Smoking is the most important preventable cause of mortality and morbidity worldwide. This nicotine addiction is mediated through the nicotinic acetylcholine receptor (nAChR), expressed on most neurons, and also many other organs in the body. Even within the ventral tegmental area (VTA), the key brain area responsible for the reinforcing properties of all drugs of abuse, nicotine acts on several different cell types and afferents. Identifying the precise action of nicotine on this microcircuit, in vivo, is important to understand reinforcement, and finally to develop efficient smoking cessation treatments. We used a novel lentiviral system to re-express exclusively high-affinity nAChRs on either dopaminergic (DAergic) or γ-aminobutyric acid-releasing (GABAergic) neurons, or both, in the VTA. Using in vivo electrophysiology, we show that, contrary to widely accepted models, the activation of GABA neurons in the VTA plays a crucial role in the control of nicotine-elicited DAergic activity. Our results demonstrate that both positive and negative motivational values are transmitted through the dopamine (DA) neuron, but that the concerted activity of DA and GABA systems is necessary for the reinforcing actions of nicotine through burst firing of DA neurons. This work identifies the GABAergic interneuron as a potential target for smoking cessation drug development.
Tobacco abuse is the worldwide leading cause of preventable morbidity and mortality. According to the World Health Organization, more than five million smokers die every year from the consequences.1 Dissecting the underlying neurobiological mechanisms of this disorder has recently made important progress with the use of genetically modified mice,2 but they remain far from being understood in detail. The identification of novel drug targets for the design of smoking cessation medication is becoming a priority in pharmacological research.3 Nicotine, the principal, if not sole, addictive component of tobacco smoke,4 exerts its reinforcing effects through its action on nicotinic acetylcholine receptors (nAChRs), a heterogeneous family of pentameric, ligand-gated ion channels.5 nAChRs situated in the mesolimbic reward system mediate nicotine-induced dopamine (DA) release in the nucleus accumbens (NAc) from midbrain dopaminergic (DAergic) neurons located in the ventral tegmental area (VTA).6 Among the different nAChRs expressed in this region, β2-containing nAChRs (β2*-nAChRs) have been shown to play a crucial role in the positive rewarding properties of nicotine, as they are the essential partner to form high-affinity receptors.7 β2*-nAChRs are expressed on both DAergic and γ-aminobutyric acid-releasing (GABAergic) interneurons present in the VTA,8 and nicotine modulates the activity of both neuronal populations.9 Pharmacological studies have addressed the complex and often opposing influences of the DA- and GABA-dependent neural systems,10 but none of them has yet elucidated the nature of their interaction, that is, whether they cooperate to produce the rewarding effect,9, 11, 12 or mediate, respectively, the acute rewarding and aversive psychological effects of nicotine.10
The complexity of nicotine's action to result in enhanced DA release has been reviewed recently.13, 14 Several independent mechanisms have been identified, or postulated. These are enhanced excitation of the DA cells,15 or their disinhibition, or modification of presynaptic terminals onto the soma of DA neurons. Previous in vitro studies proposed that the desensitization of β2*-nAChRs on GABA interneurons following nicotine exposure, resulting in decreased GABAergic neuron activity, produces a disinhibition of DA neurons, and hence increased presynaptic excitation.9, 11, 16 We have here comprehensively addressed the respective role of β2*-nAChRs expressed on DA or GABAergic VTA neurons in their response to nicotine, and the behavioral consequences. The work presented is arranged according to the following outline: using in vivo electrophysiology of single neurons in the VTA, we found surprisingly that putative GABAergic neurons do not desensitize following repeated nicotine injections as postulated by current models. We then set out to understand the physiological basis of their role in DA neuron activity. This was achieved by the selective expression of high-affinity nicotinic receptors on either DA or GABAergic interneurons, or both, and the correlation of DA firing patterns with intra-VTA self-administration behavior. We conclude that GABA-dependent burst firing of DA neurons is a crucial element in the reinforcing effect of nicotine.
Materials and methods
Viral production and analysis of transduced VTA cells
Viral particles were generated by co-transfection of HEK-293T cells by the vector plasmid, a packaging plasmid and an envelope plasmid using Lipofectamine Plus (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. At 2 days after transfection, viral particles were harvested in the supernatant, treated with DNaseI and MgCl2, filtered through 0.45 μm pores, concentrated by ultracentrifugation and resuspended in a small volume of phosphate-buffered saline (PBS). Viral stocks were stored in small aliquots at −80 °C before use. Viral titers were estimated by quantification of the p24 capsid protein using HIV-1 p24 antigen immunoassay (ZeptoMetrix Corporation, Buffalo, NY, USA) according to the manufacturer's instructions. Lentivectors were diluted in PBS before stereotaxic injection to achieve a dose corresponding to 230 ng of p24 protein in 2 μl.
Mice aged 6–12 weeks were anesthetized using ketamine/xylazine in PBS. A mouse was introduced into a stereotaxic frame adapted for use with mice. Lentivectors (enhanced green fluorescent protein (eGFP)-expressing vector: 2 μl at 50 ng p24 protein per μl; β2-expressing vector: 2 μl at 115 ng p24 protein per μl) were injected bilaterally at: antero-posterior −3.4 mm, lateral ±0.5 mm from Bregma and −4.4 mm from the surface. All procedures were carried out in accordance with European Commission directives 219/1990 and 220/1990, and approved by Animalerie centrale and Médecine du travail, Institut Pasteur. The authors carrying out surgery hold an Animal Surgery Authorization from the French Ministry of Agriculture.
Analysis of transduced VTA cells
To analyze eGFP and mCherry transduction on DAergic and GABAergic neurons, sections from vectorized DAT-Cre and GAD67-Cre mice were stained for tyrosine hydroxylase (TH) antibody. Fluorescence immunohistochemistry was performed as follows: free-floating VTA brain sections were incubated 1 h at 4 °C in a fixative solution of PBS containing 10% normal goat serum (Sigma, Lyon, France) and 0.2% Triton X-100, and then overnight at 4 °C in PBS containing a mouse anti-TH (Sigma) at 1:200 dilution, 2% normal goat serum and 0.2% Triton X-100. The next day, sections were rinsed with PBS and then incubated 3 h at room temperature with secondary antibody (Cy5-coniugated anti-mouse, Jackson Immunoresearch, London, UK) at 1:200 dilution in a solution of 2% normal goat serum in PBS. After three rinses in PBS, slices were wet-mounted using Mowiol 4-88 (Calbiochem Corporation, La Jolla, CA, USA).
Cell counting by image analysis
Anti-TH-immunostained sections from DAT-Cre mice were imaged at × 25, × 40 and × 63 with a Zeiss LSM510 confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany). A quantitative analysis of transduced neurons was performed automatically with the software identifying spot positions using a statistical analysis of the multiscale decomposition of the image. The program (QUIA; http://www.bioimageanalysis.org) detects nuclei (Cre-positive cells) by first filtering the image with an undecimated wavelet at different scales and then thresholding it with a statistical procedure,17, 18 which takes into account the noise level. To perform colocalization, we computed for each pixel of each detection the number of corresponding pixels in the green channel (eGFP-positive cells), considering a fixed threshold. The detection is labeled as colocalized if the number of green pixels reaches 50% of the area of the nuclei. For phosphorylated-extracellular signal-regulated kinase (P-ERK) immunostaining, quantifications were performed using the image analyzer software (Image-Pro Plus; Media Cybernetics, Silver Spring, MD, USA), taking into account the cells with nuclear immunofluorescence above the background.
Determination of the efficiency of DA cell transduction
The percentage of transduced TH-positive neurons in vectorized DAT-Cre mice was quantified on one of every three sections for each mouse. This results in 73±12% of DA neurons being transduced.
Epibatidine binding studies
The β−/−, β2−/− × DAT-Cre and β2−/− × GAD-Cre mice aged 12–13 weeks were injected bilaterally in the VTA. After 4 weeks of viral expression, brains were dissected, frozen in dry ice and stored at −80 °C until use. Coronal sections, 20-μm thick, were cut at −20 °C and thaw mounted on Menzel Glasser SuperFrost Plus microscope slides. Slides were incubated at room temperature with 220 pM 125I-epibatidine (NEN Perkin-Elmer, Boston, MA, USA; specific activity 2200 Ci/mmol) in 50 mM Tris (pH 7.4) for 1 h. After incubation, sections were rinsed twice 5 min in the same buffer and briefly in distilled water. Nonspecific binding was not distinguishable from background. Sections were exposed for 36 h to Kodak Biomax films. As reported previously,19 high-affinity 125I-epibatidine binding sites were absent from the brain of β2 knockout (KO) mice, with the exception of structures of the habenulo-peduncular system. The quantification of lentiviral restoration of functional β2*-nAChRs in the VTA was carried out by ImageJ (National Institutes of Health, Bethesda, MD, USA) on eight coronal sections for each brain. To obtain the mean value of the optical density, the same ROI was reported on every section of each brain.
In vivo electrophysiological recording of VTA DAergic and GABAergic neurons
Single unit extracellular recordings were performed in wild-type (WT), KO and VEC (vectorized) mice as detailed in the Supplementary Information. Animals were anesthetized with chloral hydrate (400 mg/kg, intraperitoneally). Spontaneously active DA and GABA neurons were identified on the basis of previously established electrophysiological criteria (see Supplementary Material and Mameli-Engvall et al.15). Intravenous injection of 0.5 mM nicotine tartrate into the saphenous vein (30 μg/kg free base in a final volume of 10 μl) was performed as described. Firing frequency was quantified over 30 s periods, expressed as a percentage of average basal firing, and means were calculated within each group.
In vitro electrophysiological recording of VTA DAergic and GABAergic neurons
Slice recordings were performed in WT, KO and VEC mice as detailed in the Supplementary Information.
In vivo microdialysis
In vivo microdialysis was carried out as described in the Supplementary Information, with a probe placed unilaterally in the shell of the NAc, coordinates from Bregma (in mm): anterior 1.34, lateral 0.7, ventral −5.4. Probe placement was verified histologically.
Intracranial self-administration procedure
Intracranial self-administration (ICSA) behavior was investigated using a procedure previously described for morphine20, 21 and cocaine,22, 23 and then adapted to nicotine.19, 24 This model is based on choice behavior in a Y-maze discrimination task between a nicotine-reinforced arm and a neutral arm. Discrimination tasks relying on choice behavior provide a more accurate assessment of reward than rate-dependent paradigms.25 This task is particularly well adapted to detect reward-related deficits in mutant mice.24, 26, 27 ICSA allows excellent temporal and spatial, region-specific delivery of nicotine for targeting specifically brain structures such as the VTA.24, 28 The present experiments were conducted in a gray Plexiglas Y-maze, see Figure 4a. Its arms were separated by an angle of 90°. The stem and the arms were 31 cm long and 12 cm high. The starting box (14 × 8 cm2) is separated from the stem by a sliding door. Each arm is composed of a sliding door at its entrance and a photoelectric cell 6 cm from its end. Mice are implanted unilaterally into the posterior VTA, using the coordinates of lentiviral injections. Before each experimental session, a stainless-steel injection cannula (outer diameter 0.229 mm, inner diameter 0.127 mm; Figure 4a) is inserted into the VTA and held in a fixed position by means of a small connector. The tip of the injection cannula projects beyond the guide cannula by 1.5 mm (Figure 4a). It is connected by flexible polyethylene tubing to the microinjection system, which houses a 5-μl Hamilton syringe containing a solution of nicotine in the artificial cerebrospinal fluid (Microperfusion Fluid; Phymep, Paris, France), the pH of which being adjusted to neutrality with NaOH 0.1 N. Mice were tested for nine consecutive days, with each daily session composed of 10 successive trials. The first session consisted in a 10-min free exploration of the Y-maze. During a second session, mice were habituated to be connected to the self-injection system, but no injection was available. In subsequent sessions, interruption of photocell beams by entering the reinforced arm of the Y-maze triggered an intra-VTA injection of nicotine that lasted 4 s (100 ng nicotine tartrate in 50 nl of artificial cerebrospinal fluid, resulting in a nicotine concentration of 4 mM as referred to the base). As photobeams are located only 6 cm from the end of each arm, mice had to enter completely into an arm to end a trial. Therefore, partial entrance into the arm did not count as a choice. Entering completely the non-reinforced arm produces no injection, and entering either arm terminated the trial. The animal's movements are detected using an optical system and transmitted to a computer, which rotated in turn the injector in the same direction, thus avoiding twisting. The number of times mice trigger a self-injection within the nicotine-reinforced arm (the number of self-administrations) and the time between the opening of the starting box door and triggering of self-injections (self-injection latency) were recorded by a computer running a program specifically developed for ICSA (YAA; Imetronic, Pessac, France). The mean of one daily session provided one graph plot. Both WT and mutant mice exhibited a similar number of intra-VTA injections of vehicle (artificial cerebrospinal fluid, data not shown) that was not different from chance level (50%). All groups were constituted of at least n=7 and up to n=10 subjects per group. Following thionin-stained histological control of stereotaxic implantation, only mice with a self-injection cannula located within the VTA were included in the statistical analysis. For the ICSA, all data were analyzed by repeated measures analysis of variance using StatView 5.0 (Abacus Concepts, Piscataway, NJ, USA), followed by post-hoc Fisher's test for which the level of statistical significance was set at P<0.05.
In vivo electrophysiology of the nicotine response in the VTA
We wanted to obtain a comprehensive understanding of nicotine responses in the microcircuits of the VTA. As most previous studies were carried out in slice preparations, we decided to address the nicotinic-cholinergic regulation of the VTA using in vivo electrophysiology.15 There, all afferents to the VTA are preserved, and in particular the cholinergic input from the tegmentum, a key player in VTA physiology29, 30 (see Figure 1). Intravenous administration of nicotine (30 μg/kg) in naive animals resulted in a rapid and pronounced increase in the firing rate of DA cells (Figure 2a), and also of putative GABAergic interneurons (Figure 2b). This latter result is incompatible with a rapid, desensitization-dependent decrease in GABA activity postulated by the disinhibition hypothesis. We then performed paired injections of nicotine at intervals of 5 or 15 min. If disinhibition were the main mechanism underlying the DA cell response, subsequent nicotine injections would further enhance receptor desensitization in GABA neurons, resulting in an increase in the consecutive peak response of DA cells to nicotine. At a 15-min interval, when no accumulation of nicotine in blood and brain was possible due to nicotine pharmacokinetics,31 we observed no change in the amplitude of the DA response to nicotine (Figure 2c). However, at the interval of 5 min, allowing accumulation of nicotine upon injections, DA neurons exhibited a decrease in the amplitude of nicotine response (Figure 2d). Therefore, in vivo, the immediate firing rate increase of DAergic cells in response to an acute injection of nicotine entails DA excitation, and a competition with inhibition, rather than desensitization of nAChRs on GABAergic interneurons.
Modeling the in vivo response to nicotine
To corroborate the proposed direct excitation mechanism of nicotine action, we used a minimal model of the VTA microcircuit to account quantitatively for the recorded responses of DA and GABA neurons, a model modified from Graupner and Gutkin.32, 33 The model reflects the cholinergic (ACh) afferents to the DA and GABA cells in the VTA, local inhibition of DA cells by GABA neurons as well as the pharmacology of the nAChRs (Figure 1). In this minimal model, we chose to leave aside a potential contribution from local glutamatergic neurons as described very recently, as no data exist as to their expression of nicotinic receptors.34 Nicotine-evoked changes are thus exclusively mediated by β2*-nAChRs expressed somatically on both DA and GABA neurons. This is in line with experiments and modeling studies suggesting that the ß2-containing subtype plays the dominant role in nicotinic modulation of DA signaling.7, 15, 32, 35
The model, see Supplementary Information, fitted to the recordings of DA and GABA firing rate responses (green and blue traces in Figure 2), respectively, shows that nicotine drives both β2*-nAChR activation and desensitization. The early response to nicotine in DA and GABA neurons, for example, to a single nicotine injection, is dominated by direct receptor activation that boosts the firing rates in both neuron types (Figures 2a and b). Subsequent nicotine injections further recruit receptor desensitization giving rise to an accommodation in the peak response of DA cells to nicotine (Figures 2c and d). This accommodation is stronger for shorter inter-injection intervals (compare Figure 2c for 15-min intervals and Figure 2d for 5 min). In addition, in our model nicotine-dependent receptor desensitization in GABA neurons decreases their firing rate below baseline after nicotine clearance, thereby producing a prolonged elevated DA response in the model (see Figures 2c and d).
Experimental and model data suggest that receptor desensitization does not govern the immediate DA response and confirms the predominant role of early excitation due to acute nicotine. The model results furthermore suggest that long-lasting disinhibition might give rise to a secondary delayed and prolonged elevated DA activity after nicotine clearance (Figures 2c and d). Finally, these results also suggest that both DA and GABA pathways are activated by nicotine. Dissociating the consequences of the activation of the two pathways by acute nicotine on firing of identified DA cells would confirm the respective role of β2*-nAChRs on DA and GABA neurons. A simulation of two variants of the model, that is, where we set β2*-nAChRs to be expressed only on DA neurons, or only on the GABA neurons, illustrates the potential benefit of this experiment. The first model gives rise to a rapid DA activity boost through direct nicotine-evoked excitation of DA neurons alone (Figure 2e). The second yields only a nicotine-evoked immediate inhibition of the DA firing rate (Figure 2f). This predicts that a differential expression of β2*-nAChRs on the DA or on the GABA neuron would favor excitation versus inhibition of the DA firing rates.
A novel system for the cell-type-specific expression of nicotinic receptors
To test these respective in vivo roles of the DAergic and GABAergic neurons in the microcircuits involved in nicotine reinforcement, we used a Cre recombinase-activated lentiviral expression vector18 (Figure 3a) to drive specific β2*-nAChR re-expression in DA or GABAergic neurons of the VTA. We created two new mouse lines expressing Cre recombinase only in DAergic neurons (β2−/− × DAT-Cre mice),36 or only in GABAergic neurons (β2−/− × GAD67-Cre mice).18, 37 [125I]Epibatidine autoradiography on coronal brain slices demonstrated the recovery of high-affinity receptors at the injection site in the VTA and in axon terminals in the NAc (Figure 3b) for vectorized β2−/− × DAT-Cre mice (referred to as β2DA−VEC), but only in the VTA for β2−/− × GAD67-Cre mice (β2GABA−VEC) injected with the floxed lentivirus. Sections of these mice labeled with epibatidine were quantified, and as shown in Supplementary Figure S2, restoration of between 70 and 85% of high-affinity binding sites compared to WT mice were obtained. We injected mice with a GFP-expressing lentivector (Figure 3a)18 to confirm that transgene expression was completely restricted only to DAergic cells of DAT-Cre mice, and to GABAergic cells of GAD67-Cre mice (Figure 3c), and to quantify the efficiency of the re-expression. We found that 75% of DA neurons re-express the gene with this approach, see Materials and methods. This correlates well with the restoration observed in the binding studies. We also established slice electrophysiology of vectorized mice, see Supplementary Figure S3. We were thus able to demonstrate in an independent set of experiments that vectorization of either the DA or GABA neurons fully restores responsiveness of high-affinity nAChRs.
We then compared nicotine-evoked modification of firing pattern in DAergic neurons of WT mice with those of ß2−/− (see Supplementary Material for ß2−/− group definition), β2DA−VEC and β2GABA−VEC mice (Figures 3d and e). Selective re-expression of the β2 subunit in DAergic neurons only resulted in an increased firing frequency after systemic nicotine injection, whereas β2GABA−VEC mice exhibited a decrease (Figure 3e). The duration of the DA cell inhibition in β2GABA−VEC tallies well with the duration of GABAergic excitation observed in WT animals (Figure 2b). As with the WT mice (Figures 2c and d), repeated nicotine injections were carried out as shown in Supplementary Figure S4. Similar to WT mice, β2DA–VEC exhibited a decrease in the peak response of DA neurons to consecutive injections of nicotine (Supplementary Figure S4b). This is in accordance with model data suggesting that successive nicotine injections further recruit receptor desensitization and give rise to an accommodation in the peak response of DA cells to nicotine. Results suggesting a similar phenomenon were obtained in β2GABA−VEC mice (Supplementary Figure S4c). Taken together, these data indicate that ß2*-nAChRs are desensitized by nicotine at the level of both the DA and GABA neurons. However, in WT mice, the net effect of this desensitization, at a timescale on the order of minutes, results in a reduction of the peak response of DA cells, and not in a strong disinhibition.
These experiments dissociate the competitive effect of direct nicotine excitation revealed in β2DA−VEC and the indirect inhibition through GABAergic nAChR excitation. We had previously shown that the nicotine-elicited increase in the tonic activity of DA cells is abolished in β2−/− mice, and is restored when the β2 subunit is re-expressed in all neurons of the VTA.15, 27 To exclude any possible contribution from other cell types in the VTA, for example, recently described glutamatergic interneurons (see Figure 1), we created a third mouse line crossing the DAT-Cre and GAD67-Cre lines together onto a β2−/− background. Injecting theses double transgenic lines with the floxed β2-expressing lentivirus, referred to as β2DA+GABA−VEC mice, leads to the targeted expression of the β2 subunit on both DA and GABA neurons, but not any other cell type. Injecting nicotine intravenously into these mice and recording from the DA neurons resulted in a WT pattern of responding (Figure 3e, bottom).
DA neuron transmits positive and negative motivational value
Midbrain DAergic neuron activity is associated with reward-related behaviors and positive motivation.38 Recent studies suggest that they also transmit signals related to aversive and alerting events.39 We investigated the behavioral effect of selective DA cell excitation or inhibition by nicotine using an ICSA model based on choice behavior in a Y-maze20, 21, 22, 23 (Figure 4a). In this paradigm, WT mice actively self-administer nicotine at the dose of 100 ng per injection (Figure 4b), whereas β2−/− mice do not acquire this behavior.24, 27
Here, we report that β2DA−VEC mice exhibited no acquisition of nicotine ICSA over the course of daily sessions, as evidenced by both the number of injections and latency (Figures 4c and d). However, β2DA−VEC mice behaved differently from GFP-vectorized subjects: they exhibited initially a transient approach behavior toward the nicotine-reinforced arm, but did not develop a stable, chronic nicotine self-administration over sessions (Supplementary Figures S5a and b). In contrast, β2GABA−VEC exhibited a decreased choice of the reinforced arm, which remained below chance level (Figure 4c). Preferential choice of the non-reinforced arm is typically associated with the aversive effects of drugs of abuse, as previously demonstrated with cocaine.23 To determine whether decreasing the nicotine dose would extend ICSA responding, the β2DA−VEC mice were tested further with lower concentrations of nicotine (see Supplementary Figures S5c and d). The use of a 50 ng dose led to an initial increase quickly followed by chance level responding, whereas no response was observed with the 10 ng dose. These results demonstrate that the β2 subunit expressed selectively on DA or GABA neurons of the VTA encodes, respectively, for positive or negative motivational value: direct excitation of DA neurons enhances the incentive value of nicotine, while inhibition of DA neurons through GABAergic excitation and/or direct inhibition of GABAergic efferent decreases it. However, in both cases nicotine effects were transient, as neither the positive (in β2DA−VEC mice) nor negative value (in β2GABA−VEC mice) were followed by subsequent acquisition of steady, chronic ICSA or avoidance.
A critical role for nicotine-evoked DA burst firing
To identify the mechanisms underlying stable ICSA acquisition, we studied DA release in response to nicotine and subsequent downstream signaling in DAergic target areas. In vivo microdialysis in freely moving mice was carried out to measure DA release after systemic injection of nicotine. Unlike WT mice, but similar to KO animals, the DA vectorized group did not exhibit a nicotine elicited increase in free DA (Figure 5a and Supplementary Figure S6). We then assessed the impact of this lack of DA release on DA-dependent cellular modifications in the NAc. The mitogen-activated protein kinase/ERK signaling pathway is activated by most drugs of abuse, including nicotine, within neurons of the NAc.40 Interestingly, ERK activation caused by nicotine in the NAc critically depends on DA D1 receptor activation, as it is blocked by SCH23390, a D1 receptor antagonist. Furthermore, the blockade of ERK phosphorylation prevents immediate-early gene expression and long-lasting behavioral changes associated with exposure to drugs of abuse.41 Quantifying the number of phospho-ERK-positive nuclei in NAc revealed an abolition of ERK phosphorylation in response to nicotine in β2−/− mice (Figure 5b). This indicates a critical role of β2*-nAChRs in nicotine-elicited ERK activation in the NAc. It is of interest that we failed to restore ERK activation caused by nicotine in β2DA−VEC (Figure 5b). Therefore, nicotine elicited firing rate increase in β2DA−VEC is not sufficient to elicit DA release within the NAc, nor to activate downstream signaling pathways and maintain ICSA.
We then investigated the causes of transient ICSA. Because they have distinct influences on learning and motivation, tonic to phasic activity transitions are an essential component of the DA reward response.42, 43 Indeed, bursts are more effective in enhancing transmitter release,44 and activating immediate-early genes in DA target areas.45 Optogenetic methods have recently shown that the burst firing of DA neurons is a crucial property for reward.46 We have therefore analyzed the burst structures in all of our groups by calculating the percentage of spikes within a burst (%SWB=number of spikes within burst divided by total number of spikes in a 30 s time window, see Supplementary Material for details). Our previous work had shown that lentiviral re-expression of high-affinity nAChRs in the VTA using a ubiquitous PGK promoter results in nicotine-elicited burst firing.15 However, burst analysis in β2DA−VEC and β2GABA−VEC mice revealed that acute systemic nicotine did not evoke bursting (Figure 5c). Only WT and β2DA+GABA−VEC mice, vectorized specifically in DA and GABA neurons of the VTA, exhibited a significant increase in burst firing. The analysis of spontaneous, acetylcholine-dependent activity is shown for firing rate in Figure 5d. This analysis is carried out as described in our previous publication.15 For each neuron, spontaneous activity was quantified using firing rate (mean number of spikes per second) in Hz (Figure 5d, left) and %SWB (Figure 5d, right). These values are defined on the basis of at least 5 min of spontaneous activity before nicotine injection. It confirmed the role of GABA neurons in burst generation. In β2GABA−VEC and β2DA+GABA−VEC mice the spontaneous bursting pattern observed in the WT were restored, which was not the case in β2DA−VEC. These results reveal that burst firing is a cooperative effect resulting from the interplay between direct excitation and inhibition.
We then tested a β2DA+GABA−VEC group in ICSA to determine whether habit-forming actions of nicotine could be observed in the doubly re-expressed mice. We found their nicotine-induced response to be strikingly similar to WT mice (Figure 5e). In contrast to β2DA−VEC mice, β2DA+GABA−VEC displayed a rapid and steady acquisition of nicotine ICSA. This means that burst firing can be correlated with stable ICSA acquisition.
One of the critical steps in the transition from nicotine intake to addiction is the development of nicotine-elicited habits that contribute to stimulate and maintain chronic consumption. Self-administration is a powerful behavioral paradigm to investigate important aspects of these critical early steps toward nicotine addiction.47, 48 Identifying, in vivo, the neural circuitry underlying the acquisition of nicotine self-administration provides essential clues to our understanding of nicotine addiction, and its treatment.
Novel lentiviral vectors for the cell-type-specific expression of nicotinic receptors
One of the major difficulties in analyzing the nicotinic system is that almost every neuron in the brain, and cells in most other tissues,49 expresses nAChRs. We have therefore developed a general method for the neurotransmitter-specific expression of nAChRs through lentiviral vectors in the brain and applied it to the VTA of mice, the principal target of all drugs of abuse.6, 13 In the case of nicotine addiction, the cellular organization and relevant pharmacology of the VTA is complex as different nAChRs are present on a variety of cell types: DAergic projection neurons, local GABAergic neurons and terminals from cortical, pallidal and tegmental afferents (see Figure 1). Recently, also glutamatergic neurons have been identified in the VTA, but no data exist as to their expression of nicotinic receptors.34, 50 We have focused on the contribution of nAChRs on GABAergic neurons, which has been a challenging and little explored issue.8, 51 Our system allowed us to re-express exclusively the β2 nAChR subunit, responsible for high-affinity nAChRs, on either DAergic or GABAergic cells, or both in β2DA+GABA−VEC mice, in the VTA of β2 KOs. We demonstrate that the increase or decrease of DA firing frequency caused by nicotine acting exclusively on DA or GABA cells, respectively, correlates with positive or negative incentive value to drug-associated cues. However, only the concerted activation of nAChRs on DAergic and GABAergic cells correlates with the acquisition of a steady nicotine self-administration.
A crucial role for burst firing of DAergic neurons in nicotine reinforcement
We show that a particular firing pattern of the DA neurons, burst firing, is a critical physiological mechanism required for the long-term acquisition of nicotine self-administration. Tonic to phasic transition is an essential part of the functional role of the DAergic system.43 DA cell bursts of action potentials are related to elevated DA outflow and to the signaling of novel rewards or environmental stimuli.43, 52 Disruption of NMDA receptor-dependent burst firing impairs the acquisition of several conditioned behavioral responses and learning about cues that predict salient events.53 In addition, optogenetic tools have been used recently to demonstrate that DA bursting activity is sufficient to mediate reinforcement.46 Burst firing can be triggered by a number of events.54, 55 Among these are glutamatergic receptor activation,53 activation of voltage-gated ion channels following NMDA receptor activation,56 alteration of SK conductance57, 58 and decrease of tonic GABAergic inhibition. However, our results suggest that cholinergic input, relying on the activation of α4α6β2*-nAChR expressed within the VTA,15, 27, 29, 30, 59 play the role of a permissive gate that allows these various stimuli to trigger burst firing.60, 61, 62
In this paper we show, however, that β2*-nAChRs at the level of VTA DA cells alone are not sufficient to restore bursts evoked by nicotine. Accordingly, as a functional readout, we failed to restore ERK activation induced in the NAc by nicotine in β2DA−VEC mice and an increase in DA release. These data suggest, first, that an increase in frequency due to β2*-nAChRs on DA cells can be dissociated from enhancement of bursting, and second, that cholinergic input from tegmental nuclei also has a key role on GABA cells, and on burst control.
Redefining the role for VTA GABA neurons
The role we assign to GABAergic interneurons in the VTA differs significantly from proposals formulated over the past decade on the basis of data obtained with in vitro slice preparations.9, 11, 12 Although Mansvelder and McGehee9 did show some transient activation of GABAergic neurons in slices, all subsequent models describing the acute effect of nicotine on DA cell excitability were centered on disinhibition following GABA cell desensitization, and glutamatergic activation. Our results emphasize the role of ß2*-nAChRs on DA cells to increase firing, and importantly, on the interplay between DA and GABA activation to produce burst firing, the key property of DA neurons. An increase in GABAergic activity, and not a decrease, is thus necessary to shape DA cell bursting.
Our initial results on GABA neuron activation (Figure 2b) might be seen difficult to interpret directly as a GABA-mediated inhibition for two reasons. First, in these experiments we recorded ‘putative’ GABAergic neurons. Second, GABAergic neurons in the VTA are not exclusively interneurons as some of them project to the same forebrain targets as do the DA neurons.63 Nicotine-dependent excitation could thus be an exclusive property of GABAergic neurons that project outside the VTA. To circumvent conclusively these issues, we focused our electrophysiological analyses on DA cells, and on the effect of ß2*-nAChR activation on GABAergic interneurons using cell-type-specific expression of nicotinic receptors (Figure 3). Our results demonstrate that these GABAergic interneurons are activated by nicotine, and consequently inhibit DA cell firing (see Figure 3e). This major novel finding is substantiated by unambiguous identification and labeling of the recorded DA neurons (see Figure 3d and Supplementary Information). Interestingly, the minimal computational model with nAChRs expressed only on the DA cells also reproduces the shorter time courses of the DA response in β2DA−VEC animals as compared to the WT (compare green and grey lines in Figure 2e). And in the models with GABA-nAChRs, an increase above baseline after inhibition is observed in β2GABA−VEC (compare Figure 2e to Figure 2f). This further supports the notion that late, but not short-term, DA responses are governed by receptor desensitization.
We also show that cholinergic modulation of GABAergic cells is required for spontaneous DA neuron bursting. Recent data suggest that GABAergic input can act through phasic disinhibition as an excitatory drive, promoting rebound burst firing in DA neurons.64, 65 In line with this view, our results demonstrate conclusively that re-expression of the β2 subunit in GABA neurons allows spontaneous burst firing in DA neurons. In β2GABA−VEC mice, increased GABAergic input causes increased IPSPs in DA neurons, which has been shown to produce irregular firing patterns,66 allowing us to link the restoration of spontaneous bursting in DA neurons with the cholinergic modulation of GABA neuron activity.
Finally, burst firing and irregularity would require a balance between excitation and inhibition. Strong and synchronous activation of GABAergic cells, such as those caused by nicotine in β2GABA−VEC DA neurons, cannot evoke bursting in DA cells if it is not counterbalanced by excitation. We suggest that this balance is endorsed by β2*-nAChRs expressed, respectively, at the DA and GABA level within the VTA as suggested by our results on β2DA+GABA−VEC mice.
Towards a comprehensive understanding of the in vivo effects of nicotine
There are a number of conundrums in the study of the in vivo effects of nicotine: we show that, in mice, nicotine modulates the nAChRs in a few seconds, activation of DA neurons persists for about 10 min, and nicotine-dependent axonal release of DA lasts for 1–2 h. Finally, behavioral effects such as those observed in self-administration protocols take several days to be established. We have started to tackle the mechanisms of these different timescales and their significance. In particular, our computational modeling shows that a rapid phasic increase of DA activity, and hence a DA output spike, is likely to be due to the direct excitation of receptors on DA neurons, whereas the longer lasting increase of DA firing rate by nicotine is due to desensitization of GABA neuron nAChRs. The latter may also be a mechanism that shifts the DA cell activity from tonic to bursting.67 Further computational study has suggested that receptor desensitization by nicotine may play an important role in shaping DA responses to environmental stimuli, notably reducing the impact of negative motivational stimuli and rendering DA responding to previously learned stimuli again labile.68
It is generally admitted that tobacco and nicotine are repeatedly used because they provide positive feelings, and that this is linked with the increased activity in the mesocorticolimbic DA system.69 There is some initial evidence that VTA DAergic cells encode rewarding and aversive stimuli by increases and decreases of their firing frequency, respectively.39 We show here with our experiments using β2DA−VEC and β2GABA−VEC mice that this increase in DA firing frequency caused by nicotine correlates with a positive or negative incentive value to stimuli, whereas decreased DA firing frequency and increased VTA GABA neuron activity correlates with a negative incentive value. Indeed, we cannot exclude that VTA GABA neurons may alter reward function not only by their interaction with DA neurons and their burst patterns, but also with their interaction with the downstream targets in the mesolimbic and mesocortical DA terminal fields, or the tegmentum.10 However, this result suggests that nicotine can transmit positive and negative value as a function of its action on DAergic or GABAergic neurons in the VTA. A number of different drugs, and in particular nicotine, have been postulated to act through disinhibition of the DA system.14 According to this view, the GABAergic system acts to support and enhance excitatory input to DA neurons via somatic or presynaptic activation. Our results rather suggest that, while a predominant influence of GABA would elicit mainly aversive effects,10 the co-activation of DA and GABA systems is required for the habit-forming action of nicotine.70
A novel mechanism, potentially defining a novel drug target
We elucidate, for the first time, microcircuits engaged in vivo in a critical step of drug addiction and its regulation by endogenous neurotransmitters. We propose that within the VTA, DA-GABA neuron interplay is essential for the acquisition of persistent nicotine-directed behavior. Now, we need to further establish the role of GABAergic neurons in the chronic response to nicotine. Future studies should investigate how GABAergic activity is required for the long-term maintenance of nicotine dependence and the relevant loss of control, thus identifying a potentially novel drug target. Chronic nicotine exposure upregulates specifically the amount of nAChRs containing the α4 subunit in GABAergic neurons.51 Our study thus could pave the way for the development and pre-clinical testing of smoking cessation agents, targeting the heteromeric α4β2-containing receptors on GABAergic interneurons, and not DAergic receptors, as frequently attempted by pharmaceutical companies.3
We thank Benoît Forget and Henri Korn for helpful comments on the manuscript, and Jochen Roeper for help with slice electrophysiology. This work was supported by the Institut Pasteur, Centre National de la Recherche Scientifique CNRS URA 2182, UMR 7102, UMR 5287 and ATIP programme, the Agence Nationale pour la Recherche (ANR Neuroscience, Neurologie et Psychiatrie 2005 and 2009, and ANR BLANC 2009), Neuropôle IdF, INSERM U960, the RTRA Ecole des Neurosciences de Paris and the Bettencourt Schueller Foundation. The research leading to these results has received funding from the European Union Seventh Framework Programme under Grant no. HEALTH-F2-2008-202088 (‘NeuroCypres’ project to U.M.). ST acknowledges financial support from Fondation de la Recherche Médicale (FRM), and Pasteur-Weizmann.
About this article
Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)