| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Division of Cellular and Molecular Physiology, Department of Medical Physiology, The Panum Institute, University of Copenhagen, DK-2200 Copenhagen, Denmark
| Abstract |
|---|
|
|
|---|
| I. Introduction |
|---|
|
|
|---|
-subunit leading to release of GDP
followed by binding of GTP (3). Subsequently, the GTP-bound form of the
-subunit dissociates from the receptor as well as from the stable
ß
-dimer. Both the GTP-bound
-subunit and the released
ß
-dimer can modulate several cellular signaling pathways. These
include, among others, stimulation or inhibition of adenylate cyclases
and activation of phospholipases, as well as regulation of potassium
and calcium channel activity (4). The complexity of GPCR signaling has
recently been further underlined by data indicating that GPCRs may not
solely act via heterotrimeric G proteins (5, 6, 7, 8, 9, 10). Most intriguingly, it
has been suggested that agonist-promoted phosphorylation of the
receptors by GRKs (G protein-coupled receptor kinases) (11) and
subsequent sequestration of the receptors from the cell surface (11)
are not only important for turning off signaling, but also play a key
role in switching the receptor from G protein-dependent pathways to
signaling cascades normally used by growth factor receptors (5, 6, 7, 10).
Yet another example illustrating the impressive variability of GPCR
function is the observation that human immune deficiency virus (HIV)
utilizes G protein-coupled chemokine receptors as cofactors for their
cellular entry (12, 13, 14, 15). It is thus clear that extensive experimental work performed over the last decade has uncovered multiple aspects of GPCR function and challenged many traditional paradigms (reviewed in Refs. 7, 16, 17, 18, 19, 20, 21, 22). However, it is only recently that we are beginning to gain insight into some of the most fundamental questions in GPCR function. How can, for example, so many chemically diverse hormones, neurotransmitters, and other signaling molecules activate receptors believed to share a similar overall tertiary structure? What is the nature of the physical changes linking agonist binding to receptor activation and subsequent transduction of the signal to the associated G protein on the cytoplasmic side of the membrane and to other putative signaling pathways? The goal of the present review is to specifically address these questions as well as to depict the current awareness about GPCR structure-function relationships in general.
| II. Structural Classification of G Protein-Coupled Receptors |
|---|
|
|
|---|
-helical segments connected by alternating
intracellular and extracellular loops, with the amino terminus located
on the extracellular side and the carboxy terminus on the intracellular
side (Fig. 1
|
To facilitate comparison of residues between the large number of
different receptors belonging to family A there is an obvious need to
formulate and use a common numbering scheme. Currently, three different
numbering schemes have been suggested but none of them have gained any
wide acceptance. The Schwartz and Baldwin numbering schemes are, in
principle, identical (24, 25). According to both schemes, the most
conserved residue in each helix (yellow residues in Fig. 2B
and Fig. 3
) has been given a generic number describing their predicted relative
position in a standard helix of 26 residues (24, 25). A given residue
is then described by the helix in which it is located (IVII) followed
by a number indicating its position in the helix. For example, V.16
indicates residue number 16 in TM (transmembrane segment) 5. However,
the two numbering schemes are unfortunately incompatible with one
another since they do not, except in helix 1, agree on the
relative positioning of the conserved residues in the helices (24, 25).
This problem is not apparent in the Ballesteros-Weinstein numbering
scheme (26). In this scheme, the most conserved residue in each helix
has been given the number 50, and each residue is numbered according to
its position relative to this conserved residue. For example, 6.55
indicates a residue located in TM 6, five residues carboxy terminal to
Pro6.50, the most conserved residue in helix 6 (Fig. 2B
and Fig. 3
)
(26). Since there is no general agreement at this stage in the field on
which scheme to use, all residues in this review will be indicated
according to the Schwartz scheme followed by the Ballesteros-Weinstein
number in superscript.
|
|
100
residues) extracellular amino terminus containing several cysteines,
presumably forming a network of disulfide bridges (27).
Family C receptors are characterized by an exceptionally long amino
terminus (500600 amino acids) (Fig. 1
). The receptors include the
metabotropic glutamate and
-amino- butyric acid (GABA)
receptors, the calcium receptors, the vomeronasal, mammalian pheromone
receptors, and the recently identified putative taste receptors (1, 2).
Family C receptors have, like family A and B receptors, two putative
disulfide-forming cysteines in ECL 2 and ECL 3, respectively, but
otherwise they do not share any conserved residues with family A and B
receptors (Fig. 1
). The amino terminus of the metabotropic glutamate
receptors displays remote sequence homology with bacterial periplasmic
binding proteins (PBPs), especially with the leucine/isoleucine/valine
binding protein (28). The glutamate binding site has been proposed to
be equivalent to the known amino acid binding site of PBPs; therefore,
it is believed that the amino terminus of family C receptors contains
the ligand-binding site (28, 29).
| III. Structural Probing of GPCRs |
|---|
|
|
|---|
-helices and uses retinal
as its chromophore, it has been considered a bacterial homolog of
vertebrate rhodopsin. The bacteriorhodopsin structure has accordingly
been widely used as a template for tertiary structure models of GPCRs
(31, 32, 33, 34, 35). However, bacteriorhodopsin is a proton pump, is not linked to
a G protein, and does not even display remote sequence homology with
any GPCR. Moreover, the structural information that recently has become
available for rhodopsin indicated clear differences between
bacteriorhodopsin and rhodopsin (30, 36, 37, 38, 39). Overall, the use of
bacteriorhodopsin as a template for molecular models should now be
considered obsolete.
Using electron cryomicroscopy of two-dimensional crystals, Schertler
and co-workers (36, 37, 38, 39) have succeeded in obtaining low-resolution
structures of both bovine and frog rhodopsin. In addition, a
low-resolution structure of squid rhodopsin has become available (40).
The first projection map of bovine rhodopsin at 9 Å resolution
provided the first direct insight into how the predicted seven helices
are organized relative to one another in the tertiary structure of the
receptor (36). Importantly, a very similar arrangement of the
transmembrane helices was found in the projection maps of frog and
squid rhodopsin at 7 Å and 8 Å resolutions, respectively (38, 40).
The projection maps are characterized by an arc-shaped feature, which
has been interpreted as reflecting the presence of three tilted helices
(36, 38, 40). Four additional peaks were interpreted as the remaining
four transmembrane helices (36, 38, 40). The structural information
achieved from aligning multiple receptor sequences permitted assignment
of the individual peaks in the projection maps to the individual
helices in the receptor (25, 41). As shown in Fig. 2
, it is believed
that the helices are organized sequentially in a counterclockwise
fashion as seen from the extracellular side, with helix 3 being almost
in the center of the molecule. Further insight into the packing of the
seven-helix bundle and calculation of the tilting angles of the helices
have been achieved by detailed analysis of tilted two-dimensional
crystals of bovine and frog rhodopsin, allowing generation of the first
three-dimensional maps (37, 38). The resolution of the map based on the
frog rhodopsin crystals was 7.5 Å in the plane of the membrane and
16.5 Å perpendicular to it (38). According to the map, helices 1, 2,
and 3 are tilted 2730 degrees, helix 5 is tilted 23 degrees, whereas
helices 4 and 7 are almost perpendicular to the plane of the membrane
(38). Helix 6 appears almost perpendicular to the plane of the membrane
in the cytoplasmic half but is bent toward helix 5 on the extracellular
side (38). The structure also shows that the helices are tightly packed
on the intracellular side with helices 2 and 3 packed between helix 4,
6, and 7 (38). On the extracellular side the helical arrangement opens
up and forms a cavity that serves as a binding pocket for retinal. The
cavity is lined by helices 3, 4, 5, 6, and 7 and is closed toward the
intracellular side by the tilted helix 3 (38). A recent projection map
of bovine rhodopsin with an improved resolution (5 Å) suggests that
the two-dimensional crystallography technique may lead to even more
detailed understanding of the tertiary structure of GPCRs (39).
Guided by the rhodopsin projection maps and the structural information
that has been acquired from extensive analysis of multiple GPCR
sequences, several tertiary structure models of receptors belonging to
family A have been developed over the last few years (25, 26, 41, 42)
(Fig. 2C
). The models are, of course, still somewhat uncertain but they
do provide a believable general picture of the receptor structure and
thus a reliable framework within which the structure and molecular
function of GPCRs can be further debated and experimentally explored.
Importantly, a large number of experimental studies, aimed at probing
tertiary structure relations in GPCRs, have been highly critical for
refining and validating the molecular models. First of all, this
includes identification of several distance constraints in the receptor
structure. The close proximity between TM 1 and 7 has, for example,
been established based on rescue of nonfunctional adrenergic
2/ß2 receptor chimeras
and muscarinic M2/M5
chimeras (43, 44, 45, 46). An important series of helix-helix interactions have
also been identified by engineering of histidine zinc(II) binding sites
in the neurokinin 1 (NK-1) (substance P) receptor and the
-opioid
receptor (47, 48, 49). In the NK-1 receptor bis-zinc(II) binding sites were
constructed by introducing pairs of histidines in positions predicted
to be in close proximity, and in this way it was possible to define the
proximity and orientation of TM 3 relative to TM 2 and 5 (47). The
distance constraints inferred from the engineered zinc(II) binding
sites, as well as from the rescue of nonfunctional chimeras, strongly
supported a counterclockwise organization of the seven helices as seen
from the extracellular side (45, 46, 47). Additional distance constraints
in the tertiary structure of the receptors have been identified by
formation of intramolecular disulfide bridges between engineered pairs
of cysteines in rhodopsin (50, 51) and lately in the
M3 muscarinic receptor (52). Notably, the use of
biophysical techniques has also allowed insight into tertiary structure
relationships. Turcatti et al. established a system, based
on suppression of UAG nonsense codons and the use of modified tRNAs,
allowing biosynthetic introduction of a fluorescent, unnatural amino
acid at known sites in the tachykinin NK-2 receptor during heterologous
expression in Xenopus oocytes. In this way, they were able
to define a set of distances in the tertiary structure by measurement
of fluorescence resonance energy transfer between a fluorescent peptide
antagonist and different sites containing the fluorescent amino acid
(53).
In the GnRH receptor, the proximity between TM 2 and 7 was suggested
based on identification of an evolutionary reciprocal mutation (54). In
nearly all family A receptors there is a conserved aspartic acid in TM
2, AspII.102.50 (II.10 according to the Schwartz
numbering scheme, 2.50 according to the Ballesteros-Weinstein scheme),
and a conserved asparagine in TM 7 (VII.167.49)
(Fig. 3
), but in the GnRH receptor an asparagine is found in the
corresponding position in TM 2 and an aspartic acid in TM 7. Since
replacement of the asparagine in TM 2 with aspartic acid eliminated
detectable ligand binding, but high-affinity agonist binding was
restored by additional mutation of the aspartic acid in TM 2 to
asparagine, it was proposed that the two residues are in close spatial
proximity (54). The observation is not readily compatible with the
receptor model proposed by Baldwin et al. (25). In this
model the distance between the
-carbons of the two residues is 10.4
Å, which is too large for their side chains to form a direct
hydrogen-bonding interaction (25). However, if the proposed kink at
ProVII.177.50 also causes a twisting of the
helix, the two residues can be in sufficiently close proximity to form
a direct interaction (26, 55). Remarkably, the presence of both a kink
and twist in helix 7 is experimentally supported by the observed
cysteine accessibility pattern in TM 7 (55).
Applying the substituted cysteine accessibility method to the dopamine
D2 receptor has provided further highly useful
structural information about GPCRs (55, 56, 57, 58, 59). Javitch and co-workers
(55, 56, 57, 58) have systematically substituted residues in TM 2, 3, 5, 6, and
7 with cysteine and determined their accessibility in the predicted
binding crevice by reacting with charged sulfhydryl-specific
methanethiosulfonate (MTS) derivatives. Their data have allowed mapping
of residues facing the binding crevice and estimation of the relative
orientation of individual helices (55, 56, 57, 58). The accessibility patterns
were consistent with TM 2, 3, 6, and 7 forming regular
-helices
in agreement with the predictions from the rhodopsin projection maps
(55, 56, 58). In TM 6 and 7, the data also supported the presence of
kinks corresponding to the conserved prolines,
ProVI.156.50 and
ProVII.177.50, respectively (Figs. 2
and 3
) (55, 58). The accessibility pattern in TM 5 differed from that observed in
the other helices (57). A stretch of 10 consecutive residues in the
outer portion of TM 5 were found exposed in the binding crevice, which
is inconsistent with the prediction that TM 5, like the other helices,
should form a regular helix with one side exposed and one side hidden
form the crevice (57). There is no obvious explanation for this
observation. One explanation could be that the exposed stretch of
residues is nonhelical and loop out into the lumen of the binding
crevice, making all the residues accessible to the MTS reagents.
Alternatively, the outer portion of TM 5 may be structurally flexible
and rapidly shift between different conformations, exposing different
sets of residues to the binding crevice (57). In both cases, it is of
notable interest that the exposed region contains residues believed to
form key contacts with the small-molecule agonists (60).
In rhodopsin, the application of EPR (electron paramagnetic resonance)
spectroscopy has provided information about structural features,
particularly in the cytoplasmic loop regions. Consecutive residues in
the cytoplasmic loops and the carboxy-terminal tail have been
substituted with cysteine and each of the cysteine mutants was labeled
with sulfhydryl-specific nitroxide spin labels (61, 62, 63, 64, 65). By determining
the accessibility of the attached nitroxide labels to collisions with
paramagnetic probes in solution, information about aqueous/hydrophobic
boundary zones and secondary structure relations was obtained. The
accessibility pattern in the third intracellular loop connecting TM 5
and 6 provided important evidence that these two
-helices extend two
to three turns beyond the cytoplasmic surface of the membrane (62). In
the second intracellular loop connecting TM 3 and 4, the analysis
indicated that the TM 3
-helix extends at least 1.5 turns past the
important D/ERY motif (Figs. 1
and 3
) and that much of the helix
surface at the cytoplasmic side forms contacts with protein rather than
with the lipids (61). Analysis of the "fourth intracellular loop"
between the cytoplasmic end of TM 7 and the palmitoylation site
indicated that helix 7 extends around 1.5 turns beyond the membrane
surface and that the remaining part of the loop forms very strong
tertiary contacts with the protein (64). It was therefore suggested
that the loop beyond the helix may be folded over the body of
rhodopsin, allowing interactions with residues in the first loop
between TM 1 and 2 (64).
| IV. Ligand-Binding Domains |
|---|
|
|
|---|
A. Rhodopsin and the biogenic amines
1. Rhodopsin. The photochromophore of rhodopsin
and the opsins, 11-cis-retinal, is unique among the
endogenous ligands for GPCRs in that it is covalently
attached to the receptor within a binding crevice formed by the
transmembrane helices (reviewed in Ref. 66). Through formation of a
Schiff base, 11-cis-retinal is coupled to a lysine in TM
7 (Lys296, VII.107.43). The protonated Schiff base is
paired with a glutamic acid (Glu113, III.043.28) in the
outer portion of TM 3 (67). Additional interactions are found in TM 3
between the C9 group of retinal and Gly121 (III.123.36)
(68), and between retinal and aromatic residues in the outer portion of
TM 6 (69). Upon exposure to light, 11-cis-retinal
undergoes an isomerization to all-trans-retinal, which
leads to formation of the metarhodopsin II state and thus receptor
activation (66). While all-trans-retinal behaves like
the rhodopsin agonist, 11-cis-retinal behaves as an
inverse agonist (i.e., an antagonist with negative
intrinsic activity), keeping the receptor quiescent in the absence of
light (70).
2. Classical small-molecule transmitter family A receptors.
The binding sites for the classical "small-molecule" transmitters
(epinephrine, norepinephrine, dopamine, serotonin, histamine, and
acetylcholine) are contained in a binding crevice formed by the
transmembrane helices. The residues involved in binding of agonists and
antagonists to the ß2- adrenergic receptor are
found in TM 3, 5, 6, and 7 (Figs. 2C
and 3
). The binding crevice is
deeply buried in the receptor molecule as evidenced by spectroscopic
analysis of the fluorescent antagonist carazolol bound to the
ß2 -adrenergic receptor (71). The energetically
most important interaction is most likely a salt bridge between the
charged amine of adrenergic ligands and the carboxylated side chain of
Asp113 (AspIII.083.32) in TM 3 (72) (Figs. 2C
and 3
). This aspartic acid is conserved among the biogenic amine receptors
and is thought to interact also with the positively charged head group
of dopamine (73), serotonin (74, 75), histamine (76), and acetylcholine
(77). Additional key interactions of the agonists in the
ß2-adrenergic receptor include hydrogen bonding
between the hydroxyls of the catechol ring in epinephrine and two
serines one
-helical turn apart in TM 5, Ser204
(V.095.43) and Ser207
(V.125.46) (60) (Figs. 2C
and 3
). In TM 6, Phe290
(VI.176.52) may stabilize the catechol ring (78)
while recent evidence suggests that Asn293
(VI.206.55) forms a hydrogen bind with the
ß-hydroxyl of epinephrine (79) (Figs 2C
and 3
). In the case of the
ß2-adrenergic antagonists, which are
structurally related to the endogenous agonists, evidence suggests that
they share the Asp113 (III.083.32) ionic
interaction with the agonists, but that other key interactions differ.
For aryl- oxyalkylamine antagonists, such as alprenolol and
propranolol, an asparagine in TM 7 (Asn312,
VII.067.39) has been identified as a critical
interaction point (80) (Fig. 3
). Even though the majority of ligands
for small-molecule transmitter receptors seems to bind deep within the
binding crevice, there are indications that some antagonists, which
show no structural relationship with their corresponding agonist, may
partly interact with residues closer to the surface of the membrane.
For example,
1B-antagonists, such as
phentolamine and WB4101, may interact with three residues in ECL 2
immediately adjacent to the top of TM 5 (81).
B. The binding domains for peptide ligands in peptide receptors
belonging to family A
More than 50 different neuropeptides and peptide hormones have
been identified. With only a few exceptions, these peptide messengers
all act through receptors belonging to the GPCR superfamily and, at
present, more than 100 different peptide GPCRs, including subtypes,
have been identified (1). In contrast to the general picture obtained
for the small-molecule ligands, mutational mapping of ligand-binding
sites in many of the peptide receptors has demonstrated the critical
involvement of the extracellular domains for binding of the larger
peptide ligands.
1. The tachykinin system. The mammalian tachykinins include
substance P, neurokinin A, and neurokinin B, which act at the NK-1
receptor, the neurokinin-2 (NK-2) receptor, and the neurokinin-3 (NK-3)
receptor, respectively (82). In addition, a variant of the NK-3
receptor, NK-3B, has recently been identified (83). These receptors are
homologous but display significant differences in their pharmacological
profile (82, 84, 85). The initial analyses of chimeric NK-1/NK-2
receptors and NK-1/NK-3 receptors suggested that multiple epitopes
scattered throughout the receptor structures contribute to the subtype
selectivity of the tachykinin peptides and that different receptor
domains contribute in varying degrees to the receptor specificity (84, 85). This suggests that the binding sites for the tachykinin peptides
are not fully identical (85). Exchange of extracellular loop segments
between the NK-1 and NK-3 receptors revealed the involvement of the
extracellular domains in binding of the tachykinins (86, 87).
Subsequent point-mutational analysis of the NK-1 receptor identified
three residues in the amino terminus (Asn23, Glu24, and Phe25), a
residue at the top of TM 3 (His108), and a residue at the top of TM 7
(Tyr287) as putative points of interaction for substance P (Fig. 3
)
(86, 87, 88). The importance of the loop regions in substance P binding has
been directly supported by affinity cross-linking of a photolabile and
radioactively labeled substance P analog to Met181 in the third
extracellular loop (89, 90) (Fig. 3
). At present, there is no clear
evidence that substance P, like the small-molecule ligands, enters
deeply into a binding crevice formed by the transmembrane helices.
Despite extensive mutational analysis of residues facing the putative
binding crevice, no residues have convincingly been identified as
potential sites of interactions for substance P (24, 91).
Mutational analysis of neurokinin A binding to the NK-2 receptor also demonstrated evidence for interactions with residues in the extracellular domains (92, 93). However, the residues affecting neurokinin A binding in the NK-2 receptor differed partly from the residues affecting substance P binding in the NK-1 receptor (92, 93). Moreover, mutation of residues in the transmembrane regions, e.g., Leu202 (V.095.43) in the middle of TM 5, was found to affect neurokinin A binding (92, 93). Thus, neurokinin A may partially enter the transmembrane binding crevice. In agreement with the initial chimeric studies (85), these findings indicate that there may be clear differences in the binding modes even among homologous peptides acting at homologous receptors.
Mutation of four residues situated on the same face of helix 2 in the
NK-1 receptor has been reported to substantially impair the ability of
substance P to compete for binding of radiolabeled nonpeptide
antagonists (88). It was therefore initially concluded that these
residues are involved in substance P binding. However, it has later
been shown that radiolabeled substance P itself could bind with
essentially unaffected affinity to the mutated receptors (94). The most
likely explanation is that these mutations, rather than affecting the
peptide-binding site, affect the ability of the receptor to freely
interchange between distinct receptor conformations, which bind the
nonpeptide antagonist and peptide agonist with high affinity,
respectively (94). Notably, similar observations have been done
in the
-opioid receptor (95), and recently mutation of yet another
residue in the NK-1 receptor (Gly166, IV.214.65)
has been shown to affect interconversion between different receptor
states that display distinct selectivity for the tachykinin peptides
(96). These observations underline the importance of direct
determination of binding affinity or testing second messenger coupling
ability for an agonist before it is reasonable to consider whether the
effect of a mutation reflects a real interaction between the ligand and
the receptor or is due to an indirect effect.
2. Other family A peptide receptors. For the majority of receptors studied, there is evidence for major interactions in the amino terminus and predicted extracellular loop regions. This includes the receptors for angiotensin (97, 98, 99), neuropeptide Y (100), chemokines (interleukin-8) (101), vasopressin/oxytocin (102), GnRH (103), TRH (104, 105, 106), complement factor C5A (107, 108), formyl-Leu-Met-Phe (109), somatostatin (110), opioids (111, 112, 113, 114, 115), bradykinins (116), cholecystokinin/gastrin (117, 118, 119, 120, 121), and neurotensin (122). Importantly, the significance of the extracellular domains for binding of peptide ligands has been directly documented using affinity cross-linking techniques in the GnRH receptor (103), the bradykinin B2 receptor (116), and the cholecystokinin CCK-A receptor (118, 121).
Evidence indicates that some of the peptides have additional points of interactions in the transmembrane domains and therefore, to different degrees, may enter the transmembrane binding crevice. These include both the small tripeptides TRH (123, 124) and fMLP (125) and larger peptides such as angiotensin (126, 127), endothelin (128, 129, 130), somatostatin (131, 132, 133), opioids (134), and bradykinin (135). The residues identified are found in the outer portions of TM 2, 3, 5, 6, and 7. They differ considerably among the receptors and are, except in a very few cases (128, 131), different from the key positions believed to interact with the biogenic amines. However, it is remarkable to note that almost all of the residues identified appear to be on the surface of the predicted binding crevice as assessed by the cysteine accessibility method (55, 56, 57, 58, 59). This supports a high degree of structural similarity between the receptors, even though they bind chemically very different ligands.
C. The binding domains for nonpeptide ligands in peptide receptors
belonging to family A
The large group of peptide receptors represents an impressive pool
of potential drug targets; however, until recently this has been an
almost unexplored area due to the low bioavailability and metabolic
instability of the peptide ligands. It has been a long sought goal to
develop small-molecule nonpeptide compounds that are orally active and
can act at peptide receptors with high potency. The first and most
significant discovery, indicating that this would be feasible, was the
identification in the 1970s of a family of peptides, the enkephalins
and endorphins, as the endogenous ligands of the opioid receptors
(136). Until then, the only known ligands for the opioid receptors were
nonpeptide exogenous compounds, such as morphine and naloxone. The
finding directly showed that small nonpeptide compounds can act with
high affinity at peptide receptors both as agonists and antagonists. It
is only within recent years, however, that high-affinity nonpeptide
compounds have been discovered for an increasing number of peptide
receptors and changed the peptide receptor field into a rapidly
expanding area for drug development (24). The majority of the
nonpeptide compounds [mostly antagonists but recently, in some cases,
also agonists (137)] are developed into high-affinity compounds from
"leads" identified by screening of large chemical files (24). In
almost all cases, the resulting compounds exhibit no obvious structural
similarity to the endogenous peptide ligands, despite an apparent
classical competitive mode of action and despite the ability of both
the peptide agonist and nonpeptide antagonists to bind with often
subnanomolar affinity to the same receptor (24). Interestingly, these
nonpeptide compounds have turned out to be valuable for understanding
the molecular function of GPCRs.
1. Tachykinin nonpeptide antagonists. An initial series of chimeric NK-1/NK-3 receptors provided the first evidence that the binding mode for the prototype nonpeptide NK-1 receptor antagonist, CP 96,345, was distinct from the binding mode of the endogenous agonist substance P (138). Several chimeric exchanges that dramatically affected CP 96,345 affinity did not affect binding of substance P (138). Overall, the chimeric analyses indicated that CP 96,345 and several other structurally distinct nonpeptide NK-1 receptor antagonists, but not substance P itself, interact in different ways with a domain located around TM 5 and 6 (138, 139). Moreover, data from a series of NK-1/NK-2 receptor chimeras indicated that SR 48,968, an NK-2 receptor-selective nonpeptide antagonist, has critical interactions in the same region of the NK-2 receptor (140). The different binding modes of the nonpeptide antagonists and the peptide agonists have also been supported by comparing fluorescent analogs of substance P and CP 96,345 bound to the NK-1 receptor. Most significantly, it was found that while the environment surrounding the nonpeptide antagonist was highly hydrophobic and inaccessible to hydrophilic quenchers, the peptide was directly exposed to the solvent (141).
Comprehensive point-mutational analysis has further defined the
nonpeptide antagonist binding site in the NK-1 receptor (Fig. 3
). The
residues predicted to be involved in nonpeptide antagonist binding are
located in a transmembrane crevice lined by TM 3, 5, and 6 (91, 142, 143, 144, 145, 146), although interactions for some compounds also may occur in
TM 4 (147) and 7 (88). The most well documented putative direct
interactions of the prototype compound CP 96,345 are Gln165
(IV.204.64) (147), His197
(V.055.39) (142), His265
(VI.176.52) (143, 144), Phe268
(VI.206.55) (91, 146), and Tyr272
(VI.246.59) (145). It should be emphasized here
that it is highly difficult with mutational analysis techniques to
distinguish direct interactions between the ligand and the receptor
from indirect structural effects caused by the mutation. For example,
mutation or deletion of Lys193 (V.015.35) and
Glu194 (V.025.36) substantially affect CP 96,345
binding affinity (145) (Fig. 3
). It is nevertheless unlikely that they
participate in a direct interaction since they can be interchanged
without affecting CP 96,345 affinity (145). Conceivably, these two
residues form a salt bridge that stabilizes the CP 96,345 binding
pocket (145). Two other residues, Val116
(III.123.36) and Ile290
(VII.057.38), which are nonconserved between the
human and rat receptor, have been shown to be responsible for the
species selectivity of CP 96,345 and three other structurally distinct
nonpeptide antagonists (148, 149, 150) (Fig. 3
). It was concluded that these
two residues indirectly affected the geometry of a common binding
crevice for nonpeptide ligands (148, 149, 150). However, Val116 would be
predicted to face the binding pocket and could, in fact, be involved in
a direct interaction with CP 96,345 (Fig. 3
).
Summarized, the studies on the tachykinin receptors suggest the presence of a small-molecule binding pocket, similar to the binding pocket found in the biogenic amine receptors, where structurally distinct nonpeptide compounds can be accommodated through distinct sets of interactions. Surprisingly, this binding pocket is most likely not occupied by substance P, and thus an actual overlap in the binding sites is not required for a competitive mode of action of the nonpeptide antagonists.
2. Nonpeptide ligands for other family A peptide receptors. Considerable differences in binding modes between nonpeptide antagonists and endogenous peptide agonists have been demonstrated in other peptide receptor systems as well. These include the angiotensin (97, 151, 152), opioid (153, 154), CCK/gastrin (155, 156, 157), neurotensin (122, 158), and endothelin systems (159). The general conclusions emerging are similar to the ones from the studies of the tachykinin system; hence, the small-molecule nonpeptide compounds interact with residues in the transmembrane binding crevice and, in most cases, there is no evidence that these residues are overlapping with peptide agonist binding. In the neuropeptide Y system, however, there is evidence for several overlapping contact points in the binding site for the peptide agonist and the first available nonpeptide antagonist of the Y1 receptor, BIBP 3226 (160). Similarly, nonpeptide antagonists of the endothelin ET-A and ET-B receptors may share interactions with the endothelin peptides in TM 2 and TM 3 (128, 130). On the other hand, thorough mutagenesis of 18 amino acids in the predicted transmembrane binding crevice of the ET-A receptor revealed no indication of other overlapping contact points between the nonpeptide antagonist bosentan and endothelin-1 (159).
Nonpeptide agonists have recently been discovered for the angiotensin receptors. The nonpeptide agonists of the AT-1 angiotensin receptor were found among a series of biphenylimidazole antagonists, of which some turned out to possess agonistic properties (161). Surprisingly, it appeared that the binding mode of the biphenylimidazole agonist differed both from the binding mode of the peptide agonist angiotensin, as well as that of the structurally related biphenylimidazole antagonists (161). Mutations in TM 3 and 7, known to severely affect binding of biphenylimidazole antagonists, did not affect binding of the biphenylimidazole agonist. Moreover, binding of the biphenylimidazole agonist was also unaffected by mutation of residues in the extracellular domains known to affect binding of the peptide agonist angiotensin (161).
D. Ligand-binding sites in other family A receptors
While the binding sites for eicosanoids (leukotrienes and
prostanoids) and purines mainly are contained within the transmembrane
binding crevice (reviewed in Ref. 21), high-affinity binding of
glycoprotein hormones such as LH/CG, FSH, and TSH to their receptors
occurs in the large extracellular amino terminus that characterizes
this receptor subgroup (21, 162, 163, 164, 165, 166, 167). It is believed that after the
initial binding to the extracellular domain, the amino-terminal part of
the hormone undergoes a conformational change leading to secondary
contacts with the extracellular loop regions of the membrane-associated
part of the receptor and to subsequent receptor activation (21).
The protease-activated thrombin receptors also belong to family A (168, 169). The unique activation mechanism of the thrombin receptor involves cleavage of the amino-terminal segment by thrombin (168). The resulting 33-amino acid amino terminus subsequently acts as tethered peptide ligand, which, through interactions with the extracellular loop regions of the receptor, is able to activate the receptor (170, 171).
E. Ligand-binding domains in family B receptors
Similar to peptide receptors belonging to family A, the binding
sites for peptide ligands in family B receptors involve the
extracellular domains. The large amino terminus that characterizes
family B receptors seems to play a key role for most ligands, including
secretin, VIP, pituitary adenylate cyclase-activating polypeptide
(PACAP), glucagon, glucagon-like peptide-1, PTH, and CRF (172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182).
The amino terminus is not sufficient for binding of these ligands, and
additional interactions are found in the extracellular loops (173, 175, 178, 180, 183, 184, 185, 186, 187). However, there is at present no evidence that any
of the peptides have interactions deep in a transmembrane binding
pocket. Generally, nonpeptide antagonists are still not available for
type B receptors. One exception is the CRF receptor for which a few
nonpeptide compounds have been recently developed (185, 188). Clear
evidence has already been obtained that these may bind very distinctly
from the peptide and penetrate into a transmembrane-binding crevice
(185).
F. Ligand-binding domains in family C receptors
In the metabotropic glutamate and GABA receptors, the
ligand-binding sites are contained within the large extracellular
domain characterizing family C receptors, thereby clearly
distinguishing this subclass from the biogenic amine family A receptors
(28, 189, 190). The calcium-binding site in the calcium-sensing
receptors is also found in the large amino terminus (reviewed in Ref.
191). The extracellular amino terminus of the metabotropic glutamate
receptors shares remote structural similarity with bacterial
periplasmic amino acid-binding proteins (28, 29). A high-resolution
x-ray structure of the extracellular glutamate-binding domain of an
ionotropic glutamate receptor has recently been published (192). This
structure represents the first x-ray structure of a neurotransmitter
receptor-binding domain. Based on the x-ray structure, a mechanism was
proposed for the propagation of the activation signal in the ionotropic
receptors after agonist binding (192). Whether a similar mechanism also
accounts for how the signal in metabotropic receptors is transmitted
from the extracellular domain to the receptor core region remains
elusive.
| V. Molecular Mechanisms Involved in Activation of GPCRs |
|---|
|
|
|---|
1- and
ß1-adrenergic receptors, in serum from patients
with malignant hypertension and idiopathic dilated cardiomyopathy,
respectively (194, 195). The apparent ability of these antibodies to
induce receptor activation represents an intriguing example that even
in the small-molecule biogenic amine receptors, docking of an
activating ligand in the transmembrane-binding crevice is not a
prerequisite for ligand-induced receptor activation. Additional
examples of activating antibodies, such as monoclonal antibodies
against the muscarinic receptors (196) and the bradykinin B2 receptor
(197), as well as autoantibodies directed against the extracellular
domains of the TSH receptor in Graves disease (198), also provide
strong evidence that there are multiple ways of activating GPCRs. It is
still most likely, nevertheless, that the underlying fundamental
mechanisms of activation for GPCRs have been conserved during evolution
given the ability of the receptors to activate the same intracellular
signaling pathways through the same classes of G proteins. In the
following section, our current insight into these mechanisms will be
discussed.
A. GPCRs are kept silent by constraining intramolecular
interactions
An important discovery has been the observation that many GPCRs
have a certain basal activity and thus can activate the G protein in
the absence of agonists (199, 200, 201). Interestingly, it has also been
encountered that discrete mutations are able to dramatically increase
this constitutive agonist-independent receptor activity (42, 202, 203, 204, 205).
The majority of the constitutively activating mutations were initially
identified after mutational substitutions in the C-terminal part of the
third intracellular loop of adrenergic receptors (202, 203, 204, 205), but
currently activating mutations have been identified in almost any
receptor domain in an increasing number of receptors (representative
examples in Refs. 42, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218). In a few cases, activating
mutations have been found even in the exterior part of the receptors,
such as the second extracellular loop of the TSH receptor (214) and the
third extracellular loop of the thrombin receptor (213). In the
ß2-adrenergic receptor constitutive activation
has been observed in a chimeric construct where ECL2 was substituted
with the corresponding loop of the
1B-receptor
(219). Of interest, some constitutively active mutations have arisen
naturally and have been linked to genetic diseases. This includes
mutations in the TSH receptor associated with hereditary thyroid
adenomas (208, 211, 214); mutations in the LH receptor leading to male
precocious puberty (209); and mutations in rhodopsin associated with
development of retinitis pigmentosa (210).
A crucial clue about the molecular mechanisms underlying constitutive
receptor activation came from a study carried out by Lefkowitz and
co-workers in which the naturally occurring Ala293
(VI.06.34) residue in the C-terminal part of
third intracellular loop of the
1b-adrenergic
receptor was substituted with all other possible residues. They found
that substitution of the alanine with any other residue resulted
in higher agonist-independent receptor activity (203). This led to the
suggestion that constraining intramolecular interactions have been
conserved during evolution to maintain the receptor preferentially in
an inactive conformation in the absence of agonist. Conceivably, these
inactivating constraints could be released as a part of the receptor
activation mechanism, either after agonist binding or due to specific
mutations, causing key sequences to be exposed to G protein. The
hypothesis has been indirectly supported by the recent observation that
constitutively activated ß2-adrenergic receptor
mutants are characterized by a marked structural instability and
enhanced conformational flexibility of the purified receptor proteins
(218, 220). The data imply that the mutational changes have disrupted
important stabilizing intramolecular interactions in the tertiary
structure, allowing the receptor to undergo conversion more readily
between its inactive and active state (218, 220).
Experiments performed in other receptors have also indicated that
constraining intramolecular interactions have been conserved during
evolution to keep the receptors preferentially silent in the absence of
agonists. Hsueh and colleagues (221) obtained evidence using a series
of chimeric LH/FSH receptors that stabilizing interactions between TM 5
and 6 are critical for the resistance of the FSH receptor to
constitutively activating mutations. A stabilizing role of TM 6 has
also been suggested from a random mutagenesis study in the muscarinic
M5 receptor where substitutions on one face of the helix conveyed
constitutive activity to the receptor (222). Similarly, mutation of
polar residues in TM 6 of the
-factor pheromone receptor (STE2p)
conveyed constitutive activation to this receptor (223). Molecular
modeling and analysis of naturally occurring activating mutations in
the LH receptor also strongly point to the importance of the helical
packing of TM 6 for maintaining the receptor in an inactive
configuration (224). In rhodopsin there is evidence suggesting that
opsin, the apoprotein form of rhodopsin, is maintained in an inactive
configuration by interactions between a methionine in TM 6 (Met257,
VI.056.40) and the conserved NPXXY motif in TM 7
(225), as well as by a salt bridge between Lys296
(VII.107.43) (the retinal attachment site in TM
7) and Glu113 (GluIII.043.28) (the Schiff base
counterion in TM 3) (206). Stabilizing interactions between TM 3 and 7
have also been suggested in the angiotensin AT-1 receptor between
Asn111 (III.113.35) and Tyr292
(VII.107.43) (226), and in the
1b-adrenergic receptor between the conserved
aspartic acid (Asp125, III.083.32) and Lys331
(VII.037.36) (212).
B. Protonation is a key element in GPCR activation
If receptor activation involves disruption of stabilizing
intramolecular interaction, an obvious question is how this may be
initiated after agonist binding. At present, this question cannot be
fully answered; however, substantial evidence suggests that at least
one of the key events in the activation process among family A GPCRs
involves protonation of the aspartic acid in the highly conserved D/E
RY (Glu/Asp-Arg-Tyr) motif at the cytoplasmic side of TM 3 (Fig. 3
).
The most direct evidence has been obtained by Sakmar and co-workers
(227) who compared wild-type rhodopsin and rhodopsin mutated in
position Glu134 (III.253.49) by flash photolysis,
allowing simultaneous measurement of photoproduct formation and rates
of pH changes. Their data strongly suggested that proton uptake of
Glu134 (III.253.49) accompanies formation of the
metarhodopsin II state (227). The "protonation hypothesis" has been
further supported by the observation that charge-neutralizing
mutations, which mimics the unprotonated state of the aspartic
acid/glutamic acid, cause dramatic constitutive activation of both the
adrenergic
1b-receptor and the
ß2-adrenergic receptor (42, 218, 228).
Similarly, improved coupling has been observed by mutation of the
aspartic acid in the GnRH receptor (229). Mutation of the aspartic
residue in the M1 muscarinic receptor resulted in
phosphoinositide turnover responses of the mutant that were
quantitatively similar to the wild-type despite markedly lowered levels
of expression (230). In parallel, constitutive activation was observed
in rhodopsin after mutation of the glutamic acid found in the
corresponding position of this receptor (231). Finally, it was found
that charge-neutralizing mutations of the aspartic acid (Asp130;
III.253.49) in the
ß2-adrenergic receptor are linked to the
overall conformation of the receptor (218). Thus, mutation of Asp130 to
asparagine did not only activate the receptor but also caused a
cysteine in TM 6 (Cys285, VI.126.47), which is
not accessible in the wild-type receptor, to become accessible to
methanethiosulfonate ethylammonium (MTSEA), a charged,
sulfhydryl-reactive reagent (218). This observation is consistent with
a counterclockwise rotation (as seen from the extracellular side) or
tilting of TM 6 in the mutant receptor. Importantly, this
conformational rearrangement is identical to the movement of TM 6,
which biophysical studies have indicated to be essential for
agonist-induced receptor activation (see next section).
The experimental data have been supported by molecular modeling and
computational simulations. Two distinct hypotheses have been proposed
to define the specific role of Asp/GluIII.253.49
protonation in receptor function, the "polar pocket" hypothesis
proposed by Scheer et al. (42) and the "arginine cage"
hypothesis proposed by Ballesteros et al. (229). According
to the polar pocket hypothesis, the invariably conserved
ArgIII.263.50 is in the inactive state of the
receptor constrained in a pocket formed by conserved polar residues in
TM 1, 2, and 7, including AsnI.181.50,
AspII.102.50,
AsnVII.167.49, and
TyrVII.197.52 (Fig. 3
). Upon protonation (or
mutation to alanine) of the adjacent
AspIII.253.49, the simulation indicated that the
arginine shifts out of the polar pocket leading to long-range
conformational changes in the receptor molecule (228). In their model,
they highlighted that the ionic counterpart of the arginine in the
inactive receptor state was the conserved aspartic acid in TM 2
(AspII.102.50, Fig. 3
), and that this interaction
is broken after receptor activation (228). Alternatively, based on
computational simulations in the GnRH receptor, the arginine-cage
hypothesis suggests that the ionic counterpart of
ArgIII.263.50 in the inactive state of the
receptor could be the adjacent AspIII.253.49 and
not AspII.102.50 (229). It was hypothesized that
during receptor activation, AspIII.253.49 becomes
protonated and that AspII.102.50 substitutes for
AspIII.253.49 in forming an ionic interaction
with ArgIII.263.50 (229). Thus, an ionic
interaction between ArgIII.263.50 and
AspII.102.50 was associated with the active
receptor state instead of with the inactive state as proposed by Scheer
et al. (42). An indirect support for this alternative
hypothesis is the observation in several GPCRs that mutations, which
eliminate the charged character of AspII.102.50
and in this way conceivably destabilize the Asp-Arg interaction, also
disturb functional coupling of the receptor (232, 233, 234, 235, 236). Spectroscopic
experiments in rhodopsin have also indicated that
AspII.102.50 is more strongly hydrogen bonded
upon activation, consistent with its potential interaction with another
residue in the active state of the receptor (237).
C. Conformational changes involved in receptor activation
An ultimate understanding of the receptor activation mechanism
requires development of techniques that can provide insight into the
character of the physical changes accompanying transition of the
receptor from the inactive to the active state. Sheikh et
al. (238) have undertaken an approach where bis-histidine metal
ion-binding sites were generated between the cytoplasmic extensions of
TM 3 and 6 in rhodopsin. In this way, they were able to show that
cross-linking pairs of histidines with Zn2+
prevented transducin activation, providing indirect evidence that
movements of these two domains are important for activation. Recently,
they have obtained similar results in the
ß2-adrenergic receptor and in the PTH receptor
of which the latter belongs to family B (Fig. 1
) (239). This suggests
that the activation mechanism may be conserved among both family A and
family B receptors (239). Javitch and co-workers (240) have applied the
substituted cysteine accessibility method, in which specific advantage
was taken of a constitutively activated ß2
adrenergic receptor, CAM. Their main observation in CAM was that
a cysteine in TM 6 became accessible in the binding crevice to a
charged, sulfhydryl-reactive reagent (240). This indicated a
conformational rearrangement of TM 6 with CAM consistent with a
counterclockwise rotation or tilting of the helix (240). Assuming that
the conformation of CAM mimics the agonist-activated state of the
receptor, the data thus indicated that movements of TM 6 are a critical
element in the receptor activation mechanism.
Recently, biophysical techniques have also been implemented, allowing direct time-resolved analysis of conformational changes in the receptor molecule. It is not surprising that a majority of the studies initially have been carried out in rhodopsin. There are abundant natural sources of rhodopsin, and the inherent stability of the rhodopsin molecule makes it possible to produce and purify relatively large quantities of recombinant protein. Accordingly, several spectroscopic techniques have been applied to rhodopsin, including Fourier transform infrared resonance spectroscopy (FTIR) (241, 242), surface plasmon resonance (SPR) spectroscopy (243), tryptophan UV-absorbance spectroscopy (244), and EPR spectroscopy (61, 62, 64, 65). All approaches have consistently provided evidence for a significant conformational rearrangement accompanying transition of rhodopsin to metarhodopsin II. Using tryptophan UV-absorbance spectroscopy, Lin and Sakmar (244) were able to obtain the first direct evidence that photoactivation may involve relative movements of TM 3 and 6 (244). Thus, mutation of tryptophans in TM 3 and 6 eliminated the spectral differences in the UV absorbance spectra that distinguished rhodopsin from metarhodopsin II (244).
In a series of very elegant studies, carried out by Khorana, Hubbell,
and co-workers (50, 61, 62, 63, 64, 65, 245), the use of EPR spectroscopy in
combination with multiple cysteine substitutions has led to further
insight into the character of conformational changes accompanying
photoactivation of rhodopsin. Site-directed labeling of single
cysteines inserted at the cytoplasmic side of the transmembrane helices
with sulfhydryl-specific nitroxide spin labels provided evidence for
movements particularly of the cytoplasmic termination of TM 6 upon
light-induced activation of rhodopsin (50, 61, 62, 63, 64, 65). The spectroscopic
analyses also showed evidence for smaller movements in the loop
connecting TM 1 and 2 as well as at the cytoplasmic ends of TM 3 and TM
7 (61, 64, 246). Only minor or no structural changes appeared to occur
at the cytoplasmic end of TM 4 and 5 (61, 62). To investigate the
character of the conformational changes, Khorana, Hubbell, and
co-workers have taken advantage of the magnetic dipole interaction
between two nitroxide spin labels causing spectral line broadening if
the two probes are less than 25 Å apart (50). Pairs of
sulfhydryl-reactive spin labels were incorporated into a series of
double-cysteine mutants enabling measurement of changes in relative
distance between TM 3 and TM 6 (50). While the movement of TM 3 was
interpreted as relatively small, the data pointed to a significant
rigid-body movement of TM 6 in a counterclockwise direction (as viewed
from the extracellular side) and a movement of the cytoplasmic end of
TM 6 away from TM 3 (Fig. 4
) (50). Importantly, movements of TM 6 in rhodopsin upon photoactivation
have recently been additionally documented by site-selective
fluorescent labeling of cysteines inserted at the cytoplasmic
termination of the helix (247).
|
To identify the cysteines labeled with IANBD that gave rise to the
spectral changes, a series of mutant ß2
receptors with one, two, or three of the natural cysteines available
for fluorescent labeling was generated (249). The fluorescence
spectroscopy analysis of the purified and site-selectively labeled
mutants showed that IANBD bound to Cys125
(III.203.44) in TM 3 and Cys285
(VI.126.47) in TM 6 were responsible for the
observed changes in fluorescence (249). This suggests that movements of
TM 3 and 6 may occur during receptor activation (249) (Fig. 4
). The
possible spatial orientation of IANBD bound to Cys125
(III.203.44) and Cys285
(VI.126.47) in TM 6 was explored in a series of
computational simulations to define the character of the putative
movements of TM 3 and 6. In a rhodopsin-based model of the
ß2receptor, the preferred conformation of
IANBD attached to Cys125 (III.203.44), as defined
by the computational simulations, is bounded by the lipid bilayer and
the interface of TM 3 and TM 4, while the IANBD attached to Cys285
(VI.126.47) is predicted to be at the helix 6-7
interface in a boundary zone between the lipid bilayer and the more
polar interior of the protein (249) (Fig. 4