The billions of neurons in the mammalian brain communicate with each other by liberating chemical messengers that are detected by receptors on adjacent cells. These chemical messengers, or neurotransmitters, can be ions, amino acids, amino acid derivatives, or large polypeptides. Our laboratory investigates the role of specific neurotransmitters in the development and function of the mammalian nervous system by studying the consequences of removing them from the intact mouse. Because neurons depend on these molecules for communication, the loss of any particular neurotransmitter can have profound consequences on the function of the nervous system. The absence of signaling by a specific neurotransmitter is often revealed as a change in behavior or physiology. Some behavioral changes are obvious under normal conditions, but others are only apparent when an animal is challenged. Most of our current studies revolve around the roles of dopamine in mouse behavior and physiology, but we have a growing interest in the roles of glutamate and γ-aminobutyric acid (GABA) signaling.
We used gene targeting in embryonic stem cells to generate dopamine-deficient (DD) mice that lack the ability to make a critical enzyme, tyrosine hydroxylase, specifically in dopamine-producing neurons. These mice are born normally but they gradually become hypoactive, stop suckling, and expire by 4 weeks of age. Their symptoms resemble a severe form of Parkinson's disease, but unlike humans with the disease, the mice have intact dopamine-producing neurons; they just lack the ability to make dopamine. Fortunately, they can be kept alive either by hand feeding, or more commonly, by injecting them daily with L-3,4-dihydroxyphenylalanine (L-dopa), which is taken up by dopaminergic neurons and converted into dopamine. L-dopa has been used to treat Parkinson's disease for decades, but its efficacy declines as the dopamine-producing neurons die as the disease progresses. Treating DD mice with L-dopa restores locomotion and feeding for a few hours, but 24 hours later, all the dopamine has been metabolized and the mice are inactive and eat very little.
Because dopamine has been implicated in many processes, the feeding deficit of DD mice could be due to motor deficits in approaching or consuming the food, lack of normal reward processes, inability to perceive the salience of physiological and sensory signals, or a combination of the above. There are times when DD mice move about the cage as much as control mice, and when they eat they chew and swallow their food normally. These results discount motor deficits in finding or consuming food but do not distinguish between reward or salience deficits. When given the choice of water or a sweet solution, DD mice initiate bouts of feeding much less frequently than control mice, but they still have a clear preference for sucrose and consume more sugar water per bout than control mice. These results indicate that dopamine is not required for DD mice to respond to the rewarding aspects of sucrose (or saccharine). We also used DD mice to show that dopamine is not required for mice to experience or remember the rewarding properties of drugs such as morphine or cocaine. DD mice fail to respond to physiological signals that would normally elicit feeding. For example, normal mice eat in response to low glucose in the blood, to leptin deficiency, or to injection of compounds into the brain that stimulate feeding, such as neuropeptide Y; DD mice do not, however, eat in response to any of these treatments. Normal hungry mice readily learn to navigate a simple T-maze to obtain food rewards, but DD mice fail to master this task, even under conditions where they are as active as control mice; they do, however, learn where the food rewards are in the T-maze, even if they do not manifest knowledge without dopamine. Consequently, we hypothesize that dopamine is necessary for mice to respond to salient environmental or physiological signals to initiate goal-directed behaviors, such as seeking and consuming food, but it is not necessary for the experience of reward or learning to associate rewards with the environment in which they are encountered.
To determine where in the brain dopamine is required for normal feeding, we use gene therapy approaches to restore dopamine production to specific brain regions. Our most recent strategy involves using a new line of DD mice that have an inactive allele of the tyrosine hydroxylase gene that can be activated by the action of Cre recombinase. After canine adenovirus engineered to express Cre recombinase is injected into the striatum, it is retrogradely transported and activates the endogenous tyrosine hydroxylase gene in midbrain neurons that project to that striatal region. After one bilateral injection of virus into the dorsal striatum, the animals eat enough to sustain themselves without further L-dopa injections for the rest of their lives. We hypothesize that dopamine action in the dorsal striatum strengthens relevant cortical signals to facilitate initiation of goal-directed behaviors; e.g., dopamine may help integrate hunger signals with salient sensory inputs such as sight and smell of food to initiate food-seeking behaviors. Dopamine signaling restricted to the dorsal striatum is also sufficient to restore several cognitive functions, e.g., the ability to recognize a novel object or to remember how to navigate in a water maze.
Although L-dopa restores feeding by DD mice, direct-acting dopamine receptor agonists do not, even though both stimulate robust locomotion. Dopamine receptor agonists actually inhibit feeding by normal mice. Amphetamine-like drugs that release dopamine into the synapse also inhibit feeding by mice, and they have been used as diet pills to suppress appetite for decades. We suggest that the transient occupancy of dopamine receptors in response to phasic release of dopamine may be critical for feeding, while chronic occupancy of receptors after treatment with amphetamine or dopamine agonists may be inhibitory.
We have also made mice that lack NMDA receptors selectively in dopamine neurons. As a consequence, these mice have deficits in learning to associate pleasurable events with the environment in which they occur. For example, these knockout mice have impaired conditioned place preference for food, they are slow to learn the T-maze task, and they are slow to press a lever for food rewards. We suggest that the lack of NMDA receptors compromises burst-firing activity by dopamine neurons, that the consequent lack of transient spikes in dopamine release fails to signal salience of events in critical brain regions such as the hippocampus, and that this hinders learning.
A small population of neurons in the hypothalamus that make neuropeptide Y and agouti-related protein (NPY/AgRP neurons) has been implicated in the control of feeding because injection of either of these peptides into the brain stimulates feeding. However, inactivation of the genes for these peptides has little effect on feeding. We devised a strategy that allows us to kill these neurons by selectively expressing the receptor for diphtheria toxin (DT) in those cells and then injecting the mice with DT. Adult mice stop eating within a few days after these NPY/AgRP neurons are killed. The mice not only stop initiating feeding activity but also refuse to consume food placed directly in their mouth. We suspect that GABA is the critical neurotransmitter that NPY/AgRP neurons make because we can prevent starvation by continuous infusion of a GABA receptor agonist. We subsequently infused the GABA receptor agonist directly into different brain regions and discovered that the parabrachial nucleus is a critical region where excess neuronal activity leads to starvation. Interestingly, the parabrachial nucleus is activated in response to gastrointestinal malaise (food poisoning); so, it is likely that killing AgRP neurons mimics that physiological response.
We now know of two neural signaling pathways that become critical for feeding: the dopamine pathway to the dorsal striatum and a signaling pathway from the NPY/AgRP neurons. Our next challenge is to deduce how these signaling pathways facilitate feeding.
These experiments provide examples of how we combine genetics, gene therapy, and pharmacology to obtain insight into how and where specific neurotransmitters control behavior.