January 19th, 2009
(L to R) Venki N. Murthy, Antoniu L. Fantana, Edward R. Soucy, Markus Meister, and Dinu F. Albeanu
Our brains generally represent the outside world using ordered maps. For example, neighboring points in visual space activate neighboring points on the retina, and this relation is preserved through several subsequent stages of processing. This sort of systematic projection of an external variable onto the brain's surface is found in most sensory systems - vision, audition and touch. There is even a plausible theory behind it: Most brain computation is local, relying on short connections between nearby cells. This is necessary because the connections between neurons occupy most of the volume available to the brain, and long-distance connections require more of this volume. In this way, the arrangement of sensory information in a given brain region, namely the sensory map, can reveal something about which neuronal signals are combined in subsequent circuits.
In the case of vision, touch, and hearing, we have some intuition about which sensory stimuli should be mapped close to each other. This is much less obvious for olfaction. Odorant molecules are diverse, and differ along many independent parameters such as size, charge-distribution, bond-saturation and three-dimensional structure. Mice and rats use about 1,000 odorant-receptor types to probe this space of chemicals. Receptor neurons in the nose project their axons to the brain and form anatomical units called glomeruli, arrayed in a layer on the surface of the olfactory bulb. Each glomerulus receives input from a single receptor type, whose ligand-binding properties determine the spectrum of odor responses. In this way, the layout of glomeruli on the bulb forms a two-dimensional map of odors.
In this study, we set out to chart this odor map. First we asked how precise the map is. We recorded the odor responses of glomeruli on the dorsal bulb in rats and mice to a diverse battery of several hundred odors. Many of the glomeruli could be identified uniquely by their odor response spectrum, which allowed us to recognize the same glomerulus across animals. In this way we found that the layout of the map is precisely reproducible: Across individuals the position of a given glomerulus varies by only 1 glomerular spacing. Compared to the size of the map, this represents a remarkable developmental precision of 1 part in 1000.
Second, we asked whether the layout of glomeruli is systematically related to their odor sensitivities. By analogy to the other sensory modalities, one might expect that nearby glomeruli should have similar odor sensitivities. We were surprised to find this was not the case: The odor response spectra of two neighboring glomeruli were as dissimilar as those of distant glomeruli. This seemingly haphazard layout of sensory properties is a marked distinction from other brain maps, and one might have speculated that it results from developmental errors in the targeting of axons. However, given the remarkable precision of targeting we observed, the local diversity of glomeruli is a reliable feature rather than a random flaw of the odor map. How and what the brain computes with such a fractured map remains to be explored.
Read more in Nature Neuroscience
Read more in Harvard University Gazette Online