How the brain is organized to respond appropriately to experience and produce complicated behaviors has implications for injury, disease, and development, and even for the design of mechanical brains. To approach this large area of investigation, we use the fruit fly Drosophila melanogaster, because of its wealth of genetic tools and because its brain, while simpler than a mammal's, is nevertheless able to drive complicated behaviors such as flight and learning. We anticipate that the principles we derive from these analyses will hold true in higher organisms.
Our basic approach is to make rough structure-function maps for a variety of different behaviors. We rapidly and reversibly increase or decrease neural activity in distinct subsets of the fly brain and determine the kinds of behavioral defects that result. Our initial screens identified clusters of neurons that are capable of instigating behaviors as diverse as seizures and courtship. We continue to screen for groups of neurons that disrupt specific behaviors, to narrow the number of neurons affected in any given experiment, and to analyze the projection patterns and connectivity of critical neurons already identified. Ultimately we seek genetic control of the different components that make up a neural circuit governing a defined behavior. Then we will study how specific neurons—and the genes expressed in them—contribute to the circuit. By comparing the organization of circuits controlling different behaviors we hope to determine, for example, whether circuits tend to be dedicated or recycled, how much of the brain is involved in different behaviors, and whether there are redundant circuits that drive the same behavior. (The groundwork for this research was supported by a Helen Hay Whitney postdoctoral fellowship, a L'Oreal Women in Science Award, grants from the National Institutes of Health to Barry Ganetzky, and HHMI.)
Spatial Control of Gene Expression
To dissect neural circuits, we need to control gene expression in small numbers of neurons in the brain. The UAS-GAL4 system allows us to access the same groups of neurons reproducibly and to modify those neurons. We now have a large collection of GAL4 lines, each of which drives gene expression in a different subset of neurons. We made 400 new enhancer-trap lines that express GAL4 in response to different enhancers or control loci: the neurons in each of these patterns share expression of the "trapped" gene. To target specific classes of neurons (those that use a particular neurotransmitter, for example), we isolated enhancers from genes expressed exclusively in those classes—the vesicular neurotransmitter transporters that package transmitters into synaptic vesicles—and designed GAL4 fusions.
Although these GAL4 lines are a great resource for initial mapping, they are often expressed in many neurons in the brain, making it difficult to identify the minimal critical neurons within the larger patterns. We have taken advantage of GAL80, a protein that blocks the function of GAL4, to narrow the number of cells where GAL4 is able to trigger gene expression. Combining specific GAL4 and GAL80 lines increases our ability to target small groups of neurons reproducibly. As with the GAL4s, we have a large collection of enhancer-trap (random) GAL80s and GAL80s designed to subtract out the effects of particular types of neurons. One advance we have made enables us to separate the contributions of the brain and the thoracic ganglion (roughly the spinal cord equivalent) to behavioral deficits. Combining these tools with those that group neurons based on their lineage (MARCM and Flp), we are winnowing down the neurons essential for specific behavioral functions. At Janelia, we will take advantage of the extensive GAL4 collection being generated by Gerald Rubin (Director, Janelia Farm Research Campus) and explore additional ways to target even smaller numbers of neurons.
A side benefit of our GAL4 enhancer-trap collection is that in "trapping" genes with specific expression patterns, we may identify genes with roles in establishment and function of the relevant circuits. We can readily use the inserted transposable element to identify and generate mutations in these genes. Thus, although our goal is to map the neurons important for behavioral function, our toolkit includes a way to isolate genes that may play fundamental roles as well.
Temporal Control of Neural Activity
One essential aspect of our work is that we want to look at the role of particular neurons in a developmentally normal adult brain. This requires the ability to alter neural activity late in life and very rapidly to prevent adaptation and compensation. We have made extensive use of UAS-shibirets1, a reagent generated by Toshihiro Kitamoto (University of Iowa) that blocks synaptic vesicle recycling, and thus chemical neurotransmission, at elevated temperature. We generated a complementary tool, UAS-seizurets2, that increases neural activity at elevated temperature by interfering with a potassium channel critical for repolarizing neurons after action potentials. The ability to change neural activity reciprocally has been a powerful technique for identifying neurons governing particular behaviors.
Although we will continue to use these tools, we are also improving them. Specifically, we are interested in blocking electrical—as well as chemical—neural activity and in designing ways to change activity only in neurons using particular neurotransmitters. Temperature is a terrific trigger, but other options are also being explored. Some of these new tools are already being developed, and others are being designed, in collaboration with Loren Looger (HHMI, JFRC). Our lab also plans to electrophysiologically characterize the normal properties of neurons identified in the behavioral screens and assay precisely how our activity-altering transgenes change them.
Additionally, we have used a range of techniques to change neural activity, including some techniques that alter neural function throughout development. By comparing the effects of permanently blocking activity with those of acutely blocking it, we have identified some neurons whose loss can be compensated for. We are exploring the frequency of this phenomenon and attempting to determine which neurons pick up the slack.
Visualization of Neural Projections and Connectivity
The same GAL4 lines that drive the neural activators and inhibitors can be used to drive fluorescent reporters targeted to different parts of neurons. We have examined the projection patterns of many groups of neurons and have identified some brain regions that may be critical for particular behaviors. It is often difficult to identify precisely which regions of expression different brains or different GAL4 patterns have in common. We use alignment and registration techniques such as the Standard Fly Brain developed by Martin Heisenberg (University of Würzburg) to compare fluorescence patterns in different brains, and we are collaborating with Eugene Myers (HHMI, JFRC) to automate the alignment and analysis.
We are now looking at synaptic and dendritic reporters to try to understand how information moves between parts of the brain and how different patterns affecting the same behavior are connected. To validate possible connections, we are designing and testing a range of potential trans-synaptic tracers, some of which are being built in collaboration with Loren Looger.
While we are primarily interested in the acute effects on behavior of changing activity in particular neurons, our tools also allow us to explore how neuronal projections change in response to altered activity. For example, some neurons may expand their dendritic arbors in response to hyperactivity, while others may retract them. This has implications for human diseases such as epilepsy that can be either progressive or stable. We will explore whether the identity of the neurons or the type of hyperactivity affects the morphology of the neurons involved.
It may be possible to use neural activity itself to trace which neurons are firing during particular behaviors. We are testing existing and novel reporters that are based on gene expression triggered by neural activity. We plan to explore real-time reporters of calcium influx, pH changes during vesicle release, and voltage changes in the neural membranes as well.
Initially our screens were limited to behavioral changes that could be observed during a 10-minute period in a glass vial at elevated temperature. This introduced us to a myriad of peculiarities of fly behavior and led to identification of regions implicated in seizures, paralysis, coordination, and courtship. We are now refining this assay to look directly for brain regions that control more specific behaviors. We have at least two assays in development and hope to establish others in collaboration with Roland Strauss, Vivek Jayaraman, and Michael Reiser (HHMI, JFRC). Comparing the brain regions critical for a variety of specific behaviors will give us insights into the logic of brain organization. Ultimately we hope to have independent genetic control of the different constituents of a neural circuit that drives a quantifiable behavior. We can then explore how particular types of neurons, and the genes expressed in them, help them fulfill their roles as parts of the circuit.
We understand that mapping a brain is an ambitious task, and we hope that our efforts will contribute to the larger, community-wide endeavor. The maps we make and the tools we generate can be used to address a wide range of scientific questions, including many beyond the scope of our current focus. We feel strongly that tools are meant to be shared and will freely distribute our reagents. We look forward to discussing our research with interested scientists; please do not hesitate to contact us with your questions and suggestions.