The human brain contains ~1011 neurons, each making ~103 synapses with other neurons. These 1014 synaptic connections enable us to sense, think, remember, and act. How is this vast number of neurons organized into circuits to process information? How are these circuits assembled during development? To address these questions, we use model neural circuits in the less numerically complex brains of the fruit fly (~105 neurons) and mouse (~108 neurons) and combine state-of-the-art molecular genetics and viral techniques with physiological and behavioral approaches.
Organization of the Fly and Mouse Olfactory Circuits
The olfactory circuits of flies and mice share remarkable similarities and offer salient advantages for investigating their structure, function, and development. In the fly, olfactory receptor neurons (ORNs) expressing the same odorant receptor project their axons to the same glomeruli in the antennal lobe. Projection neurons (PNs) send dendrites to individual glomeruli and relay olfactory information via their axons to the mushroom body and lateral horn (Figure 1), higher-order centers that mediate learned and innate olfactory behavior, respectively. By systematically labeling individual neurons, we found that terminal arborizations of the same PN type are variable in the mushroom body but exhibit striking stereotypy in the lateral horn. PN axon terminals that represent food odors are spatially segregated from those that represent mating pheromones. Using Ca2+ imaging and optogenetics, we found that lateral horn-projecting inhibitory PNs (iPNs) selectively suppressed food-related odor responses but spared pheromone signal transmission. Thus, anatomical segregation and iPNs function together to ensure that downstream circuits differentially process odors of distinct biological significance.
Mouse ORNs expressing the same odorant receptor also project their axons to the same glomeruli in the olfactory bulb. Using rabies virus–mediated trans-synaptic tracing, we found that individual cortical neurons receive input from broadly distributed glomeruli to integrate information from multiple odorants. Whereas the cortical amygdala preferentially receives dorsal olfactory bulb input, the piriform cortex samples the olfactory bulb without obvious bias, suggesting that the mouse olfactory cortex may also use differentially organized input for mediating learned and innate behavior.
Assembly of the Fly Olfactory Circuit
The assembly of the fly olfactory system requires precise glomerular targeting of axons from each of the 50 ORN types and dendrites from each of the 50 PN types. We have used this neural circuit to investigate the general principles by which wiring specificity is established during development. We found that PN dendrites pattern the antennal lobe prior to ORN axon arrival. Global graded cues and local binary determinants collaborate to pattern PN dendrites. ORN axon targeting also employs a multistep process involving trajectory choice, axon-axon repulsion, and synaptic partner matching to establish one-to-one connections between cognate ORNs and PNs. We recently identified a pair of evolutionarily conserved Teneurins as the first synaptic partner–matching molecules. Teneurins are highly expressed in select PN-ORN pairs and instruct synaptic partner matching through homophilic attraction. In addition, basal-level Teneurin expression is required for proper ORN-PN synapse development.
To decipher the cell surface code that specifies olfactory circuit assembly, we are performing high-resolution RNA interference–based screens in different parts of the antennal lobe to identify new factors that instruct PN dendrite targeting, ORN axon targeting, and ORN-PN matching. We are also investigating the synaptic maturation process and exploring the roles of glia in the assembly and maintenance of the olfactory circuit. These studies will provide a more comprehensive understanding of how wiring specificity is achieved in this model neural circuit.
Explorations of Mammalian Neural Development
Over the past decade, we have extended our tools and concepts developed in flies to mice. We are particularly interested in how wiring specificity of mammalian neural circuits is achieved by the interplay between molecular determinants and neuronal activity. We are exploring the function of the mammalian homologs of wiring-specificity molecules we discovered in flies, and we are testing candidate molecules that likely play important roles in activity-dependent wiring. For example, we found a cell-autonomous role for the NMDA receptor, a key coincidence detector for correlated neuronal activity, in patterning dendrites of barrel cortex stellate cells.
We recently found that sparse but not global knock out of the neurotrophin receptor TrkC in cerebellar Purkinje cells reduced dendrite complexity (Figure 2). Removal of the TrkC ligand NT-3 from granule cells, the major presynaptic partners of Purkinje cells, suppressed the Purkinje dendrite defects caused by sparse TrkC disruption. These findings suggest that growing dendrites require NT-3 from their presynaptic partners during competitive dendrite morphogenesis. We are exploring the relationship between neurotrophin signaling and neuronal activity in this process.
Development of Genetic Tools
We have a continued interest in creating genetic tools to probe neural circuit organization and assembly. For example, the MARCM (mosaic analysis with a repressible cell marker) method in flies and MADM (mosaic analysis with double markers) method in mice allow the visualization and genetic manipulation of isolated single neurons (Figures 1 and 2). These tools enabled many experiments described above and are widely used by the research community.
Recently, we have extended the rabies-mediated trans-synaptic tracing tools developed by Ed Callaway and colleagues (Salk Institute for Biological Studies). By modifying helper viruses, we have reduced background labeling and enhanced the efficiency of trans-synaptic tracing, so that we can trace whole-brain and local input to genetically and anatomically specified neuronal types (Figure 3). We have also developed tools to initiate rabies tracing from neurons of a particular type and with a particular projection pattern. We are applying these tools to explore the input-output organization of the monoamine neuromodulatory systems, which regulate diverse behavior through their widespread projections.
Genetic dissection of neural circuits relies on access to specific neuronal populations defined by cell type, location, or axonal projection. We recently developed an approach in the mouse to obtain genetic access to neurons that were activated by defined stimuli, with the use of an immediate early gene promoter to drive a drug-inducible Cre recombinase. We developed a conceptually analogous approach in the fly. Genetic access to neurons on the basis of their activity, in combination with tools for labeling, tracing, recording, and manipulating neurons, offers a powerful approach for understanding how neural circuits process information and generate behavior.
These studies have also been supported by grants from the National Institutes of Health and the Simons Foundation.
As of February 23, 2015