Genetic Variation in a Human Odorant Receptor Alters Odour Perception

Abstract

Human olfactory perception differs enormously between individuals, with large reported perceptual variations in the intensity and pleasantness of a given odour. For instance, androstenone (5alpha-androst-16-en-3-one), an odorous steroid derived from testosterone, is variously perceived by different individuals as offensive ("sweaty, urinous"), pleasant ("sweet, floral") or odourless. Similar variation in odour perception has been observed for several other odours. The mechanistic basis of variation in odour perception between individuals is unknown. We investigated whether genetic variation in human odorant receptor genes accounts in part for variation in odour perception between individuals. Here we show that a human odorant receptor, OR7D4, is selectively activated in vitro by androstenone and the related odorous steroid androstadienone (androsta-4,16-dien-3-one) and does not respond to a panel of 64 other odours and two solvents. A common variant of this receptor (OR7D4 WM) contains two non-synonymous single nucleotide polymorphisms (SNPs), resulting in two amino acid substitutions (R88W, T133M; hence 'RT') that severely impair function in vitro. Human subjects with RT/WM or WM/WM genotypes as a group were less sensitive to androstenone and androstadienone and found both odours less unpleasant than the RT/RT group. Genotypic variation in OR7D4 accounts for a significant proportion of the valence (pleasantness or unpleasantness) and intensity variance in perception of these steroidal odours. Our results demonstrate the first link between the function of a human odorant receptor in vitro and odour perception.

Full text

LETTERS
Genetic variation in a human odorant receptor alters
odour perception
Andreas Keller
1
*, Hanyi Zhuang
2
*, Qiuyi Chi
2
, Leslie B. Vosshall
1
& Hiroaki Matsunami
2,3
Human olfactory perception differs enormously between indivi-
duals, with large reported perceptual variations in the intensity
and pleasantness of a given odour. For instance, androstenone
(5a-androst-16-en-3-one), an odorous steroid derived from tes-
tosterone, is variously perceived by different individuals as offens-
ive (‘‘sweaty, urinous’’), pleasant (‘‘sweet, floral’’) or odourless
1–3
.
Similar variation in odour perception has been observed for several
other odours
4–6
. The mechanistic basis of variation in odour per-
ception between individuals is unknown. We investigated whether
genetic variation in human odorant receptor genes accounts in part
for variation in odour perception between individuals
7,8
. Here we
show that a human odorant receptor, OR7D4, is selectively acti-
vated in vitro by androstenone and the related odorous steroid
androstadienone (androsta-4,16-dien-3-one) and does not respond
to a panel of 64 other odours and two solvents. A common variant of
this receptor (OR7D4 WM) contains two non-synonymous single
nucleotide polymorphisms (SNPs), resulting in two amino acid
substitutions (R88W, T133M; hence ‘RT’) that severely impair
function in vitro. Human subjects with RT/WM or WM/WM geno-
types as a group were less sensitive to androstenone and androsta-
dienone and found both odours less unpleasant than the RT/RT
group. Genotypic variation in OR7D4 accounts for a significant
proportion of the valence (pleasantness or unpleasantness) and
intensity variance in perception of these steroidal odours. Our
results demonstrate the first link between the function of a human
odorant receptor in vitro and odour perception.
We investigated the hypothesis that polymorphisms in odorant
receptors contribute to variability in human odour perception by
combining a cell-based assay technique to identify active ligands
for odorant receptors
9
with an olfactory psychophysical study of a
diverse population of human subjects
10
. A total of 66 odorants were
used to measure both odorant receptor responses in vitro and
psychophysical responses in human subjects, with a focus on odorous
steroids because the perception of these odours is exceptionally vari-
able
3,11
. We cloned a panel of 335 putative human odorant receptors,
representing more than 85% of the odorant receptors with full open
reading frames, and expressed them in Hana3A cells, an HEK293T-
derived cell line stably expressing accessory factors for odorant re-
ceptor expression
9,12
. We screened for androstenone-mediated
stimulation with a luciferase reporter
9
. Among the receptors tested,
that encoded by OR7D4 showed the strongest responses to andros-
tenone (Fig. 1a). Recent expression analysis shows that OR7D4 is
selectively expressed in human nasal epithelium
13
. Several other
receptors showed smaller responses to androstenone that may be
relevant in vivo (Supplementary Table 1), but these were not inves-
tigated further. The failure of a specific odorant receptor to respond
in this assay must be interpreted with caution because it may reflect a
failure of the odorant receptor to be functional in the assay rather
than a lack of sensitivity to the tested odour.
A search for polymorphisms in OR7D4 in SNP databases and
sequencing the coding region of OR7D4 in 391 subjects identified
13 non-synonymous SNPs in this receptor, with four occurring at
*These authors contributed equally to this work.
1
Laboratory of Neurogeneti cs and Behaviour, The Rockefeller University, 1230 York Avenue, New York, New York 10065, USA.
2
Department of Molecular Genetics and Microbiology,
and
3
Department of Neurobiology, Duke University Medical Centre, Research Drive, Durham, North Carolina 27710, USA.
Androstenone
O
H
O
Androstadienone
(–)-Menthol
(+)-Menthol
1-Butanol
2-Butanone
2-Decenal
2-Ethylfenchol
2-Methoxy-4-methylphenol
4-Methylvaleric acid
Ambrette
Androstenone
Anise
Banana
Bourgeonal
Butyl acetate
Butyric acid
Hexanoic acid
Cedarwood
Cineole
Cinnamon
Cis-3-hexen-1-ol
Citral
Citronella
Decyl aldehyde
Diacetyl
Diallyl sulphide
Diphenyl ether
Ethyl vanillin
Ethylene brassylate
Eugenol
Eugenol acetate
Eugenol methyl ether
Fenchone
Fir
Galaxolide
Androstadienone
Geranyl acetate
Guaiacol
Heptaldehyde
Heptyl acetate
Hexyl butyrate
Isobornyl acetate
Isobutyraldehyde
Isobutyric acid
Isoeugenol
Isovaleric acid
Jasmine
Lime
Linalool
Methanethiol
Methyl salicylate
Nonyl aldehyde
Nutmeg
Octyl acetate
Octyl aldehyde
Orange
Pentadecalactone
Phenyl acetaldehyde
Pyrazine
(R)-Carvone
(R)-Limonene
Sandalwood
Spearmint
Terpineol
Terpinyl acetate
Undecanal
Vanillin
Paraffin oil
Propylene glycol
None
Odorant
c
1.0
0.5
0
0.1
0
Normalized
response
OR families
1234 5678910111213515256
a
Normalized
response
1.0
0.5
0
OR7D4
30 µM androstenone
Residue
change
refseq
P79L
S84N
R88W
T133M
Allele
frequency
0.786
0.040
0.013
0.157
0.157
b
Figure 1
|
OR7D4 is selectively activated by androstenone and
androstadienone. a, Hana3A cell luciferase assays of 335 unique human
odorant receptors; the concentration of androstenone used was 30 mM.
Numbers and coloured bars indicate different OR families.
b, Allele
frequencies of common variants. refseq, reference sequence.
c, OR7D4 RT
(black columns) and WM (red columns) tested against 66 odours and 2
solvents (30 mM, or 1/30,000). Normalized responses are shown as means
and s.e.m. (n 5 4).
Vol 449
|
27 September 2007
|
doi:10.1038/nature06162
468
Nature
©2007
Publishing
Group

frequencies greater than 1% (Supplementary Table 2). Two non-
synonymous polymorphisms in complete linkage disequilibrium in
this population occurred at the highest frequency and led to two
amino-acid changes (R88W and T133M; Fig. 1b). We refer to the
most common allele of this receptor, or the reference sequence, as RT,
and to the other as WM.
We investigated the ligand specificity of RT and WM receptor
variants in vitro with a panel of 66 odours and 2 solvents. OR7D4
RT responds selectively to androstenone and the closely related
odorous steroid androstadienone but shows no response to any other
stimuli tested (Fig. 1c, top). OR7D4 WM shows no response to any
compound at the concentrations tested here (Fig. 1c, bottom). Dose–
response curves with RT and WM show that the paired SNPs in the
WM variant, which affect amino acids in extracellular loop 2 and
intracellular loop 2 (Fig. 2a), severely impair function (Fig. 2b). We
generated odorant receptors with each of the SNPs and found that
OR7D4 R88W and OR7D4 T133M retained an intermediate level of
function, suggesting that both residues are important for OR7D4
function (Fig. 2b).
We examined two other SNP variants found at frequencies
greater than 1%, which led to amino acid changes P79L and S84N,
respectively (Fig. 1b and 2a). P79L and S84N possess residues R88
and T133 and are referred to by the single variant residue. Analysis of
P79L function in vitro showed severely impaired function at all con-
centrations of either steroidal odour tested (Fig. 2c). In contrast,
OR7D4 S84N showed remarkable sensitivity to both odours in vitro,
exceeding the activity of the common RT variant at every concentra-
tion tested, with a concentration giving half-maximal response
(EC
50
) to androstadienone less than one-fifth of that of the RT vari-
ant (Fig. 2c).The functional differences between these two variants
are not due to the amino-terminal epitope tag (Supplementary Fig.
1). The other non-synonymous substitutions in OR7D4 result in
varying receptor functions (Fig. 2d).
OR7D4 is situated in a cluster of seven intact odorant receptor
genes, but we found that none of the polymorphisms of the six intact
odorant receptors in the OR7D4 gene cluster showed significant
linkage with the OR7D4 SNPs (Supplementary Fig. 2a and Supple-
mentary Table 3). We tested the responses of all the major variants of
odorant receptors in the OR7D4 cluster and found that none showed
responses to androstenone and androstadienone exceeding that of
the impaired WM variant (Supplementary Fig. 2b, c).
The chimpanzee OR7D4 orthologue differs from the human RT
reference sequence at five amino acid residues: at the S84N substi-
tution also found in humans and at four additional non-synonymous
substitutions not found in humans (V26I, G171V, G227R and
K232E). A dose–response analysis of the chimpanzee OR7D4 ortho-
logue in vitro showed robust responses to both steroidal odours,
exceeding the activity of the human S84N variant (Fig. 2c; compare
purple and green curves).
What accounts for the functional differences between OR7D4 var-
iants? We found no obvious difference in subcellular distribution or
expression level in permeabilized Hana3A cells expressing RT, WM,
P79L or S84N (Supplementary Fig. 3). Western blot analysis con-
firms that all are expressed at comparable levels (Fig. 2e), and RT,
WM and P79L have similar low levels of surface staining as measured
by flow cytometry of live cells stained to reveal the N-terminal epi-
tope (Fig. 2f). S84N showed considerably more surface expression
(Fig. 2f), suggesting that the increased function of this variant may
stem from enhanced stability at the cell surface or from enhanced
cell-surface trafficking.
We next asked whether variation in OR7D4 is correlated with
variation in the perception of androstenone and androstadienone
measured in human subjects. Psychophysical data on 391 subjects
performing three different tasks were collected: subjects rated the
perceived intensity and valence of 66 different odours at two con-
centrations (Supplementary Fig. 4); detection thresholds were mea-
sured to androstenone and androstadienone in a subset of subjects,
and to three control odours in all subjects
14,15
(Supplementary Fig. 5);
subjects profiled four odours with 146 semantic labels
10,16
(see
Supplementary Methods).
Psychophysical data on the subjects were subsequently divided
according to genotype and assessed for the influence of OR7D4 geno-
type on perceptual phenotype (Supplementary Table 2). Of the 66
odours and two solvents rated by RT/RT and RT/WM subjects, only
androstenone and androstadienone showed a significant effect of
genotype (Fig. 3a and Supplementary Fig. 6). The steroids were rated
as less intense by the RT/WM group (Fig. 3a); the proportion of
RT/WM subjects rating the high concentration of androstenone as
‘‘extremely weak’’ was fourfold that of RT/RT subjects (Supplemen-
tary Fig. 6). This phenotype was specific for these two compounds, as
the perception of all the other odours was not affected by OR7D4
genotype (Fig. 3a, b). The significant effect of OR7D4 genotype on
steroidal odour intensity perception was replicated in both males and
females (Supplementary Fig. 7) and in the largest racial category,
Caucasians (Supplementary Fig. 8). Although the WM allele strongly
affected androstenone intensity perception even in heterozygous
subjects, the group of WM/WM subjects showed an even stronger
effect on intensity ratings, rating both steroidal odours as less intense
EC
50
(µM)
RT
R88W
T133M
2.9
8.9
10
WM
>30
RT
S84N
Chimp
WM
EC
50
(µM)
0.10
0.52
3.3
>30
P79L
>30
1.0
EC
50
(µM)
Normalized response
[Androstenone] (M)
010
–7
10
–6
10
–5
[Androstadienone] (M)
[Androstenone] (M) [Androstadienone] (M)
010
–7
10
–6
10
–5
0
0.5
1.0
Normalized response
0
0.5
1.0
0
0.5
1.0
0
0.5
RT
R 88W
T133M
12
17
>30
WM
>30
0
10
–7
10
–6
10
–5
10
–8
0
10
–7
10
–6
10
–5
10
–9
10
–8
RT
S84N
Chimp
WM
EC
50
(µM)
0.91
4.2
12
>30
P79L
>30
c
Normalized
response
1.0
0.5
0
RT
WM
D52G
S75C
P79L
S84N
R88W
H131Q
T133M
M136I
C139Y
L162P
A279D
L292M
C139R
0 µM
3 µM
Androstenone
d
b
a
ef
100
10
1
10
2
Counts
80
60
40
20
0
10
0
PE fluorescence
WM
P79L
S84N
RT
No receptor
RT
WM
S84N
P79L
80
60
50
40
120
kDa
Anti-GFP
Anti-Rho
NH
2
COOH
R88W
S84N
P79L
T133M
30 µM
Figure 2
|
Functional characterization of OR7D4 polymorphisms. a, OR7D4
snake plot with amino acid changes indicated.
b, c, Dose–response curves and
EC
50
values of OR7D4 RT, WM, R88W and T133M (b) and of OR7D4 RT,
WM, P79L and S84N and chimpanzee OR7D4 (
c) to androstenone (left) and
androstadienone (right).
d, Activity of 13 SNP variants compared with that of
RT and WM variants. In
b–d, normalized responses are shown as means
and s.e.m. (n 5 4–6).
e, Western blot analysis of whole-cell lysates from
HEK293T cells transfected with OR7D4 RT, WM, P79L or S84N and co-
transfected with green fluorescent protein (GFP).
f, Flow cytometry analysis
of cell-surface OR7D4 RT, WM, P79L and S84N expression as measured by
intensity of phycoerythrin (PE) signal among GFP-positive cells.
NATURE
|
Vol 449
|
27 September 2007 LETTERS
469
Nature
©2007
Publishing
Group

than the group of RT/RT subjects (Fig. 3b). The few subjects among
our population carrying the less frequent P79L and S84N variants
were examined, and RT/P79L subjects as a group showed a trend to
perceive both androstenone and androstadienone as less intense than
the group of RT/RT subjects (Fig. 3c). Conversely, RT/S84N subjects
as a group showed a trend to rate both odours as more intense
(Fig. 3c). However, these differences measured in subjects with the
rare alleles were not statistically significant.
Detection thresholds of 121 subjects were determined for both
steroidal odours (Fig. 3d, e). RT/WM subjects as a group had a higher
detection threshold—and were therefore less sensitive—to both
compounds than the group of RT/RT subjects (Fig. 3d) but had
normal thresholds to three control odours (data not shown). The
threshold for androstenone in the RT/WM group was 11-fold, and
that for androstadienone 16-fold, of that in the RT/ RT group. In
addition, 46% of RT/WM subjects but only 28% of RT/RT subjects
could not detect the highest concentration of androstenone we pro-
vided (P , 0.05; x
2
test). Detection thresholds were also obtained
from the few RT/P79L, WM/WM, WM/P79L and RT/S84N subjects
(Fig. 3e and Supplementary Fig. 9). The proportion of RT/P79L
subjects unable to detect the highest concentration of androstenone
and androstadienone provided was more than double that of RT/RT
subjects (P , 0.05; x
2
test) (Fig. 3e). It is unclear why the loss of one
functional allele has such a profound effect on the median detection
threshold. The same trend was found with the WM/WM and WM/
P79L groups (Supplementary Fig. 9), whereas the RT/S84N group
was more sensitive to both steroids, with lower detection thresholds
than the group of RT/RT controls. However, these differences were
not statistically significant (Fig. 3e).
We next examined whether variation in OR7D4 affects the percep-
tion of androstenone and androstadienone odour quality. The RT/
WM group rated both steroidal odours as less unpleasant than the
RT/RT group (Fig. 4a, b and Supplementary Fig. 10), such that the
proportion of RT/WM subjects rating the high concentration of
androstadienone as ‘‘extremely unpleasant’’ was less than half of that
of RT/RT subjects (Supplementary Fig. 10). None of the other 64
odours or the solvents showed a statistically significant difference
between the genotypes (Fig. 4a), and the effect was statistically sig-
nificant for both steroidal odours in both males and females (Sup-
plementary Fig. 7). The group of subjects carrying the impaired RT/
P79L variant showed a trend to rate both androstenone and andros-
tadienone as less unpleasant than RT/RT controls (Fig. 4c), whereas
the opposite was found in the RT/S84N group carrying a more sens-
itive variant of OR7D4 (Fig. 4c); however these differences were not
statistically significant.
We asked subjects to assess androstenone odour quality by profil-
ing this odour with a standard set of 146 semantic descriptors (see
Supplementary Methods)
10,16
. All descriptors used by more than 10%
of the subjects were analysed and descriptor usage in individuals with
different genotypes was compared. Of the 74 descriptors used for
androstenone, pentadecalactone, vanillin and the solvent propylene
glycol, only five differed significantly by genotype (see Supplemen-
tary Methods). OR7D4 RT/WM subjects were more likely to rate
vanillin as smelling ‘‘honey’’, ‘‘sweet’’ and ‘‘vanilla’’ (Fig. 4d, e),
Terpinyl acetate
Vanillin
Cinnamon
Galaxolide
Diacetyl
1-Butanol
Anise
Butyl acetate
Citronella
Hexyl butyrate
Isobornyl acetate
Orange
Undecanal
4-Methylvaleric acid
Geranyl acetate
2-Methoxy-4-methylphenol
Banana
Butyric acid
Cineole
Citral
Diphenyl ether
Eugenol methyl ether
Guaiacol
Heptyl acetate
Jasmine
Pentadecalactone
(R)-Carvone
2-Ethylfenchol
Eugenol
(R)-Limonene
(–)-Menthol
2-Decenal
Ethyl vanillin
Eugenol acetate
Isobutyraldehyde
Isobutyric acid
Isoeugenol
Linalool
Methyl salicylate
Octyl acetate
Paraffin oil
Propylene glycol
Spearmint
Terpineol
Cis-3-hexen-1-ol
Ethylene brassylate
Fenchone
Fir
(+)-Menthol
Bourgeonal
Heptaldehyde
Isovaleric acid
Phenyl acetaldehyde
Sandalwood
2-Butanone
Decyl aldehyde
Hexanoic acid
Diallyl sulphide
Nonyl aldehyde
Nutmeg
Octyl aldehyde
Pyrazine
Methanethiol
Lime
Ambrette
Cedarwood
Androstadienone
Androstenone
RT/P79L (n = 29)
RT/S84N (n = 7)
b
a
*
*
*
Odorant
More intense in RT/WM
Less intense in RT/WM
Difference in
odour intensity ran
k
*
*
*
c
d
RT/RT (n = 242)
RT/WM (n = 96)
WM/WM (n = 10)
***
*
***
*
Androstenone
Androstadienone
Propylene glycol
0
20
40
60
–20
RT/RT (n = 242)
*
Relative change in odour intensity rank
Androstenone
Androstadienone
Propylene glycol
Pentadecalactone
Vanillin
Pentadecalactone
Vanillin
0
20
40
60
–20
Less
intense
More
intense
Less
intense
More
intense
20
20
10
10
0
0
0
2
4
6
8
0
2
4
6
8
5
10
15
20
25
RT/RT
median
RT/WM
median
*
0
5
10
15
20
25
RT/WM
median
RT/RT
median
*
e
Androstenone detection
threshold
Androstadienone detection
threshold
RT/
S
84
N
me
dian
RT/
P
79L
me
dia
n
No. of subjects
R
T/P
79
L
m
ed
ia
n
RT/S
8
4
N
m
ed
ia
n
H
ig
her conce
ntr
ation
Higher
conce
ntration
H
igher concent
ration
Higher concen
tratio
n
Figure 3
|
OR7D4 variation affects androstenone and androstadienone
intensity perception. a
, Differences in median odour intensity ranking of 66
odours and 2 solvents between OR7D4 RT/WM and RT/RT groups. Data for
two different odour concentrations were pooled.
b, Change in odour
intensity ranking relative to solvent of four odours for the RT/RT, RT/WM
and WM/WM groups (
b) and for the RT/RT, RT/P79L and RT/S84N groups
(
c). The whisker plots show the median rank (normalized to the median rank
of the RT/RT group), the first and third quartile and the upper and lower
limits. Significance was assessed in
a–c with a Mann–Whitney U-test with a
Bonferroni correction. Before correction: asterisk, P , 0.00073; two
asterisks, P , 0.00014; three asterisks, P , 1.47 3 10
25
. After correction:
asterisk, P , 0.05; two asterisks, P , 0.01; three asterisks, P , 0.001.
d, e, Detection thresholds measured in RT/RT (n 5 47) and RT/WM subjects
(n 5 49) (
d) and in RT/P79L (n 5 12) and RT/S84N (n 5 3) subjects
(
e) plotted as the number of subjects detecting the odour at a given binary
dilution (x-axis concentrations are binned from left to right: the first bar
represents binary dilution 6, the subsequent 15 bars represent binary
dilutions 7–21, and the last bar represents dilutions 22–27). Significance was
assessed with a Mann–Whitney U-test (asterisk, P , 0.05).
LETTERS NATURE
|
Vol 449
|
27 September 2007
470
Nature
©2007
Publishing
Group

42% less likely to consider androstenone ‘‘sickening’’ and 129% more
likely to rate it as smelling like ‘‘vanilla’’ than RT/RT subjects (Fig. 4f,
g). There was no effect of OR7D4 genotype on odour profiling of
pentadecalactone or propylene glycol (data not shown).
Last, we performed a non-parametric regression analysis
17
to
estimate the fraction of the phenotypic variance in steroid odour
perception attributable to OR7D4. We found that OR7D4 genotype
in our population explained 19% and 39% of the variance in the
valence and intensity ratings of the steroid odours, respectively
(P , 0.0001; see Supplementary Methods).
We identify OR7D4 as a significant heritable factor influencing
androstenone and androstadienone perception, thus providing the
first reported link between genetic polymorphisms in an odorant
receptor gene and altered perception of the ligands that activate this
receptor. As predicted by the theory of combinatorial coding
18
,we
find that polymorphisms in the OR7D4 protein-coding sequence
alone do not fully account for specific anosmia to androstenone
and androstadienone. We think it likely that additional human odor-
ant receptors sensitive to androstenone and androstadienone remain
to be discovered.
Previous work indicated that sensitivity to androstenone
19,20
, and
to several other odours
21
, is modulated by non-genetic effects such as
central processing
22
or peripheral sensitization
23
that might obscure
any underlying genetic influences on androstenone perception.
About half of the subjects initially unable to detect androstenone
were able to smell this compound after being exposed to androste-
none daily for six weeks
19
. Although it has been assumed that olfact-
ory induction is a phenomenon that occurs only in individuals with
specific anosmia to androstenone, more recent studies have found
that induction of enhanced olfactory sensitivity seems to be a general
phenomenon affecting several odorants
21
. The role of OR7D4 in this
sensitization can now be tested in subjects chronically exposed to
androstenone.
In this study we investigated only the olfactory percept reported
when odorous steroids were sniffed, but olfactory exposure to
androstenone and androstadienone has also been shown to induce
several physiological responses in both men and women
24,25
. The
identification of an odorant receptor gene that is strongly correlated
with the perception of these odours will permit future analysis of
olfactory-induced autonomic responses in humans.
METHODS SUMMARY
Heterologous expression of human odorant receptors. A total of 423 human
odorant receptors, including 335 predicted functional receptors, were cloned.
The chimpanzee OR7D4 orthologue was cloned from chimpanzee genomic
DNA (Coriell Cell Repositories). Odorant receptors containing the first 20
amino acids of human rhodopsin
26
in pCI (Promega) were expressed in the
Hana3A cell line together with a short form of mRTP1 called RTP1S (M37 to
*
*
*
*
*
*
RT/P79L (n = 29)
RT/S84N (n = 7)
c
b
a
Odorant
Less pleasant in RT/WM
More pleasant in RT/WM
0
10
20
10
20
Androstadienone
Androstenone
(–)-Menthol
Banana
Fir
Jasmine
Methyl salicylate
Eugenol methyl ether
Cedarwood
Eugenol
Isoeugenol
Linalool
(+)-Menthol
Anise
Cis-3-hexen-1-ol
Pyrazine
Fenchone
Isobutyraldehyde
Isovaleric acid
Pentadecalactone
Terpineol
Vanillin
Citral
1-Butanol
2-Methoxy-4-methylphenol
Butyric acid
Cinnamon
Diacetyl
Diphenyl ether
Ethyl vanillin
Galaxolide
Isobornyl acetate
Isobutyric acid
Octyl aldehyde
Orange
Propylene glycol
Spearmint
2-Ethylfenchol
2-Butanone
4-Methylvaleric acid
Ethylene brassylate
Lime
(R)-Carvone
Ambrette
Cineole
Decyl aldehyde
Diallyl sulphide
Eugenol acetate
Guaiacol
Hexyl butyrate
Paraffin oil
(R)-Limonene
Sandalwood
Geranyl acetate
2-Decenal
Heptyl acetate
Methanethiol
Nutmeg
Phenyl acetaldehyde
Terpinyl acetate
Butyl acetate
Hexanoic acid
Citronella
Nonyl aldehyde
Octyl acetate
Bourgeonal
Heptaldehyde
Undecanal
Difference in
odour valence rank
e
f
*
*
Used less by RT/WM
Descriptors of androstenone odour
0
10
10
*
Used more by RT/WM
Sickening
Heavy
Putrid foul
Bitter
Rancid
Urine
Ammonia
Sweaty
Cleaning fluid
Dirty linen
Sharp pungent
Chemical
Musky
Musty mouldy
Aromatic
Stale
Light
Fragrant
Rotten fruit
Sweet
Vanilla
d
Used more by RT/WM
Used less by RT/WM
Descriptors of vanillin odour
0
10
10
Difference in descriptor usage (%)
*
*
*
Light
Cologne
Fruity other
Buttery fresh
Cool
Aromatic
Fragrant
Cinnamon
Floral
Perfumery
Incense
Caramel
Coconut
Heavy
Musky
Chocolate
Maple syrup
Almond
Molasses
Honey
Sweet
Warm
Vanilla
RT/RT (n = 242)
RT/WM (n = 96)
WM/WM (n = 10)
0
20
40
60
***
Androstenone
***
Androstadienone
Propylene glycol
Pentadecalactone
Vanillin
–20
More
unpleasant
Less
unpleasant
More
unpleasant
Less
unpleasant
RT/RT (n = 242)
0
20
40
60
Relative change in odour valence rank
Androstenone
Androstadienone
Propylene glycol
Pentadecalactone
Vanillin
–20
‘Vanilla’‘Sweet’
Odour profiled:
vanillin
0
50
100
‘Honey’
*
*
*
RT/RT (n = 242)
RT/WM (n = 96)
WM/WM (n = 10)
g
Descriptor usage (%)
‘Sickening’
‘Vanilla’
Odour profiled:
androstenone
0
50
100
**
*
RT/RT (n = 242)
RT/WM (n = 96)
WM/WM (n = 10)
Figure 4
|
OR7D4 variation affects androstenone and androstadienone
quality perception. a, Differences in median odour valence ranking for the
same odours and genotypes as in Fig. 3a.
b, c, Change in odour valence
ranking for the same odours and genotypes as in Fig. 3b, c. Significance in
a–c was assessed with a Mann–Whitney U-test with Bonferroni correction.
Before correction: asterisk, P , 0.00073; two asterisks, P , 0.00014; three
asterisks, P , 1.47 3 10
25
. After correction: asterisk, P , 0.05; two asterisks,
P , 0.01; three asterisks, P , 0.001. The whisker plots show the median rank
(normalized to the median rank of the RT/RT group), the first and third
quartile and the upper and lower limits.
d, e, Odour profiling of vanillin by
RT/RT (n 5 242), RT/WM (n 5 96) and WM/WM (n 5 10) subjects. Plotted
are the differences in descriptor usage by genotype of the 23 descriptors used
for vanillin in more than 10% of sessions (
d) and the percentage of sessions
(with 95% confidence intervals) in which the three descriptors that showed
significant differences were used (
e). f, g, Odour profiling of androstenone
for the same genotypes as in
d and e; 21 descriptors were used in more than
10% of all sessions, and two descriptors showed significant differences.
Significance was assessed in
d–g with a x
2
test with Bonferroni correction.
Before correction: asterisk, P , 0.0022 (vanillin) and P , 0.0024
(androstenone); two asterisks, P , 0.0004 (vanillin) and P , 0.0005
(androstenone). After correction: asterisk, P , 0.05; two asterisks, P , 0.01.
NATURE
|
Vol 449
|
27 September 2007 LETTERS
471
Nature
©2007
Publishing
Group

the carboxy-terminal end), which enhances functional expression of the odorant
receptors
12
. For immunocytochemistry, cells were fixed, permeabilized and
incubated with monoclonal anti-rhodopsin antibody (4D2; ref. 27), followed
by Cy3-conjugated donkey anti-mouse IgG (Jackson Immunologicals). For
fluorescence-activated cell sorting analysis, this antibody was conjugated with
phycoerythrin.
Human odorant receptor genotyping and sequencing. Venous blood was col-
lected from all subjects, and genomic DNA was prepared with the Qiagen
PAXgene blood DNA kit. Polymorphisms in OR7D4 were assayed by sequencing
and allele-specific polymerase chain reaction. Polymorphisms in the other odor-
ant receptors in the same odorant receptor gene cluster as OR7D4 were assayed
by sequencing only.
Human olfactory psychophysics. All procedures involving human subjects were
approved by the Rockefeller University Institutional Review Board. All subjects
completed two replicates of the test separated by at least 4 days. Odours were
presented in bar-coded amber vials to ensure that subjects were blind to the
identity of all odours
28
. The intensity and valence of 66 odours at two concen-
trations (‘high’ and ‘low’) and two solvents was rated on a seven-point scale.
Thresholds were calculated by using the single-staircase method with seven
reversals
14,15
. Threshold tests included both steroids as binary dilutions from
1:64 (binary dilution 6) to 1:134,217,728 (binary dilution 27). Odour profiling
used a previously established method
16
.
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 20 June; accepted 8 August 2007.
Published online 16 September 2007.
1. Wysocki, C. J. & Beauchamp, G. K. Ability to smell androstenone is genetically
determined. Proc. Natl Acad. Sci. USA 81, 4899
–
4902 (1984).
2. Gower, D. B., Nixon, A. & Mallet, A. I. in Perfumery (eds Van Toller, S. & Dodd,
G. H.) 47
–
75 (Chapman & Hall, London, 1998).
3. Bremner, E. A., Mainland, J. D., Khan, R. M. & Sobel, N. The prevalence of
androstenone anosmia. Chem. Senses 28, 423
–
432 (2003).
4. Guillot, M. Physiologie des sensations—anosmies partielles et odeurs
fondamentales. C.R. Hebd. Acad. Sci. 226, 1307
–
1309 (1948).
5. Gilbert, A. N. & Kemp, S. E. Odor perception phenotypes: multiple, specific
hyperosmias to musks. Chem. Senses 21, 411
–
416 (1996).
6. Whissell-Buechy, D. & Amoore, J. E. Odour-blindness to musk: simple recessive
inheritance. Nature 242, 271
–
273 (1973).
7. Menashe, I., Man, O., Lancet, D. & Gilad, Y. Different noses for different people.
Nature Genet. 34, 143
–
144 (2003).
8. Gilad, Y. & Lancet, D. Population differences in the human functional olfactory
repertoire. Mol. Biol. Evol. 20, 307
–
314 (2003).
9. Saito, H., Kubota, M., Roberts, R. W., Chi, Q. & Matsunami, H. RTP family
members induce functional expression of mammalian odorant receptors. Cell 119,
679
–
691 (2004).
10. Keller, A. & Vosshall, L. B. Human olfactory psychophysics. Curr. Biol. 14,
R875
–
R878 (2004).
11. Stevens, D. A. & O’Connell, R. J. Enhanced sensitivity to androstenone following
regular exposure to pemenone. Chem. Senses 20, 413
–
419 (1995).
12. Zhuang, H. & Matsunami, H. Synergism of accessory factors in functional
expression of mammalian odorant receptors. J. Biol. Chem. 282, 15284
–
15293
(2007).
13. Zhang, X. et al. Characterizing the expression of the human olfactory receptor
gene family using a novel DNA microarray. Genome Biol. 8, R86 (2007).
14. Doty, R. L., McKeown, D. A., Lee, W. W. & Shaman, P. A study of the test
–
retest
reliability of ten olfactory tests. Chem. Senses 20, 645
–
656 (1995).
15. Doty, R. L. & Laing, D. G. in Handbook of Olfaction and Gustation (ed. Doty, R. L.)
203
–
228 (Marcel Dekker, New York, 2003).
16. Dravnieks, A. Odor quality: semantically generated multidimensional profiles are
stable. Science 218, 799
–
801 (1982).
17. Cleveland, W. S. Robust locally weighted regression and smoothing scatterplots.
J. Am. Stat. Assoc. 74, 829
–
836 (1979).
18. Malnic, B., Hirono, J., Sato, T. & Buck, L. B. Combinatorial receptor codes for odors.
Cell 96, 713
–
723 (1999).
19. Wysocki, C. J., Dorries, K. M. & Beauchamp, G. K. Ability to perceive androstenone
can be acquired by ostensibly anosmic people. Proc. Natl Acad. Sci. USA 86,
7976
–
7978 (1989).
20. Dorries, K. M., Schmidt, H. J., Beauchamp, G. K. & Wysocki, C. J. Changes in
sensitivity to the odor of androstenone during adolescence. Dev. Psychobiol. 22,
423
–
435 (1989).
21. Dalton, P., Doolittle, N. & Breslin, P. A. Gender-specific induction of enhanced
sensitivity to odors. Nature Neurosci. 5, 199
–
200 (2002).
22. Mainland, J. D. et al. One nostril knows what the other learns. Nature 419, 802
(2002).
23. Wang, L., Chen, L. & Jacob, T. Evidence for peripheral plasticity in human odour
response. J. Physiol. (Lond.) 554, 236
–
244 (2004).
24. Jacob, S., Kinnunen, L. H., Metz, J., Cooper, M. & McClintock, M. K. Sustained
human chemosignal unconsciously alters brain function. Neuroreport 12,
2391
–
2394 (2001).
25. Wyart, C. et al. Smelling a single component of male sweat alters levels of cortisol
in women. J. Neurosci. 27, 1261
–
1265 (2007).
26. Krautwurst, D., Yau, K. W. & Reed, R. R. Identification of ligands for olfactory
receptors by functional expression of a receptor library. Cell 95, 917
–
926 (1998).
27. Laird, D. W. & Molday, R. S. Evidence against the role of rhodopsin in rod
outer segment binding to RPE cells. Invest. Ophthalmol. Vis. Sci. 29, 419
–
428
(1988).
28. Keller, A. & Vosshall, L. B. A psychophysical test of the vibration theory of
olfaction. Nature Neurosci. 7, 337
–
338 (2004).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements L.B.V. and A.K. thank E. Gotschlich, B. Coller, A. N. Gilbert,
I. Gomez, P. Hempstead and C. Vancil; H.M. and H.Z. thank H. Amrein, M. Cook,
M. Kubota, D. Marchuk, R. Molday, D. Tracey and R. Valdivia. This research was
supported in part by an NIH Clinical and Translational Science Award to
Rockefeller University and by grants to L.B.V. from the Irma T. Hirschl Trust, to
H.M. from the NIH, to H.Z. from an NIH National Research Servic e Award, and to
A.K. from a Marco S. Stoffel Fellowship.
Author Contributions H.Z. and H.M. screened for androstenone receptors,
identified polymorphisms, performed functional expression of receptor variants,
and genotyped the human subjects with assistance from Q.C. A.K. and L.B.V.
devised the human olfactory psychophysics study, for which A.K. supervised data
collection and analysis.
Author Information The sequences of the human OR7D4 variants are deposited in
Genbank under accession numbers EU049291
–
EU049294. Reprints and
permissions information is available at www.nature.com/reprints. The authors
declare competing financial interests: details accompany the paper on
www.nature.com/nature. Correspondence and requests for materials should be
addressed to H.M. (hiroaki.matsunami@duke.edu) and L.B.V.
(leslie@mail.rockefeller.edu).
LETTERS NATURE
|
Vol 449
|
27 September 2007
472
Nature
©2007
Publishing
Group

METHODS
Heterologous expression of human odorant receptors. For untagged odorant
receptor experiments, OR7D4 RT and S84N variants without the Rho tag were
cloned into pCI. Luciferase assays were performed as described
9
. Only a single
variant of each receptor was used in the functional screen, and it is known that
many human odorant receptor genes are highly polymorphic
7
. Because of this
and because we do not know whether each odorant receptor is capable of func-
tional expression in the cell line used, we cannot exclude the possibility that
additional human odorant receptors respond to androstenone as strongly as
OR7D4. For immunocytochemistry, cells were fixed, permeabilized and incu-
bated with monoclonal anti-rhodopsin antibody, 4D2 (ref. 27), followed by Cy3-
conjugated donkey anti-mouse IgG (Jackson Immunologicals). Western blot
analysis was performed in accordance with the Mini-Protean 2 Cell (Bio-Rad)
protocol. Enhanced chemiluminescence (ECL; Amersham) was used for detect-
ing proteins on membranes. After the initial exposure, the membrane was incu-
bated with stripping buffer (25 mM glycine-HCl pH 2, 1% SDS, 25 mM glycine,
0.036 M HCl, 1% SDS) and incubated with rabbit anti-GFP (Invitrogen). See
Supplementary Methods for detailed information.
Odours. All odours were supplied by Sigma-Aldrich, with these exceptions:
androstadienone (gift from Human Pheromone Sciences, Inc.), banana (Bell
Flavors and Fragrances), bourgeonal (Biomol), galaxolide (gift from Inter-
national Flavors and Fragrances) and (R)-carvone (Research Chemical Ltd).
The same batch and lot of each odour was used for both cell-based analysis
and human olfactory psychophysics. Detailed information on odours, odour
concentrations and perceived odour quality is provided in Supplementary
Tables 4–6.
Genotyping and sequencing of human odorant receptors. For sequencing,
human genomic DNAs were amplified, purified and sequenced with a 3100 or
3730 Genetic Analyser (ABI Biosystems) or by GeneWiz. Detailed methods are
given in Supplementary Methods.
Human olfactory psychophysics. All human subjects gave informed consent to
participate in this study and were tested in a well-ventilated room of the
Rockefeller University Hospital Outpatient Unit. Normal human subjects were
pre-screened to exclude pregnant women and those with medical conditions
causing general impairment of the sense of smell. Of the 412 subjects who
completed the study, 21 were excluded because of general anosmia (see
Supplementary Methods). The remaining 391 subjects (210 female, 181 male;
median age 34 years, age range 19–75 years) were included in the evaluation.
Detailed methods are given in Supplementary Information. Our smell tests were
purposely conducted under conditions that would not be expected to induce
odour sensitivity in our subjects. A given subject sniffed androstenone in only
two sessions, rather than the dozens of sessions spread over six to eight weeks
required for sensitization in previous studies
19
. A comparative analysis of
androstenone responses in the first and second visits does not suggest that sub-
jects became more sensitive through these brief, non-chronic exposures (data
not shown).
doi:10.1038/nature06162
Nature
©2007
Publishing
Group