PHARMACOLOGY OF G PROTEIN COUPLED RECEPTORS
Visit the link below to download the full version of this book:
https://medipdf.com/product/pharmacology-of-g-protein-coupled-receptors/
Click Download Now
xvi Preface
Class A receptors, such as the muscarinic family for which subtype selectivity
had been challenging, as well as Class C receptors (mGluR) where the
extracellular ligand binding site was not readily targeted. The special char-
acteristics of allosteric modulators provide unique pharmacological advan-
tages, such as dependence on physiological signaling, that go beyond the
simple availability of selective activators and inhibitors. The concept of
functional selectivity or biased agonists also provides a new pharmacological
toolbox to improve therapeutics directed at even the most hoary of GPCR
drug targets. By tweaking a receptor in ways that produce only a subset of its
full repertoire of responses, biased agonists can bring new specificity to old
drug targets. Further, effects that were never anticipated from classical
agonist and antagonist ligands, such as selective desensitization or unique
signal outputs, may now be possible.
Finally, mechanisms that regulate or modulate receptor function are
being rapidly elucidated and may provide novel drug targets. A number of
receptors, either normal or with genetic mutations, are known to traffic
poorly to the plasma membrane where most of their physiological actions
take place. Drugs that enhance that membrane trafficking could improve
function for patients with rare (or not so rare) genetic mutations. Alterna-
tively they could improve function for receptors (e.g., d opioid or GnRH) that
have low levels of trafficking under normal circumstances. Other regulatory
mechanisms such as synaptic localization by PDZ domains, control of sig-
naling by RGS proteins, or modulation by other scaffold proteins are also
potential sites of therapeutics development. While these may not be specific
for an individual receptor, they have a potential advantage, compared to
direct receptor-targeting ligands, whereby they may have unique tissue-
specific effects—provided that the regulatory protein has a different tissue
distribution than does the receptor itself.
In all, the past decade has provided abundant new developments in our
understanding of GPCRs. It is my hope that the cross section of those
represented in this volume will provide new insights for biologists, pharma-
cologists, and others undertaking the discovery of new therapeutics.
RICHARD R. NEUBIG
Ann Arbor, MI
Miles Congreve, Christopher Langmead, and
Fiona H. Marshall
Heptares Therapeutics, Biopark, Welwyn Garden City, Hertfordshire, United Kingdom
The Use of GPCR Structures in
Drug Design
Abstract ___________________________________________________________________________________________________________________________
Structure-based drug discovery is routinely applied to soluble targets
such as proteases and kinases. It is only recently that multiple high-resolution
X-ray structures of G protein-coupled receptors (GPCRs) have become avail-
able. Here we review the technology developments that have led to the recent
plethora of GPCR structures. These include developments in protein expres-
sion and purification as well as techniques to stabilize receptors and crystal-
lize them. We discuss the findings derived from the new structures with
regard to understanding GPCR function and pharmacology. Finally, we
examine the utility of structure-based drug discovery approaches including
homology modeling, virtual screening, and fragment screening for GPCRs in
the context of what has been learnt from other target classes.
I. Introduction _____________________________________________________________________________________________________________
The past few years have seen some major technology developments in
membrane protein crystallization that have resulted in several new X-ray
structures of G protein-coupled receptors (GPCRs). The availability of
purified functional protein outside the cell membrane has enabled new meth-
ods in studying GPCR function in reconstituted systems using novel biophys-
ical techniques. The diverse GPCR structures now available provide more
accurate homology models across the Family A class of GPCRs which can be
used to enable structure-based drug discovery (SBDD) approaches to GPCRs.
Advances in Pharmacology, Volume 62 1054-3589/11 $35.00
© 2011 Elsevier Inc. All rights reserved. 10.1016/B978-0-12-385952-5.00011-7
2 Congreve et al.
In this chapter, we review recent technology developments, the new struc-
tures available, and their application to drug discovery.
II. Technology Developments Enabling GPCR Structure
Determination _____________________________________________________________________________________________________________
Until recently, progress in obtaining structures of GPCRs was very slow,
with 7 years elapsing between the solving of the structure of bovine
rhodopsin (Palczewski et al., 2000) and the solutions of the b2-adrenergic
receptor (b2AR) (Cherezov et al., 2007; Rasmussen et al., 2007). There have
been a number of technical difficulties which are now starting to be
overcome, thereby enabling the recent increase in GPCR structures. First,
large quantities of purified functional protein are required for structural
studies. The low level of expression of GPCRs and their instability outside
the membrane environment when solubilized with detergent are still major
problem areas. For most structural programs, at least 200 mg of protein may
be required to identify and refine the crystallization conditions (Kobilka &
Schertler, 2008). However, even when sufficient protein has been produced,
this has often failed to produce diffraction quality crystals, most likely due to
the structural instability and conformational heterogeneity of GPCRs.
A number of alternative approaches have recently been developed to over-
come these issues.
A. Optimizing Expression, Purification, and Stability
GPCRs generally express at much lower levels (< 1 mg/L) than soluble
proteins (> 10 mg/L). A large number of expression systems have been eval-
uated in attempts to increase expression with the majority of success coming
from bacterial, yeast, insect, or mammalian cell expression. Although total
yield of protein is important, further considerations include whether the
protein is correctly folded, the homogeneity of the protein (e.g., with respect
to posttranslational modifications) as well as the cost and scalability of the
process. Although bacterial expression systems are cost-effective and
straightforward to use, bacteria are unable to carry out the correct posttrans-
lational modifications, such as glycosylation which may be required for
correctly folded GPCRs. The reductive environment in bacteria may also
affect the correct formation of disulphide bonds within the tertiary structure.
A number of techniques have been developed which facilitate the correct
insertion and folding of GPCRs in bacteria, such as expression of GPCRs as
fusion proteins. These approaches have allowed protein production for a
number of GPCRs (such as the neurotensin receptor and the adenosine A2A
receptor) to be achieved (Grisshammer et al., 1993; Weiss & Grisshammer,
2002). An alternative approach is to refold receptors after purification from
Receptor Structure in Drug Design 3
inclusion bodies, for example, amphipols have been used to refold receptors
such as the leukotriene receptor BLT1 (Dahmane et al., 2009).
Expression of GPCRs in yeast provides a relatively easy, robust, and
cost-effective method for structural studies, and strains such as Saccharomy-
ces cerevisiae and Pichia pastoris have been used for GPCR overexpression
(Feng et al., 2002; Huang et al., 2008). The main problem with the yeast
system is that it can be challenging to yield functional protein due to both the
nature of the yeast cell wall and that yeast (like Escherichia coli) do not have
the correct membrane lipids or glycosylation systems found in mammalian
cells.
To date, the most useful expression system for GPCR structural studies is
the baculovirus expression system in insect cells. This has been used for
almost all the GPCR structures obtained so far including the b-adrenoceptors
(Cherezov et al., 2007; Warne et al., 2008), the adenosine A2A receptor
(Jaakola et al., 2008), and the chemokine receptor CXCR4 (Wu et al.,
2010b). Insect cells provide a more native intracellular environment for
GPCRs than bacteria or yeast and, although they do differ in membrane
lipid composition from mammalian cells, the percentage of correctly folded
GPCR in insect cells is higher than in mammalian cells (Errey & Marshall,
2010). A key advantage is the ability to scale up protein production, although
this is partially offset by the higher cost and time taken for virus generation.
Mammalian cells clearly represent the most physiological environment
for GPCR expression, however, the level of receptor expression and ability to
scale up are limiting factors for structural studies. One successful example in
using such a system for GPCR structural studies is the expression of rhodop-
sin using an inducible system in HEK293 cells (Reeves et al., 2002).
Irrespective of expression system used, different GPCRs vary dramati-
cally in their level of expression. Even very closely related receptors or
orthologues from different species can be very different, and to date, it is
not clear what determines the level of receptor expression. Expression levels
can be improved by mutagenesis (Parker et al., 1991) or by directed evolution
specifically targeted to increase expression (Sarkar et al., 2008).
Once sufficient expression levels have been obtained, the next critical
step in obtaining GPCR structures is the ability to purify the protein to
homogeneity while retaining correct folding. Removal of the protein from
its membrane lipid environment involves solubilization with detergents and
herein lies one of the fundamental problems of membrane protein crystalli-
zation. Detergents that are most useful for crystallization have short chains,
but these are highly destabilizing to membrane proteins. Advances in the
use of detergents and in combinations of detergents and lipids are proving a
significant factor in the recent success stories. The optimal choice of
detergents and lipid environment varies considerably from one receptor to
the next and must be determined experimentally for each receptor (Tate,
2009). Many membrane proteins have specific binding sites for lipids or
4 Congreve et al.
cholesterol which may have to be occupied to preserve correct function
(Hanson et al., 2008; Hunte, 2005).
The instability of GPCRs in detergents compatible with crystallization
has recently been overcome by protein engineering. A key reason for the
instability and the difficulty in obtaining diffracting crystals is the inherent
flexibility of GPCRs due to the existence of multiple conformational states.
In order for receptors to crystallize, they must be in a single, homogenous
conformation. This can in part be obtained by the addition of ligands which
preferentially bind to a single conformation; however, it can better be
obtained through the identification of stabilizing mutations. Conformational
thermostabilization has now been carried out for several GPCRs including
the b-adrenoceptors (Serrano-Vega et al., 2008), the neurotensin receptor
(Shibata et al., 2009), and the adenosine A2A receptor (Magnani et al., 2008).
In the latter example, the receptor has been stabilized in both agonist and
inverse agonist conformations. The resultant engineered receptors have been
called StaRs (for Stabilized Receptor; Robertson et al., 2011). StaRs generally
require 4–10-point mutations to give sufficient thermostability for crystalli-
zation. In the case of the b1AR, six mutations resulted in an increase in
melting point of 21 C compared to the wild-type receptor. This receptor
was stabilized in an antagonist conformation as demonstrated by a reduced
affinity for agonist binding while displaying no change in antagonist binding
(Serrano-Vega et al., 2008). A similar pharmacology was shown for the
adenosine A2A receptor StaR (Robertson et al., 2011). This stabilization
approach has resulted in high-resolution structures of both these GPCRs
(Warne et al., 2008 and Doré et al., unpublished). Comparison of these
structures determined for the b1AR and adenosine A2A receptor with struc-
ture solutions obtained using a different approach (Cherezov et al., 2007;
Jaakola et al., 2008) shows that the thermostabilizing point mutations do not
perturb the backbone structure of the proteins.
B. New Techniques for GPCR Crystallization
The ability to generate thermostabilized receptors has greatly assisted the
crystallization of GPCRs using conventional crystallization methods. Other
approaches directed at the crystallization process itself are also being devel-
oped. During crystallization, crystal contacts are usually formed between
polar regions which protrude from the detergent micelles. A problem in
crystallizing GPCRs is that they contain relatively small hydrophilic domains
which do not present useful crystallization contacts. In addition, flexible loop
regions are often truncated. Two approaches have proved successful in
facilitating GPCR crystallization. The first is the formation of a fusion
protein through the introduction of the well-folded soluble protein T4 lyso-
zyme (T4L) into the third intracellular loop (ICL3). This serves both to
reduce the flexibility of the ICL3 region as well as increasing the polar surface
Receptor Structure in Drug Design 5
area (PSA) available for crystal lattice contacts. The T4L fusion approach has
now been successfully applied to the crystallization of the b2AR (Cherezov
et al., 2007), adenosine A2A receptor (Jaakola et al., 2008), dopamine D3
receptor (Chien et al., 2010), and the chemokine receptor CXCR4 (Wu et al.,
2010b). One potential concern with making fusions is that they will perturb
the normal structure to a greater or lesser extent and, indeed, in several cases
have been shown to alter the pharmacology by increasing agonist affinity.
In addition, the presence of the T4L fusion prevents G protein signaling and
limits the full pharmacological profiling of the engineered receptor.
Monoclonal antibody fragments have also been used successfully to
crystallize GPCRs. This approach has been used to obtain antagonist and
agonist conformations of the b2AR. In the case of the antagonist conforma-
tion, a monoclonal antibody that bound to and stabilized ICL3 was
generated by immunizing mice with the b2AR reconstituted into proteolipo-
somes (Day et al., 2007). A Fab fragment of this antibody enabled the first
crystal structure of the b2AR (Rasmussen et al., 2007), although at consider-
ably lower resolution than the subsequent T4 fusion structure (Cherezov
et al., 2007). More recently, a nanobody which binds selectively to the active
conformation of the b2AR was identified by immunizing Llamas with
purified agonist bound b2AR. This appears to mimic part of the G alpha
subunit in its ability to promote the active conformational state of the
receptor and in the way the antibody interacts with the intracellular region
of the receptor. A cocrystal structure of the receptor and nanobody (also
known as a xaperone) shows similar conformational changes seen in the
active opsin structure (Rasmussen et al., 2011).
Despite advances in thermostabilization of GPCRs and techniques to
improve crystallization, the detergent environment required to crystallize
GPCRs, using vapor diffusion methods, remains unfavorable. In order to
overcome this problem, crystallization in a more lipidic environment that
contributes to the stability of the receptor is generally required to obtain
GPCR crystals in the absence of thermostabilization. One such method is
that of lipidic cubic phase (LCP) crystallization or in meso crystallization
(Caffrey, 2009). To date, all the GPCR structures obtained using the T4
fusion approach have also required LCP crystallization. The matrix of choice
is a bicontinuous cubic phase consisting of monoolein, water, and cholester-
ol. During crystallization, nucleation is favored at the interface between the
lipid and aqueous phases. A supply of protein occurs via diffusion within
the bilayer enabling crystal growth in a highly ordered fashion. The use of the
cubic phase method greatly facilitates the crystallization of unstable GPCRs
and has in some cases given high-resolution structures. The use of LCP has
been limited by the difficulty in handling and pipetting the extremely viscous
cubic phase. This has been overcome by the development of commercially
available robotic systems specifically designed to dispense the cubic phase
into crystallization plates (e.g., TTP Labtech www.ttplabtech.com).
6 Congreve et al.
III. GPCR Structures ______________________________________________________________________________________________
A. Rhodopsin as the Prototypical Receptor
In 2000, the first X-ray diffraction structure of a GPCR was published
with the solving of the bovine rhodopsin structure from bovine retinal disk
membranes (Palczewski et al., 2000). This high quality and detailed structure
revolutionized our understanding of GPCRs at the molecular level and paved
the way for SBDD approaches for GPCRs (Costanzi et al., 2009). The 2000
structure provided a detailed picture of the ligand-binding pocket of an
inactive receptor (Fig. 1A). The ligand 11-cis retinal was clearly located in
the ligand-binding pocket making a Schiff base linkage to Lys296 in TM7.
A Rhodopsin C β 2AR-T4L E β 1AR StaR
D121
D113
S211
S203
K296
F212 S215
S207
B D F
TM4 TM3 TM2
TM5 TM1
TM6 TM7
FIGURE 1 Comparisons of antagonist and agonist bound GPCR crystal structures. (A)
Cartoon structures of dark state (inactive) rhodopsin (1F88) overlaid with active rhodopsin
(2X72) in binding site and whole proteins. (B) Inactive rhodopsin is shown in purple (dark gray)
with the active form in yellow (light gray). The antagonist ligand 11-cis retinal is shown in
turquoise (dark gray) with the agonist all-trans retinal in pink (light gray). (C) Cartoon structure
of the b2 adrenerigic receptor (2RH1) with antagonist bound (purple/dark gray chains and pink/
dark gray ligand) overlaid with the cartoon structure of the activated b2adrenerigic receptor
(3P0G) (yellow/light gray chain with turquoise/light gray ligand) focused on ligand binding site
and shown as whole protein (D). (E) Cartoon structure of the b1 adrenerigic receptor (2VT4)
with antagonist bound (purple/dark gray chains and pink/dark gray ligand) overlaid with the
cartoon structure of the agonist bound b1 adrenerigic receptor (2Y03) (yellow/light gray chain
with turquoise/light gray ligand) focused on ligand binding site and shown as whole protein (F).
Receptor Structure in Drug Design 7
Additional residues from TMs 1, 2, and 7 surround the Schiff base while the
b-ionone ring interacts with side chains from TM5 and TM6.
For many years, the structures of the inactive dark-state rhodopsin
provided the only approach for GPCR homology modeling (Patny et al.,
2006). Several of these have been used for virtual screening and have been
able to identify known antagonists from libraries of compounds, but there
has been limited success in identifying agonists (see reviews Barton et al.,
2007; Garland & Blaney, 2010; Topiol & Sabio, 2009). There are several
limitations of using bovine rhodopsin as a template for modeling other
GPCRs (Mobarec et al., 2009; Topiol et al., 2011). First, although rhodopsin
shares many structural features with other Family A GPCRs, the overall
homology is less that 25%. Other families of GPCRs such as the secretin,
adhesion, and metabotropic receptors have no homology with rhodopsin.
Second, rhodopsin has a very different type of ligand activation mechanism,
as retinal is covalently bound to the receptor and signaling is initially trig-
gered by its isomerization by photons of light. The lack of a diffusible ligand
means that there is no obvious entrance to the ligand-binding site which is
blocked by the second extracellular loop (ECL2; Fig. 1A). It is difficult to
understand from the rhodopsin structure how such a diverse range of
ligands, which include, for example, peptides and large proteins, might access
and activate GPCRs. Nevertheless, the majority of small molecule drugs
most likely do bind to a site in an analogous region to retinal within the
transmembrane domain.
Recently, a number of structures of the activated apoprotein form of
rhodopsin, called opsin, have been obtained (Park et al., 2008; Scheerer et al.,
2008), which include complexes with a peptide derived from the C-terminal
tail of the G protein transducin. These structures show significant movements
of the helices which are presumed to represent the conformational changes
underlying receptor activation. The key changes involve significant move-
ments of TM5 and TM6 with the cytoplasmic end of TM6 moving 6–7 Å
outward from the center. Changes are also observed in highly conserved so-
called “microswitches” (Nygaard et al., 2009) including the ionic lock
formed between Arg135 and Glu247 which is broken during activation
(Weis & Kobilka, 2008). The first liganded activated structure of rhodopsin
has recently been solved; it contains a constitutively active mutation (E113Q)
and includes the bound agonist all-trans-retinal (Standfuss et al., 2011)
(Fig. 1A and B). The structure provides insight into the stages involved in
agonist activation of GPCRs. Surprisingly, the ligand is not in fact covalently
bound to the receptor but instead may represent an intermediate step in
which the ligand is entering or exiting the receptor-binding site. A key feature
of this structure is the rearrangement of water-mediated hydrogen-bonding
networks linked to agonist binding. This results in a movement of Trp265
(6.48) (numbers in parentheses refer to the Ballesteros–Weinstein residue
nomenclature system Ballesteros et al., 1995) in the CWxP motif at the
8 Congreve et al.
bottom of the ligand-binding pocket which is transmitted through the
hydrogen-bonding network to the two most conserved residues in TM1
(Asn55 (1.50)) and TM2 (Asp83 (2.50)) and the NPxxY motif in TM7.
The hydrogen-bonding network then extends toward the G protein peptide
with water molecules hydrogen bonding to both the receptor and G protein.
In addition to advancing our knowledge of receptor activation, these new
structures will be useful for developing models of agonist binding for use in
virtual screening and drug discovery.
B. Aminergic Receptors
1. Antagonist Binding
In 2007 and 2008, the GPCR field was revolutionized by the publication
of the crystal structures of the turkey b1 and human b2 adrenoceptors
(Cherezov et al., 2007; Rasmussen et al., 2007; Warne et al., 2008). These
structures, in complex with cyanopindolol (b1) and carazolol (b2), were the
first for GPCRs with diffusible ligands and represented significant advances
beyond the various structure determinations for rhodopsin/opsin (Smith,
2010). The structure of the human b2AR was determined first at medium
resolution (3.5 Å) by complex with an antibody fragment (Rasmussen et al.,
2007) and then at higher resolution (2.4 Å) by insertion of a T4L fusion
protein into ICL3 of the receptor (Cherezov et al., 2007). In contrast, the
structure of the turkey b1AR was determined to a resolution of 2.7 Å by the
introduction of six point mutations into the receptor construct which
increased the thermostability of the protein to enable crystallization (Warne
et al., 2008). Subsequent comparison of the structure of b1AR with that of
b2AR suggested that none of the point mutations introduced resulted in
significant changes in the backbone conformation but rather served to lock
the receptor in a single (inactive) conformation which would permit
purification and structure determination. Consistent with this hypothesis,
the resulting receptor construct displayed high affinity for antagonists and
inverse agonists but markedly reduced agonist affinity (Warne et al., 2008).
Unlike the b2AR construct, which cannot interact with G proteins due to the
presence of the T4 fusion protein, the b1AR can be activated by agonists,
albeit at much higher concentrations than required for the wild-type
receptor. This is due to the reduced agonist affinity at the StaRs compared
to wild-type (Warne et al., 2008).
The structures of both b1 and b2AR displayed similar topology to that
determined for rhodopsin, with seven transmembrane helical domains and
the presence of an eighth helix at the start of the C-terminal tail. However, in
both b1 and b2AR, the ECL2 formed an a-helical structure that was not
predicted computationally and was in contrast to the beta-sheet present
in rhodopsin. Neither construct permitted resolution of ICL3 due to either
the presence of the fusion protein/antibody fragment or the deletion of part of
Receptor Structure in Drug Design 9
the loop. However, the structure of the b1AR revealed a short a-helical
segment of second intracellular loop (ICL2) which is not seen in either
rhodopsin or b2AR, despite the high level of conservation with the latter.
The visualization of this helix is important due to its proposed role as a
switch in G protein activation (Burstein et al., 1998); indeed, the structure
reveals a clear interaction with the DRY motif in TM3, notably a hydrogen
bond between Tyr149(3.60) and Asp138(3.49). The absence of this helix in
b2AR is thought to be due to crystal lattice contacts with adjacent molecules.
Although both the b1 and b2AR have been crystallized in complex with
antagonists (and, in the case of the b1AR, stabilized in an inactive conforma-
tion), neither structure shows the presence of the “ionic lock,” the salt bridge
between Arg3.50 and Glu6.30 that is seen in rhodopsin and which has been
hypothesized to hold the receptor in an inactive conformation (Cherezov
et al., 2007; Warne et al., 2008).
As one might expect, given the high level of sequence conservation
between the receptor subtypes in the transmembrane domain regions, the
binding pockets for cyanopindolol and carazolol are very similar. In the
b1AR, there are 15 amino acid residues that make contact with cyanopindo-
lol, and all these are conserved in the b2AR (Cherezov et al., 2007; Warne
et al., 2008). For both ligands, residues from TM3, TM5, TM6, TM7, and
ECL2 make contacts to form the binding pocket, most notably between the
amine group of cyanopindolol/carazolol and Asp121/113(3.32) and Asn329/
312(7.39) and also between Ser211/203(5.42) (residue numbers in b1AR
shown first) and the indole nitrogen of cyanopindolol or the carbazole of
carazolol (although the rotamer state of Ser211/203(5.42) is different in the
two structures). Despite the high degree of conservation of the binding site,
there are a number of ligands which distinguish between the two receptor
subtypes pharmacologically (e.g., CGP 20712A). Close inspection of the
binding sites reveals that although only two amino acid residues within 8 Å
of the b1AR binding site are not conserved in b2AR (Val172(4.56) and
Phe306(7.35) are Thr164 and Tyr308, respectively, in b2AR), these changes
result in subtle differences in the shape and polarity of the binding pocket
which can underlie the selectivity profile of some ligands. Interestingly, in the
b2AR, Tyr308(7.35) has been implicated in mutagenesis studies as playing as
role in the selectivity of agonists due to its ability to form a hydrogen bond
with Asn293(6.55) (Kikkawa et al., 1998). These data suggest that receptor
subtype selectivity can be achieved by exploitation of minor differences in the
primary-binding pocket or by exploration of space beyond the immediate
interactions highlighted above, for example, from nonconserved residues in
ECL2 that contribute to the binding site(s).
A further advance in the structural understanding of Family A receptors
was achieved in 2010 with the publication of the structure of the dopamine
D3 receptor in complex with the high-affinity antagonist, eticlopride
(Chien et al., 2010). As with b2AR, this was achieved by the insertion of
10 Congreve et al.
the T4-lysosyme fusion protein into ICL3, yielding a structure at a resolution
of 3.15 Å. The overall topology is similar to that observed for the b1 and
b2AR, including the presence of an a-helix in ICL2 (as previously seen in
b1AR Warne et al., 2008). However, unlike b1 and b2AR, in the dopamine
D3 receptor, ECL2 was disordered, with no apparent secondary structure,
although the portion of ECL2 that contributes to the ligand-binding site is
orientated in a similar position relative to the bound ligand. Other notable
differences from b2AR include an outward tilting of TM6 (by 3 Å) and TM7
(by 2 Å), inward tilting of approximately 3.5 Å by TM3 and TM5 at
the extracellular face, and the presence of the “ionic lock” between Arg128
(3.50) and Glu324(6.30) (Chien et al., 2010). However, given the homology
between their endogenous ligands, it is no surprise that the ligand-binding
site is similar to that of both the b1 and the b2AR, with 10 of 18 amino acids
in the primary-binding site conserved. As expected, the tertiary amine of
eticlopride forms a salt bridge with Asp110(3.32), and the aromatic ring sits
in a hydrophobic pocket formed by Phe345(6.51), Phe346(6.52), Val189
(5.39), Ser192(5.42), Ser193(5.43), Val111(3.33), and Ile183 in ECL2
(Chien et al., 2010).
Interestingly, 17 of 18 primary contact residues are conserved in the
closely related dopamine D2 receptor subtype; this corroborates the known
difficulty in designing subtype-selective ligands for the dopamine D3 recep-
tor. To rationalize this observation, homology modeling of the dopamine D2
receptor and docking of the dopamine D3-selective antagonist ligand, R22,
into the structure (D3) and the model (D2) have been performed (Chien et al.,
2010). The data suggest that while the amine group of R22 binds to Asp110
(3.32) in the primary-binding pocket, the ligand adopts an extended confor-
mation enabling the indole-2-carboxamide to act as a second pharmaco-
phore to engage a binding site formed by ECL1, ECL3, and residues at the
top of TM1, TM2, and TM7 which are largely nonconserved between the
two receptor subtypes (Chien et al., 2010). The interaction with a second
binding site is therefore thought to yield the selectivity for the dopamine D3
receptor. This “bitopic” mode of binding has been previously suggested for
the muscarinic M2 receptor partial agonist, McN-A-343 (Valant et al.,
2008), and indeed may underlie the mechanism of action of many ligands
which display high selectivity between closely related receptor subtypes.
2. Agonist Binding
Before 2011, all the diffusible ligand costructures of Family A GPCRs
were with antagonist ligands. While there have been substantial clues from
the rhodopsin/opsin system on mechanisms of agonist binding and receptor
activation (Smith, 2010), this information has been lacking for receptors with
diffusible agonist ligands.
Early clues arose from the structure of the b1AR, where docking of
adrenaline into the cyanopindolol binding pocket suggests that due to the
Receptor Structure in Drug Design 11
agonist being smaller than the antagonist, the catechol hydroxyl groups of
the ligand would be too far from the serine residues on TM5 to form
hydrogen bonds. Therefore, receptor activation must involve a contraction
of the binding site by 2–3 Å to form the interactions necessary for agonist
binding and receptor activation (Warne et al., 2008). However, this hypoth-
esis does not explain how this conformational change may result in a 5–6 Å
movement at the base of TM6 which was shown for the rhodopsin/opsin
system (Smith, 2010) (Fig. 1B).
The next steps have now been taken in understanding the mechanism of
agonist binding and at least the first stages of conformational changes that
comprise receptor activation. One recent study has described the crystal
structure of the b2AR–T4L fusion in complex with a high-affinity agonist,
BI167107, and a nanobody that mimics the actions of the cognate G protein
(Rasmussen et al., 2011). Simultaneously, cocrystal structures of partial and
full agonists at the b1AR have been determined (Warne et al., 2011); in
addition, a crystal structure of an agonist (FAUC50) irreversibly bound to
the b2AR has been solved (Rosenbaum et al., 2011). Figure 1C and D
illustrate the differences between the b2AR agonist and antagonist bound
structures in both the binding site and whole protein, while Fig. 1E and F
depicts the corresponding b1AR structures.
The data for b1AR have been determined using the receptor construct
stabilized in the inactive conformation by mutagenesis. Nevertheless, the
construct can still bind and be activated by agonists (Warne et al., 2008),
albeit at higher concentrations than for the wild-type receptor. Costructures
have been determined for partial agonists (salbutamol and dobutamine) and
full agonists (carmeterol and isoprenaline), all at resolutions of 3 Å or less
(Warne et al., 2011). The overall topology of the receptor is relatively un-
changed compared to the cyanopindolol costructure; however, there are some
notable differences in the primary binding site (Fig. 1E). The amine moiety of
all the agonist ligands (plus the b-hydroxyl for all agonists except dobutamine)
forms interactions with Asp121(3.32) and Asn329(7.39), as would be
expected. All the agonists form a hydrogen bond with Ser211(5.42); addition-
ally, the full agonists isoprenaline and carmeterol (but not the partial agonists)
form a second hydrogen bond with Ser215(5.46) in conjunction with a change
in the rotamer conformation of Ser212(5.43) to form a hydrogen bond with
Asn310(6.55) (Warne et al., 2011). The number of polar interactions formed
by the serine residues on TM5 could represent a marker for partial versus full
agonism. These interactions cause the catecholamine binding pocket to
contract by approximately 1 Å in comparison with the cyanopindolol costruc-
ture. However, it is notable that these conformational changes do not result in
the larger scale movements of TM5 and TM6 that might be expected from an
agonist-bound structure. Given that this receptor construct is one held in an
inactive conformation, it could be that these changes represent simply the first
stage of movements that result in full receptor activation.
12 Congreve et al.
Similar results have been shown for the b2AR in complex with BI167107
and a camelid antibody fragment. The fragment, or nanobody, acts as a
surrogate for the Gs protein, shifting the receptor into a high-agonist affinity
state that has allowed the structure to be solved in complex with BI167107 at
a resolution of 3.5 Å (Rasmussen et al., 2011). Unlike the b1AR study (where
low-affinity agonists can be profiled), the b2AR–T4L system requires a high-
affinity agonist, preferably with a slow off-rate (Rasmussen et al., 2011). It is
this property that enabled the structure determination for the irreversibly
binding agonist, FAC50 (Rosenbaum et al., 2011).
When compared to the carazolol–b2AR costructure, there is a clear
outward movement of TM5 and TM6, coupled with an inward movement
of TM3 and TM7 in the b2AR structure in the presence of the nanobody and
BI167107 (Fig. 1D). The 11 Å movement at the base of TM6 is comparable
to that seen for active versus inactive rhodopsin (Smith, 2010) (Fig. 1B).
However, despite these changes, the conformational effects in the binding
pocket are more subtle and comparable to those observed in the b1AR
(Rasmussen et al., 2011; Warne et al., 2011) (Fig. 1C). The agonist ligand
binds in similar way to the antagonist, carazolol, forming interactions with
Asp113(3.32) and Asn312(7.39). In addition, there are polar interactions
with serine residues on TM5 (Ser203(5.42) and Ser207(5.46)), enabled by an
inward bulging of TM5 by approximately 2 Å and smaller movements of
TM6 and TM7. Interestingly, despite the changes in the binding site and
larger movements of TM6, there was no change in the rotamer state of the
Trp286(6.48), which has been designated as the “toggle switch” for receptor
activation.
C. Adenosine A2A Receptor
The adenosine A2A receptor is an important therapeutic target in a
number of CNS diseases such as Parkinson’s disease (Shah & Hodgson,
2010). The first structure of the A2A receptor was solved using the T4L fusion
technology (Jaakola et al., 2008). T4L was inserted between Leu209 (5.70)
and Ala221 (6.23) in the ICL3 of the receptor. As with the b2AR, the
insertion of the T4L altered the pharmacology such that agonists bound to
the receptor with a higher affinity than the wild type, whereas antagonist
affinity values were in broad agreement. A2A–T4L was crystallized using the
in meso method with the addition of cholesteryl hemisuccinate (CHS) in
complex with the inverse agonist ligand ZM241385. Despite a relatively
low homology between A2A and the b adrenergic receptors (bAR)
(20–40% in the TM regions), there is close agreement in the packing of the
helices between A2A and the adrenergic receptors (root mean square devia-
tion (RMSD) 1.8–2.5 Å). A second structure of the A2A receptor has been
solved using a thermostabilization (StaR) approach (Doré et al., 2011).
This receptor, known as A2A–StaR2, includes eight thermostabilizing
