| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, March 1998, p. 1087-1100, Vol. 74, No. 3
*Department of Physiology and Biophysics, Mount Sinai School of Medicine, New York, New York 10029, and #Division of Molecular Medicine, Department of Medicine, Cornell University Medical College and the New York Hospital, New York, New York 10021 USA
 |
ABSTRACT |
|---|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
Small ligands generally bind within the seven transmembrane-spanning helices of G-protein-coupled receptors, but their access to the binding pocket through the closely packed loops has not been elucidated. In this work, a model of the extracellular loops of the thyrotropin-releasing hormone (TRH) receptor (TRHR) was constructed, and molecular dynamics simulations and quasi-harmonic analysis have been performed to study the static and dynamic roles of the extracellular domain. The static analysis based on curvature and electrostatic potential on the surface of TRHR suggests the formation of an initial recognition site between TRH and the surface of its receptor. These results are supported by experimental evidence. A quasi-harmonic analysis of the vibrations of the extracellular loops suggest that the low-frequency motions of the loops will aid the ligand to access its transmembrane binding pocket. We suggest that all small ligands may bind sequentially to the transmembrane pocket by first interacting with the surface binding site and then may be guided into the transmembrane binding pocket by fluctuations in the extracellular loops.
| |
INTRODUCTION |
|---|
|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
The thyrotropin releasing hormone receptor (TRHR)
is a member of a large family of transmembrane proteins (GPCR) for
which an interaction with an intracellular G protein is a critical part of the signal transduction pathway mediated by the receptor. It is
thought that all GPCRs have a common tertiary structure composed of
seven transmembrane helices. The topology of these membrane-bound proteins is defined by an extracellular amino terminus and an intracellular carboxy terminus. Consequently, the helices are connected
by three intracellular and three extracellular loops. As is the case
for most integral membrane proteins, and especially for GPCRs, the
three-dimensional structure of TRHR is not known. To gain understanding
about the relationship between structure and function of TRHR, we
developed a combined theoretical/experimental approach. The structural
models of the receptor are constructed and studied by theoretical
simulation methods (Laakkonen et al., 1996
), and the
structural/functional inferences are tested by experimental molecular
biological approaches (Gershengorn and Osman, 1996; Perlman et al.,
1994a,b, 1995a, 1996).
In a previous work (Laakkonen et al., 1996), we presented a de novo
model for the transmembrane domain of the receptor that was constructed
based on a generic template for GPCRs developed from an analysis of
homologous GPCR sequences (Baldwin, 1993). The template consisted of
helical axes, which provided guidelines for positioning, tilting, and
orienting the helices in a transmembrane bundle. Guided by mutagenesis
and simulation studies (Laakkonen et al., 1996; Perlman et al.,
1994a,b, 1995a, 1996), we suggested that the binding pocket of TRH
(p-Glu-His-ProNH2) consists at least of four residues that
lie within the upper half of the transmembrane core. The four residues
that were previously identified as part of the TRH binding pocket are
Tyr106 and Asn110 in helix 3, Tyr282 in helix 6, and Arg306 in helix 7, and they form specific interactions with TRH. Thus, Tyr106 forms a
H-bond to the C
O of p-Glu, and the dipole of Asn110 is antiparallel
to the CO-NH dipole of p-Glu, giving rise to a stabilizing
dipole-dipole interaction. Tyr282 in helix 6 forms a stacking or
hydrophobic interaction with the imidazole of the His residue of TRH,
and Arg306 in helix 7 forms an ionic H-bond with the CO of the
prolineamide group and the backbone of TRH. Novel mixed-mode Monte
Carlo/stochastic dynamics simulations were used to test the validity of
the complex between the hormone and the receptor and to investigate the
nature of the hormone-receptor interactions (Laakkonen et al., 1996).
An extracellular view of the space-filling model of the occupied receptor described in the previous work (Laakkonen et al., 1996) shows
that the ligand is fully buried. This raises the question of the
mechanism of TRH access to the transmembrane binding pocket, and
especially the role of the extracellular loops in occluding or helping
the access of the ligand into the binding pocket.
Recent experimental work supports the hypothesis that the extracellular
domain of the TRHR plays a role in ligand access to the binding pocket.
For example, mutations of various residues in the extracellular loops
reduced apparent binding of TRH by a significant amount (Han and
Tashjian, 1995a,b). In particular, mutation of N289 to Ala or Asp in
extracellular loop 3 reduced the affinity to TRH by approximately 10- or 100-fold, respectively. The fact that affinity of the N289D mutant
to the Pro1TRH derivative increased (>10-fold) whereas
that of the N289A mutant did not change compared with the wild-type
receptor led the authors to conclude that the interaction between p-Glu
and N289 is specific (Han and Tashjian, 1995b). Our observations that the binding pocket is entirely positioned in the transmembrane domain
(see above) and the results presented by Han and Tashjian suggest that
TRH initially interacts with residues in the extracellular loops (e.g.,
Asn289) and subsequently moves into the transmembrane binding pocket.
Recent experiments to determine the rate constants for hormone binding
(Perlman et al., 1998) supported by simulations (Colson et al., 1997)
are in agreement with this suggestion. Such a sequential binding model
rationalizes the experimental findings and highlights the importance of
the extracellular domain in molecular models of the TRH receptor, and
possibly in GPCRs in general.
In an effort to elucidate the mechanisms by which TRH accesses its
transmembrane binding pocket and the potential involvement of the
extracellular domains in such mechanisms, we have constructed a model
that encompasses the seven transmembrane helices and the three
extracellular loops of TRHR. Despite the rapidly increasing number of
molecular models of GPCRs and a systematic approach to the construction
of the transmembrane domain (Ballesteros and Weinstein, 1995), only few
models have included the intra- and extracellular loops (Dahl et al.,
1991; Findlay and Eliopoulos, 1990; Kyle et al., 1994; MaloneyHuss and
Lybrand, 1992; Sylte et al., 1996). None of these works addressed
critically the difficulties involved in the construction of the loops
in the absence of structural guidelines. Consequently, the constructed
models of the loops were not described in detail, nor was their
relevance to the binding process critically evaluated.
We present results that demonstrate the validity of the construction model. Results from annealings and long (1-ns) molecular dynamics (MD) simulations show the possible formation of a surface binding pocket, the properties of which are complementary to TRH. Finally, a quasi-harmonic analysis of trajectories from long simulations suggests the possible importance of anti-correlated conformational motions of the TRHR extracellular loops in hormone binding.
| |
MATERIALS AND METHODS |
|---|
|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
Construction of the extracellular loops
All calculations were performed using the CHARMM program version
23 (Brooks et al., 1983b). The calculations were carried out with the
all-atom parameter set par_all22_prot. The SHAKE algorithm was employed
to fix all bonds to hydrogen atoms, and the environment was represented
by a distance-dependent dielectric function.
The model receptor studied in this work consists of the seven
transmembrane helices for which the structure was obtained from previous work performed in our laboratory (Laakkonen et al., 1996) and
the three extracellular loops of which the amino acid sequence and
two-dimensional topology was described previously (Straub et al.,
1990). No consideration was made of the amino-terminal domain, because
previous experimental results have shown that most of this domain is
not essential for agonist binding (Han and Tashjian, 1995a). As the
ultimate goal of this work is to study the conformation and mobility of
the extracellular loops in relation to the ligand's access to the
transmembrane binding pocket, the seven helix bundle remained frozen in
the geometry obtained previously (Laakkonen et al., 1996).
Simulations of the relaxed system are ongoing in our laboratory.
The three extracellular loops shown in Fig.
1 were constructed as four fragments:
D85-L99 (EC1), F161-C179 (EC2A), G180-S189 (EC2B), and N289-E298 (EC3),
so that the disulfide bond between the conserved cysteines of loops 1 and 2, i.e., C98 and C179, was maintained throughout all simulations.
This bridge has been shown to be necessary in maintaining TRHR in a
high-affinity conformation (Perlman et al., 1995b). The loops were then
attached at one end to their respective target helix via a
trans peptide bond. The backbone 
and 
torsion angles
of the loops were manually rotated so that the free end of each loop
came into proximity with the other respective target helix. Adopted
Basis Newton Raphson (ABNR) minimization was performed on the initial
structure to establish a proper peptide bond between the carboxy end of
the loop and the amino terminus of the helix. Dihedral constraints were
applied to the 
angles of the loop-helix junctions to maintain the
peptide bond in a trans orientation. Once appropriate
peptide bond lengths were obtained at these junctions, the
unconstrained loops were minimized for 1000 steps.
|
Simulated annealings and molecular dynamics simulations
The minimized structure was heated to 1500 K in 14 ps followed by 7 ps of constant temperature MD simulation at 1500 K. Fourteen structures were extracted from the trajectory at 1500 K by sampling every 0.5 ps. Each structure was cooled down to 300 K in 60 ps and subsequently subjected to 100 ps of constant-temperature MD simulation at 300 K.
Fourteen energy-minimized averaged structures were obtained over the
stabilized portion of each trajectory and clustered according to their
pairwise root mean square deviations into conformational families
employing the program Xcluster (Shenkin and McDonald, 1994). One of the
structures from the major family obtained in the clustering procedure
was employed as the starting structure for a 1-ns MD simulation. The
1-ns simulation was performed at 300 K applying the SHAKE algorithm to
all hydrogen atoms. The step size was 1.0 fs, and the coordinates were
recorded every 0.1 ps. The nonbonded lists were generated using an
atom-based cutoff of 13.0 Å and updated every 25 steps.
Quasi-harmonic analysis
Normal mode analysis of the root mean square positional
fluctuations of atoms in proteins have been found to correlate well with experimentally observed properties (Brooks and Karplus, 1983a, Go
et al., 1983). The inherent limitations of normal mode analysis of a
nonharmonic force field can be overcome by the use of quasi-harmonic models (Teeter and Case, 1990) in which effective vibrational modes are
computed such that the second moments of the amplitude distribution
matches those found in a MD simulation using the complete anharmonic
force field. The resultant quasi-normal modes are a quadratic
approximation to the potential of mean force instead of the potential
energy. In the present work, the quasi-harmonic analysis was performed
with CHARMM on a 1-ns trajectory of TRHR in which the transmembrane
helices were held at fixed positions. In an effort to reduce the size
of the matrices to be calculated in the quasi-harmonic analysis, the
energy-minimized average structure of this simulation was represented
by the coordinates of the center of mass of each residue. This
structure served as the reference to orient the reduced trajectory of
the coordinates of the centers of mass. The quasi-harmonic analysis was
subsequently performed and resulted in 162 effective vibrational modes
that represent the fluctuations of 54 residues of the extracellular
loops.
Excursion distances for each center of mass were determined from the trajectory projected on the first four quasi-normal modes using Quanta (Molecular Simulations, University of York, York, UK). They represent the maximal distances traveled by each center of mass within the same effective vibrational mode.
Materials
[3H]Methyl (Me)TRH was obtained from DuPont (Wilmington, DE). myo-[3H]Inositol was obtained from Amersham Corp. (Arlington Heights, IL). TRH was from Calbiochem (La Jolla, CA) and MeTRH from Sigma Chemical Co. (St. Louis, MO). Restriction endonucleases were from New England Biolabs (Beverly, MA). The cloning vector pBluescript was from Stratagene (La Jolla, CA), and the expression vector pCDM8 from Invitrogen (San Diego, CA). Dulbecco's modified Eagle's medium and fetal calf serum were from Collaborative Research (Bedford, MA).
Cell culture and transfection
COS-1 cells were maintained and transfected as described
previously (Straub et al., 1990). In brief, cells were seeded 1 or 2 days before transfection at 0.7 × 106 to 1.5 × 106 cells/100-mm dish. Cells were transfected using the
DEAE-dextran method as described and maintained in Dulbecco's modified
Eagle's medium with 10% fetal calf serum for 1 day at which time
cells were harvested and seeded into 12-well plates at 105
cells/well in Dulbecco's modified Eagle's medium with 5% fetal calf
serum.
Mutagenesis
The full-length mouse TRHR cDNA in pBluescript (pBSmTRHR)
(Straub et al., 1990) or pCDM8 (pCDM8mTRHR) (Gershengorn and Thaw, 1991) was used for mutation. The polymerase chain reaction was used to
generate fragments containing the R185A mutation, which was subcloned
into pBSmTRHR. A fragment derived from digestion with XhoI
and NotI was then subcloned into pCDM8mTRHR. The mutation was confirmed by the dideoxy chain termination method.
Receptor binding studies
One day after reseeding into 12-well plates, binding experiments
were carried out in buffer with cells in monolayer for 1 h at
37°C as described previously (Perlman et al., 1992).
Inositol phosphate formation
One day after transfection, cells in monolayer in 12-well plates
were labeled with 1 µCi of
myo-[3H]inositol/ml. Stimulation of inositol
phosphate formation was measured 1 day later for 1 h at 37°C by
methods previously described (Perlman et al., 1994a).
| |
RESULTS AND DISCUSSION |
|---|
|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
Simulated annealings
Selection of relevant structures
Fourteen simulated annealings were conducted as described above. At the end of the cooling period, a 100-ps MD simulation at 300 K was conducted on each of the structures. The calculated root mean square deviations (RMSD) from the structure at the beginning of the 300 K simulation along each of the 14 100-ps trajectories reveal that the structures have stabilized after 20 to 40 ps and suggest that they fluctuate around an average. We calculated the average RMSDs and the standard deviations in the final 60-80 ps with respect to the 100-ps structure to demonstrate that the deviation and the fluctuations are small. The RMSDs of each of the 14 simulations were (in Å) 1.2 ± 0.1, 1.1 ± 0.1, 1.3 ± 0.3, 1.4 ± 0.5, 1.8 ± 0.3, 1.8 ± 0.3, 1.4 ± 0.3, 1.5 ± 0.2, 1.0 ± 0.3, 1.7 ± 0.2, 1.5 ± 0.3, 1.4 ± 0.1, 1.7 ± 0.3, and 1.7 ± 0.5. Consequently, the last 60- to 80-ps interval was used to obtain the energy-minimized average structures for each simulation. To examine whether the structures fall in the allowed range of conformational space, the, , and 
1 angles were measured in
the 14 energy-minimized average structures for all residues in the
extracellular loops. The Ramachandran plot is presented in Fig.
2, and an analysis performed with
PROCHECK (Laskowski et al., 1993; Morris et al., 1992) reveals that
86% of the observed values are clustered in the sterically allowed
regions. The histograms of the distributions of the 1 angles
presented in Fig. 3 show that the
majority of the residues in the loops take conformations characteristic
of non-
/
structures (McGregor et al., 1987). This analysis
confirms that the simulated annealing protocol followed by a constant
temperature simulation generates structures that are in a
conformationally allowed region and that belong to proteins without a
well defined secondary structure.
|
|
Clustering
The 14 minimized averaged structures were subsequently clustered according to their pairwise RMSDs for C values into conformational families employing the program Xcluster (Shenkin and McDonald, 1994).
At a level of 1.89-Å resolution, the cluster analysis reveals the
existence of one major and six minor families. The major family is
composed of seven members, suggesting that 50% of the structures have
a common fold of the extracellular loops. Fig.
4 shows the general fold of the
extracellular loops in two members of the major family (top) and in two
members of the minor families (bottom). Clearly, among members of the
major family, EC1 is always positioned over helices 2 and 3 and packs
against EC2A. EC2A consistently starts with a short loop at its amino
terminus and moves toward helix 1 or 7 to maintain the disulfide bridge
at the carboxy terminus of EC1. EC2B (Gly180 to Ser189) lies below EC2A
and thus closer to the helical bundle. It spans the area between
helices 3, 5, 6, and 7. Consequently, it is positioned above the
transmembrane binding pocket. The backbone of EC3 is U-shaped, but the
side chains form close contacts with those of EC2B, which occlude
access to the transmembrane binding pocket. Members of the various
minor families differ from those of the major family mostly by EC1, which often lies outside helices 1 and 2 (when viewed from the extracellular side). Consequently, EC2A moves toward helices 1 and 3 (as opposed to 1 and 7 in the major family), which results in a less
convoluted EC2B. Differences in folding patterns of the extracellular
loops clearly emerge across the various families as exemplified in Fig.
4, although the conserved patterns described above remain within
clustered groups. The conservation of folding patterns among 50% of
the structures obtained in this work suggest that, despite the short
annealing and simulation time, the simulated annealing protocol
employed here is quite satisfactory.
|
; Stanfield and Wilson,
1995). Furthermore, the interaction between a nonamer peptide and the
antibody recognition domain, Fab 17/9, was shown to induce dramatic
changes in the position of the loops and domains in the antibody (Rini
et al., 1992). Thus, an initial complex between TRH and a surface
binding site could change the interaction between the residues that
obstruct access to the transmembrane region and trigger the opening of
a pathway to access the binding pocket.
An alternative mechanism could involve a spontaneous unfolding motion
of the loops that would result in an intermittent opening of a pathway
to the transmembrane domain. A recent study performed on the access of
small ligands into the core of lysozyme (Feher et al., 1996) suggests
that relatively large-scale fluctuations of backbone atoms are required
for entry of the ligands into the core of the protein. Such a mechanism
could play a role in TRH access to its transmembrane binding pocket.
The surface binding mechanism depends on the presence of a recognition
site with properties complementary to those of TRH. The large-scale
fluctuation mechanism depends on the vibrational properties of the
loops and in particular on an anti-correlated motion of EC2B and EC3
that may open a pathway to the transmembrane binding pocket. Below we
present an analysis of the recognition elements on the surface of the
receptor. To address these questions, we have performed a 1-ns
simulation of the extracellular loops anchored to the frozen
transmembrane helices. The starting structure employed for this
simulation was one of the members that belongs to the major family
obtained from the clustering analysis described above. The structure
derived from this 1-ns simulation performed at 300 K was an
energy-minimized average structure obtained over the last 500 ps of the
trajectory. The calculated average RMSD from the structure obtained at
1 ns during the last 500 ps was 1.14 ± 0.14 Å. The small value
of the average RMSD and in particular the small standard deviation
reflect the fact that the structure has reached a stable conformation
over the last 500 ps of the trajectory. The trajectory was also used to
perform a quasi-harmonic analysis of the vibrational properties of the
loops.
Recognition elements
Characterization of surface binding properties could be divided
into two elements. One is the presence of cavities that present a
sterically confined space into which the ligand can fit. The other are
the electrostatic properties of such a cavity, which show
complementarity to the charge distribution in the ligand. These two
elements could be associated, respectively, with the entropic driving
force due to release of bound water in the cavity and due to the
enthalpic driving force generated by specific polar interactions
between the ligand and the cavity. To examine these properties we have
constructed the molecular surface of this average structure employing
the program GRASP (Nicholls, 1992). The surface curvature is presented
in Fig. 5 A. The surface has
several cavities, one of which is positioned in the interface between
loop 2 (EC2B) and loop 3 (EC3). The cavity defined by Tyr181, Lys182,
Arg185, Asn186, Tyr187, Asn289, Ser290, and Phe296 (Fig. 5
B) is consistently observed on the surface of the average
structures of members of the large clustered family (not shown here)
and persists in snapshots of the structures extracted from long MD
trajectories. This suggests the formation of a surface binding site
that can become a putative entry point into the transmembrane binding
pocket.
|
To probe the significance of the putative surface binding pocket, we
mutated three residues that define the boundaries of the pocket. R185A,
N289A, and Y181F mutant TRH receptors were constructed and tested.
Results with N289A and Y181F TRHRs are described elsewhere (Perlman et
al., in press, 1998). The affinity of N289A was 10-fold lower than that
of native TRHR. The affinity of Y181F was too low to be determined in
binding experiments and was estimated from the potency of TRH in
activating the hydrolysis of phosphatidyl inositol (EC50s). We have
shown previously (Perlman et al., 1994a,b, 1996) that the relative EC50
of a mutant receptor compared with the wild-type receptor is a good
representation of the change in relative affinities. Compared with
wild-type TRHR, the EC50 of TRH for Y181F was 3300-fold higher. The
affinity of R185A TRHR was also too low to be determined in binding
experiments and was estimated through activation experiments as
described in Materials and Methods. The EC50 of TRH for R185A TRHR was
2200 nM (1700-2800 nM, 95% confidence interval; n = 4) compared with 0.59 nM for wild-type TRHR (Perlman et al., 1997),
indicating an affinity decrease of 3700-fold. These results indicate
that mutations of Tyr181, Arg185, and Asn289 lower the affinity of TRH
to the mutant receptors and suggest that these residues are important
for binding TRH. They are consistent with the idea that Tyr181, Arg185,
and Asn289 are part of the proposed surface binding pocket.
Some of the residues of the cavity defined above have been mutated
previously by Han and Tashjian (1995a,b). Mutant receptors Y181F,
N289A, and S290A were found to exhibit a significant loss of binding
compared with native TRHR when assayed with a single dose of
N-
-methylhistidine-TRH. The binding of mutant N186A was the same as the native receptor. This further supports the suggestion that the surface binding site is defined by residues in EC2B and EC3.
To further explore the properties of this cavity, we have calculated
the electrostatic potential on the surface defined by the extracellular
loops (Fig. 5 C). The map shows that the center of the
putative surface binding pocket has a positive electrostatic potential
surrounded by areas of negative potential. The positive potential is
generated primarily by Lys182 (EC2B) whereas the surrounding negative
potential is produced by Glu298 (EC3) and polar residues in EC3 (e.g.,
Ser290 and Ser293). It is difficult to present the charge distribution
on TRH, which could demonstrate its complementarity to the
electrostatic potential of the putative binding pocket. Examination of
the TRH structure shown in Fig. 5 D reveals that the oxygens
of three of the four carbonyls point in the same direction. Such an
arrangement of negatively charged groups could be attracted by the
positive potential generated by Lys182. The structure also shows that
the positively charged groups (N---H) are distributed on the periphery
of the molecule and could be attracted by the negative potential
generated by the polar groups around the pocket. These properties could
be illustrated by the electrostatic potential on the molecular surface of TRH as shown in Fig. 5 D. The negative potential defined
by the backbone carbonyls could interact with the positive charge of
Lys182 (EC2B) and the positive potential defined by the histidine, the
pyroglutamyl, and the terminal ProNH2 could be interacting with the
negative charges of Ser290, Ser293, and Glu298 (EC3). Our finding that
the affinity of E298A TRHR is threefold lower than wild-type TRHR is
consistent with a role for Glu298 in binding (Perlman et al., 1998).
It is interesting to examine the putative binding pocket at the atomic
level. The intention of this analysis is to provide a molecular basis
for the properties of the pocket and will of course depend on the
validity of the structure of the loops. Nevertheless, the consistency
between the static and dynamic properties of the pocket (see below)
provides a point of interest. A detailed picture is shown in Fig.
6. Lys182 (EC2B) lines the bottom of the
cavity by forming H-bonds with the side chains of Ser290 (EC3) and
Glu298 (EC3), which is itself H-bonded to the OH group of Tyr181
(EC2B). This intricate hydrogen bonding network between EC3 and EC2B
results in the tight packing of the loops described above. To probe the extent of permanency of the hydrogen-bonding network, we have examined
the fraction of time the interactions between Lys182, Glu298, and
Ser290 (see Fig. 6) are interrupted in the course of a 1-ns simulation.
The Tyr181-Glu298 interaction, monitored by the
(Tyr181)-O
-H···O
(Glu298) distance,
was maintained at 
1.8 Å 99% of the time. The
(Lys182)NH···O(Glu298), and
(Lys182)NH···O
(Ser290) were
maintained 88% and 45% of the time, respectively. Thus, Glu298 in
this model of the extracellular loops is responsible for maintaining
the rigid scaffold of the putative binding pocket observed from the
snapshots of the molecular surface as a function of time. To test the
effect of the continuum dielectric representation of the solvent, we
conducted simulations with a distance-dependent dielectric function
that has a fourfold stronger dependence on distance. We have
constructed a molecular surface on the minimized average structure from
this simulation. The surface curvature and surface electrostatic
potential were nearly identical to those obtained from simulations with
= r. We conclude that the distance-dependent continuum
dielectric function did not significantly bias the simulations in favor
of polar residue packing. However, additional studies with an explicit
representation of the solvent will be needed to address the
significance of the approximate representation of the solvent.
|
The presence of a surface cavity with size and electrostatic properties complementary to those of the ligand is consistent with the suggestion that an initial recognition of TRH occurs on the surface of the receptor formed by the extracellular loops. This suggestion is supported by the observation that residues that define portions of the cavity (e.g., Tyr181, Arg185, and Asn289) also affect binding affinity. The corollary of this suggestion is that TRH binds to the surface before accessing the transmembrane binding pocket.
Dynamic analysis
Quasi-harmonic analysis
The static representation described heretofore cannot address the question of the involvement of loop motion in guiding the ligand to its binding pocket in the transmembrane domain. To address this problem, we chose to use quasi-harmonic analysis to describe loop motions and the correlated displacements between different loops. Such an approach has been taken previously in studies of deoxymyoglobin (Seno and Go, 1990a,b), lysozyme (Amadei et al., 1993; Horiuchi and Go, 1991),
thermolysin (van Aalten et al., 1995a), and retinol binding protein
(van Aalten et al., 1995b) employing normal mode analysis,
quasi-harmonic analysis, and essential dynamics. They revealed that
low-frequency motions in the essential subspace of the protein are
related to its functional behavior. For instance, Amadei et al. (1993)
show that the essential dynamics of lysozyme can be related to the
functional behavior of the protein, such as opening and closing of the
active site, whereas Horiuchi and Go (1991) extracted the motions that
represented the hinge bending of the two lobes forming the active site
of the cleft of lysozyme. van Aalten et al. (1995b) also used essential
dynamics to study differences in dynamics between the holo and apo
forms of the cellular retinol-binding protein. Their results reveal inhibition of essential motions upon ligand binding and show large correlated motions of retinol with regions of the protein, pointing to
a possible retinol entry/exit site.
In the present work, a quasi-harmonic analysis was performed on a
trajectory from a 1-ns MD simulation on the structure shown in Fig. 6,
which belongs to the major family described above. As the transmembrane
binding pocket in TRHR is located directly underneath EC2B and EC3, and
the putative surface binding site is defined by residues from these
loops, we focus our discussion on the motions of this extracellular
domain.
The amount of motion associated with the calculated eigenvectors can be
represented by the cumulative positional fluctuation defined as follows
(Amadei et al., 1993):
|
i is the eigenvalue, n is the
cumulative index of the normal modes, and N is the total
number of vibrational modes. The cumulative positional fluctuation,
n, describes the fraction of the total variance from the
average position accounted by the first n vibrational modes.
Our results show that 83% of the total variance due to motion of the
extracellular domain is described by the first 10 eigenvectors of a
total of 162. These modes carry little vibrational energy but are
responsible for large displacements. In contrast, the higher
vibrational modes, although they carry most of the zero-point
vibrational energy due to their higher frequencies, contribute very
little to the total motion.
Projection of the trajectory onto the individual eigenvectors allows
for a more detailed description of the motion accounted for by each
eigenvector. The sampling distributions for the displacements of
motions from the average along the first 10 eigenvectors in the 1-ns
trajectory are presented in Fig. 7. They
indicate that non-Gaussian distributions are found among the first few
eigenvectors. Thus, the motions described by these vibrational modes
are highly anharmonic. Anharmonic effects are of importance in
larger-scale collective motions in proteins (Ichiye and Karplus,
1991) because they are associated with low-energy vibrations and they
increase the overall amplitude of fluctuation of the protein over that predicted by a harmonic vibration (Teeter and Case, 1990). Thus, we
focus on the low-frequency modes to sample large-scale motions of
possible biological importance in the TRHR.
|
Distance analysis
To identify the residues that contribute to the large-scale motions in the low-frequency modes, we have calculated the excursion distances of the center of mass of each residue of the extracellular loops in the first four vibrational modes. They were obtained from the projected trajectories of each of the first four modes and represent the distances traveled by each center of mass within the mode and are shown in Fig. 8. In the first mode, fluctuations in excess of 2.5 Å are observed, whereas, consistent with the decrease in the fractional contribution to the total displacement, the individual contributions from the following modes significantly decrease as the frequency increases. As expected, residues identified as being part of the static recognition elements (i.e., Tyr181, Lys182, and Glu298) undergo very little motion along any of the eigenvectors because they participate in an extensive hydrogen bond network (see above). On the other hand, the stretch of amino acids I183-N186 in EC2B and F291-P295 in EC3 show large excursions exceeding 1 Å.
|
|
|
Connection between static and dynamic definition of the putative entry point
The residues in the putative binding pocket can be classified into
two different categories. One set of residues, Tyr181, Lys182, Ser290,
and Glu298, has low mobility because of the hydrogen bond interactions
among these residues and thus can be considered as a static component
responsible for TRH recognition. Another set of residues, Ile183-Asn186
in EC2B and Leu292-Pro295 in EC3, are engaged in large anti-correlated
positional fluctuations. We suggest that the interaction of TRH with
the surface binding pocket contributes to the disruption of the
interactions between the static residues, introducing a possibility of
enhancing the anti-correlated vibrational motions of the loops. Such a
mechanism will ensure the specificity of ligand recognition defined by
the static residues as well as provide a way of inducing conformational changes in the loops to open a pathway into the transmembrane pocket.
To examine whether conformational changes in the extracellular residues
can give rise to such a pathway, we have manually rotated the side
chains of residues at the rim (Tyr181 (1 by 100°) and Phe296 (1 by 100°)) and at the bottom (Lys182)
(1 by 168° and 2 by 
170°) of the
surface binding pocket. As shown in Figure 10 B, residues in
the transmembrane binding pocket (i.e., Tyr282 and Arg306) become
exposed when such rotations of the surface residues are induced.
Additional studies of the hormone-receptor complex in which TRH is
bound to the surface of TRHR will have to be conducted to elucidate the
conformational changes that can be induced by such an interaction.
| |
CONCLUSION |
|---|
|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
In the present work, we have focused our attention on the extracellular domain of the TRH receptor and showed an apparent conserved pattern in the folding of the loops through the use of simulated annealing and cluster analysis. Among the most important conclusions from this study is that a static representation of the extracellular loops does not present an apparent access pathway into the transmembrane binding pocket but suggests the formation of a putative entry point on the surface of the receptor with steric and electrostatic properties complementary to those of TRH. A dynamic representation supported by quasi-harmonic analysis shows an anti-correlated motion between EC2B and EC3, which could result in the formation of a path for TRH to access its binding pocket. Although this path appears to be occluded by Lys182, which hydrogen bonds to residues in EC3, an interaction with TRH could induce conformational changes in the surface binding pocket to expose an entrance to the transmembrane binding site.
Our results therefore suggest that both recognition of specific elements on the receptor surface and loop motion are necessary for the ligand to access its transmembrane binding pocket. These results point to a multi-step mechanism for binding of TRH to its receptor. TRH initially interacts with residues within the extracellular loops and then moves into the transmembrane binding pocket, aided by the motion of the loops. We suggest that sequential binding interactions similar to these between TRH and its receptor may occur between small ligands and all GPCRs and that small ligands may be guided into transmembrane binding pockets by extracellular loops.
| |
ACKNOWLEDGMENTS |
|---|
We thank the National Energy Research Supercomputer Center at Lawrence Livermore National Laboratory for providing computational support for this work.
This work was supported by National Institutes of Health Research Fellowship Award DK 09647 (to A.-O. Colson), National Institutes of Health Physician Scientist Award DK 02101 (to J. H. Perlman), and grant DK 43036 (to M. C. Gershengorn and R. Osman).
| |
FOOTNOTES |
|---|
Received for publication 21 April 1997 and in final form 17 November 1997.
Address reprint requests to Dr. Roman Osman, Mount Sinai School of Medicine, Department of Physiology and Biophysics, Box 1218, 1 Gustave L. Levy Place, New York, NY 10029. Tel.: 212-241-5609; Fax: 212-860-3369; E-mail: osman{at}inka.mssm.edu.
| |
REFERENCES |
|---|
|
Top
Abstract
Introduction
Materials & Methods
Results & Discussion
Conclusion
References
|
|---|
2 adrenergic receptor protein based on computer modeling studies.
J. Mol. Biol.
225:859-871[Medline].
Biophys J, March 1998, p. 1087-1100, Vol. 74, No. 3
© 1998 by the Biophysical Society 0006-3495/98/03/1087/14 $2.00
This article has been cited by other articles:
 |
|
 |
|
  S. Engel, S. Neumann, N. Kaur, V. Monga, R. Jain, J. Northup, and M. C. Gershengorn Low Affinity Analogs of Thyrotropin-releasing Hormone Are Super-agonists J. Biol. Chem., May 12, 2006; 281(19): 13103 - 13109. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Lu, W. Huang, S. Worthington, P. Drabik, R. Osman, and M. C. Gershengorn A Model of Inverse Agonist Action at Thyrotropin-Releasing Hormone Receptor Type 1: Role of a Conserved Tryptophan in Helix 6 Mol. Pharmacol., November 1, 2004; 66(5): 1192 - 1200. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Leanos-Miranda, A. Ulloa-Aguirre, T. H. Ji, J. A. Janovick, and P. M. Conn Dominant-Negative Action of Disease-Causing Gonadotropin-Releasing Hormone Receptor (GnRHR) Mutants: A Trait That Potentially Coevolved with Decreased Plasma Membrane Expression of GnRHR in Humans J. Clin. Endocrinol. Metab., July 1, 2003; 88(7): 3360 - 3367. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. G. Paterlini Structure Modeling of the Chemokine Receptor CCR5: Implications for Ligand Binding and Selectivity Biophys. J., December 1, 2002; 83(6): 3012 - 3031. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W.B. Church, K.A. Jones, D.A. Kuiper, J. Shine, and T.P. Iismaa Molecular modelling and site-directed mutagenesis of human GALR1 galanin receptor defines determinants of receptor subtype specificity Protein Eng. Des. Sel., April 1, 2002; 15(4): 313 - 323. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. J. Fromme, A. A. Katz, R. W. Roeske, R. P. Millar, and C. A. Flanagan Role of Aspartate7.32(302) of the Human Gonadotropin-Releasing Hormone Receptor in Stabilizing a High-Affinity Ligand Conformation Mol. Pharmacol., December 1, 2001; 60(6): 1280 - 1287. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A. Ballesteros, L. Shi, and J. A. Javitch Structural Mimicry in G Protein-Coupled Receptors: Implications of the High-Resolution Structure of Rhodopsin for Structure-Function Analysis of Rhodopsin-Like Receptors Mol. Pharmacol., July 1, 2001; 60(1): 1 - 19. [Abstract] [Full Text]
|
|
|
|
 |
|
 M. C. Gershengorn and R. Osman Minireview: Insights into G Protein-Coupled Receptor Function Using Molecular Models Endocrinology, January 1, 2001; 142(1): 2 - 10. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. F. O’Dowd, D. K. Lee, W. Huang, T. Nguyen, R. Cheng, Y. Liu, B. Wang, M. C. Gershengorn, and S. R. George TRH-R2 Exhibits Similar Binding and Acute Signaling but Distinct Regulation and Anatomic Distribution Compared with TRH-R1 Mol. Endocrinol., January 1, 2000; 14(1): 183 - 193. [Abstract] [Full Text]
|
|
|
|
 |
|
