Distinct CCK-2 receptor conformations associated with β-arrestin-2 recruitment or phospholipase-C activation revealed by a biased antagonist

Article abstract of DOI:10.1021/ja308784w

Seven-transmembrane receptors (7TMRs), also termed G protein-coupled receptors (GPCRs), form the largest class of cell surface membrane receptors, involving several hundred members in the human genome. Nearly 30% of marketed pharmacological agents target

Full text of DOI:10.1021/ja308784w

Article
pubs.acs.org/JACS
Distinct CCK‑2 Receptor Conformations Associated with β‑Arrestin‑2
Recruitment or Phospholipase‑C Activation Revealed by a Biased
Antagonist
Remi Magnan,† Chantal Escrieut,† Veronique Gigoux,† Kavita De,† Pascal Clerc,† Fan Niu,†
́
́
Joelle Azema,‡,⊥ Bernard Masri,§,⊥ Arnau Cordomi,† Michel Baltas,‡,⊥ Irina G. Tikhonova,*,∥
and Daniel Fourmy*,†
†EA 4552, Universite
́
de Toulouse 3, 31432 Toulouse, France
‡CNRS, LSPCMIB, UMR-5068, Toulouse, France
§INSERM U1037, Cancer Research Center of Toulouse, Toulouse, France
⊥Universite
́
de Toulouse 3, Toulouse, France
∥School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, United Kingdom
S
* Supporting Information
ABSTRACT: Seven-transmembrane receptors (7TMRs), also termed G
protein-coupled receptors (GPCRs), form the largest class of cell surface
membrane receptors, involving several hundred members in the human
genome. Nearly 30% of marketed pharmacological agents target 7TMRs.
7TMRs adopt multiple conformations upon agonist binding. Biased
agonists, in contrast to non-biased agonists, are believed to stabilize con-
formations preferentially activating either G-protein- or β-arrestin-dependent
signaling pathways. However, proof that cognate conformations of receptors
display structural differences within their binding site where biased agonism
initiates, are still lacking. Here, we show that a non-biased agonist, chole-
cystokinin (CCK) induces conformational states of the CCK2R activating
Gq-protein-dependent pathway (CCK2RG) or recruiting β-arrestin2 (CCK2Rβ) that are pharmacologically and structurally distinct.
Two structurally unrelated antagonists competitively inhibited both pathways. A third ligand (GV150013X) acted as a high affinity
competitive antagonist on CCK2RG but was nearly inefficient as inhibitor of CCK2Rβ. Several structural elements on both
GV150013X and in CCK2R binding cavity, which hinder binding of GV150013X only to the CCK2Rβ were identified. At last,
proximity between two conserved amino acids from transmembrane helices 3 and 7 interacting through sulfur−aromatic interaction
was shown to be crucial for selective stabilization of the CCK2Rβ state. These data establish structural evidence for distinct
conformations of a 7TMR associated with β-arrestin-2 recruitment or G-protein coupling and validate relevance of the design of
biased ligands able to selectively target each functional conformation of 7TMRs.
a central role in orchestrating 7TMR trafficking since β-arrestin-
bound 7TMRs are rapidly targeted to the clathrin-coated pits,
thereby promoting internalization of these receptors.6 Importantly,
therapeutic agents have been shown to activate dif-
ferentially G protein-dependent or β-arrestin-dependent signaling
pathways.7 Such ligands, triggering biased signaling, were named
biased ligands or functionally selective ligands.7
Understanding the mechanism of biased signaling can
provide more specific and efficient new drugs. It is generally
assumed that the phenomenon of biased signaling is an intrinsic
property of a ligand−7TMR complex, whereby a 7TMR exists
in several conformations, each of which being preferentially
stabilized and activated by selective ligands.8 However, despite
the recent release of several X-ray structures of agonist−7TMR
INTRODUCTION
■
Seven-transmembrane receptors (7TMRs), also termed G
protein-coupled receptors (GPCRs), form the largest class of
cell surface membrane receptors, involving several hundred
members in the human genome. Near 30% of marketed phar-
macological agents target 7TMRs.1 Biological effects triggered by
7TMRs result from the activation of both G protein-dependent
and G protein-independent intracellular signaling pathways.2
The G protein-independent activation pathway often
involves the recruitment of β-arrestins.3,4 In general, β-arrestin
recruitment is preceded by receptor phosphorylation by 7TMRs
kinases (G-protein coupled receptor kinases, abbreviated GRK)
that selectively phosphorylate serine/threonine residues of
agonist-activated 7TMRs.5 7TMR-bound arrestins have been
implicated in the activation of a number of signaling proteins
such as the MAP kinase cascade (c-RAf-1, ERK1/2, JNK3) or
non-receptor tyrosine kinases (c-Src, Yes).4 β-arrestins also play
Received: September 4, 2012
Published: January 16, 2013
© 2013 American Chemical Society
2560
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
complexes9−12 and biophysical and biochemical studies13−16
showing that different biased ligands induce or stabilize distinct
conformations, experimental evidence directly linking distinct
conformations of 7TMR to recruitment of β-arrestins or coupl-
ing to G proteins is still lacking. Furthermore, the mechanisms
whereby biased agonism initiates in the binding site of 7TMR
remain unknown.
after stimulation by Alexa F 647 alone or in the presence of the
antagonist GV150013X (not illustrated).
Receptor Binding Assay with 125I-CCK. Cells grown overnight
onto 10-cm culture dishes were transfected with 1 μg/plate (except
when mentioned) of vector containing the cDNA for the wild-type or
mutated CCK2R-Rluc using polyethylenimine. Twenty-four hours
later, cells were transferred to 24-well plates. Approximately 24 h later,
binding assays were performed using 125I-CCK according to the
protocol previously described in detail.22 IC50 values for competitors
(concentration inhibiting half of specific binding) were calculated
using the nonlinear curve-fitting software GraphPad Prism (San Diego,
CA). Mutation factors (Fmut) were calculated as IC50(mutated
CCK2R-Rluc)/IC50(wild-type CCK2R-Rluc).
In this context, we focused our study on the cholecystokinin-2
receptor, a member of family A 7TMRs, mediating central and
peripheral effects of two important structurally related neuro-
peptides, cholecystokinin (CCK) and gastrin.17 We examined
whether conformational states responsible for Gαq protein-
dependent activation of phospholipase-C or recruitment of
β-arrestins are distinguishable. For this purpose, we first com-
pared pharmacologically the CCK-induced conformational
state of the CCK2R recruiting β-arrestin-2 with that activating
phospholipase-C, using structurally related and unrelated non-
peptide ligands of the receptor. We discovered a non-peptide
ligand (GV150013X), acting as a competitive antagonist on
CCK2R inducing phospholipase-C activation but inefficient at
inhibiting CCK2R inducing β-arrestin-2 recruitment to the
CCK2R. We further identified structural elements on both
GV150013X and the CCK2R, which hinder competitive binding
of GV150013X to the orthosteric binding site of the CCK2R
state recruiting β-arrestin-2. At last, we identified two interacting
amino acids from helices 3 and 7 (M3.32 and Y7.43) as key struc-
tural elements that govern stabilization of the CCK2R state
recruiting β-arrestin-2. All together our data indicate that the
non-biased agonist, CCK, induces distinct conformations of the
CCK2R that activate phospholipase-C or recruit β-arrestin-2.
Receptor Binding Assay by Flow Cytometry. Flp-InCCK2R-
293 cells were incubated with Alexa F 647-CCK (0.1 μM) alone or in
the presence of competitors in DMEM/HEPES (20 mM) for 1 h at
37 °C in poly-L-lysine-coated 6-wells plates. Cell-associated fluorescence
was determined using a BD FACSCalibur flow cytometer, with Flp-In
293 that does not express CCK2R as a negative control.
Inositol Phosphate Production Assay. Cells were prepared as
for binding experiments and grown overnight in the presence of myo-
[2-3H]inositol (2 μCi/mL, specific activity 10−25 Ci/mmol, Perkin-
Elmer Life Sciences). Production of total inositol phosphates was
determined as described previously.22
Confocal Fluorescence Microscopy. Cells transiently trans-
fected with appropriate plasmids were incubated with relevant
agonists/antagonists and further observed by confocal fluorescence
microscopy as described previously23 using a Zeiss laser scanning
microscope (LSM-510). For quantification, images of YFP and Alexa F
647 were collected using excitation lasers (488 and 633 nm) and
double detection (windows: 504−633 nm and >650 nm) at the
indicated times. The amount of β-arrestin-2-YFP (green) co-localizing
with Alexa F 647-CCK-labeled CCK2R (red) was measured over time
using Morpho Expert Software (Explora Nova, La Rochelle, France).
Recruitment was expressed as the ratio between confocal co-localized
β-arrestin-2-YFP/Alexa F 647-CCK-labeled CCK2R and Alexa F 647-
CCK-labeled CCK2R. On average, 10−14 individually transfected cells
were analyzed on 2 separate wells, representing a duplicate.
BRET Studies. β-Arrestin-2-YFP recruitment to ligand-occupied
wild-type or mutated CCK2R-Rluc was assessed using a bio-
luminescence resonance energy tranfer (BRET) assay in a 96-well
format, as described in detail in the Supporting Information and
previously in ref 23. For titration experiments, the acceptor/donor
ratio was calculated as described previously.24
EXPERIMENTAL PROCEDURES
■
Materials. Sulfated [Thr28,Nle31]-CCK 25−33 is referred to as
CCK. 125I-Sodium (2000 Ci/mmol) and [myo-3H]inositol (5 μCi/ml)
were from GE Healthcare (Little Chalfont, Buckinghamshire, UK).
Radio-iodinated CCK (1600−2000 Ci/mmol) is referred to as
125I-CCK. Alexa Fluor 647-labeled CCK was obtained according to
the procedure described recently for glucose insulinotropic polypep-
tide and is referred to as Alexa F 647-CCK.18 The following synthetic
ligands of the CCK2R were used: PD135158,19 L365260,20 and
GV150013X.21 The cDNAs encoding CCK2R, green fluorescent protein
(GFP)-tagged CCK2R and Renilla luciferase (Rluc)-fused CCK2R, were
generated by subcloning respectively the CCK2R cDNA in pcDNA5/
FTR (Invitrogen), pEGFP-N1 (BD Biosciences Clontech), or pRluc-
N1(h) (Perkin-Elmer). Yellow fluorescent protein (YFP)-tagged β-
arrestin-2 (termed β-arrestin-2-YFP) was a generous gift from Marc
Caron (Duke University Medical Center, Durham, NC, USA). Red
fluorescent protein (RFP)-tagged β-arrestin-2 (termed β-arrestin-2-RFP)
was kindly given by Robert Lefkowitz (Duke University Medical Center).
All mutant receptor cDNAs were constructed by oligonucleotide-directed
mutagenesis (Quik Change site-directed mutagenesis kit, Stratagene,
France) using human CCK2R cDNAs cloned in the pcDNA5/FRT
vector as template. Cells used were HEK 293 cells stably expressing the
CCK2R (Flp-In CCK2R-293, Flp-In system, Invitrogen) and HEK 293T
cells transiently expressing proteins of interest. Transfections were
performed using polyethylenimine (Polyscience). The different tagged
CCK2Rs were pharmacologically characterized before use in the study.
Binding experiments using 125I-CCK indicated that wild-type CCK2R,
GFP/YFP-tagged CCK2R, and Rluc-tagged CCK2R bound CCK with
dissociation constant values (Kd) of 1.2 0.2, 1.1 0.2, and 0.9 0.2
nM, respectively, and were expressed at similar levels. Inositol
phosphate assays showed that they responded to CCK with similar
potencies (EC50 = 2.8 0.1, 2.2 0.2, and 1.4 0.4 nM, respectively)
and exactly the same efficacy. Moreover, confocal microscopy analyses
showed that Rluc-tagged CCK2R and wild-type CCK2R exhibited
non-significantly different abilities to recruit YFP-tagged β-arrestin2
Structural Modeling. Homology modeling of the CCK2R was
conducted using the Prime module of the Schrodinger software
̈
(Prime 2.0, Schrodinger, LLC, New York, NY, USA) with the default
̈
settings. To mimic the experimental conditions, and the fact that the
CCK2RG state converted to the CCK2Rβ state upon CCK binding, we
built the CCK2RG state using the crystal structure of the nanobody-
stabilized active form of the β2 adrenergic receptor (PDB code: 3P0G)
available at the time of our modeling study.25 CCK was docked into
this modeled CCK2RG on the basis of our previous mutagenesis
data,22,26−31 enabling refinement of the model. Docking of non-
peptide ligands was performed using the Glide 5.6 module of the
Schrodinger software and guided using constraints derived from results
̈
of the binding study in which we measured the effect of mutations
within the binding cavity of the CCK2R on the affinity of the ligands
(Table 1). The docking poses were reoptimized on the basis of the
obtained binding data (Table 1) using the Induced-Fit module.32 All
ligands were preprocessed with LigPrep at pH 7. The standard
precision scoring function was used in Glide. To examine the dynamics
of the complexes, we used the Dynamics module of MacroModel
(Schrodinger, LLC). Electrostatics was set to distance-dependent
̈
function and dielectric constant of 4. The backbone hydrogen bonds
of transmembrane helices were constrained to preserve the secondary
structure. The biosystems were minimized within 500 steps and
subjected to 20 ns molecular dynamics with a default setting.
OPLS2005 force field was used for all calculations. Maestro
2561
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Table 1. Effect on Mutations within the Orthosteric Binding Site of the CCK2R on Affinity of Antagonists L365260,
a
GV150013X, and GV-CH3
antagonists
GV150013X
IC50 SEM (nM)
L365260
IC50 SEM (nM)
GV-CH3
IC50 SEM (nM)
Fmut
Fmut
Fmut
WT-CCK2R
3.75 0.14
1.52 0.39***
27.7 6.6*
26.6 2.6**
1.76 0.59**
15.7 2.4***
3.7 0.2
1.0
0.4
7.3
7.0
0.5
4.2
1.0
1.1
4.1
3.8
2.8
3.1
8.7 0.4
52.8 9.5**
91.8 45.0*
5.2 0.7
1.0
6.1
115 25
239 49*
ND
1.0
2.1
T111A2.61-CCK2R
M134A3.32-CCK2R
Y189A4.60-CCK2R
L222A5.42-CCK2R
L223A5.43-CCK2R
F227A5.49-CCK2R
W346A6.48-CCK2R
N353A6.55-CCK2R
R356A6.58-CCK2R
H376A7.39-CCK2R
Y380A7.43-CCK2R
10.6*
0.6
ND
3.8
437 75*
ND
4.3 1.0**
28.7 7.4**
22.5 1.8***
6.9 1.2
0.5
ND
ND
ND
ND
ND
6.3
3.3
ND
2.6
ND
4.0 0.8
0.8
ND
15.4 2.5**
14.2 2.9**
10.5 0.3***
14.4 1.5***
45.9 14.4**
4.2 0.8
5.3**
0.5
ND
721 95**
ND
7.8 0.9
0.9
ND
ND
25.1 8.6**
2.9
ND
a
Binding experiments were performed as indicated in the Experimental Procedures section on HEK 293T cells transiently expressing wild-type or
mutated CCK2R-Rluc. Results are given as IC50 (concentration of antagonists inhibiting 50% of 125I-CCK specific binding) and mutation factors
Fmut, which were calculated as IC50(mutated CCK2R-Rluc)/IC50(wild-type CCK2R-Rluc). IC50 values are means SEM from three or four separate
determinations. Asterisks indicate p values giving significance of IC50 shifts for mutants relative to wild-type CCK2R: * 0.01 < p < 0.05; ** 0.001 <
p < 0.01; *** p < 0.001. ND = affinity was not determined.
Schrodinger utilities (Schrodinger, LLC) and VMD scripts33 were
used for the trajectories analyses.
̈
̈
IC50 = 4.06, 9.19, and 3.60 nM, respectively (Supplementary
Fig. 1). On the other hand, β-arrestin-2 recruitment to the
CCK2R was dose-dependently inhibited by PD135158 and
L365260 but not by GV150013X, suggesting that GV150013X,
despite its structural similarity with L365260, is unable to
compete with CCK for the CCK2Rβ state (Figure 2e−h).
We further ensured that L365260 and GV150013X, which
are structurally related compounds, and PD135158, a
pseudopeptidic ligand (Figure 1c), bind to the CCK2R
orthosteric binding site in a competitive manner by analyzing
the data using a Schild plot. As illustrated on Figure 3a, CCK
dose−response curves of inositol phosphate production were
right shifted in the presence of increasing concentration of
PD135158, L365260, or GV150013X and reached the same
maximal response. This indicates that all three non-peptide
ligands inhibited phospholipase-C activation competitively.
CCK dose−response curves of BRET were similarly right shifted
in the presence of increasing concentration of PD135158 or
L365260 and not in the presence of GV150013X (Figure 3b).
Thus, unlike GV150013X, both L365260 and PD135158 inhibit
β-arrestin-2 recruitment through binding to the orthosteric
binding site of the CCK2Rβ state. Comparison of dose−response
curves indicates that the potency of CCK (EC50) in BRET
experiments (10.6 1.9 nM) was lower than that of CCK alone
in inositol phosphate assays (1.4 0.37 nM). This suggested
possible switch of pathways from stimulation of phospholipase-C
through high affinity CCK binding sites to β-arrestin2 recruit-
ment through low affinity binding sites.
Statistical Analyses. Unpaired t test (Prism GraphPad software)
was used. Values for CCK2R mutants (binding, inositol phosphate
production, β-arrestin2 recruitment) were compared to that of the
wild-type CCK2R. In figures and tables, degree of confidence is
indicated by asterisks as follows: * 0.01 < p < 0.05; ** 0.001 < p <
0.01; *** p < 0.001.
RESULTS
■
The CCK2Rβ State Recruiting β-Arrestin-2 Is Pharma-
cologically Distinct from the CCK2RG State Activating
Phospholipase-C. Binding of cholecystokinin (CCK) to the
CCK2R (Figure 1a) activates a Gαq signaling pathway, namely
phospholipase-C activation,34 which we monitored using the
inositol phosphate assay in HEK 293 cells (Figure 1b). The con-
formational state of the CCK2R responsible for Gαq-dependent
stimulation of phospholipase-C upon CCK stimulation was
named CCK2RG (Figure 1b). Moreover, under CCK stimula-
tion, β-arrestin-2 is rapidly recruited to the CCK2R. This recruit-
ment was followed by observing translocation of cytoplasmic
β-arrestin-2-GFP to activated CCK2R with fluorescent CCK, and
by measuring specific increase of BRET signal between CCK2R-
Rluc and β-arrestin-2-YFP.23 The state of the CCK2R recruit-
ing β-arrestin-2 upon CCK stimulation was named CCK2Rβ
(Figure 1b).
To initiate molecular and structural comparison of the CCK-
induced CCK2Rβ and CCK2RG states, we investigated whether
the binding cavity of the two functional states, involving
residues from transmembrane helices (Figure 1a and ref 22)
could show differences. Thus, the ability of three standard
molecules (PD135158, L365260, and GV150013X) (Figure 1c)
to inhibit CCK-stimulated production of inositol phosphates or
β-arrestin-2 recruitment was assessed in HEK 293 cells. As
shown in Figure 2a−d, CCK-stimulated inositol phosphate
production was dose-dependently inhibited by PD135158,
L365260, and GV150013X. These results are in agreement with
binding data indicating that 125I-CCK binding to HEK cells
expressing the CCK2R was dose-dependently and fully
inhibited by PD135158, L365260, and GV150013X with
Since CCK2R-induced β-arrestin-2 recruitment is a molec-
ular event preceding CCK2R internalization and trafficking,23
we then investigated the effects of non-peptidic ligands on
CCK-stimulated internalization of the CCK2R. GFP-tagged
CCK2R was located at the cell membrane and was internalized
under CCK stimulation (Figure 4a,b). PD135158 or L365260
fully inhibited CCK-induced internalization of GFP-tagged
CCK2R (Figure 4c,d). Remarkably, GV150013X, which did not
stimulate internalization by itself, only weakly inhibited in-
ternalization of GFP-tagged CCK2R (Figure 4f,g). We next in-
vestigated whether the absence of inhibitory effect of GV150013X
2562
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 1. Representation of CCK2R and assays used to study states signaling through Gαq coupling or recruiting β-arrestin-2. (a) Serpentine
representation of the human CCK2R. Amino acids previously identified as participating in the orthosteric binding site28 and/or activation site29,34,66
(black full circles), as well as the endocytosis motif in the C-terminal region23 (black letters in red circles), are depicted and double-numbered (main
number, regular numbering for proteins; superscript numbers, Ballesteros−Weinstein numbering). (b) Diagrams illustrating the principle of assay for
the CCK2RG state signaling through Gαq coupling or the CCK2Rβ state recruiting β-arrestin-2 by measurement of inositol phosphate production or
BRET, respectively. (c) Chemical structures of CCK2R non-peptide ligands used in the study.
on CCK2Rβ state (whereas this compound fully competed with
be due to the presence of a population of CCK2R binding sites
125I-CCK in binding experiments, Supplementary Fig. 1) could
recognizing CCK, albeit with a low affinity, but not recognizing
2563
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 2. Effect of antagonists showing inability of GV150013X to inhibit CCK-induced β-arrestin-2 recruitment to the CCK2R. (a−d) Inositol
phosphate (Ins-P) accumulation upon stimulation for 60 min with 10 nM CCK alone or in the presence of increasing concentrations of PD135158,
L365260, or GV150013X. Experiments were carried out with Flp-InTM CCK2R-293 cells expressing the wild-type CCK2R. Results are expressed as
the percent of inositol phosphate production achieved with 10 nM CCK alone. (e−h) BRET between β-arrestin-2 and CCK2R upon stimulation for
300 s with 100 nM CCK in the absence or in the presence of increasing concentrations of PD135158, L365260, or GV150013X added at the same
time, before BRET measurements. Experiments were carried out with HEK 293T cells transiently expressing the wild-type CCK2R-Rluc and
β-arrestin-2-YFP. Results are expressed as net BRET as described in the Experimental Procesures. Data are the mean SEM of three or four
independent experiments, each determination being performed in triplicate.
GV150013X. For this purpose, we performed confocal micro-
copy observations of HEK cells incubated with Alexa F 647-
CCK in the presence of GV150013X at a saturating concentra-
tion. As shown on Figure 4g, fluorescent CCK still abundantly
internalized in the presence of GV150013X. We next per-
formed binding experiments using Alexa F 647-CCK in order
to probe such low affinity binding sites. Interestingly, these
binding experiments enabled identification of a fraction of CCK
binding sites (>30% of total binding sites) which were not
occupied by GV150013X even at 10 μM concentration (Figure 5),
thus suggesting a relationship between the fraction of labeled
CCK2R in the presence of GV150013X and internalization
(Figure 4g). Hence, CCK2R binding sites presenting a low affinity
for CCK and unable to bind GV150013X are identified in HEK
cells. These CCK2R binding sites likely correspond to CCK2Rβ
state recruiting β-arrestin2.
This first set of pharmacological results allows us to propose
the view that CCK-stimulated formation of the CCK2RG state
stimulating phospholipase-C is prevented in a competitive
fashion by the non-peptide ligands PD135158, L365260, or
GV150013X. The CCK2RG state is converted to the CCK2Rβ
state recruiting β-arrestin-2 upon CCK binding, probably GRK
action, and subsequently undergoes internalization. The forma-
tion of CCK2Rβ-β-arrestin2 complex upon CCK stimulation is
inhibited by PD135158 or L365260 and not by GV150013X.
This striking result led us to investigate the molecular basis for
the absence of inhibitory effect of GV150013X on CCK-induced
formation of CCK2Rβ state recruiting β-arrestin-2.
Structural Basis for Distinct Behavior of Benzodiaze-
pine-like Antagonists on the CCK2Rβ State Recruiting
β-Arrestin-2 or the CCK2RG State Activating Phospho-
lipase-C. At this point, we focused our study on the two
benzodiazepine-like ligands, GV150013X, with a bulky adaman-
tane group, and L365260, with a methyl group at the similar
position, as shown in Figure 1c. Comparison of the two struc-
tures led us to hypothesize that the adamantane ring may hinder
competitive binding of GV150013X to the CCK2Rβ state and
not to the CCK2RG state. Following this hypothesis, we postu-
lated that exchange of amino acids in contact with the adaman-
tane group within the orthostric binding site of the CCK2R by a
residue having a smaller side chain or chemical substitution of
adamantane by a smaller group would restore ligand binding at
the CCK2Rβ state and therefore, inhibition of CCK-stimulated
recruitment of β-arrestin-2.
For this purpose, the CCK2RG state of CCK2R was homo-
logy modeled using the crystal structure of the nanobody-
stabilized active form of the β2 adrenergic receptor.25 The model
was further refined on the basis of the mutagenesis data used to
2564
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 3. Analysis of antagonist effects using schild plot shows that GV150013X interacts competitively with CCK2RG state stimulating
phospholipase-C but could not competitively interact with the CCK2Rβ state recruiting β-arrestin-2. (a) Effect of antagonists on CCK-induced
inositol phosphate production. Flp-InTM CCK2R-293 cells expressing the wild-type CCK2R were stimulated for 60 min with increasing
concentrations of CCK alone (indicates on x axis) or in combination with PD135158, L365260, or GV150013X at indicated concentrations. Results
are expressed as the percent of inositol phosphate production achieved with CCK alone and are the mean SEM of four independent experiments,
each determination being performed in triplicate. pA2 = 7.82 0.23, 7.46 0.07, and 7.38 0.16 were calculated using GraphPad Prism software
for PD135158, L365260, and GV150013X, respectively. (b) Effect of antagonists on BRET between β-arrestin-2 and CCK2R. HEK 293T cells
transiently expressing wild-type CCK2R-Rluc and β-arrestin-2-YFP were stimulated for 300 s with increasing concentrations of CCK alone
(indicated on x axis) or in combination with PD135158, L365260, or GV150013X at indicated concentrations before BRET measurements. Results
are expressed as net BRET as previously described.23 Data are the mean SEM of 4−6 independent experiments, each determination being
performed in triplicate. pA2 = 7.79 0.08 and 8.15 0.04 were calculated using GraphPad Prism software for PD135158 and L365260, respectively.
validate the key residues interacting with CCK in our previous
works22,28,29,31 (Supplementary Fig. 2). We then docked
GV150013X and L365260 to the refined model of the
CCK2RG state. Docking experiments predicted most probable
binding poses of GV150013X and L365260 that are likely
distinct (Figure 6). More precisely, molecular docking sug-
gested that GV150013X binds between helices 3−6 and is
surrounded by Y1894.60, N3536.55, and R3566.58 likely through
binding affinity of both ligands for the CCK2R. Thus, results
from site-directed mutagenesis and molecular docking agree
with the view of a distinct ligand positioning within the binding
pocket of the CCK2RG, though, there is significant overlapping
(Supplementary Fig. 3).
From the proposed GV150013X binding pose, we hypo-
thesized that amino acids M1864.57, L2225.42, L2235.43, N3536.55
,
and R3566.58, which were close to the adamantane group of
GV150013X in the CCK2RG binding site, could hinder binding
of the antagonist to the CCK2Rβ state recruiting β-arrestin-2.
To test this hypothesis we mutated these residues to alanine
and measured β-arrestin-2 recruitment by BRET. Remarkably,
GV150013X dose-dependently inhibited CCK-stimulated
hydrogen bonds and by M1343.32, M1864.57, L2235.43, W3466.48
,
H3767.39, and Y3807.43 via hydrophobic contacts. On the other
hand, docking suggested that L365260 binds between helices
2−3 and 6−7, establishing hydrogen bonds with T1112.61
,
Y3807.43, and N3536.55 and hydrophobic interactions with
M1343.32, L2235.43, F2275.47, and Y3807.43
.
β-arrestin-2 recruitment to mutants M186A4.57 and L222A5.42
,
Since the docking experiments predicted distinct poses of
L365260 and GV150013X into the orthosteric binding site of
the CCK2R, we experimentally verified whether the mutation
of amino acids of the binding cavity could support (or refute)
the docking data. Binding results with CCK2R mutants (Table 1)
indicated that alanine mutations of residues Y1894.60, R3566.58, or
H3767.39 stronger affected the affinity of GV150013X than that of
L365260. In contrast, alanine mutation of residues T1112.61 or
F2275.47 stronger affected the binding affinity of L365260 than that
whereas it was almost ineffective in inhibition of β-arrestin-2
recruitment to the wild-type CCK2R (Figure 7a and Table 2).
On the other hand, GV150013X dose-dependently inhibited
CCK-induced inositol phosphate production with an identical
high potency on mutants M186A4.57 and L222A5.42 and on wild-
type CCK2R (Figure 7b and Table 2). Control experiments
indicate that mutants M186A4.57 and L222A5.42 recruit β-arrestin-2
or stimulated inositol phosphate in response to CCK stimulation
with the same efficacy as the wild-type CCK2R indicating that the
two mutants were fully converted to either CCK2Rβ or CCK2RG
states following CCK stimulation (Supplementary Fig. 4).
of GV150013X. At last, mutation of residues M1343.32, L2235.43
N3536.55, or Y3807.43 to alanine almost equally affected the
,
2565
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 4. Absence of competitive inhibition of CCK-stimulated internalization of CCK2R by GV150013X. Confocal microscopy images of HEK
293T cells transfected with CCK2R-GFP: (a) Cells were not stimulated with CCK and cells were incubated at 37 °C with CCK (100 nM) alone for
30 min (b) or with CCK (100 nM) plus PD135158 (c) or L365260 (d). In (e) cells were incubated for 30 min with GV150013X alone (1 μM), and
in (f) with CCK (100 nM) plus GV150013X (10 μM). In (g) cells were incubated for 30 min with Alexa F 647-CCK (0.1 μM) in the presence of
GV150013X (10 μM). Antagonists were added at the same time as CCK. Images are representative of at least 10 separate observations (cell number,
n > 5) on two different batches of transfected cells (N = 2).
L222A5.42 or on wild-type CCK2R and this potency was about 50,
85, and 113 fold lower than that of GV150013X, respectively
(Figure 7 and Supplementary Fig. 5 and Table 2). GV150013X at
10 μM also inhibited CCK-stimulated β-arrestin-2 recruitment
to mutants N353A6.55 and R356A6.58 but not to L223A5.43
(Supplementary Fig. 6). Note that the activity of L365260 and
PD135158 on the mutants was unchanged compare to the wild-
type CCK2R (Supplementary Fig. 6).
To summarize, pharmacological data show that adamantane
substitution by methyl group in GV150013X enables the
antagonist to interact with CCK2Rβ state whereas this chemical
change dramatically decreases the potency of the antagonist
on CCK2RG state (opposite effects caused by chemical sub-
stitution). Furthermore, mutation of residues M1864.57 and L2225.42
to alanine in the CCK2R enables GV150013X to interact with
CCK2Rβ state and also increases the potency of GV-CH3 on
CCK2Rβ state whereas the mutation of these residues do not affect
significantly potency of the antagonists on CCK2RG state.
Given that pharmacological studies suggested that sub-
stitution of adamantane in GV150013X by methyl enables GV-
CH3 to bind to the CCK2Rβ state, the binding pose of GV-CH3
in the CCK2R was explored by molecular docking and binding
Figure 5. Evidence for the existence of a low-affinity CCK binding site
experiments with the selected CCK2R mutants, involving
T1112.61, M1343.32, Y1894.60 and R3566.58 (Table 1). Our data
which does not recognize GV150013X. Inhibition of binding was
carried out by incubating Alexa F 647-CCK (0.1 μM) in the presence of
suggest that the position of GV-CH3 in the CCK2R binding site
increasing concentrations of CCK or GV150013X for 60 min, as
partially overlaps that of GV150013X, in particular, the diazepine-
described in the Experimental Procedures. Specific binding was assayed
by flow cytometry, and results are expressed as the percent of maximal
dione moiety likely interacts with residues R3566.58 and Y1894.60 as
shown on Figure 6c. However, GV-CH3 seats deep in the ortho-
steric binding cavity and the 1-phenylurea moiety, similar to the
1-phenylurea moiety of L365260, interacts with helices 3 and 6
and is close to T1112.61. This pose of GV-CH3 into the binding
site may explain the low potency of GV-CH3 for the CCK2RG
state as well as its low potency as inhibitor of β-arrestin recruit-
ment on CCK2Rβ state (Table 2).
Therefore, steric hindrance between the adamantane group
of GV150013X and residues M1864.57, L2225.42 of the CCK2R
(and to less extent, N3536.55 and R3566.58) most likely explains
in part inability of GV150013X to competitively interact with
the CCK2Rβ state. Interestingly, the fact that this steric hindrance
was observed with the CCK2Rβ state recruiting β-arrestin-2 and
not with the CCK2RG state stimulating phospholipase-C suggests that
transition of the CCK2R from the CCK2RG state to the CCK2Rβ
specific binding in the absence of competitor, and are the mean SEM
of three or four experiments.
Furthermore, we synthesized the analogue of GV150013X
(abbreviated as GV-CH3) in which the adamantane group was
substituted by a methyl group for further assessment of the
adamantane role in preventing β-arrestin-2 recruitment. We
found that unlike GV150013X, GV-CH3 dose-dependently
inhibits CCK-stimulated β-arrestin-2 recruitment to the wild-type
CCK2R, although with a low potency (Figure 7c and Table 2).
Interestingly, antagonist potency of GV-CH3 was increased on
mutants M186A4.57 and L222A5.42 relative to wild-type CCK2R
(Figure 7, Supplementary Fig. 5, and Table 2). In contrast, GV-
CH3 dose-dependently inhibited CCK-induced inositol phosphate
production with an identical potency on mutants M186A4.57 and
2566
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 6. Docking poses of benzodiazepine-like antagonists in the modeled CCK2R structure. Binding sites of (a) GV150013X, (b) L365260, and
(c) GV-CH3 are shown with the backbone of CCK2R in a curved line and the side chains of the amino acids forming the binding site in stick. The
details of docking procedure are given in the Experimental Procedures. The schematic representation of the interactions within the binding site, with
highlighted hydrogen bonds, is depicted on corresponding panels d, e, and f. On these panels, residues forming hydrogen bonds are in blue circles,
residues forming the hydrophobic pocket are in yellow circles, and residues contributing to stabilization of the CCK2Rβ state are in red and green
circles. Numbering according to Ballesteros−Weinstein nomenclature is shown in superscript.
is accompanied by changes in the structure of the orthosteric binding
pocket.
Y380A7.43 mutants in the presence of saturated levels of
β-arrestin-2-YFP (Supplementary Fig. 8).
Proximity of Met1343.32 and Tyr3807.43 Is Pivotal for
Stabilization of the CCK2Rβ State Recruiting β-arrestin-2.
The experimental data above suggest that the orthosteric
binding site of the CCK2R, which is composed of amino acids
from transmembrane helices, display structural differences in
the CCK2Rβ state versus the CCK2RG state. We further in-
vestigated whether residues from transmembrane helices could
be differentially involved in the stability of CCK2Rβ and CCK2RG
states. We focused our attention on M1343.32−Y3807.43 interac-
tions. Indeed, in the 3D-homology model of the CCK2RG state,
the Cα atoms of M1343.32 and Y3807.43 are at a distance of 10 Å
and the sulfur atom of M1343.32 and the center of aromatic ring
of Y3807.43 are at a distance of 6.5 Å (Supplementary Fig. 2).
We predicted that M1343.32 can interact with Y3807.43 through a
S−HO hydrogen bond or a sulfur−aromatic interaction in the
CCK2Rβ state, and this interaction could contribute to stabilization
of this state recruiting β-arrestin-2. We tested this hypothesis by
constructing several mutants affecting positions 3.32 and 7.43 in
the CCK2R.
Remarkably, M134A3.32 and Y380A7.43 mutants responded to
CCK-induced production of inositol phosphates with the same
efficacy as the wild-type CCK2R (Figure 8h). A shift of potency
reflecting contribution of residue Y380A7.43 to CCK binding
affinity29 was, however, observed with both inositol phosphate
production and BRET dose−response curves. Control experi-
ments indicated that expression levels of the mutants were
similar to the wild-type CCK2R (Supplementary Fig. 9).
Therefore, the two amino acids, M1343.32 and Y3807.43 appear
to be much more involved in efficacy of CCK-stimulated β-arrestin-2
recruitment to the CCK2R than in efficacy of phospholipase-C
stimulation.
Additional mutants were constructed to study the interaction
between M1343.32 and Y3807.43 and its role in CCK2Rβ state
stabilization. The possibility of H-bond interaction between
M1343.32 and Y3807.43 was first explored. BRET results
(Supplementary Fig. 10), show that mutant Y380F7.43 recruited
β-arrestin-2 similarly to the wild-type CCK2R ruling out the
possibility of H-bond interaction between M1343.32 and
Y3807.43. On the other hand, mutant M134C3.32 showed high
basal BRET in the presence of β-arrestin-2-YFP relative to the
wild-type CCK2R, suggesting a constitutive recruitment of
β-arrestin-2-YFP (Figure 8g). However, with this mutant, BRET
signal could not be increased upon CCK stimulation. In fact,
confocal microscopy observation of cells expressing the mutant
M1343.32C indicated that it was not detected at the cell surface
but was present in numerous intracellular vesicles containing
β-arrestin-2-YFP (Figure 8f). As expected, this mutant did not
exhibit inositol phosphate production (Figure 8h). Thus, to
Confocal microscopy quantification of β-arrestin-2-YFP
recruitment to activated CCK2R mutants labeled with Alexa F
647-CCK (Figure 8 and Supplementary Fig. 7) showed that
M134A3.32 and Y380A7.43 mutants poorly recruited β-arrestin-2-
YFP relative to the wild-type CCK2R (85 and 55% decreases,
respectively, Figure 8e). Correspondingly, mutants M134A3.32
and Y380A7.43 exhibited dramatic decreases in the maximal net
BRET signal relative to the wild-type CCK2R (75 and 45%
decreases, respectively) (Figure 8g). These maximum BRET
decreases were confirmed by titration curves of M134A3.32 and
2567
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 7. Mutations within orthosteric binding site of CCK2R or substitution of adamantane group by methyl in GV150013X restore antagonist
activity on CCK2Rβ state. M1864.57 or L2225.42 of CCK2R, potentially in contact with the bulky adamantane group of GV150013X when docked in
the binding site of modeled CCK2RG (Figure 6a), were mutated to alanine. (a) Mutations M186A4.57 and L222A5.42 restore the ability of
GV150013X to inhibit BRET between β-arrestin-2 and CCK2R. (b) In contrast, mutations M186A4.57 and L222A5.42 have minor effects on the
potency of GV150013X on CCK-induced production of inositol phosphates. (c) Substitution of the adamantane group in GV150013X by a methyl
group enables the antagonist GV-CH3 to inhibit BRET between β-arrestin-2 and wild-type CCK2R. (d) In contrast, substitution of the adamantane
group in GV150013X by a methyl group dramatically decreases the potency of the antagonist to inhibit CCK-induced inositol phosphate production.
Experiments were carried out as for Figure 2. Values of IC50 (corresponding to potency) are reported in Table 2. Note that due to solubility limits in
the incubation medium, GV150013X could not be tested at concentrations higher than 10 μM.
Table 2. Effects of Adamantane Substitution in GV150013X and M186A4.57 or L222A5.42 Mutations in CCK2R on the Potency of
a
the Antagonist on CCK2Rβ State Recruiting β-Arrestins or CCK2RG State Activating Phospholipase-C
IC50 SEM (nM)
CCK-induced β-arrestin-2 recruitment
CCK-induced PLC activation
GV150013X
GV-CH3
GV150013X
GV-CH3
WT
NC
13500 4300
230 80
39.9 5.4
25.8 9.2
23.2 7.2
1958 440
2260 730
2710 520
M186A4.57
L222A5.42
300 60
∼2500
800 200
a
CCK2Rβ recruiting β-arrestins and CCK2RG state activating phospholipase-C were monitored by BRET and inositol phosphate measurement,
respectively. Results are given as IC50 (potency), corresponding to concentrations of antagonists inhibiting half of CCK(10 nM)-induced responses.
Dose−response curves for both biological assays are given in Figure 7 and Supplementary Fig. 4. Values are mean
experiments. NC = value cannot be calculated.
SEM of three separate
summarize, modeling predictions validated by experimental
results with M134A3.32 and Y380A7.43 mutants support the hy-
pothesis that the interaction between M1343.32 and Y3807.43 is
required for the CCK-induced CCK2Rβ state recruiting
β-arrestin-2. Furthermore, normal recruitment of β-arrestin-2
by Y380F7.43 together with high level recruitment of β-arrestin-2 by
2568
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
Figure 8. M1343.32 and Y3807.43 in the CCK2R are differently involved in the stabilization of the CCK2Rβ state recruiting β-arrestin-2 or the
CCK2RG state stimulating phopholipase-C. (a−d) Schematic representation of β-arrestin-2 recruitment to wild-type and mutated CCK2R at
positions M1343.32 and Y3807.43. (e) Confocal microscopy analyses showing that mutation M134A3.32 or Y380A7.43 decreased by 85% or 55%
recruitment of β-arrestin-2 to CCK2R, respectively. (f) Confocal microscopy image showing (1) absence of co-localization between CCK2R-GFP
(in green) and β-arrestin-2-RFP (in red) before CCK stimulation; (2) recruitment of β-arrestin-2-RFP to CCK2R-GFP (yellow spots) upon CCK
stimulation (20 min, 10 nM CCK); and (3) intracellular co-localization of M134C3.32-GFP and β-arrestin-2-RFP (yellow) in the absence of CCK or
(4) in the presence of CCK. (g) BRET dose−response curves showing decreases of 75 or 45% of maximal CCK-induced BRET signal between
β-arrestin-2 and M134A3.32 or Y380A7.43 mutants as well as high basal BRET signal between β-arrestin-2 and M134C3.32. (h) Dose−response curves
for inositol phosphate production showing absence of effect of mutations M134A3.32 or Y380A7.43 on maximal CCK-induced phospholipase-C
activation. Maximal inositol production with the mutants is expressed as the percent of maximal production achieved with the wild-type CCK2R.
Note that shifts in potencies observed for BRET and inositol phosphate curves with mutant Y380A7.43 reflect involvement of residue Y380A7.43 in
interaction with CCK in the binding site.22 Experiments in (e) were carried out with HEK 293T cells coexpressing transiently wild-type or mutated
CCK2R and β-arrestin-2-YFP. Recruitment of β-arrestin-2-YFP to activated receptors (labeled by Alexa F 647-CCK) was quantified at 5 min of
stimulation as described in the Experimental Procedures. (f) Experiments were carried out with HEK 293T cells coexpressing transiently CCK2R-
GFP or M134C3.32-GFP and β-arrestin-2-RFP. (g) Experiments were carried with HEK 293T cells expressing transiently wild-type or mutated
CCK2R-Rluc and β-arrestin-2-YFP. (h) Experiments were carried with HEK 293T cells expressing transiently wild-type or mutated CCK2R-Rluc.
Asterisks indicate the p values giving the significance of β-arrestin-2-YFP recruitment by mutants relative to wild-type CCK2R: ** 0.001 < p < 0.01;
*** p < 0.001.
M134C3.32 mutant in the absence of CCK strongly suggest
that residue M1343.32 or C1343.32 and Y3807.43 are able to form
a sulfur−aromatic interaction that likely holds the receptor in the
CCK2Rβ state in the presence or in the absence of CCK,
respectively.
the CCK2R (Supplementary Fig. 2), might play as catalyzing
role, forcing the side chains of M1343.32 and Y3807.43 to adopt a
favorable conformation to promote sulfur−aromatic interac-
tion. The distance between the aromatic ring and sulfur atom
remains short and stable (4 Å) in the simulations of the un-
occupied M134C3.32 mutant (Figure 9a), suggesting that the
Cys-Tyr distance and the polarity of the S−H group can
stabilize the sulfur−aromatic interaction in the absence of CCK.
In Figure 9b, we show the typical molecular dynamics struc-
tures of the wild-type CCK2R and M134C3.32 mutant, from
which one can see the face position of the Cys and the distant
position of Met to the Tyr ring.
We have further examined the possibility of sulfur−aromatic
interaction conducting the molecular dynamics simulations of
our CCK2R model in the ligand free form, occupied with CCK
and the ligand free form of M134C3.32 mutant. Molecular
dynamics simulations represent a theoretical and computational
method that uses Newton’s law of motion to predict protein
plasticity in interactions and function. The simulation data
indicate that the sulfur−aromatic distance between M1343.32
and Y3807.43 is shorter and more stable with the average value
of 4.5 Å in the presence of CCK than in the unoccupied wild-
type CCK2R (Figure 9a), suggesting that CCK stabilizes
M1343.32−Y3807.43 contacts. Notably, our experimentally validated
model of an agonist−receptor interactions built in the previous
studies22,29 suggests that side chains of residue Met/Leu of
CCK which are the closest to residues M1343.32 and Y3807.43 of
DISCUSSION
■
The newly recognized ability of β-arrestins to serve as signal
transducers for 7TMRs and the discovery of β-arrestin or G
protein biased ligands represent innovative paradigms in 7TMR
signaling with potential therapeutic interest.7,35 Here, taking
CCK2R as a 7TMR model system which both couples to Gαq
and recruits β-arrestins, we provide converging pharmacological
2569
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
binding cavity of the CCK2R, which is composed of residues
from transmembrane helices, is modified in the CCK2Rβ state
relative to the CCK2G state. Third, mutation of two interacting
amino acids from helices 3 and 7, namely M1343.32 and Y3807.43
produced CCK2R variants, which either poorly recruit β-arrestin-2
after CCK stimulation or recruit β-arrestin-2 in the absence
of CCK.
The key role of M1343.32−Y3807.43 interaction in the stability
of the CCK2Rβ state through S−aromatic interaction, and
possible changes in relative position and/or orientation of helix
3 and 7 are in line with reports that associate conformational
changes in helix 7 of the β2-adrenergic receptor and β-arrestin-
biased agonists stimulation.13,15 Interestingly, interaction
between M1343.32 and Y3807.43 is observed in CCK2Rβ state
recruiting β-arrestin-2 while an equivalent constraint is found in
most crystallized members of class A 7TMRs at the level of the
ligand binding pocket.11,15,25,36−48 The nature of this constraint
and its role in the stabilization of different signaling conforma-
tions would have evolved in different sub-families. Indeed,
M1343.32−Y3807.43 interaction in CCK2R corresponds to the
D3.32-Y7.43 hydrogen bond in aminergic receptors (http://lmc.
uab.cat/gmos). Furthermore, in aminergic receptors, like in
CCK2R, ligands bind very close to these residues and pre-
sumably regulate contacts between helices 3 and 7. The
existence of S−aromatic interactions is well documented in
several proteins,49,50 but rarely in 7TMRs. However, analysis of
crystal structures revealed that S−aromatic contacts occur
much more frequently than would be expected from random
association.51 Estimates for the S−aromatic interaction energies
give comparable values to aromatic−aromatic interactions or
N−H···O hydrogen bonds.49,52
Figure 9. Sulfur−aromatic interactions between residues in position
1343.32 and 3807.43 in the CCK2R. (a) Time dependence of the
distance between S atom of M1343.32 and the center of the mass of the
aromatic ring of Y3807.43 in the unoccupied CCK2R (blue), in the
complex CCK2R−CCK (red), and in the unoccupied mutant
M134C3.32 (cyan) from 20 ns molecular dynamics. The sulfur−
aromatic interaction is likely stably present in the CCK2R.CCK
complex (4.5 Å), whereas the interaction is poorly seen in the
unoccupied CCK2R (>5 Å). In the unoccupied M134C3.32 mutant, the
distance between the aromatic ring and sulfur atom remains short and
constant without needing a ligand. (b) Superimposition of typical
molecular dynamics structures of the CCK2R wild-type and M134A3.32
mutant. The receptor superimposition is shown from the side view.
The backbone of the wild-type and mutant receptor is shown in white
and green, respectively, and only the side chains at positions 3.32 and
7.43 are visualized in the stick-like form. The image of the receptor
typical structure shows that the sulfur atom of cysteine faces the
aromatic ring of Y3807.43, whereas the sulfur of M1343.32 is above the
aromatic ring of Y3807.43
.
and molecular evidence that CCK-induced conformational
states of the CCK2R recruiting β-arrestin-2 (termed CCK2Rβ
state) and that activating phospholipase-C through Gαq coupl-
ing (termed CCK2RG state) are distinct. Differences were
shown at the CCK2R orthosteric binding site formed by trans-
membrane helices of the CCK2R. This is of prime importance
since biased agonism initiates at the binding site of receptors.
Our conclusion is based on strong converging pharmacological
and molecular data. First, BRET results, which account for
β-arrestin-2 recruitment to the CCK2R showed that CCK-
induced formation of CCK2Rβ state is competitively inhibited
by the two structurally unrelated molecules PD135158 and
L365260 but not by GV150013X, whereas CCK-stimulation of
CCK2RG state is competitively inhibited by the three mol-
ecules. In agreement with these results, GV150013X is almost
inefficient to competitively block CCK-induced CCK2R inter-
nalization, a process, which was previously demonstrated to
involve β-arrestin-2 recruitment.23 Second, CCK2R binding sites
presenting a low affinity for CCK and to which GV150013X is
unable to compete efficiently with CCK were identified in addition
to high affinity CCK binding sites which can bind GV150013X
with a high affinity. The low affinity CCK binding sites likely
corresponded to CCK2Rβ state recruiting β-arrestin2.
Furthermore, our attempts at using structural modeling and
site-directed mutagenesis to identify structural elements which
most likely hinder binding of GV150013X to CCK-induced
CCK2Rβ state showed that two mutations within the ortho-
steric binding site of CCK2R (M186A4.57 or L222A5.42) or sub-
stitution of adamantane group by methyl in GV150013X re-
stores part of the antagonist activity on the CCK2Rβ state,
whereas these mutations or chemical substitution cause either
minor effects or dramatic decreases on antagonist potency on
CCK-induced formation of CCK2RG, respectively. All together
the data strongly support that the shape of the orthosteric
Although the current study does not provide 3-D structures
of CCK2Rβ and CCK2RG states, it is foreseeable that dif-
ferences at the binding site level and the relative position of
helices 3 and 7 are accompanied by conformational changes
in intracellular parts of the receptor. The specific structural
changes and mechanism remain to be identified. As a second
major predictable structural difference, CCK2R is likely
phosphorylated at specific positions in the CCK2Rβ state, but
not in the CCKR2G state, since the phosphorylation by GRK is
subsequent to activation of G protein-dependent signals.5
However, we previously showed that phosphorylation of the
C-terminal region in the CCK2R upon CCK stimulation favors
β-arrestin-2 recruitment, but is dispensable.23 Hence, binding of
β-arrestin-2 to CCK2Rβ state compulsorily involves other
regions of the receptor (most likely intracellular parts), the
conformation of which should be different in CCK2Rβ state
relative to CCK2RG state. This view agrees with the accepted
mechanism of recognition between 7TMRs and arrestins in-
volving two main sites of interaction on both partners, which
were termed “activation-recognition” and “phosphorylation-
recognition” sites.53,54 The phosphorylation-recognition site of
arrestins discriminates phosphorylated from unphosphorylated
7TMRs. So, this phosphorylation-recognition is unable to select
a particular 7TMR.55,56 The activation-recognition site in arrestins
discriminates active from inactive 7TMRs and interaction through
this site seems to be sufficient for activation of arrestins. Indeed,
phosphorylation-deficient mutants of 7TMRs were shown to cause
similar conformational changes in β-arrestin-2 as the wild-type
corresponding 7TMRs.57
How changes identified in the CCK2Rβ state versus CCK2RG
state occur is an unresolved issue. However, as we previously
simulated for transition from inactive CCK2R to CCK2RG, 19 it
2570
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
is predictable that coordinated motions of helices involving
several networks of interactions cause changes within the bind-
ing cavity, which propagate to intracellular portions of the
receptor forming binding site of β-arrestin-2. It is also plausible
that binding of β-arrestin-2 to the phosphorylated region of
CCK2R and subsequently to the activation-recognition site
contributes to formation and stabilization of the CCK2Rβ state.
Hence, the fact that the CCK2Rβ state appears to be stabilized
by slightly higher concentrations of CCK than does the CCK2RG
state might result in part from β-arrestin binding which, as an
allosteric regulator transmits conformational changes to the
orthosteric binding site. This view is in line with structural and
pharmacological data whereby binding affinity of agonists for
7TMR, including CCK receptors, is augmented by hetero-
trimeric G protein coupling, which, therefore, acts as allosteric
regulators on the binding site.9,58 It is worthy to note that
changes in the position of some residues in the orthosteric
binding site of CCK2Rβ state relative to CCK2RG state hinder
binding of the non-peptide antagonist GV150013X whereas
these changes less significantly affect CCK recognition. This
observation is in line with data showing that a synthetic ligand
can be sensitive to minor changes within the orthosteric bind-
ing site of a 7TMR whereas the natural agonist is not or is less
sensitive to such changes. For instance, canine and human CCK2R
recognize CCK with strictly identical affinities but display very
different affinities toward a synthetic antagonist. The presence of a
val in helice VI in canine CCK2R and Ile at the same position in
human CCK2R (so two residues with very similar side-chains)
explains the different pharmacological behaviors.59 At last, the
existence of CCK2Rβ state distinct from CCK2RG state must be
discussed in light of accumulating data indicating that 7TMRs
can exist both as monomer, homodimers or larger oligomeric
forms functionally selective in the cell surface membrane.60−62
Thus, it is not excluded that CCK2Rβ and CCK2RG states cor-
respond to distinct structural entities. An agonist dose-dependent
switch from G protein-coupled to G protein-independent signaling
was previously reported for β2-adrenergic receptor.63 Indeed,
β2-adrenergic receptor signals through Gαs upon low con-
centrations of agonist to activate MAP kinase pathway, whereas
it signals also via a Src-dependent G protein-independent path-
way upon higher agonist concentrations. The authors proposed
a mechanism whereby high concentrations of agonists stabilize
dimers of β2-adrenergic receptor, providing structural features
necessary for direct activation of Src by bringing two molecules
of Src into proximity, and allowing their intermolecular
autophosphorylation and activation.63
CONCLUSION
■
This study provides converging pharmacological and molecular
evidence that a 7TMR, the CCK2R, can adopt distinct con-
formations to either trigger G protein signaling or β-arrestin-2
recruitment. These conformations are distinguishable at the
orthosteric binding site which is defined by transmembrane
helices. An antagonist of the CCK2R, GV150013X, was identified
as a biased antagonist on β-arrestin-2 recruitment pathway. Two in-
teracting amino acids, M3.32 and Y7.43, play key roles in formation/
stabilization of a state recruiting or β-arrestin-2. These findings shall
have future impact not only on CCK2R pharmacology but also on
7TMRs in general, and they validate the relevance of the design of
biased ligands able to selectively target each functional conformation
of 7TMRs.
ASSOCIATED CONTENT
■
S
* Supporting Information
Supplementary figures, legends to those figures, and supple-
mentary methods. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
daniel.fourmy@inserm.fr; i.tikhonova@qub.ac.uk
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We greatly appreciate the gifts of plasmids encoding YFP-tagged
β -arrestin-2 from Marc Caron (Duke University Medical Center)
and GFP-tagged β-arrestin-2 from Robert Lefkowitz (Duke
University Medical Center). We thank Dr Romina D’Angelo
from Cellular Imaging Facility Rangueil (I2MC/TRI Plateform)
for excellent technical support and advice on confocal microscopy
and image analysis, and Chantal Zedde and Isabelle Fabing
(CNRS, LSPCMIB, UMR-5068) for purification of GV-CH3. The
work was supported by Grants from Association pour la Recherche
contre le Cancer (ARC No. 4870) and Ligue Nationale Contre le
Cancer (comite
Salut Carlos III.
́
31). A.C. is supported by a grant from Instituto de
REFERENCES
■
(1) Hopkins, A. L.; Groom, C. R. Nat. Rev. Drug Discov. 2002, 1, 727.
(2) Rajagopal, S.; Rajagopal, K.; Lefkowitz, R. J. Nat. Rev. Drug
Discov. 2010, 9, 373.
Two major consequences of our findings can be anticipated
regarding the CCK2R. First, since GV150013X is not able to
inhibit CCK-induced recruitment of β-arrestin-2, this com-
pound should not be able to efficiently block CCK2R-induced
β-arrestin-2-dependent signaling pathways and associated
events, as illustrated with CCK2R internalization. This might
be of high importance owing to expression of the CCK2R in a
large variety of cancers and its recognized contribution to
tumoral progression as well as increasing role of β-arrestin1/2-
dependent signaling pathways to cancer progression.35,64−66
Second, determination of the structure of CCK2R both in the
CCK2Rβ and CCK2RG states is required to enable design of
new ligands of desired pharmacological profiles. In particular, a
β-arrestin1/2-biased CCK2R agonist, would be very helpful to
delineate consequences of new signaling pathways downstream
of the CCK2R in cellular and animal models of cancer.
(3) Lefkowitz, R. J.; Shenoy, S. K. Science 2005, 308, 512.
(4) DeFea, K. A. Cell Signal. 2011, 23, 621.
(5) Ferguson, S. S. Pharmacol. Rev. 2001, 53, 1.
(6) Hanyaloglu, A. C.; von Zastrow, M. Annu. Rev. Pharmacol.
Toxicol. 2008, 48, 537.
(7) Whalen, E. J.; Rajagopal, S.; Lefkowitz, R. J. Trends Mol. Med.
2011, 17, 126.
(8) Vaidehi, N.; Kenakin, T. Curr. Opin. Pharmacol. 2010, 10, 775.
(9) Rasmussen, S. G.; DeVree, B. T.; Zou, Y.; Kruse, A. C.; Chung, K.
Y.; Kobilka, T. S.; Thian, F. S.; Chae, P. S.; Pardon, E.; Calinski, D.;
Mathiesen, J. M.; Shah, S. T.; Lyons, J. A.; Caffrey, M.; Gellman, S. H.;
Steyaert, J.; Skiniotis, G.; Weis, W. I.; Sunahara, R. K.; Kobilka, B. K.
Nature 2011, 477, 549.
(10) Xu, F.; Wu, H.; Katritch, V.; Han, G. W.; Jacobson, K. A.; Gao,
Z. G.; Cherezov, V.; Stevens, R. C. Science 2011, 332, 322.
(11) Lebon, G.; Warne, T.; Edwards, P. C.; Bennett, K.; Langmead,
C. J.; Leslie, A. G.; Tate, C. G. Nature 2011, 474, 521.
2571
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
(12) Warne, T.; Moukhametzianov, R.; Baker, J. G.; Nehme, R.;
Edwards, P. C.; Leslie, A. G.; Schertler, G. F.; Tate, C. G. Nature 2011,
469, 241.
(37) Chien, E. Y.; Liu, W.; Zhao, Q.; Katritch, V.; Han, G. W.;
Hanson, M. A.; Shi, L.; Newman, A. H.; Javitch, J. A.; Cherezov, V.;
Stevens, R. C. Science 2011, 330, 1091.
(13) Granier, S.; Kim, S.; Shafer, A. M.; Ratnala, V. R.; Fung, J. J.;
Zare, R. N.; Kobilka, B. J. Biol. Chem. 2007, 282, 13895.
(14) Kahsai, A. W.; Xiao, K.; Rajagopal, S.; Ahn, S.; Shukla, A. K.;
Sun, J.; Oas, T. G.; Lefkowitz, R. J. Nat. Chem. Biol. 2011, 7, 692.
(15) Liu, J. J.; Horst, R.; Katritch, V.; Stevens, R. C.; Wuthrich, K.
Science 2012, 335, 1106.
(16) Rahmeh, R.; Damian, M.; Cottet, M.; Orcel, H.; Mendre, C.;
Durroux, T.; Sharma, K. S.; Durand, G.; Pucci, B.; Trinquet, E.; Zwier,
J. M.; Deupi, X.; Bron, P.; Baneres, J. L.; Mouillac, B.; Granier, S. Proc.
Natl. Acad. Sci. U.S.A. 2012, 109, 6733.
(17) Dufresne, M.; Seva, C.; Fourmy, D. Physiol. Rev. 2006, 86, 805.
(18) Yaqub, T.; Tikhonova, I. G.; Lattig, J.; Magnan, R.; Laval, M.;
Escrieut, C.; Boulegue, C.; Hewage, C.; Fourmy, D. Mol. Pharmacol.
2010, 77, 547.
(19) Hughes, J.; Boden, P.; Costall, B.; Domeney, A.; Kelly, E.;
Horwell, D. C.; Hunter, J. C.; Pinnock, R. D.; Woodruff, G. N. Proc.
Natl. Acad. Sci. U.S.A. 1990, 87, 6728.
(20) Bock, M. G.; DiPardo, R. M.; Evans, B. E.; Rittle, K. E.; Whitter,
W. L.; Veber, D. E.; Anderson, P. S.; Freidinger, R. M. J. Med. Chem.
1989, 32, 13.
(21) Ursini, A.; Capelli, A. M.; Carr, R. A.; Cassara, P.; Corsi, M.;
Curcuruto, O.; Curotto, G.; Dal Cin, M.; Davalli, S.; Donati, D.;
Feriani, A.; Finch, H.; Finizia, G.; Gaviraghi, G.; Marien, M.;
Pentassuglia, G.; Polinelli, S.; Ratti, E.; Reggiani, A. M.; Tarzia, G.;
Tedesco, G.; Tranquillini, M. E.; Trist, D. G.; Van Amsterdam, F. T. J.
Med. Chem. 2000, 43, 3596.
(38) Choe, H. W.; Kim, Y. J.; Park, J. H.; Morizumi, T.; Pai, E. F.;
Krauss, N.; Hofmann, K. P.; Scheerer, P.; Ernst, O. P. Nature 2011,
471, 651.
(39) Granier, S.; Manglik, A.; Kruse, A. C.; Kobilka, T. S.; Thian, F.
S.; Weis, W. I.; Kobilka, B. K. Nature 2012, 485, 400.
(40) Haga, K.; Kruse, A. C.; Asada, H.; Yurugi-Kobayashi, T.;
Shiroishi, M.; Zhang, C.; Weis, W. I.; Okada, T.; Kobilka, B. K.; Haga,
T.; Kobayashi, T. Nature 2012, 482, 547.
(41) Hanson, M. A.; Roth, C. B.; Jo, E.; Griffith, M. T.; Scott, F. L.;
Reinhart, G.; Desale, H.; Clemons, B.; Cahalan, S. M.; Schuerer, S. C.;
Sanna, M. G.; Han, G. W.; Kuhn, P.; Rosen, H.; Stevens, R. C. Science
2012, 335, 851.
(42) Jaakola, V. P.; Griffith, M. T.; Hanson, M. A.; Cherezov, V.;
Chien, E. Y.; Lane, J. R.; Ijzerman, A. P.; Stevens, R. C. Science 2008,
322, 1211.
(43) Murakami, M.; Kouyama, T. Nature 2008, 453, 363.
(44) Shimamura, T.; Shiroishi, M.; Weyand, S.; Tsujimoto, H.;
Winter, G.; Katritch, V.; Abagyan, R.; Cherezov, V.; Liu, W.; Han, G.
W.; Kobayashi, T.; Stevens, R. C.; Iwata, S. Nature 2011, 475, 65.
(45) Thompson, A. A.; Liu, W.; Chun, E.; Katritch, V.; Wu, H.;
Vardy, E.; Huang, X. P.; Trapella, C.; Guerrini, R.; Calo, G.; Roth, B.
L.; Cherezov, V.; Stevens, R. C. Nature 2012, 485, 395.
(46) Warne, T.; Serrano-Vega, M. J.; Baker, J. G.; Moukhametzianov,
R.; Edwards, P. C.; Henderson, R.; Leslie, A. G.; Tate, C. G.; Schertler,
G. F. Nature 2008, 454, 486.
(47) Wu, B.; Chien, E. Y.; Mol, C. D.; Fenalti, G.; Liu, W.; Katritch,
V.; Abagyan, R.; Brooun, A.; Wells, P.; Bi, F. C.; Hamel, D. J.; Kuhn,
P.; Handel, T. M.; Cherezov, V.; Stevens, R. C. Science 2010, 330,
1066.
(22) Foucaud, M.; Marco, E.; Escrieut, C.; Low, C.; Kalindjian, B.;
Fourmy, D. J. Biol. Chem. 2008, 283, 35860.
(23) Magnan, R.; Masri, B.; Escrieut, C.; Foucaud, M.; Cordelier, P.;
Fourmy, D. J. Biol. Chem. 2011, 286, 6707.
(24) Masri, B.; Salahpour, A.; Didriksen, M.; Ghisi, V.; Beaulieu, J.
M.; Gainetdinov, R. R.; Caron, M. G. Proc. Natl. Acad. Sci. U.S.A. 2008,
105, 13656.
(48) Manglik, A.; Kruse, A. C.; Kobilka, T. S.; Thian, F. S.;
Mathiesen, J. M.; Sunahara, R. K.; Pardo, L.; Weis, W. I.; Kobilka, B.
K.; Granier, S. Nature 2012, 485, 321.
(49) Meyer, E. A.; Castellano, R. K.; Diederich, F. Angew. Chem., Int.
(25) Rasmussen, S. G.; Choi, H. J.; Fung, J. J.; Pardon, E.; Casarosa,
P.; Chae, P. S.; Devree, B. T.; Rosenbaum, D. M.; Thian, F. S.;
Kobilka, T. S.; Schnapp, A.; Konetzki, I.; Sunahara, R. K.; Gellman, S.
H.; Pautsch, A.; Steyaert, J.; Weis, W. I.; Kobilka, B. K. Nature 2011,
469, 175.
Ed. 2003, 42, 1210.
(50) Chakrabarti, P.; Bhattacharyya, R. Prog. Biophys. Mol. Biol. 2007,
95, 83.
(51) Zauhar, R. J.; Colbert, C. L.; Morgan, R. S.; Welsh, W. J.
Biopolymers 2000, 53, 233.
(26) Gales, C.; Poirot, M.; Taillefer, J.; Maigret, B.; Martinez, J.;
Moroder, L.; Escrieut, C.; Pradayrol, L.; Fourmy, D.; Silvente-Poirot,
S. Mol. Pharmacol. 2003, 63, 973.
(52) Ringer, A. L.; Senenko, A.; Sherrill, C. D. Protein Sci. 2007, 16,
2216.
(53) Gurevich, V. V.; Gurevich, E. V. Trends Pharmacol. Sci. 2004, 25,
(27) Silvente-Poirot, S.; Escrieut, C.; Gales, C.; Fehrentz, J. A.;
Escherich, A.; Wank, S. A.; Martinez, J.; Moroder, L.; Maigret, B.;
Bouisson, M.; Vaysse, N.; Fourmy, D. J. Biol. Chem. 1999, 274, 23191.
(28) Foucaud, M.; Archer-Lahlou, E.; Marco, E.; Tikhonova, I. G.;
Maigret, B.; Escrieut, C.; Langer, I.; Fourmy, D. Regul. Pept. 2008, 145,
17.
105.
(54) Gurevich, V. V.; Gurevich, E. V. Pharmacol. Ther. 2006, 110,
465.
(55) Vishnivetskiy, S. A.; Gimenez, L. E.; Francis, D. J.; Hanson, S.
M.; Hubbell, W. L.; Klug, C. S.; Gurevich, V. V. J. Biol. Chem. 2011,
286, 24288.
(29) Marco, E.; Foucaud, M.; Langer, I.; Escrieut, C.; Tikhonova, I.
(56) Zhan, X.; Gimenez, L. E.; Gurevich, V. V.; Spiller, B. W. J. Mol.
G.; Fourmy, D. J. Biol. Chem. 2007, 282, 28779.
Biol. 2011, 406, 467.
(30) Tikhonova, I. G.; Marco, E.; Lahlou-Archer, E.; Langer, I.;
Foucaud, M.; Maigret, B.; Fourmy, D. Curr. Top. Med. Chem. 2007, 7,
1243.
(57) Shukla, A. K.; Violin, J. D.; Whalen, E. J.; Gesty-Palmer, D.;
Shenoy, S. K.; Lefkowitz, R. J. Proc. Natl. Acad. Sci. U.S.A. 2008, 105,
9988.
(31) Langer, I.; Tikhonova, I. G.; Travers, M. A.; Archer-Lahlou, E.;
Escrieut, C.; Maigret, B.; Fourmy, D. J. Biol. Chem. 2005, 280, 22198.
(32) Sherman, W.; Day, T.; Jacobson, M. P.; Friesner, R. A.; Farid, R.
J. Med. Chem. 2006, 49, 534.
(58) Huang, S. C.; Fortune, K. P.; Wank, S. A.; Kopin, A. S.; Gardner,
J. D. J. Biol. Chem. 1994, 269, 26121.
(59) Beinborn, M.; Lee, Y. M.; McBride, E. W.; Quinn, S. M.; Kopin,
A. S. Nature 1993, 362, 348.
(33) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14,
27.
(34) Gales, C.; Kowalski-Chauvel, A.; Dufour, M. N.; Seva, C.;
Moroder, L.; Pradayrol, L.; Vaysse, N.; Fourmy, D.; Silvente-Poirot, S.
J. Biol. Chem. 2000, 275, 17321.
(60) Fung, J. J.; Deupi, X.; Pardo, L.; Yao, X. J.; Velez-Ruiz, G. A.;
Devree, B. T.; Sunahara, R. K.; Kobilka, B. K. EMBO J. 2009, 28, 3315.
(61) Milligan, G. Br. J. Pharmacol. 2009, 158, 5.
(62) Urizar, E.; Yano, H.; Kolster, R.; Gales, C.; Lambert, N.; Javitch,
J. A. Nat. Chem. Biol. 2011, 7, 624.
(35) Luttrell, L. M.; Gesty-Palmer, D. Pharmacol. Rev. 2010, 62, 305.
(36) Cherezov, V.; Rosenbaum, D. M.; Hanson, M. A.; Rasmussen, S.
G.; Thian, F. S.; Kobilka, T. S.; Choi, H. J.; Kuhn, P.; Weis, W. I.;
Kobilka, B. K.; Stevens, R. C. Science 2007, 318, 1258.
(63) Sun, Y.; Huang, J.; Xiang, Y.; Bastepe, M.; Juppner, H.; Kobilka,
B. K.; Zhang, J. J.; Huang, X. Y. EMBO J. 2007, 26, 53.
(64) Watson, S. A.; Grabowska, A. M.; El-Zaatari, M.; Takhar, A. Nat.
Rev. Cancer 2006, 6, 936.
2572
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573

Journal of the American Chemical Society
Article
(65) Fourmy, D.; Gigoux, V.; Reubi, J. C. Gastroenterology 2011, 141,
814.
(66) Gales, C.; Sanchez, D.; Poirot, M.; Pyronnet, S.; Buscail, L.;
Cussac, D.; Pradayrol, L.; Fourmy, D.; Silvente-Poirot, S. Oncogene
2003, 22, 6081.
2573
dx.doi.org/10.1021/ja308784w | J. Am. Chem. Soc. 2013, 135, 2560−2573