Activity of Striatal Direct and Indirect Pathways During Action Performance The striatum and downstream nuclei of the basal ganglia participate in control of action initiation, sequencing and termination, as well as the learning of these processes. The medium spiny projection neurons (MSNs) in striatum send inhibitory output to via two pathways, the direct and indirect pathways. The circuitry connecting direct and indirect pathways to cortical output indicates that direct pathway activation will increase activation of cortical neurons, while the indirect pathway inhibits these neurons. Thus, it has been suggested that the direct pathway serves as a go signal that increases action output, and the indirect pathway has a counteracting no-go function. Gross manipulation of activity in the two pathways has produced results consistent with this idea. However, to date there is no information on the activity of direct and indirect pathway neurons during performance of actions or action sequences. Working with Drs. Rui Costa of the Champalimaud Neuroscience Institute and Steven Vogel of the Laboratory of Molecular Physiology at NIAAA, we have developed a fiber optic system to perform Time Correlated Single Photon Counting (TCSPC) fluorimetry in the striatum in vivo. Coupling this technique with selective expression of a fluorescent calcium indicator protein (GCAMP3) in either direct or indirect pathway MSNs allowed us to measure calcium transients reflective of neuronal activation in the two pathways while an awake, freely-moving mouse performs a lever-pressing task. Fluorimetry and imaging experiments indicate the GCAMP3 transients occur when neurons are driven by strong synaptic input and fire short bursts of action potentials, consistent with MSN firing patterns observed during lever pressing tasks in vivo. We have observed spontaneous calcium transients in mice during performance of the lever pressing task, and transient frequency increases markedly when mice initiate a movement contraversive to the hemisphere where transients are measured. The precise timing of photon detection with the TCSPC system allowed us to determine that the increase in transients precedes movement initiation, suggesting that this neuronal activity predicts sequence onset, and may well contribute to initiation of the movement sequence. Similar timing and magnitude of calcium transient increases are observed in both direct and indirect pathway MSNs. Our findings suggest that the direct and indirect pathways work in concert during action sequence initiation, and thus they are not gross go/no-go drivers. The concerted activity is more likely necessary for proper action sequencing and/or simultaneous activation of desirable action sequences and suppression of competing sequences. The TCSPC system should have widespread utility for real-time measurement of activity in identified neuronal subtypes throughout the brain, and will allow us to correlate this activity with behavior in a temporally precise fashion. Dissecting endocannabinoid and CB1 receptor roles in goal-directed and habitual actions Instrumental learning allows animals to acquire new behaviors to adapt to environmental demands. Research in behavioral psychology has identified two separable processes that control instrumental learning and performance. One process, termed action-outcome (A-O, also known as goal-directed) learning, involves behavior that is sensitive to the proximal outcome of the action. The learning process known as stimulus-response (S-R, or habitual) learning is expressed as behavior that is driven by the context, independent of the proximal outcome. Previous studies from our laboratory group indicated that CB1 endocannabinoid receptors play a crucial role in S-R learning. Interestingly, recent findings suggest that chronic treatment with delta-9-tetrahydrocannabinol (THC, the predominant psychoactive ingredient in cannabis plants and a CB1 partial agonist) also disupts S-R learning. To produce a tonic increase in endocannabinoids, we have treated mice with inhibitors of monoacyl glycerol lipase (MAGL) or fatty acid amide hydrolase (FAAH), the enzymes that catalyze degradation of the endocannabinoids 2-arachidonoyl glycerol and arachidonolyethanolamide, respectively. Treatment with blockers of either enzyme disrupts S-R learning. Altogether, the findings to date suggest that CB1 activation produced by natural patterns of endocannabinoid release during learning are important for establishment of habitual action strategies. However, prolonged activation of CB1 that is not properly timed with training disrupts proper habit formation. We have also been examining roles of CB1 receptors present on different cellular elements within cortico-basal ganglia circuitry in A-O and S-R instrumental learning. To do this we have employed mice expressing loxP-flanked CB1 receptor genes, and interbred them with mice expressing the cre recombinase in different cortical and striatal neuronal subtypes. We can also inject a virus that induces Cre expression in different brain regions of the mice to knock out CB1 in specific brain regions and neuronal subtypes. Mice are then trained on an instrumental lever-pressing task in two separate environments, using reward schedules that foster A-O learning (random ratio training) or S-R learning (random interval training). Using a variety of breeding and viral expression approaches, we have gathered evidence that reducing CB1R expression in the orbitofrontal cortex (OFC) disrupts S-R learning. The CB1 receptors are almost exclusively expressed on presynaptic terminals, and thus our findings indicate that the receptors important for S-R learning most likely reside on terminals in brain regions to which the OFC projects. We have also injected Cre-inducing virus into the dorsolateral striatum to reduce CB1 expression in MSNs within this region that is implicated in SR learning. This manipulation did not alter task performance or habit learning when measured early in training, but testing after more extended training showed strong A-O-based performance and no evidence of use of an S-R strategy. These new findings are intriguing, and suggest that removing endocannabinoid signaling at axon terminals of MSNs (either in striatum or at downstream targets) weakens S-R control of behavior in a time or experience-dependent manner.
Showing the most recent 10 out of 48 publications
