Neural Ensemble Codes in Drosophila Olfaction
Our research program aims to understand how neural ensembles represent and process information. Information is inevitably tied to a physical carrier - a pencil mark on paper, a charge on a capacitor, an action potential or synaptic event in a nervous system. Any information processing operation, be it a measurement or a computation, transforms one such representation into another, according to a set of rules embodied in the physical properties of the system.

While these rules are fairly well understood for single neurons encoding simple stimulus features, neural representations of complex stimuli, their inter-relations, and their behavioral significance remain largely mysterious. It is presumed that these representations are distributed over neural ensembles - groups of neurons in transient functional linkage - and written in a code that involves the spatial locations of active cells or synapses and the times at which activity occurs. Due to a paucity of experimental approaches, however, even seemingly elementary facts about ensemble codes are unknown: the sizes of ensembles and their dynamics, the nature of the functional linkage among ensemble members, or the features which demarcate co-active ensembles.

A genetically encoded optical sensor reports synaptic activity in living neurons

Even in favorable neuroanatomical circumstances, recording ensemble signals presents a serious challenge. Electrophysiological methods are generally limited to a few neurons at a time, while synthetic indicator dyes face problems of access and specificity, particularly in functionally intact systems. Our strategy to overcome these difficulties relies on protein-based sensors that provide direct optical images of neural activity. Since these molecules are encodable in DNA, they can be introduced into intact animals by genetic manipulation, and their expression pattern can be tailored to include - exclusively and at the same time comprehensively - the neurons of interest.

Reading an ensemble code (as opposed to merely recording ensemble activity) requires knowledge not only of ensemble signals but also of sensory input and behavioral output. The olfactory system of the fly, Drosophila melanogaster, can serve as such a Rosetta stone. Information transduced by odorant receptors in the insect’s antennae (corresponding to the nose of vertebrates) is mapped onto the antennal lobes (corresponding to the olfactory bulb) and mushroom bodies (corresponding to the olfactory cortex). Each of these structures can be marked selectively with a genetically encoded sensor, and each can be viewed in a living animal by optical microscopy. Representations of odors and odor blends can thus be analyzed in virtual isolation at successive stages of an intact processing cascade, allowing the transformation rules which apply between stages to be deduced.

Once the territory is charted, questions about the mechanics of ensembles as well as their information content become meaningful. These questions may be posed in the form of mutations with defined neural or behavioral phenotypes. Although a large number of such mutations have been described in Drosophila, efforts to seamlessly relate mutant molecule to mutant behavior have often failed. This is not surprising: to do so would require knowledge not only of how mutation of a particular gene alters its protein product and the physiology of cells in which it is expressed, but also how inclusion of cells with altered physiology in a neural ensemble affects ensemble properties. It is this difficult and elusive last step that we seek to accomplish.

Our strategy is similar in spirit to the successful dissection of complex metabolic pathways by altering one element at a time. It is hoped that a careful analysis of the effects of such alterations will produce an understanding of brain function that transcends the boundaries which presently divide neuroscience into molecular, cellular, and systems branches. A not unimportant by-product of our efforts, the indicator molecules we generate to visualize neural activity will continue to prove valuable in many other areas, most notably the process of drug discovery.

Questions about/problems with this site? Please e-mail the webmaster.

© Copyright 2007 Kinship Foundation. All rights reserved.