Chem. Senses 28: 87-104,
2003
© Oxford University Press 2003
RESEARCH PAPERS |
Sensitivity-dependent Hierarchical Receptor Codes for Odors
Life Electronics Laboratory, National Institute of Advanced Industrial Science and Technology, 3-11-46 Nakoji, Amagasaki, 661-0974, Japan
Correspondence to be sent to: Dr Takaaki Sato, Tissue Engineering Research Center, AIST-Amagasaki, National Institute of Advanced Industrial Science and Technology, 3-11-46 Nakoji, Amagasaki, Hyogo 661-0974, Japan. e-mail: taka-sato{at}aist.go.jp
Abstract
In order to comprehend the strategy of odor encoding by odorant receptors,
we isolated 2740 mouse receptor neurons from four olfactory epithelial zones
and classified them in terms of their sensitivities and tuning specificities
to a chiral pair of odorants, S(+)-carvone (caraway-like odor) and
R(-)-carvone (spearmint-like odor). Our approach revealed that the
majority of receptors at the lowest effective stimulus concentration
represented the principal odor qualities characteristic of each enantiomer by
means of the principal odor qualities of the odorants for which the receptors
were most sensitive. The chiral-non-discriminating receptors became 3.7 times
of R(-)-carvone-sensitive receptors in the subpopulations when the
stimulus concentration was increased 10-fold. More than 80% of the responsive
receptors (an estimated 70 ± 
types) exhibited overlapping
sensitivities between the enantiomers. The signals from the non-discriminating
receptors may be reduced to decode the characteristic odor identity for
R(-)-carvone in the brain over an adequate range of stimulus
strengths. The information processing of odors appears to involve the
selective weighting of the signals from the most sensitive receptors. An
analysis of the overall receptor codes to carvones indicated that the system
employs hierarchical receptor codes: principal odor qualities are encoded by
the most sensitive receptors and lower-ranked odor qualities by less sensitive
receptors.
Key words: calcium imaging, odorant, odor discrimination, odor quality, olfactory receptor neuron, sensory information processing
Introduction
The mammalian olfactory system can discriminate subtle differences in the
molecular structures of a great number of volatile organic compounds,
presented either as pure chemicals or blended as complex mixtures. This
remarkable capability is exemplified by the discrimination of a pair of
enantiomeric carvones which possess the same molecular structures except for
the chiral portion, but evoke distinct odor qualities. A spearmint-like fresh
herbal odor is induced by R(-)-carvone (rCa), whereas a
caraway-like fresh herbal scent appears in S(+)-carvone
(sCa) (Friedman and Miller,
1971
; Leitereg et
al., 1971; Boelens et
al., 1993). The basic question is what mechanism enables us
to discriminate different odors. The importance of the molecular features of
an odorant has been indicated by psychochemical theory
(Braun and Kröper, 1937;
Moncrieff, 1949;
Amoore, 1970;
Beets, 1982;
Ohloff, 1986), by the
responses of olfactory receptor neurons (ORNs) or functionally expressed
odorant receptors (ORs) (Sato et
al., 1994; Bozza and
Kauer, 1998; Malnic et
al., 1999; Touhara et
al., 1999; Kaluza and
Breer, 2000; Kajiya et
al., 2001; Bozza et
al., 2002) and by the responses of glomeruli or neurons in
the olfactory bulb (Mori et al.,
1992,
1999; Rubin and Katz,
1999,
2001;
Johnson and Leon, 2000;
Uchida et al., 2000;
Linster et al., 2001;
Wachowiak and Cohen, 2001;
Spors and Grinvald, 2002).
The discovery of a family of 
1000 mammalian genes encoding ORs
(Buck and Axel, 1991) led to
an understanding of the response properties of individual ORs
(Krautwurst et al.,
1998; Zhao et al.,
1998; Malnic et al.,
1999; Touhara et al.,
1999; Wetzel et al.,
1999; Araneda et al.,
2000; Kajiya et al.,
2001; Bozza et al.,
2002), the homogeneity of ORs expressed in single ORNs
(Chess et al., 1994;
Malnic et al., 1999;
Touhara et al., 1999;
Rawson et al., 2000;
Kajiya et al., 2001),
the alternative expression of ORs in one of the four olfactory epithelial
zones (Ressler et al.,
1993; Buck, 1996),
the distribution of 1296 ORs in chromosomes
(Zhang and Firestein, 2002),
candidate odorant-binding sites in ORs
(Singer and Shepherd, 1994;
Singer, 2000;
Floriano et al.,
2000), the OR-type-specific convergent projections from the
olfactory epithelium to the bulb (Ressler
et al., 1994; Mombearts et al., 1996;
Wang et al., 1998;
Tsuboi et al., 1999;
Serizawa et al.,
2000) and the OR-type-dependent clusters of pyramidal cells in the
olfactory cortex (Zou et al.,
2001). Odor encoding in the periphery has been interpreted in
terms of overlapping combinatorial receptor codes
(Malnic et al.,
1999). However, a simple combinatorial receptor code may not be
sufficient to represent actual odor qualities. The olfactory system may employ
additional strategies to encode and decode odor information.
We suppose that the brain genetically functions to find common information and unique information between different subjects, based on an informational factorization at receptors. We suppose that the identity of an odorant can be decomposed into a unique odor quality, transduced by specific binding to unique ORs, plus one or more additional odor qualities common to multiple odorants. The additional odor qualities would be transduced by various subsets of common ORs with highly overlapped sensitivities to multiple odorants. We postulate that the relative strengths of the different odor qualities elicited by an odorant are primarily dependent on the relative sensitivities of different classes of ORs to that odorant. This is in contrast to a simple combinatorial coding model which assigns equal weights to all ORs activated by an odorant. To test our hypothesis regarding odor encoding in ORs, we clarified the overlap and differences among the receptors activated by subsets of odorants, including the carvone enantiomers sCa and rCa, which possess unique and shared elements of odor quality.
Our analysis of the sensitivity profiles of ORs responsive to enantiomeric carvones reveals that the most sensitive receptors which occupy minor subpopulations can represent the principal odor qualities of the odorant. Here, we describe sensitivity-dependent hierarchical receptor codes for odors. The most sensitive receptors encode the principal odor qualities and the less sensitive receptors encode supplementary odor qualities of the odorant. The receptors preferentially sensitive to either sCa or rCa could represent the unique odor qualities of caraway or spearmint, respectively; those equally sensitive to both enantiomers could represent the fresh herbal quality common to both. In the hierarchical receptor codes, multiple supplementary odor qualities encoded by common receptors or less sensitive receptors are reduced in an overall odor identity representation by signal weighting governed by the most sensitive receptors.
Materials and methods
Determination of odorant tuning specificities of ORNs
In order to understand the concrete receptor codes for the odor qualities
of interesting odorants, we employed a previously reported experimental
strategy (Malnic et al.,
1999). The odorant responsiveness of the ORs was examined in
isolated ORNs using an intracellular calcium imaging assay (Ca-imaging assay)
on living cells (Sato et al.,
1994; Malnic et al.,
1999; Touhara et al.,
1999) and the amino-acid sequences of the ORs were identified in
the assayed ORNs using single-cell RT-PCR
(Malnic et al.,
1999). These methods and recent modifications are briefly
described as follows. Mice (BALB/c, CBA, or NZB) were anesthetized by Ketalar
50 (50 mg/ml ketamine, 5 µl/g, i.p.; Sankyo Pharmaceutical) and killed. The
animals were treated in accordance with the guidelines for the care and use of
laboratory animals of the National Institute of Health and of the AIST-Kansai
Animal Experiment Committee. The ORNs on a lateral surface of a small piece of
one of the four olfactory epithelial zones of mice were isolated on coverslips
using the tissue-printing method (Hirono
et al., 1992) after sequential enzymatic treatments with
trypsin, trypsin inhibitor and DNase I. The DNase treatment was added to
remove olfactory mucus and damaged ORNs from the epithelial pieces. The
isolated ORNs were loaded with fura-2 and then exposed to different odorants
(Aldrich) for 4 s. The cellular responses were monitored in intracellular
Ca2+ increases (Ca2+ responses) by observing the fura-2
fluorescence decreases at 510 nm, excited at 380 nm. Ca2+ responses
to 6-14 stimuli sequentially applied at intervals of 35 s were recorded
every 10-15 min.
To determine the relative sensitivity of ORNs/ORs to various test odorants,
we compared the relative response amplitudes of the odorant-induced
Ca2+ responses. The odorant-induced Ca2+ increases are
mediated by three pathways: Ca2+ influxes via cyclic-AMP-dependent
channels (Kurahashi, 1990;
Sato et al., 1992;
Leinders-Zufall et al.,
1998) and voltage-dependent Ca2+ channels
(Hirono et al.,
1991), and Ca2+ release
(Sato et al., 1991;
Zufall et al., 2000).
All pathways are theoretically activated more intensively as the receptor
potentials of the ORNs become greater. In the typical odorant-induced
Ca2+ responses, whose maximum intracellular Ca2+
concentrations were tens to hundreds of nanomolar
(Malnic et al.,
1999), reproducible Ca2+ influxes were predicted even
in insufficient recovery of the intracellular Ca2+ concentration,
because the extracellular Ca2+ concentration was millimolar which
could be sufficiently high to produce the differential concentration for
driving the Ca2+ influxes. In fact, we previously showed that four
successively repeated identical stimulations at intervals of 35 s evoked
responses with amplitudes of 100, 98, 80, and 77% of the initial ones in
insufficient recovery levels (Sato et
al., 1994). Furthermore, fluorescence intensity excited at
380 nm monotonically decreases with increasing Ca2+ concentration.
Although the responses were gradually reduced in the fluorescence changes
during the recordings, we could observe Ca2+ responses to the most
sensitive odorants with response amplitudes of >60% of the initial
responses in most ORNs. Considering these properties, we were able to compare
the relative sensitivity of ORs by comparing the net fluorescence decreases
normalized by the baseline fluorescence after the subtraction of the
background signals in the somata.
We employed the following criteria to judge whether the fluorescence changes were odorant-induced responses or artifacts. The Ca2+ responses in the soma of depolarized healthy neurons should be mediated mainly by the depolarization-induced activation of voltage-dependent Ca2+ channels, intracellular Ca2+ release and the Ca2+ extrusion system. These mechanisms predict a rapid increase of Ca2+ concentration, followed by a slow decay of the Ca2+ concentration, in the soma of ORNs. The kinetics of the Ca2+ concentration change should be determined by the kinetics of the relative transduction components, which are characteristically differentiated cell by cell. In fact, the odorant-induced responses showed a rapid initial decrease of fluorescence with reasonable onset latency, followed by a relatively slow exponential recovery to baseline in a cell-dependent manner. The initial fluorescence decrease could be fit to a logistic curve (b/(1 + c·exp(-at))) and the time constants of the fluorescence recovery could be reproducible with a statistical variation in each ORN. The fluorescence changes exhibiting these kinetics were subjected to analysis. Responses which started with slow fluorescence decreases, or whose time constants of exponential recovery differed from the averages by more than two standard deviations in each ORN, were attributed to artifacts, spontaneous changes or incomplete activations of the signal cascade and were rejected.
The odorant tuning specificities of individual ORNs were determined in
terms of their relative sensitivities to the test odorants, at least by the
sets of the most sensitive odorants. At a fixed concentration, the relative
sensitivity of an ORN was defined as low for odorants that evoked responses
with amplitudes of <60% of the largest response across all odorants. The
`first' (most sensitive) odorants were defined to induce responses at the
lowest concentration among all test odorants with amplitudes of >60% of the
maximum at the same concentration. The `second' odorants were the second-most
sensitive odorants, less sensitive than the first odorants and more sensitive
than the third odorants, in this definition. The validity of this
classification of ORN types is supported by the conservation of relative
sensitivities of M71-expressing ORNs to two odorants
(Bozza et al., 2002).
Previous studies have shown that most individual ORNs express only a single OR
gene (Chess et al.,
1994; Malnic et al.,
1999; Rawson et al.,
2000). Furthermore, the relative sensitivities of functionally
expressed exogenous ORs to subsets of odorants were coincident with those of
ORNs in which the mRNAs of the ORs were amplified after the physiological
assays (Touhara et al.,
1999; Kajiya et al.,
2001). Based on these results, we made the assumption that the
odorant specificities of ORNs represent those of ORs with the cAMP-mediated
signal amplifier.
It is necessary to minimize the intensity and time of the exposure of the
ORNs to ultraviolet light and also to minimize intracellular Ca2+
increases, in order to prevent the intracellular fura-2 and the mRNAs of ORs
from photo-bleaching and degrading in the ORNs before performing the
single-cell RT-PCR for OR identification. The success rate of single-cell
RT-PCR was reduced by a factor of 2.3 after the Ca-imaging assays
(Malnic et al.,
1999). Considering all these factors, we employed a weak
excitation light illumination, an interstimulus interval of 35 s and
single trials for each stimulus with the following two exceptions. When the
preceding stronger stimulus severely prevented the following weak stimulus
from evoking a response clear enough to evaluate the relative sensitivity, the
weak stimulus was again applied to the ORNs in the next recording. When the
odorant sensitivities severely decreased during a series of recordings, a set
of candidates for the most sensitive odorants were again assayed in the single
recording. Because all of the ORNs responded to 2 s 145.6 mM KCl stimulation
(hk) and most of them responded to 4 s 2 mM IBMX (ibmx),
which generates the receptor potential via cAMP increase, cell viability was
tested using these stimuli in the first set of recordings and near the end of
the experiment.
Chemicals
All chemicals for the odorants were commercially purchased from Aldrich. The stimulants used were: carvone (Ca), 5-isopropenyl-2-methyl-2-cyclohexenone; (-)menthone (mn), (2S,5R)-2-isopropyl-5-methylcyclohexanone; R(+)-pulegone (pu), (R)-p-menth-4(8)-en-3-one; isopulegol (ip), 2-isopropenyl-5-methylcyclohexanol; menthol (me), 2-isopropyl-5-methylcyclohexanol; R(+)-limonene (lim), (R)-1-methyl-4-(1-methylethyl)-cyclohexene; isoamyl acetate (am), 3-methylbutyl acetate; vanillin (va), 4-hydroxy-3-methoxybenzaldehyde; o-vanillin (ova), 3-methoxysalicyl-aldehyde; geraniol (ge), 3,7-dimethyl-(E)-2,6-octadiene-1-ol; nerol (ne), 3,7-dimethyl-(Z)-2,6-octadiene-1-ol; hexanoic acid (mc6); heptanoic acid (mc7); octanoic acid (mc8); nonanoic acid (mc9); 1-hexanol (mh6); 1-heptanol (mh7); 1-octanol (mh8); 1-nonanol (mh9); indole (in), 1-H-indole; triethylamine (ta); isovaleric acid (iv), 3-methylbutylic acid; potassium chloride (hk); and IBMX (ibmx), 3-isobutyl-1-methylxanthine.
The supplied reagents of S(+)-carvone (sCa) and
R(-)-carvone (rCa) contained small amounts of impurities. In
rCa, the major impurities were -terpineol and
cis-2,6-dimethyl-5,6-epoxy-1,7-octadien-3-ol, whose peak areas in gas
chromatography equipped with a flame ionization detector were 0.17 and 0.15%
of the total peak areas, respectively. The remaining 99.68% was occupied by
the peak of rCa. In sCa, the major impurities consisted of a
mixture of -terpineol + 
-terpineol, DH-isocarveol,
neoiso-DH-carveol, a mixture of ß-elemene + DH-carvone, a mixture of
DH-carveol + trans-6-p-menthen-2-ol +
cis-1,2-p-menthadien-8-ol, longifolene and DH-isocarvone,
whose peak areas in gas chromatography were 1.13, 0.46, 0.36, 0.31, 0.21, 0.21
and 0.17%, respectively. The peak area for rCa occupied 96.74% of the
total peak areas. Although we cannot exclude the possibility that these or
other minor ingredients, not the carvones, might have induced the responses in
a minority of assayed ORNs, we assumed that they did not severely distort the
assay of OR responsiveness for determining the most sensitive odorants in the
present study.
Isolation of OR cDNAs from single ORNs
Each ORN was transferred to a tube and the cDNAs were prepared from the
3' ends with pd(T)25-30 or Anchor T primer
(TATAgAATTCgCggCCgCTCgCgA(T)24) and then amplified as described
previously (Matsunami and Buck,
1997; Malnic et al.,
1999), but with a substitute SuperScript II (Gibco BRL) in some
ORNs. Briefly, we performed two-step PCR: the first PCR for all cDNA using the
AL1 primer (ATTggATCCAggCCgCTCTggACAAAATATgAATTC-(T)24 or Anchor T
primer, and the second PCR using OR-specific degenerate primers. The primers
used to amplify a region of the transmembrane domain 3 (TM3)-TM6 were P26
(GCITA(C/T)GA(C/T)CGI-TA-(C/T)GTIGCIATITG) and P27
(ACIACIGAIAG(G/A)-TGIGAI(G/C)C(G/A)CAI-GT). The purified PCR products for the
TM3—TM6 region were subcloned into pCR 2.1 or pCR II-TOPO vector
(Invitrogen) and sequenced using a gel DNA-sequencer (DSQ-2000L; Shimadzu Co.,
Japan). The dendrogram was generated using the Clustal X program
(Thompson et al.,
1997). The amino acid sequence of the TM3—TM6 region of each
OR was used for the sequence comparisons.
Behavioral assay for evaluating odorant discrimination capability of mice
We employed a Y-maze behavioral assay for determining whether or not mice
can discriminate sCa from rCa. The animals were treated in
accordance with the guidelines for the care and use of laboratory animals of
the National Institute of Health and of the AIST-Kansai Animal Experiment
Committee. The basic design of the Y-maze was similar to that previously
reported (Yamazaki et al.,
1999). In order to simplify the apparatus, we used negative
pressure generated by a dry vacuum pump downstream of the Y-maze instead of
positive pressure by a fan upstream of the Y-maze and omitted gates to control
the starts and stops of each mouse's run. The Y-maze was made of glass tubes
80 mm in inner diameter (i.d.). The negative pressure was weakened between the
Y-maze body and the vacuum pump by adjusting the opening of the leak line so
that the flow rate of influx fresh air was 0.7 ml/min at each arm of the maze.
The air influxes and effluxes passed through a 10 mm i.d. glass-tube port
centered at each terminal cap of the maze. At the glass-tube port of each
Y-maze arm, the influx air was odorized by being passed through a small
stainless steel mesh basket holding cotton balls immersed with diluted
odorants. Each mouse ran along the body of the maze for 18 cm from the
starting area and selected one arm of 45 cm length for the subsequent run. The
reward for odor concordance was a drop of water in a small glass funnel placed
inside the terminal cap of the arm of the maze. If the mouse went straight to
the funnel on the discordance odor side, the terminal cap with the funnel was
removed before the mouse started to drink the water. When identical odors were
presented at both arms, one of the two sets of odorized cotton balls had been
selected for the concordance before the series of the assay. The terminal caps
with the cotton balls were randomly exchanged between two arms. The glass
funnels were also sometimes exchanged between the two terminal caps. These
randomizations made the mouse pick each arm presenting the identical odor at
an even chance. The odorants were diluted to target concentrations in
propylene glycol. Identical odorant concentrations were prepared for both
arms. The concordance odorant was fixed to sCa for carvone enantiomer
discrimination in this study.
Results
Classification of odorant tuning specificities of ORNs
In our physiological assay system, the order of the relative sensitivities of individual ORNs among various odorants was conservative in repeated stimulations, at least for determining the most sensitive odorants, even though the sensitivities gradually decreased during the measurements. Figure 1 shows an example of the reproducibility of fluorescence changes normalized as shown in the Materials and Methods. The traces indicate changes in fura-2 fluorescence intensity excited at 380 nm, where a fluorescence decrease corresponds to a Ca2+ increase. The order of letters (A—D) represents the order of recordings. At a concentration of 100 µM, the ORN responded to mc7, mc8 and mc9 with identical net fluorescence changes, indicating the identical Ca2+ increases at the maximum (Figure 1A). As shown in Figure 1B,C, differences in the relative sensitivities appeared clearer in the response amplitude at a low odorant concentration than at a high concentration. The response reduction rate in the identical stimuli generally became greater with increasing number of responses between two recordings. At 100 µM, all net response amplitudes in fluorescence changes similarly decreased to 62% of the initial responses after several recordings (Figure 1A,D). At 10 µM, the responses to the three odorants were reduced into 83, 91 and 90% of the respective responses in the preceding recording, but the order of response amplitude was conservative between the two successive recordings (Figure 1B,C). Single or two successive recordings for object odorants helped us determine whether the relative sensitivities were clearly different or not. Based on these observations, we set the relative response amplitude to a borderline at 60% of the largest responses in order to separate relative sensitivities from those with clear differences.
|
First, we determined the odorant tuning specificities of the carvone-responsive ORNs for a series of test odorants. The test odorants were selected to test our hypothesis regarding odor identity decomposition into unique odor qualities by unique ORs and multiple common odor qualities by common ORs. In addition to the carvone enantiomers sCa and rCa, we tested odorants chosen for the similarities of their odor qualities to the odors of the carvones. For example, R(+)-pulegone (pu), (-)menthone (mn), isopulegol (ip), menthol (me), isoamyl acetate (am), nerol (ne), R(+)-limonene (lim) and normal aliphatic alcohols (mh6-9) with a 6-9 carbon chain were selected for the common fresh and/or herbal odor qualities possessed by both carvones. The first four odorants were also similar to rCa in that they had a minty odor. All of the 11 odorants above, as well as vanillin (va), o-vanillin (ova) and geraniol (ge), present somewhat sweet odors, as do both carvones. We also included some compounds with very different odors: indole (in), triethylamine (ta) and isovaleric acid (iv). For a statistical comparison with our previous data, normal aliphatic acids (mc6-9) with a 6-9 carbon chain were also included from our previous series.
Most ORs for carvones were responsive to multiple odorants, as previously
reported for other ORs (Sato et
al., 1994; Zhao et
al., 1998; Malnic et
al., 1999; Touhara et
al., 1999; Araneda et
al., 2000; Kajiya et
al., 2001; Bozza et
al., 2002). The responses of an ORN expressing an OR,
car-b85, are shown as odorant-induced intracellular Ca2+
increases (Figure 2). The
odorant tuning specificities were determined by relative response amplitudes
among the odorants. The cell viabilities were confirmed at the beginning of
the measurements (Figure 2A)
and at almost the end (Figure
2F) by KCl and IBMX stimulations. The criteria for responses
enabled us to identify small responses, such as those to 1 µM rCa
or sCa (Figure 2E).
The lowest (threshold) responsive concentration for an ORN was 1 µM, where
rCa, sCa, mn and pu induced responses with relative net
fluorescence decreases of 44, 25, 95 and 100%. Based on the relative response
amplitudes, the sensitivity of the ORN was determined as pu, mn >
rCa, sCa. At 10 µM, ip induced a response in the ORN with
a relative response amplitude of 48%
(Figure 2F), normalized to a
10-µM-pu response (Figure
2D), while 100 µM me also induced a response of 58% of
that evoked by 100 µM mn
(Figure 2B). Taken together,
the sensitivity among the test odorants was pu, mn > rCa,
sCa > ip, me. The response amplitudes of sCa and
rCa at 100 µM were reduced to 63 and 62% of the initial ones
(Figure 2A,E), similar to the
reduction in mc-induced responses shown in
Figure 1. This result indicated
that the sensitivity decrease was not too severe to compare the relative
response amplitudes in our criteria. Consequently, the odorant tuning
specificity of the ORN expressing the OR, car-b85, was determined as
pulmn sCa/rCa > ip/me. In the definition in this study
(described in Materials and methods), pulmn were the first odorants
(the most sensitive odorants) and sCa/rCa were the second odorants
(the second-most sensitive odorants). All time constants of the recovery
phases of the responses varied within two standard deviations (27.2 s) from
the average of 45.9 s.
|
Three ORNs expressing the identical ORs in the amino acid sequences showed
similar odorant tuning specificities (Figures
3 and
4). The ORNs expressing ORs,
car-b153 and car-b158, came from the same preparation and
the third ORN expressing an OR, car-b86, came from a different
preparation. The three ORNs were identical in having the odorant tuning
specificities of sCa > pu. Small fluorescence decreases
induced by 100 µM rCa in all three cells were not considered valid
responses but incomplete activations, because the time constants of the
recovery phases differed from the averages by 15 standard deviations in
individual ORNs (Figure 3; see
Materials and methods). The sensitivity of the car-b86 ORN was lower
than those of the other two ORNs by approximately half an order of magnitude.
The response of the car-b86 ORN to 100 µM pu was 50% of
that to 100 µM sCa and no clear responses were evoked by 10 µM
sCa, whereas the responses of the ORNs with car-b153 or
car-b158 to 100 µM pu were 88 and 100% of those to 100
µM sCa, respectively; clear responses were evoked only by
sCa at 10 µM. The ratio of the net fluorescence changes of IBMX
versus KCl in the car-b86 ORN at the beginning of the measurements
was smaller than those of the car-b153 and the car-b158 ORNs
at the beginning of the measurements and similar to those of the two ORNs at
the end of the measurements (Figure
3A,F,E'). This ratio is an indicator of signal amplification
by the cyclic-AMP second messenger system in individual ORNs. Thus, the
sensitivity of the car-b86 ORN might be reduced throughout its
receptive range, compared to the car-b153 and the car-b158
ORNs, but with little change in tuning specificity. Consequently, the odorant
tuning specificities of the same types of ORs/ORNs could be quite similar
except for differences in absolute sensitivities in some cases, as reported
previously (Bozza et al.,
2002). Based on these observations, we classified the odorant
responsiveness of ORs/ORNs by the first odorants and, supplementarily, the
second odorants.
|
|
To adequately sample nearly all types of carvone ORNs, we assayed 2740 ORNs
which is 2.7 times the estimated 1000 types of ORs
(Buck and Axel, 1991). Out of
2740 ORNs, 263 ORNs (9.6%) were responsive to sCa and/or
rCa. Out of 263 carvone ORNs, 234 ORNs were responsive to
sCa and 222 ORNs were responsive to rCa. First, the odorant
tuning specificities of ORNs were classified into four primary classes in
terms of their most sensitive odorants among the carvones and other odorants,
i.e. sCa-sensitive (n = 59, 22% of 263 ORNs),
rCa-sensitive (n = 49, 19%), sCa/rCa-equi-sensitive
(n = 100, 38%) and other-odorant-sensitive (n = 55, 21%).
They could be further divided into >41 subclasses by the combination of the
first and second odorants in the tuning specificities (data not shown).
However, most odorants, except for sCa or rCa, were not
tested against all ORNs. For example, pu and mn were not
tested in 105 carvone ORNs (40% of 263). Therefore, some additional subclasses
may emerge and the subpopulation of subclasses may change after exhaustive
assays have been completed.
Another possible source of variation of the ORN subpopulation of each subclass might be heterogeneous sampling of the four OR expression zones in the olfactory epithelium. We roughly estimated the largest variation of sampling rates among the four epithelial zones to be a factor of three for zone 1 versus zone 3, where the expected sampling number of ORNs was 320. Because the grand average of the subpopulations of OR types in a zone was 250 for a total of 1000 types of ORs throughout the four zones, the ORNs assayed in this study were expected to cover most types of ORs. We concluded that the largest subpopulation of ORs is the sCa/rCa equi-sensitive type and the second largest, the sCa-sensitive type. Our finding that the number of sCa/rCa-discriminating ORs was only about one-half the number of sCa/rCa-non-discriminating ORs indicates the need for an additional mechanism for combining receptor signals to produce the characteristically distinct perceived odors of the carvones.
In order to estimate how many ORs are responsive to sCa and
rCa, we tried to clone ORs from ORNs by single-cell RT-PCR using
degenerate primers, after completing the physiological assays for odorant
responsiveness. We obtained OR cDNA products from 28/103 ORNs responsive to
carvones. They included four pairs and two trios of the same ORs, resulting in
a total of 20 types of ORs. Because most of the ORN subclasses were subjected
to single-cell RT-PCR in a random manner (data not shown), we employed an
OR-type multiplicity factor of 28/20 in this study. This ratio produced an
estimated 70 ± types of carvone ORs from a 9.6% subpopulation
of responsive ORNs in 1000 ORs. The RT-PCR success rate was 5/6 with no
Ca-imaging assays. These rates before or after the Ca-imaging assays were
similar to those of 18/26 and 14/47 reported previously
(Malnic et al.,
1999). This accordance indicates that our single-cell RT-PCR would
amplify a large variety of ORs as well as in the previous study.
Sensitivity profiles of carvone-responsive ORs among test odorants
Figure 4 shows the
sensitivity profiles of 28 sCa- or rCa- responsive ORNs (20
types) and one mn-sensitive ORN that was responsive to neither
sCa nor rCa. The sensitivities are demonstrated by the
threshold concentration for each odorant. In most cases, the set of ORs with
identical sequences showed quite similar sensitivity profiles for test
odorants, as described above. As shown previously
(Malnic et al.,
1999), one odorant is recognized by multiple ORs, in a manner in
which different odorants are recognized by different combinations of ORs with
some overlaps. In this study, we focused the difference in the classified
sub-populations of sensitive ORNs as described in the next section.
The TM3—TM6 regions of 18 types of carvone ORs were compared with 97 ORs previously identified in mice (Figure 5). Seven of the 18 carvone ORs were grouped in the same cluster, indicating greater similarity to one another than to other ORs. Out of a possible 153 pairs among 18 ORs, 125 pairs (82%) possessed 35-45% amino acid identities among the TM3—TM6 regions and two pairs showed higher identities of >77%.
|
Additionally, we would like to mention an interesting observation: three
ORs marked by asterisks in Figure
4— car-c10, car-n272 and
car-b154—showed a synergistic effect of rCa and
sCa. The indicated thresholds were observed for sCa after
rCa stimulation with an interstimulus interval of 35 s. When
sCa was applied prior to rCa, the threshold concentrations
of sCa were elevated by more than one order of magnification. This
phenomenon suggests the existence of an intermediate activation state, in
addition to the resting states and activated states of some ORs.
Receptor codes for odors of enantiomeric odorants sCa and rCa
How is odor quality encoded in odorant receptors? To discover the strategy used to encode odor qualities in ORs, we evaluated overlap rates of the activated receptors between each of the carvones and the test odorants and compared them with overlaps of submodal odor qualities of the odorants (Table 1). The receptor overlap rates were highest between sCa and rCa (>80%), followed by pu, mn and ip in decreasing order. Based on the uniformly weighted summation of ORN signals in a simple combinatorial receptor coding scheme, the observed overlap of >80% most likely leads to expected, almost-identical odors for both odorants, which are inconsistent with the perceived characteristic odors of sCa and rCa.
|
More detailed comparisons are described as follows. The principal strategies of odor encoding and decoding may be similar across mammalian species. Because the submodal odor qualities have not been determined in mice, the odor qualities of the odorant were substitutively represented by those perceived by humans. The odorants sCa, rCa and pu, whose receptors overlap with the carvone receptors ranging from 51 to 87% in cross-comparison, commonly possess three submodal odors of `fresh', `herbal' and `sweet'. However, the odorants mh7 and mh8, which also possess `fresh', `herbal' and `sweet' odors, shared common receptors in the receptor codes with carvones only at 19 or 21%. Such a smaller receptor overlap rate might represent the perceived strengths of `fresh', `herbal' and `sweet' odors in mh7 and mh8 to a weaker degree than in sCa, rCa and pu. In contrast, the receptor overlap rate between the carvones and va, which presents a more intense `sweet' odor than the carvones, was only 2%. This small overlap rate may be attributed to an existence of different `sweet' odor qualities or a difference between species. The overlaps of the carvone receptors were, respectively, 13 and 21% with the receptors for mh6 and mh9, aliphatic alcohols that also possess a `sweet' odor and a `herbal' or `fresh' odor. Additionally, the odorants mn, ip, am, ova, ge and ne, showing receptor overlap rates ranging from 7 to 43%, also present a `sweet' odor. This finding suggests that the common ORs for these odorants may partially contribute to a common `sweet' odor quality among the odorants.
In contrast, the fraction of carvone ORNs that were responsive to the aliphatic acids mc6 or mc7 (with no clear odor qualities shared with the carvones) was less than one-sixth of the fraction responsive to the aliphatic alcohols mh6 or mh7 (with three common odor qualities, as described above). This was so, despite the similarity of the molecular structures of mc and mh, except for the different terminal functional groups of carboxyl versus hydroxyl. These results could lead to the following hypothesis: as the overlapped subpopulation of carvone ORNs among test odorants increased, the common odor qualities increased in number and/or intensity. In other words, a common odor quality among multiple odorants may be encoded by subsets of common ORs in the manner of a weighted and integrated population. Furthermore, a small surplus in the receptor overlap rate (4%) was observed in the rCa—mn pair with respect to the sCa—mn pair. It is likely that the surplus carvone ORs may contribute to a `minty' odor characteristic of rCa but not to that of sCa. We observed some ORs that were mn-sensitive and sCa/rCa-non-responsive (Figure 4). Such mn-unique ORs should contribute to the composition of an mn-unique odor quality.
Three odorants (mc8, mc9 and ta) that smelled quite
differently (repulsive, nut-like and fishy, respectively) from the carvones
induced responses commonly in 10% of the carvone ORNs and most of their
sensitivities were not particularly high. This result leads to the hypothesis
that more sensitive ORs contribute strongly to the representation of the
respective odor qualities and less sensitive ORs contribute only weakly. If
the ORN sampling procedure in our experiments was relatively unbiased, then
the odorant sensitivities of our assayed ORNs should provide a good estimate
of the subpopulation of carvone ORs in the entire olfactory epithelium. Based
on this assumption, we compared the actual numbers of the classified ORNs
sensitive to sCa or rCa at 1 or 10 µM.
Our analysis revealed that a difference between the receptors classified by the most sensitive odorants characteristically appeared at the lowest effective stimulus concentrations and the sCa/rCa-non-discriminative ORNs became dominant at 10 times the lowest concentration (Table 2). At 1 µM, rCa activated four rCa-sensitive ORNs, two sCa/rCa-equi-sensitive ORNs and one of each of the other five types of ORNs, whereas sCa activated two sCa/rCa-equi-sensitive ORNs, two sCa-sensitive ORNs, two rCa-sensitive ORN and one of each of the other three types of ORNs. Thus, the four rCa-sensitive ORNs were the largest subpopulations among the classified rCa-most-sensitive ORNs, whereas the two sCa-sensitive ORNs and the two sCa/rCa-non-discriminative ORNs were the largest subpopulations among the classified sCa-most-sensitive ORNs. The rCa-sensitive ORNs activated by 1 µM sCa were not included in the most sensitive types, because the responses were smaller in the percentage maximum amplitude than those of the sCa/rCa-equi-sensitive ORNs. Interestingly, the principal odor qualities of the target odorants were represented by the principal (common) odor qualities of the most sensitive odorants of the most sensitive ORs which occupied the largest subpopulation at the lowest effective stimulus concentration, although the murine OR codes applied to the odor qualities perceived by humans. This observation led to the hypothesis that the signals of the most sensitive ORs of the largest subpopulation may govern the signal processing for the odor identity representation in the brain.
|
By increasing the stimulus concentration up to 10 µM, 19 sCa-sensitive ORNs and three rCa-sensitive ORNs were recruited by sCa, 14 rCa-sensitive ORNs and six sCa-sensitive ORNs were recruited by rCa, in addition to 39 sCa/rCa-equi-sensitive ORNs. The largest subpopulation of the sCa/rCa-equi-sensitive ORNs, which were one subclass of the most sensitive ORNs for sCa, could explain the stronger `herbal' or `sweet' quality, which is common to sCa and rCa, in the odor of sCa, compared to rCa. However, based on the simple combinatorial receptor coding scheme, it is difficult to understand the fact that the rCa-sensitive ORNs, which were approximately one-third of the sCa/rCa-equi-sensitive ORNs activated by 10 µM rCa, formed a `minty' odor of rCa more intensively than the `herbal' quality of rCa. These results strongly indicate that the signals from the most sensitive ORs govern olfactory information processing by selectively composing OR signals that represent principal odor qualities of the target odorants, instead of a simple combinatorial integration of receptor signals to synthesize odor qualities. This also suggests that 1-µM-rCa-activated signals may feed into an inhibitory circuit that suppresses the less sensitive and different submodal signals in the brain. The receptors most sensitive to the odorant provide the signaling framework that defines the unique odor quality of the target odorant and the multiple subsets of ORs less-sensitive to the odorant contributes to the supplementary odor qualities of the target odorant. Thus, the sensitivity-dependent hierarchical receptor codes can well describe odor identity encoding and decoding in the olfaction.
Carvone enantiomer discrimination in mice
sCa and rCa have been reported to be discriminative in
humans, monkeys and rats (Friedman and
Miller, 1971; Leitereg et
al., 1971; Boelens et
al., 1993; Laska et
al., 1999; Linster et
al., 2001), but not in mice. In order to determine whether
mice can discriminate sCa from rCa, we conducted behavioral
assays using a Y-maze. One NZB mouse and one CBA mouse were initially trained
to discriminate amyl acetate from mineral oil. After the concordance trials
occupied >80% of each trial set (12 or 24 trials), the mice were tested to
discriminate sCa from rCa. The concordance trials
immediately increased to >80% (Figure
6). This value was significantly different from even chance in the
2 test (P < 0.05), indicating that the mice could
discriminate sCa from rCa. The mice successively performed
significant discriminations for more than eight sets of trials, even when the
odorant concentrations were gradually reduced to 1/4000 v/v step by step. The
NZB mouse demonstrated a decrease in the concordance trial rate at 1/40 000
v/v and the following 1/400 000 v/v (Figure
6A). The mouse could not discriminate sCa from the
identical sCa at 1/100 v/v or 1/2000 v/v in this Y-maze assay.
However, the NZB mouse demonstrated a significant concordance trial rate for
discriminating sCa from rCa at 1/80 000 v/v after the
sCa versus sCa trials
(Figure 6A).
|
The CBA mouse also demonstrated significant results in the Y-maze
behavioral assays for discriminating sCa from rCa
(Figure 6B). The CBA mouse lost
discriminative performance for a pair of sCa and rCa at
1/400 000 v/v and 1/4 000 000 v/v. The concordance trial rates recovered to
>80% by elevating the concentrations up to 1/20 v/v. After the mouse
demonstrated >80% concordance performances, the significance of concordance
again showed a decrease to near the level of P = 0.5
(2 test) at a concentration of 1/400 000 v/v. The concordance
trial rates for the following identical pair trial of sCa further
decreased to the level of even chance. These results were preliminary in
evaluating the lowest concentration for discrimination between sCa
and rCa. However, the results demonstrated that mice can discriminate
sCa from rCa at concentrations >1/4000 v/v, where most
carvone ORs of mice might be activated intensively.
Discussion
Odorants seem difficult to discern because of their small molecular size and slight structural differences across many compounds. How is odor quality encoded during olfactory reception? The biological systems are generally hierarchically structured and derived from layers of simpler functional modules. Salient coding pathways in neural systems may be predetermined by genetic hard-wiring and modified via experience-driven plasticity. In the present study, we explored the structure of pathways encoding odor quality at the peripheral level by undertaking a detailed analysis of the receptor codes for a chiral pair of molecules, the S(+)- and R(-)-carvone, in the sensitivities and the subpopulations classified by the odorant tuning specificities of ORNs. In our data analysis and interpretations, we have made the working assumption that there is a general correspondence between human ORs and respective murine ORs regarding odorant tuning specificity and, simultaneously, that there are no large differences in OR members among the three mouse strains CBA, BALB/c and NZB. The principal strategies of odor encoding and decoding may be similar across mammalian species. Surprisingly, the hierarchical receptor code hypothesis can explain the human-perceptive odor qualities characteristic of the carvone enantiomers in murine receptor codes. This result suggests that our assumptions may be not greatly different from nature. As learned from the Human Genome Project, the expected number of human ORs is about one-third that of murine ORs, after omitting pseudogenes. The large family size of murine ORs may indicate that odor discrimination capabilities should be generally higher in mice than in humans. The validity of our working hypothesis is still subject to future verification by functional assays of human ORs.
Differential responses to carvone enantiomers have been observed previously
in the assays of ORNs/ORs (Krautwurst
et al., 1998; Ma and
Shepherd, 2000), in the assays of limited subsets of glomeruli in
the olfactory bulb (Taniguchi et
al., 1992; Rubin and Katz,
1999,
2001) and entirely in the
olfactory bulb at a single concentration
(Linster et al.,
2001). However, the differences in their global receptor codes
have remained unclear. We attempted to screen nearly all types of ORs/ORNs
responsive to the carvone enantiomers by assaying ORNs numbering more than
twice the expected 1000 types of ORs from all four epithelial zones. We found
that 9.6% of the cells responded to one or both carvones. More than 80% of
carvone-responsive ORs demonstrated overlapping sensitivities between the
enantiomers. This finding is difficult to reconcile with a simple
combinatorial scheme in which the responses of all ORs are equally weighted.
Such a scheme predicts enantiomers with closely similar, not clearly
distinctive, odor qualities at moderate stimulus concentrations. Their odors
would be dominated by the common odor quality encoded by the
non-discriminating ORs, which occupy some 40% of carvone ORs and outnumber the
discriminating ORs 2-fold. Contrary to this prediction, the present study
demonstrates that mice can discriminate sCa from rCa at
relatively high concentrations in the Y-maze assay.
Furthermore, the receptors that were activated at the lowest effective stimulus concentration represented the principal odor qualities of each enantiomer by means of the principal odors of the most sensitive odorants for the receptors. These characteristic receptors in the receptor codes became a minority in the subpopulation at 10 times the lowest effective concentration. The combinatorial scheme predicts a dramatic change in the principal odor qualities of rCa, corresponding to the shift of the major subclass along the odorant concentration increase. To produce the divergent and constant perceptual odor qualities of the two chiral forms, we postulate that the olfactory system imposes non-uniform weighting on the OR inputs, granting higher priority to the more specific, more sensitive, chiral-discriminating ORs. This would allow minor subpopulations of chiral-discriminating ORs to govern the odor quality and counterbalance numerically superior sets of chiral-non-discriminating ORs. In general, we hypothesize that single-odorant-sensitive ORs contribute primarily to unique odor qualities and that multiple-odorant-equi-sensitive ORs contribute to the encoding of common odor qualities, and that odor identity is represented by composing OR signals for submodal odors of the target odorant in a sensitivity-dependent hierarchical manner.
The number of 1-µM-rCa-sensitive ORs was twice that of the
1-µM-sCa-sensitive ORs, although the total sample size was small.
If the human OR system is an analogous to the murine OR system except for the
large member size and a small subpopulation of ORs most sensitive to
rCa strongly contributes to the sensitivity to rCa, one can
easily understand the higher sensitivity to rCa compared to
sCa observed in the perceptual threshold
(Polak et al., 1989).
Various specific anosmias are defined as the significant elevation of sensory
thresholds for particular odorants (Amoore
and Steinle, 1991). In our hierarchical sensitivity model,
specific anosmias arise naturally as the genetic disruption of
odor-identity-governing ORs, or as a reduction in their weighted central
integration. Interestingly, the defect degrees of human specific anosmias were
almost twice as great in rCa in the logarithmic threshold than in
sCa (Pelosi and Viti,
1978). It is more difficult to explain specific anosmias to
odorants with tens of responsive receptors in terms of the combinatorial model
of odor coding.
Odorant receptor signals converge onto OR-type-specific glomeruli in the
olfactory bulb. The convergent pathways genetically form a stereotypic map of
OR signal inputs to the brain. Odorant-characteristic glomerular maps
activated near the threshold concentration in the olfactory bulb have also
been reported for the other odorants
(Wachowiak and Cohen, 2001).
They also reported significant overlaps of glomerular maps between different
odorants at higher concentrations, as shown here. The signals generated by ORs
most sensitive to the most intense compounds in a complex odor stimulus should
arrive first at neurons anywhere in the main olfactory pathway. Filtering by
inhibitory networks in the central pathways should favor the first input
signals over the later ones from the relatively less-sensitve ORs. These first
input signals from the most sensitive ORs are expected to play an important
role in maintaining the characteristic odor representation over a wide range
of stimulus concentrations. The fast and slow inhibitions in the piriform
cortex (Satou et al.,
1982a,
b) may play an important role
in such a signal filtering. The possible mechanisms for emphasizing
odorant-unique signals include the dendrodendritic synaptic inhibition in the
olfactory bulb (Yokoi et al.,
1995). The dynamic optimization of odor representations as mitral
cell activity patterns (Friedrich and
Laurent, 2001; Spors and
Grinvald, 2002) might also be related to the enhancement system of
characteristic signals for given odorants.
The subpopulations of pyramidal cells, which receive signals from a single
type of OR and formed a few stereotypical clusters, were calculated as 2.4 and
3.7% (Zou et al.,
2001). The subpopulations of 30 times the 0.1% corresponding
to that of a single OR suggest that the pyramidal cells in each cluster may
summate the signals of different types of ORs by synchronized inputs
(Kashiwadani et al.,
1999) and/or by the association fiber system
(Haberly, 2001). The pyramidal
cells may represent submodal odor qualities, such as the primary odors
investigated by Amoore and other groups. The signal summations of selected ORs
may be organized by odorant tuning specificity.
Almost half of the carvone-responsive ORNs could not discriminate
enantiomeric structures, while the remaining half could do so. Single
carvone-responsive ORs could recognize diverse combinations of odorants with
various sensitivities. Furthermore, we have found that
aliphatic-odorant-sensitive ORNs/ORs could discriminate the molecular length
of an odorant at a resolution of a single carbon atom in a dose-dependent
manner (Sato et al.,
1994; Malnic et al.,
1999). This suggests that target interaction sites should include
two sites around the major axial terminals of an odorant molecule. Considering
these results, each OR may interact with and recognize odorant molecules at
multiple sites in such a manner that different odor molecules select different
combinations of interaction sites in a given OR's odorant-binding pocket. Some
receptor sites tightly bind different parts of a given odorant molecule,
endowing the receptor with a high sensitivity and specificity for the unique
three-dimensional profile of intermolecular interactions with that odorant.
Other sites may not bind the given odorant, but may loosely bind other
odorants, conferring lower sensitivity to those stimuli. This principle of
`tight-and-loose profiling' results in well-defined odorant tuning curves, or
tuning surfaces in a multidimensional odorant space. This hypothesis is also
supported by psychochemical studies (Braunand Kröper, 1937;
Moncrieff, 1949;
Amoore, 1970;
Beets, 1982;
Ohloff, 1986).
A schematic diagram of receptor codes for odors is presented in Figure 7. Most hydrogen atoms are omitted and one of the possible molecular structures of nonanol is shown. The ORs are classified into subfamilies such as the sCa-sensitive, rCa-sensitive, or sCa/rCa-equi-sensitive types by the most sensitive odorants (Figure 7A). Each subfamily consists of multiple members whose tuning specificities are partially different. Each OR is tuned to a specific three-dimensional configuration of intermolecular interaction sites. Sensitivity decreases as differences from the tuned intermolecular interaction profiles increase. Consequently, one OR can discriminate one pair of odorants while another cannot. Odorant-unique ORs contribute mainly to the generation of an odorant-unique odor quality, according to their specific sensitivities. Multiple odorant equi-sensitive ORs contribute to the generation of common odor qualities among the multiple odorants. Common ORs could also assist in emphasizing the signals of unique ORs via inhibitory circuits.
|
Figure 7B illustrates the
same hypothesis in a different way. As the concentration of odorant increases,
each odor quality becomes more intense by recruiting new members of respective
OR subclasses, in addition to the signal growth in each OR. Each of submodal
odor qualities is represented by composing the signals of multiple different
ORs. The principal odor qualities of the odorant are encoded by the most
sensitive ORs and additional less-sensitive similar tuning-type ORs. As shown
in Figure 7C, if all ORs most
sensitive to odorant C are more sensitive to odorant A
(odor X) and their subpopulation is small, we may recognize odorant
C as an indeterminate odor (i.e. odor X-like). If the
odorant is biologically very important for the animals to discriminate from
other odorous stimulants, a subset of ORs of lower-ranked sensitivity to the
odorant may play a dominant role in composing odor representation signals
instead of the most sensitive receptors. In color vision, the hierarchical
receptor code may explain `yellow' as the principal color common to the
long-wave-sensitive receptor and the medium-wave-sensitive receptor, based on
the significantly overlapping wavelength range between the two cones
(Bowmaker and Dartnall, 1980).
Various `red' colors are also explained as those principally encoded by the
long-wave-sensitive cones which are most sensitive to `red'.
We possess three remote sensory systems—vision, audition and olfaction. We have also developed recording systems that help us to recognize, remember and share transient experiences pertaining to the first two modalities. However, we have yet to develop an olfaction-compatible information recording system that shares transient odor information. In order to develop olfactory informational devices that have the same sophistication as video and audio systems, it is important to gather knowledge on the detailed profiles of odor receptor codes in the olfactory system. Data gathered by the type of extended approach applied in this study may facilitate the future development of innovative devices for recording and presenting olfactory experiences in a universal coordinate system.
Acknowledgments
This work was supported by grants from METI, Japan (T.S.) and from MEXT, Japan (H.H.). We would like to thank Drs Shigeru Yamane, Tetsuo Moriya and Kenji Kawano for their directorial support concerning the research budget, Drs Linda Buck, Bettina Malnic and Hiroaki Matsunami for their help in establishing the single-cell RT-PCR system in our laboratory and Dr Kunio Yamazaki for his helpful advice on Y-maze behavioral assays. We would also like to thank Mahoko Takeuchi for her technical assistance with the Y-maze assays, Drs Tsuneyoshi Kanisawa, Makoto Emura, Kazuhiko Tokoro and Takahiro Ishikawa for kindly agreeing to conduct the analysis of chemical impurities using GC-MS at Takasago International Corp. Our heartfelt thanks also go to Drs Graeme Lowe, Gary Beauchamp and Daniel Funeriu for their critical discussion and their helpful comments in rewriting the manuscript. Finally, we would like to thank Drs Kotaro Takahama and Masamine Tekebayashi for their helpful discussion regarding the analogy to color vision. H.H. and J.H. contributed equally to this work. H.H. is a Domestic Research Fellow of the Japan Society for the Promotion of Science.
References
Amoore, J.E. (1970) Molecular Basis of Odor. Charles C. Thomas.
Amoore, J.E. and Steinle, S. (1991) A graphic history of specific anosmia. In Wysocki, C.J. and Kare, M.R. (eds), Chemical Senses Vol.3: Genetics of Perception and Communication. Marcel Dekker, pp.331 -351.
Araneda, R.C., Kini, A.D. and Firestein, S. (2000) The molecular receptive range of an odorant receptor. Nat. Neurosci., 3,1248 -1255.[CrossRef][ISI][Medline]
Arctander, S. (1969) Perfume and Flavor Chemicals: Aroma Chemicals. Steffen's Arctander's Publications, Las Vegas, NV.
Beets, M.G.J. (1982) Odor and stimulant structure. In Theimer, E.T. (ed.), Fragrance Chemistry. Academic Press, pp. 77-122.
Boelens, M.H., Boelens, H. and van Gemert, L.J. (1993) Sensory properties of optical isomers.Perfum Flavor , 18,1 -16.
Bowmaker, J.K. and Dartnall, H.J. (1980)
Visual pigments of rods and cones in a human retina. J.
Physiol., 298,501
-511.
Bozza, T.C. and Kauer, J.S. (1998)
Odorant response properties of convergent olfactory receptor neurons.J. Neurosci.
, 18,4560
-4569.
Bozza, T.C., Feinstein, P., Zheng, C. and Mombaerts,
P. (2002) Odorant receptor expression defines functional
units in the mouse olfactory system. J. Neurosci.,22
, 3033-3043.
Braun, J.V. and Kröper, H. (1937) Geruch und Konstitution (I. Mitteil.). Ber. Dtsch. Chem. Ges., 62,2880 -2885.
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][ISI][Medline]
Buck, L. (1996) Information coding in the vertebrate olfactory system. Annu. Rev. Neurosci.,19 , 517-544.[CrossRef][ISI][Medline]
Chess, A., Simon, I., Cedar, H. and Axel, R. (1994) Allelic inactivation regulates olfactory receptor gene expression. Cell, 78,823 -834.[CrossRef][ISI][Medline]
Floriano, W.B., Vaidehi, N., Goddard, W.A., III, Singer,
M.S. and Shepherd, G.M. (2000) Molecular
mechanisms underlying differential odor responses of a mouse olfactory
receptor. Proc. Natl. Acad. Sci. USA,97
, 10712-10716.
Friedman, L. and Miller, G.J. (1971)
Odor incongruity and chirality. Science,172
, 1044-1046.
Friedrich, R.W. and Laurent, G. (2001)
Dynamic optimization of odor representations by slow temporal patterning
of mitral cell activity. Science,291
, 889-894.
Haberly, L.B. (2001) Parallel-distributed
processing in olfactory cortex: new insights from morphological and
physiological analysis of neuronal circuitry. Chem.
Senses, 26,551
-576.
Hirono, J., Sato, T., Tonoike, M. and Takebayashi, M. (1991) Increases in cytoplasmic free calcium concentration and membrane electrical responses to odor in isolated olfactory receptor neurons. Proc. Jap. Symp. Taste Smell,25 , 193-196.
Hirono, J., Sato, T., Tonoike, M. and Takebayashi, M. (1992) Simultaneous recording of [Ca2+]i increases in isolated olfactory receptor neurons retaining their original spatial relationship in intact tissue. J. Neurosci. Methods, 42,185 -194.[CrossRef][ISI][Medline]
Johnson, B.A. and Leon, M. (2000) Modular representations of odorants in the glomerular layer of the rat olfactory bulb and the effects of stimulus concentration. J. Comp. Neurol., 422,496 -509.[CrossRef][ISI][Medline]
Kajiya, K., Inaki, K., Tanaka, T., 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.
Kaluza, J. and Breer, H. (2000) Responsiveness of olfactory neurons to distinct aliphatic aldehydes.J. Exp. Biol. , 203,927 -933.[Abstract]
Kashiwadani, H., Sasaki, Y.F., Uchida, N. and Mori,
K. (1999) Synchronized oscillatory discharges of
mitral/tufted cells with different molecular receptive ranges in the rabbit
olfactory bulb. J. Neurophysiol.,82
, 1786-1792.
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][ISI][Medline]
Kurahashi, T. (1990) The response induced
by intracellular cyclic AMP in isolated olfactory receptors of the newt.J. Physiol.
, 430,355
-370.
Laska, M., Liesen, A. and Teubner, P.
(1999) Enantioselectivity of odor perception in squirrel
monkeys and humans. Am. J. Physiol.,277
, R1098-R1103.
Leinders-Zufall, T., Greer, C.A., Shepherd, G.M. and
Zufall, F. (1998) Imaging odor-induced calcium
transients in single olfactory cilia: specificity of activation and role in
transduction. J. Neurosci., 18,5630
-5639.
Leitereg, T.J., Guadagni, D.G., Harris, J., Mon, T.R. and Teranishi, R. (1971) Evidence for the difference between the odours of the optical isomers (+)- and (-)-carvone.Nature , 230,455 -456.[CrossRef][Medline]
Linster, C., Johnson, B.A., Yue, E., Morse, A., Xu, Z., Hingco,
E.E., Choi, Y., Choi, M., Messiha, A. and Leon, M.
(2001) Perceptual correlates of neural representations evoked
by odorant enantiomers. J. Neurosci.,21
, 9837-9843.
Ma, M. and Shepherd, G.M. (2000)
Functional mosaic organization of mouse olfactory receptor neurons.Proc. Natl. Acad. Sci. USA
, 97,12869
-12874.
Malnic, B., Hirono, J., Sato, T. and Buck, L. (1999) Combinatorial receptor codes for odors.Cell , 96,713 -723.[CrossRef][ISI][Medline]
Matsunami, H. and Buck, B.L. (1997) A multigene family encoding a diverse array of putative pheromone receptors in mammals. Cell, 90,775 -784.[CrossRef][ISI]