficiently as if it were un-occupied. To visualize the dynamics of HA-MC4R-GFP at the cellsurface, we used total internal reflectance fluorescence (TIRF)microscopy. By this approach, HA-MC4R-GFP appeared as dif-

fuse fluorescence and as brighter puncta (Fig. 8A), as is consistentwith data obtained previously (24). Time-lapse experiments car-ried out over an interval of 3 min indicated that, although most ofthe HA-MC4R-GFP puncta remained visible at the cell surfaceand changed their position minimally, some puncta appeared, andothers disappeared (Fig. 8B). We labeled these appearing/dis-appearing puncta as “mobile GFP puncta.” In the N2AMC4R cellskept in the basal state, the percentage of mobile puncta eitherappearing or leaving the field of vision was the same, 50% (Fig.8C, lanes 1 and 2). This observation is consistent with our previousfinding that the receptor recycles constitutively so that, at thesteady state in the absence of agonist, the same number ofreceptors arrives at and leaves the plasma membrane. However,when α-MSH was added to the medium, the percentages of mo-bile GFP puncta appearing and disappearing from the field ofvision were significantly different, with fewer puncta appearing atthe plasma membrane (Fig. 8C, lanes 3 and 4). This result isconsistent with our finding that binding to the extracellular agonistinhibits recycling of MC4R back to the cell surface (24). In theN2AMC4R cells coexpressing ER–α-MSH, the percentage of mo-bile GFP puncta appearing and disappearing from the plasmamembrane was the same regardless of the addition of α-MSH tothe medium (Fig. 8C, lanes 5–8). These observations indicate thatMC4R in a complex with ER–α-MSH does not become desensi-tized because the receptor recycles efficiently to the plasmamembrane. The result is consistent with the concept that a differ-ence in the conformation of MC4R in a complex with intracellularand extracellular α-MSH, respectively, underlies specific proper-ties, such the ability to cycle efficiently to the cell surface, as shownin Fig. 8, and resistance to AgRP antagonism at the cell surface, asshown in Fig. 7.

MC4R in a Complex with Intracellular α-MSH, Unlike MC4R in a Complexwith Extracellular α-MSH, Does Not Route to Lysosomes. As describedabove, we have found that when MC4R initially binds to α-MSHin the ER, the receptor does not become desensitized underprolonged exposure to extracellular α-MSH (Fig. 6) and cycles tothe plasma membrane as efficiently as the unoccupied receptor(Fig. 8). These observations suggest that the complex MC4R/in-tracellular α-MSH, unlike MC4R/extracellular α-MSH, does not

Fig. 8. At the cell surface, MC4R in a complex with intracellular α-MSH hasdynamics similar to those of the unoccupied receptor. (A) A frame froma TIRF time-lapse experiment (0–3 min) showing puncta of HA-MC4R-GFP atthe cell surface of live N2AMC4R cells. (Scale bar, 1.5 μm.) (B) Puncta corre-sponding to HA-MC4R-GFP and visualized during a 3-min interval are dividedin three categories: White arrows indicate puncta remaining stably at thecell surface; red arrows indicate puncta disappearing from the cell surface;and green arrows indicate puncta appearing at the cell surface. A detaileddescription of the TIRF experiments and quantification can be found inSI Methods. (Scale bar, 1.5 μm.) (C) N2AMC4R cells were either mock-trans-duced or transduced with ER–α-MSH–plenti6 lentiviral particles. Cells wereimaged in the presence or absence of 100 nM α-MSH. Quantification wasdone from three independent experiments as described in SI Methods.[mock, number of cells (n) = 13; mock + 100 nM α-MSH, n = 16; ER–α-MSH, n =15; ER–α-MSH + 100 nM α-MSH = 6]. Data are expressed as mean ± SD. ***P <0.001; one-way ANOVA; ns, nonsignificant.

Fig. 9. MC4R in a complex with intracellular α-MSH does not route tolysosomes. (A) GT1-7 cells were cotransfected with LAMP1-GFP, WT HA-MC4R, and either empty vector (EV) or ER–α-MSH. Cells were incubated withanti-HA antibodies for 1 h at 37 °C. Immunofluorescence staining, confocalimaging, and quantification of the data were done as described in SI Methods.LAMP1-GFP (GFP, green fluorescence) and HA-MC4R at the cell surface andendosomes (CS + end; anti-HA, Cy3, red fluorescence) are shown. Whiteindicates colocalization pixels. Arrows indicate the portion of the imagesenlarged in the Insets. (B) The graph shows the quantification of theexperiments in A [EV, number of cells (n) = 16; EV + 100 nM α-MSH, n = 24;ER–α-MSH, n = 20]. Quantification was done from two independentexperiments. Data are expressed as mean ± SD. ***P < 0.001; one-wayANOVA; ns, nonsignificant.

E4740 | www.pnas.org/cgi/doi/10.1073/pnas.1219808110 Granell et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

route to lysosomes when internalized. To evaluate this possibility,we transiently transfected hypothalamic GT1-7 and N2A cellswith HA-MC4R and the lysosomal marker lysosomal-associatedmembrane protein 1 LAMP1-GFP. Cells were treated with ex-tracellular α-MSH or were left untreated in the presence of anti-HA antibody for 1 h at 37 °C to label the entire pool of receptorsconstitutively trafficking between the plasma membrane and theendosomal compartment (24). The localization of this pool ofrecycling receptors was visualized by staining fixed and per-meabilized cells with secondary antibodies conjugated to Cy3 (Fig.9A and Fig. S7A, red fluorescence). In GT1-7 and N2A cells nottreated with α-MSH, colocalization of HA-MC4R with LAMP1-GFP was ∼10% (Fig. 9B and Fig. S7B). Conversely, when GT1-7and N2A cells were incubated with α-MSH, colocalization of HA-MC4R with the lysosomal marker increased by approximatelysixfold. Importantly, when HA-MC4R was coexpressed with ER–α-MSH, the amount of receptor colocalizing with LAMP1-GFPdid not increase significantly compared with cells that were nottreated with α-MSH. These results indicate that, although thebinding of MC4R to ER–α-MSH along the biosynthetic pathwaystabilizes an active conformation of the receptor, it does not in-duce routing to lysosomes. These data are consistent with theefficient cycling of the receptor in a complex with the intracellularα-MSH to the cell surface observed in Fig. 8.It has been reported the N-glycosylation site at position 187 in

the β-2 adrenergic receptor is necessary for desensitization byrouting the receptor to the lysosomes after exposure to the agonist(58). This observation suggested that the change in N-glycosylationof MC4R caused by the binding to ER–α-MSH results in aninability of MC4R to become desensitized. However, all theHA-MC4R-GFP receptors mutated at N-glycosylation sites,including N17Q, disappeared from the plasma membrane inresponse to extracellular α-MSH to the same extent as the WTreceptor (Fig. S7B). Thus, the change in N-glycosylation ofMC4R resulting from coexpression with ER–α-MSH does notseem, by itself, to be sufficient to confer receptor resistanceto desensitization.

ConclusionsThe most important result of this paper is that MC4R can exist ina fully active conformation that does not become desensitizedand that such a conformation is achieved when MC4R formsa complex with α-MSH in the ER. Under these conditions, thereceptor is poised to signal constantly at the same amplitude asMC4R acutely exposed to extracellular α-MSH. Constant sig-naling by MC4R in a complex with the intracellular agonist likelyoccurs outside the biosynthetic pathway because inhibition ofprotein synthesis does not affect this function, which insteadappears to be dependent on an acquired resistance to de-sensitization. In this respect, we find that MC4R in a complexwith intracellular α-MSH cycles efficiently to the cell surface anddoes not route to lysosomes, similar to unoccupied MC4R andunlike MC4R in a complex with the extracellular α-MSH (22–25,57). Our data suggest that the resistance of MC4R to de-sensitization in a complex with intracellular α-MSH most likelyis caused by the receptor’s being stabilized in an active confor-mation different from that of the receptor bound to extracellularα-MSH. This conclusion is based on data gathered using ourtemporally resolved FRET-based assays to measure real-timechanges in intracellular cAMP levels. Using this assay, we showthat AgRP antagonizes signaling by MC4R in a complex withextracellular α-MSH, as is consistent with previous reports (12–16). However, when the receptor/agonist complex is formedalong the secretory pathway, AgRP does not attenuate MC4Rsignaling at the cell surface, thus suggesting a change in theconformation of the receptor. Such change in conformationappears to be an intrinsic property of the receptor folded in thepresence of α-MSH because it remains insensitive to antagonism

by AgRP, even when intracellular α-MSH is replaced with ex-tracellular α-MSH. We find that intracellular α-MSH interactswith WT-MC4R in the ER because N-glycosylation at Asn17 ofthe receptor is abolished. Consistent with that finding, expressionof the intracellular α-MSH promotes folding of an obesity-linkedvariant of MC4R that otherwise is retained in the ER, allowing itto exit the biosynthetic pathway and signal constantly. For otherGPCRs, it has been found that different classes of drugs can alterthe active conformation of the receptors so that they becomedesensitized to different extents (36–39, 59). Our results suggestthat, at least for MC4R, the cellular localization where agonistbinding initially takes place affects the conformation and thedesensitization properties of the receptor. In this respect, proteinfolding in the ER is dependent on the surrounding environment(60), which here is modified by coexpression of the MC4R li-gand. Several cell-permeable inverse agonist/antagonists havebeen shown to increase the exit of obesity-linked variants ofMC4R from the biosynthetic pathway (32–34); here we offerevidence that targeting an agonist to the ER can have a similareffect. Our data suggest that a membrane-permeable small-molecule agonist would be a better approach than an antagonistto rescue MC4R function of variants because it would lead toconstant/maximal signaling by the rescued receptors, as shownhere, and it would not have to be displaced by α-MSH to stim-ulate receptor activity (29).It has been reported that peripheral administration of potent

MC4R agonists to mice that are obese because of prolongedtreatment with high-fat diet leads to tachyphylaxis after few daysof treatment, an effect that may occur by compensatory up-regulation of anorexigenic hormones such as AgRP and/or bydesensitization of the receptor by prolonged exposure to theagonist (26, 27). The data in this paper indicate that intracellulardelivery of α-MSH blunts both desensitization of MC4R andantagonism by AgRP. This work offers evidence that small-molecule or peptide-based agonists that are able to reach the ERmay constitute a therapeutic approach to promote constant sig-naling by MC4R, and perhaps by other GPCRs, by stabilizingthem in an active conformation that is resistant to de-sensitization.

MethodsReagents, antibodies, and constructs, fluorescence microscopy, determinationof HA-MC4R-GFP at the cell surface by enzyme-linked immunoassay, immu-noprecipitation of HA-MC4R-GFP, and phosphorylation of AMPK are describedin SI Methods.

Cell Culture, Transfection, and Transduction. Neuroblastoma N2A (AmericanType Culture Collection) and immortalized hypothalamic GT1-7 cells werecultured in DMEM with 10% (vol/vol) FBS and penicillin/streptomycin. (Fordetails, see SI Methods.)

Assay to Determine cAMP. Intracellular cAMP was measured by using thedirect cAMP enzyme immunoassay kit from Enzo Life Sciences following themanufacturer’s instructions. (For details, see SI Methods.)

Deglycosylation of MC4R. PNGase F was purchased from New England Biolabs.(For details, see SI Methods.)

Statistical Analysis. Data are expressed as mean ± SD. The experiments wereperformed in triplicate and done at least three times. Data were analyzed byusing GraphPad Prism version 5.0 Software and were compared using theStudent t test and one-way ANOVA as indicated.

ACKNOWLEDGMENTS.We thank Dr. Richard I. Weiner (University of California,San Francisco) for the kind gift of GT1-7 cells; Dr. Jennifer Lippincott-Schwartz(National Institutes of Health) for the kind gift of LAMP1-GFP; Dr. Kees Jalink(The Netherlands Cancer Institute) for the kind gift of TEpacVV; Dr. SameerMohammad for the construction of ER–γ+α-MSH; and Dr. Paul Miller, Dr. AlanDiekman, and Dr. Faith K. Cragle for comments on the experiments and on themanuscript. This work was supported by National Institutes of Health GrantsR01-DK080424 (to G.B.) and F30-DK095569 (to B.M.M.).

Granell et al. PNAS | Published online November 18, 2013 | E4741

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0

1. Cone RD (2005) Anatomy and regulation of the central melanocortin system. NatNeurosci 8(5):571–578.

2. Tao Y-X (2010) The melanocortin-4 receptor: Physiology, pharmacology, and patho-physiology. Endocr Rev 31(4):506–543.

3. Williams KW, Elmquist JK (2012) From neuroanatomy to behavior: Central integrationof peripheral signals regulating feeding behavior. Nat Neurosci 15(10):1350–1355.

4. Huszar D, et al. (1997) Targeted disruption of the melanocortin-4 receptor results inobesity in mice. Cell 88(1):131–141.

5. Vaisse C, Clement K, Guy-Grand B, Froguel P (1998) A frameshift mutation in humanMC4R is associated with a dominant form of obesity. Nat Genet 20(2):113–114.

6. Yeo GS, et al. (1998) A frameshift mutation in MC4R associated with dominantly in-herited human obesity. Nat Genet 20(2):111–112.

7. Hinney A, et al. (1999) Several mutations in the melanocortin-4 receptor gene in-cluding a nonsense and a frameshift mutation associated with dominantly inheritedobesity in humans. J Clin Endocrinol Metab 84(4):1483–1486.

8. Yeo GS, et al. (2003) Mutations in the human melanocortin-4 receptor gene associ-ated with severe familial obesity disrupts receptor function through multiple mo-lecular mechanisms. Hum Mol Genet 12(5):561–574.

9. Mountjoy KG, Mortrud MT, Low MJ, Simerly RB, Cone RD (1994) Localization of themelanocortin-4 receptor (MC4-R) in neuroendocrine and autonomic control circuits inthe brain. Mol Endocrinol 8(10):1298–1308.

10. Kishi T, et al. (2003) Expression of melanocortin 4 receptor mRNA in the centralnervous system of the rat. J Comp Neurol 457(3):213–235.

11. Balthasar N, et al. (2005) Divergence of melanocortin pathways in the control of foodintake and energy expenditure. Cell 123(3):493–505.

12. Graham M, Shutter JR, Sarmiento U, Sarosi I, Stark KL (1997) Overexpression of Agrtleads to obesity in transgenic mice. Nat Genet 17(3):273–274.

13. Mizuno TM, Mobbs CV (1999) Hypothalamic agouti-related protein messenger ribo-nucleic acid is inhibited by leptin and stimulated by fasting. Endocrinology 140(2):814–817.

14. Ollmann MM, et al. (1997) Antagonism of central melanocortin receptors in vitro andin vivo by agouti-related protein. Science 278(5335):135–138.

15. Wortley KE, et al. (2005) Agouti-related protein-deficient mice display an age-relatedlean phenotype. Cell Metab 2(6):421–427.

16. Xiang Z, et al. (2006) Pharmacological characterization of 40 human melanocortin-4receptor polymorphisms with the endogenous proopiomelanocortin-derived agonistsand the agouti-related protein (AGRP) antagonist. Biochemistry 45(23):7277–7288.

17. Nijenhuis WA, Oosterom J, Adan RA (2001) AgRP(83-132) acts as an inverse agonist onthe human-melanocortin-4 receptor. Mol Endocrinol 15(1):164–171.

18. Haskell-Luevano C, Monck EK (2001) Agouti-related protein functions as an inverseagonist at a constitutively active brain melanocortin-4 receptor. Regul Pept 99(1):1–7.

19. Luquet S, Perez FA, Hnasko TS, Palmiter RD (2005) NPY/AgRP neurons are essential forfeeding in adult mice but can be ablated in neonates. Science 310(5748):683–685.

20. Gropp E, et al. (2005) Agouti-related peptide-expressing neurons are mandatory forfeeding. Nat Neurosci 8(10):1289–1291.

21. Ren H, et al. (2012) FoxO1 target Gpr17 activates AgRP neurons to regulate foodintake. Cell 149(6):1314–1326.

22. Shinyama H, Masuzaki H, Fang H, Flier JS (2003) Regulation of melanocortin-4 re-ceptor signaling: Agonist-mediated desensitization and internalization. Endocrinol-ogy 144(4):1301–1314.

23. Gao Z, et al. (2003) Agonist-dependent internalization of the human melanocortin-4receptors in human embryonic kidney 293 cells. J Pharmacol Exp Ther 307(3):870–877.

24. Mohammad S, et al. (2007) Constitutive traffic of melanocortin-4 receptor in Neuro2Acells and immortalized hypothalamic neurons. J Biol Chem 282(7):4963–4974.

25. McDaniel FK, et al. (2012) Constitutive cholesterol-dependent endocytosis of mela-nocortin-4 receptor (MC4R) is essential to maintain receptor responsiveness toα-melanocyte-stimulating hormone (α-MSH). J Biol Chem 287(26):21873–21890.

26. Blüher S, et al. (2004) Responsiveness to peripherally administered melanocortins inlean and obese mice. Diabetes 53(1):82–90.

27. Pierroz DD, et al. (2002) Effects of acute and chronic administration of the melano-cortin agonist MTII in mice with diet-induced obesity. Diabetes 51(5):1337–1345.

28. Krishna R, et al. (2009) Potent and selective agonism of the melanocortin receptor 4with MK-0493 does not induce weight loss in obese human subjects: Energy intakepredicts lack of weight loss efficacy. Clin Pharmacol Ther 86(6):659–666.

29. Ulloa-Aguirre A, Michael Conn P (2011) Pharmacoperones: A new therapeutic ap-proach for diseases caused by misfolded G protein-coupled receptors. Recent PatEndocr Metab Immune Drug Discov 5(1):13–24.

30. Tao YX, Segaloff DL (2003) Functional characterization of melanocortin-4 receptormutations associated with childhood obesity. Endocrinology 144(10):4544–4551.

31. Granell S, Mohammad S, Ramanagoudr-Bhojappa R, Baldini G (2010) Obesity-linkedvariants of melanocortin-4 receptor are misfolded in the endoplasmic reticulum andcan be rescued to the cell surface by a chemical chaperone. Mol Endocrinol 24(9):1805–1821.

32. Ward N-A, Hirst S, Williams J, Findlay JBC (2012) Pharmacological chaperones increasethe cell-surface expression of intracellularly retained mutants of the melanocortin 4receptor with unique rescuing efficacy profiles. Biochem Soc Trans 40(4):717–720.

33. Fan ZC, Tao YX (2009) Functional characterization and pharmacological rescue ofmelanocortin-4 receptor mutations identified from obese patients. J Cell Mol Med13(9B):3268–3282.

34. René P, et al. (2010) Pharmacological chaperones restore function to MC4R mutantsresponsible for severe early-onset obesity. J Pharmacol Exp Ther 335(3):520–532.

35. Granell S, et al. (2012) A novel melanocortin-4 receptor mutation MC4R-P272L asso-ciated with severe obesity has increased propensity to be ubiquitinated in the ER inthe face of correct folding. PLoS ONE 7(12):e50894.

36. Keith DE, et al. (1996) Morphine activates opioid receptors without causing theirrapid internalization. J Biol Chem 271(32):19021–19024.

37. Yu Y, et al. (1997) Mu opioid receptor phosphorylation, desensitization, and ligandefficacy. J Biol Chem 272(46):28869–28874.

38. Raehal KM, Schmid CL, Groer CE, Bohn LM (2011) Functional selectivity at the μ-opioidreceptor: Implications for understanding opioid analgesia and tolerance. PharmacolRev 63(4):1001–1019.

39. Hoffmann C, Ziegler N, Reiner S, Krasel C, Lohse MJ (2008) Agonist-selective, receptor-specific interaction of human P2Y receptors with beta-arrestin-1 and -2. J Biol Chem283(45):30933–30941.

40. Holder JR, Bauzo RM, Xiang Z, Haskell-Luevano C (2002) Structure-activity relation-ships of the melanocortin tetrapeptide Ac-His-DPhe-Arg-Trp-NH(2) at the mousemelanocortin receptors. 1. Modifications at the His position. J Med Chem 45(13):2801–2810.

41. Hadley ME, Haskell-Luevano C (1999) The proopiomelanocortin system. Ann N Y AcadSci 885:1–21.

42. Castrucci AM, Hadley ME, Wilkes BC, Hruby VJ, Sawyer TK (1989) Melanotropinstructure-activity studies on melanocytes of the teleost fish, Synbranchus marmor-atus. Gen Comp Endocrinol 74(2):209–214.

43. Chapman KL, Kinsella GK, Cox A, Donnelly D, Findlay JBC (2010) Interactions of themelanocortin-4 receptor with the peptide agonist NDP-MSH. J Mol Biol 401(3):433–450.

44. Hruby VJ, et al. (1987) alpha-Melanotropin: The minimal active sequence in the frogskin bioassay. J Med Chem 30(11):2126–2130.

45. Kornfeld R, Kornfeld S (1985) Assembly of asparagine-linked oligosaccharides. AnnuRev Biochem 54:631–664.

46. Jones J, Krag SS, Betenbaugh MJ (2005) Controlling N-linked glycan site occupancy.Biochim Biophys Acta 1726(2):121–137.

47. Srinivasan S, et al. (2004) Constitutive activity of the melanocortin-4 receptor ismaintained by its N-terminal domain and plays a role in energy homeostasis in hu-mans. J Clin Invest 114(8):1158–1164.

48. Janovick JA, et al. (2003) Structure-activity relations of successful pharmacologicchaperones for rescue of naturally occurring and manufactured mutants of the go-nadotropin-releasing hormone receptor. J Pharmacol Exp Ther 305(2):608–614.

49. Morello JP, et al. (2000) Pharmacological chaperones rescue cell-surface expressionand function of misfolded V2 vasopressin receptor mutants. J Clin Invest 105(7):887–895.

50. VanLeeuwen D, Steffey ME, Donahue C, Ho G, MacKenzie RG (2003) Cell surfaceexpression of the melanocortin-4 receptor is dependent on a C-terminal di-isoleucinesequence at codons 316/317. J Biol Chem 278(18):15935–15940.

51. Mellon PL, et al. (1990) Immortalization of hypothalamic GnRH neurons by geneticallytargeted tumorigenesis. Neuron 5(1):1–10.

52. Büch TR, Heling D, Damm E, Gudermann T, Breit A (2009) Pertussis toxin-sensitivesignaling of melanocortin-4 receptors in hypothalamic GT1-7 cells defines agouti-re-lated protein as a biased agonist. J Biol Chem 284(39):26411–26420.

53. Khong K, Kurtz SE, Sykes RL, Cone RD (2001) Expression of functional melanocortin-4receptor in the hypothalamic GT1-1 cell line. Neuroendocrinology 74(3):193–201.

54. Minokoshi Y, et al. (2004) AMP-kinase regulates food intake by responding to hor-monal and nutrient signals in the hypothalamus. Nature 428(6982):569–574.

55. Damm E, Buech TR, Gudermann T, Breit A (2012) Melanocortin-induced PKA activa-tion inhibits AMPK activity via ERK-1/2 and LKB-1 in hypothalamic GT1-7 cells. MolEndocrinol 26(4):643–654.

56. Klarenbeek JB, Goedhart J, Hink MA, Gadella TWJ, Jalink K (2011) A mTurquoise-based cAMP sensor for both FLIM and ratiometric read-out has improved dynamicrange. PLoS ONE 6(4):e19170.

57. Overton JD, Leibel RL (2011) Mahoganoid and mahogany mutations rectify theobesity of the yellow mouse by effects on endosomal traffic of MC4R protein. J BiolChem 286(21):18914–18929.

58. Mialet-Perez J, Green SA, Miller WE, Liggett SB (2004) A primate-dominant thirdglycosylation site of the beta2-adrenergic receptor routes receptors to degradationduring agonist regulation. J Biol Chem 279(37):38603–38607.

59. Vilardaga JP (2006) Switching modes for G protein-coupled receptor activation. NatChem Biol 2(8):395–396.

60. Roth DM, Balch WE (2011) Modeling general proteostasis: Proteome balance inhealth and disease. Curr Opin Cell Biol 23(2):126–134.

E4742 | www.pnas.org/cgi/doi/10.1073/pnas.1219808110 Granell et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

26,

202

0