Chemical Senses Vol. 29 No. 7 © Oxford University Press
2004; all rights reserved
Predicted 3-D Structures for Mouse I7 and Rat I7 Olfactory Receptors and Comparison of Predicted Odor Recognition Profiles with Experiment
Materials and Process Simulation Center (MC: 139–74), California Institute of Technology, Pasadena, CA 91125, USA
Correspondence to be sent to: William A. Goddard, Materials and Process Simulation Center (MC: 139–74), California Institute of Technology, Pasadena, CA 91125, USA. e-mail: wag{at}wag.caltech.edu
 |
Abstract |
|---|
 Top Abstract IntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
The first step in the perception of an odor is the activation of one or more olfactory receptors (ORs) following binding of the odorant molecule to the OR. In order to initiate the process of determining how the molecular level receptor-odorant interactions are related to odor perception, we used the MembStruk computational method to predict the three-dimensional (3-D) structure of the I7 OR for both mouse and rat. We then used the HierDock ligand docking computational method to predict the binding site and binding energy for the library of 56 odorants to these receptors for which experiment response data are now available. We find that the predicted 3-D structures of the mouse and rat I7 OR lead to predictions of odorant binding that are in good agreement with the experimental results, thus validating the accuracy of both the 3-D structure and the predicted binding site. In particular we predict that heptanal and octanal both bind strongly to both mouse and rat I7 ORs, which conflicts with the older literature but agrees with recent experiments. To provide the basis of additional validations of our 3-D structures, we also report the odorant binding site for a new odorant (8-hydroxy-octanal) with a novel functionality designed to bind strongly to mouse I7. Such validated computational methods should be very useful in predicting the structure and function of many other ORs.
Key words: G protein coupled receptor, hydrogen bonds, molecular dynamics, transmembrane domain
|
Introduction |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
The early stage in odorant detection involves binding of the odorant molecule to an olfactory receptor (OR; Buck and Axel, 1991
;
Lancet et al., 1993)
followed by activation of the OR through release of the G-protein fragments. Each
olfactory sensory neuron expresses only one OR type, but a particular OR can respond to
multiple odorants. A particular ligand can also elicit response from multiple ORs. This
leads to a unique combination of OR responses for each odorant (Malnic et al., 1999). Thus the mammalian olfactory
system uses a combinatorial response to discriminate thousands of odorants (Sicard and Holley, 1984;
Malnic et al., 1999;
Kajiya et al., 2001).
ORs belong to the superfamily of membrane bound G-protein coupled receptors (GPCRs;
Buck and Axel, 1991;
Mombaerts, 1999). There are 913 ORs
in mice (Godfrey et al.,
2004) and 339 ORs in humans (Malnic
et al., 2004) making it an extremely forbidding task to elucidate
experimentally (or computationally) the details by which odorants activate each of the
ORs. There is almost no molecular level information on how and where each odorant binds
to the OR or when and how this leads to their activation. The major impediment to
obtaining this molecular level information is that there is no experimental
three-dimensional (3-D) structural information available for any OR of any species of
life! Indeed, considering all forms of life, there is an experimental 3-D structure for
only a single GPCR, bovine rhodopsin (Grigorieff
et al., 1996). This is because it has not yet been possible to
obtain crystals suitable for diffraction studies of these membrane bound proteins,
despite years of intense effort.
Consequently, we have developed computational techniques (MembStruk) suitable for
predicting the 3-D structures of GPCRs. The original version of MembStruk1.0, was
validated on bacteriorhodopsin and used for prediction of structure of OR-S25 (Floriano et al., 2000). Subsequent
improved version of MembStruk2.0 (with optimization of rotational orientations of the
helices) have been validated for bovine rhodopsin, where it leads to a CRMS (coordinate
root-mean-square) error of 2.8 Å in the transmembrane (TM) domains (Vaidehi et al., 2002;
Trabanino et al., 2004)
compared to the crystal structure (Poincelot
et al., 1970;
Palcezwski, 2000). We have also
validated the MembStruk2.0 predicted structures for human ß1 and ß2 adrenergic
receptors (Vaidehi et al.,
2002;
Freddolino et al., 2004) and
human dopamine D2 receptor (Kalani et
al., 2004). Since no experimental structural data are available for
direct validation of our predicted structures for these systems, we used the HierDock2.0
method (Floriano et al.,
2000;
Vaidehi et al., 2002) to
predict the binding sites of ligands to the predicted 3-D structures of these GPCRs.
These binding sites were then compared to the numerous experimental mutation and binding
studies carried out in developing subtype specific agonist and antagonist
pharmaceuticals. We found that the predicted binding site of these ligands all agree
quite well with all available experimental mutation data. This validation of the
techniques gives us confidence to now apply these techniques (MembStruk and HierDock) to
the more complex problem of ORs, where all available information suggests much less
selectivity than for the rhodopsin, adrenergic and dopamine receptors.
There has been some progress in determining which odorants lead to activation of
specific mammalian ORs. However, the experiments are laborious and results are available
on only a few ORs for a relatively small library of odorants (Kiefer et al., 1996;
Bozza and Kauer, 1998;
Krautwurst et al., 1998;
Zhao et al., 1998;
Duchamp-Viret et al., 1999;
Malnic et al., 1999;
Mori et al., 1999;
Rubin and Katz, 1999;
Araneda et al., 2001;
Kajiya et al., 2001).
Consequently, we have chosen to apply the MembStruk and HierDock methods to these few
more well studied ORs as the first step in approaching the much more complicated task of
elucidating the structures and function for the whole set of mammalian ORs.
Our first report on the structure and ligand binding for an OR (Floriano et al., 2000) was for the S25 mouse OR
where it was known that only 2 of 24 simple aliphatic odorants were agonists for this
ORS25. Here, we correctly predicted that the two known cases do bind much more strongly
than the other 22 odorants and report a more complete validation by comparing the
calculated binding energies of 56 odorants to the intracellular Ca2+
imaging measurements to the rat and mouse I7 OR. Prior to publication of these
experimental results (Bozza et al.,
2002), we arranged to carry out a blind test of our methods. Tom Bozza and
Peter Mombaerts (Rockefeller University) sent us the names of the 56 odorants (shown in
Table
1) for which they had measured the
intracellular calcium influx response for the I7 OR both rat and mouse, but they provided
no experimental data until after we reported to them our calculated binding sites and
energies, reported herein. We predicted the structure and odorant binding energies of
R-I7 and M-I7 using MembStruk1.0 and HierDock2.0. They then provided us with the list of
experimental agonists for these two ORs, which then was published (Bozza et al., 2002). As shown in this paper, the
calculated binding energies correlate well, but not perfectly to the experimental
activation profiles (correctly showing that binding to aldehydes is favored while binding
to such chemical classes as acids and alcohols are not favored). In addition our
predictions confirmed in advance the result that both rat and mouse I7 receptor are
activated by both heptanal and octanal.
|
After making these blind predictions, we made significant improvements to the MembStruk structure prediction methods for our studies on biogenic amine receptors, where there are large amounts of experimental data on mutations and ligand binding affinities. These improved methods have now been applied to mouse and rat I7 ORs, leading to results that are in significantly improved agreement with experimental measurements of the intracellular calcium imaging results. Based on our best predictions of the structure and binding site, we have designed three new odorants with two functional groups that we predict will bind to mouse and rat I7 receptors. Experimental tests on these compounds would provide additional tests on how well the theory can be trusted for predictions prior to experiment.
The mouse I7 (M-I7) and rat I7 (R-I7) ORs both contain 301 residues. They have
95% sequence identity, differing by only 15 residues, four of which are located in
the TM region (see Figure
1). Despite the high similarity of
M-I7 and R-I7, their odorant activities are somewhat different (Krautwurst et al., 1998;
Zhao et al., 1998;
Wetzel et al., 1999;
Araneda et al., 2000;
Bozza et al., 2002;
Levasseur et al., 2003).
These differences and similarities in odor recognition make M-I7 and R-I7 good candidates
to test how well our modeling techniques can discriminate odor differentiation resulting
from slight changes in sequence. Previous modeling of R-I7 based on bacteriorhodopsin
structure was reported by
Singer (2000).
|
|
Results and discussion |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
We report here the predictions for the 3-D atomic-level structures of the M-I7 and R-I7 ORs, the binding sites for the odorants that activate these receptors and the relative binding energies for the odorants in these sites. We find results that correlate well with the experimental intracellular calcium ion influx measurements (Araneda et al., 2000
;
Bozza et al., 2002; T. Bozza,
personal communication). Prediction of the 3-D structure of M-I7 and R-I7 ORs
The details of the methods used for predicting the structure and function of M-I7 and R-I7 ORs are described in an appendix (see Supplementary material). However, in the next section a brief outline of the methods as applied to M-I7 and R-I7 are given.
Predicted structures for M-I7 and R-I7
To predict the TM regions we aligned the sequences for M-I7 and R-I7 along with 21
other rat and mouse ORs that had similar homology and these alignments were used to
predict the TM region based on hydropathicity profiles (Trabanino et al., 2004). The predicted TM regions
for M-I7 are compared in Figure
1. Using the predicted TM regions, we
applied the MembStruk1.0 method to predict the 3-D structures. Two sets of structures
were predicted using MembStruk1.0 and subsequently MembStruk2.0 methods. The first
structures using MembStruk1.0 (described in
Floriano et al., 2000) are
denoted as preM-I7 and preR-I7. These structures were used for the predictions made in
the blind study, prior to the publication of the experimental odorant activation assays
(Bozza et al., 2002).
Subsequently, the improved version of MembStruk2.0 method was used, as described in an
Appendix (http://www.wag.caltech.edu/gpcr/i7/appendix.html)
and in
Vaidehi et al. (2002), to
obtain the final structures of M-I7 and R-I7. These improvements in MembStruk2.0 used the
calculated potential energy to determine the optimum rotational orientation of the
helices, rather than just the hydrophobic moment as in MembStruk1.0. These were motivated
by studies we were doing on the structures of dopamine and adrenergic receptors
(Freddolino et al., 2004;
Kalani et al., 2004). In
addition, we used the predicted structure of M-I7 and the high sequence homology between
R-I7 and M-I7 to build a homology model for R-I7 based on the predicted M-I7 structure as
template. Below we refer to this as the R-I7(hom) structure.
For each of these 3-D structures we applied HierDock2.0 (Vaidehi et al., 2002) to predict the odorant
binding sites and binding energies of the 62 molecule odorant library. Since the results
using as preM-I7 and preR-I7 were obtained prior to knowledge of the experimental
results, we consider it valuable to report them here. Hence, the methods used to these
results are described in detail in an appendix (http://www.wag.caltech.edu/gpcr/i7/appendix.html).
However, the main body of the results and the analysis of the binding sites have been
done with the predictions with R-I7(hom) and M-I7 that are in better agreement with
experiment.
Comparison of the Predicted M-I7 and R-I7(hom) Structures
A standard way to compare different structures for the same protein is to determine
the coordinate root-mean-square (CRMS) difference between the structures (after matching
the center of mass and moments of inertia). However, since CRMS is an average quantity it
does not have the discrimination required to understand how the differences in structure
might affect function. Consequently, we developed the ‘MembComp’ method for
comparing the structural features of two GPCRs. Here, we start with the reference plane
intersecting hydrophobic center (Trabanino et
al., 2004) of each TM helix of the final structure, compare such helical
characteristics as helical bends and tilts (Filizola et al., 1998;
Trabanino et al., 2004) for
each of the 7 TM domains. These helical properties are summarized in Table
2 and Figure
2. These results show that the M-I7
and R-I7(hom) structures are remarkably similar with only slight differences in their
hydrophobic moments, as expected by the 95% sequence identity. The main chain
atoms in the TM region differ by a CRMS of only 1.5 Å.
|
|
Table 3 compares the structural features between M-I7 and bovine rhodopsin (which have a sequence identity of 11%), while Figure 2 compares them graphically. The CRMS difference between these structures is 6.22 Å, with the largest differences in the hydrophobic moments of TM6 and a salt bridge from TM6 to the IC2 loop in rhodopsin. With such a large difference in structure, we expect that using bovine rhodopsin as the template for homology structure predictions may not lead to useful predicted structures for the ORs.
|
Experimental methods to determine odorant activation profiles
The odorant activation profiles for the M-I7 and R-I7 ORs were determined
experimentally by Dr Tom Bozza of Rockefeller University, using fura-2 calcium imaging in
acutely dissociated olfactory sensory neurons. KCl and forskolin were used as positive
control stimuli (see
Bozza et al., 2002). These
experiments were carried out by first grouping the odorants into six sets or mixtures
(A–F) as shown in Table
1. Then for those mixtures with a
positive response, the mixtures were separated into individual components to determine
which odorant was causing the activation of the OR (Bozza et al., 2002). We had no access to these
experimental results, nor did we make use of literature data (Araneda et al., 2000), until the
predictions of binding energies using the preM-I7 and preR-I7 predicted structures were
completed and sent to Bozza.
Predicted binding site and binding energy for odorants in M-I7 and R-I7 ORs
Identification of the putative odorant binding site
We predicted the putative binding site for each of the 62 test odorants to both R-I7
and M-I7, by using HierDock2.0 to scan the entire receptor structure. The first step was
to partition the entire receptor into 13 overlapping binding regions containing all the
internal voids and surface accessible voids present in the predictions protein
structures. Then we applied the HierDock2.0 protocol, described in
Vaidehi (2002) and summarized in an
Appendix (http://www.wag.caltech.edu/gpcr/i7/appendix.html),
to docking the potential odorants to each of these regions. The best binding region of
these 13 regions for all of the test odorants was found to be located between TM helices,
3, 4, 5 and 7 in both the OR structures.
Binding energies of odorants in the preM-I7 and preR-I7 structures
Having located the binding region, the HierDock2.0 protocol (detailed athttp://www.wag.caltech.edu/gpcr/i7/appendix.html)
was again used to dock all 62 odorants in this putative binding region for both M-I7 and
R-I7 and to calculate the binding energy. The calculated binding energies of odorants in
the initial preR-I7 and preM-I7 structures are shown in Table
4. The odorants with the best binding
energies have their energies shaded darkest (predicted binding energy greater than 30
kcal/mol, none seen in this table) while the second best are shaded lightest (predicted
binding energies >25 kcal/mol and <30 kcal/mol), followed by (for 20–25
kcal/mol) and (for 15–20 kcal/mol). As indicated in Table
4, the pattern of predicted binding
energies is in fair agreement with experimental intracellular calcium concentration
measurements for both R-I7 and M-I7 structures. For example, we predicted that both
heptanal and octanal bind strongly to both M-I7 and R-I7 in disagreement with the
published experiments (Krautwurst et
al., 1998), but as we learned later this does agree with the new
experiments (Bozza et al.,
2002). Overall, 25% of the odorant predicted to be good binders
(shaded lightest) were confirmed by experiment, while 17% of the odorants with
medium affinity predictions were confirmed, 10% of the weak binding odorants and
7% of the unmarked were observed to be agonists. Comparison of the calculated
binding energies to the experimental data available in literature (Araneda et al., 2000;
Levasseur et al., 2003), we
find that 50% of the good binding odorants (shaded lightest) are also found to be
agonists with experiments. For example, nonanal and decanal were shown to be agonists for
R-I7 (Araneda et al., 2000;
Levasseur et al., 2003),
which is in agreement with the calculated binding energies in Table
4. The blind predictions correctly
concluded that aldehydes would be the main group activating R-I7 and M-I7.
|
Binding energies of the odorants in the R-I7(hom) and M-I7 refined structures
Although the MembStruk1.0 calculations led to results in fair agreement with the measured activation profiles for R-I7 and M-I7 receptors, there were several false positives (e.g. lilial, lyral and benzaldehyde) in the prediction. Later, the MembStruk1.0 method was improved while we were predicting the structure and function for the dopamine and adrenergic receptors, for which there is abundant mutation data available to validate the predicted binding sites.
After completing the computational results in the blind test, we applied the improved MembStruk2.0 method to again predict the 3-D structures of the I7 receptors. Then we used these new I7 structures with HierDock2.0 to predict the binding site and binding energy for the 62 odorants. The calculated binding energies of the odorants for the improved structures are shown in Table 5.
|
To simplify comparisons the calculated binding energies of the 62 ligands were categorized into seven grades. They are: class A, the best binding odorants with binding energies ranging from 30 to 40 kcal/mol (energies shaded darkest in Tables 4 and 5); class B (25–30 kcal/mol; shaded lightest); class C (20–25 kcal/mol); class D (15–20 kcal/mol); class E (10–15 kcal/mol no color); class F (0–10kcal/mol, no color); and class G, negative binding energy indicating no binding, no color.
Comparing to experiment (see Table 5) we find the following.
- Among class A, M-I7 has six aldehydes (of which four were
observed experimentally to be agonists) and one ketone (not an agonist experimentally).
While R-I7 has five aldehydes, of which all but decanal were observed experimentally to
be agonists by Bozza, while decanal was also observed to be agonist by
Araneda et al. (2000) and one
ester (not an agonist experimentally).
- Among class C, M-I7 has one alcohol, four esters and three ketones none of which
were observed to be agonists experimentally, while R-I7 has four aldehydes, two alcohols,
two esters and three ketones, none of which were agonists.
- Among classes D–G, were the other 38 odorants for M-I7 and 37 odorants for
R-I7, none of which were observed to be agonists.
Overall there is good agreement between the calculated binding energies and measured intracellular calcium response. Thus 62% (69% including decanal in I7 rat) of class A odorants were observed to be agonists experimentally, while 33% (39% including nonanal in I7 rat) of class B odorants and none of the five lower binding classes (with 75% of the odorants) were observed to be agonists. Clearly, the predictions identified aldehydes as the prominent binders to I7, which correlates well with the experimental observation that only aldehydes activate these ORs. Most of the experimental agonists (56% including decanal and nonanal for I7 rat) are in the top predicted binders shaded darkest (predicted binding energy >30 kcal/mol). The rest of the experimental agonists (44%) are the next best binders shaded lightest.
The false positives in the calculations could be due to (i) inaccuracies in the calculation of the binding energies such as no explicit inclusion of entropy or room temperature effects or (ii) the fact that some of these odorants predicted as false positives do bind but may not activate the ORs and could act as antagonists.
The available experimental data involves measuring the increase in calcium ion concentration in individual olfactory sensory neurons, which is a measure of activation by the odorant and not just the binding whereas the theory calculates binding site and binding energy of the odorant but not the activation process of the ORs. Strong binding is a necessary but not a sufficient condition for activation and hence our calculated binding energies should best be compared to measured binding constants. Unfortunately, such data is scarce and are not yet available for these mammalian ORs. Thus some odorants predicted to have good binding energies may not bind in the correct configuration to activate the OR serving perhaps as an antagonist rather than an agonist.
For example, we predict lilial and lyral to be in the top (red) group of good
binders, whereas the experiments did not find them to activate the receptors (Bozza et al., 2002). There are two
possible explanations for this discrepancy between binding energy and measured
activation: (i) the experiments tested these odorants only in a mixture; this makes the
comparison of theory with experiment ambiguous, since a mixture might contain an
antagonist ligand that would compete with an agonist in the mixture; and (ii) the other
possibility is that lilial and lyral themselves could be antagonists to these rat and
mouse I7 ORs. In the third section, we discuss competitive binding experiments that could
test if some odorants predicted to be top binders are not observed to activate because
they are antagonists or because that are agonists but inhibited by antagonists.
Residues predicted to be directly involved in binding of odorants to the R-I7 and M-I7 OR structures
Figure
3 shows the predicted binding sites
for octanal in M-I7 and R-I7(hom). Octanal was predicted as a good binder and shown
experimentally to be an agonist for both M-I7 and R-I7 (Araneda et al., 2000;
Bozza et al., 2002). Figure
3A,C indicate the binding pocket depth
as 
10 Å deep from the extracellular surface. This is similar to the
epinephrine-binding pocket of the beta-adrenergic receptor (ßAR;
Strader et al., 1989;
Freddolino et al., 2004) and
other ORs (Vaidehi et al.,
2002) and to 11cis-retinal pocket in bovine rhodopsin (Palcezwski et al., 2000). These figures
show that the ligand binding pocket is located in between TM helices 3, 4 and 6. The
residues making direct contact with the odorant are in the hypervariable region in the
sequence alignment of ORs (Buck and Axel,
1991;
Singer et al., 1995a,b;
Mombaerts, 1999;
Pilpel and Lancet, 1999), consistent
with their involvement in differential odor binding for different OR subtypes.
|
The details of the binding site of octanal in M-I7 and R-I7 structures are shown in Figure 4A,B, respectively. We find that Lys 164 is hydrogen bonded to the polar moiety for all the positive agonists, making it one of the critical residues for the binding of aldehydes. This could be directly tested experimentally by mutating this residue to uncharged polar residues (Tyr, Thr), which might switch receptor specificity toward odorants with polar but uncharged functional groups (say alcohols or ketones) or by mutating it to a nonpolar residue, which should lead to a dramatically different binding profile (or possibly to misfolding). Other residues that are involved in binding are: Ile 255, Ala 258, Ala 259, Ser 280 and Tyr 283. As detailed later in the text, mutating these residues might modulate the length of the alkyl chain recognized by these receptors. Tables 6 and 7 show the distances of the residues in the binding site of R-I7(hom) and M-I7 structures for the aldehydes predicted to be the best binders.
|
|
|
Description of binding sites of odorants with good binding energies
Citral, citronellal (+) and (–), heptanal, hexanal, nonanal and trans-cinnamaldehyde to M-I7. The binding site and orientation of citral, citronellal (+) and (–), heptanal, hexanal, nonanal and trans-cinnamaldehyde were all the same as octanal (Figure 4A). This is shown in Figure 4C for citral (yellow), nonanal (lime) and trans-cinnamaldehyde (orange). For each ligand, the long axis of the odorant is parallel to the membrane. In all these agonists the aldehyde functional group makes a hydrogen bond to Lys 164. The size of the odorant that can fit sufficiently near Lys 164 to hydrogen bond is modulated by Cys 114, Cys 117 and Phe 205. The length of the odorant binding in this mode is limited by the Ile 255, Ala 258, Ala 259, Ser 280 and Tyr 283. This suggests the residues that might be mutated to modify the binding profile and thereby validate our predictions.
Octanal, citral, citronellal (+) and (–), heptanal, hexanal, nonanal
and trans-cinnamaldehyde to R-I7(hom). The predicted binding site of these eight
ligands in R-I7(hom) (see Figure
4B) has the aldehyde functional group
hydrogen bonded to Lys 164. The binding site near Lys 164 is shaped into a narrow groove
lined by the residues: Cys 114, Cys 117, Phe 205 and Ile 209 which is very similar in
R-I7(hom) and M-I7. Table
8 shows the differences in the
binding pocket for the experimentally observed agonists and for decanal. The main
difference between these two receptors is that Leu 110 is closer to the binding pocket in
M-I7 while Phe 205 and Ile 209 are farther away from the binding pocket in M-I7. This may
explain why citronellal binds more strongly to M-I7. The residues near Lys 164 form a
groove that is narrower in R-I7(hom) than the corresponding groove in M-I7 and the
residues that limit the length of the ligand: Ile 255, Ala 258, Ala 259, Ser 280 and Tyr
283 are generally closer in R-I7(hom), which may explain why the longer ligand nonanal is
experimentally observed in M-I7 and not in R-I7(hom). However, there is a report that
nonanal is experimentally observed but with a weaker response (Araneda et al., 2000). These distances differ just
slightly for each ligand and the long axis of the odorant is again perpendicular to the
membrane. Since our calculations indicate similar binding constants for these ligands, it
could be that any differences observed experimentally might arise from other factors such
as the ease of activation following binding of agonist which might be affected by
residues remote from the active site.
|
Decanal to M-I7 and R-I7(hom) . We find that decanal binds to a site in M-I7 and R-17(hom) very similar to octanal; however, decanal must twist along its long axis (horizontal) in order to fit into the binding site. This is due to Ile 255, Ala 258, Ala 259, Ser 280 and Tyr 283 that hinder the length of this aldehyde (see Figure 4D for M-I7). The initial experimental results (Bozza et al., 2002
)
did not find activation by decanal, but as discussed below, experiments done after the
calculations show that it does lead to activation but is slower than heptanal in I7
mouse. Indeed,
Araneda et al. (2000) also
find that decanal activates I7 rat. The twisting of the molecule to fit the binding site
could cause strain and could be the cause for the slow activation, which caused it to be
missed as an agonist in the initial experiments. Lilial to M-I7 and R-I7(hom) . We predict that lilial binds strongly, but it was not found experimentally to be a positive agonist. Indeed the predicted binding site for lilial is quite different than for the observed agonists, being nearly vertical (see Figure 4E for M-I7). This vertical binding (parallel to the membrane) of this odorant is stabilized by the hydrophobic residues: Leu 106, Phe 109, Leu 110, Ile 168, Phe 205, Phe 262 and Ile 263. These residues form a hydrophobic tunnel that might act as a path for the aldehydes to enter into the binding pocket. In the bovine rhodopsin crystal structure, the extracellular loop II is closed down into the TM region with 11cis-retinal bound. With lilial bound in its vertical binding site, this loop cannot close in the same way, perhaps explaining why lilial does not activate the OR. This speculation that lilial may serve as a competitive antagonist was tested experimentally (see below) and found not to be the case.
Lyral to M-I7 and R-I7(hom) . We predict that lyral binds strongly, but it was not found experimentally to be an agonist. Indeed the predicted binding configuration for lyral is quite different than the observed agonists. The binding site is similar to the M-I7 octanal site with the exception that the aldehyde functional group of lyral is hydrogen bonded to Ser 280 while the alcohol functional group at the other end is hydrogen bonded to Lys 164 (see Figure 4F). Although this reversed binding site leads to a good predicted binding energy, its reversed orientation may be responsible for its inability as a positive agonist. This may indicate that strong binding to Lys 164 is necessary for activation. Thus lyral may serve as a competitive antagonist.
Summary of binding studies. To summarize the results on binding studies, we
used HierDock to predict the most probable binding site of octanal for the M-I7 and
R-I7(hom) structures and to predict the binding of all 62 odorants to this binding site.
The corresponding binding energies are shown in Table
5, where we find a good comparison
with the experiments. Again both M-I7 and R-I7 are predicted to bind both heptanal and
octanal. As discussed above, some of the experiments in literature had indicted that
heptanal activates M-I7 but not R-I7 while octanal activates R-I7 but not M-I7
(Krautwurst et al., 1998),
but later experiments (Bozza et al.,
2002) find that both lead to activation. The calculated binding energies also
agree with literature that nonanal and decanal activate I7 rat (Araneda et al., 2000;
Levasseur et al., 2003).
Lys164 forms a hydrogen bond with the aldehyde group of the aldehyde agonists, This was
also previously observed by
Singer (2000).
Agonists, antagonists, binding of mixtures
A difficulty in comparing the calculated binding energies directly to the experimental
activation data is that a strongly bound odorant could be an agonist (eliciting
intracellular calcium ion influx) or an antagonist (preventing activation of the OR).
However, most experiments on ORs detect only agonists. Consequently, we are particularly
concerned about comparing the calculated binding energies of odorants to experiments done
only on mixtures, since a mixture containing an antagonist might mask the activation by
an agonist (Cromarty and Derby, 1998).
There could also be cases where two ligands both interact with the same receptor, which
is outside the scope of our current studies. Antagonists could be sought experimentally
by competitive binding studies of suspected antagonists against known agonists. This
might identify OR inhibitors that could impair the detection of specific odorants. We
have compared the predicted binding energies only to the experimental agonists that have
been tested as individual odorants. For cases in which only mixtures were known to not
elicit activation of the ORs, we did not assume that the single components are
non-binders.
|
Proposed competitive experiments and verification |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
Based on the first generation of predicted structures (pre-RI7 and pre-MI7) for rat and mouse I7, we predicted three ligands: decanal, lyral and lilial to have good binding energies (within the top 10%) which were not observed agonists to these ORs. Since cinnamaldehyde (an observed agonist) was predicted to bind in a similar location and structure as lilial and lyral and with a similar binding energy, we speculated that lilial and lyral might be antagonists. Similarly heptanal (an observed agonist) is predicted to bind in location and structure similar to decanal and with a similar binding energy, but again decanal did not elicit activation response. At that stage, we proposed three experiments that could be done to test for competitive binding to M-I7: (i) cinnamaldehyde versus lilial; (ii) cinnamaldehyde versus lyral; and (iii) decanal versus heptanal.
Competitive activation assays for decanal and lilial
The proposed competitive experiments were carried out to test the above suggestions (Bozza, personal communication). Specifically, Bozza tested whether decanal or lilial can inhibit responses to the known I7 agonists heptanal or cinnamaldehyde, respectively.
In the new experiments on decanal using a variety of concentrations, it was found to
be an agonist but slower than heptanal or cinnamaldehyde. Concentration of the ligand can
affect the binding affinity (Levasseur et
al., 2003) and thus suggesting that nonanal and decanal are weaker
agonists to I7 rat (Araneda et al.,
2000).
However, the experimental results showed that lilial does not behave as a robust inhibitor of cinnamaldehyde for mouse I7 OR (Bozza, personal communication). Thus the predicted binding of lilial must be assumed to be a false positive, while experiments show that decanal does activate the receptor, as predicted by the theory.
|
Filtering false positives with moments of inertia |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
Since the competitive experiments suggested that lilial neither agonizes nor antagonizes I7, we suspected that there might be a size restriction on which ligands could bind and activate the I7 mouse OR, as also discussed in Araneda et al. (2000
). Such a
restriction might result from difficulties in the odorant successfully diffusing into the
binding site. To test this idea we calculated the moments of inertia for the final bound
structures of all the aldehydes to the M-I7 model (see Table
9). The moments of inertia were
calculated by assigning each atom with it’s atomic weight and then finding the axis
that correspond to the highest distribution of the mass of the ligand. These numbers
represent the general size of the ligand, since the larger the number the farther away
from the axis the density is. Since the agonists all have a small first moments of
inertia number, this means that the binding site prefers a long narrow shape opposed to a
rounder or fatter one. This correlates well with the observations that molecular length
is critical for rat I7 that are found in
Araneda et al. (2000). This
first small moments of inertia component is aligned with the long axis of the ligand. We
found that the two smaller moments of inertia for lilial and lyral are larger than those
for the odorants compounds observed to be agonists to M-I7. Indeed, Figure
5 shows that comparing these two
moments with the binding energy scores leads to a contour map (Figure
5) in which all false positives are
well separated from the true positives for the M-I7 profile.
|
|
Based on these results we defined the new scoring function in equation (1) that combines these two moments of inertia with the docking energy score. This equation was developed to fit the preferred moments of inertia trend observed in Figure 5. Those ligands observed to be in the right shape (by moments of inertia) were weighted to keep most of their original energy scores, while those that are farther from the right shape are given increasingly larger penalties to the original energy score. Sorting the aldehydes with this new weighted score puts all the observed agonists at the top (Table 9), plus it suggests that decanal is a weaker agonist. This provides an empirical relation that can be used to testing for new agonists.
|
|
where a is the weighted energy score, b is the bonding energy score, and c1 and c2 are the first and second moments of inertia respectively.
After determining that this expression works for aldehydes, we applied it to the
other molecules in the odorant library for both mouse and rat I7 (see Table
0). We found a good correlation to
experimental activation measurements (now including decanal as a positive agonist for
mouse I7). Thus we find that 100% of class A (eight compounds), 90%
(including the literature results for decanal and nonanal for I7 rat) of class B (10
compounds) and none of the poor binder classes were observed (106 compounds) to be
agonists. Of course, the use of such an empirical relation to predict the agonists is not
fully satisfactory. Thus we will continue to search for improved atomistic methods that
predict correctly the ligands that activate these receptors without the use of empirical
data.
|
|
Discussion |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
The correlation between the calculated binding energies and the measured experimental calcium ion influx suggests that the combination of experimental functional assays with OR structure prediction will make it possible to identify potential odors for other ORs. Even more important the knowledge of the detailed binding site suggests site-directed mutations experiments that would validate the predictions. Indeed the theory could be used to determine mutations that would increase the selectivity for particular odorants or even to modify the ORs to be selective against new odorants. Theoretical predictions provide an atomic level understanding of the odorant binding to ORs. This might be used to enhance the development of biosensors for the fragrance and food industries, industrial and environmental safety and explosives and narcotics detection.
Additional experiments to directly test the predictions made here would be most useful. Such comparisons could help develop knowledge based methods to predict the function of GPCRs in terms of pharmacaphore models that might accelerate the predictions of the response patterns of new odorants.
Proposed experiments
The in-depth analysis of the dimensions of the binding site of odorants in the final structure of M-I7 discussed above shows that Lys 164, Cys 117 and Ser 280 are main contributors to ligand binding. Indeed, the Lys 164 might well play an essential role beyond the binding mode we have studied. We suggest that Lys 164 might form a Schiff’s base with the aldehyde agonists just as is known to occur with 11cis-retinal in bovine rhodopsin. Thus the first step of noncovalent binding which we find to strongly prefer aldehyde, might position the aldehyde for a subsequent formation of the Schiff’s base, which could be responsible for the changes in conformation that lead to activation. If such chemical events play a special role in activation, it could have a significant impact on how we think about the binding in ORs and we strongly suggest experimental tests of these highly speculative suggestions. For example, mutating Lys164 to such polar groups as Arg or His might still bind an aldehyde but would not form a Schiff base. Similarly mutation to Asn or Gln or even Ser, Thr, or Tyr might still bind an aldehyde but would not accommodate the covalent attachment. This might explain the preference of I7 towards aldehydes.
Lys 164, Cys 117 and Ser 280 along with Phe 205 and Phe 109 cap the width of the
binding pocket, forming a pocket that is 8 Å long and 4 Å deep (see
Figure
6). Using this predicted binding
pocket, we now consider the design of novel odor agonists that should bind strongly to
the mouse I7 receptor and may lead to activation. We considered several multi-functional
potential ligands, which we subjected to the HierDock2.0 protocol. The best of these
suggested potential odorants (8-hydoxy-octanal) has two chemical functional groups, with
character very different than the known agonists for I7. 8-hydroxy-octanal has the
following strong interactions with I7: Lys 164 to the aldehyde functional group and Ser
280 to the alcohol group. Thus experiments on the binding of this compound would serve as
a good test on the value of the theory to predict binding and activity.
|
|
Summary and conclusions |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
We have used MembStruk2.0 and HierDock2.0 methods to predict the structures and odorant binding sites of 56 odorants in two closely related ORs: mouse and rat I7. The predicted binding site of odorants is located in TM domains 3, 4 and 6. In particular Lys164, Phe109, Cys114, Cys117 and Ile255 of TM4 are predicted to be involved in recognition of octanal and other aldehydes in the I7 receptor. This suggests that mutation experiments could be used to test further our predictions. Thus the mutation of Lys164 should dramatically change the recognition profile of M-I7 and R-I7.
The calculated binding energy of octanal and heptanal to both M-I7 and R-I7 are nearly equal and hence we predicted that both heptanal and octanal would activate the receptors. This was subsequently confirmed by experimental measurements on the intracellular calcium concentration influx. Also out of the top 10% of the calculated best binding odorants, 62% were observed to be agonists experimentally and out of the next 15% in the binding energy list, 33% were observed to be agonists. None of the bottom 75% of the worst binders was observed to be agonist. This provides an overall validation of the predicted structures for these proteins and of the methods.
The results presented here demonstrate significant progress toward predicting structure and function of olfactory receptors (and other GPCRs). Each of these predictions can be directly tested experimentally. Development of the atomistic structural models for ORs with specific binding requirements for specific odorants to provide information that could be valuable in making the connection between binding, processing to the cortex, to eventually perception and psychological response. Understanding these relationships could have significant impact on the fragrance and food industries and might be useful in developing artificial olfaction sensors.
Indeed as the accuracy of the predicted OR structures are validated, it should be practical and useful to predict the 3-D structures of all 913 mouse ORs and all 339 human ORs. Then it would be practical to predict the binding of large odorant libraries to all olfactory receptors to obtain overall binding profiles that could be most useful in tracing through the processing connecting molecular recognition to odorant recognition.
|
Supplementary material |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
Supplementary material can be found at: http://www.chemse.oupjournals.org/
|
Acknowledgements |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
We thank Dr Thomas Bozza and Professor Peter Mombaerts of Rockefeller University for their suggestion of the blind test on I7 and their assistance in this collaboration. Our special thanks to Dr Thomas Bozza for conducting some competitive assay experiments proposed by us. This work was initiated with support by ARO-MURI (Dr Robert Campbell) with some of the HierDock work supported by NIH (BRGRO1 andGM625523). The computational facilities used here were supported by an IBM-SUR grant and by an ARO-DURIP grant. Other facilities of the Materials and Process Simulation Center used in this project are supported also by DOE (ASCI ASAP), General Motors, ChevronTexaco, ONR, NSF (CHE and MRI), ARO, ONR, Beckman Institute, Seiko-Epson, NIH HD and Asahi Kasei.
|
References |
|---|
|
TopAbstractIntroductionResults and discussionProposed competitive experiments...Filtering false positives with...Discussion Summary and conclusionsSupplementary materialAcknowledgementsReferences |
|---|
Araneda, R.C., Kini, A.D. and Firestein, S. (2000) The molecular receptive range of an odorant receptor. Nat. Neurosci., 3, 1248–1255.[CrossRef][Web of Science][Medline]
Araneda, R.C., Mermet, N., Verjat, T., Angulo, J.F. and Radicella. J.P. (2001) Expression of Kin17 and 8-OxoG DNA glycosylase in cells of rodent and quail central nervous system. Brain Res. Bull., 56, 139–146.[CrossRef][Web of Science][Medline]
Bower, M., Cohen, F.E. and Dunbrack, R.L., Jr (1997) Prediction of protein side-chain rotamers from a backbone-dependent rotamer library: a new homology modeling tool. J. Mol. Biol., 267, 1268–1282.[CrossRef][Web of Science][Medline]
Bozza, T. and Kauer, J.S.
(1998) Odorant response properties of convergent olfactory
receptor neurons.
J. Neurosci., 18,
4560–4569.
Bozza, T., Feinstein, P., Zheng, C.
and Mombaerts, P. (2002) Odorant receptor
expression defines functional units in the mouse olfactory system.
J. Neurosci., 22,
3033–3043.
Buck, L. and Axel, R. (1991) A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell, 65, 175–187.[CrossRef][Web of Science][Medline]
Connolly, M.L. (1983)
Solvent-accessible surfaces of proteins and nucleic acids.
Science, 221,
709–713.
Cromarty, S.I. and Derby, C.D. (1998) Inhibitory receptor binding events among the components of complex mixtures contribute to mixture suppression in responses of olfactory receptor neurons of spiny lobsters. J. Comp. Physiol. A, 183, 699–707.[CrossRef][Medline]
Ding, H.Q., Karasawa, N. and Goddard, W.A., III (1992a) Atomic level simulations on a million particles: the cell multipole method for Coulomb and London nonbond interactions. J. Chem. Phys., 97, 4309–4315.[CrossRef]
Ding, H.Q., Karasawa, N. and Goddard, W.A., III (1992b) Atomic level simulations on a million particles: the cell multipole method for Coulomb and London nonbond interactions. Chem. Phys. Lett., 196, 6–10.[CrossRef][Web of Science]
Donnelly, D. (1993) Modelling alpha-helical transmembrane domains. Biochem. Soc. Trans., 21, 36–39.[Web of Science][Medline]
Duchamp-Viret, P., Chaput. M.A.
and Duchamp, A. (1999) Odor response properties of
rat olfactory receptor neurons.
Science, 284,
2171–2174.
Eisenberg D., Weiss R.M. and Terwilliger, T.C. (1984) The hydrophobic moment detects periodicity in protein hydrophobicity. Proc. Natl Acad. Sci. USA, 8, 140–144.
Ewing, T.A. and Kuntz, I.D. (1997) Critical evaluation of search algorithms for automated molecular docking and database screening. J. Comput. Chem., 18, 1175–1189.[CrossRef][Web of Science]
Filizola, M., Perez, J.J. and Carteni-Farine, M. (1998) BUNDLE: a program for building the transmembrane domains of G-protein-coupled receptors. J. Comput. Aided Mol. Des., 12, 111–118[CrossRef][Web of Science][Medline]
Floriano, W.B., Vaidehi, N., Singer,
M.S., Shepherd, G.M. and Goddard, W.A., III (2000)
Molecular mechanisms underlying differential odor responses of a mouse
olfactory receptor.
Proc. Natl Acad. Sci. USA, 97,
10712–10716.
Freddolino, P.L., Kalani, M.Y.,
Vaidehi, N., Floriano, W.B., Hall, S.E., Trabanino, R.J., Kam, V. and
Goddard, W.A., III (2004) 3-D structure for human b2
adrenergic receptor and the binding site for agonists and antagonist.
Proc. Natl Acad. Sci. USA, 101,
2736–2741
Gasteiger, J. and Marsili, M. (1980) Iterative partial equalization of orbital electronegativity––a rapid access to atomic charges. Tetrahedron, 36, 3219–3228.[CrossRef][Web of Science]
Ghosh, A., Rapp, C.S. and Friesner, R.A. (1998) Generalized Born model based on a surface integral formulation. J. Phys. Chem. B, 102, 10983–10990.[CrossRef]
Godfrey, P.A., Malnic, B. and
Buck, L.B. (2004) The mouse olfactory receptor gene
family.
Proc. Natl Acad. Sci. USA, 101,
2156–2161.
Grigorieff, N., Ceska, T.A., Downing, K.H., Baldwin, J.M. and Henderson, R. (1996) Electron-crystallographic refinement of the structure of bacteriorhodopsin. J. Mol. Biol., 259, 393–421.[CrossRef][Web of Science][Medline]
Jain, A., Vaidehi, N. and Rodriguez, G. (1993) A fast recursive algorithm for molecular dynamics simulation. J. Comp. Phys., 106, 258–268.[CrossRef]
Kajiya, K., Inaki, K., Tanaka, M.,
Haga, T., Kataoka, H. and Touhara, K. (2001).
Molecular bases of odor discrimination: reconstitution of olfactory
receptors that recognize overlapping sets of odorants.
J. Neurosci., 21,
6018–6025.
Kalani, M.Y., Vaidehi, N., Freddolino
P.E., Floriano W.B., Hall, S.E., Trabanino, R.J., Kam V. and Goddard,
W.A., III (2004) Structure and function of human
D2L receptor.
Proc. Natl Acad. Sci. USA, 101,
3815–3820.
Kiefer, H., Krieger, J., Olszewski, J.D., von Heijne, G., Prestwich, G.D. and Breer, H. (1996) Expression of an olfactory receptor in Escherichia coli: purification, reconstitution and ligand binding. Biochemistry, 35, 16077–16084.[CrossRef][Medline]
Krautwurst, D., Yau, K.W. and Reed, R.R. (1998) Identification of ligands for olfactory receptors by functional expression of a receptor library. Cell, 95, 917–926.[CrossRef][Web of Science][Medline]
Lancet, D., Sadovsky, E. and
Seidemann, E. (1993) Probability model for molecular
recognition in biological receptor repertoires: significance to the olfactory
system.
Proc. Natl Acad. Sci. USA, 90,
3715–3719.
Levasseur, G., Persuy, M., Grebert, D., Remy, J., Salesse, R. and Pajot-Augy, E. (2003) Ligand-specific dose–response of heterologously expressed olfactory receptors. Eur. J. Biochem., 270, 2905–2912.[Web of Science][Medline]
Lim, K.-T., Brunett, S., Iotov, M., McClurg, R.B., Vaidehi, N., Dasgupta, S., Taylor, S. and Goddard, W.A., III (1997) Molecular dynamics for very large systems on massively parallel computers: the MPSim program. J. Comput. Chem., 18, 501–521.[CrossRef][Web of Science]
Malnic, B., Hirono, J., Sato, T. and Buck, L.B. (1999) Combinatorial receptor codes for odors. Cell, 96,713–723.[CrossRef][Web of Science][Medline]
Malnic, B., Godfrey, P.A. and
Buck, L.B. (2004) The human olfactory receptor gene
family.
Proc. Natl Acad. Sci. USA, 101,
2584–2589.
Mathiowetz, A.M., Jain, A., Karasawa, N. and Goddard, W.A., III (1994) Protein simulations using techniques suitable for very large systems: the cell multipole method for nonbond interactions and the Newton–Euler inverse mass operator method for internal coordinate dynamics. Proteins, 20, 227.[CrossRef][Web of Science][Medline]
Mayo, S.L., Olafson, B.D. and Goddard, W.A., III (1990) DREIDING: a generic force field for molecular simulations. J. Phys. Chem., 94, 8897–8909.[CrossRef]
Mombaerts, P. (1999)
Seven-transmembrane proteins as odorant and chemosensory receptors.
Science, 286,
707–711.
Mori, K., Nagao, H. and
Yoshihara, Y. (1999) The Olfactory Bulb: Coding and
Processing of Odor Molecule Information.
Science, 286,
711–715.
Palcezwski, K., Kumasaka, T. Hori, T.,
Behnke, C., Motoshima, H., Fox, B., Trong, I., Teller, D., Okada, T., Stenkamp,
R., Yamamoto, M. and Miyano, M. (2000) Crystal
structure of rhodopsin: a G-protein-coupled receptor.
Science, 289,
739–745.
Pilpel, Y. and Lancet, D. (1999) The variable and conserved interfaces of modeled olfactory receptor proteins. Protein Sci., 8, 969–977.[Web of Science][Medline]
Poincelot, R.P., Millar, P.G., Kimbel, R.L, Jr and Abrahamson, E.W. (1970) Determination of the chromophoric binding site in native bovine rhodopsin. Biochemistry, 9, 1809–1816.[CrossRef][Medline]
Rappé, A.K. and Goddard, W.A., III (1991) Charge equilibration for molecular dynamics simulations. J. Phys. Chem., 95, 3358.[CrossRef]
Rubin, B.D. and Katz, L.C. (1999) Optical imaging of odorant representations in the mammalian olfactory bulb. Neuron, 23, 499–511.[CrossRef][Web of Science][Medline]
Schertler, G.F.X. (1998) Structure of rhodopsin. Eye, 12, 504–510.
Sicard, G. and Holley, A. (1984) Receptor cell reponses to odorants—similarities and differences among odorants. Brain Res., 292, 283–296.[CrossRef][Web of Science][Medline]
Singer, M.S. (2000)
Analysis of the molecular basis for octanal interactions in the expressed
rat I7 olfactory receptor.
Chem. Senses, 25,
155–165.
Singer, M.S., Weisinger-Lewin, Y., Lancet, D. and Shepherd, G.M. (1995a) Positive selection moments identify potential functional residues in human olfactory receptors. Receptors Channels, 4, 141–147.
Singer, M.S., Oliveira L., Vriend, G. and Shepherd, G.M. (1995b) Potential ligand-binding residues in rat olfactory receptors identified by correlated mutation analysis. Receptors Channels, 3, 89–95.[Web of Science][Medline]
Strader, C.D., Sigal, I.S. and Dixon R.A.F. (1989) Structural basis of beta-adrenergic receptor function. FASEB J., 3, 1825–1832.[Abstract]
Trabanino, R.J., Hall, S.E., Vaidehi, N., Floriano, W.B. and Goddard, W.A., III (2004) First principles predictions of the structure and function of G-protein coupled receptors: validation for bovine rhodopsin. Biophys. J., 86, 1904–1921.[Web of Science][Medline]
Vaidehi, N., Jain, A. and Goddard, W.A., III (1996) Constant tempeature constrained molecular dynamics: the Newton–Euler inverse mass operator method. J. Phys. Chem., 100, 10508.[CrossRef]
Vaidehi, N., Floriano, W.B., Trabanino,
R., Hall, S.E., Freddolino, P., Choi, E.J., Zamanakos, G. and Goddard,
W.A., II (2002) Prediction of structure and function of G
protein-coupled receptors.
Proc. Natl Acad. Sci. USA, 99,
12622–12627.
Vriend, G. (1990) WHAT IF: a molecular modeling and drug design program. J. Mol. Graph., 8, 52–56.[CrossRef][Web of Science][Medline]
Wetzel, C.H., Oles, M., Wellerdieck,
C., Kuczkowiak, M., Gisselmann, G. and Hatt, H. (1999)
Specificity and sensitivity of a human olfactory receptor functionally
expressed in human embryonic kidney 293 cells and Xenopus Laevis oocytes.
J. Neurosci., 19,
7426–7433.
Zamanakos, G. (2001) A fast and accurate analytical method for the computation of solvent effects in molecular simulations. PhD thesis, California Institute of Technology, Pasadena, CA.
Zhao, H., Ivic, L., Otaki, J.M.,
Hashimoto, M., Mikoshiba, K. and Firestein, S. (1998)
Functional expression of a mammalian odorant receptor.
Science, 279,
237–241.
Accepted June 17, 2004
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. S. Steiger, A. E. Fidler, J. C. Mueller, and B. Kempenaers Evidence for Adaptive Evolution of Olfactory Receptor Genes in 9 Bird Species J. Hered., December 4, 2009; (2009) esp105v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Reisert and D. Restrepo Molecular Tuning of Odorant Receptors and Its Implication for Odor Signal Processing Chem Senses, September 1, 2009; 34(7): 535 - 545. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Moreno-Estrada, F. Casals, A. Ramirez-Soriano, B. Oliva, F. Calafell, J. Bertranpetit, and E. Bosch Signatures of Selection in the Human Olfactory Receptor OR5I1 Gene Mol. Biol. Evol., January 1, 2008; 25(1): 144 - 154. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. A. Goddard III and R. Abrol 3-Dimensional Structures of G Protein-Coupled Receptors and Binding Sites of Agonists and Antagonists J. Nutr., June 1, 2007; 137(6): 1528S - 1538S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Abaffy, A. Malhotra, and C. W. Luetje The Molecular Basis for Ligand Specificity in a Mouse Olfactory Receptor: A NETWORK OF FUNCTIONALLY IMPORTANT RESIDUES J. Biol. Chem., January 12, 2007; 282(2): 1216 - 1224. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. McClendon, N. Vaidehi, V. W. T. Kam, D. Zhang, and W. A. Goddard III Fidelity of seryl-tRNA synthetase to binding of natural amino acids from HierDock first principles computations Protein Eng. Des. Sel., May 1, 2006; 19(5): 195 - 203. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Spijker, N. Vaidehi, P. L. Freddolino, P. A. J. Hilbers, and W. A. Goddard III Dynamic behavior of fully solvated beta2-adrenergic receptor, embedded in the membrane with bound agonist or antagonist PNAS, March 28, 2006; 103(13): 4882 - 4887. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. C. Lai, M. S. Singer, and C. J. Crasto Structural Activation Pathways from Dynamic Olfactory Receptor-Odorant Interactions Chem Senses, November 1, 2005; 30(9): 781 - 792. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||