Despite its tiny size, the fruitfly brain is staggeringly intricate. So teasing apart how it remembers things — even a simple line pattern — is a daunting task. Progress is being made, thanks to genetic innovations.
Neuroscientists these days have a satisfactory understanding of how individual neurons work and of how they communicate with their immediate neighbours. By contrast, understanding at the next level of organization is hazier; for example, how neurons form functional circuits, how these circuits encode behaviour and particularly how experience changes the activity and connectivity in circuits to alter behaviour.
In this fog, a natural question is: how simple a system can one study profitably? Molluscs such as the marine snail Aplysia have yielded much insight because they have large, simple neuronal circuits. However, these animals perform only very basic behaviours. Insects, on the other hand, often have intricate neural circuits and complex stereotyped behaviours, such as the dance language of the honeybee. But their neurons have seemed too small and tangled for conventional analyses. Advances in genetic techniques have overcome this problem of scale, and in this issue Liu et al. (page 551)1 use these ingenious methods to begin to dissect finely how the fruitfly Drosophila learns visual patterns.
The authors take advantage of ‘jumping genes’, which can hop about the fly genome and splice themselves into chromosomes at random points. A jumping gene can be tailored to carry along another gene of interest — a transgene. If the transgene (called ‘TG-1’, say) is jumped into the DNA near a naturally occurring gene, it is usually expressed in the same tissues as the natural gene2. This jumping-gene method is now highly developed in Drosophila, so that large numbers of fly stocks are available by mail, each with TG-1 expressed in a different tissue, neuroanatomical structure or subgroup of neurons.
Nowadays, TG-1 usually encodes GAL4, a gene regulator that can be used specifically to turn on a second transgene (‘TG-2’) only in the requisite tissues or cell types. TG-2 will be a particular gene variant or mutation of interest, and it will have genetics such that it can be transferred into or out of a fly's genome via simple genetic crosses3. So it is now possible, by mail order and simple fly breeding, to express almost any interesting gene in an infinite variety of tissues or cells (Fig. 1). TG-2 can encode a gene product that identifies the selected neurons by dye or fluorescence, or it can be a protein that kills them, blocks transmission from them or overstimulates them. If TG-2 makes a protein that fluoresces in response to markers of neuronal activity such as elevated calcium or membrane voltage changes, then one can watch the neurons in action4,5.
Liu et al.1 used this method to dissect minutely the circuits involved in learning visual patterns. In their behavioural assay6, the fly faces a choice between two visual figures (for example, upright and inverted T shapes) and is punished by mild heating when it turns to one of them. Not surprisingly, the fly ends up facing the other figure; more surprisingly, it remembers to face the second figure in a subsequent test without heat reinforcement7.
Learning in Drosophila depends substantially on the Rutabaga enzyme. This enzyme is a calcium–calmodulin-dependent adenylyl cyclase8 that could be a molecular site for the convergence of signals from the cued stimulus and the reinforcement. Different types of learning require Rutabaga in different areas of the brain. Odour-discrimination learning, for instance, requires the enzyme found in brain structures called the mushroom bodies9. What Liu et al. show is that to learn specific visual patterns, a fly must have Rutabaga in a brain region known as the central complex. They used mutant flies that lacked Rutabaga, and then switched Rutabaga on in a selection of specific patterns using the system above, and tested to see which flies remembered the visual cues. Strikingly, discrimination between different features — the angular orientation of a bar or its vertical position on the fly's retinal field — requires Rutabaga in different sets of central-complex neurons, which project to different layers of the ‘fan-shaped body’ (FB), a sub-structure of the central complex (Fig. 2). Those neurons that use Rutabaga to store vertical elevation extend into FB layer 5; those that encode contour orientation project to FB layer 1.
How does this work advance the field? Memory has been plausibly localized to small groups of neurons before, in Aplysia, but that was memory simply to increase or decrease the strength of an innate reflex — gill withdrawal after electric shock. Liu et al. localize memories much richer in information about higher-order visual properties — memories that presumably require the cells' integration into finely tuned circuits. Can we understand the circuits? Probably not for a while. The neuroanatomy of the central complex has been systematically studied, and several stereo-typed classes of neuron have been categorized and mapped there10. The organization of the central complex is repetitive and regular, but fiendishly intricate. Understanding how the circuitry encodes angular orientation, for example, will require either new methods or clever, and lucky, anatomical insight.
A more tractable question is: how does Rutabaga bring about changes in the appropriate central-complex cells to encode the memories, and what are the relevant changes? If the changes are structural, there are methods that will pinpoint them11. Finding electrical changes in the cells is a knottier problem, because the central-complex cells are small and located deep inside the fly brain. Clues may come from other mutants deficient in learning, work on Aplysia, or studies of larger, more superficial cells in the Drosophila brain5. What Liu et al. make very clear is that genetic trickery has converted Drosophila from one of the worst organisms for functional neuroanatomy to one of the better ones. Indeed, for localizing individual, complex memories it may be the best.
Liu, G. et al. Nature 439, 551–556 (2006).
O'Kane, C. J. & Gehring, W. Proc. Natl Acad. Sci. USA 84, 9123–9127 (1987).
Brand, A. H. & Perrimon, N. Development 118, 401–415 (1993).
Suh, G. S. et al. Nature 431, 854–859 (2004).
Yu, D. et al. Cell 123, 545–557 (2005).
Heisenberg, M. & Wolf, R. J. Comp. Physiol. A 130, 113–130 (1979).
Dill, M. et al. Nature 365, 751–753 (1993).
Sziber, P. P. et al. Cell 37, 205–215 (1984).
Zars, T. et al. Science 288, 672–675 (2000).
Hanesch, U., Fischbach, K. F. & Heisenberg, M. Cell Tissue Res. 257, 343–366 (1989).
Lee, T. & Luo, L. Trends Neurosci. 24, 251–254 (2001).
About this article
Science China Life Sciences (2010)
Journal of Neurogenetics (2009)
Proceedings of the National Academy of Sciences (2007)