The scale of the problem seems astronomical. How do cells in a developing nervous system "know" how to make synapses, or connections, to other nerve cells, or neurons, to create the proper wiring needed in the adult brain? It is estimated that a fruit fly's 250,000 neurons create millions of synapses, while the 100 billion neurons in the human brain form 100 trillion synapses. An organism's ability to buzz about or think abstract thoughts relies on the integrity of the circuitry among all these neurons.
For almost 30 years, Larry Zipursky has been searching for the soldering molecules nature uses to link up nerve cells in the brain of the model organism, Drosophila melanogaster, commonly known as the fruit fly. It has been a meandering quest. But about a decade ago, Zipursky discovered Dscam, a family of proteins, and changed the way neuroscientists think about the processes involved in wiring.
Thousands of different Dscam proteins exist. Each type of Dscam protein binds to the same type but not to another type. While it has been generally believed that specific proteins must act to promote wiring between different cells, the Dscam proteins act in a different way. They prevent a neuron from connecting to itself in a "self-avoidance" process vital to creating proper brain circuitry.
Zipursky combines biochemistry and genetics in his work. "Emotionally, I'm a biochemist," Zipursky says, "but intellectually, I am a geneticist." Biochemistry experiments yield results more quickly than the laborious crosses that genetics entails, he explains. With genetics, scientists make changes or mutations to the genetic code of a model organism—such as a bacterium, virus, fruit fly, or mouse—and determine, often after several generations, how the alteration affects a trait, such as eye color or even a behavior. Biochemical techniques reveal the changes at a molecular level.
Zipursky became intrigued by nervous system wiring in the early 1980s as a postdoctoral fellow in Seymour Benzer's laboratory at the California Institute of Technology. Benzer had used genetics to study fruit fly behavior and wanted to understand circuitry formation. "The idea was that genetics would give us wiring mutants to identify the genes, and then we could understand how the proteins encoded by the genes function to allow neurons to recognize each other," Zipursky says.
Finding wiring genes was difficult. But while analyzing a candidate wiring gene in his lab at the University of California, Los Angeles, Zipursky made an important, but unrelated, discovery. Through studies on the fly visual system, he showed at the molecular level one way cells talk to each other during the formation of the nervous system. He identified the bride of sevenless gene, which is in one cell type in the visual system and encodes a protein that binds to the product of the sevenless gene on another neighboring precursor cell. Once both proteins interact, the precursor nerve cell becomes a specific type of visual system neuron.
It took until the late 1990s, however, before Zipursky made progress in identifying wiring genes. His first breakthrough was the dreadlocks mutant, which affects how axons in photoreceptors, or light-sensitive neurons, align in the fly brain. Axons are long projections from a nerve cell body. "Normally axons line up like combed hair strands," Zipursky explains. "In dreadlocks mutants, the hair clumps and the axons form abnormal bundles."
The sequence of the dreadlocks gene revealed it codes for a linker protein, which, in this case, connects cell surface proteins, such as receptors, to signaling molecules inside the cell that control the cell's activity. Zipursky then set out to find the proteins to which Dreadlocks binds on the cell surface and internally. He soon found Pak, an internal protein that modulates the cell's scaffolding at the axon's growing end. Subsequently, his group focused on the fly version of the cell surface protein Dscam (in humans called Down syndrome cell adhesion molecule).
The group characterized the Dreadlocks, Dscam, and Pak interaction and discovered a strange aspect of the Dscam gene: It codes for 38,016 possible proteins through an extreme case of alternative splicing. Genes contain a linear DNA sequence that the cellular machinery ultimately reads to make a protein. But some genes, such as Dscam, contain multiple versions of certain sections of the gene, which are preferentially spliced in and allow for variations in parts of the protein.
Recently, Zipursky and colleagues determined one way Dscam helps pattern the formation of neuronal connections. Each neuron expresses a unique medley of Dscam proteins, giving a cell a distinct identity. Typically, each neuron will extend many dendrites that fan out like tree branches. Dendrites from many different neurons intermingle. "Branches from the same neuron avoid each other and it turns out that the unique combination of Dscam proteins on the surface of a neuron allows neurons to tell the difference between branches from the same neuron, which they will avoid, from the branches from other cells to which they connect," Zipursky says.
Recognition occurs because the variable parts of the Dscam molecules allow them to bind to each other and subsequently tell the sister branches of the same neuron to then grow away from each other. Although vertebrates have only one type of Dscam, recent studies suggest that they also use the protein to prevent specific neurons from connecting to themselves.
How the nerve cell separates from itself after binding is unknown and is an active area of research in Zipursky's laboratory. How many different types of Dscam are necessary to wire the fly brain is another line of inquiry. Although Dscam plays a profound role in wiring, other proteins are acting to allow neurons to recognize each other, too. Zipursky and his colleagues study several of these and continue the search for others. "The diversity of Dscam is humbling," Zipursky says, "but it is not the only player. You just have to look at the anatomy of the brain to see how complicated wiring is."