GABA_B Receptor Functions in the Mesolimbic Dopamine System

Abstract

GABA_B receptors are expressed in neurons of the dopamine system where they bidirectionally modulate activity and release of glutamate, GABA, and dopamine itself. Dopamine has many functions including signaling of salient external stimuli and prediction errors that optimize decision-making, as well as motivation and initiation of movement. GABA_B receptors thus exert a second-order modulation, with effects on locomotion, motivation, and reward learning. Moreover, recent findings indicate that neuronal activity may induce a plasticity of GABA_B receptor signaling. In this chapter, we review the structural and functional features of GABA_B receptor signaling in the dopaminergic system, from subcellular specialization to plasticity and fine-tuning of mesolimbic circuits. Beyond a physiological role GABA_B receptors may also affect disease, such as addiction. GABA_B receptors may therefore constitute an interesting target for pharmacological interventions to treat this condition.

1 The Midbrain DA System

1.1 Projection- and Input-Specific DA Populations

The midbrain dopamine (DA) system has its origin in two major nuclei: the Ventral Tegmental Area (VTA ) and the Substantia Nigra compacta (SNc), which are anatomically and functionally distinct. The VTA DA neurons , close to the midline at the caudal end of the midbrain, project most notably to the Nucleus Accumbens (NAc) and Prefrontal Cortex (PFC), forming the mesocorticolimbic system. Increased DA activity is typically associated with positive outcome, whereas DA inhibition is aversive (Schultz 1998; Tan et al. 2012; van Zessen et al. 2012). DA neurons of the SNc are located lateral to the VTA , project to the dorsal striatum (the mesostriatal system) where they modulate locomotion. DA neurons also receive reciprocal inputs from their striatal target; the dorsal striatum mostly projects to the SNc, the NAc preferentially targets the VTA . Additionally, nuclei associated with motivational states like the dorsal raphe and the lateral hypothalamus selectively target the VTA over the SNc, further differentiating these two populations. In comparison, both DA nuclei receive equal projections from the cortex (Watabe-Uchida et al. 2012).

1.2 VTA Microcircuit

The VTA contains three cell types: DA, γ-aminobutyric acid (GABA)-ergic, and glutamatergic. The DA neurons form 80 % of the total population. The GABAergic neurons are fewer and provide a major inhibitory input to DA neurons (Johnson and North 1992a). The functional relevance of this microcircuit was initially demonstrated by showing that μ-opioid receptors specifically inhibit GABA neurons , leading to the disinhibition of DA neurons . Lastly, a small population of glutamatergic neurons forms local synapses with DA and GABA neurons (Dobi et al. 2010).

The recent development of transgenic mouse lines and optogenetics has allowed investigators to manipulate neuronal populations of the VTA in vivo and measure their impact on behavior, further validating the functional organization of the VTA microcircuit. For example, direct activation of DA neurons is sufficient to drive reward-related behaviors (Tsai et al. 2009; Adamantidis et al. 2011). In contrast, activation of GABA neurons inhibits DA firing and is sufficient to drive aversive behavioral responses (Tan et al. 2012; van Zessen et al. 2012). Finally, activation of the local glutamatergic population excites DA neurons and is rewarding (Wang et al. 2015).

1.3 Inputs to VTA Neurons

In the VTA , both DA and GABAergic populations receive quantitatively similar proportions of excitatory and inhibitory inputs from outside the VTA (Beier et al. 2015). Major excitatory inputs include the frontal cortex, central amygdala, hippocampal septum, lateral habenula, and dorsal raphe, while major inhibitory inputs comprise the Nac, ventral pallidum, and globus pallidus. However, specific inputs from a given region may be qualitatively biased toward one cell type. For example, the GABAergic projection from the NAc preferentially inhibits VTA GABA neurons , and eventually disinhibits DA neurons (Xia et al. 2011; Bocklisch et al. 2013). Therefore, the ultimate DA output is controlled by balanced modulation of both DA and GABA neurons (Fig. 8.1).

Fig. 8.1

Cell-type-specific GABA_B receptor signaling in the VTA . (a) VTA microcircuit depicting GABA neuron (green) inhibition onto DA neuron (red), and cell-type-specific expression of proteins involved in GABA_B receptor signaling. (b) Example voltage clamp recordings of maximal baclofen-evoked currents (300 μM) in DA (red, upper panel) and GABA (green, lower panel) neurons. Currents are blocked by GABA_B receptor antagonist CGP54626 (2 μM). (c) Dose–response curve for baclofen in DA (red) and GABA (green) neurons, showing the difference in EC50. Insets indicate relevant proteins and their effect on coupling efficiency. (d) Schematic representation of DA (red) and GABA (green) neuron firing rate to increasing concentrations of baclofen [Traces in (b) reproduced from Labouèbe et al. 2007]

1.4 GABA_B Receptors in the Midbrain DA System

GABA_B receptors are enriched across DA neurons of the midbrain, and also found in VTA GABA neurons , whereas their presence in VTA glutamatergic neurons remains to be established. It has been proposed that local and long-range inhibitory inputs to DA neurons have a specialization with synapses that contain only GABA_A or GABA_B receptors, respectively (Sugita et al. 1992; Johnson and North 1992b; Cameron and Williams 1993). However, recent investigations have suggested the presence of GABA_A receptors at all inhibitory synapses in DA neurons . For example, optogenetic activation of long-range projections from the NAc evokes GABA_A receptor-mediated inhibitory postsynaptic currents (IPSCs) in VTA DA neurons (Bocklisch et al. 2013). Additionally, in vivo electrical stimulation of inhibitory nuclei targeting the SNc evokes a combined GABA_A, GABA_B receptor-mediated inhibitory postsynaptic potential (IPSP) (Brazhnik et al. 2008). Whether VTA GABA neurons can activate GABA_B receptors in DA neurons remains to be investigated. Dendritic GABA_B receptors are typically extrasynaptically located (Boyes and Bolam 2003; Koyrakh et al. 2005), therefore a single pulse of stimulation evokes mostly GABA_A receptor-mediated currents in the VTA . Higher intensity and frequency stimulation is usually required to drive GABA spillover outside the synaptic cleft and reach extrasynaptic GABA_B receptors, suggesting that only sustained GABA release engages GABA_B receptor signaling, independently of the synaptic input.

2 GABA_B Receptor Signaling in the Midbrain

2.1 Common Features of GABA_B Receptor Signaling

2.1.1 GABA_B Receptor Structure and Main Effectors

Functional GABA_B receptors are heterodimers that require the assembly of one GABA_B1 isoform (1a or 1b) with GABA_B2 (Jones et al. 1998; White et al. 1998; Kaupmann et al. 1998; see also Chap. 4 of this book). The GABA_B1a subunit differs from GABA_B1b by the expression of two sushi domains, which guide trafficking to axon terminals. Accordingly, GABA_B1a/2 receptors are mainly found presynaptically, whereas postsynaptic compartments mostly express GABA_B1b/2 dimers (Vigot et al. 2006). Because GABA_B receptor subunit composition is similar across most neurons, any variability in GABA_B receptor signaling is likely to be accounted for by associated proteins.

Both GABA_B1a/2 and GABA_B1b/2 similarly couple to G_i/o protein. Gα inhibits the adenylate cyclase pathway while Gβγ gates ion channels. In presynaptic boutons, GABA_B receptors decrease neurotransmitter release via Gβγ-mediated closing of Voltage-Gated Ca²⁺ (CaV) channels and direct interaction with the release machinery. Postsynaptically, the major effect of Gβγ is the opening of G protein-activated inwardly rectifying K⁺ channels (GIRK, also known as Kir3), which hyperpolarize the membrane and decrease neuronal excitability (Bettler et al. 2004; Lüscher et al. 1997). (For a more detailed description of GABA_B receptor signaling , see Chap. 6 of this book).

2.1.2 Macromolecular Signaling Complex and Associated Proteins

Accumulating evidence suggests that GABA_B receptors and associated proteins form macromolecular complexes to promote the specificity of spatial and temporal control of signaling (Doupnik et al. 2004). Electron microscopy, co-immunoprecipitation, as well as bioluminescence and fluorescence resonance energy transfer experiments initially revealed close interactions between GABA_B and GIRK subunits, suggesting the existence of those complexes (David et al. 2006; Fowler et al. 2007; Ciruela et al. 2010). These approaches also identified the interaction of GABA_B receptors and GIRK channels with Regulator of G protein signaling (RGS) proteins (Fowler et al. 2007). RGS provide powerful modulation of the coupling efficiency between GPCR and effectors by accelerating the activation and deactivation kinetics of G proteins, and have also been proposed to influence signaling desensitization, a process of signal attenuation upon receptor activation (Mutneja et al. 2005). For most G protein-coupled receptors (GPCR), desensitization is expressed by rapid agonist-induced internalization of the receptor. However GABA_B receptors behave differently, and desensitization mechanisms remain elusive (Couve et al. 2002). More recently, affinity purification assays combined with quantitative mass spectrometry from native membrane preparations allowed the identification of four KCTD (K⁺ Channel Tetramerization Domain-containing) proteins as auxiliary subunits of the GABA_B receptors (Schwenk et al. 2010). KCTDs bind to the GABA_B2 subunit and modulate coupling, onset kinetics and desensitization of receptor signaling (Turecek et al. 2014). Further functional proteomics analyses revealed co-assembly with unsuspected partners like HCN (hyperpolarization-activated cyclic nucleotide-gated) channels. Surprisingly, no physical interaction was found between GABA_B receptors or KCTD proteins and GIRK channels despite being major effector of GABA_B receptors (Schwenk et al. 2016).

Altogether, there are many proteins that actively shape GABA_B receptor signaling and even more possible combinations to build macromolecular complexes. Such abundance allows region and cell-type-specific variability of GABA_B receptor functional expression, which is highly relevant in the midbrain DA system.

2.2 Cell-Type-Specific GABA_B Receptor Signaling in the Midbrain

2.2.1 Differences in GABA_B Receptor Signaling in VTA DA and GABA Neurons

In the VTA both DA and GABA neurons express GABA_B receptors, however there are a number of functional differences that can be assessed by recording from VTA neurons in acute brain slices and applying baclofen, a GABA_B receptor agonist , to evoke GIRK currents. First, DA neurons have a much lower sensitivity to baclofen, measured by an EC50 value (the concentration required to reach 50 % of the maximal response) ten-fold higher than in GABA neurons . Yet, the peak current evoked by a saturating concentration of baclofen is around three-fold larger in DA neurons . Lastly, in DA neurons , baclofen-evoked currents desensitize by around 40 % over minutes, whereas GABA neurons show a steady current upon continuous baclofen application (Cruz et al. 2004). These three differences in GABA_B receptor signaling have been partially explained by differential composition and expression levels of proteins involved in GABA_B receptor macromolecular signaling complex, which we review here (Table 8.1). They also provide the mesocorticolimbic system with a bidirectional handle on DA output, which will be discussed later.

Table 8.1 Cell-type-specific expression of proteins modulating GABA_B receptor signaling in the midbrain DA system

2.2.2 Cell-Type-Specific GIRK Channel Expression

VTA DA and GABA neurons express a specific set of GIRK channel subunits. Of the four known GIRK subunits, GIRK1, 2, and 3 are found in the brain, whereas GIRK4 expression is low and probably functionally irrelevant (Wickman et al. 2000). GIRK channels assemble as heterotetramers formed by two pairs of two subunits (1/2, 1/3, 2/3), or GIRK2 homotetramers, with GIRK2 available in three isorforms, GIRKa-c (Lüscher and Slesinger 2010). GIRK2-containing channels seem to be most commonly expressed throughout the brain, as knocking-out GIRK2 ablates almost entirely GIRK-mediated currents in many different cell types, including midbrain DA and GABA neurons (Lüscher et al. 1997; Arora et al. 2010; Cruz et al. 2004; Hearing et al. 2013).

VTA GABA neurons express GIRK1, 2c, and 3 subunits, whereas VTA and SNc DA neurons do not express GIRK1, as observed by single cell RT-PCR and electron microscopy (Inanobe et al. 1999; Cruz et al. 2004; Labouèbe et al. 2007). The presence of GIRK1 partially explains why GABA_B receptors have a higher coupling efficiency in GABA neurons. Indeed, ectopic expression of GIRK1 in VTA DA neurons reduces the EC50 value for baclofen-evoked GIRK currents. Conversely, the same result is obtained by knocking out GIRK3 from DA neurons , suggesting that GIRK1 increases, whereas GIRK3 decreases the coupling efficiency of GABA_B receptors (Labouèbe et al. 2007).

Cell-type-specific GIRK channel composition is also found among the midbrain DA population. VTA DA neurons express mostly GIRK2c/3 heteromers, whereas SNc DA neurons express both GIRK2a and 2c isoforms and form GIRK2a/2c homomers (Inanobe et al. 1999). Although not directly compared, maximal currents and desensitization rate seem similar overall in both populations (VTA : Labouèbe et al. 2007; SNc: Arora et al. 2010). Accordingly, GIRK3 knockout does not affect amplitude or desensitization in SNc DA neurons , although the latter was not systematically quantified (Arora et al. 2010). Thus, it appears that GIRK channel composition broadly influences coupling efficiency, whereas a potential role for GIRK1 in desensitization remains to be explored .

2.2.3 Cell-Type-Specific RGS Expression

Another major difference observed so far is the selective expression of RGS2 in VTA DA neurons versus GABA neurons . Whereas RGS4 is present in both, RGS2 associates with the GIRK3 subunit and acts as a brake on coupling efficiency of GABA_B receptor signaling in DA neurons . Either inhibiting RGS proteins or knocking out RGS2 results in lower EC50 values for baclofen-evoked GIRK currents in DA neurons , similar to GIRK3 knockout. RGS2 and GIRK3 double knockout do not further decrease the EC50 value, suggesting that RGS2 decreases the coupling efficiency through its interaction with GIRK3 (Labouèbe et al. 2007).

2.2.4 KCTD

Notably, neither GIRK3 knockout, RGS2 knockout nor their combination in DA neurons fully recapitulates the high coupling efficiency observed in GABA neurons . Therefore, other components of the signaling complex must account for this cell-type-specific aspect GABA_B receptor signaling. The missing piece of the puzzle could very well be found in a newly identified function of KCTD proteins, as auxiliary subunits for GABA_B receptors (Schwenk et al. 2010). Affinity purification assays have identified four members of the KCTD family that interact with GABA_B2: KCTD8, 12, 12b, and 16. In cultured hippocampal neurons, all four KCTD sharpen the rise-time of GIRK currents. KCTD12 and 12b also shorten current onset, and most interestingly, strongly increase the desensitization rate. The acceleration of signal transduction is mediated by the binding of KCTDs to G proteins, keeping them close to and stabilizing GABA_B receptors. The pronounced desensitization is due to the ability of KCTD12 and 12b, but not 8 or 16, to bind activated Gβγ proteins and uncouple them from GIRK channels (Turecek et al. 2014). KCTD12 and 16 are enriched in the VTA (Metz et al. 2011), however cell-type-specific expression patterns remain so far unexplored in this same region. Selective expression of KCTD12 in DA neurons is a likely molecular candidate to explain the signature desensitizing GIRK currents in these cells, and differential KCTD expression may also explain discrepancies in coupling efficiency in GABA versus DA neurons .

2.2.5 Sorting Nexin 27

The expression of GIRK2c/3 heteromers in VTA DA neurons allows their targeting by Sorting Nexin 27 (SNX27), an endosomal protein involved in the trafficking of several G protein-coupled receptors (GPCR) (Joubert et al. 2004; Lauffer et al. 2010). SNX27 contains a PDZ domain that selectively associates with the C-terminal PDZ-binding motif on GIRK2c and GIRK3 subunits (Balana et al. 2013; Lunn et al. 2007). In mice lacking SNX27 only in DA neurons , maximal baclofen currents are blunted, presumably due to a decreased surface expression of GIRK channels (Munoz and Slesinger 2014). SNX27 is also expressed in non-DA neurons of the VTA , and might therefore fulfill a similar function in GABA neurons .

2.2.6 HCN Channels

Midbrain DA neurons , but not GABA neurons express HCN channels, which enable the pacemaker-like tonic firing of SNc DA neurons (Neuhoff et al. 2002; Khaliq and Bean 2010). HCN channels co-assemble with GABA_B2 via KCTD16. It has been suggested that GABA_B receptor activation leads to HCN channel opening over several hundreds of milliseconds, closely following the timescale of G protein signaling. HCN opening potentially decrease the amplitude and accelerates the decay of GIRK-mediated postsynaptic potentials (Schwenk et al. 2016). The mechanistic underpinnings of this phenomenon remain to be explored. For example, HCN channels are typically activated by hyperpolarization, which is driven by GIRK channels. Therefore it is possible that HCN channel opening is unrelated to GABA_B receptor signaling and simply responds to GIRK activation, providing a depolarizing drive. However HCN channels are also gated by cyclic Adenosine Monophosphate (cAMP), which is decreased by Gαi/o-mediated inhibition of adenylate cyclase, so in theory GABA_B receptor activation could shift the threshold for HCN opening toward more hyperpolarized potentials. Another possibility is that GABA_B receptor signaling directly interacts with HCN channels, as with main effectors like GIRK and CaV channels. This distinction is crucial to understanding the functional relevance of the interplay between GABA_B receptors and HCN channels. Regardless of mechanism, this DA-specific effect may also shape the unique GABA_B receptor-GIRK signaling characteristics observed in these cells.

3 GABA_B Receptor Modulation of DA Neuron Activity

3.1 DA Neuron Firing Patterns

DA neurons typically display two firing patterns: tonic firing, a regular pacemaker-like rhythm (1–8 Hz), and burst or phasic firing, characterized by several action potentials at higher frequencies (>15 Hz), also known as phasic firing (Grace and Bunney 1984a, b). Tonic firing relies on the intrinsic expression of a variety of CaV and voltage-dependent Na⁺ channels in DA neurons as well as HCN channels in the SNc (Neuhoff et al. 2002; Khaliq and Bean 2010). By contrast the transition to bursting requires glutamate release from afferent excitatory inputs and N-methyl-d-aspartate (NMDA) receptor activation (Grace et al. 2007; Zweifel et al. 2009). Tonic firing provides background concentrations of extracellular DA, whereas phasic activity drives significantly larger release of DA, necessary to drive motivated behaviors. For example, phasic, but not tonic, DA neurons stimulation is sufficient to drive conditioned place preference and operant self-stimulation (Tsai et al. 2009; Adamantidis et al. 2011). However, DA neuron inhibition below tonic firing frequencies is also necessary to signal reward prediction error and can be aversive (Schultz 1998; Tan et al. 2012). Therefore, the fine-tuning of DA firing modes allows optimal encoding of external stimuli and adequate behavioral adaptation. Here we review the molecular pathways by which GABA_B receptors in DA neurons modulate overall firing rate, occurrence of bursts and DA release. In other words, we define how GABA_B receptors shape the appropriate responsiveness of DA neurons .

3.2 GABA_B Receptor-Mediated Modulation of Firing

3.2.1 GIRK Channels

The main effect of somatodendritic GABA_B receptors results from the opening of GIRK channels. The efflux of K⁺ hyperpolarizes the membrane and decreases input resistance, all of which elevate the threshold for action potential generation. In slice recordings of SNc, bath-application of baclofen readily hyperpolarizes the membrane potential and decreases firing frequency of both DA and GABA neurons (Lacey et al. 1989). In anesthetized rats, systemic injections of baclofen decrease the firing rate and reduce the occurrence of bursts in VTA and SNc DA neurons , in a dose-dependent manner (Olpe et al. 1977). Local microiontophoresis of baclofen in the VTA yields similar results (Engberg et al. 1993; Erhardt et al. 2002). At low concentrations, this effect is probably mediated in part by activation of presynaptic GABA_B receptors at excitatory terminals onto VTA DA neurons , which display higher coupling efficiency than dendritic receptors in DA neurons . However, the effects of higher concentrations of baclofen are likely to reflect a direct hyperpolarization by DA neuron GABA_B receptors. Supporting this hypothesis, baclofen infusion in the VTA and SNc efficiently decrease DA concentrations in the striatum (Westerink et al. 1992). Similarly, baclofen and saclofen (a GABA_B receptor antagonist) decrease and increase, respectively, DA signals in the NAc, recorded in vivo in awake rats (Xi and Stein 1998). It is therefore likely that somatodendritic GABA_B receptors, by decreasing firing at the cell body, indirectly modulate DA release at the terminals.

3.2.2 CaV

Unlike many other neuron types, tonic firing in DA cells heavily relies on external Ca²⁺ and CaV channel opening (Harris et al. 1989). In midbrain DA neurons , GABA_B receptor activation decreases Ca²⁺ currents mediated by high voltage-activated CaV channels such as L-, N-, and P/Q-type (Cardozo and Bean 1995), which were shown to help generate tonic firing (Nedergaard et al. 1993; Puopolo et al. 2007). Therefore it is likely that GABA_B receptor activation also inhibits Ca²⁺ channels, potentially decreasing firing in synergy with GIRK channel opening.

3.2.3 HCN

HCN channels, recently identified as members of the extended GABA_B receptor signaling complex, are opened following GABA_B receptor activation and shorten the GIRK channel-mediated hyperpolarization. This repolarization goes physiologically against the grain of all other GABA_B receptor modulations, but may rather provide a passive feedback signal to GABA_B receptors. Additionally, HCN channels sustain tonic firing in SNc DA neurons , where their activation by GABA_B receptors could paradoxically maintain, rather than inhibit, firing, however this question remains to be investigated (Schwenk et al. 2016).

3.2.4 NMDA Receptors

Finally, GABA_B receptors can decrease Ca²⁺ current through NMDA receptors in cortical neurons, without affecting the amplitude of NMDA or α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor EPSC (Chalifoux and Carter 2010). This effect requires the Gα-mediated inhibition of adenylate cyclase and PKA activity, and is independent of GIRK and CaV channels. The existence of such a mechanism has not yet been investigated in DA neurons , but since bursting requires NMDA receptor activation, this signaling pathway might reduce tonic to phasic firing transitions. Additionally, membrane hyperpolarization through GIRK channels may increase the blockade of NMDAR by Mg²⁺ as suggested in the hippocampus, further decreasing NMDA receptor-mediated transmission (Morrisett et al. 1991; Otmakhova and Lisman 2004). However, it is not clear to which extent Ca²⁺ flowing through NMDA receptors contributes to action potential generation.

3.3 GABA_B Receptor-Mediated Modulation of DA Release

DA neurons release DA not only from axons in target regions, but also in the vicinity of their cell body, through somatodendritic vesicular release (Björklund and Lindvall 1975; Geffen et al. 1976; Jaffe et al. 1998). Numerous studies have investigated the mechanisms for these two types of DA release, mostly with fast-scan cycling voltammetry and amperometry measurements of external DA concentrations in slice preparations. These methods take advantage of the unique oxidation/reduction potential of DA to measure rapid variations of external concentration or exocytosis, respectively, with the use of carbon fiber microelectrodes. Overall, in both cases action potential-triggered Ca²⁺ influx through CaV channels seems necessary, although there may be regional or species differences in the source of Ca²⁺ and the exact channels involved (Rice et al. 1997; Ford et al. 2010; Hoffman and Gerhardt 1999; Chen et al. 2006).

3.3.1 Dendritic Release

Only a few studies have investigated the effect of GABA_B receptor on somatodendritic DA release. As previously mentioned, postsynaptic GABA_B receptors in DA neurons inhibit CaV channels to modulate firing (Cardozo and Bean 1995). Among these, N-, L-, and T-type channels drive dendritic DA release in the SNc (Beckstead et al. 2004; Kim et al. 2008). Thus, although not directly demonstrated, dendritic GABA_B receptors could inhibit CaV channels and decrease vesicular fusion. Indeed, in vivo local infusion of baclofen nearly abolishes external DA concentrations in the SNc and VTA (Westerink et al. 1992; Santiago et al. 1993b; Klitenick et al. 1992). Conversely, inhibiting GABA_B receptors in SNc slices increases voltammetry DA signals (but not in the VTA , Chen and Rice 2002).

3.3.2 Axonal Release

DA axonal release in the dorsal and ventral striatum, emanating from the SNc or VTA DA projections, relies mainly on N- and P/Q-type channels activation, with a potential minor participation of R-type channels (Phillips and Stamford 2000; Chen et al. 2006). All of these channels can be inhibited by GABA_B receptors in different brain regions and cell-types, although this has not been directly demonstrated in DA neurons . Actually, the expression and function of GABA_B receptors in DA neuron terminals remains poorly understood. GABA_B receptor messenger ribonucleic acid (mRNA) autoradiography reveals low expression in the striatum (Kaupmann et al. 1998). Electron microscopy in rats revealed the expression of GABA_B receptors at VGlut2-positive terminals in the striatum, which could belong to either thalamic excitatory neurons or VTA DA neurons (Lacey et al. 2005). In primates, rare presynaptic boutons, displaying the structural features of SNc DA terminals , also express the GABA_B1 subunit (Charara et al. 2000). At the functional level, microdialysis experiments have shown a lack of, or minimal effect of baclofen infusion in the NAc on DA contents (Santiago et al. 1993a; Westerink et al. 1994; Xi et al. 2003), and a modest decrease or increase of DA in the PFC upon GABA_B receptor activation or antagonization, respectively (Santiago et al. 1993a). A couple of more recent studies, recording electrically evoked DA voltammetry signals in the NAc (Pitman et al. 2014) and dorsal striatum (Schmitz et al. 2002) showed that at saturating concentration, baclofen only blocks 20–40 % of the total signal evoked by single pulses, whereas this number further dropped with increased stimulation frequency stimulation. This is highly divergent from the action of presynaptic GABA_B receptors in most other brain regions and cell types, which usually show high coupling efficiency and almost complete blockade of neurotransmitter release (Lüscher et al. 1997; Padgett et al. 2012). Altogether, these results suggest that GABA_B receptors are present in DA neurons terminals, but assume much less of their traditional function as gatekeepers of release.

The assessment of DA axonal GABA_B receptors modulation of neurotransmitter release has been so far difficult because of the absence of fast, DA-mediated postsynaptic current. However, it was recently discovered that DA neuron subpopulations corelease glutamate (Hnasko et al. 2010) or GABA (Tritsch et al. 2012), as measured by postsynaptic excitatory and inhibitory currents in striatal neurons, respectively. Future studies using these ionotropic responses as readouts for presynaptic regulation of release will hopefully yield more specific insight into the function of presynaptic GABA_B receptors in DA terminals.

4 Bidirectional Control of DA Neurons by GABA_B Receptor Agonists

In the VTA , GABA neurons provide a major inhibitory input to DA neurons (Johnson and North 1992a). Inhibition of GABA neurons disinhibits DA neurons , whereas GABA neuron activation shuts DA neurons down, and is sufficient to drive aversive behavioral responses (Tan et al. 2012; van Zessen et al. 2012). Therefore, modulating excitability and release from VTA GABA neurons allows for an indirect, yet powerful handle over DA neuron activity. Here, we review how adequate dosage of GABA_B receptor agonists takes advantage of the VTA microcircuit to bidirectionally modulate DA neuron activity, by selectively inhibiting GABA neurons only or both GABA and DA neurons .

4.1 Tailored Agonist Dosage for DA Neuron Disinhibition

As mentioned earlier, VTA GABA neurons express GABA_B receptors and GIRK channels, although with a much higher coupling efficiency than DA neurons , reflected by the ten-fold lower EC50 value for baclofen-mediated GIRK currents in GABA versus DA neurons (Fig. 8.1). In addition, and contrary to DA neurons , axonal GABA_B receptors in GABA neurons strongly modulate synaptic release in the VTA , with an EC50 value for baclofen similar to dendritic receptors (Cruz et al. 2004; Padgett et al. 2012). In other words, there is a window of minimal baclofen concentration (about 0.1 μM) at which dendritic and axonal GABA_B receptors are activated in GABA neurons, but not in DA neurons (Fig. 8.1). Thus, this minimal baclofen concentration inhibits GABA neurons, removes inhibition from GABA to DA neurons and therefore disinhibits DA neurons in slices. Higher concentrations then activate GABA_B receptors in DA neurons , initially normalizing and eventually decreasing DA firing (Cruz et al. 2004).

Interestingly, similar observations were reported with the club drug γ-hydroxybutirate (GHB), a low-affinity GABA_B receptor agonist . As with baclofen, the EC50 value for GIRK currents is one order of magnitude lower in GABA neurons than in DA neurons . Furthermore, saturating concentrations of GHB and baclofen evoke GIRK currents of comparable amplitude in GABA neurons, whereas GHB yields only 30 % of the maximal baclofen-evoked current in DA neurons . Therefore, a low concentration of GHB, sufficient to activate GABA_B receptors in GABA neurons but without effect in DA neurons , leads to the disinhibition of DA neurons (Labouèbe et al. 2007). For both baclofen and GHB, bidirectional control over DA neuron activity takes advantage of two key features of the VTA : First, the microcircuit formed by GABA neurons’ inhibitory innervation of DA neurons , and second, the cell-type-specific coupling efficiency of GABA_B receptor signaling.

4.2 Physiological Relevance

These experiments provide a cellular mechanism by which increasing concentrations of GABA_B receptor agonists bidirectionally modulate DA neuron activity. More specifically, at low doses, GABA neuron inhibition leads to the disinhibition of DA neurons and increases DA release throughout the brain, which is the hallmark of addictive drugs (Di Chiara and Imperato 1988). Indeed, GHB is self-administered in rodents (Martellotta et al. 1998) and has abuse potential in humans (Nicholson and Balster 2001) at concentrations leading to DA neuron disinhibition in slices. Similarly, human volunteers engaged in a gambling task show increased reinforcement learning with a low dose of baclofen (compared to a larger dose), presumably through the disinhibition of DA neurons (Terrier et al. 2011).

In contrast, higher concentrations of baclofen or GHB decrease DA neuron activity and DA release, by hyperpolarizing the membrane and increasing the threshold for burst firing. Moreover, agonist concentrations sufficient to activate GABA_B receptors in DA neurons are also likely to activate high coupling efficiency presynaptic receptors at excitatory inputs, decreasing glutamate release (Padgett et al. 2012). Accordingly, baclofen infusion in the VTA of rodents reduces self-administration of various addictive drugs (Shoaib et al. 1998; Xi and Stein 2000; Corrigall et al. 2000; see also Chaps. 14 and 15 of this book). Baclofen is also used in humans as an anticraving agent for addictive drugs like cocaine and alcohol (Ling et al. 1998; Ameisen 2005; Addolorato et al. 2009; see also Chaps. 14 and 15 of this book), an effect occasionally observed with GHB itself (Gallimberti et al. 1989).

5 Plasticity of GABA_B Receptor Signaling

While most studies on synaptic plasticity examine excitatory transmission, various forms of plasticity of GABA_A receptor-mediated transmission have been identified (Castillo et al. 2011; Kullmann et al. 2012). Only a handful of studies have characterized activity-dependent and drug-evoked plasticity of GABA_B receptor signaling. Here we review these observations with an emphasis on those occurring in the mesolimbic circuit, and discuss their induction and expression mechanisms as well as their functional implications.

5.1 Activity-Dependent Plasticity

5.1.1 Induction

The induction of all forms of activity-dependent GABA_B receptor signaling plasticity described so far requires NMDA receptor activation. Various induction protocols, like electrical stimulation of glutamate release and postsynaptic membrane depolarization in slices, or bath application of NMDA receptor agonists in cell cultures, lead to potentiation or depression of GABA_B receptor signaling. The direction of the plasticity is then specified by distinct expression mechanisms.

5.1.2 Potentiation

The first electrophysiological study to report activity-dependent synaptic plasticity of GABA_B receptors in cell cultures and acute hippocampal slices showed that a classic AMPA receptor long-term potentiation protocol potentiates GABA_B receptor-mediated IPSCs. The induction requires electrical stimulation of synaptic glutamate release paired with postsynaptic membrane depolarization, which activates NMDA receptors, increases intracellular Ca²⁺, and recruits Ca²⁺/calmodulin-dependent protein kinase II (CaMKII; Huang et al. 2005). Similar induction mechanism was recently observed in VTA DA neurons , in which sustained depolarization or burst firing leads to the potentiation of GABA_B receptor-GIRK currents (Lalive et al. 2014). Blocking the trafficking of GIRK channels by interfering specifically with PDZ-binding domains on GIRK2c or GIRK3 prevents the potentiation of GABA_B receptor-GIRK currents in DA neurons , suggesting that GABA_B-GIRK plasticity is expressed through surface delivery of heteromeric GIRK2c/3 channels in DA neurons (Fig. 8.2). Expression mechanisms in the hippocampus could additionally involve protein phosphatase (PP) 1-dependent phosphorylation of Serine (S) 9 residue on GIRK2 to modulate channel trafficking (Chung et al. 2009). Lastly, NMDA application transiently increases GABA_B receptor surface expression by recruiting 5′-adenosine monophosphate-activated kinase (AMPK)-mediated phosphorylation at S783 on the GABA_B2 subunit in cortical neuron cultures (Terunuma et al. 2010). A similar increase was also reported following glycine-mediated synaptic NMDA receptor activation, which was prevented by inhibiting protein recycling (Kantamneni et al. 2014).

Fig. 8.2

Dopamine neuron firing drives bidirectional GIRK channel plasticity. (Left) NMDA receptor and phasic activity drive Ca²⁺ entry and CaMKII activation, engaging a molecular pathway interacting with GIRK2c/3 PDZ domain and trafficking channels at the membrane. This results in the potentiation of GABA_B and D2 receptor-GIRK currents, illustrated by the example recording of a large GABA_B receptor-GIRK IPSC, evoked by electrical stimulation at −60 mV (top inset). (Right) Tonic activity or tonic NMDA receptor activation drives Ca²⁺ entry, triggering an undetermined molecular cascade eventually driving the internalization of GIRK2c/3 channels through a PDZ domain interaction, resulting in the depression of GABA_B and potentially D2 receptor-GIRK currents, as shown by the reduced amplitude of electrically evoked GABA_B receptor-GIRK IPSC (top inset). Scale: 20pA, 200 ms [Inset traces reproduced from Lalive et al. 2014]

5.1.3 Depression

In cultured hippocampal neurons, prolonged (30 min) NMDA receptor activation induces a Ca²⁺- and CaMKII-dependent depression of GABA_B receptor signaling. CaMKII specifically phosphorylates S892 on GABA_B1b, which drives the internalization of GABA_B receptors (Guetg et al. 2010). In cultured cortical neurons, a similar induction protocol depresses baclofen-evoked currents. The exact induction mechanism is not identified, however expression requires increased PP2 activity and dephosphorylation of GABA_B2 S783, leading to decreased surface expression of GABA_B receptors (Terunuma et al. 2010). It appears that dephosphorylation of GABA_B2 does not directly trigger the internalization of the receptor, but rather modifies basal recycling at the level of postendocytic sorting. This mechanism has since been implicated in GABA_B receptor plasticity in VTA GABA neurons and PFC (Padgett et al. 2012; Hearing et al. 2013) (further described in the drug-evoked plasticity section below), and in the lateral habenula following exposure to aversive stimuli (Lecca et al. 2016), suggesting a crucial and widely distributed plasticity mechanism. In VTA DA neurons , low frequency tonic firing or synaptic activation of NMDA receptors for 5 min decreases the amplitude of GABA_B receptor-GIRK IPSC (Fig. 8.2). Similarly to the burst-induced potentiation, this plasticity is expressed by trafficking of GIRK3-containing channels (Lalive et al. 2014).

5.1.4 Functional Significance

Overall, GABA_B receptor plasticity seems tightly tuned to neuronal excitation. It is striking that all of the above-described mechanisms rely on excitatory input stimulation for their induction. This may then lead to NMDA receptor activation or action potential firing, suggesting that plasticity of GABA_B receptor signaling is an adaptive response to changes in neuronal activity (Fig. 8.2). A unifying interpretation would suggest that this inhibitory plasticity is called in as a compensatory, homeostatic-like adaption, where GABA_B receptor function is potentiated following sustained activity, but reduced after periods of little activity. This hypothesis is especially supported in VTA DA neurons , where GIRK channel upregulation is functionally not restricted to GABA_B receptors, but also potentiates Dopamine 2 receptor-GIRK currents, thereby strengthening another inhibitory pathway (Lalive et al. 2014). Overall, GABA_B receptor plasticity further emphasizes the second-order nature of GABA_B receptor function in tempering neuronal excitability and responsiveness to relevant synaptic signals.

5.2 Drug-Evoked Plasticity of GABA_B Receptor Signaling

A hallmark of addictive drugs is the induction of excitatory synaptic plasticity in VTA DA neurons after a single exposure, which eventually spreads to other nuclei following repeated intake or withdrawal periods and underlies the development of addictive behaviors (Lüscher and Malenka 2011). Similarly, drugs of abuse evoke various forms of GABA_B receptor signaling plasticity (Table 8.2), however their functional significance is debated.

Table 8.2 Summary of GABA_B receptor signaling plasticity in the VTA

5.2.1 Early Observations

Early work revealed a decrease in Gα_i/o protein in the VTA and NAc following chronic but not acute cocaine treatment in rats, hinting at metabotropic signaling alterations (Nestler et al. 1990). Similar results were later obtained after withdrawal from chronic psychostimulants (Zhang et al. 2000; Xi et al. 2003; Giorgetti et al. 2002). Additionally, most GIRK knockout mice show altered drug-related behavior (Luján et al. 2014). Altogether these results pointed early on toward a potential modulation of GABA_B receptor signaling following exposure to drugs of abuse. Unfortunately the techniques employed in these studies do not allow the distinction between neuron types and axonal from dendritic GABA_B receptors, blurring functional interpretation. Patch-clamp studies in identified cell types have then helped overcome these confounds and shed light on the molecular mechanisms underlying drug-evoked plasticity of GABA_B receptor signaling.

5.2.2 Plasticity in VTA Neurons

A single exposure to psychostimulants like cocaine and methamphetamine is sufficient to evoke a decrease in GABA_B receptor-GIRK currents in VTA DA and GABA neurons (Padgett et al. 2012; Arora et al. 2011). This adaptation is also observed in DA neurons after repeated passive injections (Munoz et al. 2016) or active self-administration of psychostimulants (Sharpe et al. 2015). In both cell types, plasticity induction requires DA type 1 (D1) and DA type 2 (D2) receptor activation, however the molecular pathways underlying plasticity expression are highly different.

In DA neurons , the reduction in GABA_B receptor signaling is mediated by a decrease in GIRK channel density (but not GABA_B receptors) at the membrane (Arora et al. 2011). Interestingly, in DA neurons lacking the GIRK3 subunit, psychostimulants fail to depress baclofen-evoked currents, suggesting that trafficking of GIRK2c/3 heteromers is required (Munoz et al. 2016), similar to activity-dependent GIRK plasticity described in the same cells (Lalive et al. 2014). Additionally, DA neurons lacking SNX27 display blunted baclofen-evoked currents (Munoz and Slesinger 2014). SNX27 is involved in trafficking of GIRK2c/3 channels and is sensitive to drugs of abuse (Kajii et al. 2003), and may therefore participate to GABA_B receptor signaling plasticity in DA neurons .

In GABA neurons , the decrease in GABA_B receptor signaling is also observed in axon terminals, where the coupling efficiency is reduced (Padgett et al. 2012). The reduction in GABA_B receptor-GIRK currents is due to enhanced PP2A-mediated dephosphorylation of GABA_B2 S783, a pathway previously identified for activity-dependent depression of GABA_B receptor signaling (Terunuma et al. 2010). As a result, GABA_B receptors and GIRK channels are internalized but not degraded, and acute PP2A inhibition rescues the full amplitude of baclofen-evoked currents. Furthermore, this plasticity is blocked in a mouse expressing a GABA_B2 subunit insensitive to PP2A dephosphorylation (Munoz et al. 2016). Interestingly, GIRK channels do not interact with PP2A but are also removed from the membrane, possibly because of their proximity to GABA_B receptors (Padgett et al. 2012).

Drugs other than psychostimulants also evoke GABA_B receptor signaling plasticity. Chronic GHB or morphine treatment increases the coupling efficiency of GABA_B receptors to GIRK channels in VTA DA neurons . This specific effect is mediated by the downregulation of RGS2, a protein acting as a brake on G protein signaling. Consistent with this, the coupling efficiency is increased in RGS2 knockout mice, and drug treatment has no further effect (Labouèbe et al. 2007).

5.2.3 Plasticity of VTA Afferents

In VTA -projecting PFC neurons, repeated cocaine injection decreases baclofen-evoked currents in a D1 receptor-dependent fashion (Hearing et al. 2013). GABA_B receptors are internalized following PP2A-dependent S783 GABAB2 dephosphorylation, similar to what was described in VTA GABA neurons (Padgett et al. 2012). It is not known whether GABA_B receptor signaling is also altered in axons projecting to the VTA . Interestingly, chronic morphine increases the coupling efficiency of presynaptic GABA_B receptors at excitatory terminals onto DA neurons (Manzoni and Williams 1999). However, this was observed with nonspecific electrical stimulation and may not reflect GABA_B receptor function at all axon terminals.

5.2.4 Functional Significance

In DA neurons , the reduction in maximal baclofen-evoked current after psychostimulant exposure is paralleled with a mild decrease in the ability of GABA_B receptors to block neuronal activity (Munoz et al. 2016). In other words, DA neurons are partially relieved from the rule of GABA_B receptors. However, in the same conditions synaptically evoked GABA_B receptor GIRK IPSCs, which reflect a more physiological mode of GABA_B receptor activation, are unaffected (Padgett et al. 2012). This suggests that despite GIRK channel internalization, GABA_B receptor signaling is still fully functional under physiological levels of activation. In stark contrast, synaptically evoked GABA_B receptor-GIRK IPSC in VTA GABA neurons are absent following drug exposure, arguing for a physiologically relevant functional deficiency. Accordingly, GABA_B receptor activation, even with saturating concentrations of baclofen, is unable to decrease firing (Padgett et al. 2012). Therefore, the VTA microcircuit is altered: GABA neuron activity is less likely to be inhibited, and its inhibitory control over DA neurons may be strengthened. Similarly, the increase in coupling efficiency in DA neurons after morphine treatment empowers GABA_B receptors to modulate DA neuron excitability. Indeed, DA neurons are now inhibited by a minimal concentration of baclofen (0.1 μM) that would only affect GABA neurons in control conditions (Labouèbe et al. 2007). To fully predict the net functional effect of these forms of plasticity, two key questions need to be answered. First of all, do these plasticities represent a uniform and simultaneous adaptation following exposure to any drug of abuse, like it has been suggested for excitatory plasticity (Lüscher and Ungless 2006), or are they tailor-cut to specific drugs? For example, a decrease in GABA_B receptor GIRK current amplitude in DA neurons , as seen after psychostimulant exposure (Arora et al. 2011), might be compensated for by an increase in coupling efficiency, as observed after morphine treatment (Labouèbe et al. 2007), resulting in no functional difference. Second, what is the actual concentration of endogenous agonist to which GABA_B receptors are exposed in vivo? In other words, thus far it has been difficult to infer the physiological function of GABA_B receptors from maximal baclofen-evoked current amplitude.

5.2.5 Drug-Evoked Plasticity: Compensatory or Contributory?

Drug-evoked excitatory synaptic plasticity is believed to progressively modify circuit function and eventually lead to the development of addictive behaviors (Lüscher and Malenka 2011). Repeated drug exposure increases DA neuron activity and excitability (White 1996; Henry et al. 1989). However, it is not clear whether GABA_B receptor signaling plasticity subserves or counteracts these circuit changes. The loss of GABA_B receptor function in VTA GABA neurons argues for a compensatory mechanism, promoting GABA release onto DA neurons to limit their activity (Padgett et al. 2012). However, mimicking the drug-evoked plasticity by downregulating GIRK channels in VTA DA neurons or PFC increases the acute locomotor response to psychostimulant, suggesting an active participation to the behavioral effects of addictive drugs (Munoz and Slesinger 2014; Hearing et al. 2013).

Over recent years GABA_B receptor plasticity has emerged as a cellular mechanism relevant to neuronal activity and circuit function. Activity-dependent and drug-evoked adaptations occur in different cell types and regions, concurrently with excitatory plasticity. Whereas excitatory synaptic plasticity is usually perceived as the primary neural correlate of experience, GABA_B receptor plasticity rather appears as a secondary adaptation that may either stabilize or derail neuronal excitability, especially in DA neurons . Future studies will have to causally assess the contribution of GABA_B receptor signaling plasticity to neuronal activity and behavioral alterations in the context of addiction disease.

6 Conclusions

GABA_B receptors engage diverse signaling pathways to hyperpolarize somatodendritic compartments and reduce probability of neurotransmitter release, thus exerting a pre- and postsynaptic modulation of neural activity. In the VTA , the cell-type-specific composition of GABA_B receptor complexes determines the signaling efficacy and allows for bidirectional control of DA output through inhibition of local interneurons. GABA_B receptor agonists may thus lead to disinihibition at low concentration, while inhibiting DA neurons at higher doses. This may explain the reinforcing effects of baclofen and GHB and associated risk for addiction. This model also accounts for the anti-craving effect of high doses of baclofen.

In DA neurons , where dynamic switching between pauses, tonic, and phasic firing occurs with experience, GABA_B receptors control the excitability and adjust responsiveness to relevant synaptic inputs. Furthermore, GABA_B receptor signaling can undergo plasticity and adapts to changes in neuronal activity, which may constitute a compensatory mechanism to maintain physiological neuronal activity. Addictive drugs also alter GABA_B receptor signaling throughout the mesolimbic system , however the relevance of this plasticity for the development of addictive behavior remains elusive.

Altogether, GABA_B receptors are effective and dynamic modulators of the DA system, and form a target with potential for pharmacological interventions in humans.