The central nervous system (CNS) integrates environmental sensory information (sight, sound, smell, taste, and touch) with signals from the body (sensory information from internal organs and hormones) to generate appropriate movements. The CNS can learn to associate particular sensory cues with subsequent events to facilitate appropriate responses—either approach or avoidance behaviors. Our laboratory uses genetic manipulations in the mouse to examine the neural circuits (the wiring diagrams) involved in these responses and the signaling molecules (neurotransmitters/neuromodulators) that are used by the neurons in the circuit. The neural circuits that mediate essential behaviors that do not require thought are likely to be hardwired, although still subject to modulation. Complete neural circuits have not been defined for most basic behaviors in mammals. However, genetic and viral tools are being developed that are promoting rapid progress. We are using these techniques to decipher neural circuits that promote or inhibit feeding behavior.
Neurons in an area of the brain called the arcuate region of the hypothalamus (ARC) integrate hormonal (insulin, leptin, and ghrelin) and neuronal inputs to modulate food intake and metabolism as a means of maintaining adequate energy supplies for bodily needs. One population of neurons in the ARC that has received considerable attention expresses γ-aminobutyric acid (GABA), neuropeptide Y, and agouti-related protein (AgRP) as neurotransmitters/neuromodulators. Because AgRP is expressed exclusively in these neurons, we refer to them as AgRP neurons. These neurons become active when an animal is hungry and promote feeding behaviors by releasing their transmitters in various regions of the brain to activate the next neurons in the circuit. Genetic manipulation of AGRP neurons is relatively easy because one can target the expression of new genes to the Agrp gene locus by homologous recombination in embryonic stem cells and then derive mice carrying that genetic modification. If Cre recombinase is targeted to these AgRP neurons, then a virus carrying a Cre-dependent effector gene can be injected into the ARC and the effector protein will be expressed only in AgRP neurons. One useful effector gene is channelrhodopsin (ChR2),a light-activated ion channel that causes neurons to release transmitters when activated by a laser connected to an optic fiber inserted just above the ARC. Photo activation of AgRP neurons during the day when mice are usually sleeping stimulates robust feeding behaviors. Scott Sternson, a group leader HHMI's Janelia Farm Research Campus, has exploited this clever technique. We performed a converse experiment by making mice that expressed the human diphtheria toxin receptor from the Agrp locus and then administered diphtheria toxin, which killed the AgRP neurons. As might be expected, if these neurons are important for feeding, the demise of AgRP neurons in adult mice led to starvation.
To understand why the mice starve after ablation of AgRP neurons, we assumed that loss of inhibitory signaling by these AgRP neurons resulted in hyperactivity of postsynaptic neurons elsewhere in the brain, which then promoted anorexia. Through a series of experiments, we discovered that hyperactivity of neurons in a brain region called the parabrachial nucleus (PBN) was responsible for the anorexia. We exploited the location of the hyperactive neurons in the PBN and coincidence with the expression pattern of the Calca gene encoding calcitonin gene–related protein (CGRP) to make a mouse that expresses Cre recombinase from the Calca locus. Viral delivery of Cre-dependent ChR2 to the PBN of the Calca-Cre mice followed by photoactivation revealed that excitation of these CGRP-expressing neurons inhibited feeding by hungry mice. Furthermore, we used another strategy to chronically inhibit these neurons and prevented starvation after ablation of AgRP neurons, providing strong evidence that loss of AgRP neurons promotes starvation by activating the CGRP neurons.
These CGRP neurons are known to relay sensory information to the forebrain. They are normally activated by visceral malaise (e.g., food poisoning), nausea (e.g., motion sickness), satiety, and probably many other conditions that lead to anorexia. Thus, activation of these neurons provides a brake on normal feeding activity and presumably protects mice from dangerous environmental events. The phenomenon of conditioned taste aversion, in which ingestion of a novel food is followed by visceral illness and consequent aversion to consuming that food in the future, depends on this circuit. This is a long-lasting, one-trial learning experience that is of obvious value to a foraging animal. Photoactivation of CGRP neurons coincident with presentation of a novel food is sufficient to establish an aversion to eating that food. The CGRP neurons also mediate the aversive effects of a foot shock: blockade of CGRP neuron function attenuates the ability of a mouse to associate the foot shock with the location in the environment where it occurred.
By selective expression of ChR2-mCherry in CGRP neurons, we visualized axon projections to the bed nucleus of the stria terminalis (BNST) and to the lateral capsule region of the central nucleus of the amygdala (lcCeA). Photoactivation of the terminals in the lcCeA inhibited feeding by hungry mice, whereas activation of terminals in the BNST had no effect. Current efforts are directed toward identifying the target neurons in the lcCeA and determining where they project their axons. We also intend to identify neurons in the hindbrain that directly activate CGRP neurons. These efforts should help to define the neural circuit that leads from the viscera to the amygdala and beyond.
Ablation of AgRP neurons in adult mice results in starvation by activating the CGRP neurons. We have discovered numerous ways to prevent the hyperactivation of CGRP neurons and thus prevent starvation, including genetic downregulation of glutamatergic signaling onto CGRP neurons, pharmacological activation of GABA signaling onto CGRP neurons, and prior treatment with lithium chloride. Remarkably, after a week or so of these interventions, the mice survive without their AgRP neurons, suggesting that some form of adaptation has taken place such that the appetite-enhancing role of the AgRP neurons is no longer necessary for adequate feeding. We suspect that adaptation involves synaptic plasticity within the CGRP neurons and plan to use electrophysiological techniques to discern the mechanisms.
Our research is supported in part by grants from the National Institutes of Health and the Klarman Family Foundation.
As of February 28, 2014