Functional units in the olfactory system
At each stage of odor coding, the olfactory system is divided into anatomical subdivisions that seem to serve distinct functions. Although the olfactory (piriform) cortex has long been known to have anterior and posterior subdivisions with different local architectures (1), the function of these anatomical units has been poorly understood. In this issue of PNAS, Kadohisa and Wilson (2) report that anterior and posterior piriform cortices are modified differently by olfactory experience, suggesting that the principles of functional domain organization in the olfactory system extend to higher levels of processing, where the subdivisions may have separate roles involving odorant discrimination and odor generalization.
Each olfactory sensory neuron in the olfactory epithelium expresses a single type of odorant receptor that binds odorants on the basis of their molecular features (3, 4). Different types of receptors are expressed in different parts of the nose, some in separate organs (5), or in different compartments or zones of the main olfactory epithelium (6). In the main epithelium, most odorant receptor expression zones may be functional anatomical units, because they are organized orthogonally to airflow, establishing an interaction between chromatographic separation of odorants across the nasal mucosa and receptor specificity to establish the unique activity patterns across the epithelium that are evoked by different odorant molecules (6).
There are specialized peripheral organs in the nose that project to separate small sets of olfactory bulb glomeruli (5), while the zones of the olfactory epithelium project to corresponding zones within the glomerular layer of the bulb (ref. 6 and Fig. 1). The axons of sensory neurons expressing the same odorant receptor converge into glomeruli, which are organized into modular clusters that respond differentially to aspects of shared odorant chemistry, such as functional groups, hydrocarbon structure, and/or chemical properties, that are determined by the whole molecule (7). Glomerular modules in the ventral half of the bulb have responses organized such that larger molecules activate more ventral glomeruli, a pattern that probably reflects chromatographic separation of odorants in the epithelium (8).
Molecular features in odorant chemicals such as geranyl acetate bind to odorant receptors expressed in zones of the olfactory epithelium. The sensory neurons that express the same odorant receptor gene converge into glomeruli of the olfactory bulb to produce patterns of major response foci such as those shown here in a 3D rendering of 2-deoxyglucose uptake evoked by geranyl acetate (only the medial surfaced is shown). Mitral cells associated with these glomeruli project in diffuse but patchy patterns onto both the anterior and the posterior piriform cortices. As shown by Kadohisa and Wilson (2), odor experience causes responses in the anterior piriform cortex to become more narrowly tuned to particular odorants, perhaps supporting the type of odorant discrimination responsible for the specific rose odor of geranyl acetate, whereas responses in the posterior piriform cortex become more broadly tuned, perhaps supporting the type of odor classification responsible for the general floral odor shared by geranyl acetate and odorants of distinct chemical structure.
These anatomical units exist in the olfactory system, but have they been shown to have critical functions in olfactory coding? Although early bulb ablation studies, most of which were not guided by knowledge of actual odorant responses, questioned the relevance of domain organization in the olfactory bulb, more recent interventive experiments have confirmed the critical involvement of spatially restricted regions in the perception of particular odorants. Pharmacological blockade of dorsal bulbar activity interferes with experience-dependent modification of responses to odorants whose activity involves those dorsal regions, but it does not affect responses to odorants activating only the ventral part of the bulb (9). Chemical ablation of particular domains in the epithelium diminishes responses to odorants expected to be represented in those regions but does not diminish responses to odorants expected to activate intact regions (10). Finally, antibodies to a receptor activated by octanal reduce both behavioral and focal neuronal responses to that odorant.† Certainly, there is now very clear evidence for critical functions of anatomical domains in the olfactory system, and this evidence casts doubt on alternative hypotheses of odor coding that exclusively involve temporal coding or hypotheses that depend on an extreme distribution of responses.
The mitral cells that report activity to the piriform cortex each receive input from only one glomerulus but then amplify, sharpen, and filter the signal before sending it to the piriform (11, 12, ‡). Mitral cells project to a broad set of higher brain regions that have rarely been investigated for their function but that are likely to serve distinct roles in odor-directed behavior. The largest projection is to the anterior and posterior piriform cortices, which have distinct cytoarchitecture and connectivity (1). Mitral cells receiving primary input from the same odorant receptors project in a patchy pattern across the piriform cortex (13), and the cells that are activated by different odorant receptors project to different areas in the anterior piriform cortex, such that different odorants elicit different activity patterns in that structure (14). The responses to different chemical features are thereby thought to be brought back together in the piriform cortex for the perception of odor identity (14).
Wilson and Stevenson (15) have pointed out that odor perception is likely to be more complex than the simple, faithful representation of external odorants in the brain. The final perception of odor that is represented in the brain should involve an interaction between sensory information being relayed from the periphery and some central neural representation of what already has been learned about various odors (15). Thus, an experienced cortex may respond differently to the same odorant than would a naive cortex. In testing for such experience-dependent changes, Kadohisa and Wilson (2) found that pyramidal neurons in the anterior piriform cortex do indeed become more highly tuned to an odorant mixture when rats are trained to distinguish it from its components. They further found that equivalent neurons in the posterior piriform cortex actually become more broadly tuned after the same experience, thus revealing that different parts of the olfactory cortex are differently modified by the same experience (2). Interestingly, another recent report (16) using functional MRI (fMRI) and odorant cross-habituation showed that the anterior piriform cortex in humans seems to classify odorants according to chemical similarities, whereas the posterior piriform cortex classifies odorants according to overall similarities in perceived odors (ref. 16 and Fig. 1). Together, these findings suggest the possibility that the anterior piriform cortex may function in discriminating closely related odorants, for example, distinguishing geranyl acetate (rose) from methyl anthranilate (orange blossom), whereas the posterior piriform cortex may function to represent perceptual similarities (floral).
There is now very clear evidence for critical functions of anatomical domains in the olfactory system.
As is the case for most important scientific findings, the article by Kadohisa and Wilson (2) raises a number of questions. Would different reinforcement contingencies during the odorant experience lead to a different type of change in either cortex? Do the changes occur equally for every set of odorants that might be experienced, or are they specifically relevant to the mixtures and components that were tested in the study? Do the piriform changes affect only the odorants that were actually experienced, or is there some transference to odorants of similar chemistry or similar perceived odor? What is the mechanism by which these differences arise? Are distinct bulbar domains involved in producing the different types of cortical responses? What are the differences in the projections of mitral cells to the anterior and posterior pirifom that could underlie these differences in perception? Will there be new anatomical areas identified that underlie other parallel-processed aspects of perception (17)? Would pharmacological interventions or physical lesions affecting the anterior piriform cortex affect discrimination learning without affecting the learning of perceptual categories? Answers to such questions are likely to be found in the coming years.
Recent neural network approaches seem to have addressed one of these questions. These analyses have been applied to determine which aspects of rat glomerular layer activity patterns are used by the piriform to predict either odorant chemical features or human odor descriptors, with the unexpected result that the parts of the pattern associated with the chemistry of a given odorant are often adjacent to, but rarely overlap with, the parts of the pattern predicting the perceived odor of the compound.§. Thus, the anterior piriform cortex may be attending to different sets of activated mitral cells (those predicting odorant chemical features) than the posterior piriform cortex (which might be attending to mitral cells predicting perceived odor). These findings may provide one aspect of the mechanism underlying the functional distinctions in the piriform.
Footnotes
- *To whom correspondence should be addressed. E-mail: mleon{at}uci.edu
-
Author contributions: M.L. and B.J. wrote the paper.
-
The authors declare no conflict of interest.
-
See companion article on page 15206.
-
↵ † Deutsch, S., Apfelbach, R. (2006) Abstr. Assoc. Chemoreception Sci. Annu. Meeting 28:261 (abstr.).
-
↵ ‡ Cleland, T., Johnson, B., Leon, M., Linster, C. (2006) Abstr. Assoc. Chemoreception Sci. Annu. Meeting 28:75 (abstr.).
-
↵ § Madany Mamlouk, A., Teehankee, A., Schuh, E., Martinetz, T., Leon, M., Johnson, B. A. (2006) Abstr. Eur. Chemoreception Res. Org. Meeting 17:234 (abstr.).
- © 2006 by The National Academy of Sciences of the USA
