All animals have evolved innate behaviors to respond rapidly to specific sensory stimuli that threaten or promote the survival of the species. In higher vertebrates, these behaviors are often associated with emotions. My laboratory is using molecular genetic tools to functionally map neural circuits involved in innate emotional behaviors, in both mice and flies. In mice, we are focused on neural circuits underlying fear, pain, and aggression. In flies, we are studying innate defensive responses to aversive stimuli such as carbon dioxide and wind, as well as more complex behaviors such as aggression. Our ultimate objectives are twofold: to provide a neural circuit-level description of emotional behaviors, in order to provide a framework for understanding how genes and environmental factors (such as stress) cause psychiatric disorders; and to understand the evolution of "emotional" behaviors, using Drosophila as a model system.
Dedicated Primary Sensory Pathways for Innate Defensive Responses
The neural circuits that mediate innate defensive behavioral responses begin at the periphery with primary sensory neurons. How specifically "tuned" are such primary sensory neurons to the sensory stimuli that release defensive behaviors? In flies, we have studied two defensive behaviors: avoidance of carbon dioxide (CO2), and "freezing" behavior (immobility) caused by wind (wind-induced suppression of locomotion, or WISL). In both cases, we have identified specific sensory neurons that appear dedicated to the detection of these stimuli.
When mechanically traumatized, flies release a mixture of odorants—including CO2 and other, as-yet unidentified compounds—that promote avoidance in other flies. In collaboration with Richard Axel (HHMI, Columbia University) and the late Seymour Benzer (California Institute of Technology), we identified a specific set of primary olfactory sensory neurons that detect CO2 and that are not activated by any other odorants thus far tested. Functional inhibition of these neurons, using a genetically encoded inhibitor of nerve transmission, is sufficient to block behavioral avoidance of CO2 in freely moving flies. Conversely, activation of these neurons with blue light, using the light-sensitive cation channel channelrhodopsin-2 (ChR2) from Chlamydomonas, causes avoidance of blue light (which is normally attractive to flies). These data suggest that avoidance behavior may be "hardwired" to the activation of CO2-sensitive neurons and that these neurons (in contrast to other classes of olfactory sensory neurons) are specialized for CO2 responses. Future studies are directed at mapping the neural pathway that translates CO2 detection into avoidance behavior.
When flies are exposed to wind at speeds ≥1 m/sec, they abruptly cease walking and remain immobilized until the wind dies down. Using genetically encoded calcium indicators to measure neuronal activity, together with specific enhancer trap lines generated and characterized by Kei Ito (University of Tokyo), we identified specific sets of wind-responsive neurons in the fly antenna. Surprisingly, these neurons are located in the same sensory organ, Johnston's organ (JO), that is used for hearing. They are, however, distinct in their intrinsic properties from sound-sensitive neurons also located within JO. Genetic ablation of the wind-sensitive subset of JO neurons abolishes wind-induced locomotor arrest, without impairing hearing. Thus, the distinct innate behavioral responses elicited by sound (e.g., female responses to male courtship song) and wind (WISL) are mediated by distinct classes of mechanosensory neurons "tuned" to distinct types of air particle movements. Future studies will be aimed at understanding how and where in the brain the detection of wind is translated into the arrest of walking behavior.
The skin, the body's largest sensory organ, detects different types of painful stimuli, such as thermal and mechanical. We can tell without looking whether we have been stuck by a pin or burnt by a match. How does the brain know what kind of painful stimulus the body is sensing? The primary sensory neurons that detect noxious (painful) stimuli are called nociceptors. In vertebrates, individual nociceptors can be activated by both noxious mechanical and thermal (heat) stimuli. This has suggested that different noxious stimulus modalities are discriminated by the brain only at higher levels of information processing. We have discovered that genetic ablation of one specific subset of nociceptors that expresses the G protein–coupled receptor (GPCR) Mrgprd causes selective deficits in sensitivity to noxious mechanical, but not thermal, stimuli. Conversely, experiments performed by Allan Basbaum (University of California, San Francisco) have shown that chemical ablation of the fibers of a different subset of sensory neurons, expressing the heat-activated ion channel TRPV1, leads to a loss of heat pain sensitivity but not a loss of mechanical sensitivity. Combined ablation of both populations yielded an additive phenotype, with no additional behavioral deficits, ruling out a redundant contribution of these populations to heat and mechanical pain sensitivity. These data, like those in the fly, indicate that different sensory systems contain specific subsets of primary sensory neurons that mediate specific defensive responses to specific stimuli or stimulus modalities.
Neural Circuits for Fear and Aggression
"Fight-or-flight" responses are among the most primitive emotional behaviors known. In vertebrates, these behaviors are mediated by deep brain structures such as the amygdala and the hypothalamus. We are applying molecular genetic tools to dissecting the neural circuitry of fear and aggression in mice, focusing on identified subpopulations of neurons located within these structures, which provide a point of entry to the circuits of interest. Our approach integrates functional manipulation of neurons in freely behaving animals with behavioral, electrophysiological, and neuroanatomic analysis. For example, we are using a method developed in collaboration with Henry Lester (Caltech) to reversibly inhibit genetically defined subsets of neurons that express a ligand-gated chloride ion channel from the nematode Caenorhabditis elegans. We are also activating the same subsets of neurons by using ChR2 and flexible fiber-optic cables to deliver illumination to deep brain structures, in collaboration with Karl Deisseroth (Stanford University). The effects of these manipulations on fear and anxiety behaviors are being investigated.
We are taking conceptually similar approaches to the study of aggression in Drosophila. This model organism affords, in principle, the ability to perform high-throughput genetic or neural circuit-level screens. However, such screens are difficult to perform in the case of aggression, which is a complex behavior that is laborious to score manually. In collaboration with Pietro Perona (Caltech), we have used machine vision-based techniques to develop software for the automated detection of different types of aggressive behavior in Drosophila. This technology, in combination with genetic tools for manipulating specific populations of neurons, should for the first time permit an unbiased, systematic approach to the functional dissection of neural circuits underlying aggressive behavior in this powerful model organism.