Ghrelin receptor conformational dynamics regulatethe transition from a preassembled to an activereceptor:Gq complexMarjorie Damiana, Sophie Marya, Mathieu Maingota, Céline M’Kadmia, Didier Gagnea, Jean-Philippe Leyrisa,1,Séverine Denoyellea, Gérald Gaibeletb, Laurent Gavaraa, Mauricio Garcia de Souza Costac, David Perahiac,Eric Trinquetd, Bernard Mouillacb, Ségolène Galandrine, Céline Galèse, Jean-Alain Fehrentza, Nicolas Floqueta,Jean Martineza, Jacky Mariea, and Jean-Louis Banèresa,2

aCNRS UMR 5247 and Faculté de Pharmacie, Institut des Biomolécules Max Mousseron, Université Montpellier 1 et 2, 34093 Montpellier Cedex 05, France;bCNRS UMR 5203, INSERM U661, and Institut de Génomique Fonctionnelle, Université Montpellier 1 et 2, 34094 Montpellier Cedex 05, France; cCNRS UMR8113 and Laboratoire de Biologie et de Pharmacologie Appliquée, Ecole Normale Supérieure de Cachan, F-94235 Cachan, France; dCisbio Bioassays, 30200Codolet, France; and eUMR 1048, Institut des Maladies Métaboliques et Cardiovasculaires, 31432 Toulouse Cedex 4, France

Edited by Robert J. Lefkowitz, Howard Hughes Medical Institute, Duke University Medical Center, Durham, NC, and approved December 8, 2014 (received forreview August 1, 2014)

How G protein-coupled receptor conformational dynamics controlG protein coupling to trigger signaling is a key but still open ques-tion. We addressed this question with a model system composedof the purified ghrelin receptor assembled into lipid discs. Combin-ing receptor labeling through genetic incorporation of unnaturalamino acids, lanthanide resonance energy transfer, and normalmode analyses, we directly demonstrate the occurrence of twodistinct receptor:Gq assemblies with different geometries whoserelative populations parallel the activation state of the receptor.The first of these assemblies is a preassembled complex with thereceptor in its basal conformation. This complex is specific of Gqand is not observed with Gi. The second one is an active assemblyin which the receptor in its active conformation triggers G proteinactivation. The active complex is present even in the absence ofagonist, in a direct relationship with the high constitutive activityof the ghrelin receptor. These data provide direct evidence of amechanism for ghrelin receptor-mediated Gq signaling in whichtransition of the receptor from an inactive to an active conforma-tion is accompanied by a rearrangement of a preassembled recep-tor:G protein complex, ultimately leading to G protein activationand signaling.

GPCR | G protein | preassembly | conformation dynamics | signaling

Gprotein-coupled receptors (GPCRs), one of the largest cellsurface receptor families, are involved in many cellular sig-

naling processes (1). Based on this property, as well as their impor-tance as drug targets, the molecular aspects of GPCR functioninghave been extensively investigated. In particular, coupling toheterotrimeric G proteins has been the focus of numerous stud-ies. Indeed, delineating the molecular mechanisms responsiblefor receptor:G protein interaction is absolutely required to betterunderstand how signaling is controlled. Recent years have seenspectacular advances that have culminated in elucidation of the3D structure of the β2-adrenergic receptor:Gs complex (2).Nevertheless, the need for further progress remains, in particularto fully understand the dynamics of this interaction. This is acrucial question, given that how the receptor interacts with its Gprotein partner governs signaling, and thus biological and path-ophysiological responses.To date, two different models for GPCR:G protein interaction

have been proposed: collision coupling and preassembly. Origi-nally, it was proposed that receptors and G proteins couple bycollision (3, 4). One of the main features of this model is thatonly activated receptors interact with G proteins. Since then,alternative models of signaling have been developed. One ofthese, the preassembly model, proposes that the receptor and theG protein make a complex even in the absence of agonist (5–8).

Discriminating between the two models is crucial. Indeed, sig-naling outputs, such as the kinetics of G protein activation, willbe significantly different depending on whether the ligand-freereceptor is always in complex with its G protein or must first beactivated by the agonist to recruit the G protein and triggersignaling. Moreover, it has been shown that GPCR conforma-tional dynamics (9–11) and signaling in the absence of ligand arekey features of GPCR functioning (12). How receptor constitu-tive activity and conformational dynamics relate to their couplingto the G protein remains an open question.Here we used the purified ghrelin receptor GHS-R1a to an-

alyze the way in which this GPCR interacts with its G proteinpartners. Ghrelin is a neuroendocrine peptide hormone that actsthrough its cognate GPCR to control important biological pro-cesses, such as growth hormone secretion, food intake, and re-ward-seeking behaviors (13). Among the GPCRs, GHS-R1a hasbeen shown to have one of the highest basal Gq activation levelsboth in vitro (10, 14) and in vivo (15, 16). The physiologicalrelevance of GHS-R1a basal activity is substantiated by the occur-rence of a natural human mutation in the GHS-R1a gene (A204Esubstitution in the second extracellular loop of the receptor)

Significance

G protein-coupled receptors (GPCRs), one of the largest cellsurface receptor families, transmit their signals through thecoupling of intracellular partners, such as the G proteins.Knowing how this coupling occurs is essential, because it gov-erns the entire signaling process. To address this open question,we used a purified GPCR as a model towhich we applied variousstate-of-the-art biochemical and biophysical approaches. Bydoing so, we provide direct experimental evidence of a signalingmechanism in which receptor conformational changes are di-rectly linked to a rearrangement of a preassembled complexbetween the receptor and its cognate Gq protein. This shedslight on the way in which a GPCR interacts with G proteins totrigger signaling.

Author contributions: S.M., N.F., J. Marie, and J.-L.B. designed research; M.D., C.M., D.G.,J.-P.L., G.G., M.G.d.S.C., B.M., S.G., C.G., and N.F. performed research; M.M., S.D., L.G., E.T.,and J.-A.F. contributed new reagents/analytic tools; M.D., S.M., C.M., D.G., G.G., M.G.d.S.C.,D.P., B.M., C.G., J.-A.F., N.F., J. Martinez, J. Marie, and J.-L.B. analyzed data; and B.M., J.-A.F.,N.F., J. Martinez, J. Marie, and J.-L.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Institut des Neurosciences de Montpellier, INSERM U1051, 34295 Mont-pellier Cedex 05, France.

2To whom correspondence should be addressed. Email: jean-louis.baneres@univ-montp1.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1414618112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1414618112 PNAS | February 3, 2015 | vol. 112 | no. 5 | 1601–1606

PHARM

ACO

LOGY

that dramatically decreases constitutive activity and is associ-ated with a short-stature phenotype (17). Along with its impor-tance in drug design, GHS-R1a is a prototype for peptide-activatedclass A GPCRs.To delineate the way in which the ghrelin receptor interacts

with G proteins, we used monomeric GHS-R1a reconstituted ina membrane-mimicking environment, lipid discs, and a combi-nation of innovative biochemical [labeling with unnatural aminoacid (UAA)] and biophysical [lanthanide resonance energy transfer(LRET) and normal mode (NM) analyses] approaches. By doingso, we provide the first direct evidence that ghrelin-mediated sig-naling involves a complex dialogue between the conformationaldynamics of the receptor and its ability to interact with the dif-ferent G protein subtypes to which it is coupled.

ResultsReceptor and Gq Labeling for LRET Measurements. Site-specific la-beling of the ghrelin receptor and its cognate Gq protein was firstrequired to monitor their interaction with LRET. The G proteinwas labeled on the free amino terminus of its αq subunit with thedonor fluorophore (Lumi4-Tb) through a classical reaction withits N-hydroxysuccinimide (NHS) derivative at neutral pH (18).This allowed specific labeling at the Gαq N terminus with ∼60%efficacy. Incomplete labeling does not affect LRET measure-ments, because only the emission of the acceptor originatingfrom energy transfer exclusively (sensitized emission) is mea-sured (19). Modification of the αq subunit did not affect theability of the G protein trimer to become activated (SI Appendix,Fig. S1A).The purified GHS-R1a was labeled through UAA technology.

To this end, the pEVOL vector (20) was used to encode p-azido-L-phenylalanine (azidoF) into the ghrelin receptor sequence inresponse to a unique amber stop codon. AzidoF was inserted inthe cytoplasmic face of GHS-R1a, where it replaced F71 at thecytoplasmic tip of TM1 (SI Appendix, Fig. S2). This modificationdid not affect either ligand binding or G protein activation (SIAppendix, Fig. S3). The azidoF-containing receptor was assem-bled as a monomer into lipid discs (10) and then labeled with thefluorescence acceptor (Alexa Fluor 488) using the strain-pro-moted alkyne-azide cycloaddition reaction (21, 22). Approxi-mately 90% of labeling was achieved under these conditions,whereas no labeling was observed with the unmodified receptor(SI Appendix, Fig. S4).

LRET-Monitored GHS-R1a:Gq Interaction. LRET measurementswere first carried out between the Gαqβ1γ2 trimer in which the αsubunit was labeled at its N terminus with the donor fluorophoreand the purified ghrelin receptor labeled with the green acceptorat the intracellular end of TM1. Measurements were done bymonitoring the decay in the acceptor-sensitized emission, so thatonly the donor and acceptor engaged in LRET were detected.This compensates for partial labeling of the G protein. In all ofthe experiments, we systematically used the monomeric receptorassembled into lipid discs, because this monomer is fully func-tional with regard to G protein activation (10).In the absence of ligand, the fluorescence decay was best fitted

by a double exponential function (Fig. 1A), with the occurrenceof two different lifetimes, τad1 and τad2 (Fig. 1B and Table 1).Because distinct emission decay lifetimes are related to distincttransfer efficiencies, and thus to distinct donor-to-acceptor dis-tances, these data indicate the occurrence of two equally popu-lated receptor:G protein assemblies with different geometries(Fig. 1C). The distance between the donor and the acceptor ineach of these assemblies was estimated using the time constantsof the acceptor-sensitized emission and the donor-only emission(Table 1). Importantly, the population associated with τad2 wasabolished in the presence of GTPγS, whereas that associatedwith τad1 was essentially unaffected (SI Appendix, Fig. S5), sug-gesting that the former is active with regard to Gq activation, butthe latter is not.We next investigated the effects of the binding of different

ligands—the neutral antagonist JMV3011 (23), the full agonistMK0677 (24), and the inverse agonist SPA (14)—on the acceptor-sensitized emission profiles. No difference between the ligand-free and antagonist-loaded receptor was observed (SI Appendix,Fig. S6A). In the presence of MK0677, the fluorescence decaywas still best fitted by a double exponential function (Fig. 1A),with lifetime values in the same range as those measured in theabsence of ligand (Table 1). A similar profile was obtained withthe natural agonist ghrelin (SI Appendix, Fig. S7), indicatingthat the two trigger similar effects. A comparison of the apo-and MK0677-loaded states showed that the distribution of τad1and τad2 changed in favor of τad2 in the presence of MK0677(Fig. 1B). This finding suggests that activation of the ghrelinreceptor by its agonist populates the receptor:G protein com-plex associated with τad2 (Fig. 1C). This complex certainly cor-responds to an active one, given that MK0677 binding is ac-companied by increased receptor-catalyzed G protein activation(SI Appendix, Fig. S3B). In contrast, no significant acceptor-sensitized emission signal could be measured in the presence of

100%

100%

apo

MK0677SPA

apo

MK0677SPA

no

rmal

ized

no

rmal

ized

A B C

D E F

Fig. 1. LRET-monitored GHS-R1a:Gq interaction. (A and D)Sensitized-emission decays from Gαqβ1γ2 and the WT re-ceptor (A) and the A204E mutant (D). The emission decayswere measured in the absence of ligand (blue dots), in thepresence of the 10 μM MK0677 (green dots), and in thepresence of 10 μM SPA (red dots). (B and E) Distribution ofthe sensitized-emission lifetimes measured for the WT re-ceptor (B) and the A204E mutant (E) and Gq in the absence orpresence of MK0677. (C and F) Schematic representation ofthe relative populations with the corresponding distancesbetween the probes. Green and red stars, donor and ac-ceptor fluorophores; green dot, MK0677, red triangle, SPA;cross on the receptor, A204E mutation. Data are presented asnormalized fluorescence intensity as a function of time andrepresent the average of four independent measurements.Absolute maximal fluorescence intensities were in the 800a.u. range.

1602 | www.pnas.org/cgi/doi/10.1073/pnas.1414618112 Damian et al.

SPA (Fig. 1A), suggesting that SPA dissociates the receptor:Gprotein complex (Fig. 1C). Consistent with this conclusion, thefluorescence decay of the donor obtained under these conditionswas similar to that measured in the presence of unlabeled receptor(SI Appendix, Fig. S8).

Effects of GHS-R1a Constitutive Activity on Its Interaction with Gq. Toassess whether the occurrence of the active complex in the ab-sence of ligand is related to the high constitutive activity of theghrelin receptor, we investigated the interaction between Gq andthe A204E mutant of GHS-R1a. As is the case in cellular systems(17, 25), the purified A204E mutant exhibits significantly de-creased constitutive activity while it can be fully activated withMK0677 (SI Appendix, Fig. S9).The mutant was labeled as described above for the WT re-

ceptor, and its interaction with the labeled α subunit in Gαqβ1γ2was monitored using LRET. In this case, the acceptor-sensitizedemission decay was still best fitted by a double exponentialfunction, but with a very major population with a lifetime decayclosely related to τad1 (Fig. 1 D and E and Table 1). Combinedwith the near absence of receptor-catalyzed GDP/GTP exchangeunder such conditions (SI Appendix, Fig. S9), this finding indicatesthat (i) τad1 is associated with an inactive complex, and (ii) thesignificant fraction of active GHS-R1a:Gq complex seen in theWT receptor is likely related to its high constitutive activity.The lifetime distribution profiles in the presence and absence

of the neutral antagonist were similar for the A204E mutant (SIAppendix, Fig. S6B). In the presence of the full agonist MK0677,the lifetime distribution was closely related to that observed withtheWT receptor, with the occurrence of two populations, a minorone (τad1) and a major one (τad2) (Fig. 1E and Table 1). Thisfinding is in agreement with the fact that theWT receptor and themutant can be equally activated by MK0677 (SI Appendix, Fig.

S9). Finally, no LRET signal could be measured in the presenceof SPA (Fig. 1 D and F).

Conformational Features of the WT Receptor and Its A204E Mutant.The differences in the coupling of the WT receptor and its A204Emutant to Gq are indicative of differences in the conformationalfeatures of these two proteins in their ligand-free state. Weassessed these differences on an experimental basis using intra-molecular LRET. In this case, the fluorescence acceptor was stillintroduced at the cytoplasmic end of TM1 as described above,whereas the donor was attached to the receptor through a uniquereactive cysteine in the intracellular tip of TM6 (SI Appendix,Fig. S2). Both the WT and A204E mutant double-labeled proteinsmaintained their ability to bind ligands and activate G proteins (SIAppendix, Fig. S3). All of the foregoing experiments were carriedout in the presence of the Gαqβ1γ2 trimer, because this affects thereceptor conformational transitions (10, 26), and thus its influenceshould be taken into account.In the absence of ligand, the acceptor-sensitized emission

decay of the WT receptor was best described by a double ex-ponential function (Fig. 2A and Table 2). This indicates that theligand-free WT GHS-R1a is characterized by the occurrence ofat least two different conformational states in essentially equalamounts, designated herein as basal and active conformationsfor Gq. The distances between fluorescent probes in these twoconformations are in the 30-Å (basal) and 40-Å (active) ranges(Table 2). Importantly, these two conformationally distinct re-ceptor populations likely are related to the two populations ofreceptor:G protein complexes; indeed, they are equally popu-lated (SI Appendix, Fig. S10). The acceptor-sensitized emissiondecay of the ligand-free A204E mutant also was best fitted bya double exponential function (Fig. 2B), but with a very majorpopulation with a lifetime value closely related to that of thebasal state for Gq (Table 2).We next assessed the effects of ligands on the distribution of

GHS-R1a conformations. As shown in Fig. 2, the intramolecularfluorescence decay profiles obtained for the WT and mutantreceptors in the presence of MK0677 were essentially undis-tinguishable, with the occurrence of a major population with alifetime value closely related to that of the active state for Gq(Table 2). Again, the occurrence of a major conformational stateclosely parallels the occurrence of a major receptor:G proteinpopulation in the intermolecular LRET measurements (SI Ap-pendix, Fig. S10). In the presence of the inverse agonist, theLRET profiles were best fitted with a single exponential functionwith a lifetime different from that of either the basal or the activeconformation (Table 2), suggesting the occurrence of a differentconformational state (designated ground conformation for Gq).Finally, in the presence of JMV3011, no significant change wasobserved in the acceptor-sensitized emission profiles of both the

Table 1. Sensitized emission lifetimes measured from Gαq (inthe Gαqβ1γ2 trimer) tagged at its N terminus with the donor andGHS-R1a tagged in TM1 with the acceptor fluorophore, andcorresponding distances between the two probes

τad1, μs D1, Å A1 τad2, μs D2, Å A2

wt, apo 599 ± 9 36.4 ± 0.3 0.47 1,071 ± 13 45.3 ± 0.4 0.53A204E, apo 586 ± 8 36.2 ± 0.2 0.95 1,091 ± 12 45.8 ± 0.3 0.05wt, MK0677 585 ± 7 36.1 ± 0.2 0.27 1,042 ± 11 44.6 ± 0.3 0.73A204E, MK0677 605 ± 9 36.8 ± 0.3 0.15 1,072 ± 12 45.3 ± 0.4 0.85wt, SPA — — — — — —

A204E, SPA — — — — — —

A is the relative population of each fraction.

A (wild type) B (A204E)

73%

39.7Å

48%

30.8Å

27%

31.1 Å

25.7Å

100%

39.9Å

52%

24.9Å

100%

9%

39.3Å

91%

30.7Å

89%

39.7Å

11%

31.1Å

apo

MK0677SPA

apo

MK0677SPA

no

rmal

ized

no

rmal

ized

Fig. 2. LRET-monitored GHS-R1a conformationaldynamics. Sensitized-emission decays from the WTreceptor (A) and A204E mutant (B) labeled with thedonor and acceptor fluorophores in TM1 and TM6(SI Appendix, Fig. S2). The emission decays weremeasured in the presence of Gαqβ1γ2 and in theabsence of ligand (blue dots), in the presence of10 μM MK0677 (green dots), and in the presence of10 μM SPA (red dots). (Insets) Schematic representa-tion of the relative populations with the corre-sponding distances between the probes. Green andred stars, donor and acceptor fluorophores; greendot, MK0677; red triangle, SPA; cross on the re-ceptor, A204E mutation. Data are presented as nor-malized fluorescence intensity as a function of timeand represent the average of four independentmeasurements. Absolute maximal fluorescence in-tensities were in the 750 a.u. range.

Damian et al. PNAS | February 3, 2015 | vol. 112 | no. 5 | 1603

PHARM

ACO

LOGY

WT receptor and the A204E mutant, indicating that this antag-onist does not affect the conformational landscape of the receptor(SI Appendix, Fig. S11).

A Structural Basis for the Changes in LRET. We performed NManalysis to assess whether the distances inferred from our ex-perimental LRET measurements are consistent with the struc-tural arrangement of a GPCR in complex with a G protein. NManalysis is considered one of the best techniques for providingkey information on the functional, collective motions of proteins(27). Because the β2-adrenergic (β2AR):Gs complex is the onlyone for which an X-ray structure is available, we used it as a model(2). It can be reasonably assumed that the general organizationof both class A receptors, GHS-R1a and β2AR, should not bestrikingly different, as was subsequently confirmed by the similaritybetween the distances measured experimentally with GHS-R1aand those inferred from the β2AR structures (see below).The two residues in β2AR that correspond to the labeled posi-

tions in GHS-R1a, F61 (TM1) and C251 (TM6), were identifiedbased on a multiple sequence alignment grouping of 200 humanGPCR sequences from the Swissprot database and 20 differentreceptor structures from the Protein Data Bank. The mean dis-tance measured between C251 (Cβ atom) and F61 (Cζ atom) throughthe 20-ns molecular dynamics trajectory was in the 39 Å range,in perfect agreement with the value of 39.9 Å measured for theactive conformation of GHS-R1a.The distance between the receptor and the G protein inferred

from LRET was more difficult to confirm because of the lack ofresidues 1–8 at the N terminus of the Gαs structure. Neverthe-less, the mean distance of ∼43 Å between T9 (Gαs; Cβ atom) andF61 (TM1; Cζ atom) all along the molecular dynamics trajectorywas again in good agreement with the ∼45 Å measured experi-mentally. Finally, an extrapolation of lacking residues at the Cterminus of TM6 in the inactive β2AR structure (28) suggestedthat the F61:C251 distance in this structure could be ∼25 Å, avalue again in agreement with the 25.7 Å measured experimen-tally for the SPA-loaded GHS-R1a.To assess whether the changes in distance inferred from LRET

are compatible with possible motions in a GPCR:G proteincomplex, we analyzed the changes in the TM1:TM6 and TM1:GαN-TER distances along the 10 lowest-frequency NMs com-puted for the β2AR:Gs complex. As shown in Fig. 3A, the plotthus obtained shows that these two distances can vary between 39Å and 30 Å for the TM1:TM6 distance and between 45 Å and 30Å for the TM1:GαN-TER distance, in full agreement with ourLRET data. Importantly, the TM1:TM6 and TM1:GαN-TERdistances were usually seen to increase or decrease simultaneously,as was the case experimentally.To exclude the possibility that the distance changes in the NM

analyses were biased by the fact that NM techniques are per-formed in vacuo, we used a new, original approach, termed mo-lecular dynamics with excited normal modes (MDeNM), toexplore the conformational space along a given normal mode di-rection when including the membrane and water (Fig. 3B and SIAppendix). These calculations confirmed that the large, collective

motion of the receptor observed in vacuo was still possible inthe presence of the membrane and surrounding water mole-cules. Importantly, although the amplitudes of most of themodes were reduced after reintroduction of the membrane,mode 16 was unaffected (Fig. 3C); thus, the motion of the re-ceptor along this mode (SI Appendix, Fig. S12 and Movie S1)could represent that observed experimentally.

GHS-R1a:G Protein Preassembly Depends on the G Protein Subtype.Because GHS-R1a can trigger activation of Gi as well as Gq (29),the foregoing data raise the question of whether preassembly iscommon to all G protein subtypes. To assess this point on an ex-perimental basis, we labeled Gαi2 with Lumi4-Tb under conditionssimilar to those used with Gαq, also with an efficacy in the 50–60%range. This modification did not affect the activity of the purifiedGi protein (SI Appendix, Fig. S1B).In contrast to what was observed with Gq, no significant ac-

ceptor-sensitized emission signal could be measured in the ab-sence of ligand for either the WT receptor or its A204E mutant(Fig. 4 A and B), indicating that that there is no preassemblybetween the ligand-free receptor and Gαi2β1γ2. This finding isconsistent with our previous study in which, in contrast to Gq, Gidid not affect the conformation of the monomeric ghrelin re-ceptor in the absence of ligand (10). In the presence of the fullagonist MK0677, however, a significant acceptor-sensitized emis-sion signal was measured (Fig. 4 A and B), indicating Gi recruit-ment on receptor activation.A possible explanation for the foregoing finding is that the

differences in the interaction mode of the ghrelin receptor withGi and Gq result in differences in the kinetics of GHS-R1a–catalyzed G protein activation. To assess this point on an exper-imental basis, we monitored the kinetics of G protein activationin HEK293T cells using BRET-based Gq and Gi protein acti-vation sensors (30). As shown in Fig. 4 C and D, significantdifferences in the activation kinetics of both G proteins subtypes

Table 2. Sensitized emission lifetimes measured from theghrelin receptor tagged with the donor and acceptorfluorophore, and corresponding distances between thecytoplasmic ends of TM1 and TM6

τad1, μs D1, Å A1 τad2, μs D2, Å A2

wt, apo 293 ± 4 30.8 ± 0.2 0.48 805 ± 8 39.9 ± 0.4 0.52A204E, apo 288 ± 3 30.7 ± 0.2 0.91 806 ± 10 39.9 ± 0.5 0.09wt, MK0677 306 ± 5 31.1 ± 0.3 0.27 793 ± 11 39.7 ± 0.5 0.73A204E,MK0677 298 ± 8 30.9 ± 0.4 0.11 795 ± 10 39.7 ± 0.4 0.89wt, SPA 113 ± 6 25.7 ± 0.3 1 — — —

A204E, SPA 118 ± 4 24.9 ± 0.2 1 — — —

A is the relative population of each fraction.

Fig. 3. Distance changes in the receptor:G protein complex. (A) Variationsof the TM1:TM6 and TM1:Gα-NTER distances along all of the energy-mini-mized conformations generated with the VMOD approach for the 10 lowest-frequency normal modes of the β2AR:Gs complex. (B) Typical large-ampli-tude motions observed during the MDeNM exploration of the mode 16computed for the B2AR:Gs complex. (C) Variations of the two TM1:TM6 andTM1:Gα-NTER distances along the concatenated simulations.

1604 | www.pnas.org/cgi/doi/10.1073/pnas.1414618112 Damian et al.

were observed with GHS-R1a in the presence of MK0677; thatis, activation of Gq occurred significantly faster than activationof Gi. In the case of GHS-R1a, the kinetics of Gi activation alsowas significantly slower than that measured for the Gi-coupledα2CAR (Fig. 4E).

DiscussionOur LRET data directly demonstrate that in the absence of li-gand and the presence of Gq, the ghrelin receptor is a dynamicprotein that displays at least two different conformations for Gq,a basal one and an active one. The agonist further stabilizes theactive conformation, whereas an inverse agonist stabilizes anadditional inactive conformational state (ground state for Gq).Associated with the basal and active receptor conformations aretwo different receptor:Gq complexes (see the model in SI Ap-pendix, Fig. S13). The first complex, associated with the activestate of GHS-R1a, is an active assembly responsible for receptor-catalyzed GDP-to-GTP exchange. The second GHS-R1a:Gqcomplex is associated with the basal receptor conformation ofthe ghrelin receptor and is characterized by a different structuralarrangement than that of the active complex. The fact that thissecond type of complex is the major one under conditions inwhich no G protein activation is observed (e.g., for the A204Emutant in the absence of agonist) and is not affected by GTPγSsuggests that it is inactive with regard to receptor-catalyzed GDP/GTP exchange. Thus, it is mechanistically distinct from the activeassembly and likely corresponds to a preassembled GHS-R1a:Gqinactive complex.Because LRET reports for proximity but not necessarily for

a direct interaction, we cannot totally exclude the possibility thatpreassembly results from an interaction of the G protein with thelipid bilayer of the nanodisc rather than with the receptor. Thepreassembled complex is readily dissociated on binding of theinverse agonist, however. Moreover, no LRET signal was observed

with lumi4-Tb–labeled Gq or Gi and empty discs containingfluorescein-labeled lipids (SI Appendix, Fig. S14). This stronglyindicates that the preassembled GHS-R1a:Gq complex results,at least to some extent, from a direct interaction of Gq with theghrelin receptor.Importantly, GHS-R1a:G protein preassembly was observed

with Gq but not with Gi. This finding is reminiscent of what hasbeen proposed for PAR1 that is preassembled to Gi but not toG12 (31). This means that the basal conformation of the ghrelinreceptor that is competent for interaction with Gq, although ina nonproductive way, is not competent for interaction with Gi,suggesting slight but nevertheless significantly different modesof interaction between the receptor and its different G proteinsubtype partners, as has been proposed for Gs and Gi (32). Apossible explanation for this may be that the affinity of theghrelin receptor in its basal conformation is lower for Gi than forGq. Regardless of the mechanistic details, this implies that GHS-R1a signaling along its different G protein-dependent pathwaysinvolves an intricate dialogue between receptor conformationaltransitions and interaction with distinct G protein subtypes.Our intramolecular LRET data demonstrate the occurrence

of two different conformations of the ghrelin receptor that areinactive with regard to Gq activation. The first of these con-formations is the basal state, which is observed for the WT re-ceptor in the absence of ligand and is the major species for theconstitutively inactive A204E mutant. The second is the groundconformation stabilized by the inverse agonist. This means thattwo distinct conformational states can give rise to a similar func-tional output (i.e., no Gq activation). Both states neverthelessdiffer in their ability to interact with the G protein. Whereas thebasal conformation is responsible for the inactive preassembledcomplex with Gq, no receptor:G protein interaction was observedwith the ground state stabilized by SPA. A possible explanation forthis finding is that the GHS-R1a basal state represents some sortof a preactivated conformation able to make a stable complex withits G partner as a first step along the Gq signaling pathway.Importantly, NM analyses using the β2AR:Gs complex as a

model are fully consistent with our intramolecular and inter-molecular LRET experiments. This strongly indicates that thedistances and their changes that we measured experimentally aretotally relevant on a structural basis. Adding consistency to ourconclusions, our NM analyses revealed that the antisymmetricrotation of the receptor on activation also promoted a coupledmotion of the G protein in which the Gα and Gβγ subunits de-viate from each other (SI Appendix, Fig. S12), in agreement withprevious FRET/BRET data (7).In contrast to the WT receptor, the constitutively inactive A204

E mutant in its apo state displays a major conformation, thebasal one. The observation that the inactive mutant is frozen in abasal inactive conformation, whereas the WT receptor oscillatesbetween this same basal state and an active state, provides anadditional piece of evidence for the model in which the absenceof constitutive activity is associated with reduced conformationaldynamics of GHS-R1a (25). Restriction of the conformationalflexibility of the A204E mutant could result from an additionalstructural constraint introduced at the level of the e2 loop onsubstitution of the A204 residue (25). Consistent with occurrenceof the basal conformation as a major state, the inactive pre-assembled complex is the major species in the case of the ligand-free, constitutively inactive A204E mutant.It has been shown that the ghrelin receptor can function as a

dimer (33). Dimerization may add an additional level of com-plexity to the interaction of GHS-R1a with its cognate G proteins.For instance, we have shown that homodimers are asymmetricassemblies with one protomer in its active state and the other inthe basal conformation for Gq (34). If the behavior of GHS-R1aprotomers with regard to interaction with G proteins is similar inthe monomer and in the dimer, then each of the protomers withinthe asymmetric dimer should have a different mode of interactionwith Gq. Thus, dimerization likely will impact the way in which

A (wild type) B (A204E)

apo

MK0677

apo

MK0677

C (GHS-R1a, Gq)

MK0677

D (GHS-R1a, Gi)

MK0677

0 10 20 30 40 50 60 70 801.1

1.2

1.3

1.4

1.5

1.6

1.7

BR

ET

ra

tio

Time (s)

E ( 2cAR, Gi)

UK 14,304

Fig. 4. LRET-monitored receptor:Gi interaction and kinetics of G proteinactivation. (A and B) Sensitized-emission decays from Gαi2β1γ2 and the WTreceptor (A) or its A204E mutant (B) labeled with the donor and acceptorfluorophores, respectively. The emission decays were measured in the ab-sence of ligand (blue dots) or in the presence of 10 μM MK0677 (green dots).Data are presented as normalized fluorescence intensity as a function oftime and represent the average of four independent measurements. Abso-lute maximal fluorescence intensities were in the 700 a.u. range. (C and D)Time changes in the BRET2 signal in HEK293T cells coexpressing Gαq-91Rluc8(C) or Gαi2-91Rluc8 (D), Gβ1, GFP10-Gγ2, and GHS-R1a after stimulation with1 μM MK0677. (E) Time changes in the BRET2 signal in HEK293T cells coex-pressing Gαi2-91Rluc8 and the α2C receptor after stimulation with 1 μM of itsUK 14,304 agonist.

Damian et al. PNAS | February 3, 2015 | vol. 112 | no. 5 | 1605

PHARM

ACO

LOGY

GHS-R1a couples to Gq, but the extent to which it will affectpreassembly remains to be delineated.In closing, our data provide a direct experimental evidence for

a specific receptor:Gq protein preassembly, as well as a plausiblemechanism for the consequences of the receptor conformationaltransitions on the assembly with one of its main signaling part-ners, Gq. Ultimately, this sheds light on the relationships be-tween the conformational dynamics of a GPCR and how it willinteract with its cognate G protein partner to trigger signaling.

Materials and MethodsUAA Labeling. The TAG amber codon was introduced by site-directed mu-tagenesis into the pET21a-α5-GHS-R1a vector (10) at the position encodingF71 of GHS-R1a. This mutation also was introduced in a cysteine-free mutant(10) with single reactive cysteine at position 255. The same modificationswere introduced in the vector encoding the A204E mutant. Incorporation ofthe modified amino acid was then carried out as detailed in SI Appendix.

Protein Labeling. Labeling of Gαq and Gαi2 on their N termini with the Lumi4-terbium (Tb) cryptate (CisBio) was carried out using the NHS derivative of thefluorophore at neutral pH (18) (SI Appendix). For labeling GHS-R1a on Cys255,the purified receptor was incubated with the fluorescence donor overnightat 16 °C in the presence of 100 μM tris(2-carboxyethyl)phosphine (SI Ap-pendix). Coupling of Alexa Fluor 488 to azidoF71 was performed by over-night incubation of the receptor with Click-IT Alexa Fluor 488 DIBO Alkyne(Life Technologies) (SI Appendix).

Spectroscopy. A cuvette-based fluorescence lifetime spectrometer with a high-powered pulsed Xe lamp as the excitation source was used for all fluorescencemeasurements (excitation 337 nm). Donor-only lifetimes were recorded usingthe labeled G protein and unlabeled receptor under the same conditions. Thefluorescence decays at 515 nm were normalized to the maximum fluorescenceand fitted to a sum of discrete exponential functions. The distances betweenthe donor and acceptor fluorophores were calculated using the LRET lifetime(τad) and donor-only lifetime (τd) using the Förster equation (35) (SI Appendix).

BRET2-Monitored G Protein Activation. Receptor andG protein sensor constructs(7, 30) were transiently cotransfected into HEK293T cells. For kinetics analyses,5 μM deep blue C (Interchim) was added before injection of the ligand (1 μM).The net BRET signal was obtained by calculating the ratio of GFP10 emission(515 ± 10 nm) over Rluc8 light emission (400 ± 10 nm) (SI Appendix).

NM Analyses. The NM analyses are described in SI Appendix.

ACKNOWLEDGMENTS. We thank P. G. Schultz and L. Supekova (The ScrippsResearch Institute, La Jolla, CA) for the pEVOL-pAzF plasmid, and S. Granierand J.-P. Pin (Institut de Génomique Fonctionnelle, Montpellier, France) forfruitful discussions. This work was supported by grants from the CentreNational de la Recherche Scientifique, Universités Montpellier 1 and 2, andAgence Nationale de la Recherche (PCV08-323163, ANR-10-BLAN-1208-01,ANR-10-BLAN-1208-02, ANR-13-BSV8-0006-01, and ANR-10-BINF-03-10). Cal-culations were carried out with the support of HPC@LR, a center of compe-tence in high-performance computing, funded by the Languedoc-Roussillonregion, the European Council, and the Université Montpellier 2.

1. Lagerström MC, Schiöth HB (2008) Structural diversity of G protein-coupled receptorsand significance for drug discovery. Nat Rev Drug Discov 7(4):339–357.

2. Rasmussen SG, et al. (2011) Crystal structure of the β2 adrenergic receptor-Gs proteincomplex. Nature 477(7366):549–555.

3. Orly J, Schramm M (1976) Coupling of catecholamine receptor from one cell with ad-enylate cyclase from another cell by cell fusion. Proc Natl Acad Sci USA 73(12):4410–4414.

4. Tolkovsky AM, Levitzki A (1978) Mode of coupling between the beta-adrenergic re-ceptor and adenylate cyclase in turkey erythrocytes. Biochemistry 17(18):3795.

5. Neubig RR, Gantzos RD, Thomsen WJ (1988) Mechanism of agonist and antagonistbinding to alpha 2 adrenergic receptors: Evidence for a precoupled receptor-guaninenucleotide protein complex. Biochemistry 27(7):2374–2384.

6. Qin K, Dong C, Wu G, Lambert NA (2011) Inactive-state preassembly of G(q)-coupledreceptors and G(q) heterotrimers. Nat Chem Biol 7(10):740–747.

7. Galés C, et al. (2006) Probing the activation-promoted structural rearrangements inpreassembled receptor-G protein complexes. Nat Struct Mol Biol 13(9):778–786.

8. Nobles M, Benians A, Tinker A (2005) Heterotrimeric G proteins precouple with Gprotein-coupled receptors in living cells. Proc Natl Acad Sci USA 102(51):18706–18711.

9. Kahsai AW, et al. (2011) Multiple ligand-specific conformations of the β2-adrenergicreceptor. Nat Chem Biol 7(10):692–700.

10. Damian M, et al. (2012) High constitutive activity is an intrinsic feature of ghrelin re-ceptor protein: A study with a functional monomeric GHS-R1a receptor reconstituted inlipid discs. J Biol Chem 287(6):3630–3641.

11. Rahmeh R, et al. (2012) Structural insights into biased G protein-coupled receptor sig-naling revealed by fluorescence spectroscopy. Proc Natl Acad Sci USA 109(17):6733–6738.

12. Bond RA, Ijzerman AP (2006) Recent developments in constitutive receptor activityand inverse agonism, and their potential for GPCR drug discovery. Trends PharmacolSci 27(2):92–96.

13. Sivertsen B, Holliday N, Madsen AN, Holst B (2013) Functionally biased signallingproperties of 7TM receptors: Opportunities for drug development for the ghrelinreceptor. Br J Pharmacol 170(7):1349–1362.

14. Holst B, Cygankiewicz A, Jensen TH, Ankersen M, Schwartz TW (2003) High consti-tutive signaling of the ghrelin receptor—identification of a potent inverse agonist.Mol Endocrinol 17(11):2201–2210.

15. Holst B, Schwartz TW (2004) Constitutive ghrelin receptor activity as a signaling set-point in appetite regulation. Trends Pharmacol Sci 25(3):113–117.

16. Petersen PS, et al. (2009) In vivo characterization of high basal signaling from theghrelin receptor. Endocrinology 150(11):4920–4930.

17. Pantel J, et al. (2006) Loss of constitutive activity of the growth hormone secreta-gogue receptor in familial short stature. J Clin Invest 116(3):760–768.

18. Pellissier LP, et al. (2011) G protein activation by serotonin type 4 receptor dimers: Evi-dence that turning on two protomers is more efficient. J Biol Chem 286(12):9985–9997.

19. Selvin PR (2000) The renaissance of fluorescence resonance energy transfer. Nat StructBiol 7(9):730–734.

20. Young TS, Ahmad I, Yin JA, Schultz PG (2010) An enhanced system for unnatural

amino acid mutagenesis in E. coli. J Mol Biol 395(2):361–374.21. Huber T, Naganathan S, Tian H, Ye S, Sakmar TP (2013) Unnatural amino acid muta-

genesis of GPCRs using amber codon suppression and bioorthogonal labeling. Methods

Enzymol 520:281–305.22. Tian H, et al. (2014) Bioorthogonal fluorescent labeling of functional G-protein–

coupled receptors. ChemBioChem 15(12):1820–1829.23. Moulin A, et al. (2008) New trisubstituted 1,2,4-triazole derivatives as potent ghrelin

receptor antagonists, 3: Synthesis and pharmacological in vitro and in vivo evalua-

tions. J Med Chem 51(3):689–693.24. Patchett AA, et al. (1995) Design and biological activities of L-163,191 (MK-0677): A

potent, orally active growth hormone secretagogue. Proc Natl Acad Sci USA 92(15):

7001–7005.25. Mokrosi�nski J, Frimurer TM, Sivertsen B, Schwartz TW, Holst B (2012) Modulation of

constitutive activity and signaling bias of the ghrelin receptor by conformational

constraint in the second extracellular loop. J Biol Chem 287(40):33488–33502.26. Louet M, et al. (2013) Conformational restriction of G-protein coupled receptors

(GPCRs) upon complexation to G-proteins: A putative activation mode of GPCRs? FEBS

Lett 587(16):2656–2661.27. Bahar I, Lezon TR, Bakan A, Shrivastava IH (2010) Normal mode analysis of biomolecular

structures: Functional mechanisms of membrane proteins. Chem Rev 110(3):1463–1497.28. Cherezov V, et al. (2007) High-resolution crystal structure of an engineered human

beta2-adrenergic G protein-coupled receptor. Science 318(5854):1258–1265.29. Camiña JP, Lodeiro M, Ischenko O, Martini AC, Casanueva FF (2007) Stimulation by

ghrelin of p42/p44 mitogen-activated protein kinase through the GHS-R1a receptor:

Role of G-proteins and beta-arrestins. J Cell Physiol 213(1):187–200.30. Saulière A, et al. (2012) Deciphering biased-agonism complexity reveals a new active

AT1 receptor entity. Nat Chem Biol 8(7):622–630.31. Ayoub MA, Trinquet E, Pfleger KD, Pin JP (2010) Differential association modes of the

thrombin receptor PAR1 with Galphai1, Galpha12, and beta-arrestin 1. FASEB J 24(9):

3522–3535.32. Rose AS, et al. (2014) Position of transmembrane helix 6 determines receptor G

protein coupling specificity. J Am Chem Soc 136(32):11244–11247.33. Schellekens H, Dinan TG, Cryan JF (2013) Taking two to tango: A role for ghrelin

receptor heterodimerization in stress and reward. Front Neurosci 7:148.34. Mary S, et al. (2013) Heterodimerization with its splice variant blocks the ghrelin re-

ceptor 1a in a non-signaling conformation: a study with a purified heterodimer as-

sembled into lipid discs. J Biol Chem 288(34):24656–24665.35. Cha A, Snyder GE, Selvin PR, Bezanilla F (1999) Atomic scale movement of the voltage-

sensing region in a potassium channel measured via spectroscopy. Nature 402(6763):

809–813.

1606 | www.pnas.org/cgi/doi/10.1073/pnas.1414618112 Damian et al.