The circuits of motivation – is it all about dopamine?
In an attempt to understand the molecular bases of motivation, early studies in the 1970s and 1980s have proposed that the neuromodulator dopamine has a central role in controlling reward-related behaviors (Gerber et al., 1981; Ungerstedt, 1971; Wise et al., 1978). Dopamine is a monoamine produced mainly by neurons within the VTA and substantia nigra pars compacta (SNc) which innervate the cerebral cortex and limbic forebrain regions (Björklund and Dunnett, 2007). Initial studies using microdialysis measurements have suggested that dopamine release within the NAc (meso-limbic projection) is associated to self-stimulation reward (Fiorino et al., 1993), while stress specifically activates the meso-cortical projection (Thierry et al., 1976). Indeed, the function of dopamine in reward and aversive behaviors largely depend on the circuit connectivity of individual dopamine neurons (Bromberg-Martin et al., 2010; Lammel et al., 2011). Furthermore, depending on their specific identity, dopamine neurons undergo different experience-driven synaptic modifications which are instrumental for behavioral adaptations underlying opposing motivational states (Pignatelli and Bonci, 2015; Volman et al., 2013).
It is now largely established that dopaminergic neurons undergo phasic changes of activity in response to unexpected salient stimuli. Seminal studies in behaving monkeys using single-unit recordings have demonstrated that dopamine neurons firing increases in phasic and burst-like manner when an unexpected reward is presented (Schultz et al., 1997). Conversely, noxious foot pinch or foot shocks predominantly inhibit dopamine neurons burst activity (Ungless et al., 2004), although some neurons also show phasic excitation (Brischoux et al., 2009). After several conditioning sessions where animals are trained to associate a cue with the delivery of a reward, the firing of dopamine neurons and the subsequent release of dopamine in the NAc no longer occur at the time of reward delivery, but shift to the cue that predicts it (Schultz et al., 1997; Stuber et al., 2008). Moreover, the magnitude and direction of dopamine responses correlate with the predicted probability of the reward. Indeed, if an expected reward fails to occur or is smaller than expected, the activity of dopamine neurons is phasically inhibited, while if the reward is larger than expected dopamine neurons will fire at the time of reward delivery (Fiorillo et al., 2003; Schultz, 1998; Schultz et al., 1997; Tobler et al., 2003). These experiments led to the idea that dopamine neurons signal a reward prediction error, defined as the difference between the outcome of an expected and actual reward (Schultz 1998; Keiflin & Janak 2015). Likewise, dopamine neurons serve as ‘sensors’ for any deviation from the expectancies, representing a teaching signal for future cue-reward associations and reward-seeking behaviors. The short bursts of dopamine neurons trigger phasic release events in the NAc, where motivational signals are considered to be translated into a motor output leading to reward seeking behaviors (Day et al., 2007; McClure et al., 2003; Mogenson et al., 1980).
More recently, a causal relationship between dopamine neurons activity and conditioned learning have been provided by temporally precise and pattern-specific optogenetic modulation of VTA dopamine neurons activity. Indeed, physiologically relevant phasic activation of dopamine neurons expressing the light-activated cation channel rhodopsin 2 (ChR2) induces conditioned place preference behaviors while their inhibition via the hyperpolarizing and light-activated chloride ion pump Halorhodopsin (NpHR) produces conditioned place avoidance (Tan et al., 2012; Tsai et al., 2009). Moreover, these learning mechanisms and reward-oriented behaviors involve transient synaptic potentiation of excitatory transmission onto dopamine neurons (Stuber et al., 2008). Similar synaptic plasticity occurs following single drug or stressful experience (Saal et al., 2003; Ungless et al., 2001). However, in the case of prolonged drug administration synaptic changes in VTA dopamine neurons and the NAc become persistent contributing to cue-induced reinstatement of drug-seeking behaviors after drug withdrawal (Chen et al., 2008; Mameli et al., 2009).
Taken together these data demonstrate an important role of dopamine neurons particularly in reward but also in aversion encoding and suggest that synaptic and circuit modifications within the meso-cortico-limbic system are instrumental for motivated behaviors. However, the dopamine system is far more complex in that it receives multiple inputs both from local GABAergic interneurons and from more distal structures, many of which also participate in motivational encoding (Fig1). Moreover, dopamine neurons present heterogeneity in terms of their physiological properties, input-output connectivity and their responses to valenced stimuli (Lammel et al., 2014, 2011; Volman et al., 2013). These aspects need to be taken into account when considering dopamine neurons function in reward or aversion.
Although motivational states can be driven by rewarding and aversive conditions, during my thesis I focused mainly on the cellular substrates devoted to aversion processing.
Neurobiological substrates of aversion
Aversion is a common term to designate the behavioral reaction of avoidance or escape in response to negative, unpleasant, painful, stressful or fearful events or stimuli. Acute and chronic exposure to such aversive stimuli produces short or long-lasting synaptic, structural and circuit modifications that can often lead to neuropsychiatric disorders including depression, anxiety, post-traumatic stress disorder and addiction (Lüscher and Malenka, 2011; Nestler and Carlezon, 2006; Russo and Nestler, 2013). Many studies have implicated the VTA dopamine system in such modifications (Berton et al., 2006; Cao et al., 2010; Chaudhury et al., 2013; Krishnan et al., 2007; Lammel et al., 2011). However, it is now known that other structures directly or indirectly innervating the VTA, and often reciprocally connected, also contribute to different aspects of aversion (Fig1). The picture is even more complex given the large heterogeneity of different cell types forming local microcircuits within the VTA and challenging our understanding of the contribution of distinct pathways to aversion encoding and avoidance behaviors. With the advances of new technologies such as the generation of specific mouse lines, viral-based circuit- and cell type-specific mapping as well as optogenetics it is now possible to interrogate the implication of distinct neuronal circuits and neuronal subtypes for different aspects of aversion.
As a matter of example, specific cell type projections from the ventral bed nucleus of the stria terminalis (BNST) to GABAergic interneurons in the medial VTA have been involved in distinct motivational states. Indeed, glutamatergic BNST neurons, activating predominantly GABA neurons in the VTA, are excited by aversive conditions such as series of foot shocks or foot shock-associated cues. Moreover, ChR2-driven optogenetic activation of this pathway produces real time avoidance and elevated anxiety states(Fig2A) (Jennings et al., 2013). This is consistent with evidence that activation of GABAergic VTA neurons inhibits dopamine neurons activity leading to conditioned place avoidance behaviors (Tan et al., 2012). In contrast, GABA neurons of the BNST predominantly inhibit GABA neurons of the VTA and are preferentially silenced by foot shock exposure and foot shock-cues. Optogenetic activation of this pathway leads to real time preference and reward seeking behaviors (Fig2A) (Jennings et al., 2013). Consistently, VTA GABA neurons inhibition drives preference, reinforces behavior and reduces anxiety states associated with previous aversive experience (Jennings et al., 2013). Similarly to this, activation of dopamine neurons of the VTA also promotes reward-related behaviors (Adamantidis et al., 2011; Tsai et al., 2009), suggesting that GABA neurons in the VTA may serve to break locally the activity of dopamine neurons and therefore control the expression of motivational states (Fig2B). These data raise the question whether other inputs onto dopamine neurons or onto GABA neurons of the midbrain can control different aspects of aversive behaviors. Indeed, a major challenge in the field is to identify the precise input-output and functional organization of the circuits of aversion.
The lateral habenula: a control station of aversion
The lateral habenula (LHb) has gained considerable attention in the last decade because of its control on midbrain monoaminergic systems as well as for its crucial implication in aversion encoding and mood disorders (Hikosaka, 2010). My PhD work has focused on unraveling some of the circuit and synaptic functions of the LHb in the context of aversion and drug experience.
Phylogeny and anatomy of the LHb
The LHb is part of the habenular complex, which is located at the posterior-dorsal end of the epithalamus close to the midline and at the border of the third ventricle (Fig3). It comprises a lateral (LHb) and medial division (MHb) which are anatomically, morphologically and functionally distinct (Aizawa et al., 2011; Andres et al., 1999; Bianco and Wilson, 2009; Kim and Chang, 2005; Sutherland, 1982). The habenula is phylogenetically conserved among vertebrates (Aizawa et al., 2011; Bianco and Wilson, 2009). In birds, reptiles and mammals the MHb and the LHb are homologous to the dorsal and ventral habenulae respectively in fish and amphibians (Aizawa et al., 2011; Amo et al., 2010). The LHb and MHb relative proportion can vary across species. In mammals the size of the LHb is typically larger than the size of the MHb. Furthermore, a comparative analysis indicates that the proportional contribution of the LHb to the total surface of the habenular complex is considerably larger in humans compared to rats (Fig4A), supporting an increasing level of anatomical and functional specialization of the LHb throughout evolution (Díaz et al., 2011). The LHb can be divided into lateral and medial portions that receive inputs in a segregated manner (Fig5B). Each of these main divisions of the LHb can be further subdivided into smaller subnuclei based on distinct and topographic cell morphology and cytoarchitecture. Indeed, a total of ten subnuclei, five within the medial and five within the lateral LHb, have been described in rats based on morphological criteria and differential immunoreactivity for cellular markers (Andres et al., 1999; Geisler et al., 2003). The medial LHb comprises an anteriour subnucleus (LHbMA), located at the most rostral portion of the LHb, a superior (LHbMS), parvocellular (LHbMPc), central (LHbMC) and marginal subnuclei (LHbMMg). The lateral LHb contains a parvocellular (LHbLPc), magnocellular (LHbLMc), oval (LHbLO), marginal (LHbLMg) and basal subnuclei (LHbLB) (Fig4B). This subnuclear organization of the LHb seems to be well preserved in rodents since it is also present in the mouse (Wagner et al., 2014). In the human LHb five subnuclei have also been described although some discrepancies exist at the level of their relative position, size, morphology and cell organization compared to rodents (Díaz et al., 2011). Despite this detailed subnuclear organization and evidence indicating certain level of input-output-specific connectivity within some of the nuclei (Fig5B and Fig6B), LHb neurons seem to be relatively homogeneous in their basic cell properties (Weiss and Veh, 2011). A genetic profiling of LHb neurons would be required in order to identify potential discrimination criteria for different LHb subnuclei and would provide a tool to assess cell type-specific functions of LHb subpopulations in distinct aspects of motivated behaviors (Lecca et al., 2014; Proulx et al., 2014).
Inputs to the LHb
The habenular complex is positioned at the highway of a major information stream in the brain, connecting the limbic forebrain and basal ganglia regions with midbrain neuromodulatory systems. It receives most of its inputs through a fiber bundle called the stria medullaris and sends its projections to monoaminergic centers through the fasciculus retroflexus (Fig3B), which altogether form the diencephalic conduction system (Sutherland, 1982). Although the MHb and LHb receive their inputs and send their outputs using the same fiber tracts, they are differently innervated and target different brain nuclei. A potential connectivity between the MHb and LHb has been debated, but has never been anatomically or functionally proven.
The MHb receives inputs from septal and diagonal band of Broca (DBB) areas and sends in turn its axons to the interpeduncular nucleus of the midbrain (Herkenham and Nauta, 1979, 1977). In contrast, the LHb receives projections from the output structure of the basal ganglia – the entopeduncular nucleus (EPN; homologue of the globus pallidus interna in primates and humans) and from limbic forebrain regions including the lateral hypothalamus (LH), lateral preoptic area (LPO), ventral pallidum (VP), BNST, DBB, lateral septum (LS) as well as feedback projections from the laterodorsal tegmentum (LDT), the VTA and the dorsal and median raphe nuclei (DRN and MRN) (Fig5A) (Herkenham & Nauta 1977; Nagy et al. 1978; Parent et al. 1981; Kowski et al. 2008; Li et al. 2011; Tripathi et al. 2013; Lammel et al. 2012; Swanson 1982; Aghajanian & Wang 1977). A projection from the PFC has also been described (Kim and Lee, 2012; Li et al., 2011; Warden et al., 2012). Some of the inputs to the LHb represent a topographical organization potentially important for LHb functions (Fig5B).
Output connectivity of the LHb
LHb neurons are almost exclusively glutamatergic and long-range projecting (Kim and Chang, 2005; Li et al., 2011; Weiss and Veh, 2011). Although some studies have suggested the existence of local GABAergic neurons in the medial LHb, functional evidence about a potential LHb microcircuit is still lacking (Li et al., 2011; Zhang et al., 2016). Anatomical and physiological studies indicate that LHb neurons send their axons mainly to GABAergic neurons in the midbrain. Indeed, tracing experiments show that neurons located mainly within the lateral LHb project to the GABAergic rostro-medial tegmental nucleus (RMTg; (Balcita-Pedicino et al., 2011; Gonçalves et al., 2012; Jhou et al., 2009b; Meye et al., 2016; Sego et al., 2014), also called tail VTA (Kaufling et al., 2009; Perrotti et al., 2005). In contrast, LHb neurons originating from the medial aspect send their axons preferentially to monoaminergic nuclei (Gonçalves et al., 2012; Sego et al., 2014), where they form functional synapses with local GABAergic neurons (Lammel et al., 2012; Weissbourd et al., 2014) as well as with dopamine neurons of the VTA (Balcita-Pedicino et al., 2011; Lammel et al., 2012) or serotonin neurons in the caudal dorsal raphe (Dorocic et al., 2014; Sego et al., 2014) (Fig 6B). Further, it has been shown that individual neurons project either to VTA or to raphe nuclei without collateralizing, suggesting that single LHb neurons have distinct output targets (Bernard and Veh, 2012; Gonçalves et al., 2012; Li et al., 2011; Maroteaux and Mameli, 2012). In addition, the LHb has been recently shown to send axons to GABA neurons within the LDT and to orexin- and melanin concentrating hormone (MCH)-expressing neurons in the LH (González et al., 2016; Lammel et al., 2012; Yang et al., 2016). While the former has also been functionally proven, it remains to be confirmed whether LHb neurons establish functional synaptic connections in the LH (Fig6A).
Given that LHb neurons are of glutamatergic phenotype and that they project to a vast majority of GABA neurons in the midbrain it is plausible that they may disynaptically inhibit dopamine or serotonin neurons potentially providing aversive signals to the midbrain.
Table of contents :
THE CIRCUITS OF MOTIVATION – IS IT ALL ABOUT DOPAMINE?
NEUROBIOLOGICAL SUBSTRATES OF AVERSION
THE LATERAL HABENULA: A CONTROL STATION OF AVERSION
PHYLOGENY AND ANATOMY OF THE LHB
INPUTS TO THE LHB
OUTPUT CONNECTIVITY OF THE LHB
FUNCTION OF THE LHB IN REWARD AND AVERSION ENCODING
ROLE OF LHB OUTPUT FOR AVERSION PROCESSING
ROLE OF INPUTS TO THE LHB FOR AVERSION PROCESSING
DYSFUNCTION OF THE LHB: IMPLICATIONS IN DEPRESSION AND ADDICTION
LHB IN DEPRESSION
LHB IN ADDICTION
PROPERTIES OF LATERAL HABENULA NEURONS
CELL MORPHOLOGY AND ELECTROPHYSIOLOGY
FAST EXCITATORY TRANSMISSION VIA IONOTROPIC GLUTAMATE RECEPTORS
FAST INHIBITORY SYNAPTIC TRANSMISSION VIA IONOTROPIC GABAA RECEPTORS
MODULATION OF FAST EXCITATORY AND INHIBITORY TRANSMISSION: ROLE OF MGLURS
SYNAPTIC PLASTICITY IN THE LHB: A CELLULAR SUBSTRATE FOR MOTIVATED STATES IN
CELLULAR MECHANISMS IN THE LHB IN DEPRESSION
CELLULAR MECHANISMS IN THE LHB IN ADDICTION
CONTEXT AND OBJECTIVES FOR THE STUDIES
I. MGLUR-LTD AT EXCITATORY AND INHIBITORY SYNAPSES CONTROLS LATERAL HABENULA OUTPUT
II. COCAINE-EVOKED NEGATIVE SYMPTOMS REQUIRE AMPA RECEPTOR TRAFFICKING IN THE LATERAL HABENULA
WHICH INPUTS UNDERGO MGLUR-ELTD AND ILTD?
IS THERE ANY OUTPUT-SPECIFICITY FOR MGLUR-ELTD AND ILTD?
WHAT IS THE BEHAVIORAL RELEVANCE OF MGLUR-LTD IN THE LHB?
INDUCTION MECHANISMS AND CIRCUIT SPECIFICITY OF COCAINE-EVOKED PLASTICITY IN THE LHB
SYNAPTIC ADAPTATIONS AFTER COCAINE WITHDRAWAL
DISTINCT SYNAPTIC ADAPTATIONS CONVERGE TO INCREASE LHB NEURONAL AND BEHAVIORAL OUTPUT
ADDITIONAL PUBLICATIONS AND CONTRIBUTIONS