While many of these neurons respond equally well to both male and female urine, a substantial subpopulation is activated exclusively by male urine. Using gas chromatography and mass spectroscopy, we discovered that a novel sulfur-containing compound mediates this specificity. A synthetic version of this chemical makes urine more attractive to female mice, providing a direct link between the chemical specificity of a group of neurons and behavioral output.

The use of natural stimuli, like urine, also has general implications for how olfactory information is coded in the brain. In contrast to prevailing views of mitral cells as only moderately selective for chemical stimuli, we find that most cells are extremely selective for individual odorants. We have extended the work with natural stimuli by engineering a “smellbot,” which can deliver hundreds of odorant stimuli, drawn from a large swath of chemical and perceptual space, within a short period of time. Even when stimulated with these large, arbitrary sets of odorants, neurons in the olfactory bulb show responses to only a small percent of stimuli. Thus, the output from the bulb appears to convey highly specific information about the nature of the olfactory stimulus, resembling the “labeled lines” that have been observed in coding of chemical stimuli in the taste system (see the abstract of Charles Zuker [HHMI, University of California, San Diego]).

Multiphoton Imaging of Olfactory Learning
An extraordinary feature of the olfactory system is its intimate and strong relationship to long-term memories. Rodents, like humans, form exceptionally durable memories based on odors and their associated meanings. How can such memories persist for days, months, even years? One long-standing hypothesis is that such long-term memories rely on significant changes in the structure and connectivity of neuronal axons and dendrites. We have used two-photon microscopy in living animals to determine whether persistent structural changes in neural circuits are required for long-term memory storage. A novel surgical approach allowed us to visualize dendritic spines—the sites of excitatory synapses—in the mammalian hippocampus, a previously inaccessible structure long linked to learning and memory. In other brain regions, such as the neocortex, spines observed over weeks come and go with considerable frequency, and their permanence is modulated by levels of neuronal activity (see also the abstract by Karel Svoboda [HHMI, Cold Spring Harbor Laboratory]). In the hippocampus, over the short term, we found that the number of spines remained constant, even in the face of massive neuronal activity induced by epileptic seizures. The ability to visualize the location and plasticity of spines in this structure, along with their component molecules, will be critical for uncovering the links between structural changes and learning.

The olfactory bulb is unusual in that certain populations of cells, especially granule cells, undergo adult neurogenesis, meaning that new neurons are constantly being inserted into the existing circuitry. Taking advantage of both viral vectors and mouse lines engineered to express a fluorescent marker (green fluorescent protein) in specific groups of neurons in the olfactory circuit, we have followed for the first time the dynamics of the insertion, development, and loss of new neurons in the living animal, using in vivo two-photon imaging. There are suggestions that enhanced activity or acquisition of a new task may lead to enhanced survival of these newly generated neurons, and we are testing this by monitoring neuronal differentiation under different levels of olfactory experience.

Molecular Correlates of Visual Processing
The mammalian visual system separates different attributes of the visual scene, such as motion and color, into distinct processing streams that originate in the retina and continue through the cortex. Although these streams have been defined both anatomically and electrophysiologically, their molecular organization has been difficult to discern. We used a custom-made ferret cDNA microarray, along with anatomical labeling and microdissection, to uncover molecular correlates of these functional streams. In addition to discovering a range of molecules, including transcription factors, that define and distinguish the visual thalamus from surrounding structures, we found one molecule, called PCP4, that is enriched in thalamic neurons that carry information about motion—the Y pathway. PCP4 was found in Y cells in the thalamus of carnivores (ferrets) and in the corresponding cell populations in primates. In contrast to other markers, PCP4 appears early in visual system development, suggesting that the functional differentiation of visual streams occurs well before visual experience. Moreover, this molecule may provide a possible tool for subsequent genetic manipulation of a specific visual pathway.

Last updated: November 28, 2005