The perception of odors requires the recognition of odorants in the periphery and more central mechanisms in the brain that process this information to allow for the discrimination of odors. The initial step in the perception of odors requires the interaction of odor molecules with receptors on sensory neurons. Individual sensory neurons in mice express only 1 of 1,500 different receptor genes. An odorant interacts with multiple distinct receptors, resulting in the activation of a unique ensemble of neurons distributed across the olfactory sensory epithelium. Discrimination among odors then requires that the brain determine which of the sensory neurons have been activated by a given odorant.
Neurons expressing a given receptor, although randomly distributed within zones of the epithelium, project with precision to two spatially invariant loci, the glomeruli in the olfactory bulb. As in other sensory systems, this pattern of projections in the bulb provides a map in the brain of receptor activation in the periphery. Imaging experiments reveal that different odors elicit different patterns of glomerular activity, providing evidence for a functional map in the olfactory bulb. Thus, a striking transformation in the representation of olfactory information is apparent in the bulb where the randomly distributed population of active neurons in the sensory epithelium is consolidated into a discrete spatial map of glomerular activity.
How is this representation transformed at the next processing center for olfactory information in the brain, the piriform cortex? The mitral cells, the output neurons of the bulb, extend an apical dendrite into a single glomerulus and project axons to several telencephalic areas with significant input onto pyramidal neurons of the piriform cortex. Pyramidal neurons are likely to integrate convergent mitral cell input from multiple glomeruli onto individual pyramidal cells. This model of convergence could transform the insular, segregated patterns of activity in the bulb into a more distributive representation. Neuronal activity in piriform cortex would no longer reflect the activation of a single receptor but the coordinated activity of the multiple receptors activated by a single odorant. Thus the convergence of like axons on individual glomeruli that characterizes the representation of olfactory information in the bulb may be transformed by the convergence of unlike axons from distinct glomeruli on single pyramidal cells.
Odor-Evoked Responses in Piriform Cortex
We have examined the representation of odors in the piriform cortex by optical imaging of neural activity. We developed a preparation that permits the recording of odor-evoked changes in fluorescent intensity across large populations of neurons in the piriform cortex by two-photon microscopy. The piriform cortex is a three-layered structure that resides on the ventrolateral surface of the cerebral hemispheres. The major projection neurons of the piriform cortex, the pyramidal cells, maintain their cell bodies in layers 2 and 3 and extend apical dendrites to layer 1 where they synapse with mitral cell afferents from the olfactory bulb.
We loaded pyramidal cells with a calcium-sensitive fluorescent dye by injecting dye into broad regions of layer 1. Optical recordings across large populations of cortical neurons demonstrate that at natural concentrations, exposure to six different odors elicits fluorescence changes in 310 percent of the pyramidal cells. The frequency of responding cells differed for different odors but remained consistent across multiple fields, as well as in different animals. Octanal, for example, activates 10 percent of the layer 2 neurons, whereas butyric acid consistently elicits fluorescent changes in fewer neurons, activating only 3 percent of the pyramidal cells. Cells responsive to a given odor within the same imaging field exhibited a range of statistically significant positive responses with fluorescence changes (ΔF/F) varying between 2 and 50 percent above baseline. Odor-evoked responses across a field of 300 cells were consistent across multiple trials. Moreover, the responses were eliminated by blockade with glutamate antagonists, suggesting that the responses we observed are synaptically mediated.
Different Odors Activate Unique Ensembles of Neurons
We observe that each odor activates a population of from 3 to 10 percent of the neurons in layer 2 of the piriform cortex. Cells responsive to a given odorant within a single field are dispersed rather than clustered. As a consequence, cells responding to different odorants were intermingled. We can discern no apparent spatial clustering within this restricted imaging field.
Analysis of the population of cells responding to a given odor reveals that each odor activates a distinct ensemble of cortical neurons but individual neurons within the ensemble can respond to different odors. For example, about 10 percent of octanal-responsive cells also respond to pinene, butyric acid, or ethylbutyrate. The representation of structurally similar odors, octanal and hexanal, exhibits somewhat greater overlap. This is, however, far less overlap than is observed for structurally related aldehydes in both sensory neurons and glomeruli. The piriform may therefore function to decorrelate similar but nonidentical inputs. This decorrelation could enable fine discrimination between odorants that elicit highly overlapping patterns of activity in both the sensory epithelium and bulb. The observation that a given neuron in piriform cortex will respond to multiple, distinct odors implies that the quality of an odor cannot be determined by the activity of an individual neuron. Rather, it is likely to be represented by an ensemble of active neurons that is unique to an individual odor.
The Spatial Distribution of Cortical Neurons Responsive to a Given Odor
A given odor activates a spatially stereotyped population of glomeruli in the olfactory bulb. Superimposed upon this invariant insular map is a chemotopy such that structurally similar odorants activate glomeruli in circumscribed regions of the olfactory bulb. We have imaged across multiple overlapping fields in both the anterior and posterior piriform cortex to determine whether different odors might activate spatially segregated populations of cortical neurons. Using the middle cerebral artery and the lateral olfactory tract as anatomic landmarks, we targeted two-photon imaging to similar regions of piriform cortex in different individuals. Imaging was performed with six odorants across three to five overlapping fields of 300 cells in multiple animals. The pattern of responsive cells encompasses the entire imaged fields in piriform cortex without evidence for segregation or clustering. We observed similar distributive representations for both structurally distinct and structurally similar odorants. The density of responsive cells in these representations of a given odorant varied two- to threefold across spatial scales of hundreds of microns, but autocorrelation analysis revealed no systematic periodic patterning. Thus, the insular, spatially segregated pattern of active glomeruli and the associated chemotopy observed in the olfactory bulb were not apparent in the odor representations in piriform cortex.
One clear implication of these observations is that the piriform does not reveal a pattern of neural activity in which one or a few cells respond to a given odorant with great specificity, the equivalent of olfactory "grandmother cells." Rather, the specificity of an odor appears to be represented by a unique ensemble of active neurons. Neurons responsive to a given odor are distributed throughout the piriform without apparent spatial segregation or spatial preference. Clustering of neurons with similar responsivity is not observed, so that cells with distinct olfactory receptive fields are interspersed. This distributive, discontinuous character of patterns of piriform responses differs from those of the cortical areas serving other sensory modalities where cells responding to similar features of a sensory image are often clustered. Rather, individual piriform cells themselves possess discontinuous receptive fields, and we observed no similarity in response properties among neighbors.
This work is supported in part by a grant from the Mathers Foundation.
As of May 30, 2012