Conformational Sensing by a Mammalian Olfactory Receptor

Article abstract of DOI:10.1002/chem.202001390

To identify odors, the mammalian nose deploys hundreds of olfactory receptors (ORs) from the rhodopsin-like class of the G protein-coupled receptor superfamily. Odorants having multiple rotatable bonds present a problem for the stereochemical shape-based

Full text of DOI:10.1002/chem.202001390

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
&
Sensory Mechanisms
Conformational Sensing by a Mammalian Olfactory Receptor
Min Ting Liu,[a, b] Mihwa Na,[a, c] Yadi Li,[a, b] Mark R. Biscoe,[a, b] and Kevin Ryan*[a, b, c]
Abstract: To identify odors, the mammalian nose deploys
hundreds of olfactory receptors (ORs) from the rhodopsin-
like class of the G protein-coupled receptor superfamily.
Odorants having multiple rotatable bonds present a prob-
lem for the stereochemical shape-based matching process
assumed to govern the sense of smell through OR–odorant
recognition. We conformationally restricted the carbon chain
of the odorant octanal to ask whether an OR can respond
differently to different odorant conformations. By using calci-
um imaging to monitor signal transduction in sensory neu-
rons expressing the mouse aldehyde OR, Olfr2, we found
that the spatial position of the C7 and C8 carbon atoms of
octanal, in relation to its ÀCHO group, determines whether
an aliphatic aldehyde functions as an agonist, partial agonist
or antagonist. Our experiments provide evidence that an
odorant can manipulate an OR through its intrinsic confor-
mational repertoire, in unexpected analogy to the photon-
controlled aldehyde manipulation observed in rhodopsin.
Introduction
work, we ask if an OR can be similarly manipulated, without
photon involvement, by the conformational sampling of an
odorant during OR binding. Understanding how ORs respond
to different conformers could help to explain how natural
product odorants with diverse carbon skeletons produce dis-
tinctive odor character, and how small variations in an odor-
ant’s carbon framework can change its odor quality.
Almost half of the approximately 800 human G protein-cou-
pled receptor (GPCR) genes encode olfactory receptors (ORs;
a.k.a. odorant receptors),[1–3] but despite continuing progress in
GPCR structure determination,[4] none of the mammalian OR
structures have been solved. Homology models abound but
structural knowledge of the ORs and odorant–OR recognition
is lacking, and thus the molecular recognition principles that
form the chemical basis of olfaction remain unknown. In lieu
of structural knowledge, research has been undertaken to cor-
relate molecular features with odor description,[5] but beyond
functional group recognition, the features responsible for acti-
vating individual ORs are unclear. In no case is the odorant–OR
recognition event harder to model or imagine than in that of
conformationally dynamic odorants having activity-defining
carbon chains. The activation of rhodopsin, which like the ORs
is a Class A GPCR, is governed by the photon-triggered cis–
trans isomerization of retinal, an extreme case of an (excited
state) conformational change during receptor binding.[6] In this
In the context of a homologous series, where the functional
group is kept constant, ORs have been observed to respond to
cognate odorants within a small range of successive carbon
chain lengths approximately centered on an optimally sized
odorant.[7–12] For example, the rodent aldehyde receptor OR-I7
(a.k.a. Olfr2, Olfr41, and MOR103-15; hereafter I7), which we
study here, responds at low micromolar concentrations to the
presence of aliphatic aldehydes having 7–10 carbon atoms
with, in most studies, the eight-carbon chain producing opti-
mal potency.[7,13–16] Considering that there is currently no struc-
tural biology information available to illuminate the molecular
details of odorant–OR complexes, the structural basis for this
trend is unknown. In addition, it is not only binding that needs
to be considered, as some shorter I7 odorant homologs bind
without activating the signal transduction process and thereby
function as competitive inhibitors, that is, antagonists.[15,17]
Among ORs studied individually thus far, there are few exam-
ples that describe in a systematic manner the molecular fea-
tures that differentiate activating odorant ligands, that is, ago-
nists, from partial agonists (sub-maximal activators) and antag-
onists.[15,17,18]
[a] M. T. Liu, Dr. M. Na, Dr. Y. Li, Prof. M. R. Biscoe, Prof. K. Ryan
Department of Chemistry and Biochemistry
The City College of New York
New York, NY 10031 (USA)
E-mail: kryan@ccny.cuny.edu
[b] M. T. Liu, Dr. Y. Li, Prof. M. R. Biscoe, Prof. K. Ryan
Ph.D. Program in Chemistry
The Graduate Center of the City University of New York
New York, NY 10016 (USA)
Rodents have over 1000 genetically distinct OR family mem-
bers, whereas humans have approximately 390.[19,20] Each
mature olfactory sensory neuron (OSN) expresses one OR
family member,[11,21–23] which consequently determines the ol-
factory response, that is, pharmacology, of that OSN. Only a
few recombinant ORs have been studied in the neuronal cell
environment. The rat I7 (rI7) receptor was the first recombinant
[c] Dr. M. Na, Prof. K. Ryan
Ph.D. Program in Biochemistry
The Graduate Center of the City University of New York
New York, NY 10016 (USA)
Supporting information and the ORCID identification number(s) for the au-
thor(s) of this article can be found under:
https://doi.org/10.1002/chem.202001390.
Chem. Eur. J. 2020, 26, 1 – 9
1
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
OR to be expressed in OSNs and the first to be deorphaned,
that is, matched with odorant agonists.[16] In that case, the rI7
expression was engineered through an adenoviral vector. In a
second recombinant system, the genetically modified UbI7
mouse, which we use here, the mouse I7 (mI7) orthologue was
put under the control of the olfactory marker protein (OMP)
transcriptional promoter.[24,25] The OMP gene is normally ex-
pressed in OSNs throughout the olfactory epithelium. In the
UbI7 mouse, the I7 mRNA is also expressed in OSNs through-
out the olfactory epithelium, though at low OR protein
levels.[26] Due to the properties of the OMP gene promoter and
expression program, not all UbI7 OSNs express I7 at high
enough levels to be functional. According to electrophysiogi-
cal[27] and calcium imaging[28] recordings, the UbI7 mouse
shows an increase in the percentage of OSNs having a func-
tional I7 response profile. The subset of cells expressing func-
tional I7 can be identified for recombinant I7 study by using
the I7 agonist octanal (in combination with the non-binding 1-
octanol, see below) as ligands. The I7 receptor has been suc-
cessfully studied in other contexts, including heterologous
cells,[17] and it has become one of the best-studied and proto-
typical mammalian ORs.
Results
Odorant design and synthesis
In previous work, we found evidence that conformationally re-
stricting the octanal C3ÀC4, C4ÀC5, and C5ÀC6 bonds by join-
ing C3 to C8, to make cyclohexylethanal (2) (Figure 1) convert-
ed octanal into an I7 antagonist (competitive binding without
activation).[15,17] Aldehyde 2 holds the octanal chain in a re-
verse-turn conformation. This result led to the hypothesis that
the octanal carbon chain must unfurl to at least 6.9 ꢁ to posi-
tion the far end of the carbon chain (C7 and C8) in a receptor
pocket where contact is required for G protein coupling and
activation. In a preliminary experiment,[15] we tested this hy-
pothesis by adding a two-carbon extension to antagonist 2, to
make 2-(4-ethylcyclohexyl)ethanal (3). Compound 3, made and
tested as a 2:1 isomer mixture (tentatively identified as 2:1
trans/cis), was slightly more potent than octanal and showed
similar efficacy (the proportional extent of activation). This
result suggested that the location of the C7 and C8 atoms of
octanal, mimicked by the 2-carbon extension, is the principal
determinant for receptor activation. However, the response to
the individual isomers in the mixture could not be measured.
Importantly, this result showed that the cyclohexyl ring in the
middle of the odorant was accommodated by the activated
form of the receptor and confirmed that the last two carbons
of octanal must occupy a distal position in relation to the alde-
hyde group for activation.
Although the rodent I7 receptor is an aldehyde receptor, re-
quiring this functional group for all ligands,[7,15] the features of
the carbon chain determine whether cells expressing I7 are ac-
tivated by an aldehyde. A simple formula has emerged to pre-
dict whether a bound aldehyde will function as an I7 agonist
or antagonist: aliphatic chains or rings composed of five or
more carbon atoms, but too short to attain a length of approx-
imately 6.9 ꢁ, from carbonyl to farthest carbon in an extended
conformation, bind without activation, whereas those extend-
ing beyond this threshold length bind with activation.[7,15,17]
Activating ligands are subject to a maximum length,[7,14,16] and
certain a,b-unsaturated aldehydes longer than 6.9 ꢁ bind with-
out activation.[7,8] Overall, these observations are consistent
with a model in which the aldehyde group anchors the ligand
through a polar interaction with I7, but also in which the
carbon chain must unfurl to beyond 6.9 ꢁ to trigger signal
transduction through I7. No study has yet precisely controlled
the spatial relationship and path between the aldehyde group
and the far end of the optimal eight-carbon chain, and thus no
study has systematically probed for agonist and antagonist
conformation(s) that satisfy the 6.9 ꢁ minimum distance re-
quirement. Here, we use the medicinal chemistry technique of
conformational restriction[29] to control the spatial relationship
between the aldehyde and the last two carbon atoms of the
octanal carbon chain, and assess the outcome in UbI7(+/+)
OSNs expressing mI7. In combination with previous studies,
our results support the hypothesis that after an aliphatic alde-
hyde binds to I7, the receptor evaluates the ligand’s attainable
conformations, and then uses this assessment as the criterion
for signal transduction. Thus, we provide evidence that some
ORs sense and report on the conformational repertoire of their
odorant ligands.
Figure 1. Conformationally restricted mimics of octanal. A) Compounds used
in this study. Red denotes bonds added to restrict the octanal conformation.
B) Newman projections looking down the C5ÀC4 bond in compounds 2 and
3 with ÀCH2CHO occupying an equatorial position. The various chair confor-
mations are assigned arbitrary letters, for example, T and U. C) Correspond-
ing Newman projections for octanal (see Figure S1D in the Supporting Infor-
mation for projections where ÀCH2CHO is axial).
&
&
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
2
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
Because the cyclohexyl group is in this structural context ac-
commodated by the activated[15] and unactivated[15,17] forms of
I7, it can be used to control the spatial relationship between
the aldehyde end (ÀCH2CHO) and the C7/C8 end (ÀCH2CH3),
both of which are needed by octanal for I7 activation. Each 3-
cis and 3-trans chair conformation is ꢀ6.9 ꢁ long and so
mimics a possible activating octanal conformation (the four
chair conformations are labeled arbitrarily as T, U, M, and L in
Figures 1 and 2, and Figures S1 and S2 in the Supporting Infor-
chair conformations ranked by relative conformational energy).
Although the intrinsically lowest-energy ligand conformation
need not be the activating conformation in any ligand–recep-
tor pair, as long as complex formation overall is favorable, in
our view the approximately 30 kJmolÀ1 energy penalty to
adopt the lowest-energy boat conformation seems unlikely in
3-trans and 3-cis. In addition, given the low affinity (ꢁ1 mm
EC50) of the octanal-I7 pair and the presence of only a single
polar functional group, the octanal conformations analogous
mation). For example, the two chair conformations having the to the twist boat conformations are also unlikely, due to
ÀCH2CHO in the equatorial position are shown in Figure 1B,
having higher energies than those corresponding to the chair
and the corresponding octanal conformations are shown in
conformations (Figure S2). The synthesis of 3-cis and 3-trans is
Figure 1C. These two chair conformations mimic different C5À outlined in Scheme 1 and described in detail in the Supporting
C6 rotamers that have different ÀCH2CH3 locations. (Figure S1
in the Supporting Information depicts the two chair conforma-
tions having the ÀCH2CHO in the axial position, along with the
corresponding octanal conformations they mimic.) The value
of using the cyclohexyl ring to position the two ends of the
octanal mimics is that when the ÀCH2CHO and the cyclohexyl
chair carbons of the 3-cis and 3-trans align (i.e., superpose),
the ÀCH2CH3 groups cannot align, such that each isomer
mimics an octanal conformation differing only in the place-
ment of the ÀCH2CH3 (Figure 2). Likewise, if the ÀCH2CH3 and
cyclohexyl groups align during binding, then the ÀCH2CHO
groups cannot align, and the mimics represent octanal confor-
mations differing only in the placement of the ÀCH2CHO. Con-
sidering that the aldehyde presumably engages in a polar in-
teraction with I7, it is more likely that binding I7 requires a
common alignment of the cyclohexyl ring and ÀCH2CHO, and
it is the placement of the ÀCH2CH3 that determines the I7 re-
sponse in our mimics. However, this need not be the case for
us to attribute a difference in the I7 response to a difference in
carbon chain conformation. In this analysis we make the rea-
sonable assumption that I7 activation does not require octanal
to adopt the higher energy conformations mimicked by the
twist-boat conformations of 3, two of which are shown in Fig-
ure S2 in the Supporting Information (along with all possible
Information.
Scheme 1. Synthetic route to 3-trans and 3-cis. Reagents and reaction con-
ditions: (a) NaH, THF, rt. (b) L-selectride, THF, À758C to À408C to rt. (c) LAH,
ether, 08C to rt. (d) PCC, CH2Cl2, rt. (e) LAH, ether, 08C to rt. (f) PCC, CH2Cl2,
rt. (g) 1. n-BuLi, PH3PCH3Br, THF, 08C to À758C; 2. BH3·THF, THF, rt.
(h) Na2CO3, H2O2, H2O, rt. (i) PCC, CH2Cl2, rt.
Preliminary testing with sensory neurons from the
UbI7(+/+) mouse
To assess the I7 response to octanal and the ana-
logues in sensory neurons, we used OSNs from the
UbI7(+/+) mouse, where both copies of the OMP gene
promoter induce transcription of the mI7 mRNA, driv-
ing transcription of the mI7 transgene throughout
the olfactory epithelium. To monitor I7 activation we
used ratiometric calcium imaging on OSNs dissociat-
ed from olfactory epithelium tissue.[28] Changes in in-
tracellular calcium levels resulting from OSN activa-
tion were monitored at ligand concentrations of 3,
10, and 100 mm. The EC50 of the I7 receptor for octa-
nal is approximately 1–2 mm so saturation should be
approached between 3 mm and 10 mm, and 100 mm
was used to represent complete saturation. In order
to score a cell as positive, and to avoid ambiguous
responses in cells with low signal-to-noise ratios, we
Figure 2. A) Depiction of the four different 3-trans and 3-cis chair conformations. B) Su-
perposition of each 3-trans with each 3-cis chair shows that when these two compounds
are tested on I7, one end, either ÀCH2CH3 or ÀCH2CHO, will occupy a different spatial po-
sition (arrows), and thereby mimic a different octanal conformation. Different I7 activity
will be attributable to this, the only difference between the two compounds.
set a stringent minimum threshold criterion of 10%
of the cell’s forskolin response, the maximum re-
sponse the cell can mount through the cAMP signal
transduction system used by OSNs.[30]
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
3
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
We first confirmed that, as observed elsewhere,[27,28] there is
a large increase in the number of OSNs responding to octanal
in UbI7(+/+) cells (e.g., 10.8% at 3 mm octanal compared to
1.4% for OSNs from wild-type (wt) mice, Figure 3A), and that
the difference was greater at lower concentrations where li-
gands of low potency and selectivity are less likely to produce
widespread activation (Figure S3).[28] Based on this comparison,
we estimate that at 3 mm octanal, 87% ([10.8–1.4]/10.8) of the
responding UbI7(+/+) OSNs were activated through the re-
combinant I7 rather than through the unidentified endoge-
nous OR that continues to be expressed by the OSN. Neverthe-
less, because the UbI7(+/+) OSNs continue to express an endog-
enous OR,[24] and a small percentage of these will by chance
be non-I7 ORs that respond to octanal, in the analysis below
we took an extra precaution to minimize the number of cells
responding through an endogenous non-I7 octanal OR. The
rodent I7 ORs are not activated by 1-octanol, while the majori-
ty of other octanal ORs are activated by 1-octanol.[7,15,16,31] For
instance, in one sampling of wt mouse OSNs, we found that
68% of wt OSNs responding to octanal responded in a similar
manner to 1-octanol.[31] Assuming that about 87% of the re-
sponding OSNs respond through I7, and about 13% respond
through a different octanal OR, by removing cells that respond
to 1-octanol we further enrich for cells responding through I7.
We cannot know exactly how many of the responding cells are
responding through I7, but it will be greater than 87%. What
is important for our experiments is that while the identity of
the responding OR cannot be identified with certainty in indi-
vidual UbI7(+/+) cells, when assessed in aggregate, greater than
87% of the octanal-responding cells respond through the I7
receptor. We consider this uncertainty to be part of the cost of
observing the effect of our analogs in individual neurons,
which provides maximal physiological relevance.
Octanal analogues 3-cis and 3-trans are I7 agonists with
different properties
Octanal, 3-cis, and 3-trans were applied individually at 3, 10,
and 100 mm to OSNs dissociated from a UbI7(+/+) mouse olfac-
tory epithelium, and their ratiometric calcium imaging respons-
es were recorded. The order of application of the three com-
pounds was varied but the order of concentrations applied
was in all cases from lowest to highest. Forskolin (10 mm) was
tested at the end of the experiment to identify the OSNs that
survived the entire experiment, and to estimate each cell’s
maximum response (i.e. 100% receptor efficacy) to allow for
normalization of all odorant responses to the forskolin re-
sponse. As described above, we also tested 1-octanol just prior
to forskolin to identify and remove from the analysis cells that
according to this criterion show evidence of responding
through an endogenous octanal receptor that is not I7. In
total, 1852 forskolin-positive OSNs were observed, and 309
gave a positive response to at least one of the three com-
pounds at one of the concentrations. After identification and
removal of the 1-octanol-responding cells (96 cells), it was
found that 173 cells responded to at least one compound at
3 mm, 182 cells responded at 10 mm, and 213 cells responded
at 100 mm. Incidentally, of the 96 cells removed because they
responded to 1-octanol, 87 cells (91%) also responded to octa-
nal, consistent with a recombinant I7-mediated response in
many of these cells. While many of these 87 cells are respond-
ing through I7 they also showed evidence of responding
through an endogenous OR that could interfere with our octa-
nal-specific experiment (in the great majority of cells, the OR
Figure 3. Survey of UbI7(+/+) mouse olfactory sensory neurons (OSNs) re-
sponding to octanal, 3-trans, and 3-cis. A) Percentage of cells that respond
to octanal at 3, 10, and 100 mm for wild-type mice (n=3) and UbI7 mouse
(n=1). The percentages were calculated relative to the number of forskolin-
responding cells (wt: 9446, UbI7: 1852). B) Calcium imaging trace of a repre-
sentative UbI7(+/+) OSN. The peak response height is color-coded according
to the heatmap scale shown in Figure S4 in the Supporting Information and
placed here above the trace to show correlation with the heatmap presenta-
tion. This cell is number 49 in the heatmap figure. C) Percentage of UbI7(+/+)
OSNs responding to the three odorants at 3 mm. A total of 173 OSNs out of
1852 forskolin-responding cells (minus 96 cells that responded to 1-octanol,
evidence of a non-I7 response) responded to at least one of the odorants.
Cells are grouped by the combination of odorants to which they responded.
Response must be ꢀ10% of the forskolin response to be scored as positive.
The number of responding cells in each group is shown in parentheses.
&
&
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
4
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
will express an endogenous OR whose pharmacology is unre-
lated to our experimental design and can be ignored.)
conformational mimics in each cell, we limited this analysis to
the OSNs that showed a response to all three aldehydes. Since
the EC50 of octanal is about 2 mm,[8,31] cells that did not respond
to octanal by 10 mm (4 cells) were omitted from this analysis
resulting in a total of 120 cells. The cell-by-cell responses of
these cells are shown in heatmap format relative to the forsko-
lin response in Figure S4. The responses of each of the
120 cells to each ligand at each concentration are shown in
box plot format in Figure 4A. The median response is marked
by the line between the 2nd and 3rd quartiles, and the aver-
age is marked by a diamond. Considering that the amount of
functional I7 will vary from cell to cell, and a small percentage
of responding cells will be responding through an endogenous
OR that is not I7, we draw our conclusions about the behavior
The calcium imaging response of a representative UbI7(+/+)
OSN from this experiment is shown in Figure 3B. A summary
of the results for all responding UbI7(+/+) OSNs at the 3 mm
concentration is shown in Figure 3C. Results at 10 and 100 mm
odorant concentrations are shown in Figure S3. The profiles at
the two higher concentrations were similar to the 3 mm results,
but because higher odorant concentrations diminish the differ-
ence between wt and UbI7(+/+) cells, and hence, diminish the
percentage of cells responding solely through mI7, we focus
on the 3 mm experiment. At 3 mm odorant, approximately half
(47.4%) of the responding cells scored a positive response (ꢀ
10% forskolin response) to octanal, 3-cis, and 3-trans, indicat-
ing that both of the conformationally restricted analogs are
mI7 agonists (the preliminary study only indicated that at least
one of them was an agonist[15]). Among the other cells scoring
a positive response to at least one of the odorants at 3 mm,
34.1% responded to both octanal and 3-trans, but not 3-cis.
Unexpectedly, 16.8% of cells responded to 3-trans alone, and
not to octanal or 3-cis. The sum of these three profiles ac-
counted for over 98% of the cells responding to at least one
odorant. The percentages for the last two categories are too
large to ascribe solely to cells that responded through an en-
dogenous aldehyde OR that is not I7; rather, some of these
cells may have very low levels of functional I7 such that only
the more potent or efficacious ligands produce a response
above threshold. If true, then at higher concentrations, where
more and more cells become activated through their endoge-
nous OR, and the distinction between wt and UbI7(+/+) cells di-
minishes, we would expect this percentage to decline, and
that is what was observed: 3-trans-only declined from 16.8 to
13.2 and 8.9%, at 3, 10, and 100 mm, respectively. The 3 mm
concentration is already 3-fold higher than the I7 EC50, such
that cells expressing a functional I7 should register a>50% re-
sponse. However, the I7 expression level in UbI7(+/+) cells is
lower than what is normal for endogenous I7.[24,26] Additionally,
in UbI7(+/+) OSNs the OMP promoter turns I7 expression on
late in the developmental program such that dissociated OSNs
not yet mature may be expressing little or no I7.[32] These fac-
tors, in addition to the heterogeneity in OSN populations ex-
pressing the same OR,[33] general OR transgene effects,[34] and
the fragile nature of live dissociated neurons, likely all contrib-
ute to cell-to-cell variations in the I7 response magnitude and
prevent the expression of optimally functional I7 in all dissoci-
ated UbI7(+/+) OSNs. Nevertheless, counting the cells showing
a positive response to the different odorants supported the
conclusion that 3-trans is a better I7 agonist–either more
potent, more efficacious or both–than octanal, and 3-cis is not
as potent (or efficacious) as octanal.
Figure 4. Estimation of 3-trans and 3-cis I7 efficacy. A) Boxplot showing the
UbI7 OSN response to three odorants at three concentrations. Lines separat-
ing Quartiles 2 and 3 are the medians; diamonds are the average response
over all cells. Only the 120 cells responding to all three compounds are in-
cluded. Dotted horizontal line indicates the forskolin response, taken to be
100% efficacy. Six hyperstimulated cells fell outside of the 200% chart area
setting but were included in the statistics. B) Bar chart showing the average
response of the responding cells according to odorant and concentration.
C) A plot of the average response versus the concentration showing that 3-
trans is the most potent and efficacious compound. Octanal and 3-cis have
similar potency but 3-cis is a partial agonist.
OR-I7 ligand efficacy: 3-trans>octanal>3-cis, a partial
agonist
We next analyzed the odorant efficacy, the degree of the cell’s
response compared to maximum activation. Since our goal
was to compare the efficacy produced by octanal and its two
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
5
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
of I7 from the averaged response of the cells after normaliza-
tion to forskolin (Figures 4B and C).
tion. In addition, the terpenoids neral and geranial, whose
double bonds impose conformational constraints different
than those studied here, have eight-carbon chains that can
extend beyond 6.9 ꢁ, yet both bind I7 without activation.[7,8]
What then is the critical I7 activation determinant built into
the carbon framework of its aldehyde agonists, and that the
receptor has evolved to detect? The available data suggested
to us that this determinant is the attainment of a preferred
conformation that orients the ligand within the receptor to
spatially align the two ends of octanal—and the path of the
carbons in between—in such a way as to stabilize the activat-
ed form of I7. We examined the octanal conformations that
can keep its ends ꢀ6.9 ꢁ apart and made 3-cis and 3-trans as
conformational mimics of the octanal chain.
It can be seen in Figure 4C that the octanal and 3-cis re-
sponse averages reached a plateau between 3 and 10 mm, con-
sistent at least in the octanal case with that ligand’s I7 EC50 of
about 1 mm.[8,31] Unexpectedly, 3-trans had reached its maxi-
mum response by 3 mm, possibly indicating a lower EC50, but
because we did not test lower concentrations we could not es-
timate the EC50 value. Strikingly, the 3-trans efficacy was 33%
higher than octanal’s at 3 mm (p value <0.00002, Welch’s t-
test), indicating it is a significantly more efficacious agonist
than octanal, which is generally considered to be a full I7 ago-
nist. In contrast, at 3 mm, the 3-cis efficacy was only about half
that of octanal (p value <0.00001). Even at the highest con-
centration, 100 mm, 3-cis never rose above 50% efficacy,
though the curve shape indicates an EC50 similar to that of oc-
tanal (Figure 4C). This behavior supports the conclusion that
3-cis is an I7 partial agonist.
The different responses of I7 to the two octanal conforma-
tional mimics demonstrate that I7 is sensitive to the spatial re-
lationship between the first two and last two carbon atoms of
octanal, here held apart by the cyclohexyl ring. In octanal, this
relationship differs dynamically as a function of conformation
along the chain. In 3-cis and 3-trans, our use of the cyclohexyl
ring limits this relationship to two different ways in the ana-
logues, thereby showing sensitivity to the placement of the
ethyl group in relation to the aldehyde end. By analogy, we
conclude that I7 must sense the conformational repertoire of
the carbon chain of octanal, but the different responses only
become evident when the chain is locked into discrete confor-
mations. Otherwise, what is observed must be a composite of
octanal’s activating and partially activating conformations—its
conformational repertoire.
The octanal feature that we propose is the determinant of I7
activation—the conformational alignment of its ends and the
path of the carbons in between—falls into the category often
referred to as odorant stereochemical “shape”.[35–37] The well-
studied Class A GPCR rhodopsin, composed of the opsin pro-
tein and its covalently bound aldehyde chromophore, 11-cis-
retinal,[6] provides a model for understanding how ligand
shape can govern olfactory GPCR activation. The 11-cis-retinal
is an inverse agonist: it not only antagonizes receptor activa-
tion by binding to the inactive form of the receptor, but also
suppresses the rate of rare ligand-independent background ac-
tivation events that most GPCRs can undergo. Absorption of a
photon causes 11-cis-retinal to isomerize to all-trans-retinal.
This sharply defined shape change converts the bound chro-
mophore from inhibitor to activator (agonist) and switches the
receptor to its activated form to initiate signal transduction. At
elevated temperature in the dark, opsin-bound retinal can un-
dergo infrequent thermal isomerization and activation without
hydrolysis of its Schiff base linkage to the protein, and without
protein denaturation.[38,39] With or without light, rhodopsin’s
ligand can change shape to switch the receptor to its activated
form.
Discussion
The I7 receptor is one of the most studied and modeled repre-
sentatives of the olfactory GPCR subfamily. Accumulating evi-
dence shows that I7 contributes to the olfactory code[11] by
binding aliphatic aldehydes with at least four carbon atoms
and the ability to adopt a binding conformation that does not
exceed 10–12 ꢁ in length. Although many natural and synthet-
ic aldehydes meet these criteria and bind I7, only a subset acti-
vate the receptor to trigger signal transduction. Thus, whereas
the aldehyde functional group is required for activation,[7] it is
not sufficient, and some feature of the carbon chain provides
the critical determinant for receptor activation. That feature
was initially described as “length”, because 7–11 carbon alka-
nals activate I7, and potency reaches a peak at octanal.[7,14,16]
Compound 2, where octanal’s carbon chain is held in a re-
verse-turn conformation by connecting C8 to C3, supports
length, or rather maximum attainable length, as a determinant
of activation because preventing the conformational unfurling
of the eight-carbon chain prevented I7 activation but permit-
ted binding.[15,17] In odorants with multiple rotatable bonds,
the attribute of length as a determinant of activation has little
meaning because length changes dynamically with conforma-
Our results support the hypothesis that I7 is manipulated by
its ligands analogously, though in a light-independent manner.
Unable to control the shape of octanal with photons, we con-
formationally restricted the middle of the structure using a cy-
clohexyl group previously found to be compatible with I7 ago-
nists and antagonists. This strategy produced two isomeric an-
alogs, 3-cis and 3-trans, that mimic distinct octanal conform-
ers. No matter which chair conformation the cyclohexyl ring
adopts in the cis and trans isomers, one of the two ring sub-
stituents, ÀCH2CHO or ÀCH2CH3, both required for I7 activation,
always occupies a different spatial position, and hence each
isomer mimics a different octanal conformation. Our experi-
ments monitored the response of I7-expressing OSNs to the
two isomers and produced evidence that, while they both
bind I7, the receptor response changed with the position of
the two carbons acting as surrogates for octanal’s C7 and C8
carbon atoms. If we assume that while bound to I7 the
CH2CHO end occupies the equatorial position in both isomers,
and also in analogue 2, where the last two carbons fold back
to become part of the cyclohexyl ring then, in analogy with oc-
tanal, rotation of the C5ÀC6 bond of octanal (Figure 1C)
&
&
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
6
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
sweeps the C7 and C8 carbon atoms of octanal from the inac-
tive conformation mimicked by antagonist 2, through the posi-
tion occupied in the partial agonist 3-cis, and finally into the
position mimicked by the full agonist 3-trans (the same con-
clusion results if the ÀCH2CHO is axial). Thus, we have three
conformational mimics, differing only in their C7ÀC8 surrogate
locations, and three graded responses from inactive to better
than full octanal activity. Tested individually, and considered to-
gether, these three compounds provide the I7 equivalent to
flipping the retinal conformation using photons to control rho-
dopsin activation. Despite the comparatively weak, non-cova-
lent, and light-independent nature of odorant-OR binding, our
results show that an OR can be manipulated by its ligand’s
conformational shape similar to the way retinal isomerization
manipulates rhodopsin.
duce flexibility, like the seco approach to odorant modifica-
tion,[41] where pieces of an odorant are excised, may owe some
of its effect to alteration of the conformational landscape.
According to observations embodied in the olfactory
code,[11] whose experimental basis included testing homolo-
gous series of acids, aldehydes, and alcohols,[8–12] an OR will re-
spond to multiple odorants, but the members in a series are
usually not equipotent. In most cases an optimal chain length
was found. The findings in the alcohol and acid series are con-
sistent with our I7 aldehyde conclusions that ORs have a pre-
ferred optimal-length agonist conformation, but will bind less
strongly, or be activated less efficaciously, to others in the
series if those incur a greater energetic or steric cost to mimic
the structure of the preferred conformation.
According to our I7 results, odorants with rotatable bonds
have conformational repertoires, among which some conform-
ers have shapes that better stabilize the activated form of the
OR and thus by which the OR senses the ligand. Rhodopsin
can be viewed as acting like an OR that has evolved to cova-
lently bind an aldehyde having one discrete, light dependent
conformation change. The two variations on the same mecha-
nism suit their different physiological roles.
Conclusions
The molecular recognition principles that form the chemical
basis of olfaction remain unknown. Stereochemical shape is as-
sumed to furnish the raw material for recognition, but most
odorants have rotatable bonds and thus, no single discernable
size or shape. We locked a common odorant chemical into dif-
ferent conformations, and found distinct conformation-depen-
dent olfactory receptor outputs, reminiscent of the changes
leading to rhodopsin activation by (photon-triggered) retinal
isomerization. Our experiments support the hypothesis that an
odorant’s conformational repertoire is among the intrinsic mo-
lecular properties by which an olfactory receptor knows its
specific odorants.
For this work, we chose the ligand-based perspective on ac-
tivation because there is yet no structural data on I7 or any
other olfactory GPCR. Based on our previous rhodopsin-based
homology model, the carbon chains likely engage in subtly dif-
ferent interactions with the hydrophobic residues of TM3–TM6,
such as F205, F262, Y264, and G111.[14]
Our experiments show that I7 monitors and reports on the
conformation of its ligands. Octanal (in one study, heptanal[13]
)
Acknowledgements
is the optimal n-alkanal I7 agonist; hexanal, nonanal, and dec-
anal are less potent.[7] We speculate that each of these agonists
must satisfy the same requirement for positioning the agonist’s
ends within the OR binding site, but the path taken by the
chain of carbon atoms in between is necessarily different.
Lower potency or efficacy will result from less favorable steric
interactions with receptor residues lining the middle of the
binding site, or a binding energy penalty for the odorant to
adopt a high energy conformation to position the ends. By
this reasoning and, given the weak binding affinities most
odorants have for their ORs, we predict it is unlikely that flexi-
ble odorants like octanal adopt intrinsically high energy con-
formations to activate their ORs. In the I7 case, our data sup-
port the hypothesis that the octanal conformation shown in
Figure 1C, analogous to that of the 3-trans U-chair, resembles
the optimal I7-activating conformation. Odorants pre-organ-
ized[40] in favorable conformation(s), that is, conformations pre-
ferred by an activated OR, should lose less entropy in attaining
the best conformer for binding the activated form of the OR,
and therefore be more potent than more flexible agonists. This
reasoning suggests an explanation for the greater I7 potency
and efficacy of 3-trans in comparison to octanal. More general-
ly, conformational restriction may provide a means to alter the
potency of odorants with rotatable bonds. There is evidence
that reducing odorant flexibility reduces the number of ORs ac-
tivated at a given concentration.[15] And approaches that intro-
This work was supported by the, U.S. Army Research Laborato-
ry and the, U.S. Army Research Office under grant number
W911NF-13-1-0148 to K.R. Additional infrastructural support
was provided through grant 3G12MD007603-30S2 from the
National Institute on Minority Health and Health Disparities.
M.R.B. acknowledges support from National Institutes of
Health grant R01GM131079. We thank Dr. L. Belluscio (NIH) for
providing UbI7 mice.
Conflict of interest
The authors declare no conflict of interest.
Keywords: GPCR · odor coding · olfactory receptors · sensory
mechanisms · signal transduction
[1] E. Block, J. Agric. Food Chem. 2018, 66, 13346–13366.
[2] C. A. de March, Y. Yu, M. J. Ni, K. A. Adipietro, H. Matsunami, M. Ma, J.
Golebiowski, J. Am. Chem. Soc. 2015, 137, 8611–8616.
[3] D. E. Gloriam, R. Fredriksson, H. B. Schioth, BMC Genomics 2007, 8, 338.
[4] W. I. Weis, B. K. Kobilka, Annu. Rev. Biochem. 2018, 87, 897–919.
[5] A. Keller, R. C. Gerkin, Y. Guan, A. Dhurandhar, G. Turu, B. Szalai, J. D.
Mainland, Y. Ihara, C. W. Yu, R. Wolfinger, C. Vens, L. Schietgat, K. De
Grave, R. Norel, G. Stolovitzky, G. A. Cecchi, L. B. Vosshall, P. Meyer, Sci-
ence 2017, 355, 820–826.
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
7
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
[6] S. T. Menon, M. Han, T. P. Sakmar, Physiol. Rev. 2001, 81, 1659–1688.
[7] R. C. Araneda, A. D. Kini, S. Firestein, Nat. Neurosci. 2000, 3, 1248–1255.
[8] R. C. Araneda, Z. Peterlin, X. Zhang, A. Chesler, S. Firestein, J. Physiol.
2004, 555, 743–756.
[29] M. P. P. de Sena M. Pinheiro, D. A. Rodrigues, R. do Couto Maia, S. Thota,
C. A. M. Fraga, Curr. Top. Med. Chem. 2019, 19, 1712–1733.
[30] K. Otsuguro, S. H. Gautam, S. Ito, Y. Habara, T. Saito, J. Pharmacol. Sci.
2005, 97, 510–518.
[9] K. Imamura, N. Mataga, K. Mori, J. Neurophysiol. 1992, 68, 1986–2002.
[10] J. F. Kaluza, H. Breer, J. Exp. Biol. 2000, 203, 927–933.
[11] B. Malnic, J. Hirono, T. Sato, L. B. Buck, Cell 1999, 96, 713–723.
[12] S. E. Repicky, C. W. Luetje, J. Neurochem. 2009, 109, 193–202.
[13] D. Krautwurst, K. W. Yau, R. R. Reed, Cell 1998, 95, 917–926.
[14] M. D. Kurland, M. B. Newcomer, Z. Peterlin, K. Ryan, S. Firestein, V. S. Ba-
tista, Biochemistry 2010, 49, 6302–6304.
[15] Z. Peterlin, Y. Li, G. Sun, R. Shah, S. Firestein, K. Ryan, Chem. Biol. 2008,
15, 1317–1327.
[16] H. Zhao, L. Ivic, J. M. Otaki, M. Hashimoto, K. Mikoshiba, S. Firestein, Sci-
ence 1998, 279, 237–242.
[17] M. T. Liu, J. Ho, J. K. Liu, R. Purakait, U. N. Morzan, L. Ahmed, V. S. Batista,
H. Matsunami, K. Ryan, Org. Biomol. Chem. 2018, 16, 2541–2548.
[18] Y. Oka, M. Omura, H. Kataoka, K. Touhara, EMBO J. 2004, 23, 120–126.
[19] P. A. Godfrey, B. Malnic, L. B. Buck, Proc. Natl. Acad. Sci. USA 2004, 101,
2156–2161.
[20] X. Zhang, S. Firestein, Nat. Neurosci. 2002, 5, 124–133.
[21] A. Chess, I. Simon, H. Cedar, R. Axel, Cell 1994, 78, 823–834.
[22] N. K. Hanchate, K. Kondoh, Z. Lu, D. Kuang, X. Ye, X. Qiu, L. Pachter, C.
Trapnell, L. B. Buck, Science 2015, 350, 1251–1255.
[31] Y. Li, Z. Peterlin, J. Ho, T. Yarnitzky, M. T. Liu, M. Fichman, M. Y. Niv, H.
Matsunami, S. Firestein, K. Ryan, ACS Chem. Biol. 2014, 9, 2563–2571.
[32] D. J. Rodriguez-Gil, D. L. Bartel, A. W. Jaspers, A. S. Mobley, F. Imamura,
C. A. Greer, Proc. Natl. Acad. Sci. USA 2015, 112, 5821–5826.
[33] E. Slankster, S. R. Odell, D. Mathew, J. Bioenergy Biomembranes 2019, 51,
65–75.
[34] M. Q. Nguyen, Z. Zhou, C. A. Marks, N. J. Ryba, L. Belluscio, Cell 2007,
131, 1009–1017.
[35] J. E. Amoore, Molecular Basis of Odor, Charles, C. Thomas, Springfield,
1970.
[36] E. Block, S. Jang, H. Matsunami, S. Sekharan, B. Dethier, M. Z. Ertem, S.
Gundala, Y. Pan, S. Li, Z. Li, S. N. Lodge, M. Ozbil, H. Jiang, S. F. Penalba,
V. S. Batista, H. Zhuang, Proc. Natl. Acad. Sci. USA 2015, 112, E2766–
2774.
[37] R. W. Moncrieff, Perfum. Essent. Oil Rec. 1949, 40, 279–285.
[38] J. Liu, M. Y. Liu, J. B. Nguyen, A. Bhagat, V. Mooney, E. C. Yan, J. Am.
Chem. Soc. 2009, 131, 8750–8751.
[39] K. W. Yau, G. Matthews, D. A. Baylor, Nature 1979, 279, 806–807.
[40] D. J. Cram, Angew. Chem. Int. Ed. Engl. 1986, 25, 1039–1057; Angew.
Chem. 1986, 98, 1041–1060.
[23] L. Tan, Q. Li, X. S. Xie, Mol. Systems Biol. 2015, 11, 844.
[24] M. K. Bennett, H. M. Kulaga, R. R. Reed, Mol. Cell. Neurosci. 2010, 43,
353–362.
[41] A. Sunderkçtter, S. Lorenzen, R. Tacke, P. Kraft, Chem. Eur. J. 2010, 16,
7404–7421.
[25] H. Zhao, R. R. Reed, Chem. Senses 2001, 26, 1110 – 1111.
[26] Z. Zhou, L. Belluscio, Cell Rep. 2012, 2, 1143–1150.
[27] C. H. Wetzel, D. Brunert, H. Hatt, Chem. Senses 2005, 30, i321-322
Suppl 1.
[28] M. Na, M. T. Liu, M. Q. Nguyen, K. Ryan, ACS Chem. Neurosci. 2019, 10,
552–562.
Manuscript received: March 20, 2020
Revised manuscript received: May 22, 2020
Version of record online: && &&, 0000
&
&
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
8
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
doi.org/10.1002/chem.202001390
Chemistry—A European Journal
FULL PAPER
&
Sensory Mechanisms
M. T. Liu, M. Na, Y. Li, M. R. Biscoe,
K. Ryan*
&& – &&
Conformational Sensing by a
Mammalian Olfactory Receptor
Rotate to activate: Precisely what mo-
lecular features do the olfactory GPCRs
sense in odorant ligands? The rodent
Olfr2 receptor binds aliphatic aldehydes
indiscriminately, but conformational re-
striction experiments on the flexible ag-
onist, octanal, correlates the C5ÀC6
rotamer position with different receptor
outputs—activation, partial activation,
or antagonism. Thus, an olfactory GPCR
can carry out conformational analysis
during odor recognition.
Chem. Eur. J. 2020, 26, 1 – 9
www.chemeurj.org
9
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ