Wiring the Drosophila Visual System

We have been studying the formation of connections between photoreceptor neurons (R cells) and their targets in the brain. The compound eye of the fly contains some 800 simple eyes called ommatidia and each ommatidium contains 8 R cells. These cells can be divided into three classes based on synaptic specificity. The R1-R6 neurons form connections in the first optic ganglion of the fly brain called the lamina. The R7 and R8 neurons extend through the lamina and make connections in two distinct layers in the second optic ganglion the medulla.

To dissect the mechanisms regulating connection specificity we have utilized two different genetic screens. For genes regulating R1-R6 connectivity we screened for flies defective in motion detection. We screened for R7 connectivity mutants by virtue of defects in their response to ultraviolet (UV) light. As many connectivity genes may also be necessary for viability or motor functions used as outputs in these assays, we have chosen to use genetic techniques that enable us to make the eye, or only a subclass of cells in the eye, homozygous for randomly induced mutations. By making the entire eye mutant and testing flies for motion detection we have identified a panel of mutants affecting R1-R6 connections. Similarly, by generating R7 neurons that are homozygous mutant and testing their response to UV light, we have identified mutations disrupting R7 connectivity.

R1-R6 neurons from the same ommatidium extend axons within a single bundle into the lamina. During subsequent development each axon projects away from the bundle and innervates a different group of postsynaptic cells. Each group of postsynaptic cells is innervated by 6 different R1-R6 neurons from 6 different ommatidia. These 6 R cells "look"at the same point in space. We have shown through genetic analysis in which the number and identity of R1-R6 cells was manipulated that interactions between R1-R6 axons are critical for these patterns of connections to emerge. Genetic screens have identified a set of genes required for the formation of these connections. Three have these have been characterized molecularly and all encode cell surface proteins. They are N-cadherin, LAR, a receptor tyrosine phosphatase, and a novel cadherin protein called Flamingo.

Although R1-R6 neurons initially terminate in the lamina largely as they do in wild type in N-cadherin, Lar and flamingo mutants, they fail to innervate their correct targets within the lamina. In N-cadherin and Lar mutants the R1-R6 axons from the same ommatidium remain together and innervate the same target. In flamingo mutants each R cell axon extends out from the bundle but innervates inappropriate targets. As both N-cadherin and Flamingo proteins are homophilic cell adhesion molecules (when these proteins are on the surface of adjacent cell membranes they bind them together), they may play an important role in mediating interactions between different R cell axons or between R cell axons and targets.

N-cadherin and Lar, but not flamingo, were also identified in genetic screens for R7-mediated behavior. Indeed, by making flies where only R7 cells were mutant for these genes it was shown that they play a remarkably specific role on connection specificity. R7 cells lacking N-cadherin or Lar, where the surrounding R cells and target cells are wild type, connect to the R8 layer rather than projecting more deeply to the R7 layer. Detailed developmental studies feasible with Lar, but not N-cadherin, revealed that in Lar mutants R7 axons extend to the R7 target layer initially, but these connections are unstable resulting in their retraction to the R8 layer. Biochemical studies in vertebrate neurons support the view that N-cadherin and Lar form a complex and histological and antibody studies have further suggested that vertebrate cadherins are important for synapse formation or stability. Our genetic studies in the fly visual system support the view that Cadherin-mediated processes are central to the formation of precise patterns of synaptic connections.

In the process of dissecting a signal transduction pathway regulating axon guidance in the larval photoreceptor neurons we, in collaboration with Jack Dickson’s lab at the University of Michigan, isolated and characterized an axon guidance receptor called Down Syndrome Cell Adhesion (Dscam). Human Dscam maps to a region of chromosome 21 associated with Down syndrome and has been speculated to contribute to brain abnormalities in this syndrome.

Drosophila Dscam acts in conjunction with two well characterized signaling molecules Dock and Pak to transmit guidance signals to the actin cytoskeleton. We proposed that in the developing embryo Dscam, Dock and Pak act together to mediate a short range interaction between a growth cone of an identified neuron, called Bolwig’s nerve, and a target with which it interacts transiently as it projects to its final target in the brain. Dscam is also required for normal patterns of neuronal connections within both the embryonic and postembryonic central nervous systems.

Multiple forms of Dscam protein are generated through extensive alternative splicing. These forms all have the same domain structures but differ in amino acid sequence. The Dscam proteins all have 10 immunoglobulin domains, 6 fibrinectin type III repeats, a transmembrane segment and a cytoplasmic domain that directly interacts with the signaling protein Dock through both SH2 and SH3 domain interactions. Dscam isoforms differ from each other by the inclusion of alternative exons that encode for three variable immunoglobulin domains. In addition, each isoform has one of two different transmembrane domains. Molecular analysis strongly supports the view that most and perhaps all of the predicted 38,016 isoforms are made in the developing animal. We speculate that these sequences confer differences in recognition underlying the enormous diversity of connection specificity. Current efforts are directed towards understanding how the different isoforms of Dscam contribute to connection specificity in the developing fly brain and whether different classes of neurons express unique receptors or a unique spectrum of receptors.

The Drosophila genome contains 3 other Dscam genes, Dscam 2, 3 and 4. They encode proteins that are predicted to share a common extracellular domain structure and highly divergent signaling domains. While the prototypical Dscam and Dscam 4 are expressed in all neurons, Dscam 2 and 3 are expressed in a subclass of neurons in the central nervous system. Dscam 2,3 and 4 do not appear to undergo extensive alternative splicing. Biochemical and genetic studies are in progress to determine the functions of Dscam 2,3 and 4 in regulating the formations of neuronal connections in Drosophila.

Schmucker, D., Clemens, J.C., Shu, H., Worby, C.A., Xiao, J., Muda, M., Dixon, J.E. and Zipursky, S.L. (2000) Drosophila Dscam is an axon guidance receptor exhibiting extraordinary molecular diversity. Cell 101, 671-684.

Clandinin, T. and Zipursky, S.L. (2000) Afferent growth cone interactions control synaptic specificity in the Drosophila visual system. Neuron 28, 427-436.

Lee, C.-H., Herman, T., Clandinin, T.R., Lee, R. and Zipursky, S.L. (2001) N-cadherin regulates target specificity in the Drosophila visual system. Neuron 30(2), 437-450.