Rui Costa., PhD. Acting Chief
National Institute on Alcohol Abuse and Alcoholism
National Institutes of Health
5625 Fishers Lane, Room TS-20
Bethesda MD 20892-9413
telephone: +1 301.443.1196
fax: +1 301.480.0466
e-mail: costarui@mail.nih.gov
Introduction
The Section on In Vivo Neural Function was established in January 2006 as a section of the Laboratory for Integrative Neuroscience of the National Institute on Alcohol Abuse and Alcoholism. We study the neurobiology of action in health and disease. To study actions is to study the way we do things, which is different than studying how we remember stimuli, or facts and events. Some actions are innate or pre-wired (like swallowing, breathing, even grooming). Others are learned through trial and error throughout life. We currently focus on understanding the processes mediating the latter.
Our overall goal is to understand how changes in molecular networks in the brain modify neural circuits to produce experience-dependent changes in actions. In order to understand how actions are learned through trial and error, we subdivided our experiments in different components, or specific goals:
Action initiation: how do we initiate and generate diverse actions (trial),
Action improvement: how do we improve the accuracy and speed of actions (through trial and error),
Actions and outcomes: how do we learn that particular actions lead to particular outcomes (goal of the action) and how do we form habits.
Understanding how we automate actions and form habits will hopefully point us to the mechanisms underlying drug seeking in addiction.
A growing body of evidence supports an important role of the basal ganglia in action initiation and selection, in skill learning, and in learning goal-directed actions and habits. Therefore, we centered our efforts on investigating the cortico-basal ganglia mechanisms underlying these three processes using an across-level approach, from molecules to circuits.
We chose to implement this integrative approach in mice because they combine the power of genetics, a mammalian brain with canonical cortico-basal ganglia loops that can generate and propagate oscillatory activity, and the possibility of accurately quantifying simple behaviors like action initiation (with EMG recordings or using inertial sensors) and stereotypic skill learning, and more elaborate behaviors like goal-directed actions.
Our research program will hopefully shed light on the mechanisms underlying the diversity of actions we perform, the automation of actions and the generalization rules or ways to do. Our research may also have important implications for understanding the relation between corticostriatal dysfunction and different neurodegenerative and psychiatric disorders.
Research Projects
Dissociable effects of dopamine on striatal firing rate and synchrony during acute dopamine dependent motor dysfunction
Our previous studies showed that dopamine depletion leads to both changes in firing rate and in synchrony in the basal ganglia (Costa et al., 2006). Since D1 and D2 receptors are preferentially expressed in striatonigral and striatopallidal medium spiny neurons, respectively, we investigated the relative contribution of lack of D1 and D2 receptor activation to the changes in striatal firing rate and synchrony observed after DA depletion. Similar to what was observed after dopamine depletion, co-administration of D1 and D2 antagonists to mice chronically implanted with multielectrode arrays in the striatum caused significant changes in firing rate, power of the local field potential oscillations, and synchrony measured by the entrainment of neurons to striatal local field potentials. However, although blockade of either D1 or D2 type receptors produced similarly severe akinesia, the effects on neural activity differed. Blockade of D2 receptors affected the firing rate of medium spiny neurons and the power of the LFP oscillations substantially, but it did not affect synchrony. In contrast, D1 blockade affected synchrony dramatically, but did not cause substantial effects on firing rate and LFP power (Fig. 1). Furthermore, D1 and D2 receptor blockade produced different effects in the medial and lateral regions of dorsal striatum, and dissociable effects on the activity of striatal interneurons. We observed no relation between firing rate changes and LFP entrainment, suggesting that these two phenomena are independent of each other, and that lack of D1 - or D2-type receptor activation an exert independent yet interactive pathological effects during the progression of Parkinson’s Disease.
Dynamic reorganization of striatal circuits via region and pathway-specific plasticity during the acquisition and consolidation of a skill
Learning to execute and automate certain actions is essential for survival. The learning of novel skills by trial and error, like riding a bicycle or playing a piano, is characterized by an initial stage of rapid improvement in performance, followed by a phase of more gradual improvements as the skills are consolidated and performance asymptotes (Kargo and Nitz, 2004; Karni et al., 1998; Miyachi et al., 2002; Miyachi et al., 1997). The different phases of skill learning have distinct behavioral and physiological hallmarks (Karni et al., 1998; Kleim et al., 2004; Muellbacher et al., 2002; Shiffrin and Schneider, 1977). For example, the early fast phase is susceptible to interference, while the later, more automatic phase is more resistant to interference (Shiffrin and Schneider, 1977). After the initial phase of acquisition, the memory of how to do things is gradually consolidated and for well-learned skills it can last a lifetime.
Previous studies have shown changes in neural activity in the striatum, the major input nucleus of the basal ganglia, during motor and procedural learning (Barnes et al., 2005; Brasted and Wise, 2004; Carelli et al., 1997; Doyon et al., 1996; Jenkins et al., 1994; Ungerleider et al., 2002) . Some studies also suggested that the striatal circuits and processes engaged during the early and late phases of skill learning may differ (Costa et al., 2004; Miyachi et al., 2002; Miyachi et al., 1997). For example, the dorsomedial or associative striatum (DMS, roughly homologous to the caudate in primates), which receives input primarily from association cortices such as the prefrontal cortex (McGeorge and Faull, 1989; Voorn et al., 2004), seems to be preferentially involved in the initial stages of visuomotor learning and during the rapid acquisition of action- outcome contingencies (Miyachi et al., 2002; Miyachi et al., 1997; Yin et al., 2005b). On the other hand, the dorsolateral or sensorimotor striatum (DLS, roughly homologous to the putamen in primates), which receives inputs from sensorimotor cortex (McGeorge and Faull, 1989; Voorn et al., 2004), is critical for the more gradual acquisition of habitual and automatic behavior (Miyachi et al., 2002; Miyachi et al., 1997; Yin et al., 2004).
We have recently recorded the neural activity in the DMS and DLS striatal regions during the different stages of skill learning in vivo, and found that the task-related activity in these striatal regions differed during the acquisition and consolidation of a novel skill, with the DMS or associative striatum being engaged during the early phase, and the DLS or sensorimotor striatum during the late phase. We confirmed the differential involvement of these striatal regions in the different stages of skill learning using selective excitotoxic lesions of the dorsal striatum. We also investigated whether the changes in striatal neural activity observed during skill learning could be mediated by synaptic plasticity or excitability changes in medium spiny projection neurons in the dorsal striatum by using an ex vivo approach, and found that learning was accompanied by long-lasting changes in glutamatergic transmission. These changes evolved dynamically during the different phases of skill learning: changes in the DMS were predominant early in training, while changes in the DLS evolved only after extensive training. In summary, our previous studies indicated that during the automation or consolidation of a skill, there is extensive potentiation of glutamatergic transmission in medium spiny neurons from the dorsolateral striatum, and that this region is necessary for the performance of automatized actions and skills.
Neurons from the dorsolateral striatum, however, do not all project to the same downstream basal ganglia structures. Medium spiny neurons projecting preferentially to the substantia nigra (striatonigral or ‘direct’ pathway), and MSNs projecting to the external globus pallidus (striatopallidal or ‘indirect’ pathway) have different dopamine receptor expression, different physiological properties, and different plasticity mechanisms (Gerfen et al., 1990; Kreitzer and Malenka, 2007; Shen et al., 2008; Shen et al., 2007). Until recently, LTP induction in the striatum was thought to always depend on D1-receptor activation (Kerr and Wickens, 2001), which suggested that it would occur preferentially in striatonigral neurons, which preferentially express D1 receptors, and less in striatopallidal neurons, which express predominantly D2 receptors. However, recent studies have shown that LTP can occur in both types of neurons, just by different mechanisms (Shen et al., 2008). Therefore, in order to understand the mechanisms underlying the consolidation and automatization of skills, it is crucial to determine whether the long-lasting potentiation observed in DLS after extended training occurs in striatonigral or striatopallidal neurons, or in both. We can do this by recording from MSNs in the DLS of D2- EGFP mice and D1-EGFP mice that are extensively trained on the rotarod or naïve (Fig. 2). These mice allow the visualization of neurons that express D1 receptors, which are almost exclusively striatonigral neurons, and neurons that express D2 receptors, which are almost exclusively striatopallidal neurons, respectively (Fig. 2). Using the D2-EGFP mice we obtained preliminary data indicating that extensive rotarod training resulted only in a slight potentiation in the non-D2 expressing neurons (putative striatonigral neurons, or direct pathway), but in a much greater potentiation in the D2-expressing neurons (striatopallidal or indirect pathway) in the DLS of extensively trained animals (Fig. 2). Concomitantly, the performance of the skill became less dependent upon the activation of D1 receptors.
These findings demonstrate that, as a skill becomes automatized, region-specific and pathway-specific plasticity sculpt the circuits involved in the performance of the skill, and could elucidate why in Parkinson’s disease voluntary movements are more affected than automatized movements.
Molecular and circuit mechanisms underlying goal-directed actions and habits
We can learn to perform particular actions to obtain specific outcomes in our environments through a process of trial and error. Initially, these actions are goal–directed, and their performance is highly sensitive to changes in the incentive value of the outcome, and also to changes in the contingency between the action and the outcome. With repetition, however, actions can become not only more efficient but also more automatic and habitual (Dickinson, 1985; Foerde et al., 2007; Miyachi et al., 1997). Previous studies in rats have shown that extended training on an instrumental task where animals lever press for particular food reinforcements can lead to a shift from goal-directed responding, which is sensitive to changes in the value of the outcome, to habitual responding which is insensitive to outcome devaluation and can be elicited by antecedent stimuli (Adams, 1982; Adams and Dickinson, 1981). Interestingly, shifts from goal-directed to habitual responding can be produced not only by extended training, but also by different schedules of reinforcement, with random interval schedules favoring the formation of habits compared with random ratio schedules (Adams and Dickinson, 1981; Dickinson, 1985; Dickinson et al., 1983). Using an operant lever pressing task and a reversible devaluation paradigm by sensory-specific satiety before a probe test, we showed that in mice random interval schedules also favor habit formation compared with random ratio schedules.
The neuroanatomical circuits that support the learning and the performance of goal-directed actions are different than those supporting the formation of habits (Balleine and Dickinson, 1998; Yin and Knowlton, 2006). The acquisition of goal-directed actions appears to rely on the associative cortico-basal ganglia circuit involving the dorsomedial or associative striatum (Yin et al., 2005a; Yin et al., 2005b), while the formation of habits depends upon the dorsolateral or sensorimotor striatum.
What are the molecular mechanisms underlying goal directed actions and habit formation? Dopamine may have multiple roles in this process (Costa, 2007; Hitchcott et al., 2007; Wickens et al., 2007). Amphetamine sensitization has been shown to lead to increased predisposition for habit formation (Nelson and Killcross, 2006). Endocannabinoid release in the striatum is modulated by dopamine signaling (Giuffrida et al., 1999; Kreitzer and Malenka, 2005; Yin and Lovinger, 2006) and is necessary for the induction of long-term depression (LTD) (Gerdeman and Lovinger, 2001; Gerdeman et al., 2002). Endocannabinoid signaling through the cannabinoid receptors type 1 (CB1) has been implicated in reward and addiction (Caille et al., 2007; Cossu et al., 2001; De Vries et al., 2001; Di Marzo et al., 2001; Gerdeman et al., 2003; Hansson et al., 2007; Houchi et al., 2005; Sanchis-Segura et al., 2004; Wang et al., 2003). The expression of CB 1 receptors in the brain displays an interesting gradient across the striatum, with a very high level of expression in the dorsolateral striatum (Gerdeman et al., 2003; Herkenham et al., 1991), at both excitatory and inhibitory terminals (Uchigashima et al., 2007). Interestingly, recent studies have shown that amphetamine sensitization depends upon endocannabinoid signaling in the dorsal striatum (Corbille et al., 2007).
We therefore decided to investigate if endocannabinoid signaling is involved in habit formation using mice with genetically targeted mutations in the CB1 gene (Zimmer et al., 1999), and found that CB1 mutant mice had decreased predisposition for habit formation (Fig. 3). Furthermore, we confirmed that CB 1 blockade specifically during training also blocked habit formation (Hilario et al., 2007), indicating that the effects observed in CB1 mutant mice were not due to developmental changes. Our results indicate that endocannabinoid signaling through CB1 receptors is critical for habit formation.
The role of NMDA receptors in dopamine neuron bursting and habit formation
At the molecular level, dopamine (DA) has been implicated in both voluntary actions (Hitchcott et al., 2007) and habits (Faure et al., 2005), and we have recently uncovered that dopaminergic neurons increase firing rate and burst just before animals perform a particular action that leads to a specific outcome. Since the findings of Schultz (Ljungberg et al., 1992; Schultz, 1986), researchers have been hypothesizing about what mechanisms underlie the ability of dopaminergic neurons to burst in response to unexpected rewards and reward predicting stimuli. NMDA receptors, which are critical for LTP of the glutamatergic transmission onto dopaminergic cells, seem to be critical for dopaminergic neuron bursting (Kitai et al., 1999; Overton and Clark, 1997). Our data indicate that deletion of the essential NMDA subunit NR1 in DA neurons dramatically decreases bursting, and consequently impairs phasic increases in firing in DA neurons before lever pressing.
However, dopaminergic neurons from the Ventral Tegmental Area (VTA) and the Substantia Nigra pars compacta (SNc) project to different regions of the striatum and the cortex, and the specific role of VTA and SNc dopamine in goal-directed actions and habits has not been clarified. We therefore generated animals which lack the essential NR1 subunit of the NMDA by crossing a TH-Cre line with animals with the NR1 subunit flanked by loxp sites (Fig. 4). Our data suggests that deletion of NR1 in VTA and SNc DA neurons specifically leads to distinct behavioral phenotypes: whereas bursting in SNc DA neurons seems to be critical for habit formation, bursting in VTA DA neurons seems to promote habitual behavior instead of goal- directed actions.
New tools
Mouse genetics tools to investigate the functional microcircuitry cortico-basal ganglia circuits.
In order to be able to selectively ablate particular cell types within the cortico-basal ganglia circuitry we are taking advantage of mice generated by Dr. Waisman (iDTR – inducible diphtheria toxin receptor), in which Cre-mediated excision of a STOP cassette renders the cells able to express the diphtheria toxin receptor, and therefore makes those cells sensitive to injections of diphtheria toxin (Buch et al., 2005)
Although the iDTR system has great spatial resolution, it lacks temporal resolution and it is not reversible. In order to be able to activate or inactivate neurons with great temporal precision and in specific neuronal populations, we have generated mice carrying light activated channels (channelrhodopsin2 and halorodhopsin) in which expression of the channels is dependent on Cre-mediated excision of a STOP cassette. Channelrhodopsin2 is a light-gated cation channel obtained by Chlamydomonas reinhardtii, while halorhodopsin is a light-driven chloride pump homologous to bacteriorhodopsin obtained from Natronomonas pharaonis. Expression of these channels in neurons permits an exquisite control of their activity (activation or inactivation) with light (Boyden et al., 2005; Zhang et al., 2007).
In vivo detection of changes in gene expression in the brain of awake behaving mice using fiber optics.
To monitor real time local gene expression in awake and freely moving mice at different stages of behavioral training, in collaboration with Dr. Vogel’s section, we have designed a fiber optics system, which uses the principles of Time Correlated Single Photon Counting (TCSPC) to allow acquisition of emission spectra as well as lifetimes. The system is based on delivering pulse laser light of 473 nm through a single mode optical fiber to excite a fluorescent protein, while a multimode fiber is used to collect the excitation emission. These optic fibers can be implanted in the brain, and we have been able to detect GFP expression in subcortical brain structures, like the striatum of reporter mice. We are currently using Arc-GFP knock-in mice (Wang et al., 2006) to monitor on-line the expression of the immediate early gene Arc (activity-regulated cytoskeleton¬associated protein), during amphetamine sensitization.
New microelectrode arrays for targeting multiple cortical and subcortical structures.
We are constantly building new types of microelectrode arrays to be able to record from different brain structures. We mainly probe new types of wire (platinum plated tungsten, platinum iridium) or new shapes to target simultaneously cortical and subcortical, or different subcortical structures.
The generation of these new tools will not only allow us to pursue our research interests, but also benefit all researchers interested in similar problems.
Publications 2007-2008:
Yin HH, Prasad-Mulcare S, Hilario MR, Clouse E, Davis MI, Hansson A, Lovinger DM, Costa RM. Dynamic reorganization of striatal circuits during the acquisition and consolidation of a skill.
Nature Neurosci 12:333-41, 2009.
Hilario MR, Costa RM. High on Habits.
Frontiers Neurosci 2,2:208-21, 2008.
Groszer M, Keays DA, Deacon RMJ, De Bono JP, Prasad-Mulcare S, Gaub S, Baum MG, French CA, Nicod J, Coventry JA, Enard W, Fray M, Brown SDM, Nolan PM, Paabo S, Channon PM, Costa RM, Eilers J, Ehret G, Rawlins JNP, Fisher SE. Impaired synaptic plasticirty and motor learning in mice with a point mutation implicated in human speech deficits.
Current Biol 18: 354-62, 2008.
Cui Y, Costa RM, Murphy GG, Elgersma Y, Zhu Y, Gutmann DH, Parada LF, Mody I, Silva AJ. Neurofibromin regulation of Ras/ERK signaling modulates GABA release and learning.
Cell 135:549-60, 2008.
Hilario MR, Clouse E, Yin HH, Costa RM. Endocannabinoid signaling is critical for habit formation.
Frontiers Integr Neurosci 1:6, 2007.
Prior key publications:
Costa RM, Lin SC, Sotnikova TD, Cyr M, Gainetdinov RR, Caron MG, Nicolelis MAL. Rapid alterations in corticostriatal ensemble coordination during acute dopamine-dependent motor dysfunction.
Neuron 52:359-69, 2006.
Costa RM, Drew C, Silva AJ. Notch to Remember.
Trends Neurosci 28, 429-35, 2005.
Costa RM, Cohen D, Nicolelis MAL. Differential corticostriatal plasticity during fast and slow motor skill learning in mice.
Current Biol 14:1124-34, 2004.
Costa RM, Honjo T, Silva AJ. Learning and memory deficits in Notch mutant mice.
Current Biol 13: 1348-54, 2003.
Costa RM, Silva AJ. Mouse models of Neurofibromatosis type I: Bridging the GAP.
Trends Mol Med 9: 19-23, 2003.
Costa RM, Federov NB, Kogan JH, Murphy GG, Stern J, Ohno M, Kucherlapati R, Jacks T, Silva J. Mechanism for the learning deficits in a mouse model of neurofibromatosis type
1. Nature 415:526-30, 2002.
Costa RM, Yang T, Huynh DP, Pulst SM, Viskochil DH, Silva AJ, Brannan CI. Learning deficits, but normal development and tumor predisposition, in mice lacking exon 23a of the Neurofibromatosis type I gene.
Nature Genet 27: 399-405, 2001.
Updated: May 2009