- Email: [email protected]
Olfactory coding in the mammalian olfactory bulb
Brain Research Reviews 42 (2003) 23–32 www.elsevier.com / locate / brainresrev
Review
Olfactory coding in the mammalian olfactory bulb Michael Leon*, Brett A. Johnson Department of Neurobiology and Behavior, University of California, Room 2205 MH, Irvine, CA 92697 -4550, USA Accepted 21 January 2003
Abstract There have been a number of recent approaches to the study of olfactory coding, each of which has its advantages and disadvantages. In the present review, we discuss our own work on this topic, which has involved mapping uptake of [ 14 C]2-deoxyglucose across the entire glomerular layer of the rat main olfactory bulb in response to systematically selected pure odorant molecules. Our strategy to understand the olfactory code has involved four approaches. In the first, we determined whether the system encodes odorants in their entirety, or whether it encodes odorants by representing combinations of molecular features that add together to comprise a neural picture of each odorant. Multiple odorant features appeared to be coded by multiple receptors. Our second strategy examined the ways that such features are represented. We stimulated rats with odorants that differed greatly in their molecular structure to be able to identify a set of odorant feature response domains. Our third approach asked how odorants with very small differences in molecular structure are coded, and we found systematic differences in the representation of such features within response domains. Finally, we were able to predict odor perception from the neural representations of odorants that differed in only a single aspect of their structure. Using these strategies, we have been able to learn some of the rules by which the olfactory code operates. These rules have allowed us to predict where previously unmapped molecules would be represented and how differences in molecular representations affect olfactory perceptions. 2003 Elsevier Science B.V. All rights reserved. Theme: Sensory systems Topic: Olfactory senses Keywords: Olfactory; Odorant; Glomerulus; Olfactory bulb; Coding; Response domain
Contents 1. Introduction ............................................................................................................................................................................................ 2. Functional organization of the olfactory system ......................................................................................................................................... 3. Are odorants coded by a combination of receptors?.................................................................................................................................... 4. What constitutes a molecular feature? ....................................................................................................................................................... 5. How does the brain code odorants that are very similar in structure? ........................................................................................................... 6. What is the critical molecular feature that produces the chemotopic response? ............................................................................................. 7. Can we predict the neural response from the structure of the molecule? ....................................................................................................... 8. Can we predict odor perception from the neural response to the molecular structure? ................................................................................... 9. Signal vs. noise in the olfactory system ..................................................................................................................................................... 10. Learning and the signal to noise ratio ...................................................................................................................................................... 11. Noise, signal and concentration............................................................................................................................................................... 12. Relational responses to odorants ............................................................................................................................................................. 13. Odorant significance .............................................................................................................................................................................. 14. Temporal aspects of mammalian olfactory coding .................................................................................................................................... 15. Discrimination mechanism .....................................................................................................................................................................
*Corresponding author. Tel.: 11-949-824-5343; fax: 11-949-824-2447. E-mail address: [email protected] (M. Leon). 0165-0173 / 03 / $ – see front matter 2003 Elsevier Science B.V. All rights reserved. doi:10.1016 / S0165-0173(03)00142-5
24 24 24 25 26 26 27 28 29 29 29 29 30 30 30
24
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
16. Summary .............................................................................................................................................................................................. Acknowledgements ...................................................................................................................................................................................... References...................................................................................................................................................................................................
1. Introduction A code is a set of rules by which information is transposed from one form to another. In the case of olfactory coding, it would describe the ways in which information about odorant molecules is transposed into neural responses. We thought that when we understood that code, we might be able to predict the odorant molecule from the neural response, the neural response from the odorant molecule, and the perception of the odorant molecule from the neural response. The following paper reviews our approach to the issue of olfactory coding. Because a critical or comprehensive assessment of other approaches to the study of the olfactory code are not provided, the reader may want to consult one of several recent reviews for a more complete perspective on this area [6,12,16,26–28,51].
2. Functional organization of the olfactory system Olfactory transduction in the rat starts with perhaps a thousand different types of olfactory receptors located on the cilia of a large number of olfactory receptor neurons that comprise the olfactory epithelium [3,53]. Each of these neurons probably expresses a single type of olfactory receptor gene and homologous receptor neurons project to a small number of olfactory glomeruli paired on the medial and lateral aspects of the olfactory bulb (Fig. 1) [31,33,38,46,50]. The glomeruli are dense synaptic bundles in which many homologous olfactory receptor neurons connect to second-order neurons. A small number of mitral cells emanate from each glomerulus and project to a number of regions, including the olfactory cortex. The projections of mitral cells receiving input from homologous olfactory receptor neurons form reliable discrete clusters in different regions of the olfactory cortex [54]. Interneurons mediate inhibition between glomeruli and between mitral cells [42].
3. Are odorants coded by a combination of receptors? There are many different olfactory receptors in rodents [3,53], but there are orders of magnitude more odorants to which they respond [1,7,34]. It is therefore likely that most odorants are coded by a combination of receptors such that a unique combination of responses would describe any particular molecule [37]. One way this could occur is by
30 31 31
simple feature extraction in which different receptors independently recognize different parts of the same odorant molecule. To test this combinatorial, feature-extraction hypothesis, we compared the responses to pairs of odorants that either shared or did not share a molecular feature to which the receptors might respond [24]. Specifically, we exposed rats to one of four esters sharing a straight-chain hydrocarbon domain, with two of these molecules sharing an isoamyl feature and two sharing an ethyl feature. All of these odorants were exposed at the same vapor concentrations to insure that any difference in response was due to the qualitative differences among odorants, rather than to a difference in the number of molecules to which they were exposed. We also exposed odorants at relatively low concentrations to avoid seeing low-affinity binding to receptors. Glomerular responses were assessed across the entire lamina using [ 14 C]2-deoxyglucose autoradiography [24]. We developed a method in which discrete measurements of uptake are taken at systematic angle increments around equally spaced coronal bulb sections. The data from individual sections then is merged into arrays and standardized for differences in bulb size. The anatomically standardized data arrays can be subjected to a variety of transformations, plotted as color-coded contour charts, and compared statistically to test specific hypotheses concerning the impact of odorant chemistry on activity patterns. We previously discussed the relative advantages of this approach, which include the use of an unanesthetized, freely respiring animal and the ability to analyze the entire glomerular layer [21,22]. At the same time, the technique is unable to resolve the temporal dynamics of the olfactory response, and it also is unable to compare responses to multiple stimuli in the same animal. There is no question that different techniques must be used to test different hypotheses regarding olfactory coding. We found that each odorant evoked a unique glomerular pattern that was consistent across rats. The pattern became more complex as the complexity of the molecule increased, suggesting that the larger molecules may have more features that are coded. Focal glomerular responses were also duplicated on the medial and lateral aspects of the bulb in the same anatomical relationship as homologous sensory neuron projections. Most importantly, the two ethyl esters shared a very specific focus of glomerular activity that was not seen for the other two odorants. Similarly, the two isoamyl esters shared a different focus of glomerular activity that was not seen for the other two odorants. These data strongly support the idea that multiple
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
25
Fig. 1. (A) A representation of the convergence of olfactory receptor neurons onto a glomerulus in the olfactory bulb. (B) The medial and lateral paired projections of homologous olfactory receptor neurons shown at both an anterior and a posterior coronal section of the olfactory bulb. (C) The projection of a mitral cell that receives input from a single glomerulus to a pyramidal cell in the olfactory cortex, shown in a sagittal section. AOB, accessory olfactory bulb.
receptors can code odorant molecules as a combination of their constituent features.
4. What constitutes a molecular feature? To determine which characteristics of an odorant can constitute a molecular feature recognizable by odorant
receptors, we began a research program in which panels of odorants that were systematically different in chemical structure were used as stimuli. While we did not know a priori what would constitute a molecular feature that could be recognized by olfactory receptors, it seemed possible that chemical functional groups might constitute such a feature. If a chemical functional group is a coded feature, then exposing rats to molecules that differed only in their
26
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
functional group should differentiate their glomerular responses based on that molecular feature. Therefore, we exposed rats to odorants with the same straight-chain hydrocarbon structure, but with different functional groups: an acid, a ketone, an ester, an alcohol, and an aldehyde. We then compared glomerular activity patterns to determine whether functional groups could constitute molecular features that are recognized by odorant receptors. We found that each odorant stimulated a unique combination of glomerular clusters that we call modules [21]. The modules describe a group of glomeruli in which we reliably find responses to a particular odorant feature. Again, the glomerular module activity was found both medially and laterally for each odorant. Most importantly, odorants with different functional groups stimulated different glomerular modules. These data raised the possibility that olfactory receptors are sensitive to chemical functional groups as part of the combination of molecular features that describe an odorant. Uchida et al. [49], observing a limited glomerular population, subsequently suggested that functional groups play a role in activating two glomerular domains, and Araneda et al. [2] then found that substituting functional groups eliminated the specific response of an identified receptor to otherwise identical odorant ligands. The presence of only one molecular representative of each functional group in this study, however, raised the possibility that the different glomerular responses that we observed were due to the specific structures of the molecules that we tested, rather than specifically to the functional group on these molecules. To test the generality of these conclusions we exposed rats to a wide variety of odorants that shared functional groups, but had very different hydrocarbon structures and numbers of carbons [19]. If molecules with the same functional groups, but very different hydrocarbon configurations stimulated the same glomerular modules, then we could conclude that the functional group itself may be a critical molecular feature to which the receptors specifically responded. We exposed groups of rats to one of 54 odorants with a number of representatives of the different functional groups that we previously described. The clustered, often overlapping responses were circled to indicate the possibility of a response domain (Fig. 2). As before, each odorant could be described by a unique set of glomerular module responses that were duplicated medially and laterally. As predicted, molecules with specific functional groups stimulated those areas that were previously stimulated by molecules with the same functional groups, but with different hydrocarbon configurations. Even the smallest representatives of molecular functional groups had responses in the modules activated by the larger molecules with that functional group. These data provide strong support for the notion that olfactory receptors recognize specific functional groups as molecular features. While the
anterior modules had response domains sensitive to functional groups, we found that the large posterior responses varied with the hydrocarbon structure of the odorant molecules.
5. How does the brain code odorants that are very similar in structure? We noted that responses to different odorants differed in their exact positions within certain modules. We therefore considered the possibility that this variation was due to some systematic difference among the odorants. To address that issue, we varied one aspect of the odorant molecules systematically and then determined the glomerular response pattern in a particular response domain. We tested the response to a homologous series of organic acids in which the number of carbons in the aliphatic chain were systematically incremented [25]. We found that responses to these organic acid odorants were clustered in the paired modules that responded to the acid functional group [25]. The responses within these modules overlapped across odorants, and the focus of activity moved ventrally with additional carbons in the aliphatic structure. Similar response patterns have since been reported for aliphatic aldehydes and alcohols [14,32,39,49]. These data support the idea that certain glomerular responses to such odorants are organized chemotopically and that differences in the responses of neighboring glomeruli may underlie differences in the perception of these very similar odorants.
6. What is the critical molecular feature that produces the chemotopic response? Olfactory receptors cannot actually respond directly to the number of odorant carbons, but they are likely to be differentially responsive to a molecular feature such as hydrophobicity, length, or volume, which covaries with carbon number. Since one of the goals of studying odor coding is to determine the true relationships between stimulus and neural response, we determined which of these properties was critical in producing the systematic changes in their representation. Specifically, we determined whether the change in location of the response focus in the responsive module evoked by acid odorants was due to incremental changes in hydrophobicity, molecular length, or molecular volume. To distinguish among these possibilities, we exposed groups of rats to organic acids that had the same number of carbons, but differed in their hydrocarbon structures, thereby allowing the acids to differ independently in hydrophobicity, length, and volume [22]. We found that the only molecular property that strongly correlated with the location of the activity focus in the responsive module was molecular length, suggesting that this molecular feature is the principal determinant of the
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
27
Fig. 2. Outlined, paired modules have a preferential response for the ketone functional group, regardless of odorant hydrocarbon structure [19].
chemotopic response. Using a different approach, Araneda et al. [2] found that a specific olfactory receptor was also particularly sensitive to odorant molecular length.
7. Can we predict the neural response from the structure of the molecule? Another measure of our understanding the odor code is whether we can predict neural activity in the bulb from the structure of molecules. We noted that molecules without
any oxygen moiety activated more ventral and caudal glomerular regions than were activated by medium-sized odorants containing oxygen atoms [19]. This response pattern presented an opportunity to predict how odorants would be represented in the bulb. Specifically, we predicted that odorant features with no oxygen moiety would also activate caudal and ventral areas of the bulb. To test this prediction, we exposed rats to pinene and santalol, two odorants containing dense hydrocarbon features without oxygen, and found that both stimulated caudal, ventral modules. These data demonstrate that we can predict
28
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
certain novel neural response patterns from odorant molecular features.
8. Can we predict odor perception from the neural response to the molecular structure? If odors are coded using particular patterns of glomerular activity, it should be possible to predict aspects of odor perception based on these activity patterns. If we could find odorants differing in a single molecular feature that induces a simple difference in their glomerular response, then we would predict that such odorants would be discriminated. However, if there were odorants with differences in molecular structure, but without differences in their evoked glomerular activity, then such odorants should not be discriminated. To test this hypothesis, we used pairs of enantiomers (optical isomers), which have the same number of carbons, the same functional groups and the same hydrocarbon structure, except that they are mirror images of each other [29]. The use of such molecules limits the possible basis for discrimination to a single
aspect of the molecules, their stereoconfiguration. We then predicted the difficulty with which odorant pairs could be discriminated, based on the glomerular activity patterns. While the enantiomers of carvone evoked both shared and statistically different modular responses, the modular glomerular activity patterns evoked by the enantiomers of both terpinen-4-ol and limonene were not statistically different from each other [29] (Fig. 3). As predicted, rats spontaneously discriminated between the enantiomers of carvone in a cross-habituation paradigm. Conversely, the enantiomers of both limonene and terpinen-4-ol were not spontaneously discriminated. Therefore, we are able to predict the perception of odorant molecules based on the neural activity that they evoke, which is a step toward understanding the olfactory code. Patterns of glomerular activity measured using our 2-DG mapping technique also predicted quantitative similarities in odor perception across the series of aliphatic acids that differed incrementally in carbon number [25]. Cleland et al. [8] tested the behavioral responses to those same odorants using the cross-habituation assay to assess spontaneous discriminations by rats among a structurally similar
Fig. 3. (Right) Our current model of modular responses in the olfactory bulb. Apparent modular responses to odorants were outlined and superimposed on one another to identify modules used in the representations of multiple odorants. Shown here are those modules that were activated by more than three of the 54 odorants studied. For a great majority of the odorants, whenever a module was identified in the lateral aspect of the bulb, a module of similar activity was detected in the medial bulbar aspect. The corresponding lateral and medial modules are labeled here by using the same letter, lower case for the lateral representation and upper case for the medial representation. (Left) Mean maximal z-score response in each module evoked by the enantiomers of carvone, limonene, and terpinen-4-ol are shown on the left. Asterisks indicate significant differences (P,0.01) between enantiomers in individual modules. The lack of activity in module i for terpinen-4-ol reflects a negative z-score for both enantiomers ([29], used with permission).
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
series of odorants. The relative differences among the spontaneous discriminations was characterized by a behavior dissimilarity index and that number was correlated with a calculated dissimilarity index that characterized the differences across glomerular response patterns to the different odorants that had been reported by Johnson et al. [25]. The correlation coefficient was 0.926, indicating a particularly close relationship between glomerular responses and perception.
29
criminate between each pair of enantiomers, and it took them only somewhat longer to learn to discriminate odorant pairs that were not discriminated spontaneously in the cross-habituation assay [30]. It is also interesting to note that the spread of the small differences across the entire glomerular layer suggest that it would be difficult to make an olfactory bulb lesion large enough to interfere with discriminations that are differentially reinforced.
10. Learning and the signal to noise ratio 9. Signal vs. noise in the olfactory system These data indicate that we can predict the odorant molecule from the neural response, the neural response from the odorant molecule, and the odor perception from the neural response. It may be particularly important to make these predictions successfully in order to have confidence that the responses that we have identified are neural signals. We regard a neural signal as a neural response that carries critical information regarding the sensory stimulus, while neural noise may be coincident with the signal, but not carry such information. The ability to discriminate between noise and signal in the olfactory system may be particularly important because it seems to be a particularly noisy sensory system. All odorants sensed in the environment are experienced against a background of other odorants. All odorants would be expected to have at least some level of contamination. One also should expect interactions between odorants and the chemicals and enzymes in olfactory mucous. In addition, there may be low-affinity responses in the receptor array. Finally, there appear to be numerous secondary and tertiary responses in the olfactory system that may correlate with the signal, but may not carry information about the odorant. Therefore, it would be possible to mistake noise for signal in an analysis of olfactory responses if we did not establish through these successful predictions that we are studying the signal. One example of neural noise can be seen in an alternate analysis of the glomerular responses to limonene and terpinen-4-ol. Recall that these odorants did not differ when we assessed their major modular responses. However, when we compared these enantiomers for activity at every point in our data array, we noted many small regions of potential difference between them [30]. We can regard these differences as noise because the rats ignored them when asked to make a spontaneous discrimination between enantiomers [29]. At the same time, we reasoned that rats might be able to use the small differences in glomerular response to make discriminations between these odorants if it became important for them to do so. In essence, they may be able to turn noise into a signal if differential responses were reinforced, rather than being spontaneous. Indeed, we found that differential reinforcement induced rats to dis-
We have also shown that reinforced learning in young animals can alter the focal glomerular response by enhancing it [9,20,23,47]. The resulting increase in signal-to-noise ratio in the glomerular layer would be expected to increase olfactory acuity, and that is exactly what Fletcher and Wilson [13] have reported for adult rats.
11. Noise, signal and concentration There is another means by which noise can become signal. A small proportion of odorants change their olfactory perception with increasing concentration [1,11]. Pentanal recruits new glomerular foci with increasing concentration [21], and humans report that this odorant changes in perception from an ethereal odor to a strong, noxious odor. Such a change in perception would be expected with the excitation of new modules, as the background noise of the contaminants becomes a signal with increasing concentration. Indeed, the likely contaminant in this case is the oxidation product of pentanal, which would be pentanoic acid, and the addition of the strong noxious odor of pentanoic acid added to the ethereal scent of pentanal would account for the change in olfactory perception that has been noted.
12. Relational responses to odorants Most odorants are perceived as having the same qualitative odor across a wide range of concentrations. But while perception remains constant, the absolute number of activated glomeruli in any particular focus increases in number with increasing concentration. Because homologous olfactory receptor neurons with closely related olfactory receptor genes project to neighboring glomeruli [10,48], normal ligand / receptor relationships would predict such a broadening of the local response within a response domain with increasing odorant concentration [15,45]. At the same time, one would expect the olfactory perception to change if there were new glomeruli stimulated by the same odorant when exposed at a high concentration. However, when the glomerular responses are reported as relative to background (noise) responses, in
30
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
our case as z scores, the pattern of response remains constant over the range of concentrations that we presented [21]. These data suggest that upstream neurons may respond to the relative responses emanating from the bulb, a mechanism that would help to filter responses to noise, particularly as concentrations increase. One can imagine, for example, that the olfactory bulb projection areas may respond to glomerular activity only in relation to the response of one aspect of the olfactory system that may provide odorant concentration information.
13. Odorant significance It is possible that rats have evolved larger responses to some odorant features because those responses characterize odorants that have special significance. Indeed, when we tested the response of rats to two natural odorants that are produced by other rats [17], pentanoic acid and methylbutyric acid, we found large, continuous response foci in the glomerular layer [25]. At the same time, we found that two other acids that are never found in nature evoked a number of small discontinuous responses in the lamina [25].
14. Temporal aspects of mammalian olfactory coding In order for combinatorial coding to function in the olfactory system, there almost certainly must be a critical role for a temporal mechanism. If the olfactory receptor population essentially breaks up odorants into their constituent features for purposes of coding, the information regarding these features must be reassembled in the olfactory cortex [51,54]. It would make sense for the cortex to regard only those responses from the bulb that arrive synchronously to be part of the combination of features that describe a particular molecule. Schoppa and Westbrook [40,41] have shown that mitral cells emerging from a single glomerulus in rats fire synchronously, thereby providing the basis for a synchronous stimulus arriving at the olfactory cortex for individual odorant features. The synchronous response emanating from each glomerulus would also be expected to serve as an amplification mechanism for each type of olfactory receptor neuron [7]. The synchrony of the responses to multiple features is probably driven by respiration in the rat. Indeed, the olfactory receptor neuron, glomerular and mitral cell responses are focused on the periodic inspiration peak in the respiratory cycle [4,5,36,43,44], thereby presumably driving all odorant-related responses to arrive in the cortex simultaneously. Such a system would provide a temporal filtering device that would eliminate responses arriving outside that synchronized temporal window as neural noise.
15. Discrimination mechanism The specificity of the glomerular response appears to be preserved in the mitral cells. We found that the spatial patterns of activity in the external plexiform layer and granule cell layer, which are secondary to activation of mitral cells, closely resemble patterns in the glomerular layer [25]. Just as we found for aliphatic acids in the glomerular layer of the rat olfactory bulb, mitral cells in the dorsomedial rabbit bulb have also been found to be highly tuned in their response to such odorants within a restricted range of carbon number [18,35]. The chemotopic arrangement of glomeruli and mitral cells responding to odorants of most similar carbon number may ensure that tuning suppresses responses to the most similar odorants [25]. Lateral inhibition between mitral cells apparently associated with neighboring glomeruli seems to underlie odorant tuning in the bulb [52]. The interactions in the bulb therefore appear to amplify and sharpen responses to the olfactory stimulus, thereby facilitating the ability of the olfactory cortex to resolve differences among odorants.
16. Summary Our data point to a number of coding strategies that appear to be needed to encode the large number of olfactory stimuli that vary in numerous ways across a variety of dimensions and parameters. First, most odorants appear to be coded by a combination of their molecular features. Second, the molecular features appear to use an identity code in which specific neurons transmit specific information within the system. Third, the system seems to be capable of using spatial relationships among odorants to facilitate its ability to encode information regarding closely related odorants. Fourth, there appears to be a dynamic code, in which noise can be converted to signal by increasing concentration or with differential reinforcement. Fifth, there may be a relational code in which glomerular responses are considered in relation to response across the glomerular layer. Finally, there appears to be a significance code that has assigned differential numbers of glomeruli to be responsive to specific molecules that may be particularly important to that species. To provide a first approximation of the olfactory code, we have chosen to make the assumption that the olfactory system uses a relatively straightforward strategy for coding odorant molecules. That is, we have assumed that olfactory coding involves the use of different olfactory receptors that respond to different molecular features of odorants. We have obtained significant support for this perspective. It is possible that there are emergent responses within the system, or that there are network properties that code information by using variations in responses across the entire system, or that the glomerular responses are shaped by centrifugal input, or that there are temporal dynamics
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
that hold the key to olfactory quality. However, there is no comparable body of data at this time to support such complex hypotheses that has the same level of predictive power as the model that emerges from our simple approach. At the same time, we have just begun to examine the extraordinarily wide range of odorant molecules and it is certainly possible that there are additional coding mechanisms that are involved in producing the perceptions of molecules that have yet to be studied.
[18]
[19]
[20]
[21]
Acknowledgements This work was supported by DC03545.
[22] [23]
References
[24]
[1] S. Arctander, Perfume and Flavor Chemicals (Aroma Chemicals), Allured Publishing Company, Carol Stream, IL, 1994. [2] R.C. Araneda, A.D. Kini, S. Firestein, The molecular receptive range of an odorant receptor, Nat. Neurosci. 3 (2000) 1248–1255. [3] L.B. Buck, R. Axel, A novel multigene family may encode odorant receptors: A molecular basis for odor recognition, Cell 65 (1991) 175–187. [4] M.A. Chaput, EOG responses in anesthetized freely breathing rats, Chem. Senses 25 (2000) 695–701. [5] M. Chalansonnet, M.A. Chaput, Olfactory bulb output cell temporal response patterns to increasing odor concentrations in freely breathing rats, Chem. Senses 23 (1998) 1–9. [6] T.A. Christensen, J.G. Hildebrand, Pheromonal and host-odor processing in the insect antennal lobe: how different?, Curr. Opin. Neurobiol. 12 (2002) 393–399. [7] T.A. Cleland, C. Linster, How synchronization properties among second-order sensory neurons can mediate stimulus salience, Behav. Neurosci. 116 (2002) 212–221. [8] T.A. Cleland, A. Morse, E.L. Yue, C. Linster, Behavioral models for odor similarity, Behav. Neurosci. 116 (2002) 222–231. [9] R. Coopersmith, M. Leon, Enhanced neural response to familiar olfactory cues, Science 225 (1984) 849–851. [10] S. Conzelmann, D. Malun, H. Breer, J. Strotmann, Brain targeting and glomerulus formation of two olfactory neuron populations expressing related receptor types, Eur. J. Neurosci. 14 (2001) 1623–1632. [11] A. Dravnieks, Atlas of Odor Character Profiles. ASTM Data Series DS 61, Philadelphia, 1985. [12] R.W. Friedrich, Real time odor representations, Trends Neurosci. 25 (2002) 487–489. [13] M.L. Fletcher, D.A. Wilson, Experience modifies olfactory acuity: acetylcholine-dependent learning decreases behavioral generalization between similar odorants, J. Neurosci. 22 (2002) RC201. [14] H.U. Fried, S.H. Fuss, S.I. Korsching, Selective imaging of presynaptic activity in the mouse olfactory bulb shows concentration and structure dependence of odor responses in identified glomeruli, Proc. Natl. Acad. Sci. USA 99 (2002) 3222–3227. [15] K.M. Guthrie, C. Gall, Functional mapping of odor-activated neurons in the olfactory bulb, Chem. Senses 20 (1995) 271–282. [16] L.B. Haberly, Parallel-distributed processing in olfactory cortex: new insights from morphological and physiological analysis of neuronal circuitry, Chem. Senses 26 (2001) 551–576. [17] T. Høverstad, T. Midtvedt, T. Bøhmer, Short-chain fatty acids in intestinal content of germfree mice monocontaminated with Es-
[25]
[26] [27] [28] [29]
[30]
[31] [32] [33]
[34] [35]
[36]
[37] [38]
[39] [40]
[41] [42]
31
cherichia coli or Clostridium difficile, Scand. J. Gastroenterol. 20 (1985) 373–380. K. Imamura, N. Mataga, K. Mori, Coding of odor molecules by mitral / tufted cells in rabbit olfactory bulb I. Aliphatic compounds, J. Neurophysiol. 68 (1992) 1986–2002. B.A. Johnson, S. Ho, Z. Xu, J.S. Yihan, S. Yip, E.E. Hingco, M. Leon, Functional mapping of the rat olfactory bulb using diverse odorants reveals modular responses to functional groups and hydrocarbon structural features, J. Comp. Neurol. 449 (2002) 180–194. B.A. Johnson, M. Leon, Spatial distribution of [ 14 C] 2-deoxyglucose uptake in the glomerular layer of the rat olfactory bulb following early olfactory preference learning, J. Comp. Neurol. 376 (1996) 557–566. B.A. Johnson, M. Leon, Modular glomerular representations of odorants in the rat olfactory bulb and the effects of stimulus concentration, J. Comp. Neurol. 422 (2000) 496–509. B.A. Johnson, M. Leon, Odorant molecular length: One aspect of the olfactory code, J. Comp. Neurol. 426 (2000) 330–338. B.A. Johnson, C.C. Woo, H. Duong, V. Nguyen, M. Leon, A learned odor evokes an enhanced Fos-like glomerular response in the olfactory bulb of young rats, Brain Res. 699 (1995) 192–200. B.A. Johnson, C.C. Woo, M. Leon, Spatial coding of odorant features in the glomerular layer of the rat olfactory bulb, J. Comp. Neurol. 393 (1998) 457–471. B.A. Johnson, C.C. Woo, E.E. Hingco, K.L. Pham, M. Leon, Multidimensional chemotopic responses to n-aliphatic acid odorants in the rat olfactory bulb, J. Comp. Neurol. 409 (1999) 529–548. J.S. Kauer, On the scents of smell in the salamander, Nature 417 (2002) 336–342. S. Korsching, Olfactory maps and odor images, Curr. Opin. Neurobiol. 12 (2002) 387–392. G. Laurent, Olfactory network dynamics and the coding of multidimensional signals, Nat. Neurosci. Rev. 3 (2002) 884–895. C. Linster, B.A. Johnson, A. Morse, E. Yue, Z. Xu, E.E. Hingco, Y. Choi, M. Choi, A. Messiha, M. Leon, Perceptual correlates of neural representations evoked by odorant enantiomers, J. Neurosci. 21 (2001) 9837–9843. C. Linster, B.A. Johnson, A. Morse, E. Yue, M. Leon, Spontaneous vs. reinforced olfactory discriminations, J. Neurosci. 22 (2002) 6842–6845. B. Malnic, J. Hirono, T. Sato, L. Buck, Combinatorial receptor codes for odors, Cell 96 (1999) 713–723. ¨ M. Meister, T. Bonhoeffer, Tuning and topography in an odor map on the rat olfactory bulb, J. Neurosci. 21 (2001) 1351–1360. P. Mombaerts, F. Wang, C. Dulac, S.K. Chao, A. Nemes, M. Mendelsohn, J. Edmonson, R. Axel, Visualizing an olfactory sensory map, Cell 87 (1996) 675–686. R.W. Moncrieff, The Chemical Senses, Leonard Hill, London, 1967. K. Mori, N. Mataga, K. Imamura, Differential specificities of single mitral cells in rabbit olfactory bulb for a homologous series of fatty acid odor molecules, J. Neurophysiol. 67 (1992) 786–789. B.D. Philpot, E.M. Lyders, P.C. Brunjes, The NMDA receptor participates in respiration-related mitral cell synchrony, Exp. Brain Res. 118 (1998) 205–209. E.H. Polak, Multiple profile-multiple receptor site model for vertebrate olfaction, J. Theor. Biol. 40 (1973) 469–484. K.J. Ressler, S.L. Sullivan, L.B. Buck, Information coding in the olfactory system: Evidence for a stereotyped and highly organized epitope map in the olfactory bulb, Cell 79 (1994) 1245–1255. B.D. Rubin, L.C. Katz, Optical imaging of odorant representations in the mammalian olfactory bulb, Neuron 23 (1999) 499–511. N.E. Schoppa, G.L. Westbrook, AMPA autoreceptors drive correlated spiking in olfactory bulb glomeruli, Nat. Neurosci. 11 (2002) 1194–1202. N.E. Schoppa, G.L. Westbrook, Glomerulus-specific synchronization of mitral cells in the olfactory bulb, Neuron 31 (2001) 639–651. M.T. Shipley, M. Ennis, Functional organization of olfactory system, J. Neurobiol. 30 (1996) 123–176.
32
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
[43] E.C. Sobel, D.W. Tank, Timing of odor stimulation does not alter patterning of olfactory bulb unit activity in freely breathing rats, J Neurophysiol. 69 (1993) 1331–1337. [44] H. Spors, A. Grinvald, Spatio-temporal dynamics of odor representations in the mammalian olfactory bulb, Neuron 34 (2002) 301–315. [45] W.B. Stewart, J.S. Kauer, G.M. Shepherd, Functional organization of rat olfactory bulb analyzed by the 2-deoxyglucose method, J. Comp. Neurol. 185 (1979) 715–734. [46] J. Strotmann, S. Conzelmann, A. Beck, P. Feinstein, H. Breer, P. Mombaerts, Local permutations in the glomerular array of the mouse olfactory bulb, J. Neurosci. 20 (2000) 6927–6938. [47] R.M. Sullivan, M. Leon, Early olfactory learning induces an enhanced olfactory bulb response in young rats, Brain Res. 392 (1986) 278–282. [48] A. Tsuboi, S. Yoshihara, N. Yamazaki, H. Kasai, H. Asai-Tsuboi, M. Komatsu, S. Serizawa, T. Ishii, Y. Matsuda, F. Nagawa, H. Sakano, Olfactory neurons expressing closely linked and homologous odor-
[49]
[50]
[51] [52]
[53] [54]
ant receptor genes tend to project their axons to neighboring glomeruli on the olfactory bulb, J. Neurosci. 19 (1999) 8409–8418. N. Uchida, Y.K. Takahashi, M. Tanifuji, K. Mori, Odor maps in the mammalian olfactory bulb: domain organization and odorant structural features, Nat. Neurosci. 10 (2000) 1035–1043. ˜ R. Vassar, S.K. Chao, R. Sitcheran, J.M. Nunez, L.B. Vosshall, R. Axel, Topographic organization of sensory projections to the olfactory bulb, Cell 79 (1994) 981–991. D.A. Wilson, Receptive fields in the rat piriform cortex, Chem. Senses 26 (2001) 577–584. M. Yokoi, K. Mori, S. Nakanishi, Refinement of odor molecule tuning by dendrodendritic synaptic inhibition in the olfactory bulb, Proc. Natl. Acad. Sci. USA 92 (1995) 3371–3375. X. Zhang, S. Firestein, The olfactory receptor gene superfamily of the mouse, Nat. Neurosci. 5 (2002) 124–133. Z. Zou, L.F. Horowitz, J.P. Montmayeur, S. Snapper, L.B. Buck, Genetic tracing reveals a stereotyped sensory map in the olfactory cortex, Nature 414 (2001) 173–179.
Review
Olfactory coding in the mammalian olfactory bulb Michael Leon*, Brett A. Johnson Department of Neurobiology and Behavior, University of California, Room 2205 MH, Irvine, CA 92697 -4550, USA Accepted 21 January 2003
Abstract There have been a number of recent approaches to the study of olfactory coding, each of which has its advantages and disadvantages. In the present review, we discuss our own work on this topic, which has involved mapping uptake of [ 14 C]2-deoxyglucose across the entire glomerular layer of the rat main olfactory bulb in response to systematically selected pure odorant molecules. Our strategy to understand the olfactory code has involved four approaches. In the first, we determined whether the system encodes odorants in their entirety, or whether it encodes odorants by representing combinations of molecular features that add together to comprise a neural picture of each odorant. Multiple odorant features appeared to be coded by multiple receptors. Our second strategy examined the ways that such features are represented. We stimulated rats with odorants that differed greatly in their molecular structure to be able to identify a set of odorant feature response domains. Our third approach asked how odorants with very small differences in molecular structure are coded, and we found systematic differences in the representation of such features within response domains. Finally, we were able to predict odor perception from the neural representations of odorants that differed in only a single aspect of their structure. Using these strategies, we have been able to learn some of the rules by which the olfactory code operates. These rules have allowed us to predict where previously unmapped molecules would be represented and how differences in molecular representations affect olfactory perceptions. 2003 Elsevier Science B.V. All rights reserved. Theme: Sensory systems Topic: Olfactory senses Keywords: Olfactory; Odorant; Glomerulus; Olfactory bulb; Coding; Response domain
Contents 1. Introduction ............................................................................................................................................................................................ 2. Functional organization of the olfactory system ......................................................................................................................................... 3. Are odorants coded by a combination of receptors?.................................................................................................................................... 4. What constitutes a molecular feature? ....................................................................................................................................................... 5. How does the brain code odorants that are very similar in structure? ........................................................................................................... 6. What is the critical molecular feature that produces the chemotopic response? ............................................................................................. 7. Can we predict the neural response from the structure of the molecule? ....................................................................................................... 8. Can we predict odor perception from the neural response to the molecular structure? ................................................................................... 9. Signal vs. noise in the olfactory system ..................................................................................................................................................... 10. Learning and the signal to noise ratio ...................................................................................................................................................... 11. Noise, signal and concentration............................................................................................................................................................... 12. Relational responses to odorants ............................................................................................................................................................. 13. Odorant significance .............................................................................................................................................................................. 14. Temporal aspects of mammalian olfactory coding .................................................................................................................................... 15. Discrimination mechanism .....................................................................................................................................................................
*Corresponding author. Tel.: 11-949-824-5343; fax: 11-949-824-2447. E-mail address: [email protected] (M. Leon). 0165-0173 / 03 / $ – see front matter 2003 Elsevier Science B.V. All rights reserved. doi:10.1016 / S0165-0173(03)00142-5
24 24 24 25 26 26 27 28 29 29 29 29 30 30 30
24
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
16. Summary .............................................................................................................................................................................................. Acknowledgements ...................................................................................................................................................................................... References...................................................................................................................................................................................................
1. Introduction A code is a set of rules by which information is transposed from one form to another. In the case of olfactory coding, it would describe the ways in which information about odorant molecules is transposed into neural responses. We thought that when we understood that code, we might be able to predict the odorant molecule from the neural response, the neural response from the odorant molecule, and the perception of the odorant molecule from the neural response. The following paper reviews our approach to the issue of olfactory coding. Because a critical or comprehensive assessment of other approaches to the study of the olfactory code are not provided, the reader may want to consult one of several recent reviews for a more complete perspective on this area [6,12,16,26–28,51].
2. Functional organization of the olfactory system Olfactory transduction in the rat starts with perhaps a thousand different types of olfactory receptors located on the cilia of a large number of olfactory receptor neurons that comprise the olfactory epithelium [3,53]. Each of these neurons probably expresses a single type of olfactory receptor gene and homologous receptor neurons project to a small number of olfactory glomeruli paired on the medial and lateral aspects of the olfactory bulb (Fig. 1) [31,33,38,46,50]. The glomeruli are dense synaptic bundles in which many homologous olfactory receptor neurons connect to second-order neurons. A small number of mitral cells emanate from each glomerulus and project to a number of regions, including the olfactory cortex. The projections of mitral cells receiving input from homologous olfactory receptor neurons form reliable discrete clusters in different regions of the olfactory cortex [54]. Interneurons mediate inhibition between glomeruli and between mitral cells [42].
3. Are odorants coded by a combination of receptors? There are many different olfactory receptors in rodents [3,53], but there are orders of magnitude more odorants to which they respond [1,7,34]. It is therefore likely that most odorants are coded by a combination of receptors such that a unique combination of responses would describe any particular molecule [37]. One way this could occur is by
30 31 31
simple feature extraction in which different receptors independently recognize different parts of the same odorant molecule. To test this combinatorial, feature-extraction hypothesis, we compared the responses to pairs of odorants that either shared or did not share a molecular feature to which the receptors might respond [24]. Specifically, we exposed rats to one of four esters sharing a straight-chain hydrocarbon domain, with two of these molecules sharing an isoamyl feature and two sharing an ethyl feature. All of these odorants were exposed at the same vapor concentrations to insure that any difference in response was due to the qualitative differences among odorants, rather than to a difference in the number of molecules to which they were exposed. We also exposed odorants at relatively low concentrations to avoid seeing low-affinity binding to receptors. Glomerular responses were assessed across the entire lamina using [ 14 C]2-deoxyglucose autoradiography [24]. We developed a method in which discrete measurements of uptake are taken at systematic angle increments around equally spaced coronal bulb sections. The data from individual sections then is merged into arrays and standardized for differences in bulb size. The anatomically standardized data arrays can be subjected to a variety of transformations, plotted as color-coded contour charts, and compared statistically to test specific hypotheses concerning the impact of odorant chemistry on activity patterns. We previously discussed the relative advantages of this approach, which include the use of an unanesthetized, freely respiring animal and the ability to analyze the entire glomerular layer [21,22]. At the same time, the technique is unable to resolve the temporal dynamics of the olfactory response, and it also is unable to compare responses to multiple stimuli in the same animal. There is no question that different techniques must be used to test different hypotheses regarding olfactory coding. We found that each odorant evoked a unique glomerular pattern that was consistent across rats. The pattern became more complex as the complexity of the molecule increased, suggesting that the larger molecules may have more features that are coded. Focal glomerular responses were also duplicated on the medial and lateral aspects of the bulb in the same anatomical relationship as homologous sensory neuron projections. Most importantly, the two ethyl esters shared a very specific focus of glomerular activity that was not seen for the other two odorants. Similarly, the two isoamyl esters shared a different focus of glomerular activity that was not seen for the other two odorants. These data strongly support the idea that multiple
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
25
Fig. 1. (A) A representation of the convergence of olfactory receptor neurons onto a glomerulus in the olfactory bulb. (B) The medial and lateral paired projections of homologous olfactory receptor neurons shown at both an anterior and a posterior coronal section of the olfactory bulb. (C) The projection of a mitral cell that receives input from a single glomerulus to a pyramidal cell in the olfactory cortex, shown in a sagittal section. AOB, accessory olfactory bulb.
receptors can code odorant molecules as a combination of their constituent features.
4. What constitutes a molecular feature? To determine which characteristics of an odorant can constitute a molecular feature recognizable by odorant
receptors, we began a research program in which panels of odorants that were systematically different in chemical structure were used as stimuli. While we did not know a priori what would constitute a molecular feature that could be recognized by olfactory receptors, it seemed possible that chemical functional groups might constitute such a feature. If a chemical functional group is a coded feature, then exposing rats to molecules that differed only in their
26
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
functional group should differentiate their glomerular responses based on that molecular feature. Therefore, we exposed rats to odorants with the same straight-chain hydrocarbon structure, but with different functional groups: an acid, a ketone, an ester, an alcohol, and an aldehyde. We then compared glomerular activity patterns to determine whether functional groups could constitute molecular features that are recognized by odorant receptors. We found that each odorant stimulated a unique combination of glomerular clusters that we call modules [21]. The modules describe a group of glomeruli in which we reliably find responses to a particular odorant feature. Again, the glomerular module activity was found both medially and laterally for each odorant. Most importantly, odorants with different functional groups stimulated different glomerular modules. These data raised the possibility that olfactory receptors are sensitive to chemical functional groups as part of the combination of molecular features that describe an odorant. Uchida et al. [49], observing a limited glomerular population, subsequently suggested that functional groups play a role in activating two glomerular domains, and Araneda et al. [2] then found that substituting functional groups eliminated the specific response of an identified receptor to otherwise identical odorant ligands. The presence of only one molecular representative of each functional group in this study, however, raised the possibility that the different glomerular responses that we observed were due to the specific structures of the molecules that we tested, rather than specifically to the functional group on these molecules. To test the generality of these conclusions we exposed rats to a wide variety of odorants that shared functional groups, but had very different hydrocarbon structures and numbers of carbons [19]. If molecules with the same functional groups, but very different hydrocarbon configurations stimulated the same glomerular modules, then we could conclude that the functional group itself may be a critical molecular feature to which the receptors specifically responded. We exposed groups of rats to one of 54 odorants with a number of representatives of the different functional groups that we previously described. The clustered, often overlapping responses were circled to indicate the possibility of a response domain (Fig. 2). As before, each odorant could be described by a unique set of glomerular module responses that were duplicated medially and laterally. As predicted, molecules with specific functional groups stimulated those areas that were previously stimulated by molecules with the same functional groups, but with different hydrocarbon configurations. Even the smallest representatives of molecular functional groups had responses in the modules activated by the larger molecules with that functional group. These data provide strong support for the notion that olfactory receptors recognize specific functional groups as molecular features. While the
anterior modules had response domains sensitive to functional groups, we found that the large posterior responses varied with the hydrocarbon structure of the odorant molecules.
5. How does the brain code odorants that are very similar in structure? We noted that responses to different odorants differed in their exact positions within certain modules. We therefore considered the possibility that this variation was due to some systematic difference among the odorants. To address that issue, we varied one aspect of the odorant molecules systematically and then determined the glomerular response pattern in a particular response domain. We tested the response to a homologous series of organic acids in which the number of carbons in the aliphatic chain were systematically incremented [25]. We found that responses to these organic acid odorants were clustered in the paired modules that responded to the acid functional group [25]. The responses within these modules overlapped across odorants, and the focus of activity moved ventrally with additional carbons in the aliphatic structure. Similar response patterns have since been reported for aliphatic aldehydes and alcohols [14,32,39,49]. These data support the idea that certain glomerular responses to such odorants are organized chemotopically and that differences in the responses of neighboring glomeruli may underlie differences in the perception of these very similar odorants.
6. What is the critical molecular feature that produces the chemotopic response? Olfactory receptors cannot actually respond directly to the number of odorant carbons, but they are likely to be differentially responsive to a molecular feature such as hydrophobicity, length, or volume, which covaries with carbon number. Since one of the goals of studying odor coding is to determine the true relationships between stimulus and neural response, we determined which of these properties was critical in producing the systematic changes in their representation. Specifically, we determined whether the change in location of the response focus in the responsive module evoked by acid odorants was due to incremental changes in hydrophobicity, molecular length, or molecular volume. To distinguish among these possibilities, we exposed groups of rats to organic acids that had the same number of carbons, but differed in their hydrocarbon structures, thereby allowing the acids to differ independently in hydrophobicity, length, and volume [22]. We found that the only molecular property that strongly correlated with the location of the activity focus in the responsive module was molecular length, suggesting that this molecular feature is the principal determinant of the
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
27
Fig. 2. Outlined, paired modules have a preferential response for the ketone functional group, regardless of odorant hydrocarbon structure [19].
chemotopic response. Using a different approach, Araneda et al. [2] found that a specific olfactory receptor was also particularly sensitive to odorant molecular length.
7. Can we predict the neural response from the structure of the molecule? Another measure of our understanding the odor code is whether we can predict neural activity in the bulb from the structure of molecules. We noted that molecules without
any oxygen moiety activated more ventral and caudal glomerular regions than were activated by medium-sized odorants containing oxygen atoms [19]. This response pattern presented an opportunity to predict how odorants would be represented in the bulb. Specifically, we predicted that odorant features with no oxygen moiety would also activate caudal and ventral areas of the bulb. To test this prediction, we exposed rats to pinene and santalol, two odorants containing dense hydrocarbon features without oxygen, and found that both stimulated caudal, ventral modules. These data demonstrate that we can predict
28
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
certain novel neural response patterns from odorant molecular features.
8. Can we predict odor perception from the neural response to the molecular structure? If odors are coded using particular patterns of glomerular activity, it should be possible to predict aspects of odor perception based on these activity patterns. If we could find odorants differing in a single molecular feature that induces a simple difference in their glomerular response, then we would predict that such odorants would be discriminated. However, if there were odorants with differences in molecular structure, but without differences in their evoked glomerular activity, then such odorants should not be discriminated. To test this hypothesis, we used pairs of enantiomers (optical isomers), which have the same number of carbons, the same functional groups and the same hydrocarbon structure, except that they are mirror images of each other [29]. The use of such molecules limits the possible basis for discrimination to a single
aspect of the molecules, their stereoconfiguration. We then predicted the difficulty with which odorant pairs could be discriminated, based on the glomerular activity patterns. While the enantiomers of carvone evoked both shared and statistically different modular responses, the modular glomerular activity patterns evoked by the enantiomers of both terpinen-4-ol and limonene were not statistically different from each other [29] (Fig. 3). As predicted, rats spontaneously discriminated between the enantiomers of carvone in a cross-habituation paradigm. Conversely, the enantiomers of both limonene and terpinen-4-ol were not spontaneously discriminated. Therefore, we are able to predict the perception of odorant molecules based on the neural activity that they evoke, which is a step toward understanding the olfactory code. Patterns of glomerular activity measured using our 2-DG mapping technique also predicted quantitative similarities in odor perception across the series of aliphatic acids that differed incrementally in carbon number [25]. Cleland et al. [8] tested the behavioral responses to those same odorants using the cross-habituation assay to assess spontaneous discriminations by rats among a structurally similar
Fig. 3. (Right) Our current model of modular responses in the olfactory bulb. Apparent modular responses to odorants were outlined and superimposed on one another to identify modules used in the representations of multiple odorants. Shown here are those modules that were activated by more than three of the 54 odorants studied. For a great majority of the odorants, whenever a module was identified in the lateral aspect of the bulb, a module of similar activity was detected in the medial bulbar aspect. The corresponding lateral and medial modules are labeled here by using the same letter, lower case for the lateral representation and upper case for the medial representation. (Left) Mean maximal z-score response in each module evoked by the enantiomers of carvone, limonene, and terpinen-4-ol are shown on the left. Asterisks indicate significant differences (P,0.01) between enantiomers in individual modules. The lack of activity in module i for terpinen-4-ol reflects a negative z-score for both enantiomers ([29], used with permission).
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
series of odorants. The relative differences among the spontaneous discriminations was characterized by a behavior dissimilarity index and that number was correlated with a calculated dissimilarity index that characterized the differences across glomerular response patterns to the different odorants that had been reported by Johnson et al. [25]. The correlation coefficient was 0.926, indicating a particularly close relationship between glomerular responses and perception.
29
criminate between each pair of enantiomers, and it took them only somewhat longer to learn to discriminate odorant pairs that were not discriminated spontaneously in the cross-habituation assay [30]. It is also interesting to note that the spread of the small differences across the entire glomerular layer suggest that it would be difficult to make an olfactory bulb lesion large enough to interfere with discriminations that are differentially reinforced.
10. Learning and the signal to noise ratio 9. Signal vs. noise in the olfactory system These data indicate that we can predict the odorant molecule from the neural response, the neural response from the odorant molecule, and the odor perception from the neural response. It may be particularly important to make these predictions successfully in order to have confidence that the responses that we have identified are neural signals. We regard a neural signal as a neural response that carries critical information regarding the sensory stimulus, while neural noise may be coincident with the signal, but not carry such information. The ability to discriminate between noise and signal in the olfactory system may be particularly important because it seems to be a particularly noisy sensory system. All odorants sensed in the environment are experienced against a background of other odorants. All odorants would be expected to have at least some level of contamination. One also should expect interactions between odorants and the chemicals and enzymes in olfactory mucous. In addition, there may be low-affinity responses in the receptor array. Finally, there appear to be numerous secondary and tertiary responses in the olfactory system that may correlate with the signal, but may not carry information about the odorant. Therefore, it would be possible to mistake noise for signal in an analysis of olfactory responses if we did not establish through these successful predictions that we are studying the signal. One example of neural noise can be seen in an alternate analysis of the glomerular responses to limonene and terpinen-4-ol. Recall that these odorants did not differ when we assessed their major modular responses. However, when we compared these enantiomers for activity at every point in our data array, we noted many small regions of potential difference between them [30]. We can regard these differences as noise because the rats ignored them when asked to make a spontaneous discrimination between enantiomers [29]. At the same time, we reasoned that rats might be able to use the small differences in glomerular response to make discriminations between these odorants if it became important for them to do so. In essence, they may be able to turn noise into a signal if differential responses were reinforced, rather than being spontaneous. Indeed, we found that differential reinforcement induced rats to dis-
We have also shown that reinforced learning in young animals can alter the focal glomerular response by enhancing it [9,20,23,47]. The resulting increase in signal-to-noise ratio in the glomerular layer would be expected to increase olfactory acuity, and that is exactly what Fletcher and Wilson [13] have reported for adult rats.
11. Noise, signal and concentration There is another means by which noise can become signal. A small proportion of odorants change their olfactory perception with increasing concentration [1,11]. Pentanal recruits new glomerular foci with increasing concentration [21], and humans report that this odorant changes in perception from an ethereal odor to a strong, noxious odor. Such a change in perception would be expected with the excitation of new modules, as the background noise of the contaminants becomes a signal with increasing concentration. Indeed, the likely contaminant in this case is the oxidation product of pentanal, which would be pentanoic acid, and the addition of the strong noxious odor of pentanoic acid added to the ethereal scent of pentanal would account for the change in olfactory perception that has been noted.
12. Relational responses to odorants Most odorants are perceived as having the same qualitative odor across a wide range of concentrations. But while perception remains constant, the absolute number of activated glomeruli in any particular focus increases in number with increasing concentration. Because homologous olfactory receptor neurons with closely related olfactory receptor genes project to neighboring glomeruli [10,48], normal ligand / receptor relationships would predict such a broadening of the local response within a response domain with increasing odorant concentration [15,45]. At the same time, one would expect the olfactory perception to change if there were new glomeruli stimulated by the same odorant when exposed at a high concentration. However, when the glomerular responses are reported as relative to background (noise) responses, in
30
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
our case as z scores, the pattern of response remains constant over the range of concentrations that we presented [21]. These data suggest that upstream neurons may respond to the relative responses emanating from the bulb, a mechanism that would help to filter responses to noise, particularly as concentrations increase. One can imagine, for example, that the olfactory bulb projection areas may respond to glomerular activity only in relation to the response of one aspect of the olfactory system that may provide odorant concentration information.
13. Odorant significance It is possible that rats have evolved larger responses to some odorant features because those responses characterize odorants that have special significance. Indeed, when we tested the response of rats to two natural odorants that are produced by other rats [17], pentanoic acid and methylbutyric acid, we found large, continuous response foci in the glomerular layer [25]. At the same time, we found that two other acids that are never found in nature evoked a number of small discontinuous responses in the lamina [25].
14. Temporal aspects of mammalian olfactory coding In order for combinatorial coding to function in the olfactory system, there almost certainly must be a critical role for a temporal mechanism. If the olfactory receptor population essentially breaks up odorants into their constituent features for purposes of coding, the information regarding these features must be reassembled in the olfactory cortex [51,54]. It would make sense for the cortex to regard only those responses from the bulb that arrive synchronously to be part of the combination of features that describe a particular molecule. Schoppa and Westbrook [40,41] have shown that mitral cells emerging from a single glomerulus in rats fire synchronously, thereby providing the basis for a synchronous stimulus arriving at the olfactory cortex for individual odorant features. The synchronous response emanating from each glomerulus would also be expected to serve as an amplification mechanism for each type of olfactory receptor neuron [7]. The synchrony of the responses to multiple features is probably driven by respiration in the rat. Indeed, the olfactory receptor neuron, glomerular and mitral cell responses are focused on the periodic inspiration peak in the respiratory cycle [4,5,36,43,44], thereby presumably driving all odorant-related responses to arrive in the cortex simultaneously. Such a system would provide a temporal filtering device that would eliminate responses arriving outside that synchronized temporal window as neural noise.
15. Discrimination mechanism The specificity of the glomerular response appears to be preserved in the mitral cells. We found that the spatial patterns of activity in the external plexiform layer and granule cell layer, which are secondary to activation of mitral cells, closely resemble patterns in the glomerular layer [25]. Just as we found for aliphatic acids in the glomerular layer of the rat olfactory bulb, mitral cells in the dorsomedial rabbit bulb have also been found to be highly tuned in their response to such odorants within a restricted range of carbon number [18,35]. The chemotopic arrangement of glomeruli and mitral cells responding to odorants of most similar carbon number may ensure that tuning suppresses responses to the most similar odorants [25]. Lateral inhibition between mitral cells apparently associated with neighboring glomeruli seems to underlie odorant tuning in the bulb [52]. The interactions in the bulb therefore appear to amplify and sharpen responses to the olfactory stimulus, thereby facilitating the ability of the olfactory cortex to resolve differences among odorants.
16. Summary Our data point to a number of coding strategies that appear to be needed to encode the large number of olfactory stimuli that vary in numerous ways across a variety of dimensions and parameters. First, most odorants appear to be coded by a combination of their molecular features. Second, the molecular features appear to use an identity code in which specific neurons transmit specific information within the system. Third, the system seems to be capable of using spatial relationships among odorants to facilitate its ability to encode information regarding closely related odorants. Fourth, there appears to be a dynamic code, in which noise can be converted to signal by increasing concentration or with differential reinforcement. Fifth, there may be a relational code in which glomerular responses are considered in relation to response across the glomerular layer. Finally, there appears to be a significance code that has assigned differential numbers of glomeruli to be responsive to specific molecules that may be particularly important to that species. To provide a first approximation of the olfactory code, we have chosen to make the assumption that the olfactory system uses a relatively straightforward strategy for coding odorant molecules. That is, we have assumed that olfactory coding involves the use of different olfactory receptors that respond to different molecular features of odorants. We have obtained significant support for this perspective. It is possible that there are emergent responses within the system, or that there are network properties that code information by using variations in responses across the entire system, or that the glomerular responses are shaped by centrifugal input, or that there are temporal dynamics
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
that hold the key to olfactory quality. However, there is no comparable body of data at this time to support such complex hypotheses that has the same level of predictive power as the model that emerges from our simple approach. At the same time, we have just begun to examine the extraordinarily wide range of odorant molecules and it is certainly possible that there are additional coding mechanisms that are involved in producing the perceptions of molecules that have yet to be studied.
[18]
[19]
[20]
[21]
Acknowledgements This work was supported by DC03545.
[22] [23]
References
[24]
[1] S. Arctander, Perfume and Flavor Chemicals (Aroma Chemicals), Allured Publishing Company, Carol Stream, IL, 1994. [2] R.C. Araneda, A.D. Kini, S. Firestein, The molecular receptive range of an odorant receptor, Nat. Neurosci. 3 (2000) 1248–1255. [3] L.B. Buck, R. Axel, A novel multigene family may encode odorant receptors: A molecular basis for odor recognition, Cell 65 (1991) 175–187. [4] M.A. Chaput, EOG responses in anesthetized freely breathing rats, Chem. Senses 25 (2000) 695–701. [5] M. Chalansonnet, M.A. Chaput, Olfactory bulb output cell temporal response patterns to increasing odor concentrations in freely breathing rats, Chem. Senses 23 (1998) 1–9. [6] T.A. Christensen, J.G. Hildebrand, Pheromonal and host-odor processing in the insect antennal lobe: how different?, Curr. Opin. Neurobiol. 12 (2002) 393–399. [7] T.A. Cleland, C. Linster, How synchronization properties among second-order sensory neurons can mediate stimulus salience, Behav. Neurosci. 116 (2002) 212–221. [8] T.A. Cleland, A. Morse, E.L. Yue, C. Linster, Behavioral models for odor similarity, Behav. Neurosci. 116 (2002) 222–231. [9] R. Coopersmith, M. Leon, Enhanced neural response to familiar olfactory cues, Science 225 (1984) 849–851. [10] S. Conzelmann, D. Malun, H. Breer, J. Strotmann, Brain targeting and glomerulus formation of two olfactory neuron populations expressing related receptor types, Eur. J. Neurosci. 14 (2001) 1623–1632. [11] A. Dravnieks, Atlas of Odor Character Profiles. ASTM Data Series DS 61, Philadelphia, 1985. [12] R.W. Friedrich, Real time odor representations, Trends Neurosci. 25 (2002) 487–489. [13] M.L. Fletcher, D.A. Wilson, Experience modifies olfactory acuity: acetylcholine-dependent learning decreases behavioral generalization between similar odorants, J. Neurosci. 22 (2002) RC201. [14] H.U. Fried, S.H. Fuss, S.I. Korsching, Selective imaging of presynaptic activity in the mouse olfactory bulb shows concentration and structure dependence of odor responses in identified glomeruli, Proc. Natl. Acad. Sci. USA 99 (2002) 3222–3227. [15] K.M. Guthrie, C. Gall, Functional mapping of odor-activated neurons in the olfactory bulb, Chem. Senses 20 (1995) 271–282. [16] L.B. Haberly, Parallel-distributed processing in olfactory cortex: new insights from morphological and physiological analysis of neuronal circuitry, Chem. Senses 26 (2001) 551–576. [17] T. Høverstad, T. Midtvedt, T. Bøhmer, Short-chain fatty acids in intestinal content of germfree mice monocontaminated with Es-
[25]
[26] [27] [28] [29]
[30]
[31] [32] [33]
[34] [35]
[36]
[37] [38]
[39] [40]
[41] [42]
31
cherichia coli or Clostridium difficile, Scand. J. Gastroenterol. 20 (1985) 373–380. K. Imamura, N. Mataga, K. Mori, Coding of odor molecules by mitral / tufted cells in rabbit olfactory bulb I. Aliphatic compounds, J. Neurophysiol. 68 (1992) 1986–2002. B.A. Johnson, S. Ho, Z. Xu, J.S. Yihan, S. Yip, E.E. Hingco, M. Leon, Functional mapping of the rat olfactory bulb using diverse odorants reveals modular responses to functional groups and hydrocarbon structural features, J. Comp. Neurol. 449 (2002) 180–194. B.A. Johnson, M. Leon, Spatial distribution of [ 14 C] 2-deoxyglucose uptake in the glomerular layer of the rat olfactory bulb following early olfactory preference learning, J. Comp. Neurol. 376 (1996) 557–566. B.A. Johnson, M. Leon, Modular glomerular representations of odorants in the rat olfactory bulb and the effects of stimulus concentration, J. Comp. Neurol. 422 (2000) 496–509. B.A. Johnson, M. Leon, Odorant molecular length: One aspect of the olfactory code, J. Comp. Neurol. 426 (2000) 330–338. B.A. Johnson, C.C. Woo, H. Duong, V. Nguyen, M. Leon, A learned odor evokes an enhanced Fos-like glomerular response in the olfactory bulb of young rats, Brain Res. 699 (1995) 192–200. B.A. Johnson, C.C. Woo, M. Leon, Spatial coding of odorant features in the glomerular layer of the rat olfactory bulb, J. Comp. Neurol. 393 (1998) 457–471. B.A. Johnson, C.C. Woo, E.E. Hingco, K.L. Pham, M. Leon, Multidimensional chemotopic responses to n-aliphatic acid odorants in the rat olfactory bulb, J. Comp. Neurol. 409 (1999) 529–548. J.S. Kauer, On the scents of smell in the salamander, Nature 417 (2002) 336–342. S. Korsching, Olfactory maps and odor images, Curr. Opin. Neurobiol. 12 (2002) 387–392. G. Laurent, Olfactory network dynamics and the coding of multidimensional signals, Nat. Neurosci. Rev. 3 (2002) 884–895. C. Linster, B.A. Johnson, A. Morse, E. Yue, Z. Xu, E.E. Hingco, Y. Choi, M. Choi, A. Messiha, M. Leon, Perceptual correlates of neural representations evoked by odorant enantiomers, J. Neurosci. 21 (2001) 9837–9843. C. Linster, B.A. Johnson, A. Morse, E. Yue, M. Leon, Spontaneous vs. reinforced olfactory discriminations, J. Neurosci. 22 (2002) 6842–6845. B. Malnic, J. Hirono, T. Sato, L. Buck, Combinatorial receptor codes for odors, Cell 96 (1999) 713–723. ¨ M. Meister, T. Bonhoeffer, Tuning and topography in an odor map on the rat olfactory bulb, J. Neurosci. 21 (2001) 1351–1360. P. Mombaerts, F. Wang, C. Dulac, S.K. Chao, A. Nemes, M. Mendelsohn, J. Edmonson, R. Axel, Visualizing an olfactory sensory map, Cell 87 (1996) 675–686. R.W. Moncrieff, The Chemical Senses, Leonard Hill, London, 1967. K. Mori, N. Mataga, K. Imamura, Differential specificities of single mitral cells in rabbit olfactory bulb for a homologous series of fatty acid odor molecules, J. Neurophysiol. 67 (1992) 786–789. B.D. Philpot, E.M. Lyders, P.C. Brunjes, The NMDA receptor participates in respiration-related mitral cell synchrony, Exp. Brain Res. 118 (1998) 205–209. E.H. Polak, Multiple profile-multiple receptor site model for vertebrate olfaction, J. Theor. Biol. 40 (1973) 469–484. K.J. Ressler, S.L. Sullivan, L.B. Buck, Information coding in the olfactory system: Evidence for a stereotyped and highly organized epitope map in the olfactory bulb, Cell 79 (1994) 1245–1255. B.D. Rubin, L.C. Katz, Optical imaging of odorant representations in the mammalian olfactory bulb, Neuron 23 (1999) 499–511. N.E. Schoppa, G.L. Westbrook, AMPA autoreceptors drive correlated spiking in olfactory bulb glomeruli, Nat. Neurosci. 11 (2002) 1194–1202. N.E. Schoppa, G.L. Westbrook, Glomerulus-specific synchronization of mitral cells in the olfactory bulb, Neuron 31 (2001) 639–651. M.T. Shipley, M. Ennis, Functional organization of olfactory system, J. Neurobiol. 30 (1996) 123–176.
32
M. Leon, B. A. Johnson / Brain Research Reviews 42 (2003) 23–32
[43] E.C. Sobel, D.W. Tank, Timing of odor stimulation does not alter patterning of olfactory bulb unit activity in freely breathing rats, J Neurophysiol. 69 (1993) 1331–1337. [44] H. Spors, A. Grinvald, Spatio-temporal dynamics of odor representations in the mammalian olfactory bulb, Neuron 34 (2002) 301–315. [45] W.B. Stewart, J.S. Kauer, G.M. Shepherd, Functional organization of rat olfactory bulb analyzed by the 2-deoxyglucose method, J. Comp. Neurol. 185 (1979) 715–734. [46] J. Strotmann, S. Conzelmann, A. Beck, P. Feinstein, H. Breer, P. Mombaerts, Local permutations in the glomerular array of the mouse olfactory bulb, J. Neurosci. 20 (2000) 6927–6938. [47] R.M. Sullivan, M. Leon, Early olfactory learning induces an enhanced olfactory bulb response in young rats, Brain Res. 392 (1986) 278–282. [48] A. Tsuboi, S. Yoshihara, N. Yamazaki, H. Kasai, H. Asai-Tsuboi, M. Komatsu, S. Serizawa, T. Ishii, Y. Matsuda, F. Nagawa, H. Sakano, Olfactory neurons expressing closely linked and homologous odor-
[49]
[50]
[51] [52]
[53] [54]
ant receptor genes tend to project their axons to neighboring glomeruli on the olfactory bulb, J. Neurosci. 19 (1999) 8409–8418. N. Uchida, Y.K. Takahashi, M. Tanifuji, K. Mori, Odor maps in the mammalian olfactory bulb: domain organization and odorant structural features, Nat. Neurosci. 10 (2000) 1035–1043. ˜ R. Vassar, S.K. Chao, R. Sitcheran, J.M. Nunez, L.B. Vosshall, R. Axel, Topographic organization of sensory projections to the olfactory bulb, Cell 79 (1994) 981–991. D.A. Wilson, Receptive fields in the rat piriform cortex, Chem. Senses 26 (2001) 577–584. M. Yokoi, K. Mori, S. Nakanishi, Refinement of odor molecule tuning by dendrodendritic synaptic inhibition in the olfactory bulb, Proc. Natl. Acad. Sci. USA 92 (1995) 3371–3375. X. Zhang, S. Firestein, The olfactory receptor gene superfamily of the mouse, Nat. Neurosci. 5 (2002) 124–133. Z. Zou, L.F. Horowitz, J.P. Montmayeur, S. Snapper, L.B. Buck, Genetic tracing reveals a stereotyped sensory map in the olfactory cortex, Nature 414 (2001) 173–179.