Molecular Mechanism Underlying Partial and Full Agonism Mediated by the Human Cholecystokinin-1 Receptor*

The cholecystokinin-1 receptor (CCK1R) is a G protein-coupled receptor (GPCR) that regulates important physiological functions. As for other GPCRs, the molecular basis of full and partial agonism is still far from clearly understood. In the present report, using both laboratory experiments and molecular modeling approaches, we have investigated the partial agonism mechanism of JMV 180, on the human CCK1R. We first showed that efficacy of the CCK1R to activate phospholipase C is dependent on the correct orientation of the C-terminal end of peptidic ligands toward residue Phe330 of helix VI. We have previously reported that a single mutation of Met121 (helix III) markedly reduced the receptor-mediated inositol phosphate production upon stimulation by CCK. Computational simulations predicted that residue 121 affected orientation of the C-terminal end of CCK, thus suggesting that the molecular complex with a reduced inositol phosphate production observed with the mutated CCK1R resembles that resulting from binding of JMV 180 to the WT-CCK1R. Pharmacological, biochemical, and functional characterizations of the two receptor·ligand complexes with decreased abilities to signal were carried out in different cell types. We found that they presented the same features, such as total dependence of inositol phosphate production to Gαq expression, single affinity of binding sites, insensitivity of binding to non-hydrolyzable GTP, absence of GTPγ[S35] binding following agonist stimulation, similarity of dose-response curves for amylase secretion, and incapacity to induce acute pancreatitis in pancreatic acini. We concluded that helices VI and III of the CCK1R are functionally linked through the CCK1R agonist binding site and that positioning of the C-terminal ends of peptidic agonists toward Phe330 of helix VI is responsible for extent of phospholipase C activation through Gαq coupling. Given the potential therapeutic interest of partial agonists such as JMV 180, our structural data will serve for target structure-based design of new CCK1R ligands.

G protein-coupled receptors (GPCRs) 1 regulate a wide range of physiological processes by transmitting signals into cells in response to stimuli such as light, Ca 2ϩ , odorants, amino acids, nucleotides, peptides, or proteins (1). Structurally, GPCRs are integral membrane proteins composed of seven transmembrane helices connected by extracellular and intracellular loops, an N-terminal extracellular domain, and a C-terminal intracellular region. GPCRs are by far the most successful drug targets as evidenced by the fact that more than 50% of marketed drugs treat diseases by acting on some 20 GPCRs (2).
Compounds that stimulate GPCRs are classified as either full agonists or partial agonists. Partial agonists are of therapeutic interest because of their low propensity to exert sideeffects and to desensitize their target (3). The actions of agonists and partial agonists are initiated at the ligand binding site on the receptor. Binding sites in GPCRs are physically distant from the site of functional coupling to G proteins, which resides at the intracellular face of the GPCRs (4 -8). Consequently, triggering of G protein activation by agonists implies that agonist will induce conformational change of the target receptor from its resting state to an active state (4, 9 -11). In the context of this simplified two-states model for GPCRs activation, full and partial agonism may be explained by ligand induced stabilization of different receptor conformations having distinct affinities for a given G protein or distinct abilities to activate it. Alternatively, full and partial agonists may stabilize the receptor in the same active conformation, but the two types of ligands will generate quantitatively different amounts of this active receptor conformation due to their distinct intrinsic activities.
The cholecystokinin-1 receptor (CCK1R) is a member of group I in the superfamily of GPCRs (12). CCK1R-mediated effects include control of gallbladder contraction, pancreatic exocrine secretion, gastric emptying and gut motility, and satiety (13). Among available synthetic agonists of the CCK1R, JMV 180 is a CCK-related peptide (Fig. 1), which can act as a full agonist, a dual agonist/antagonist, or a partial agonist depending on species, tissue, or biological event considered (14 -18). Interestingly, on the contrary to CCK administered at high doses, JMV 180 does not cause acute pancreatitis in rats and can antagonize CCK-induced pancreatitis (19). With re-* This work was supported by the Association pour la Recherche sur le Cancer (Grants 4430 and 3282) and by the Ligue contre le Cancer. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Recipient of a grant from the Association pour la Recherche Contre le Cancer.
In the recent period, we have devoted much effort toward mapping of the human CCK1R binding site for the full agonist CCK, because we reasoned that knowledge of receptor binding site is required for the understanding of ligand action. In the current work, we investigated the molecular mechanism of JMV 180 partial agonism on the CCK1R using a multidisciplinary approach. We found that helices VI and III of the CCK1R are functionally linked through the CCK1R agonist binding site, and positioning of the C-terminal ends of peptidic agonists toward Phe 330 of helix VI is responsible for the extent of phospholipase C activation through G␣ q coupling.
Computer Modeling-The model of the CCK1R was built as described previously (24). A starting conformation of the JMV 180 molecule was firstly built and refined by 10,000 steps of energy minimizations using Insight II software (Accelrys, San Diego, CA). The consistent valence forcefield was used in the Discover Accelrys molecular mechanics and molecular dynamics program. This starting conformation was next used for a simulated annealing conformational sampling using molecular dynamics in vacuo. The dielectric constant used here was a distancedependent one to stimulate roughly the electrostatic shielding due to the solvent. The simulated annealing method used consisted of 100 loops of slow cooling, each one leading to a low energy conformer. Each loop starts by fixing the temperature to 1,000 K, followed by 5,000 molecular dynamics steps of 1 fs each. The temperature was then decreased by steps of 100 K: by decreasing the temperature this way every 5,000 molecular dynamics steps, after 40,000 steps the temperature of the system correspond to 300 K. The final conformation obtained at the end of this process was energy-refined and after storage was used to start a new simulation at 1,000 K with a slow cooling as described above. This procedure produced 100 minimized and different conformations of JMV 180. All these 100 conformers were used for the docking procedure using the GOLD algorithm (25). The "chem score" scoring function was used. GOLD ranked the conformational samples in such a way that we decided to examine only the 10 top solutions. From these 10 solutions, only 2 were found in accordance with pharmacological results (see "Results"). The two resulting complexes were submitted to annealing molecular dynamics calculations followed by energy minimization, the backbone of the receptor being first fixed. Then, all the constraints were removed excepting hydrogen bonds within helices, and finally the whole receptor-ligand complex was submitted to energy minimization. For JMV 172 and (Met)-CCK, models of the peptide and of the complexes were obtained as described here for JMV 180. The whole modeling was carried out with Accelrys software Insight II modules on a Silicon Graphics O 2 station.
Site-directed Mutagenesis and Transfection of COS-7 Cells-Mutant receptor cDNAs were constructed by oligonucleotide-directed mutagenesis (QuikChange TM site-directed mutagenesis kit, Stratagene) using the human CCK1R cDNAs cloned into pRFENeo vector as template. Mutations were confirmed by automated sequencing of both cDNA strands (Applied Biosystems). COS-7 cells (1.6 ϫ 10 6 ) were plated onto 10-cm culture dishes and grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal calf serum. After overnight incubation, cells were transiently transfected with 0.1-0.5 g/plate of pRFENeo vectors containing the cDNA for the wild-type or mutated CCK1R, using a modified DEAE-dextran method. Cells were transferred to 24-well plates at a density of 20,000 -80,000 cells/well 24 h after transfection, depending on the transfected mutant and experiment to be performed.

Membrane Preparation, Receptor Binding, and Inositol Phosphates
Assays-Plasma membranes from COS-7 and from pancreatic acinar cells were prepared as described previously (26.). Both binding and inositol phosphates measurements were rigorously performed as described previously (26).
Construction of Recombinant Adenoviruses-Recombinant adenoviruses encoding the wild-type or mutant human CCK1R, or encoding the mouse G␣ q subunit, were produced according to the method described previously (27). Briefly, the wild-type or mutant human CCK1R cDNAs, and the mouse G␣ q cDNA, were cloned into the shuttle vector pAdTrack-CMV, linearized, and co-transformed into Escherichia coli strain BJ5183 along with the adenoviral backbone vector, pAd-Easy-1. Recombinants were selected for kanamycin resistance and confirmed by restriction endonuclease analyzes. Finally, linearized recombinant plasmids were transfected into an adenovirus packaging cell line, HEK293 cells. Recombinant adenoviruses were collected 7-12 days after infection and were concentrated using a CsCl gradient. Under the control of distinct cytomegalovirus promoters, these plasmids express the protein of interest and, in parallel, green fluorescent protein. The titer of the viral stocks was estimated by green fluorescent protein assays. Viruses were stored in a 2ϫ storage buffer (10 mM Tris, pH 8.0, 100 mM NaCl, 0.1% BSA, and 50% glycerol, filter sterilized) at Ϫ80°C.
Preparation of Acini and Infection with Viruses-The pancreas was excised from freely feed male CCK1R Ϫ/Ϫ mice (26). Pancreas was washed in KREBS buffer containing 0.2% bovine serum albumin, 0.1 mg/ml soybean trypsin inhibitor, and 115 mM NaCl, 4.8 mM KCl, 0.5 mM CaCl 2 (2H 2 O), 4.8 mM MgSO 4 (7 H 2 O), 1.2 mM KH 2 PO 4 , 24.5 mM HEPES, 2 mM L-glutamine, 5 mM D-glucose, and essential and nonessential amino acids. The pH was adjusted to 7.4 and equilibrated with 100% O 2 before use. The acini were injected with collagenase and enzymatic digestion was then achieved at 37°C under mechanical shearing. Then, acini were wash two times by sedimentation through 0.2% BSA in KREBS buffer, and acini of medium size were selected. Finally, acini were purified two times by sedimentation through 1% BSA in KREBS buffer. The acini of CCK1R Ϫ/Ϫ mice suspended in KREBS buffer 1% BSA were infected with virus encoding human CCK1R wild type or M121V-CCK1R mutant. All viruses were used at ϳ10 7 plaque-forming units/mg of protein. Virus was added to acini in 2 ml of Krebs buffer for 10 min. The acini were then diluted with KREBS to 10 ml, transferred to 100-mm dishes, and incubated 4 h at 37°C in a humidified 5% CO 2 atmosphere.
Analysis of Amylase Secretion and Pancreatic Acinar Cells Morphology-Infected acini were washed and resuspended in KREBS buffer containing 1% BSA at ϳ1 mg/ml. Aliquots of 2 ml were incubated with an increasing concentration of CCK or JMV 180 for amylase assay and with a high concentration of CCK (10 Ϫ6 M) or JMV 180 (3 ϫ 10 Ϫ5 M) for morphology studies, at 37°C for 30 min, under gently shearing. Incubation was terminated by centrifugation of aliquots for 1 min at 400 rpm. The concentration of amylase in the medium was measured using Infinity TM amylase reagent (Sigma Diagnostics®). Results were expressed as percentage of maximal amylase release. For morphological studies, acinar pellets were fixed 20 h in 100% alcohol, 40% formaldehyde, and 17 M acetic acid (75/25/25), embedded in paraffin, hematoxylin and eosin stained, and visualized by optic microscopy.

JMV 180, a High Affinity Agonist of the CCK1R with a Low
Intrinsic Activity-In COS-7 cells expressing the human CCK1R receptor, (Nle)-CCK and (Met)-CCK ( Fig. 1) indistinguishably stimulated production of inositol phosphates with a potency (ED 50 ), of 0.42 Ϯ 0.10 nM and a maximal stimulation of 15-to 25-fold the basal value (Fig. 2). In the same cells, JMV 180-induced production of inositol phosphates reached a max- imum (efficacy) representing 9.3 Ϯ 1.1% of that achieved by (Nle)-CCK, the ED 50 being 17.8 Ϯ 6.3 nM. On the other hand, non-sulfated CCK, which is known to bind to the CCK1R with 500-fold lower affinity than sulfated CCK, stimulated inositol phosphate formation with a very weak potency (ED 50 : 215 Ϯ 18 nM) but an efficacy that represented 75% of that of CCK (Fig.  2). Partial agonism of JMV 180 in COS-7 cells was maintained at its value of efficacy for expression levels of CCK1R ranging from 2 to 6.8 pmol/10 6 cells (not illustrated), indicating that partial agonism was independent of expression ratio between CCK1R and its intracellular effectors. Thus, partial agonism of JMV 180 to stimulate inositol phosphate formation appears to be due to its weak intrinsic activity.
Identification of CCK1R Residues Involved in the JMV 180 Binding Site-We postulated that the weak intrinsic activity of JMV 180 should be due to the inability of this compound to interact with some amino acids of the CCK1R binding site playing a key role in the activation process. On the other hand, JMV 180 displays a strong structural homology with CCK ( Fig.  1), suggesting that binding sites for JMV 180 and CCK are likely overlapping. We therefore tested the effect of mutation of residues previously shown to be part of CCK1R binding site, on molecular recognition of JMV 180. Amino acids of the CCK1R binding site for CCK are shown on the serpentine representation of the CCK1R (Fig. 3). Binding studies with labeled JMV 180 allowed identification of three main categories of mutants (Table I). A first series of CCK1R mutants bound [ 3 H]JMV 180 with identical or very similar affinities to the wild-type CCK1R Table I) These mutants were previously shown to respond to CCK with strong drops of potency as the mutated residues are involved in CCK1R binding site (Table II) (26, 28 -30). Owing to structural homology between JMV 180 and CCK, defective or decreased [ 3 H]JMV 180 binding to mutants F107A, M195L, R197M, R336M, and I352A can be interpreted as an involvement of those residues in the binding site of JMV 180. Additional experiments to evaluate contribution of these residues to recognition of JMV 180 were performed, by measuring inositol phosphate production induced by the mutants in response to JMV 180. However, most of the mutants did not produce sufficient amounts of inositol phosphates to allow accurate analysis (not illustrated). As a partial agonist, JMV 180 was able to inhibit CCK1R stimulations by the full agonist CCK. We took advantage of this behavior to evaluate contribution of mutated amino acids to recognition of JMV 180, by measuring potency of JMV 180 to competitively inhibit CCK-induced production of inositol phosphates on mutants. Inhibition curves indicated that JMV 180 was 15.9-, 20.4-, 127.3-, 124.7-, and 9.5-fold less potent to inhibit CCKinduced formation of inositol phosphates on mutants F107A, M195L, R197M, R336M, and I352A than on the WT-CCK1R, respectively ( Fig. 4 and Table II). The F mut values show that some residues subjected to mutation are similarly involved in recognition of CCK and JMV 180, others are differentially involved, and some are involved in activity of CCK but not in that of JMV 180. This supports partial overlapping of the  In the two models, JMV 180 presented very similar positions, but the orientation of the C-terminal phenyl group was remarkably different, suggesting that this ligand may adopt two conformations within the receptor groove (Fig. 5, A and B). Trp of JMV 180 approximated Phe 107 (distance: 2.87 Å); the Tyr aromatic ring was at 6.70 Å from Met 195 sulfide, a distance that was ideal for quadrupole-quadrupole interactions (31); the sulfate moiety of JMV 180 and the penultimate Asp carboxylate residue of JMV180 paired with Arg 197 and Arg 336 guanidium side chains, respectively (distances: 2.36 and 1.90 Å). The comparison of CCK⅐CCK1R and JMV 180⅐CCK1R complexes (Fig.  5, A-C) revealed very similar positions of the two peptides in the receptor groove, except for their C-terminal moieties. Indeed, in the CCK⅐CCK1R complex, the phenyl side chain of the  Experimental Evidences That Different Positions of C-terminal Ends of JMV 180 and CCK in the CCK1R Binding Site Are Responsible for Their Distinct Efficacies-We further experimentally verified if the C-terminal aromatic rings of JMV 180 and CCK were differently positioned in the CCK1R binding site and if these distinct positions were responsible for their different intrinsic activities. We first incorporated a bulky residue (Phe) at position 329 in the CCK1R, to determine whether this mutation could affect recognition of CCK and JMV 180. The I329F mutant, which did not bind radiolabeled CCK, responded to CCK stimulations with a 600-fold decreased potency. This result, which can be explained by steric hindrance of Phe 329 , is in accordance with the stretched position of the C-terminal aromatic ring of CCK (Fig. 6A). In contrast, I329F mutant bound [ 3 H]JMV 180 with an affinity that was similar to that of the WT-CCK1R (k d : 8.3 Ϯ 0.6 nM, versus 9.3 Ϯ 1.4 nM, n ϭ 3) (Fig. 6B) thus supporting a strong preference for a folded conformation of JMV 180 in the CCK1R binding site.
In a previous work, a folding of the C-terminal aromatic ring of CCK toward Met 121 side chain was found in the modeled (M121V)-CCK1R⅐CCK complex. Moreover, this complex was experimentally shown to lack any biological activity (26). In contrast, only a stretched conformation of CCK was found in fully active CCK⅐CCK1R complex. Therefore, we hypothesized that the relative position of the C-terminal aromatic ring of ligands within the bottom of the receptor pocket could determine the extent of CCK1R activation. In line with this view, JMV 180⅐CCK1R complex in which JMV 180 displays a folded position of its C-terminal end (Fig. 5A) should correspond to an inactive one, whereas JMV 180⅐CCK1R complex exhibiting a stretched conformation of JMV 180 should represent the active one (Fig. 5B). Accordingly, the lack of preference for a stretched positioning of the JMV 180 C-terminal aromatic ring could explain both the absence of effect of Phe 329 mutation on JMV 180 recognition and its weak efficacy.
To experimentally verify that the stretched conformation of bound-JMV 180 corresponded to the active one, we searched, in Jean Martinez's library of peptides, for a peptide having a C-terminal phenyl group constrained in a stretched position. JMV 172, which has two phenyl moieties at its C-terminal end, fulfilled such a criterion (Fig. 1). The JMV 172⅐CCK1R complex obtained from modeling studies suggested that this peptide should mimic both conformations of JMV 180 at the same time, because one C-terminal aromatic ring was folded and the other was stretched and located toward Phe 330 (Fig. 7A). JMV 172 inhibited CCK binding to COS-7 cells expressing the CCK1R with a 500-fold lower potency than JMV 180 (K i : 1.6 Ϯ 0.3 m, Hill coefficient 0.99, not illustrated) but stimulated production of inositol phosphates with a maximum that was 3.8-fold higher than that of JMV 180 (Fig. 7B). Therefore, constraining the phenyl moiety of a CCK ligand toward Phe 330 increased the level of agonist efficacy on phospholipase C activation.
To test the critical role of Phe 330 in CCK1R activation, we exchanged Phe 330 for an Ala. This mutation abolished inositol phosphate responses of CCK1R to both JMV 180 and JMV 172 and diminished that to CCK by 43% (not illustrated). Thus, although weakly involved in binding, Phe 330 strongly contributed to CCK1R activation. These results represent an additional support for a crucial role of the T-shape interaction between the C-terminal phenyl moiety of peptidic ligands and Phe 330 of CCK1R in the activation process of phospholipase-C.
Both CCK and JMV 180 Stimulate Production of Inositol Phosphate through G␣ q Coupling-Agonist-directed trafficking of coupling, a concept whereby different agonists of the same GPCR can functionally induce distinct G protein coupling, have been documented for a number of GPCRs (32). Moreover, several types of phospholipase C have been identified, including ␤1, ␤2, ␤3, and ␤4, which can be stimulated either by G␣ q/11 or ␤␥ dimers (33). In the case of CCK1R, stimulation of phospholipase C via a pertussis toxin-sensitive (PTX) G protein was reported for cat gallbladder muscle (34), although this receptor is recognized to predominantly couple to G␣ q (16). We therefore examined the possibility that partial agonism of JMV 180 could result from CCK1R coupling to a G protein distinct from that involved in CCK stimulations. Pretreatment with PTX of COS-7 cells expressing the WT-CCK1R did not significantly affect inositol phosphate responses to both CCK and JMV 180.
In control experiments with COS-7 cells expressing the ␣ 2adrenergic receptor reported to weakly stimulate phospholipase C through PT-sensitive G proteins, a 75% decrease of inositol phosphate production was observed in presence of PTX (not illustrated). This first series of experiments, showing that PTX-sensitive G proteins are not involved in human CCK1Rmediated activation of phospholipase C in COS-7 cells strengthened the hypothesis that both CCK and JMV 180 stimulate phospholipase C through coupling to a PTX-insensitive G protein, presumably G q .
We then examined if JMV 180-induced production of inositol phosphates was abolished by triple mutation of amino acids Lys 308 -Lys 309 -Arg 310 located at the C-terminal end of the third intracellular loop of the CCK1R (Fig. 3). Indeed, the triplebasic sequence Lys 308 -Lys 309 -Arg 310 , that is conserved in the CCK2R, has been shown to be without any importance on G␣ i /PLA2 activation, but its mutation abolished coupling to G␣ q /␣ 11 /PLC (35). As illustrated on Fig. 8A, triple mutation K308M/K309T/R310L fully inhibited ability of the CCK1R to induce inositol phosphate production in response to JMV 180 in COS-7 cells. This triple basic residues mutation altered response of the WT-CCK1R to CCK by 60% (Fig. 8A). In control binding experiments, the triple mutant bound radiolabeled CCK to a single and low affinity class of binding sites (K d ϭ 15.3 Ϯ 0.7 nM, B max : 0.99 Ϯ 0.04 pmol/10 6 cells) and had its affinity for radiolabeled JMV 180, which remained unaffected (K d ϭ 8.2 Ϯ 0.6 nM, B max : 2.3 Ϯ 0.2 pmol/10 6 cells for the triple mutant versus 9.7 Ϯ 0.6 nM, B max : 13.0 Ϯ 0.8 pmol/10 6 for the WT-CCK1R).
As a second approach, we expressed CCK1R in EF-88 cells derived from ␣q/11 Ϫ/Ϫ mice. Attempts to transfect CCK1R cDNA-containing plasmid using different protocols were unsuccessful, confirming previous reports (36). This led us to use adenoviral infections. In absence of co-infection with adenovirus encoding mouse G␣ q , EF-88 cells expressing CCK1R did not respond to both JMV 180 and to CCK. In contrast, EF-88 cells co-expressing G␣ q and CCK1R responded to both CCK and JMV 180, and efficacy of JMV 180 represented 6.1 Ϯ 0.5% of that of CCK (Fig. 8B).
At this step of the study, our results strongly support that the full agonist, CCK, and the partial agonist, JMV 180, stimulate production of inositol phosphates through coupling to the same G protein, most likely G q . Moreover, they indicate that positioning of C-terminal ends of the peptidic ligands in the CCK1R binding site is responsible for their different efficacies to stimulate inositol phosphate production.

Partial Agonism of (Met)-CCK on the M121V-CCK1R Involves a Molecular Mechanism Similar to Partial Agonism of JMV 180
on the WT-CCK1R-A set of previous results suggested that activation of the CCK1R, and positioning of the C-terminal phenyl moiety of the peptidic ligand, can be regulated by exchanging residues in the vicinity of Met 121 of the receptor (26). In particular, although M121V-CCK1R mutant did not respond to (Nle)-CCK, partial stimulation was recovered if (Met)-CCK was used. This led us to consider that partial agonism of (Met)-CCK on the M121V-CCK1R could involve a molecular mechanism similar to that of JMV 180 on the WT-CCK1R.
We compared molecular and functional features of (Met)-CCK⅐M121V-CCK1R complex with that of JMV180⅐CCK1R complex. Automated docking of (Met)-CCK into the binding site of M121V-CCK1R yielded two complexes in which the C-terminal aromatic ring of CCK Phe was either in a stretched or in a folded conformation (not shown), as found for JMV180⅐CCK1R complex (Fig. 5, A and B). The second feature of JMV180⅐CCK1R complex was its inability to activate phospholipase C via a PTX-sensitive G protein but its functional coupling to G␣ q . Experiments with PTX-treated cells expressing M121V-CCK1R indicated that inositol phosphate production in response to (Met)-CCK was not affected (not illustrated). Triple mutation of Lys 308 -Lys 309 -Arg 310 in M121V-CCK1R completely inhibited partial stimulation by (Met)-CCK (Fig. 9A). So, integrity of Lys 308 -Lys 309 -Arg 310 motif appeared as identically critical for activation of both M121V-CCK1R by (Met)-CCK and CCK1R by JMV 180. Furthermore, as for stimulation of inositol phosphate production by JMV 180 in COS-7 cells expressing the WT-CCK1R, stimulation of M121V-CCK1R by (Met)-CCK was totally dependent on re-expression of G␣ q in EF-88 cells (Fig. 9B).
A third characteristic of JMV180⅐CCK1R complex, which is in good accordance with its coupling to G␣ q and known to be little sensitive to GTP (37), was reflected by binding of [ 3 H]JMV 180 to COS-7 cells expressing the WT-CCK1R. Indeed, a single affinity class of binding sites was found, confirming previous findings of another group, with a peptide analogous to JMV 180 (38). Furthermore, when performed on plasma membranes from the same cells, this binding of behave similarly with respect to action of GTP, further support that the two complexes are functionally dependent on the same ␣-subunit of G protein, most likely G␣ q . On the other hand, binding properties of CCK⅐CCK1R complex is in accordance with previously reported CCK-induced coupling of the CCK1R to several G-protein subunits, including G␣ s and G␣ i (12).
Finally, two additional recognized features of JMV 180 are the shape of its dose-response curve for amylase secretion in rat pancreatic acini and its inability to induce acute pancreatitis in rats. We investigated whether responses of pancreatic acini expressing M121V-CCK1R to stimulation by (Met)-CCK presented these features. Pancreatic acini were prepared from mice deficient for CCK1R gene expression, to be used as host cells. They were infected with adenovirus encoding the M121V-CCK1R, for comparison with acini infected with the virus encoding the wild-type human CCK1R. As shown in Fig. 10A, mouse acini expressing the wild-type human CCK1R secreted amylase in response to both (Nle)-CCK and (Met)-CCK according to a biphasic dose-response curve, the concentration giving 50% of maximal response (ED 50 ) being of ϳ56 pM. JMV 180 stimulated amylase release with the same maximum as CCK, but the dose-response curve was monophasic, ED 50 being 23.8 Ϯ 18.3 nM (Fig. 10B). Similarly, (Met)-CCK stimulated amylase release from acini expressing the M121V-CCK1R according to a monophasic curve, ED 50 being 42.6 Ϯ 11.3 nM. In contrast, (Nle)-CCK did not stimulate amylase release from acini expressing the M121V-CCK1R, a result that is in line with inability of M121V-CCK1R to respond to (Nle)-CCK.
Because marked differences were observed in the biological response profiles of pancreatic acini, we analyzed binding characteristics of the different biological models. CCK binding to acini expressing the WT-CCK1R occurred on two affinity sites having dissociation constants: K d1 ϭ 0.22 Ϯ 0.03 nM and K d2 ϭ 60.2 Ϯ 10.3 nM, respectively (not shown). CCK binding to acini expressing M121V-CCK1R occurred on a single affinity of binding sites having a K d ϭ 37.4 Ϯ 6.6 nM. Thus, comparison between the potency of the agonist and its affinity revealed that, as previously observed on rat pancreatic acini, potency of CCK for amylase secretion (56.8 Ϯ 9.4 pM) is 10-fold higher compared with high affinity dissociation constant of CCK for the WT-CCK1R (K d1 ϭ 0.22 Ϯ 0.03 nM), reflecting intracellular signal amplification (16). In contrast, potency of JMV 180 was close to dissociation constant value of the partial agonist for the receptor (23.8 Ϯ 18.3 and 9.3 Ϯ 1.4 nM). Similarly, potency of (Met)-CCK (D 50 : 42.6 Ϯ 11.3 nM) was in agreement with its affinity for M121V-CCK1R mutant (K d ϭ 37.4 Ϯ 6.6 nM). However, we could not verify precisely that dose-response curves obtained with JMV 180 (Fig. 10B) and (Met)-CCK were monophasic due to limited concentration usable in the assays.
As mentioned before, JMV 180, at the difference of CCK, is unable to induce acute pancreatitis at high doses in rats (19). This was tested on CCK1R Ϫ/Ϫ mice pancreatic acini expressing the wild-type or the M121V-CCK1R. As illustrated on Fig. 11, CCK at a high concentration induced membrane and cellular damage typical of acute pancreatitis (Fig. 11B). By contrast, JMV 180, and (Met)-CCK did not alter morphology of pancreatic acini expressing the WT-CCK1R and M121V-CCK1R, respectively, indicating that the two complexes behave identically (Fig. 11, C and D). DISCUSSION The aim of the current work was to investigate the mechanism that, at the ligand binding site of a GPCR specimen of family I, the human CCK1R, is responsible for partial activation of phospholipase C, taken as an enzyme effector of this receptor.
Using both an in silico approach of molecular modeling, and pharmacological and biological analysis of CCK1R mutants, we provide evidence that the binding sites for the partial agonist, JMV 180, and for the full agonist, CCK, are overlapping. However, the two ligands have their C-terminal aromatic rings differently positioned. Indeed, all results supported that the Phe aromatic ring of CCK interacts with aromatic ring of Phe 330 of helix VI (stretched position). In contrast, in the modeled JMV 180⅐CCK1R complex, the phenyl ester moiety of JMV 180 appeared either in a stretched position toward Phe 330 , or in a folded position within a hydrophobic pocket surrounding Met 121 of helix III. Experimental results showing absence of hindrance effect of Ala 329 3 Phe mutation on JMV 180 binding, indicated that, in the majority of JMV180⅐CCK1R complexes, JMV 180 is most likely in a folded conformation. Based on these results, we hypothesized that JMV 180 is a partial CCK agonist, because it does not show a preference for a stretched positioning of its C-terminal phenyl ethyl ester. According to binding data demonstrating a single class of affinity sites for JMV 180 on the CCK1R, the stretched and folded positions appears to generate complexes of equal affinity. Alternatively, the two complexes could be undiscernible by conventional binding experiments, due to the fact that complexes with JMV 180 in a folded position may represent the majority of total JMV 180⅐CCK1R complexes. The demonstration that JMV 172 stimulated inositol phosphate production more efficiently than JMV 180 despite its very low affinity for the CCK1R, provided a strong experimental proof that correct orientation of the Cterminal aromatic ring toward Phe 330 is critical for agonist efficacy. In line with this last result, Phe 330 on helix VI appeared to be essential for phospholipase C activation, especially by partial agonists. Results with JMV 172 also indicate that the CCK1R is still active when occupied by a peptide having at the same time one phenyl moiety in a folded position and the second one in a stretched conformation. Accordingly, the folded conformation of ligands, which is unable to activate the CCK1R, does not prevent its activation. Our current findings on the molecular mechanism whereby JMV 180 partially activates CCK1R were compared with previous findings with M121V-CCK1R mutant (26). Indeed, the position of the C-terminal aromatic ring of (Nle)-CCK was shown to adopt a folded position while bound to M121V-CCK1R, and the resulting complex was unable to activate phospholipase C (26). On the other hand, (Met)-CCK could adopt a stretched position while bound to M121V-CCK1R (as observed in the fully active CCK⅐CCK1R complex) and partially activated phospholipase C. However, this modeling result was unable to explain partial agonism. Automated docking of (Met)-CCK to M121V-CCK1R performed in the present study, yielded two conformations for (Met-CCK), thus suggesting again that partial stimulation of phospholipase C is due to lack of preference for a stretched conformation of the ligand. We further investigated mechanisms underlying partial agonism of JMV 180 and that whereby (Met)-CCK partially stimulated M121V-CCK1R mutant by comparing pharmacologically and functionally the (Met)-CCK⅐M121V-CCK1R complex with that of JMV 180⅐CCK1R complex for which much data exist in the literature. Strikingly, we found that the two types of complexes presented rigorously the same features such as total dependence of inositol phosphate production to G␣ q expression, single affinity of binding sites, insensitivity of binding to non-hydrolyzable GTP, absence of GTP␥[S 35 ] binding following agonist stimulation, similarity of dose-response curves for amylase secretion from pancreatic acini, and incapacity to induce acute pancreatitis in pancreatic acini. Such experimental results obtained in different cell types do not provide a demonstration that the same molecular mechanism is the origin of partial activation of the two complexes. Nevertheless, they indicate that the proposed mechanism at the binding site level is highly probable and involves G␣ q .
To explain the distinct locations of the ligand C-terminal aromatic ring within the bottom of the CCK1R binding site, one must take into account the relative importance of regions and amino acids, which compose the binding site of the CCK1R in relation to the structure of the ligands. In fact, the binding site of CCK on the CCK1R is composed of amino acids of the first and second extracellular loops (Phe 107 , Arg 197 , and Met 195 ) and of amino acids of the upper third of transmembrane helices that form a pocket surrounding the C-terminal tripeptide of CCK. This binding pocket clearly involves 5 amino acids of helix VI (Arg 336 , Asn 333 , Trp 326 , Ile 329 , and Phe 330 ), 3 amino acids of helix VII (Ile 352 , Leu 356 , and Phe 360 ), 2 amino acids of helix III (Val 125 and Met 121 ), 1 amino acid of helix V (Phe 218 ), and amino acids Leu 51 and Leu 53 (helix I), and Cys 94 (helix II). In this binding pocket, the region that energetically contributes the most to anchoring of CCK is helix VI. In particular, two hydrophilic residues, Arg 336 and Asn 333 , form strong bonds with the penultimate Asp and the C-terminal amide of CCK, respectively (29). Each of these two contact points accounts for 1,000-to 10,000-fold in association constant of CCK (29). Unlike Arg 336 and Asn 333 , Phe 330 weakly contributes to CCK affinity but instead plays a role in the occurrence of different affinity states of the CCK1R and behaves as an activation switch (Ref. 26 and this study). In the JMV 180⅐CCK1R-mod- eled complex, at the difference of CCK⅐CCK1R complex, the C-terminal end of JMV 180, which lacks the amide function, was not anchored to the Asn 333 side chain. Furthermore, the phenyl ethyl ester of JMV 180 is more flexible than the aromatic ring of CCK Phe. These structural differences, together with the shape of the binding site consisting in two communicating hydrophobic pockets, allow C-terminal phenyl moiety of JMV 180 to move laterally from a stretched position to a folded position. In the case of (Met)-CCK⅐M121V-CCK1R complex, (Met)-CCK, at the difference of JMV 180, is bound through its C-terminal amide to Asn 333 side chain. However, our data showed that the presence of Val residue at position 121 on helix III leads the aromatic ring of C-terminal Phe of CCK to adopt either a folded or a stretched conformation. Based on these modeling data, Met 121 appears as a key residue important for CCK1R activation, as previously stated, most likely because it is a critical regulator of hydrophobic interactions, which can take place within the deepest part of CCK1R binding site. The fact that (Nle)-CCK did not activate M121V-CCK1R, whereas (Met)-CCK partially activated this mutant was previously interpreted as an indication that amino acid at position 121 on helix III and amino acid Met/Nle of CCK are so close that they are interchangeable to regulate physico-chemical properties of the binding cavity (26). Thus, helices III and VI of the CCK1R can be considered as functionally linked through the ligand binding site. It is worthy to note that in the three-dimensional model, Phe 330 , which is T-shape linked with the C-terminal phenyl of CCK ligands in the active liganded CCK1R, also formsinteractions with Phe 218 of helix V. In a previous study, exchange of Phe 218 for an Ala was shown to be important for the occurrence of different affinity states of the CCK1R and for CCK1R expression at the cell surface (26). Mutation of Phe 218 also dramatically decreased efficacy of CCK1R responses to CCK, suggesting that Phe 218 is another key residue for CCK1R activation. Based on these different data, we suggest that CCK1R activation involves interactions between Cterminal phenyl moiety of agonists and Phe 330 of the CCK1R, which modify the relative positioning of helices VI and V both linked to the third intracellular loop of the receptor.
Data from the present study describing a critical role of transmembrane helix VI in the activation process of the CCK1R agree with conclusions reached with several other GPCRs indicating that movement of this helix is part of the activation process of GPCRs in general. Pioneering studies using electron paramagnetic resonance analysis of spin-labeled rhodopsin have revealed proximity of helices III and VI and their body movement upon light activation. Furthermore, disulfide cross-linking of these helices prevented activation of the rhodopsin-coupled G protein, transducin (39). In rhodopsin too, engineered metal-ion binding sites between transmembrane helices to restrain activation-induced conformational change in specific locations confirmed that helices III and VI are in close proximity and moved relative to one another to activate transducin (40). Using a cysteine-reactive fluorophore incorporated in mutants of a G s -coupled GPCR, the ␤ 2 -adrenoreceptor, conformational changes involving helices III and VI were detected upon agonist activation (41). Results from experiments of metal binding sites bridging of helices III and VI applied to two GPCRs with distant amino acid sequences, the ␤ 2 -adrenoreceptor and the parathyroid hormone receptor, highlighted the evolutionary conservation of the activation switch mechanism in GPCRs (42). A set of data suggests that not only helices III and VI move following agonist activation. For instance, in the C5a receptor, motion of helix VII has been proposed to explain activation switch with mutants (43).
The interesting question of whether CCK1R conformational state induced upon JMV 180 binding is the same as or is distinct from that induced upon CCK binding could not be answered in the current study. Nevertheless, at this stage of our investigations, the achieved conclusion that JMV 180 is a partial agonist on phospholipase C activation, because this peptide shows a poor preference for a stretched positioning of its C-terminal phenyl moiety toward Phe 330 of helix VI, whereas the full agonist, CCK, only exhibits a stretched conformation, is in favor of a single active conformational state induced by both JMV 180 and CCK. Accordingly, partial agonism of JMV 180 at CCK1R can be explained by the "two states model for GPCR activation" and by the low intrinsic capability of the peptide to stabilize the active state of the CCK1R. This view is not inconsistent with pharmacological and biological observations suggesting that CCK could induce different conformational states, which each couple to different G protein subunits, whereas JMV 180 could not (15). However, available modeling techniques still remain unable to discriminate between such conformational states. Direct evidence of graded conformational changes induced by ligands of distinct efficacies were only obtained with the ␤ 2 -adrenergic receptor in which a fluorescent probe was incorporated at the bottom of helix VI. In this GPCR, agonists of the ␤ 2 -adrenoreceptor having different intrinsic efficacy, such as full and partial agonists, were shown to induce and/or stabilize distinct conformational states of the receptor (44 -46).
To conclude, in this study using both experimental and molecular modeling approaches, we demonstrate that positioning of the C-terminal end of peptidic agonists within the binding site of the CCK1R is responsible for the extent of phospholipase C activation through G␣ q coupling. Our findings may have general impact toward the understanding of the functioning of other peptide GPCRs. Moreover, given the potential therapeutic interest in partial agonists such as JMV 180, our structural data will serve for screening and for rational design of new CCK1R ligands.