The olfactory bulb (OB) is the first relay station of the central olfactory system in the mammalian brain and contains a few thousand glomeruli on its surface. Because individual glomeruli represent a single odorant receptor, the glomerular sheet of the OB forms odorant receptor maps. This review summarizes the emerging view of the spatial organization of the odorant receptor maps. Recent studies suggest that individual odorant receptors are molecular-feature detecting units, and so are individual glomeruli in the OB. How are the molecular-feature detecting units spatially arranged in the glomerular sheet? To characterize the molecular-feature specificity of an individual glomerulus, it is necessary to determine the molecular receptive range (MRR) of the glomerulus and to compare the molecular structure of odorants within the MRR. Studies of the MRR mapping show that 1) individual glomeruli typically respond to a range of odorants that share a specific combination of molecular features, 2) each glomerulus appears to be unique in its MRR property, and 3) glomeruli with similar MRR properties gather together in proximity and form molecular-feature clusters. The molecular-feature clusters are located at stereotypical positions in the OB and might be part of the neural representation of basic odor quality. Detailed studies suggest that the glomerular sheet represents the characteristic molecular features in a systematic, gradual, and multidimensional fashion. The molecular-feature maps provide a basis for understanding how the olfactory cortex reads the odor maps of the OB.
The olfactory system plays a key role in the daily life of mammals. Major roles of the olfactory system include identification of foods; judging the edibility; detection of impending danger such as predator; recognition of their mate, parents, and offspring; and detecting signals for a variety of social behaviors including the maintenance of territories.
The olfactory information is carried in a vast variety of odor molecules (odorants), small volatile compounds with molecular mass <300 Da. Each object, an apple for example, typically emits hundreds of different odorants. It has been estimated that more than 400,000 different compounds are odorous to the human nose (31). More surprisingly, no two compounds have ever been found to have exactly the same odor quality (9), suggesting that the olfactory system can detect and discriminate more than 400,000 different compounds. How does the olfactory system recognize and discriminate numerous different odorants? This is one of the basic and challenging questions in the field of olfactory research.
In addition to the question of odorant discrimination, olfactory research has been challenging the difficult puzzle of the molecular structure-odor relationship (85). Since the discovery of the methods for determining the molecular structure of odorants, chemists, especially fragrance/flavor chemists, have noted a variety of correlations between the molecular structure of odorants and their perceived “odor.” In addition, psychophysical studies of olfaction were pursued to elucidate the rules for odorant structure-odor relationships (3, 9, 63, 81). It is surprising that there exist relationships between the molecular structure and “odor,” in spite of the large gap between the two. The structure of odorants is at the molecular level in the range of nanometers, whereas the perceived odor of the odorants is the subjective description of human olfactory sense.
The knowledge of the structure-odor relationship is fragmental and far from complete. Furthermore, postulated relationship rules contain many exceptions. Nonetheless, studies on the structure-odor relationship have accumulated a large number of specific examples and provided theoretical frameworks of individual specific examples that can be tested experimentally in the olfactory system of both human and animals. These studies thus raised a notion that the information of molecular features detected and processed at the levels of olfactory sensory neurons and olfactory bulb (OB) may play a key role in the perception of odor quality of odorants in the brain. Thus olfactory research faces another important question: What is the neural basis for the molecular structure-odor relationship? To this end, the use of panels of odorants with systematic variation of molecular structure for sensory stimulation is crucial.
The OB is the first relay station of the central olfactory system in the mammalian brain (Fig. 1) and has a cortical structure with distinct layers and numerous glomerular modules (66, 100). Individual sensory neurons in the olfactory epithelium project a single axon to a single glomerulus in the OB. The glomerular sheet of the OB thus forms a map of olfactory axon terminals. The axonal projection of olfactory sensory neurons to the glomeruli is precisely organized in such a way that olfactory sensory neurons expressing a given odorant receptor (OR) converge their axons to a few topographically fixed glomeruli (Fig. 1) (13, 62, 68, 83, 122). Since individual glomeruli represent a single OR, the glomerular sheet of the OB forms a map of ORs.
How are individual odorants encoded in the glomerular OR map in the OB? To address this question, it is necessary to record and map the odorant-evoked activity in the glomerular sheet of the OB. Because individual mitral/tufted cells in the OB send a single primary dendrite to a single glomerulus, the glomerular OR map can be studied also by examining odorant-evoked responses of mitral/tufted cells. In the 1950s, Adrian (1) recorded odorant-evoked neuronal activity in the mammalian OB, and first demonstrated the odorant selectivity of responses of individual mitral/tufted cells. Spatial mapping of the odorant-evoked activity was first performed by the Shepherd's group using the 2-deoxyglucose (2-DG) uptake method (97, 114). These pioneering 2-DG studies demonstrated that each odorant elicits a characteristic spatial pattern of glomerular activities in the OB and provided the first information regarding the relationship between molecular structure of odorants and the spatial pattern of glomerular activities. In addition, the 2-DG studies showed a first hint that the odorant-induced activity is represented symmetrically between lateral and medial hemispheres of the OB (100).
Since then, a variety of methods have been used to spatially map the odorant-induced activities of glomeruli or mitral/tufted cells in the vertebrate OB. The mapping methods include 2-DG uptake (39, 41, 42, 52, 97, 111, 114); optical imaging with voltage-sensitive dyes (23, 45, 46, 112); electrophysiological recording from individual mitral/tufted cells (37, 44, 65); expression of immediate early genes such as c-fos (29, 78, 91), c-jun (7), zif268 (38), and activity-regulated cytoskeleton-associated (ARC) genes (56, 57); optical imaging of intrinsic signals (10, 36, 59, 88, 116, 120); expression of phosphorylated ERK (61); imaging with calcium-sensitive dyes (24, 124); imaging using pHluorin (11); and fMRI (128, 129).
The results clearly showed that a given odorant activates a specific combination of glomeruli that are located and distributed at stereotypical positions in the OB. Different odorants activate a distinct combination of glomeruli. The odorant-specific spatial positions of activated glomeruli are conserved across different animals of the same species. The principle of odor mapping is expressed widely by different vertebrate species (25, 76) and by invertebrates such as honeybees (26, 90), moths (32, 51), and flies (49, 125). Thus, either in the OB or in the antennal lobe, the insect analog of the vertebrate OB, the principle of odor mapping is conserved.
The mapping methods can be classified into two groups in terms of advantages and disadvantages in the mapping of odorant-induced activities in the OB. The first group of methods has the advantage of the ability to map the responses over the entire OB and includes the methods using 2-DG uptake, immediate early gene expression, phosphorylated ERK expression, and fMRI. Studies with these methods addressed the question about how individual odorants are represented in the space of entire glomerular sheet of the OB. For many different odorants, these studies revealed the odorant-specific spatial pattern of glomerular activities in the OB. These studies showed also that the odorant-induced activity maps are arranged in a symmetric fashion; one map (lateral map) in the rostro-dorso-lateral hemisphere of the OB and the other (medial map) in the caudo-medial hemisphere. However, except for the fMRI method, these methods map the response only to a single odorant in each animal.
The second group of methods can map responses to many different odorants in the same OB of an animal. These methods include optical imaging of intrinsic signals, optical imaging with Ca2+-sensitive dyes or voltage-sensitive dye, imaging with pHluorin, and fMRI. With the use of these methods, it is possible to determine the range of odorants that activate an individual glomerulus [molecular receptive range (MRR) of the glomerulus] (68) and to examine the spatial representation of the MRR property in the glomerular sheet of the OB. However, again except for the fMRI method, these methods allow us to map only the exposed surface of the OB. Because of this technical difficulty, we have been successful, so far, to image only the dorsal and postero-lateral surfaces of the OB.
The two groups of methods complement each other and have begun to reveal the basic spatial organization of the odor map in the OB. For the extensive knowledge of the spatial representation of individual odorants, see the recent reviews by Leon and Johnson (41, 52), Xu et al. (127), and Shepherd et al. (100). Detailed data of the spatial representation of individual odorants in the OB can be seen at the websites of the Leon laboratory (http://leonlab.bio.uci.edu/index.html), the Restrepo laboratory (http://www.uchsc.edu/rmtsc/restrepo/), and the OdorMapDB (http://senselab.med.yale.edu/senselab/OdorMapDB/default.asp).
Instead of reviewing the whole knowledge about the spatial representation of individual odorants, this review focuses on the question about the spatial arrangement of represented ORs in the OB. We review the emerging ideas regarding the spatial representation of MRR properties in the glomerular sheet of the mammalian OB. Recent studies with the optical imaging method have accumulated information on the MRR map, although the regions that are explored with this method cover only a half of the lateral hemisphere of the OB. The results so far obtained suggest that the mammalian OB represents systematically the characteristic molecular features of odorants, which reflect the systematic spatial representation of ORs. We suggest that the systematic representation of characteristic molecular features may participate in the initial grouping and processing of odorant molecular information at the level of the OB. The information carried by neuronal activities in the molecular-feature map of the OB may be utilized and processed at higher levels in the central olfactory system to form the neuronal representation of the olfactory image of objects and to perceive the odor quality of objects.
II. ODORANT MOLECULAR FEATURES AND PERCEIVED “ODOR”
The relationship between the molecular structure of odorants and their subjectively perceived “odor” has been one of the central questions of the fragrance/flavor chemists and organic chemists. Based on the assessment of odor similarity in relation to the molecular structure, they noted two different but overlapping categories of odorant molecular features (3, 9, 63, 81, 85). One category of odorant molecular feature that has been studied is a polar functional group that contains an oxygen, or a nitrogen, or a sulfur atom. The other category of odorant molecular feature is the molecular profile that is formed mainly by the overall molecular shape.
The polar functional groups are known as “osmophores” (63) and strongly influence the perceived odor quality. As shown in Figure 2A, a homologous series of aliphatic acids share the same functional group (carboxyl group, -COOH) and have similar odor quality (pungent, sour, fatty, and rancid). A homologous series of aliphatic alcohols share the hydroxyl group (-OH) and have similar “fresh and sweet” note. However, odors of aliphatic acids are distinct from those of aliphatic alcohols. Thus the comparison of odors between the two homologous series suggests a major role of polar functional groups for the odor quality perception.
Polar functional groups that have a strong influence on the perceived odor include carboxyl group (-COOH), aldehyde group (-CHO), hydroxyl group (-OH), ketone group (>C=O), ester group (-COO-), amino group (-NH2), isothiocyanate group (-N=C=S), and thiol group (-SH) among others (Fig. 2B).
The molecular profile is equivalent to molecular shape or three-dimensional structure and is known to have a strong influence on perceived odor. The benzene ring (Fig. 2C) is one of the characteristic molecular profiles shared by many benzene-family odorants. The similarity of perceived odor of benzene-family hydrocarbon odorants (e.g., benzene, o-, m-, and p-xylenes, toluene, etc.) suggests the importance of the benzene-ring profile in odor quality perception. Cyclic terpene hydrocarbons (e.g., limonene and pinene) share a molecular profile (Fig. 2C) and have citrusy or woody odor in common. Because hydrocarbons lack any polar functional group, the molecular profile may play a critical role for perception of the odor of hydrocarbon odorants.
In odor molecules, polar functional groups and molecular profiles overlap each other. A specific combination of polar functional groups and molecular profiles may determine the basic framework of molecular features of odorants that are critical for their perceived odor. For example, furan and its derivatives have a specific combination of polar functional groups and molecular profiles, as shown in Figure 3. The furans share the caramellic odor (Fig. 3).
What is the molecular origin of the molecular feature-odor relationship? One plausible explanation is the following. The polar functional groups and molecular profiles are essential for odorants to bind ORs and thus to activate olfactory sensory neurons (4, 43, 54, 103, 133). ORs are seven-transmembrane G protein-coupled receptors (12). Knowledge of the three-dimensional structure of other G protein-coupled receptors such as rhodopsin (77, 79) has provided a template for constructing computational models of ORs to investigate interactions between features of odorants and specific amino acid residues (odorant receptive sites) within transmembrane domains (20, 43, 55, 80, 103–106, 121). The distinct structure of the odorant-receptive site in each OR recognizes a specific combination of polar functional groups, molecular profiles, and other molecular features and appears to determine the range of ligand odorants that are effective in activating the OR (4, 103). Because of the critical role of the molecular features in the odorant-OR binding, they are also named “molecular determinants” or “odotopes” (99).
Recent studies show that individual ORs bind to a range of odorants having similar molecular features (54, 133). The range of odorants that bind to an OR is called the MRR (67, 68) of the OR. We define the characteristic molecular features of an OR as those molecular features shared by the odorants effective in activating the OR. Thus, at the molecular level, an individual OR presumably functions as a molecular-feature detecting unit. Because mice have a repertoire of ∼1,000 OR genes, odorants are thought to be initially probed by ∼1,000 different molecular-feature detecting units.
The logic of molecular-feature detection at the OR level (molecular level) is converted to the cellular and neuronal network levels such that signals of distinct combinations of odorant molecule features are processed in parallel by a large number of molecular-feature detecting cells and modules (6, 66, 98).
At the cellular level, an individual sensory neuron in the olfactory epithelium may function as a molecular-feature detecting unit. Because each olfactory sensory neuron expresses a single OR, it may respond to a range of odorants having similar combinations of molecular features (4). Therefore, the characteristic molecular features of an olfactory sensory neuron can be determined by examining the MRR and molecular features shared by the odorants within the MRR.
At the neuronal circuit level, an individual glomerulus in the OB may function as a molecular-feature detecting unit. Because individual glomeruli represent a single OR, it is possible to determine the MRR and characteristic molecular features of each glomerulus by the same token.
In summary, initial olfactory processing at the levels of ORs, olfactory sensory neurons, and olfactory glomeruli is characteristic in such a way that odorant molecular features are processed in parallel by a large number of molecular-feature detecting units. The information regarding the molecular-features may be utilized by the higher olfactory systems to form the conscious perception of odor quality.
III. TWO-DIMENSIONAL REPRESENTATION OF ODORANT RECEPTOR MAPS IN THE OLFACTORY BULB
The topographic arrangement of columnar modules in the sensory maps of the neocortex provided a basis for understanding the function of sensory systems in the brain (33, 34, 70). The mammalian OB has a cortical structure and contains a large number of glomerular modules (66, 68, 100). The estimated number of glomeruli in each OB is ∼1,800 in the mouse (87), ∼2,400–4,200 in the rat (58, 86), and ∼6,300 in the rabbit (86). An individual glomerular module consists of several thousands converging olfactory axons, ∼10–20 mitral cells, and ∼50–70 tufted cells (2, 86). Individual glomeruli represent a single OR (62, 83, 122). The glomerular sheet of the OB forms OR maps.
The mouse has a repertoire of ∼1,000 types of ORs. Each OB contains ∼1,800 glomeruli (87). Individual ORs are represented by a few (typically two) glomeruli. In rats and mice, each OB contains two mirror-image maps of ORs, the lateral and medial maps (62, 72, 83, 122). For a majority of ORs, each OR is represented by two sites: one site in the lateral map and a corresponding site in the medial map. For a small group of ORs, however, each OR is represented by a single glomerulus in one site (115). These glomeruli are located at the ventral midline of the OB.
To illustrate the spatial arrangement of the glomerular sheet that covers the oval OB, it is convenient to transform the three-dimensional glomerular sheet to a two-dimensional plane map with a defined scale and a coordinate reference system. Such a flattened map is useful for comparing and compiling data obtained by different mapping methods and in different individual animals. The map proved to be useful also for detecting the distortion of odor maps in Semaphorin 3A-deficient mice (118). Several methods have been proposed for the two-dimensional representation of odor maps in the mammalian OB, each having advantages and disadvantages (40, 50, 114, 129) (see also the websites of Restrepo's lab and OdorMapDB). For the readers who are not familiar with the three-dimensional anatomy of the mammalian OB, however, we recommend first to refer to the three-dimensional representation of odor maps, as shown, for example, by Leon's group (41).
We have generated the two-dimensional glomerular maps of rat and mouse OB as follows (Fig. 4) (38, 72, 118). Consecutive frontal sections of the OB are Nissl-stained or immunohistochemically labeled for NCAM to clearly see the glomeruli. In each section, we first define as reference points the dorsomedial edge (arrow heads) of the mitral cell layer. The glomeruli are then traced. The trace of the ring of glomeruli is flattened by opening it along its ventral edge. The traces are then aligned from anterior (top) to posterior (bottom) using the dorsomedial edge as a reference line to make an overall unrolled map of glomeruli.
The constructed map has a dorsal center view. The flattened map indicates symmetrical mirror-image arrangement of glomeruli into the lateral and medial maps (Fig. 4). A flattened map with ventral center view can also be formed by cutting the glomerular trace at the dorsomedial edge and using the ventral edge as a marker for the alignment of the glomerular traces.
IV. INDIVIDUAL GLOMERULI REPRESENT A SPECIFIC COMBINATION OF MOLECULAR FEATURES
A. MRR Property of Individual Glomeruli
To examine whether individual glomeruli in the OB function as molecular-feature detecting units, it is necessary to determine the MRR of individual glomeruli and to compare the molecular structure of odorants that are effective in activating the glomeruli. For this purpose, it is essential to use a large panel of stimulus odorants with a systematic variation of molecular features. It is also critical to select those methods that enable us to examine the response of individual glomeruli in the same OB of an animal to numerous odorants in the panel.
Optical imaging of intrinsic signals is one of the useful methods to examine the MRR property of individual glomeruli (10, 36, 59, 88, 116, 120). It has been shown with this method that some glomeruli can discriminate even between enantiomers (89). Although the intrinsic signal-imaging method has several weak points that include small signal-to-noise ratio and the lack of histological correlation of the signal with glomeruli, the method provides a convenient way of recording the glomerular responses in the same OB to many different odorants (typically ∼70). Using the intrinsic signal-imaging method and a panel of systematically changing odorants, we thus examined the MRR property of a large number of glomeruli that were located in the dorsal, dorso-lateral, and postero-ventrolateral surfaces of the rat OB (36, 116, 117, 120). Comparison of MRR properties among the imaged glomeruli revealed the following.
Individual glomeruli typically respond to a range of odorants that share a specific combination of molecular features.
Each glomerulus in the lateral map appears to be unique in its MRR property.
Glomeruli with similar MRR properties are located in proximity and form molecular-feature clusters.
Results 1 and 2 are consistent with the notion that individual glomeruli represent a single OR and that different glomeruli in each map of the OB may represent different ORs. These results suggest that each glomerulus can be characterized by its MRR property and by its characteristic molecular features. In other words, individual glomeruli represent the glomerulus-specific combination of molecular features. Result 3 will be discussed in detail in section v.
B. MRR Property of Individual Mitral/Tufted Cells
Mitral/tufted cells are principal neurons in the OB and project their axons to the olfactory cortex. Individual mitral/tufted cells extend a single primary dendrite to a single glomerulus and within the glomerulus receive excitatory synaptic inputs from many olfactory axons. Thus the MRR property of an individual mitral/tufted cell may strongly reflect the MRR property of its own glomerulus. Studies with single-unit recording from mitral/tufted cells basically support this hypothesis, although the MRR property of individual mitral/tufted cells is determined not solely by the MRR property of its own glomerulus but is modified by synaptic interactions within the local circuit of the OB (67, 68, 130).
Odorant-response specificity of individual mitral/tufted cells was examined using the systematic panel of odorants and extracellular single-unit recording in the rabbit OB (37, 44, 65) and in the rat OB (73, 117). Although there are many reports on the odorant-response specificity of mitral/tufted cells, it is critical to use a large panel of systematically changing odorants and to have the knowledge of odor maps in the OB (as described in sects. v–vii) to examine the MRR property of recorded mitral/tufted cells. Our results show that individual mitral/tufted cells in the dorsomedial region of the OB respond to a subset of aliphatic acids and aliphatic aldehydes with a similar hydrocarbon chain length. Mitral/tufted cells in the ventromedial region of rabbit OB respond to a range of aliphatic and aromatic compounds with similar molecular structures. Thus individual mitral/tufted cells typically respond to a range of odorants that share a specific combination of molecular features. These studies indicate that individual mitral/tufted cells are tuned to a specific combination of molecular features.
A recent study by Lin et al. (53) supports the idea that individual mitral cells in the mouse OB function as a molecular-feature detector. Urine contains numerous odorous compounds with a variety of molecular structures. Using the gas chromatography single-unit recording method, they demonstrated that individual urine-responsive mitral cells respond selectively to only one (or two) compound among many components of urine.
V. CLUSTERING OF GLOMERULI DETECTING A SIMILAR COMBINATION OF MOLECULAR FEATURES
As stated in section i, this review focuses on the question about the spatial arrangement of MRR representation in the glomerular sheet of the mammalian OB. We think that our recent studies (36, 116, 117, 120) provide the most extensive mapping to date of the relation between molecular structure and its neural representation at the level of individual glomeruli in the mammalian OB. We thus summarize these studies in this section.
Figure 5A shows examples of MRR properties of glomeruli at the dorsal surface of the rat OB. For the detailed MRR properties of individual glomeruli, readers may refer to the data in Takahashi et al. (116) for the dorsal and dorsolateral surfaces and in Igarashi and Mori (36) for the postero-ventro-lateral surface of the OB. Examination of the relationship between the spatial position of glomeruli and their MRRs clearly indicates that glomeruli with similar MRR tend to locate in a close proximity and form clusters (Fig. 5, A and B). The clusters of glomeruli can be seen more clearly by the spatial mapping of characteristic molecular features using the method of superimposing three-dimensional structures of effective odorants, as shown in section vii. We tentatively classify four clusters of glomeruli (clusters A–D) in the dorsal surface, three clusters (clusters E–G) in the dorsolateral surface, and two clusters (clusters H and I ) in the postero-ventrolateral surface of the rat OB (Fig. 6). The MRR property of glomeruli in each cluster will be described in detail below. These descriptions form the basis for the spatial mapping of the MRR property in the glomerular sheet of the OB. The molecular structure of stimulus odorants is shown in Figure 2A and Figure 7.
A. Cluster A
Glomeruli in cluster A are located at the anteromedial part of the dorsal surface (Fig. 6). Because it is relatively easy to image the dorsal surface, the MRR property of glomeruli in cluster A has been most thoroughly studied among the nine clusters.
1. Oxygen-containing functional groups
Glomeruli in cluster A respond to aliphatic acids and aliphatic aldehydes. When examined with a homologous series of aliphatic acids and aldehydes, individual glomeruli respond to subsets of these odorants with a similar carbon chain length. These glomeruli respond also to aliphatic acids and aldehydes with a branched hydrocarbon structure or with one or more double bonds in the carbon chain.
The cluster A glomeruli respond also to a subset of esters with a relatively long carbon chain of acid-part. Glomeruli in the most posterior part of cluster A respond also to diketones. A subset of cluster A glomeruli responds also to benzaldehyde and benzoic acid. Thus the molecular features of odorants that are effective in activating the cluster A glomeruli share either a carboxyl group (-COOH), a diketone group [-(CO)(CO)-], or an ester group (-COO-). These are functional groups that have two oxygen atoms located in proximity. In addition, odorants having a single carbonyl group (>C=O) at the end of the molecule are effective in activating the cluster A glomeruli.
2. Nitrogen-containing functional groups
A large subset of cluster A glomeruli responds also to alkylamines that contain an amino group (-NH2) (117). These results suggest that ORs represented by the subset of cluster A glomeruli respond to both fatty acids with carboxyl group (-COOH) and alkylamines with amino group (-NH2). Because it is difficult to conceive that a single part in the odorant receptive site of OR binds to both carboxyl and amino groups, we speculate that the represented ORs can recognize both carboxyl and amino groups at different parts of their odorant-receptive site. If that is the case, these ORs resemble fish ORs that detect amino acids [H2N-CH(R)-COOH] (15, 22, 24). This raises the possibility that the alkylamine-responsive glomeruli in cluster A represent fishlike class I ORs (see sect. vi) (17, 132).
3. Molecular profiles
Based on the carbon chain length of the effective aliphatic acids, we tentatively classify the cluster A glomeruli into three subclusters: A-1, A-2, and A-3 (Fig. 5B). Glomeruli in A-1 respond selectively to aliphatic acids and aldehydes with a long carbon chain, those in A-2 to acid and aldehydes with a carbon chain of middle size, and those in A-3 selectively to acids and aldehydes with a short carbon chain. However, this is not the case for the carbon chain length of the homologous series of alkylamines (117).
Benzoic acid and benzaldehyde (Fig. 12B) have a molecular profile of a benzene ring and an oxygen-containing functional group (carboxyl group or aldehyde group) in their molecular structure. These odorants activate several glomeruli in the subclusters A-2 and A-3.
B. Cluster B
Cluster B is located in the anterior region of the lateral part of the dorsal surface of the OB (Fig. 6).
1. Oxygen-containing functional groups and molecular profiles
Glomeruli in cluster B respond to aliphatic alcohols with a relatively long carbon chain and to a wide range of aliphatic ketones. The cluster B glomeruli also respond to a subset of esters. Whereas cluster A glomeruli respond to a subset of esters with a relatively long carbon chain of acid part, cluster B glomeruli preferentially respond to a different subset of esters with a relatively long carbon chain of alcohol-part. A majority of cluster B glomeruli respond also to anisole and its derivatives that have a methoxy group (-O-CH3) and a carbon side chain arranged at the para-position of the benzene ring.
Thus the molecular features of odorants that are effective in activating the cluster B glomeruli are a combination of elongated carbon chain structures with a hydroxyl group (-OH), an alkoxyl group (-O-R), or a carbonyl group (>C=O) attached at one side of the molecule.
The glomeruli in the posterior part of the cluster B respond to a wider range of aliphatic ketones and aliphatic alcohols than those in the anteromedial part. We thus tentatively classify cluster B glomeruli into two subclusters: B-1 (the anteromedial glomeruli) and B-2 (the posterolateral glomeruli) (Fig. 5B).
C. Cluster C
Glomeruli in cluster C are located at the central region of the lateral part of the dorsal surface (Fig. 6).
1. Oxygen-containing functional groups and molecular profiles
Glomeruli in cluster C respond to phenol family odorants, molecules having a hydroxyl group attached to the benzene ring. Many of the cluster C glomeruli respond also to phenyl ethers. Thus odorants effective in activating cluster C glomeruli share the molecular features composed of a benzene ring with a hydroxyl group, a methoxy group, or an ethoxy group (-O-CH2-CH3). Thus the characteristic molecular features of those glomeruli include the combination of the benzene-like hydrocarbon profile and the hydroxyl or alkoxyl group.
Based on the detailed MRR properties and the positions of glomeruli, we tentatively divide the cluster C into four subclusters: C-1, C-2, C-3, and C-4 (Fig. 5B). In addition to phenol family odorants, C-1 glomeruli respond to aliphatic alcohols with a relatively short carbon chain (4OH-6OH). C-2 glomeruli invariably respond to salicylaldehyde (Fig. 12B), a phenol derivative with a carbonyl group attached to the ortho-position. Glomeruli in subcluster C-3 respond relatively selectively to phenols and phenyl ethers. Glomeruli in subcluster C-4 respond also to short aliphatic ketones and aliphatic ethers.
D. Cluster D
Cluster D is located at the caudal region of the lateral part of the dorsal surface (Fig. 6).
1. Functional groups and molecular profiles
Glomeruli in cluster D tend to respond to wide structural classes of odorants. Most notable are their responses to a variety of ketones: aliphatic ketones, aliphatic-aromatic ketones, diketones, and cyclic ketones. Compared with glomeruli in cluster B, the cluster D glomeruli tend to respond to a subset of aliphatic ketones with relatively short carbon side chains (Fig. 5B). We tentatively classify the cluster D glomeruli into two subclusters: D-1 and D-2. Medially located subcluster D-1 glomeruli respond to creosol and eugenol, phenol derivatives with a methoxy group attached at the ortho-position and a carbon side chain at the para-position. Laterally located subcluster D-2 glomeruli respond to aldehydes, alcohols, and ethers in addition to ketones.
E. Cluster E
Cluster E glomeruli are located at the dorsal-most part of the lateral surface of the OB (Fig. 6). We could not characterize the odorant-response specificity of cluster E glomeruli because these glomeruli did not systematically respond to any classes of odorants in our panel.
F. Cluster F
Cluster F is located at the rostroventral part of the dorsolateral surface of the OB (Fig. 6).
1. Oxygen-containing functional groups
Glomeruli in cluster F are activated by aliphatic ketones. A subset of glomeruli responds also to secondary alcohols, phenyl ethers, diketones, aliphatic-aromatic ketones, and cyclic ketones. We tentatively divide cluster F glomeruli into two subclusters: anteriorly located F-1 and posteriorly located F-2 (116). Compared with F-1 glomeruli, F-2 glomeruli tend to respond to a wide range of aliphatic ketones and cyclic ketones.
2. Molecular profiles
In addition to odorants with oxygen-containing functional groups, odorants effective in activating cluster F glomeruli include hydrocarbon odorants that do not have any polar functional group. Among the hydrocarbon odorants, the cluster F glomeruli preferentially respond to terpene hydrocarbons. Thus the terpene hydrocarbon structure (Fig. 2C) may be one of the main determinants for the activation of the cluster F glomeruli.
G. Cluster G
Cluster G is located at the ventrocaudal part of the dorsolateral surface of the OB (Fig. 6).
1. Oxygen-containing functional groups
Many glomeruli in cluster G respond to phenyl ethers, diketones, aliphatic ketones with relatively short side chains, aliphatic-aromatic ketones, cyclic ketones, and ethers. The cluster G glomeruli are tentatively parceled into two subclusters: G-1 and G-2. Ventrally located G-1 glomeruli respond to aliphatic alcohols, while dorsally located G-2 does not respond to them.
2. Molecular profiles
Glomeruli in cluster G invariably respond to hydrocarbons as well as odorants with oxygen-containing functional groups. Interestingly, cluster G glomeruli preferentially respond to benzene family hydrocarbon odorants rather than to terpene hydrocarbons. Thus the benzene ring structure (Fig. 2C) could be one of the main determinants for the activation of cluster G glomeruli.
H. Cluster H
Cluster H glomeruli extend from the posterocentral region to the posterodorsal region of the postero-ventrolateral surface of the OB (Fig. 6).
1. Molecular profiles
Individual glomeruli in cluster H respond to one or more of the benzene-family odorants. However, these glomeruli show no or only a weak response to cyclic terpene hydrocarbons. In addition, some of the cluster H glomeruli respond also to open-chain hydrocarbons.
2. Oxygen-containing functional groups
Most of the glomeruli in cluster H respond also to one or more of the benzene derivatives with a polar functional group, such as anisole (which has a methoxy group on the benzene ring), phenol (a hydroxyl group), benzaldehyde (an aldehyde group), or acetophenone (a carbonyl group) (Fig. 7).
Thus the characteristic molecular features of odorants effective in activating glomeruli in cluster H appear to be benzene-related hydrocarbon skeleton.
I. Cluster I
Cluster I is located at the anteroventral region of the postero-ventrolateral surface of the OB (Fig. 6).
1. Molecular profiles
Glomeruli in cluster I respond to at least one of the cyclic terpene hydrocarbons. Many of the cyclic terpene hydrocarbon-responsive glomeruli in cluster I respond also to a subset of benzene-family odorants and open-chain hydrocarbons.
2. Oxygen-containing functional groups
Several glomeruli in the ventral part of cluster I respond also to the cyclic terpene ketones (for example, d-pulegone or d-carvone) or cyclic terpene alcohols (for example, d-menthol).
Thus the characteristic molecular features of odorants in cluster I contain the hydrocarbon skeleton shared by terpene hydrocarbons, terpene ketones, and terpene alcohols.
Figure 8 summarizes the response specificity of clusters A–G in terms of chemical classes of odorants. The glomeruli responsive to a similar combination of chemical classes are located in close proximity and form molecular-feature clusters. The results shown in this section suggest that ORs represented in a same molecular-feature cluster or subcluster may selectively respond to odorants that have similar combination of molecular features. The molecular-feature clusters A–I are located at stereotypical positions in the glomerular sheet.
Although the panel of odorants we used is relatively large and covers several chemical classes, it nevertheless contains only a small subset of odorants in each class. In addition, the panel lacks important classes of odorants that include thiols, sulfides, lactones, pyridines, pirazines, and musks (Fig. 9). By expanding the panel of stimulus odorants, it is therefore necessary to refine the description of MRR properties of individual glomeruli and individual clusters.
J. Comparison With Maps Obtained With Other Methods and in Other Species
With regard to the idea of molecular-feature clusters, it is interesting to compare our maps with the intrinsic signal imaging maps reported by other laboratories (10, 59, 88). Because the latter studies are designed to address questions other than the molecular-feature clusters, it is difficult to compare the reported maps of odorant-induced glomerular activity with our maps of MRR. Nevertheless, the reported maps at the dorsal surface of the rat and mouse OB are consistent with the idea of cluster organization, especially with regard to the position and odorant specificity of cluster A.
Using the 2-DG method, Johnson et al. (40) collected a huge number of measurements of glomerular activities in response to a large panel of odorants with systematic variation of the molecular structure. With the 2-DG method, it is possible to map the glomerular response only to a single odorant in each animal. Therefore, they compiled the maps from many different rats and found that almost the entire glomerular sheet is covered by glomerular “modules” (clusters of nearby glomeruli) and that each module responds to a particular molecular feature. Because of the difference in the measurement method and the definition of glomerular clusters, it is difficult to compare our maps of molecular-feature clusters with the maps of the glomerular “modules,” especially in terms of the shape and boundary of each cluster. However, the overall comparison in terms of odorant specificity and spatial position suggests that the glomerular “modules” in the 2-DG studies are similar in general to the molecular-feature clusters in our study. For example, glomerular “modules” a and c (40) may correspond to clusters A and D in our study, respectively. The assembly of glomerular “modules” d, e, f, h, j, k, and l may roughly correspond to the assembly of clusters H and I. Thus the two different methods presumably shed light on the same basic spatial arrangement of glomeruli from different angles.
The idea of molecular-feature clusters is consistent with the evidence that olfactory sensory neurons expressing homologous odorant receptor genes project their axons to neighboring glomeruli (119). The molecular-feature clusters are found also in the fish OB (25, 47, 76). Furthermore, recent studies indicate that a primitive form of molecular-feature clusters is present also in the larval and adult antenna lobe of insects (16, 49, 51, 90). These results suggest that the clustering of glomeruli that detect similar molecular features is the basic spatial organization of the OB (and antennal lobe) and is conserved across a wide variety of species.
VI. ZONES AND CLUSTERS IN THE OLFACTORY BULB
Most ORs are classified into four zonal subsets (84, 123). Because individual glomeruli in the OB represent a single OR, glomeruli can be classified into zonal subsets (66, 68). With the use of an antibody that recognizes a cell adhesion molecule OCAM, it is possible to immunohistochemically distinguish glomeruli that receive input from zone 1 ORs (OCAM-negative glomeruli) from those that receive input from zones 2–4 ORs (OCAM-positive glomeruli) (131). In the flattened glomerular sheet of the OB (Fig. 4), OCAM-negative glomeruli are largely segregated from OCAM-positive glomeruli (64, 72, 95, 131). OCAM-negative glomeruli are located at the rostro-dorsal part of the OB (zone 1 of the OB), while OCAM-positive glomeruli are distributed at the latero-ventral and postero-medio-ventral parts of the OB (zones 2–4 in the OB, Figs. 1 and 4).
By combining the optical imaging method and the OCAM immunohistochemistry, we examined the spatial relationship between bulbar zones and the molecular-feature clusters in the rat OB (Fig. 10) (36, 116). Most glomeruli in clusters A-D are in the OCAM-negative zone 1, whereas those in clusters F-I are in the OCAM-positive zones 2–4. The boundary between zone 1 and zones 2–4 is located near the lateral margin of clusters B, C, and D.
Figure 10 shows a schematic diagram illustrating the spatial map of the glomerular sheet of the OB. In each cluster, one or two representative examples of the MRR of individual glomeruli are shown by the structural formulas of the odorants (indicated by an arrow and surrounding line). Thick color lines on the structural formulas indicate the common and characteristic molecular features that are shared by effective odorants for individual glomeruli.
Comparison of the MRR property indicates a clear difference in the characteristic molecular features of individual glomeruli between the group of clusters A-D (in zone 1) and that of clusters F-I (in zones 2–4). Odorants effective in activating glomeruli in clusters A-D (in zone 1) have one or more polar functional group(s) in common. Thus the presence of a specific polar functional group tends to be of primary importance to activate glomeruli in zone 1. In contrast, odorants effective in activating glomeruli in clusters F-I (in zones 2–4) have one or more molecular profile(s) in common, suggesting that the overall molecular shape tends to be critical in activating glomeruli in zones 2–4. For example, characteristic molecular features of glomeruli in clusters H and I are the hydrocarbon profiles or relatively nonpolar parts of odorants (Fig. 10).
OR genes are classified into two broad families, class I and class II (27, 132). Class I ORs (fish type, present in both fish and tetrapods) tend to recognize relatively hydrophilic compounds with one or more polar functional group(s), whereas tetrapod-specific class II ORs tend to recognize more hydrophobic compounds (48, 54, 60). Class I ORs are expressed in zone 1 of the olfactory epithelium, while class II ORs are expressed in all the four zones (14, 17, 54). Thus glomeruli in zone 1 represent both class I ORs and class II ORs, while glomeruli in zones 2–4 are most likely to represent class II ORs. Thus the difference in the characteristic molecular features between the groups of clusters A-D (in zone 1) and F-I (in zones 2–4) might correlate in part to the zonal difference in the expression of class I and class II ORs. Further studies are necessary to examine the spatial relationships among the clusters, zones, and OR classes in the glomerular OR maps of the OB.
VII. TOPOGRAPHIC MAP OF CHARACTERISTIC MOLECULAR FEATURES
As described in sections iv, v, and vi, comparison of the MRR property among many glomeruli in the dorsal and lateral surfaces of rat OB reveals the following results.
Individual glomeruli typically respond to a range of odorants that share a specific combination of molecular features.
Each glomerulus (in the lateral map) appears to be unique in its MRR property.
Glomeruli with a similar MRR property locate in proximity and form molecular-feature clusters.
Results 1 and 2 suggest that each glomerulus represents a specific combination of molecular features. They thus raise the possibility that each glomerulus can be characterized by determining the specific combination of molecular features that are shared by the odorants that are effective in activating the glomerulus. We define the term characteristic molecular features of a glomerulus as the specific combination of molecular features that are shared by odorants that are effective in activating the glomerulus (116).
One of the practically feasible methods for deducing the characteristic molecular features of an individual glomerulus is to superimpose the three-dimensional structure of odorants that are effective in activating the glomerulus and to determine the combination of molecular features that the effective odorants share. Figure 11 shows two examples of this method. It should be noted that the superposition of effective odorants is not intended to describe the odorant receptive site structure, but to effectively examine the spatial representation of MRR properties. Mapping the deduced characteristic molecular features on the glomerular sheet will make it easier to see the spatial arrangement of MRR maps.
Odorant molecular features include overall molecular profiles. Because open-chain aliphatic compounds are very flexible and can have many different molecular conformations, it is difficult to deduce the three-dimensional structure of these compounds. In contrast, aromatic compounds have at least one benzene ring or benzene-like ring that assumes a fixed disklike conformation and thus has a semirigid molecular structure. Therefore, aromatic compounds are more suitable for superimposition of the three-dimensional structure than the open-chain aliphatic compounds. We focused our analysis of determining the characteristic molecular features of individual glomeruli first in cluster C and its neighboring clusters because cluster C glomeruli respond to phenol, its derivatives, phenylethers, and methoxybenzenes, all having a rigid benzene ring.
Figure 12 shows a map of the characteristic molecular features in the cluster C and neighboring clusters in the rat OB. To facilitate the analysis, the presumed critical parts of the combination of characteristic molecular features are labeled by color in this figure. For example, blue shadows indicate a combination of molecular features composed of a single methoxy (-O-CH3) group or a single ethoxy group (-O-CH2CH3) attached to a benzene ring profile. Yellow shadows show a combination of molecular features composed of a single hydroxyl group (-OH) attached to a benzene ring. Red surroundings indicate a combination of molecular features composed of a single hydroxyl group and a single alkoxyl group (-O-R) arranged at the ortho-position around a benzene ring. The combination of molecular features composed of a single hydroxyl group and a single carbonyl group (>C=O) arranged at the ortho-position around a benzene ring is shown by pink surroundings.
By the comparison of the characteristic molecular features of individual glomeruli (Fig. 12A), we noted a systematic and gradual change in the represented characteristic molecular features along the axes of the glomerular sensory maps of the OB. This idea originated from the observation that two neighboring clusters or subclusters show a similarity in terms of the characteristic molecular features of constituent glomeruli. For example, as indicated by pink surroundings, subcluster C-2 glomeruli selectively respond to salicyl aldehyde that has a hydroxyl group and a carbonyl group arranged at the ortho-position (Fig. 12B). Thus the characteristic molecular features of the subcluster C-2 glomeruli include two oxygen atoms in proximity. At the region medial to the subcluster C-2, there are many glomeruli in cluster A that respond to aliphatic acids (-COOH), diketones [-(CO)(CO)-], and esters (-COO-). Because these odorants also have two oxygen atoms in proximity, the results suggest that the characteristic molecular features of C-2 glomeruli partially resemble those of neighboring cluster A glomeruli.
In terms of the combination of molecular profile and functional group, salicyl aldehyde closely resembles benzoic acid and benzaldehyde (Fig. 12B). In accordance with this observation, several glomeruli in subclusters A-2 and A-3 respond to benzoic acid and benzaldehyde, and subclusters A-2 and A-3 directly appose the salicyl aldehyde-responsive subcluster C-2. This observation also suggests a similarity of the characteristic molecular features among neighboring subclusters C-2, A-2, and A-3.
The systematic and gradual change in the represented characteristic molecular features is also seen among subclusters within cluster C. As indicated by red surroundings, many glomeruli in subcluster C-2 respond to methoxyphenols (e.g., guaiacol, creosol, eugenol; see Fig. 7) as well as salicyl aldehyde. The area occupied by the methoxyphenol-responsive glomeruli extend antero-laterally from subcluster C-2 to the medial part of subcluster C-3 and even to a part of the subcluster C-1. Thus the neighboring glomeruli in subclusters C-2, C-3, and C-1 share the molecular features.
The similarity of the characteristic molecular features can be seen further along the axis that crosses the boundary between cluster C and anteriorly adjacent cluster B. As indicated by yellow shadows, the common characteristic molecular features in the cluster C glomeruli include a hydroxyl group attached to a benzene ring profile (phenols in Fig. 7). In addition, the subcluster C-1 glomeruli invariably respond to relatively short aliphatic alcohols such as butanol and hexanol. At the region more anterior to subcluster C-1, there are many glomeruli in cluster B that strongly respond to aliphatic alcohols with a relatively long carbon chain. Furthermore, a few C-1 glomeruli respond also to phenyl ethers that have methoxy group and a carbon side chain attached at the para-position (shown by blue shadows). A majority of the cluster B glomeruli also respond to phenyl ethers (Fig. 11E). These results suggest that the C-1 glomeruli and neighboring cluster B glomeruli partially share the characteristic molecular features.
Thus the characteristic molecular features change gradually and systematically in the chains of subclusters A-2, A-3, and C-2; in the medial part of C-3 and C-1; and in cluster B. Because a systematic and gradual change in the characteristic molecular features occurs also along the chains of subclusters C-4, D-1, and D-2, we speculate that the systematic and gradual representation of characteristic molecular features is the property that is not unique to cluster C and neighboring clusters but is shared by many different clusters.
As described above, we observed multidimensional change in the represented characteristic molecular features along the axis across clusters A, C-2, C-3, C-1, and B. In one segment along this axis, we observed a change of a particular molecular feature, but in the other segment we noted a change of different molecular features. In other words, a specific axis in the glomerular sheet does not appear to represent a particular molecular feature. This is consistent with the observation that the change in the carbon chain length in homologous series of aliphatic compounds causes a gradual shift of the positions of activated glomeruli only in a particular segment along an axis in the glomerular sheet. In the different segments of the same axis, carbon chain length does not appear to be systematically represented.
In a wide variety of species, olfactory sensory neurons with related functional specificity map to related spatial positions in the OB or in the antennal lobe (47, 49, 51, 90, 119). The hypothesis that the glomerular sheet of the rat OB topographically represents the characteristic molecular features in a systematic, gradual, and multidimensional fashion is consistent with the above finding. This raises the possibility that the topographic representation of characteristic molecular features might be conserved across different species. In addition, the hypothesis of topographic representation raises the issue whether there are boundaries between different molecular-feature clusters. Clustering of glomeruli with a similar MRR property does not necessarily indicate that different clusters are segregated by specific boundaries.
VIII. MOLECULAR-FEATURE CLUSTERS MAY PARTICIPATE IN ODOR QUALITY PERCEPTION
As described in sections v–vii, glomeruli in the OB form molecular-feature clusters and subclusters. Glomeruli in a same cluster or subcluster tend to represent a similar combination of molecular features. The molecular feature clusters are located at stereotypical positions in the glomerular sheet, and their overall spatial arrangement is conserved across individual animals. Because odorants having a similar combination of molecular features tend to have similar “odor quality” to the human nose, the above observations raise the possibility that the molecular-feature clusters of glomeruli might be part of the representation of basic odor quality. Under this supposition, we summarized the possible relationship between the molecular-feature clusters and basic odor quality. It should be noted, however, that the basic odor quality is described in subjective words and is difficult to define precisely.
A. Cluster A and Odor Quality
Odorants effective in activating the cluster A glomeruli include aliphatic acids, diketones, and aldehydes. These odorants contain “fatty, rancid, sour, and pungent” note in common (Fig. 2). Aliphatic acids with a relatively short carbon chain (C2-C3) tend to have “pungent” and “sour” notes, those with a middle size carbon chain (C4-C5) have “sour” and “rancid” notes, and those with a relatively long carbon chain (C6-C10) have “rancid” and “fatty” notes in common. Because the carbon chain length of aliphatic acids is represented in part by overlapping glomeruli whose position shifts gradually from subcluster A-3 through A-2 to A-1 as their carbon number increases, a correlation is noted among the molecular structure of aliphatic acids, their perceived odor quality, and spatial position of their activated glomeruli within cluster A. The carbon chain length of aliphatic aldehydes is also systematically represented along the posteromedial-to-anterolateral axis within the cluster A. In accordance with this, aliphatic aldehydes with a relatively short carbon chain tend to have “pungent” note, whereas those with a long carbon chain tend to have “fatty” note in common. In addition, diketones that have “pungent” note in common activate many glomeruli within subcluster A-3. These results suggest that cluster A might be part of the representation of the basic odor quality that participates in the “pungent, sour, rancid, fatty” odor quality.
Alkylamines activate a large subset of glomeruli in cluster A and have a fatty, fishy, disagreeable odor. As will be discussed in section ix, this part of cluster A might participate in the representation of fatty, fishy off-flavor (117).
B. Cluster B and Odor Quality
Odorants effective in activating glomeruli in cluster B are aliphatic alcohols with a relatively long carbon chain and a wide range of aliphatic ketones. The odor quality shared by these odorants is “floral and fruity” note. Some aliphatic alcohols and aliphatic ketones also contain “green and grassy” note in common. In addition, many cluster B glomeruli respond to anisole family odorants which share “anisic” note. Thus cluster B may be part of the representation of the basic odor quality that participates in the “floral, fruity, and anisic” and “green and grassy” odor quality.
C. Cluster C and Odor Quality
Glomeruli in cluster C invariably respond to phenol family odorants. Phenol family odorants contain “phenolic and medicinal” note in common, suggesting that cluster C may represent part of the “phenolic and medicinal” odor quality. Further analysis suggests that individual subclusters might differ in the basic odor quality representations. The subcluster C-1 glomeruli respond not only to phenols but also to aliphatic alcohols with a relatively short carbon chain (mainly C4-C6). These short aliphatic alcohols have “chemical” note in common. Subcluster C-2 glomeruli are strongly activated by salicylaldehyde, which have “almondlike” note in addition to “phenolic” note. Many subcluster C-3 glomeruli respond to methoxyphenols, which have “spicy” note as well as “phenolic and medicinal” note. Subcluster C-4 glomeruli are activated by aliphatic ketones with relatively short side chains and ethers in addition to phenols. Short aliphatic ketones and ethers tend to have “ethereal” note in common.
D. Cluster D and Odor Quality
The cluster D glomeruli respond to a relatively wide range of structural classes of odorants in our panel. Odorants effective in activating subcluster D-1 glomeruli are mainly cyclic ketones and methoxyphenols, while those of subcluster D-2 glomeruli are cyclic ketones. Because these odorants tend to have “spicy” and “minty” notes in common, the representation of basic odor quality by the cluster D glomeruli may be part of the representation of “spicy” and “minty” odor quality.
E. Cluster H and Odor Quality
The cluster H glomeruli tend to respond to benzene derivatives that have no, one, or two side chain(s) at ortho- or meta-position around the benzene ring. Because these odorants have gassy and kerosene-like odor in common, the cluster H glomeruli might participate in the representation of “gassy” and “kerosene-like” odor quality.
F. Cluster I and Odor Quality
The cluster I glomeruli tend to respond to cyclic terpene hydrocarbons in addition to the benzene derivatives. Because cyclic terpene hydrocarbons have “citrusy” or “woody” odor in common, these glomeruli in cluster I may participate in the representation of “citrusy” or “woody” odor quality, in addition to the “gassy” or “kerosene-like” odor quality.
The above results suggest a working hypothesis that a specific molecular-feature cluster (or subcluster) might represent a cluster-specific basic odor quality. Because a given odorant typically activates more than one cluster (or subcluster) in the OB, the quality of the odorant may be ascribed to the complex combination of more than one cluster-specific basic odor quality. The results obtained by optical imaging studies fit well to this notion. For example, odor quality of hexanol is composed of “fruity,” “chemical,” and “fatty” notes. Optical imaging data showed that hexanol activated glomeruli within cluster B, cluster AB, and subcluster C-1. 1 Cluster B represents in part “fruity and floral,” cluster AB “fatty” and subcluster C-1 “phenolic, medicinal, and chemical” notes. Thus the map of the molecular-feature clusters in the OB might provide the neuronal basis for the relationship between the odorant molecular structure and the subjectively perceived odor quality.
It should be noted, however, that behavioral studies showed almost no change in the discrimination among odorants (e.g., a homologous series of fatty acids) even after a lesion of a large part of the OB (108–110). These studies underscore the notion that individual odorants are typically coded by two or more clusters and that the activated clusters are arranged both in the lateral and medial maps of the OB. Animals with the OB lesion may be able to perform odorant discrimination based on the information only from the remaining clusters. Because of the difficulty in assessing the odor quality perception in animals, we think that it is not yet known whether or not the animals with the OB lesion perceive the same odor quality. The present hypothesis that different clusters are involved in coding particular odor quality predicts that the animals with the OB lesion show the change in odor quality perception. Further studies are necessary to understand the neuronal systems and the logic responsible for odor quality perception.
Studies of axonal projection of mitral/tufted cells to the olfactory cortex indicate that a given region of the olfactory cortex receives converging inputs from several different clusters of glomeruli (19, 30, 94, 96, 107, 134). Thus olfactory cortex may integrate signals arising from different clusters (74, 126, 135). The knowledge of the molecular-feature clusters and their possible relationship to the basic odor quality may be indispensable for understanding the way of integrating signals from different clusters at the level of the olfactory cortex.
IX. MAPPING THE NATURAL ODOR IN THE MOLECULAR-FEATURE MAPS OF THE OLFACTORY BULB
What is the functional significance of the systematic molecular-feature maps in the OB? It is conceivable that the olfactory nervous system has evolved to detect and discriminate odors that are important for the animal's life such as food odors and predator odors. Therefore, a key step to investigate the functional significance of the molecular feature maps is to examine how the behaviorally significant odors are spatially represented in the sensory maps of the OB. In this section, we review two examples of the mapping of behaviorally significant odors: spoiled food smells and urine odors.
A. Spatial Representation of Spoiled Food Smells in the Odor Maps
One of the important roles of the olfactory system in animal life is to recognize and discriminate between safe, nutritious foods and spoiled foods that contain microorganisms or bacterial toxins. How are spoiled food smells encoded in the odor maps of the OB? Amine odorants are produced by the bacterial decarboxylation of free amino acids in meat and fish (5, 18, 28, 82, 102) and are one of the major causes of the unpleasant fatty, fishy off-flavors of spoiled foods. Amines thus signal the presence of bacterial toxins in the foods, and detection of amine odors leads to the avoidance of spoiled foods. Other major ingredients of the fatty, fishy off-flavors of spoiled foods are aliphatic acids and aliphatic aldehydes that are generated by lipid oxidation of meat and dairy products (5, 8, 8a, 21, 35, 69, 75, 113).
We mapped glomerular responses to amine odorants in the dorsal and dorsolateral surfaces of the rat OB using the intrinsic signal-imaging method and found that the amine-responsive glomeruli were clustered in the anteromedial region (cluster A) of the dorsal OB. Glomeruli in the anteromedial part are known to respond to aliphatic acids and aliphatic aldehydes. Thus we performed detailed analysis of the MRRs of these glomeruli using a systematic panel of stimulus odorants containing a homologous series of aliphatic acids, aliphatic aldehydes, and amines (117). The results show that many glomeruli in the anteromedial region respond not only to amines but also to aliphatic acids and aldehydes. In agreement with the result of imaging study, the single-unit recording study shows that individual mitral/tufted cells in the anteromedial region respond to amines in addition to aliphatic acids and aliphatic aldehydes (117). Since amines, aliphatic acids, and aliphatic aldehydes are major causes of fatty, fishy off-flavors emitted by spoiled foods, the results suggest that the antero-medial amine-responsive glomerular cluster may be involved in detecting the spoiled food smells and participate in the representation of fatty, fishy off-flavor.
In cooking, a variety of spices (e.g., fennel and clove) are commonly used to add flavor and reduce any unpleasant fatty, fishy off-flavor. Optical imaging shows that odors of fennel and clove activate clusters of glomeruli that are located near the alkylamine-responsive cluster (117). Single-unit recording shows that spike discharges of about half of alkylamine-responsive neurons are clearly suppressed by fennel and clove. These results suggest that the masking of spoiled food smells is mediated, in part, by lateral inhibitory connections in the odor maps of the OB.
B. Spatial Representation of Urine Odors in the Mouse Olfactory Bulb
Another important role of the olfactory system is conspecific recognition including olfactory recognition of their mate, parents, and offspring. Mapping studies using immediate early gene expression revealed that urine odor activates clusters of glomeruli in the ventro-lateral region of the mouse OB (92, 93). Recently, Katz and colleagues (53) addressed the issue of how volatile compounds in urine were encoded by individual mitral/tufted cells in the mouse OB. They fractionated the volatile compounds in urine using gas chromatography and recorded the single-unit responses of mitral/tufted cells to the fractionated components. The results show that urine-responsive mitral/tufted cells are clustered in two restricted regions of the OB. These two urine-responsive clusters might reflect the mirror-imaged symmetric maps of ORs in the OB. More importantly, urine-responsive mitral/tufted cells show high odorant selectivity. Most urine-responsive neurons in the cluster selectively respond to a single eluted volatile component among more than 100 different components. Thus individual mitral/tufted cells may function as a molecular-feature detector and, in most cases, do not combine information of more than two different component odorants of urine.
This review summarizes the emerging view of the spatial organization of the odor maps in the mammalian OB. Individual glomeruli and mitral/tufted cells can be viewed as molecular-feature detectors and selectively respond to a specific combination of molecular features. Thus information carried by individual odorants is fractionated according to the ORs that they bind and is handled by a specific combination of molecular-feature detectors in the OB. Even a complex combination of many different odorants can be represented by the activity of a specific combination of glomeruli at the level of the OB.
Mapping studies indicate that glomeruli and mitral/tufted cells that respond to a similar combination of molecular features locate in proximity and form molecular-feature clusters. The molecular-feature clusters are located at stereotypical positions in the OB. The molecular-feature clusters of glomeruli might be part of the representation of basic odor quality. Detailed analyses of MRR of individual glomeruli suggest that the glomerular sheet represents the characteristic molecular features in a systematic, gradual, and multidimensional fashion. Further studies of detailed mapping of glomerular MRR properties are necessary to understand the manner of spatial representation of characteristic molecular features over the entire sheet of glomeruli.
A single object, a cabbage, for example, emits dozens of different odorants. To reproduce cabbage “odor” using the component odorants, it is necessary to combine at least a few component odorants, each having distinct molecular features. This suggests that the central olfactory system has neural mechanisms for combining and integrating information from at least a few molecular feature detectors. Olfactory cortex is supposed to play an important role in integrating signals from different molecular-feature-detecting units (74, 134, 135). A future challenge thus includes the understanding of the functional role of the olfactory cortex for the formation of olfactory image of objects. How does the olfactory cortex integrate signals from different molecular-feature detectors and clusters in the OB? How does the olfactory cortex read the molecular-feature maps of the OB?
Murakami et al. (71) recently noted the behavioral state-dependent gating of olfactory information flows at the level of the olfactory cortex. In urethane-anesthetized rats, neocortical electroencephalogram shows a periodical alternation between two states: a slow-wave state (SWS) and a fast-wave state (FWS). Single-unit recordings from olfactory cortex neurons showed robust spike responses to adequate odorants during FWS, whereas they showed only weak responses during SWS. The state-dependent change in odorant-induced responses was observed in a majority of olfactory cortex neurons, but in only a small percentage of OB neurons. This suggests that olfactory cortex does not always read the odor maps in the OB and underscores the importance of internal activity in determining the mode of information processing in the olfactory cortex and higher centers in the central olfactory system.
Since the discovery of ORs by Buck and Axel in 1991 (12), we have witnessed a rapid progress in the understanding of the olfactory system. However, the rapidly accumulating knowledge is mostly regarding the logic of odorant discrimination at the levels of olfactory sensory neurons and the OB. Olfactory cortex still remains to be a relatively unexplored region in terms of processing of odorant molecular information. We think that future researches seek to understand the specific functional roles of the olfactory cortex and higher olfactory centers for perception, learning and memory of food odors, social odors, and predator odors. The basic knowledge of the molecular-feature maps of the OB would be indispensable for the future research of the central olfactory system.
This work was supported by Grants-in-Aid for Creative Scientific Research (to K. Mori) from the Japanese Society for the Promotion of Science (JSPS). Y. K. Takahashi and K. M. Igarashi were supported by JSPS Research Fellowships for Young Scientists.
We thank the members of the Department of Physiology in the University of Tokyo for discussion and anonymous referees for helpful suggestions.
Address for reprint requests and other correspondence: K. Mori, Dept. of Physiology, Graduate School of Medicine, Univ. of Tokyo, 7–3-1 Hongo, Bunkyo-ku, Tokyo 113–0033, Japan (e-mail:).
- Copyright © 2006 the American Physiological Society