Acute cocaine exposure occludes long-term depression in ventral tegmental area GABA neurons

Summary

The ventral tegmental area (VTA) in the midbrain is essential in incentive salience of reward behavior. Drugs of abuse increase midbrain dopamine cell activity and/or dopamine levels, and can alter endogenous VTA glutamate plasticity, leading to addiction or dependence. VTA dopamine cells are regulated by local inhibitory GABA cells, which exhibit a form of pre-synaptic cannabinoid receptor 1-dependent long-term depression of their glutamatergic inputs. Our current aim was to determine cocaine’s influence on VTA GABA cell glutamate plasticity and circuity. Using whole cell voltage-clamp electrophysiology in VTA slices of GAD67-GFP knock-in mice, we recorded excitatory inputs on GABA cells. Acute and chronic injections of cocaine were sufficient to occlude long-term depression. The plasticity could be reversed to the naïve state however, as long-term depression was observed following a 7-day abstinence from acute cocaine exposure. Furthermore, chronic cocaine decreased AMPA/NMDA ratios, compared to vehicle injection controls, the opposite change noted in dopamine cells. Collectively, our data suggest the cellular mechanism of cocaine-mediated synaptic modification that may result in dependence/withdrawal could involve changes in glutamate input to VTA GABA circuitry in addition to VTA dopamine cells, and therefore VTA GABA cells may play a larger role, possibly in a synergistic manner with dopamine, in overall cocaine-induced circuit changes than previously known.

Introduction

The mesolimbic system is critical for motivated behaviors behind reward. Essential in this system is the ventral tegmental area (VTA) composed of dopamine (DA) cells that signal reward, and neighboring gamma-aminobutyric acid (GABA) cells that among other features provide tonic inhibition of DA cells (van Zessen et al., 2012). The dopamine hypothesis suggests that increased dopamine results in learned reward responses and/or reward prediction error (Eshel et al., 2016). The mesolimbic system is heavily involved in dependence development in animals or “addiction”. For example, the common feature of drugs of abuse is that they increase VTA DA cell activity or synaptic DA levels. Subsequently, substance abuse leads to behavioral dependence and continued drug use in spite of the negative consequences. A key element in the mesolimbic system mediating dependence is synaptic re-arrangements, known as synaptic plasticity, that outlive the acute effects of the drug (Conrad et al., 2008; Ungless et al., 2001). This initiates adaptive changes that lead to drug-seeking behavior (Engblom et al., 2008). Indeed, all addictive psychoactive substances examined induce long-term changes of excitatory synapses of VTA DA neurons, while non-addictive psychoactive substances do not (Mameli and Luscher, 2011).

Cocaine specifically has a robust and profound effect on modifying plasticity of glutamatergic input to DA cells in every study examining this circuit in the VTA to date (Mameli and Luscher, 2011). A common form of endogenous plasticity is glutamatergic long-term potentiation (LTP), normally induced by NMDA receptors that enhance synaptic AMPA receptor numbers. Synaptic changes mediated by cocaine on glutamatergic terminals to DA cells include induction of long-term potentiation (Niehaus et al., 2010), increases in AMPA/NMDA ratios (Ungless et al., 2001), decreases in AMPA rectification indices (Creed et al., 2016), and altered synaptic AMPA receptor subtype expression (Isaac et al., 2007; Liu and Zukin, 2007). These synaptic changes occur even after a single dose cocaine exposure (Ungless et al., 2001). However, these synaptic alterations can revert to the naïve state following 1 week or more of abstinence after a single exposure (Ungless et al., 2001).

In addition, cocaine can indirectly alter endocannabinoid circuit equilibrium (Luchicchi et al., 2010; Mereu et al., 2015; Wang et al., 2015), a system also involved in synaptic plasticity. The principle endocannabinoid pathway components include cannabinoid receptor 1 (CB1) and its endogenous ligands 2-arachidonylglyerol (2-AG) and anandamide (Heifets and Castillo, 2009). Cocaine-induced increases in endocannabinoid release alter eCB plasticity (Mereu et al., 2015), and occluded eCB-mediated long-term depression (LTD) of glutamatergic synapses in the nucleus accumbens (Fourgeaud et al., 2004). While the role of cocaine-induced plasticity modification of synaptic glutamate inputs to VTA DA cells is well studied, the effect of acute/chronic cocaine on glutamatergic VTA GABA cell LTD is unknown, though this mechanism is eCB-dependent.

We recently identified this novel LTD of glutamatergic inputs to VTA GABAcells was dependent on presynaptic CB1 receptors that were activated by postsynaptically produced 2-AG following metabotropic glutamate receptor 5 activation (Friend et al.,2017). Δ9-tetrahydrocannabinol, the active ingredient in marijuana, mimicked thisplasticity following acute slice application, and occluded LTD following chronic intraperitoneal injections. As cocaine commonly modifies VTA glutamatergic synapsesand endocannabinoid signaling, our goal was now to examine the possible effect ofcocaine on eCB-dependent glutamatergic LTD onto VTA GABA cells.

Results

NMDA receptors are required for LTD

VTA GABA neurons were positively identified using a GAD67-GFP knock-in mouse line. These fluorescently labeled cells were patched using whole-cell voltage-clamp electrophysiology and we recorded glutamate excitatory post-synaptic currents (EPSCs). A 100Hz high frequency stimulus (HFS) protocol was used to induce a CB1-dependent LTD of glutamate inputs to VTA GABA cells as reported previously (Friend et al., 2017). In the VTA, plasticity of DA cells often require DA receptor (D1/D2) activation, while many forms of glutamatergic plasticity require NMDA receptors. As glutamate LTD of VTA GABA cells remains to be fully investigated, we examined the potential involvement of NMDA and D2 receptors in this plasticity. We observed HFSLTD was blocked in the presence of the NMDA receptor antagonist APV (Figure 1A), but not by D2 receptor antagonist eticlopride (Figure 1B).

Figure 1. Long-term depression (LTD) in VTA GABA cells is NMDA-dependent, and blocked by in vivo cocaine exposure.

A. In the presence of the NMDA receptor antagonist APV (50 μM), HFS-induced LTD is significantly blocked (n=9; p<0.05; compared to control LTD in Friend et al., 2017; t value = −10.73) in GAD67-GFP positive VTA GABA cells, demonstrating LTD dependence on NMDA receptors. Arrow indicates 100Hz high frequency stimulus (HFS). B. D₂ dopamine receptor antagonist eticlopride (10 μM) failed to block HFS-induced LTD (n=7; p>0.1 compared to control LTD in Friend et al., 2017; t value = −1.67), suggesting the D₂ receptor is not involved in LTD. C. Chronic (7–10 consecutive days) intraperitoneal injections of cocaine (15mg/kg) blocked HFS-induced LTD (n=6; p>0.5; f = 0.88). D. LTD remains present in mice chronically injected with saline (n=6; p<0.05; compared to chronic cocaine in C; t value = −10.99). E. A single injection of cocaine was sufficient to prevent LTD, and was not significantly different from chronic cocaine injection, (n=7; p>0.4; t value = 0.85) F. To determine whether cocaine-occluded HFS-induced LTD can revert to the naïve state, mice were given a single injection of cocaine followed by a period of abstinence. We again observed significant depression (n=7; p<0.05; compared to acute cocaine injected; t value = 17.47) following 7 days of abstinence, illustrating acute cocaine occlusion of HFS-induced LTD is reversible. All scale bars 100 pA, 10 msec. Plots, mean with s.e.m.

Cocaine occludes LTD

In addition, drug-induced alterations of VTA plasticity correlate to induction and duration of dependence/withdrawal, and behavioral changes. While glutamate plasticity to DA cells is modified by cocaine, the effect of cocaine on the VTA GABA cells, which compose ~30% of VTA cells, is relatively unexplored. To investigate the role of cocaine in the inhibitory circuit with a focus on glutamate LTD of VTA GABA cells, mice were exposed to cocaine in vivo via intraperitoneal injection and plasticity was examined 24 hours after the last injection day. Following chronic cocaine injections (15mg/kg), HFS failed to induce LTD (Figure 1C), which was significantly different from the chronic saline vehicle injections where LTD remained (Figure 1D). Similarly, even a single cocaine injection 24 hours prior to recording, blocked LTD (Figure 1E). As single-injection cocaine-evoked alterations in DA cell plasticity can be reversed to the naïve state following a period of 1-week abstinence, we investigated whether the LTD in question could be recovered as well. To do this, mice were injected with a single dose of cocaine, followed by 7 days of abstinence before recording. In this scenario, HFS-induced LTD recovered (Figure 1F) to a level similar to naïve mice (Friend et al., 2017), suggesting synaptic changes following acute cocaine exposure persist for less than 7 days. These studies suggest cocaine can reversibly influence plasticity of VTA GABA neuron LTD, similar to DA neuron LTP following a single exposure.

Cocaine decreases AMPA/NMDA ratio

AMPA/NMDA ratio and AMPA receptor subunit composition changes are additional indicators of drug-evoked synaptic alterations. To examine this we recorded VTA GABA neurons between −70 and +40 mV in the presence and absence of APV with spermine added to the intracellular solution. The AMPA/NMDA ratio from control mice exposed to chronic saline (7–10 days) averaged 1.32 ± 0.18, while mice treated with chronic cocaine (7–10 days) had a significantly decreased AMPA/NDMA ratio, averaging 0.87 ± 0.08 (Figure 2A).

Figure 2. Cocaine decreases AMPA/NMDA ratio.

A. Alterations in the ratio of AMPA/NMDA receptor current provides additional evidence for glutamate synapse plasticity. Chronic cocaine injections significantly decreases AMPA/NMDA ratios (n=12 cells; p<0.05; t value = 3.77; open circles represent individual experiment values while filled circles indicate overall average) compared to saline injected controls (n=11). Insets represent two individual experiments of saline (left) or chronic cocaine (right) injections. B. Current-voltage (IV) plots for chronic cocaine and saline treated mice illustrate no significant difference in rectification indices at positive potentials between the cocaine-treated group (n=10) and saline-treated control group (n=11; p>0.7; two-way ANOVA; f = 1.50). Arrowhead: Note no change in rectification. C. Measuring the ratio of the current at −70mV to +40mV is another way to determine changes in AMPA receptor rectification. No significant change was observed between chronic cocaine (n=10) and saline treatments (n=11; p>0.5; t value = 0.44). D. The CB1 agonist WIN55,212-2 still induced depression of synaptic glutamate currents following chronic cocaine exposure (n=6; p> 0.3 compared to control WIN55,212-2 induced depression, Friend et al., 2017; t value = 0.89).

In order to determine whether cocaine-induced decreased AMPA/NMDA ratio was accompanied with an alteration in AMPA receptor subunit expression, we analyzed current-voltage (IV) plots to look for changes in rectification that are indicative of GluA2-lacking AMPA receptor subunits. No significant difference in the rectification of the AMPA receptor currents was observed among the GABA neurons from the chronic cocaine injected mice compared to saline-injected control. This suggests the change in AMPA/NMDA ratio was not accompanied by a change in AMPA receptor subunit expression, but an overall reduction in AMPA receptor numbers (Figure 2B). Ratios of AMPA currents at −70/+40mV supported this finding as they also reveal AMPA receptor subunit composition changes (Figure 2C), and were not significantly different between control and cocaine-treated groups. Collectively, these suggests there is no change to AMPA receptor subunit composition.

To identify mechanistically whether cocaine-occluded LTD is via either CB1 receptor desensitization/internalization or inhibition of postsynaptic production of endocannabinoids initializing LTD, we examined CB1 receptor activity following chronic cocaine exposure. The CB1 agonist WIN 55,212-2 continued to induce depression of glutamate synaptic currents following chronic cocaine injections (Figure 2D). This illustrates CB1 receptors are viable, and that inhibition of postsynaptic production of endocannabinoids is the likely cause of LTD occlusion by cocaine.

Discussion

We present novel actions of cocaine on glutamate receptor expression and plasticity of the excitatory inputs to VTA GABA cells, which implicate cocaine in modifying the GABAergic reward circuit in addition to the dopaminergic circuit that could play an additional role in cocaine’s reward/withdrawal effects. Collectively, we demonstrated LTD of VTA GABA cells is NMDA receptor-dependent and occluded by both acute and chronic cocaine intraperitoneal injections. Furthermore, chronic cocaine decreases synaptic AMPA receptor numbers. This is the first demonstration of cocaine-induced changes to the VTA inhibitory circuit via altering excitatory inputs to VTA GABA cells.

Cocaine’s influence on excitatory inputs to VTA DA neurons is well known where glutamate plasticity is altered just 24-hours after a single intraperitoneal injection of cocaine (Creed et al., 2016; Lüscher, 2013; Ungless et al., 2001). Also, in every study to date, cocaine modified glutamatergic input plasticity to VTA DA cells (Mameli and Luscher, 2011). However, cocaine’s role in altering glutamate input plasticity to VTA GABA cells is yet to be investigated. The lack of studies is likely because plasticity of glutamate inputs to VTA GABA cells was only recently identified (Friend et al., 2017). In addition, as cocaine occluded LTP of nucleus accumbens medium spiny neuron GABAergic projections to VTA GABA cells (Bocklisch et al., 2013), this further illustrates the potential implications of cocaine on the VTA GABA circuit in drug-dependence, justifying our current study examining their glutamate inputs. Our results are consistent with the timing of acute cocaine-induced glutamate LTP of DA neurons, as also only a single dose was required to occlude LTD of glutamate inputs to VTA GABA cells. We also note a reversal of LTD occlusion after 7 days withdrawal, following a single cocaine exposure. This is noteworthy due to its consistency with the reacquisition of LTP in DA neurons after ~7 days of withdrawal, following a single dose of cocaine (Borgland et al., 2004; Ungless et al., 2001). This illustrates withdrawal time points are similar in both DAergic and GABAergic VTA circuits. There is also prior precedence for cocaine modulating plasticity via CB1 receptors, where cocaine altered endocannabinoid-induced plasticity of GABA inputs onto VTA DA cells (Pan et al., 2008) and occluded CB1-dependent LTD in the nucleus accumbens (NAc) after a single dose (Fourgeaud et al., 2004). Our data support cocaine modification of endocannabinoid plasticity. Mechanistically, likely either cocaine induces LTD and thus we cannot initiate further LTD (occlusion), or it inhibits/blocks the ability of this synapse to exhibit LTD possibly by decreasing endocannabinoid production. As CB1 agonist continue to depress synaptic glutamate responses after chronic cocaine exposure, it suggests the latter is true. One potential option is that cocaine could decrease surface mGluR5 as noted in other brain regions (Fourgeaud et al., 2004), which could cause this LTD occlusion as this CB1-dependent form of LTD requires mGluR5-mediated production of 2-AG. The mechanisms of LTD occlusion is a matter of current/future investigation.

As cocaine also modulates synaptic AMPA receptor expression levels and receptor subtype, as measured by AMPA/NMDA ratios, IV plot of AMPA receptors, and −70/+40 mV AMPA receptor current ratios (Lüscher and Malenka, 2011), we examined AMPA receptor expression and subtypes using these techniques. Changes in the AMPA receptor subunit composition are implicated in cocaine cravings and relapse (Conrad et al., 2008). Cocaine specifically signals for the insertion of GluA2-lacking AMPA receptors in DA neurons (Engblom et al., 2008), which are unique in their ability to conduct calcium (Brown et al., 2010), and cause an increase in AMPA/NMDA ratios (Ungless et al., 2001). As there are several reasons for AMPA/NMDA ratio changes, additional analyses are critical to perform to discriminate between them (Mameli and Luscher, 2011). For example, the AMPA/NMDA ratio increase following a single injection of cocaine (Ungless et al., 2001) are now attributed to a replacement of synaptic GluA2 receptors with GluA2-lacking receptors (Lüscher and Malenka, 2011; Mameli et al., 2007) versus an increase in AMPA receptor numbers. Regarding prior work on VTA GABA cells, changes in AMPA/NMDA ratio were not noted in VTA GABA cells in response to a single cocaine injection (Ungless et al., 2001), though chronic cocaine exposure was never examined. However, as repeated cocaine-exposure altered synaptic AMPA receptor numbers of NAc GABAergic medium spiny neurons (Thomas et al., 2001); we determined it worth examining chronic cocaine injections as to their relevance on synaptic AMPA receptor expression of VTA GABA cells. Our data demonstrated an unexpected result as chronic cocaine decreased AMPA/NMDA ratio, in contrast to single injection cocaine-induced increase in this ratio in DA neurons, which illustrate increased AMPA/NMDA ratio. This illustrates that VTA GABA cells require repeated cocaine exposure compared to DA neurons for synaptic AMPA receptor modification, and that this would likely have the same synergistic effect on DA cell activity via disinhibition. Our IV plot and −70/+40 mV ratio data for AMPA receptors suggest AMPA receptor subtypes are not changing in response to cocaine, unlike DA neurons, but are truly a reduction in AMPA receptor numbers. Collectively, this data illustrates reduced synaptic AMPA receptor numbers following chronic cocaine administration. This would result in less excitation of VTA GABA neurons in the circuit by glutamate subsequent to cocaine exposure, and reduce inhibition onto VTA DA neurons, or disinhibtion.

It is of interest to note that there is a variety of AMPA receptor currents ranging from rectified to non-rectified in both saline and cocaine treated mice in our study, suggesting variations between different VTA GABA cells in the number of GluA2-lacking (rectifying) and GluA2-containing (non-rectifying) AMPA receptors, unlike DA neurons that mainly express GluA2-containing receptors under naïve conditions. These variations are easily noted in the −70/+40 ratio data (see fig. 2C). One last caveat to consider is that the AMPA/NMDA ratio could be changing due to increases in synaptic NMDA receptor numbers (i.e. as others noted cocaine likely caused decreased NMDA receptors in DA cells (Mameli et al., 2011)). While this cannot be completely ruled out, it is less likely and any change in NMDA or AMPA receptors is evidence of a psychoactive drug evoking synaptic and reward circuit adaptations via VTA GABA cells.

There are several potential implications of cocaine impact via VTA GABA cells on reward/dependence. First, as prior studies noted increased GABA cell activity inhibits DA cells activity and DA release (Tan et al., 2012; van Zessen et al., 2012); a depression of excitatory inputs onto GABA cells could disinhibit DA release and increase reward. Thus, this LTD could be an additional means whereby the rewarding effects of cocaine are mediated. As cocaine occludes this LTD, it is also possible that this mechanism mediates some of the withdrawal effects of cocaine. In the NAc, chronic cocaine similarly decreased the AMPA/NMDA ratio and occluded LTD, which was suggested to lead to a behavioral component of addiction (Thomas et al., 2001). There is also evidence of cocaine relevance to behavior via the GABA circuit. For example, cocaine-seeking is reduced by drugs depressing GABA levels (Gardner et al., 2002), and chronic cocaine reduces the strength of GABA inhibition to facilitate DA cell LTP (Liu et al., 2005). Indeed, it was proposed that cocaine alteration of LTD of direct GABAergic inputs onto VTA DA neurons might contribute to incentive sensitization of cocaine-associated cues (drug-associated incentive salience) (Pan et al., 2008). Our report illustrates that cocaine modification through VTA GABA neurons, along with prior known impact of cocaine on DA neurons, may synergistically have a similar effect.

In summary, drug-induced plasticity alterations in the mesocorticolimbic system are a consideration of importance as drugs of abuse often change the ‘rules’ of synaptic plasticity (Mameli et al., 2011). In essence, abused drugs induce aberrant forms of plasticity causing a change in brain circuitry that remains fixed, long after the drug is removed from the system, likely mediating reinforcing properties of the drug. While literature on drug-induced plasticity changes of VTA GABA circuit is relatively sparse, this study illustrates cocaine-induced effect on GABA cell glutamate synapses. This illustrates that future considerations should turn more focus towards understanding cocaine’s impact on the GABA circuit in addition to the dopamine circuit in drug-dependence with the former also potentially involved in drug-dependence and/or withdrawal. Collectively, cocaine modifies plasticity of glutamate inputs and glutamate receptor number to VTA GABA cells, thus demonstrating an additional mechanism by which cocaine modulates the reward circuit.

Methods

All experiments were performed in accordance with Institutional Animal Care and Use Committee (IACUC) protocols and followed National Institute of Health guidelines for the care and use of laboratory animals. IACUC protocols for all experiments were approved by the Brigham Young University Institutional Animal Care and Use Committee.

Most experiments were performed as described previously (Friend et al., 2017). Briefly, male and female CD-1 GAD67-GFP knock-in mice were used to identify GABA cells in the VTA. Mice (15–35 day old) were anesthetized with isoflurane and decapitated with a rodent guillotine. Brains were rapidly removed and sectioned horizontally on a vibratome at 300 μm. Recordings began following at least one hour after cutting while tissue was stored in oxygenated artificial cerebral spinal fluid composed of 119 mM NaCl, 26 mM NaHCO3, 2.5mM KCl, 1 mM NaH2PO4, 2.5mM CaCl2, 1.3mM MgSO4, and 11mM glucose at 34°C for 1 hour and then room temperature. Picrotoxin (100μM) was added to the recording solution to block GABA currents.

The VTA was visualized using an Olympus, BX51W1 microscope with a 40x water immersion objective. Patch pipette resistance was ~2.5–4.5 MΩ. Cells were patched with a glass pipette filled with internal solution composed of 117mM cesium gluconate, 2.8mM NaCl, 20mM HEPES, 5mM MgCl2, 0.6 mM EGTA, 2 mM ATP, 0.3 mM GTP and 1mM QX-314 (pH 7.28, 275–285 mOsm). Cells were recorded in voltage clamp mode at −65 mV. While generating AMPA/NMDA ratios and IV plots, spermine (0.1mM) was included in the internal solution to avoid polyamine dialyzation and maintain its block of GluA2-lacking AMPA receptors at depolarized potentials, thereby determining the subunit composition of synaptic AMPA receptors.

Currents were measured using Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA). Signals were filtered at 4 kHz and digitized with an Axon 1440A digitizer (Molecular Devices) connected to a Dell personal computer with pClamp 10.2 or 10.7 Clampfit software (Molecular Devices).

Plasticity was induced with the application of two high frequency stimuli (100Hz, 1-second duration, 2 times 20 second apart). To obtain the AMPA/NMDA ratio, cells were first held at −70mV during whole cell recording. Prior to the recording, at least 10 minutes was given for spermine to dialyze into the cell; the holding potential was then raised to +40mV and APV (50μM) was applied subsequently to isolate AMPA currents. Immediately after the AMPA currents were recorded, the holding potential was moved back gradually from +40 to −70 mV in +20 mV steps in order to create IV plots. During data analysis, NMDA currents were calculated by digital subtraction of currents before and after APV application at +40 mV. The −70/+40 ratio was calculated based on the average currents at these holding potentials. For statistics, Student’s t test was used to compare between two groups, while ANOVA was employed to compare within an individual experiment or single group of experiments. Differences were considered significant at p < 0.05. No differences were noted between males and females during data analysis.

Picrotoxin was purchased from Abcam, APV and QX-314 were purchased from Tocris, and spermine was purchased from Sigma. Cocaine HCl was purchased from Sigma or supplied by the NIDA Drug Supply Program. Cocaine was administered via intra-peritoneal injections at 15 mg/kg. All salts were purchased from Fischer, Sigma or Mallinkrodt-Baker.