This is a preprint of an article published in

The Journal of Comparative Neurology, 1998, 393:457-471.

© 1998 Wiley-Liss, Inc

Spatial Coding of Odorant Features in the Glomerular Layer of the Rat Olfactory Bulb

BRETT A. JOHNSON*, CYNTHIA C. WOO, AND MICHAEL LEON

Department of Psychobiology, University of California, Irvine, CA 92697-4550

Number of text pages = 34

Number of figures = 10

Abbreviated title: Spatial coding of odorant features

Associate Editor: Jon H. Kaas

Indexing terms: chemical senses; deoxyglucose; esters; mapping; metabolic activity

Send proofs and reprint requests to: Brett A. Johnson

Department of Psychobiology

University of California, Irvine

2205 BioSci II

Irvine, CA 92697-4550

Telephone: (714) 824-7303

Fax: (714) 824-2447

Grant sponsor: NICHD Grant number: HD24236

ABSTRACT

In order to determine whether olfactory receptors recognize molecular features of odorants, rather than entire odorant chemicals, and to determine if such molecular features are represented spatially in the glomerular layer of the olfactory bulb, we used metabolic mapping of [14C]2‑deoxyglucose uptake in rats exposed to equal vapor concentrations of odorants differing systematically in chemical structure. The odorants were ethyl acetate, ethyl butyrate, isoamyl acetate, and isoamyl butyrate. Statistical analysis of anatomically standardized arrays of uptake revealed that each ester produced a characteristic spatial pattern of activity in the glomerular layer. The patterns were similar in different rats exposed to the same odorant, and their complexity increased with increasing odorant carbon number. This finding suggests that the presence of more potentially recognized molecular features is associated with a greater number of activated receptors. Individual regions of the glomerular layer responded specifically to isoamyl esters, and other regions preferred ethyl esters. Regions of similar specificity occurred in lateral and medial aspects of the bulb, the medial representation being more caudal and ventral than the lateral one. This pattern correlates with projections of olfactory sensory neurons expressing the same putative olfactory receptor gene. The patterns overlapped greatly in the posterolateral and posteromedial glomerular layer, a finding one should predict, given the large overlap in chemical structure across the aliphatic esters. Thus, molecular features appear to be encoded spatially in the glomerular layer, and the identity of the odorant may be determined by a subsequent decoding of the combination of molecular features represented in the glomerular layer.

A central goal in the study of olfaction is to determine how volatile chemicals and mixtures of chemicals lead to the sensation and perception of unique odors. A vital clue to the possible mechanism of odor coding has come from the discovery of a superfamily of putative olfactory receptor genes expressed by receptor neurons in the olfactory epithelium (Buck and Axel, 1991). The putative receptor proteins are homologous to G protein-coupled hormone and neurotransmitter receptors (Buck and Axel, 1991), and they support inositol trisphosphate responses in odorant-stimulated transfected insect cell lines (Raming et al., 1993). It has been estimated that there are 500-1,000 distinct putative olfactory receptor genes (Kishimito et al., 1994) and that each olfactory sensory neuron may express only one of these genes (Chess et al., 1994).

What is recognized by an olfactory receptor protein? Most receptors are actually detectors of molecular features present in their natural ligands. Thus, systematic chemical modification of the natural ligand typically reveals elements of chemical structure that are necessary for interaction with the receptor; other features of the molecule can be modified without much effect on the activation of that receptor by the ligand (see, for example, Dean, 1987). Neurotransmitter receptors are typically only exposed to a single potential ligand, so that the feature detection principle is mainly of pharmacological interest. However, olfactory receptors are located in an environment where they are exposed to a far greater number of potential ligands, and it has been widely proposed that the olfactory receptors function in odor coding by detecting which molecular features are present in an odorant or odorant mixture; the particular combination of receptors activated by the features of an odorant would yield its distinctive odor (Buck and Axel, 1991; Shepherd, 1991; Imamura et al., 1992; Mori et al., 1992; Katoh et al., 1993; Kauer and Cinelli, 1993; Shepherd, 1994; Ressler et al., 1994; Vassar et al., 1994; Sato et al., 1994; Buck, 1996). Therefore, each pure odorant molecule could be considered to contain a combination of molecular features, and any given feature could be shared by multiple, molecularly similar odorants. In addition, any individual odorant could activate several olfactory receptor proteins, each recognizing a single feature of the molecule.

An example of four molecules that have both common and unique elements of chemical structure is shown in Figure 1. These aliphatic esters all have in common both an ester bond and units of hydrocarbon structure present in the smallest molecule, ethyl acetate. Any olfactory receptor protein that recognizes this portion of the molecule that is common to this chemical family and that is unaffected by steric hindrance from the additional structure present in the other three molecules might respond to all four esters. The two isoamyl esters in Figure 1 share an additional, branched hydrocarbon structure on the "O-side" of the ester bond that is lacking from the ethyl esters, and it is possible that there are receptors that recognize this additional structure as a feature while other receptors may recognize the ethyl group specifically. Similarly, the two butyrates share an additional two carbons on the "C-side" of the ester bond, and this may be a feature recognized specifically by another subclass of olfactory receptor proteins.

How would a combinatorial code of odor quality be relayed from the olfactory receptor proteins to the central nervous system? Axons from olfactory receptor neurons bearing the same receptor sequence converge into a very limited number of glomeruli (two to five) in the main olfactory bulb (Ressler et al., 1994; Vassar et al., 1994; Mombaerts et al., 1996); these axons invariably project to both lateral and medial glomeruli, with the latter being located more caudally and ventrally than the lateral projection (Ressler et al., 1994; Vassar et al., 1994; Sullivan and Dyer, 1996; Mombaerts et al., 1996). These glomeruli appear consistent in location across animals (Ressler et al., 1994; Vassar et al., 1994; Mombaerts et al., 1996). The glomerular convergence of projections from olfactory receptor neurons expressing the same olfactory receptor gene suggests that an early step in central odor coding might involve a spatial element, as it does in other sensory systems (Somjen, 1972).

Given this information, the pattern of glomerular activation should be determined by which olfactory receptor proteins recognize a specific molecular feature of an odorant molecule(s). Thus, each pure odorant should generate a characteristic spatial pattern of glomerular activation. Any two odorants sharing a molecular feature recognized by a single receptor should activate the same glomeruli. One may also predict that simple, small odorants should generate simpler patterns of glomerular activation than complex, large odorants of the same chemical family, because complex odorants should have additional features that could activate additional receptor proteins. Finally, if the putative olfactory receptor proteins are true feature detectors, then the glomerular activity pattern should be represented both laterally and medially, given that olfactory receptor neurons expressing the same putative olfactory receptor gene project to both lateral and medial glomeruli.

There is ample evidence from studies of 2‑deoxyglucose (2-DG) uptake (Stewart et al., 1977; Coopersmith et al., 1986; Royet et al., 1987; Bell et al., 1987) and c-fos expression (Guthrie et al., 1993) that distinct odorants with greatly different chemical structures generate different patterns of glomerular activation. However, the other predictions of the proposed combinatorial mechanism for odorant coding require a systematic investigation of odorants differing in more well-defined elements of chemical structure.

In order to test these predictions, we have analyzed [14C]2-DG uptake across most of the glomerular layer throughout the bulb to generate anatomically standardized maps of activity in rats exposed to equal vapor-phase concentrations of aliphatic esters differing discretely in chemical structure on either side of the ester bond (Fig. 1). These odorants should give the olfactory system the opportunity to display specific recognition of distinct molecular features. Our results provide strong support for the proposed combinatorial mechanism for encoding of odorant features and suggest that there may be a map of molecular features across the bulbar surface that resembles the tonotopic, somatotopic, and visual coding present in other sensory systems (Somjen, 1972).

MATERIALS AND METHODS

Animals

A total of 35 Wistar rats (postnatal day 20 or 21) from seven litters were used in this study. Each litter was culled to six males and two females on the day after birth. On the test day, the entire litter and dam were transferred to a clean cage for at least 1 hour prior to the first odorant exposure in order to reduce carryover of odors from soiled home cages into the test apparatus. Each litter contributed five male rats to the study, one for each odorant condition. The first rat was exposed to ultra-zero grade air as an unstimulated control. The order of exposure to the four esters was varied across the different litters. All procedures involving rats were approved by the UC Irvine animal care committee.

Odor exposures

Rats were given a subcutaneous injection of [14C]2-DG (Sigma, St. Louis, MO; 0.2 mCi/kg) and immediately were placed into a clean, 1-liter mason jar for 45 minutes. The odorant entered and the exhaust exited through the lid during that period. Following odor presentation, rats were immediately decapitated, and their brains then were removed and frozen in isopentane at ‑45°C.

Odorants were diluted by using a flow dilution olfactometer. Ultra-zero grade air from a cylinder fitted with a brass/Teflon regulator was bubbled at a total flow rate of 150 ml/min through a 250-ml gas washing bottle containing 200 ml of odorant (purity > 98%). The outlet from the gas washing bottle was split between a vent and a Gilmont size 1 flowmeter fitted with a flow regulator. This regulator was set so that the appropriate flow rate of saturated odorant was mixed with another stream of ultra-zero grade air to give a final flow rate of 2 liters/min entering the exposure chamber. The apparatus was equilibrated with each odorant for at least 15 minutes prior to the exposure. All tubing and connections leading to the exposure chamber were made of Teflon, Kynar, glass, or brass to minimize reactivity with the odorants. Clean exposure chambers and tubing were used for each odorant to minimize carryover from the previous odor.

We considered it to be crucial that the rats were exposed to the same concentrations of different odorant molecules in order to be able to compare responses across different odorants. To accomplish this goal, we equalized odorants by vapor phase concentrations rather than by dilution of saturated odorant vapor. The vapor pressures of the four esters used in this study were estimated from their boiling points using the equation of Hass and Newton (1975) to give values at 22°C and 1 atm. The resultant vapor pressures varied greatly from one another (in mm Hg, ethyl acetate: 94, ethyl butyrate: 14, isoamyl acetate: 5.4, isoamyl butyrate: 0.81). Therefore, each odorant was diluted to give a final partial pressure of 0.57 mm Hg, which corresponds to a vapor phase concentration of 75 parts per million. The dilutions were 1/1650 for ethyl acetate, 1/250 for ethyl butyrate, 1/94 for isoamyl acetate, and 1/14 for isoamyl butyrate.

Analysis of 2-DG uptake

The following procedures are refinements of those used in our previous study of early olfactory learning (Johnson and Leon, 1996). Olfactory bulbs were sectioned coronally at 20‑µm thickness with a cryostat. The first section was collected on a 22 x 22-mm coverglass and immediately placed on a slide warmer at 60°C for rapid dehydration. The second was collected on a gelatin-subbed microscope slide and thaw-mounted. The third section was discarded. This collection procedure was repeated until the entire olfactory bulb had been sectioned. Coverglasses were taped to cardboard and juxtaposed to Kodak SB5 autoradiography film for 10 days along with 14C-standards (ARC-146A; American Radiolabeled Chemicals, Inc., St. Louis, MO) previously calibrated to tissue equivalents (nCi/mg) of isotope. Sections on slides were stained by using cresyl violet in order to locate anatomical hallmarks along the anterior-posterior dimension, to evaluate the angle of section, and to match up regions of focal 2‑DG uptake with individual glomeruli. The first section containing a mitral cell layer was identified, and this section was numbered as section #1 on the adjacent autoradiographic image. The last section containing a medial mitral cell layer also was identified and labeled on the autoradiogram. Analyses of glomerular layer 2‑DG uptake were confined between these hallmarks, inclusively. The first section containing a subependymal zone and the first section containing an accessory olfactory bulb were also recorded. Each hallmark was separately determined for right and left bulbs of each brain.

Films were coded prior to analysis. The code was such that the litter number remained apparent, but that the identity of the odorant was hidden. Autoradiograph sections were visualized in pseudocolor and analyzed by using MCID/M1 software and a Sony, model X6-77 CCD camera (Imaging Research Inc., St. Catharines, Ontario, Canada). Measurements were obtained in units of nCi/mg by calibration to the 14C-standards. A protractor was centered in the core of each bulb section, and samples of 9 x 9 pixels (12 x 12 µm) were taken in the glomerular layer at 48 fixed angle increments chosen to give equidistant samples in the largest sections. Starting with the dorsal-most sample and progressing laterally, the angles were (in degrees): 0, 6, 12, 19, 26, 33, 40, 47, 54, 62, 70, 80, 90, 100, 110, 118, 126, 133, 140, 147, 154, 161, 168, 174, 180, etc., the angle spacing being symmetrical for the lateral and medial glomerular layers. We sampled uptake at each of these positions, whether this uptake was low, moderate, or high. Therefore, our sampling procedure was not restricted to areas showing only the highest levels of uptake. Sample collection continued in this manner for each bulb section until the lateral glomerular layer disappeared, being replaced by the anterior olfactory nucleus. Then, the center of the protractor was placed at a fixed distance from the ventral and medial extents of the bulb being analyzed in order to continue measurements from the medial glomerular layer, which continued a considerable distance more caudally than the lateral glomerular layer. The point for centering the protractor was chosen from the last section that was judged to have a lateral glomerular layer. For sample locations judged not to have a glomerular layer due to this phenomenon, tears, or tissue folds, measurements were taken from unexposed areas of the film adjacent to the section. These measurements were consistently < 50 nCi/mg and less than any glomerular layer measurement. Data for each section were imported into an Excel spreadsheet to give an array (48 measurements x as many sections as existed between the anterior and posterior limits of the analysis). Both bulbs of each brain were analyzed in this matter to yield 70 such arrays.

Transformation of the arrays

The arrays of raw data from individual rats were transformed prior to analysis. First, the measurements described above that did not correspond to the glomerular layer were deleted. When these were due to missing values in individual sections due to tears or folds in the tissue, they were replaced with the average of values taken at the same angle in the preceding section and in the subsequent section. At this point, the arrays of uptake were converted to values relative to measurements taken within the bulbar core. We then printed color-coded contour charts to visualize differences in glomerular activity across the bulb, as we have reported previously (see Johnson and Leon, 1996 for additional details regarding this method). The charts represent the "analysis space" used in these studies, and they differ both from actual bulb space and from the rolled-out maps used by others in describing spatial distributions of activity in the bulb (Stewart et al., 1979). Although the actual bulb increases in perimeter from anterior to posterior, these charts are consistent in size along this dimension. Also, the actual perimeter of the largest coronal section is larger than the anterior-posterior length of the bulb, whereas these charts represent the anterior-posterior length as being greater. The size of glomeruli differ around the lamina. Our sampling procedure was not intended to select for either larger or smaller glomeruli, and we can not assess the relative contribution of larger compared to smaller glomeruli either to the patterns of activity or to odor coding.

For formal analysis, additional transformations were performed. In order to correct for departures from a true coronal angle of section, the Nissl-stained sections were assessed where the lateral mitral cell layer gave way to the anterior olfactory nucleus. Since the absence of the entire lateral mitral cell layer occurs abruptly (personal observation), we judged that the disappearance of the lateral mitral cell layer more ventrally or dorsally indicated an angle of sectioning that was different from a true coronal plane. Accordingly, the angles at which the lateral mitral cell layer disappeared were recorded as a function of section number for each bulb using the protractor. If there was evidence of non-true coronal sectioning, the affected measure numbers were moved back in the section arrays so that measurements corresponding to the location of the last lateral mitral cell layer would be contained in a single column (section) of the array.

Arrays then were transformed to equalize the number of sections between anterior-posterior hallmarks. This standardization was accomplished by inserting mock sections evenly spaced between the hallmarks. These mock sections were given values that were the averages of those at the same angles in the immediately preceding and immediately subsequent sections. After expansion, section #15 was the first section in each bulb containing the subependymal zone, section #60 was the first to contain the accessory olfactory bulb, and section #110 was the final section of the array (Fig. 2). These anatomically standardized arrays of nCi/g from the two bulbs of a given animal were then averaged to give a single array for the animal. To correct for different amounts of 2-DG injected into different rats, the arrays were subjected to a z-score transformation where the average uptake across the array was subtracted from the value of each cell of that array, and these differences were divided by the standard deviation of the values across the array. These standardized arrays were used for statistical analyses of pattern dissimilarities and for generating arrays which were averaged across animals that were exposed to the same odorant. The use of z-score transformations is well accepted for the analysis of patterns of activity independently of the absolute amplitudes of activity (Royet et al., 1987; Kent and Mozell, 1992).

Correspondence between fields of activity and individual glomeruli

The number of individual glomeruli underlying high uptake foci was analyzed in five fields, the boundaries of which were delineated from the high 2-DG uptake regions seen on contour charts of z-scores averaged across rats exposed to the same odorant. In order to obtain actual section numbers, the focal high uptake region for each field was identified on an anatomically standardized contour chart of glomerular/core uptake for the right bulb of each animal. This region then was relocated on the corresponding uncorrected contour chart of relative uptake. Two sections were selected from within the darkest focal region in the field, and we recorded the section numbers and the angles at which the focal uptake was encountered. The section numbers for the two corresponding adjacent Nissl-stained sections also were recorded.

Each Nissl-stained section was digitized (Image software, NIH) and contrast-enhanced such that the boundaries of glomeruli were clearly visible. The outline of the section and the outlines of all glomeruli within the region of interest were carefully traced on an acetate sheet. The corresponding autoradiograph section then was digitized at the same magnification and pseudocolor-enhanced to visualize easily the high-uptake foci. The tracing of the Nissl-stained section then was overlaid on the autoradiograph image to align the outlines of the two sections, which is similar to the method used previously (Woo and Leon, 1991). The high uptake focus within the field of interest was traced on the acetate sheet, and the individual glomeruli underlying the focus were counted and recorded. In most cases, it was possible to determine a discrete number of glomeruli, but in some cases, it was difficult to discriminate between one and two glomeruli. In these cases, a range of glomerular numbers was recorded. For statistical purposes, the lower and higher ends of any ranges were averaged for each section. The glomerular number for a given section was then averaged with the number for the other section from that bulb. These averages were compared across high-uptake 2-DG fields by using a one-way ANOVA where the statistical unit was a single field in a single animal.

RESULTS

Distribution of uptake in individual bulbs

Our method for mapping 2‑DG uptake across the glomerular layer resulted in standardized arrays that could be visualized as contour charts. Figure 3 shows examples of these charts for individual bulbs of ten rats from two litters. The pattern seen in any individual bulb was very closely matched by the pattern in the other bulb from the same rat. This bilateral symmetry of response has been described and statistically documented by others (Royet et al., 1987), and was not further analyzed in the current study.

Our first analysis was a qualitative one. At the time of analysis, we were blind to the odorant that had been presented to each rat, but we were aware of which rats came from the same litter. First, we determined whether the glomerular activation patterns increased in complexity in proportion to the number of potentially recognizable features contained within the pure odorants. Therefore, we ranked the pure odorants in order of increasing complexity based on the increasing number of carbon atoms across the odorants (air < ethyl acetate < ethyl butyrate < isoamyl acetate < isoamyl butyrate). Then, for each litter, we made a qualitative determination of which contour maps were more complex than others, and we ranked the contour maps on that basis. Second, we adjusted our rankings in order to match rats in the different litters that displayed similar contour maps. By this two-step procedure, we were able to identify correctly the odorant to which 33 of the 35 rats were exposed. The single error of transposition within a litter involved a rat exposed to isoamyl butyrate that had a low glomerular layer/core ratio and a rat exposed to ethyl butyrate that had an unusually high glomerular layer/core ratio. Therefore, the complexity of the contour maps appeared to match well the inferred complexity of the odorant molecules, and rats exposed to the same odorant appeared to have similar patterns of glomerular layer activity.

To continue the qualitative analysis, the glomerular-layer uptake in air-exposed rats was typically much lower than in odorant-exposed rats from the same litter (Fig. 3). The pattern evoked by ethyl acetate was remarkably simple in individual bulbs, being characterized by sharply defined foci of activity within the posterior portions of the midlateral and midmedial glomerular layer. Typically, there were two to three such foci in each of these regions (pink or red in Fig. 3). At the other extreme, isoamyl butyrate typically evoked a very large number of foci distributed over much of the glomerular layer. About a third to a half of the glomerular layer in individual bulbs seemed to respond to some degree to this odorant when compared to the air-exposed rats from the same litter (Fig. 3). Ethyl butyrate and isoamyl acetate both evoked an intermediate number of high-uptake regions. However, there were two regions of uptake (one midlateral and one midmedial) seen in the isoamyl acetate-exposed rats (arrows in Fig. 3) that rarely had any equivalent in the ethyl butyrate-exposed rats. Uptake in these two regions also was observed in the isoamyl butyrate-exposed rats (arrows in Fig. 3). Despite the overall qualitative similarity in patterns across rats exposed to the same odorant, the patterns were never exactly the same between any two rats.

Quantitative analyses of pattern differences between individual rats

The contour maps representing glomerular uptake as a ratio of bulbar core uptake were necessary to demonstrate the difference in magnitude of the response in odorant-stimulated versus air-exposed rats, but we discovered that this expression of relative uptake presented several problems. Entire litters yielded higher values of this ratio than did other litters, and the overall uptake in individual rats also appeared to be low or high when compared either to littermates or to rats in other litters exposed to the same odorant (see, for example, the low overall uptake in the isoamyl acetate-evoked pattern in Litter 4, Fig. 3). Therefore, for the remainder of our analyses, we averaged the arrays of uptake (nCi/g) for the left and right bulbs of a given animal and converted these values to z-scores relative to the average and standard deviation of uptake across the array. This transformation has been used by many others studying patterns of activity (Royet et al., 1987; Kent and Mozell, 1992), and it resulted in more uniform values both within and across litters in the current study.

In order to obtain a representation of pattern differences and similarities that would lend itself to a statistical evaluation of our data, we calculated indices of pattern dissimilarity similar to those used by Kent and Mozell (1992). In this analysis, the data underlying the contour map for each individual rat was compared to that of each other rat in the study. Each array was subtracted from each other array and the values in the resulting difference arrays then were converted to absolute values. The numbers in a given absolute value array were averaged to yield a single positive value that would be low for pairs of rats exhibiting similar patterns and would increase with increasing dissimilarity of the two patterns.

Because there were 35 rats in the study, 595 pairs of rats were compared to assess their relative similarity, with a distribution of the 595 values that was roughly Gaussian (Fig. 4). If the patterns of glomerular-layer 2-DG uptake are both odorant-specific and conserved across different rats, then the values of pattern dissimilarity resulting from comparisons of pairs of rats exposed to the same odorant (n = 105) should be lower than those derived from comparisons of pairs of rats exposed to different odorants (n = 490). As shown in the histograms of Figure 4, this prediction was clearly fulfilled. A Mann-Whitney U-test comparing same-odorant pairs and different-odorant pairs revealed that the difference between these groups was statistically significant (U = 12240, p < 0.0001).

The distribution of pattern dissimilarity values from same-odorant comparisons possessed at least two modes, one at 0.64 and one at 0.79, the latter of which was similar to the mode for the different odorant comparisons (Fig. 4). This higher mode was entirely populated by comparisons of air-exposed rats (18 out of 21 pairs gave values > 0.70) and isoamyl butyrate-exposed rats (14 of 21 pairs with values > 0.70). The values for these rats typically were even greater when they were compared with rats exposed to different odorants.

To determine whether the pattern generated by each individual odorant was different from that evoked by each other individual odorant, we performed a similar analysis. Pattern dissimilarities from same-odorant paired comparisons for odorant 1 (n = 21) were combined with same-odorant paired comparisons for odorant 2 (n = 21). This set of values (n = 42) then was compared to the set of pattern dissimilarities arising when rats exposed to odorant 1 were paired with rats exposed to odorant 2 (n = 49). This was done for every possible pair of odorants in the study. As shown in Figure 5, the different-odorant values were always higher than the same-odorant values. This finding indicates that for every pair of odorants, the patterns visualized in the contour maps were more similar between rats exposed to the same odorant than they were between rats exposed to different odorants. In each case, the difference was significant (Mann-Whitney U-tests: air vs ethyl acetate, U = 459, p < 0.0001; air vs ethyl butyrate, U = 232, p < 0.0001; air vs isoamyl acetate, U = 134, p < 0.0001; air vs isoamyl butyrate, U = 173, p < 0.0001; ethyl acetate vs ethyl butyrate, U = 619, p = 0.0011; ethyl acetate vs isoamyl acetate, U = 134, p < 0.0001; ethyl acetate vs isoamyl butyrate, U = 347, p < 0.0001; ethyl butyrate vs isoamyl acetate, U = 513, p < 0.0001; ethyl butyrate vs isoamyl butyrate, U = 500, p < 0.0001; isoamyl acetate vs isoamyl butyrate, U = 582, p = 0.0004). Therefore, each of the four odorants evoked a clearly characteristic pattern of glomerular activity.

Maps of uptake averaged across animals exposed to the same odorant

To determine if there were specific regions of the bulb that responded to different odorant features, the z-score arrays for all rats exposed to the same odorant were averaged, and these arrays were visualized and compared to rats exposed to different odors using color-coded contour charts (Fig. 6). A number of apparently specific fields emerged in these averaged maps. There were two fields that appeared to be evoked more by the two ethyl esters than by the isoamyl esters; one such field was located in the dorsomedial glomerular layer, and the other was found in the anterior, dorsal glomerular layer (Fig. 6, black arrows). There were two fields that appeared to be evoked by the two isoamyl esters but not by the two ethyl esters; one of these fields was midlateral, and one was midmedial, (Fig. 6, straight, white arrows). Both the ethyl-specific field and the isoamyl-specific field represent evidence for spatial coding of molecular features in the glomerular layer. There also were two fields, one ventrolateral and one ventromedial, that appeared to be specific to isoamyl butyrate-exposed rats (Fig. 6, curved, white arrows).

Activity in a large expanse of the posterior, midlateral glomerular layer overlapped appreciably in the rats exposed to the different esters. However, there appeared to be subtle differences between the locations of activity within this large field. Ethyl acetate stimulated posterior portions of the field to a similar extent as anterior portions, whereas the other odorants stimulated the anterior portions more than the posterior portions (Fig. 6). A similar observation applied to uptake within the posterior, midmedial glomerular layer (Fig. 6).

Indices of pattern dissimilarity were calculated for pairs of these averaged arrays in the same manner as for the individual bulbs. As shown in Figure 7, ethyl butyrate and isoamyl acetate were found to yield the most similar patterns (index of 0.33), probably due to the similarity in the pattern of uptake over the posterior portions of both the lateral and the medial glomerular layer. Ethyl acetate and isoamyl butyrate yielded the most distinct patterns of the odorant-exposed animals (index of 0.47). All comparisons using the air-exposed animals gave values > 0.5. The values of pattern dissimilarity were correlated with the differences in the number of carbons present in the odorants (r = 0.86, F[1,4] = 11.5, p < 0.05; Fig. 7).

Statistical comparisons within individual fields

Although fields of uptake in the averaged maps appeared to be specific to odorant features and sets of odorants, it remained possible that the uptake in these regions was dominated by values from individual rats and was not representative of the odorant-evoked uptake across different rats. To determine how the individual rats varied within each of these fields, comparisons were conducted using single-factor ANOVAs. Because the main interest of the study was to locate specific regions differing between odorants, rather than to document changes between odorant- and air-exposed rats, the air-exposed rats were excluded from this analysis. Seven fields were defined on the basis of the average maps as shown in Figure 8, left.

Within Fields 1 and 2, we determined the anterior-posterior centers of the z-score values (the section where the sum in the anterior direction was equivalent to the sum in the posterior direction). To do this, values at different angles within the field in a given section were summed, and a running sum then was generated along the sections within the field. We divided the value of the running sum at each section by the grand sum, such that the most anterior section of the field yielded the lowest value and the most posterior section of the field yielded a value of 1. We then recorded the section numbers for the two sections most closely bracketing a value of 0.5 (the center of activity). The actual location of the center of activity was calculated by interpolating between these two sections. These real number values then were subjected to a single factor ANOVA across odorants. The center of activity was significantly different across odorants for both Field 1 (F[3,24] = 11.31, p < 0.0001) and Field 2 (F = 7.81, p < 0.001). Ethyl acetate-exposed rats possessed uptake centered at a relatively more posterior position within both of these fields than did rats exposed to the other odorants (Fig. 8).

In order to assess whether the maximal activity within Fields 3 through 7 differed across rats exposed to the different odorants, we determined for each bulb the maximal z-score value within the boundaries shown in Figure 8, left. Maxima should represent the highest measurement (e.g., the most active glomerulus) within the field for a given bulb. The values of the two bulbs were averaged for a given rat, and these values were compared in ANOVAs. All of these comparisons yielded significant results (Field 3: F[3,24] = 12.85, p < 0.0001; Field 4: F = 11.93, p < 0.0001; Field 5: F = 15.95, p < 0.00001; Field 6: F = 19.78, p < 0.00001; Field 7: F = 10.50, p < 0.0005). In each case where an ANOVA yielded significant results, nonparametric Kruskall-Wallis tests also resulted in p < 0.001. Therefore, individual regions of the glomerular layer do indeed respond similarly in rats exposed to the same odorant, but differently in rats exposed to odorants possessing different molecular features.

Within Field 3, the uptake was greatest in ethyl acetate-exposed rats (Fig. 8). Fields 4 (lateral) and 5 (medial) were remarkably similar in their specificities, with the uptake in rats exposed to isoamyl acetate or isoamyl butyrate exceeding the uptake in rats exposed to either ethyl acetate or ethyl butyrate (Fig.8). Fields 6 (lateral) and 7 (medial) also were remarkably similar in specificity, with isoamyl butyrate evoking the greatest response, followed by isoamyl acetate, ethyl butyrate, and ethyl acetate (Fig. 8).

Relationships between fields of uptake and individual glomeruli

In order to estimate the number of glomeruli contributing to the fields of response seen in the contour charts, foci of highest 2‑DG uptake in individual bulbs were located within regions corresponding to Fields 4 and 5 in both isoamyl acetate- and isoamyl butyrate-exposed rats, to Fields 6 and 7 in isoamyl butyrate-stimulated rats, and to Field 3 in both ethyl acetate- and ethyl butyrate-exposed rats. Nissl-stained sections adjacent to the autoradiography sections then were investigated to visualize individual glomeruli. Figure 9 shows representative examples of adjacent sections for each of these odorant and field combinations.

The mean number of glomeruli associated with each of the analyzed fields is illustrated in Figure 10. A single-factor ANOVA revealed that the numbers of glomeruli across these 8 odorant/field combinations were significantly different (F[7,45] = 8.0, p < 0.0001). Fields that showed similar odorant specificities (Fields 4 and 5, and Fields 6 and 7) also were associated with similar numbers of glomeruli for any given odorant exposure. Isoamyl butyrate-evoked foci of 2‑DG uptake within Fields 4 and 5 involved fewer glomeruli than did isoamyl acetate-evoked foci within the same fields. The isoamyl butyrate foci within Fields 6 and 7 were associated with a very low number of glomeruli in any given coronal section. In some cases (1/4 to 1/2 of the analyzed foci), these foci aligned with single glomeruli. Ethyl acetate-evoked foci of 2‑DG uptake within Field 3 also were associated with very few glomeruli. In ethyl butyrate-stimulated rats, a greater number of glomeruli were associated with focal responses in this field.

In individual rats, we also evaluated the relative dorsal-ventral location of high-uptake foci within paired fields (Fields 4 and 5 in isoamyl acetate-exposed and isoamyl butyrate-exposed rats, and Fields 6 and 7 in isoamyl butyrate-exposed rats). In most cases, the medial field of the pair clearly was located more ventrally than the lateral field of the pair. This is evident for Fields 4 and 5 in the isoamyl acetate-exposed rat shown in Figure 9, as well as for Fields 6 and 7 in the isoamyl butyrate-exposed rat (Fig. 9). Although the foci within the paired fields occasionally overlapped in their dorsal-ventral extents, the lateral foci were never more ventral than the medial foci.

DISCUSSION

Different odorants evoke distinct patterns of glomerular activity

Previous studies of glomerular activity using 2‑DG or c-fos in situ hybridization have found that odorants with greatly different functional groups, e.g., isoamyl acetate and camphor (Stewart et al., 1979), cyclohexanone and peppermint extract (Coopersmith et al., 1986), propionic acid and limonene (Bell et al., 1987), amyl acetate and isovaleric acid (Royet et al., 1987), and isoamyl acetate and peppermint extract (Guthrie et al., 1993), evoke different patterns of activity in the glomerular layer. The present results indicate that different spatial patterns of glomerular activation also can be seen for members of the same chemical class that differ only slightly in composition. The patterns were distinct across odorants and consistent across animals, so that rats exposed to the same odorant could be identified subjectively. Calculations of objective indices of pattern dissimilarity confirmed the distinctiveness of odorant-evoked activity. These results suggest that the patterns of activity evoked in the glomerular layer might contain sufficient information to allow for the decoding and discrimination of unique odors by the olfactory bulb.

Units of the patterns correlate with chemical features

By averaging patterns from rats exposed to the same odorant, fields of response were identified that correlated with particular odorants or odorant features. Detailed analyses of the uptake within these regions verified that the variance within rats exposed to a given odorant was less than that between rats exposed to different odorants. Fields were found that responded more to isoamyl esters than to ethyl esters (Fields 4 and 5), and that were activated more by isoamyl butyrate (Fields 6 and 7) or ethyl acetate (Field 3, posterior portions of Fields 1 and 2).

The four odorants chosen for this study differ from one another in discrete units on either side of the ester bond. It therefore is possible to hypothesize the chemical features responsible for each field of activation. For example, the isoamyl esters differ from the ethyl esters by the additional presence of three carbons in a branched structure on the O-side of the ester bond (stippled box in Fig. 1). The response of Fields 4 and 5 therefore may indicate the recognition of a portion of this structure by particular receptor proteins. It remains uncertain whether it is the branched structure, the number of carbons on the O-side of the ester bond, or the entire number of carbons in the molecule, that is/are responsible for the differential response. Investigation of odorants bridging the gap between the structure of either ethyl acetate and isoamyl acetate or ethyl butyrate and isoamyl butyrate in a step-by-step fashion should help to determine the minimal stimulus required for the response. Similarly, the specificity of Fields 6 and 7 suggests a graded response that increases from ethyl acetate to isoamyl butyrate; this possibility could be confirmed by investigating the intermediate compounds.

There is no guarantee that the optimal stimulus for any of these fields was present in our sample of four esters. Other regions of the glomerular layer (e.g., Fields 1 and 2) accumulated more 2-DG than Fields 3 through 7 (Fig. 3), resulting in maximal z‑score values of 4 to 5 as opposed to ~3 (Fig. 8). If all glomeruli are capable of the same uptake of 2‑DG, then it is possible that other odorants could have led to additional uptake within Fields 3 through 7. Alternatively, the current odorants may indeed represent the best stimuli for receptors that could require a higher odorant concentration for an optimal response.

The ability to relate magnitude of response to discrete molecular features suggests the ultimate prospect of determining a map of odorant chemistry across the bulb surface, much like the tonotopic maps in auditory cortex, the somatotopic maps in somatosensory cortex, and the visual field maps in visual cortex (Somjen, 1972). A thorough rendering of this "chemotopic" map will need to include investigations of numerous factors, for example, functional groups, cis-trans isomers, hydrocarbon length and structure, steric factors, stereochemistry, ring substitution, and the influence of odorant concentration.

It should be noted that studies of brain activity typically focus on regions of increased activity as being critical for understanding differential responses, but it could be the case that decreases in activity or moderate changes in activity carry important encoded information within the olfactory system. We restricted our post-hoc statistical analyses to fields showing moderate to high activity in maps of uptake averaged across rats exposed to the same odorants and revealed clear differences in the encoded glomerular-layer pattern of activity produced by closely related chemicals. However, it remains possible that other lower-activity areas of the glomerular layer could also show more subtle, feature-specific changes across odorants.

Representations of responses in both lateral and medial glomerular layers

Units of response that correlated with odorants or odorant features typically occurred in pairs, with one situated laterally and one medially. The medial field invariably was located more caudally than the lateral one, and typically was positioned more ventrally (Fig. 9), which is consistent with the projection patterns of olfactory receptor neurons expressing the same putative olfactory receptor gene (Ressler et al., 1994; Vassar et al., 1994; Mombaerts et al., 1996). The pairs of Fields 1 and 2, Fields 4 and 5, and Fields 6 and 7 each displayed remarkable similarities in odorant specificities (Fig. 8). Field 3 also may have an anterior, dorsal equivalent activated by ethyl esters. This region was located just anterior to the first subependymal (around section 10) and at at angles of 0° to 12° (Fig. 6). However, because it is difficult to maintain complete tissue integrity at the dorsal extremity of coronal sections, there was an insufficient number of rats contributing values in this location for a statistical validation of the specificity of this field. The pairs of Fields 4 and 5 and Fields 6 and 7 also had a remarkably similar number of glomeruli associated with the region of focal 2‑DG uptake in single coronal sections (Fig. 10). The spatial relationship along medial-lateral, rostral-caudal, and dorsal-ventral axes between these pairs of similarly specific foci offers compelling, albeit indirect, support for the functional relevance of the putative olfactory receptor genes, given the similar projection patterns of the olfactory sensory neurons that express the same gene. It nevertheless remains possible that these correlated pairs of foci are actually independent. Further studies on specificity using additional odorants or direct demonstrations of co-localization of paired foci with olfactory receptor mRNA may be required for a more definitive answer to this question.

Relationships between numbers of glomeruli and fields of 2‑DG uptake

High-uptake foci within the isoamyl butyrate-specific Fields 6 and 7 were associated with very few glomeruli in any given coronal section (Figs. 9 and 10). In some cases, the foci appeared to involve only a single glomerulus in a section. This would be the predicted consequence of the activation of a single class of olfactory receptor protein expressed by olfactory receptor neurons whose projections converge into a single glomerulus. High-uptake foci within Field 3 activated by ethyl acetate also involved very few glomeruli. In contrast, most odorant and field combinations involved foci of 2‑DG uptake that were associated with more than one glomerulus in any given coronal section. Our ability to find a close correspondence between individual glomeruli and high-uptake foci for the isoamyl butyrate-specific Fields 6 and 7 suggests that the larger numbers of glomeruli associated with other foci is not merely due to a technical limitation. A more likely explanation is that the odorants activated numerous, adjacent glomeruli at the odorant concentrations we employed.

The fact that more activated glomeruli are associated with Fields 4 and 5 in isoamyl acetate-stimulated rats than in isoamyl butyrate-stimulated rats suggests that isoamyl acetate activates additional glomeruli situated near those responding to isoamyl butyrate. Similarly, the increased number of activated glomeruli within Field 3 in ethyl butyrate-exposed rats in comparison to ethyl acetate-exposed rats suggests that additional glomeruli responding more specifically to ethyl butyrate are located near glomeruli responding to ethyl acetate. These data provide further support for previous proposals that adjacent glomeruli receive projections from olfactory receptors with related specificities. These proposals have come from studies where increases in the concentration of a given odorant lead to increased sizes of foci of either 2‑DG uptake or c‑fos mRNA hybridization signal (Stewart et al., 1979; Bell et al., 1987; Guthrie and Gall, 1995), and from studies where local iontophoresis of antagonists believed to reduce lateral inhibition from adjacent glomeruli resulted in a broadening of the receptive fields of mitral cells stimulated with a series of aliphatic aldehydes (Yokoi et al., 1995).

Overlapping responses in the posterolateral glomerular layer

In a series of studies where field potentials were recorded across a large array of electrodes implanted in the posterolateral bulb (1/6 to 1/8 of the total bulb), Freeman and coworkers were unable to isolate any particular subset of electrodes containing more odor-specific information than any other subset (Freeman and Skarda, 1985; Freeman and Baird, 1987; Freeman and Grajski, 1987). Their analysis employed n-amyl acetate as one of the odorants. In the corresponding part of the bulb, we also have found that the four esters evoke glomerular activity that is only subtly different in spatial distribution for the different odorants (Fig. 6). In our previous study of 2‑DG uptake evoked by the odor of peppermint extract, responses also were seen in this area (Johnson and Leon, 1996). Peppermint extract is a complex mixture of odorants, dominated by ethanol (80%) and stereoisomers of menthol, menthyl acetate, and menthone.

It seems possible that the responses in the posterolateral glomerular layer, as well as in the posteromedial layer, are evoked by olfactory receptors with greatly overlapping specificity. With respect to the four esters used in the present study, almost the entire structure of ethyl acetate is contained in all of the odorants. This shared structure is a larger molecular feature than any of the specific molecular features that distinguish the odorants from each other. Thus, one might expect that the region of overlap in the patterns evoked by the four esters would be greater than the regions that discriminate the particular features that distinguish these odorants. Given that the ester bond is a common characteristic of fruit odors, and that fruits are likely to be important food sources for rats in the wild, it may be that numerous receptors and a great area of the olfactory bulb are devoted to the detection of this molecular feature. However, it also is possible that receptors mapping to this region respond to some attribute of odorant chemistry that is not easily defined in terms of discrete features. For example, the receptors mapping to these regions may be general hydrophobicity detectors and/or they may reflect some chromatographic property of the odorants as they distribute across the nasal epithelium (Mozell and Jagodowicz, 1973), either of which could explain the graded shift in the location of maximal response from the small, relatively hydrophilic ethyl acetate to the larger, relatively hydrophobic isoamyl butyrate. It also is possible that within the olfactory sensory neurons that project to the posterior bulb, the subset of olfactory receptor proteins respond to virtually any odor and may function in generalized, non-specific odorant detection rather than in specific odorant responses. Although conclusive evidence for the determinants of responses within the posterior bulb certainly will require investigations of many more odorants, the present results strongly support the existence of spatial coding in other regions of the bulb and may explain failures to detect spatial coding of odors in the posterolateral bulb.

Individual variation

We have obtained clear, statistically significant evidence for similar patterns of activity in rats exposed to the same odorants. However, there was individual variation in the activity patterns across animals; no individual rat showed the exact same activity pattern that emerged upon averaging across all rats exposed to that odorant, and no pair of individual rats displayed exactly the same pattern of activity. Our perception was that individual rats had unique high-uptake foci of activity in addition to those shared by the other rats exposed to the same odorant (compare Fig. 3 to Fig. 6). A profound limitation of the [14C]2-deoxyglucose method used here is that it can only be performed once for a given animal. We therefore could not assess whether these unique foci appear reliably in a given animal upon exposure to the odorant, or whether they indicate spurious activity that, for example, may relate to odors brought with the animal to the test apparatus. In our previous analysis, we found evidence for high uptake foci in air-exposed rats that were also seen in odorant-exposed rats taken from the same home cage (Johnson and Leon, 1996). However, in the current study, home cage odors were intentionally reduced and ultra-zero grade air was used as a vehicle, and we observed very few high-uptake foci in the air-exposed animals. We therefore can not eliminate the possiblity that odor coding within an individual animal may involve unique regions of response in addition to those that are present in the majority of the rats. Given the evidence for specific anosmias in humans (Amoore, 1974), and for allelic inactivation of certain receptor genes in rodents (Chess et al., 1994), it appears possible that there is individual variation in the receptor repertoire available for odor coding.

Odorant features and olfactory coding

In conclusion, the present results are consistent with a combinatorial mechanism of olfactory coding wherein unitary responses of olfactory receptors to odorant features would produce spatial patterns of bulbar activity that are characteristic for a given odorant. We found that different pure but closely related odorants generated distinct spatial patterns of glomerular response. The spatial response increased in complexity with an increase in the complexity of the odorant. Units of the spatial pattern corresponded to individual odorants with discrete chemical features. The units were present in lateral and medial pairs such as would be predicted from the spatial patterns of the projections of olfactory receptor neurons transcribing the same putative olfactory receptor gene. Finally, our systematic method for analyzing 2‑DG uptake, and our systematic choices of odorants, allow for the visualization of a rudimentary "chemotopic" map of the glomerular layer that might be refined in future studies of odorants possessing additional discrete features. Thus, the principles used in odor coding may be analogous to those used in other sensory systems, where categories of sensory stimuli such as frequencies of sounds, locations of visual objects, and locations of somatosensory stimulation are represented spatially at an early stage of central nervous system processing (Somjen, 1972).

ACKNOWLEDGMENTS

We thank Edna Hingco for assistance with data collection and analysis, and Dr. Garr Updegraff for writing software to facilitate the generation of arrays from individual data files.

LITERATURE CITED

Amoore, J.E. (1974) Four primary odor modalities of man: Experimental evidence and possible significance. In D.A. Denton and J.P. Coghlan (eds): Olfaction and Taste V. New York, NY: Academic Press, pp. 283-289.

Bell, G.A., D.G. Laing, and H. Panhuber (1987) Odour mixture suppression: evidence for a peripheral mechanism in human and rat. Brain Res. 426:8-18.

Buck, L.B. (1996) Information coding in the vertebrate olfactory system. Annu. Rev. Neurosci. 19:517-44.

Buck, L.B. and R. Axel (1991) A novel multigene family may encode odorant receptors: A molecular basis for odor recognition. Cell 65:175-187.

Chess, A., I. Simon, H. Cedar and R. Axel (1994) Allelic inactivation regulates olfactory receptor gene expression. Cell 78:823-834.

Coopersmith, R., S.R. Henderson, and M. Leon (1986) Odor specificity of the enhanced neural response following early odor experience in rats. Dev. Brain Res. 27:191-197.

Dean, P.M. (1987) Molecular Foundations of Drug-Receptor Interaction. Cambridge: University Press.

Freeman, W.J. and C.A. Skarda (1985) Spatial EEG Patterns, non-linear dynamics and perception: the Neo-Sherringtonian view. Brain Res. Rev. 10:147-175.

Freeman, W.J. and B. Baird (1987) Relation of olfactory EEG to behavior: Spatial analysis. Behav. Neurosci. 101:393-408.

Freeman, W.J. and K.A. Grajski (1987) Relation of olfactory EEG to behavior: Factor analysis. Behav. Neurosci. 101:766-777.

Guthrie, K.M., A.J. Anderson, M. Leon, and C. Gall (1993) Odor-induced increases in c-fos mRNA expression reveal an anatomical "unit" for odor processing in olfactory bulb. Proc. Natl. Acad. Sci. USA 90:3329-3333.

Guthrie, K.M. and C.M. Gall (1995) Functional mapping of odor-activated neurons in the olfactory bulb. Chem. Senses 20:271-282.

Hass, H.B. and R.F. Newton (1975) Correction of boiling points to standard pressure. In R.C. Weast (ed.) Handbook of Chemistry and Physics. 56th ed. Cleveland, Ohio: CRC Press, pp.

Imamura, K., N. Mataga, and K. Mori (1992) Coding of odor molecules by mitral/tufted cells in rabbit olfactory bulb. I. Aliphatic compounds. J. Neurophysiol. 68:1986-2002.

Johnson, B.A. and M. Leon (1996) Spatial distribution of [14C]2-deoxyglucose uptake in the glomerular layer of the rat olfactory bulb following early odor preference learning. J. Comp. Neurol. 376:557-566.

Katoh, K., H. Koshimoto, A. Tani, and K. Mori (1993) Coding of odor molecules by mitral/tufted cells in rabbit olfactory bulb. II. Aromatic compounds. J. Neurophysiol. 70:2161-2175.

Kauer, J.S. and A.R. Cinelli (1993) Are there structural and functional modules in the vertebrate olfactory bulb? Microsc. Res. Tech. 24:157-167.

Kent, P.F. and M.M. Mozell (1992) The recording of odorant-induced mucosal activity patterns with a voltage-sensitive dye. J. Neurophys. 68:1804-1819.

Kishimoto, J., H. Cox, E.B. Keverne, and P.C. Emson (1994) Cellular localization of putative odorant receptor mRNAs in olfactory and chemosensory neurons: a non radioactive in situ hybridization study. Mol Brain Res. 23:33-39.

Mombaerts, P., F. Wang, C. Dulac, S.K. Chao, A. Nemes, M. Mendelsohn, J. Edmondson, and R. Axel (1996) Visualizing an olfactory sensory map. Cell 87:675-686.

Mori, K., K. Imamura, and N. Mataga (1992) Differential specificities of single mitral cells in rabbit olfactory bulb for a homologous series of fatty acid odor molecules. J. Neurophysiol. 67:786-789.

Mozell, M.M. and M. Jagodowicz (1973) Chromatographic separation of odorants by the nose: retention times measured across in vivo olfactory mucosa. Science 181:1247-1249.

Raming, K., J. Krieger, J. Strotmann, I. Boekhoff, S. Kubick, C. Baumstark and H. Breer (1993) Cloning and expression of odorant receptors. Nature 361:353-356.

Ressler, K.J., S.L. Sullivan, and L.B. Buck (1994) Information coding in the olfactory system: Evidence for a stereotyped and highly organized epitope map in the olfactory bulb. Cell 79:1245-1255.

Royet, J.P., G. Sicard, C. Souchier, and F. Jourdan (1987) Specificity of spatial patterns of glomerular activation in the mouse olfactory bulb: computer-assisted image analysis of 2‑deoxyglucose autoradiograms. Brain Res. 417:1-11.

Sato, T, J. Hirono, M. Tonoike, and M. Takebayashi (1994) Tuning specificities to aliphatic odorants in mouse olfactory receptor neurons and their local distribution. J. Neurophys. 72:2980-2989.

Shepherd, G.M. (1991) Computational structure of the olfactory system. In J. Davis and H. Eichenbaum (eds): Olfaction as a Model System for Computational Neuroscience. Cambridge, MA: MIT Press, pp. 3-41.

Shepherd, G.M. (1994) Discrimination of molecular signals by the olfactory receptor neuron. Neuron 13:771-790.

Somjen, G. (1972) Sensory Coding in the Mammalian Nervous System. New York: Appleton-Century-Crofts.

Stewart, W.B., J.S. Kauer, and G.M. Shepherd (1979) Functional organization of rat olfactory bulb analysed by the 2-deoxyglucose method. J. Comp. Neurol. 185:715-734.

Sullivan, S.L. and L. Dyer (1996) Information processing in mammalian olfactory system. J. Neurobiol. 30:20-36.

Vassar, R., S.K. Chao, R. Sitcheran, J.M. Nunez, L.B. Vosshall, and R. Axel (1994) Topographic organization of sensory projections to the olfactory bulb. Cell 79:981-991.

Woo, C.C. and M. Leon (1991) Increase in a focal population of juxtaglomerular cells in the olfactory bulb associated with early learning. J. Comp. Neurol. 305:49-56.

Yokoi, M., K. Mori, and S. Nakanishi (1995) Refinement of odor molecule tuning by dendrodendritic synaptic inhibition in the olfactory bulb. Proc. Natl. Acad. Sci. USA 92:3371-3375.

FIGURE LEGENDS

Fig. 1. Chemical structures of the odorants used in this study. The stippled boxes indicate the additional feature present in the isoamyl esters that is lacking from the ethyl esters, and the open boxes indicate the additional feature of the butyrates that is lacking from the acetates.

Fig. 2. Schematic of a sagittal bulb section showing locations of the anterior-posterior hallmarks used in this study. AOB, accessory olfactory bulb; GL, glomerular layer; MCL, mitral cell layer; SEZ, subependymal zone.

Fig. 3. Color-coded contour charts displaying the spatial distribution of relative 2‑DG uptake (glomerular layer/bulbar core) for single bulbs of each animal from two litters. Section #1 is the first to contain a mitral cell layer, section #15 the first to contain a subependymal zone, section #60 the first to contain an accessory olfactory bulb, and section #110 the last to contain a mitral cell layer on the medial aspect of the bulb (Fig. 2). Measure #1 is dorsal, measure #13 is midlateral, measure #25 is ventral, measure #37 is midmedial, and measure #48 again approaches dorsal. The blank region in the upper right of each of these charts indicates that the lateral glomerular layer is missing from these sections, having been replaced by the anterior olfactory nucleus and lateral olfactory tract. Individual values of uptake are at the intersections of gridlines. Equal intervals of relative uptake are assigned distinct colors; cool colors represent low uptake and warm colors, high uptake. The highest interval (3.5 to 4.0) is represented by dark red. Arrows indicate foci of 2‑DG uptake that were reliably present for both isoamyl acetate and isoamyl butyrate, but that rarely were seen for the other odorant conditions.

Fig. 4. Histograms of pattern dissimilarities. Indices of pattern dissimilarity were calculated for each pair of rats in the study (595 total), and then were subdivided into cases where the pair was exposed either to the same odorant (n = 105) or to different odorants (n = 490). Note that the values for same-odorant comparisons are lower than those for different-odorant comparisons (Mann-Whitney U test, p < 0.0001).

Fig. 5. Values of same-odorant versus different-odorant pattern dissimilarities calculated for each pair of odorant conditions. "Same" indicates rats exposed to odorant 1 compared to other rats exposed to odorant 1, and rats exposed to odorant 2 compared to other rats exposed to odorant 2. "Different" indicates rats exposed to odorant 1 compared to rats exposed to odorant 2. The mean value ± the standard error are shown. For each odorant pair, the values for different-odorant comparisons were significantly larger than the values for same-odorant comparisons (Mann-Whitney U tests, p < 0.002). Ea, ethyl acetate; eb, ethyl butyrate; iaa, isoamyl acetate; iab, isoamyl butyrate.

Fig. 6. Color-coded contour charts displaying z-score values averaged across rats exposed to the same odorants. Black arrows indicate fields of uptake that were higher for the ethyl esters; straight, white arrows denote fields that were more pronounced for the isoamyl esters; and curved, white arrows denote fields that showed apparent specificity for isoamyl butyrate.

Fig. 7. Correlation between pattern dissimilarities calculated for pairs of averaged arrays and differences in total number of carbons between odorant pairs. The line is the result of least-squares linear regression.

Fig. 8. Analyses of uptake in individual fields of the averaged maps. The left panel shows the boundaries used to define each field superimposed on gray scale-coded contour charts of z-scores averaged across rats exposed to the same odorants. Field 3 corresponds to a region indicated by black arrows in Figure 6; Fields 4 and 5 correspond to the regions denoted by straight, white arrows; and Fields 6 and 7 correspond to regions indicated by curved, white arrows. A-P center is the section number containing the anterior-posterior coordinate of the centroid of the response within Fields 1 and 2. Error bars denote standard errors across animals.

Fig. 9. Correspondence between fields of 2-DG uptake and individual glomeruli. Portions of coronal bulb sections are shown to illustrate the alignment between foci of high 2-DG uptake and glomeruli in adjacent Nissl-stained sections for isoamyl acetate Fields 4 and 5, isoamyl butyrate Fields 6 and 7, ethyl acetate Field 3, and ethyl butyrate Field 3. On the left of each series is a grayscale contrast-enhanced image of the Nissl-stained section, on the right is a pseudocolor-enhanced image of the adjacent 2-DG section, and in the middle is an overlay of the two images. Warm colors denote higher 2-DG uptake and cool colors denote lower uptake. The arrows point to the foci located within the respective fields. The examples shown for isoamyl acetate 4 and 5 are from one animal, and those for isoamyl butyrate 6 and 7 are from another animal. Note that the number of glomeruli associated with the high uptake foci is similar for isoamyl acetate 4 (lateral) and isoamyl acetate 5 (medial), and for isoamyl butyrate 6 (lateral) and isoamyl butyrate 7 (medial). In addition, fewer glomeruli are associated with isoamyl butyrate 6 and 7 than with isoamyl acetate 4 and 5, and the ethyl butyrate response in Field 3 involves more glomeruli than the ethyl acetate response. The scale bar (under the top, left series) equals 1.0 mm.

Fig. 10. Number of glomeruli associated with foci of 2‑DG uptake in individual fields. The number of glomeruli underlying focal 2‑DG uptake were evaluated in Nissl-stained sections adjacent to the autoradiograph sections. Fields were defined as illustrated in Figure 8, and the odorants were isoamyl acetate (IAA), isoamyl butyrate (IAB), ethyl acetate (EA), and ethyl butyrate (EB). The mean number of glomeruli ± standard error across rats are shown. The number of glomeruli were found to differ significantly across these odorant/field conditions (p < 0.0001, single factor ANOVA). Note the similarities in the number of associated glomeruli for any given odorant within fields showing similar odorant specificity, i.e., Fields 4 and 5, and Fields 6 and 7. Note also the difference between odorants in the number of associated glomeruli for a given field, i.e., IAA vs IAB in Fields 4 and 5, and EA vs EB in Field 3.