Abstract
Drugs of abuse cause changes in the prefrontal cortex (PFC) and associated regions that impair inhibitory control over drug-seeking. Breaking the contingencies between drug-associated cues and the delivery of the reward during extinction learning reduces relapse. Vagus nerve stimulation (VNS) has previously been shown to enhance extinction learning and reduce drug-seeking. Here we determined the effects of VNS-mediated release of brain-derived neurotrophic factor (BDNF) on extinction and cue-induced reinstatement in male rats trained to self-administer cocaine. Pairing 10 d of extinction training with VNS facilitated extinction and reduced drug-seeking behavior during reinstatement. Rats that received a single extinction session with VNS showed elevated BDNF levels in the medial PFC as determined via an enzyme-linked immunosorbent assay. Systemic blockade of tropomyosin receptor kinase B (TrkB) receptors during extinction, via the TrkB antagonist ANA-12, decreased the effects of VNS on extinction and reinstatement. Whole-cell recordings in brain slices showed that cocaine self-administration induced alterations in the ratio of AMPA and NMDA receptor-mediated currents in Layer 5 pyramidal neurons of the infralimbic cortex (IL). Pairing extinction with VNS reversed cocaine-induced changes in glutamatergic transmission by enhancing AMPAR currents, and this effect was blocked by ANA-12. Our study suggests that VNS consolidates the extinction of drug-seeking behavior by reversing drug-induced changes in synaptic AMPA receptors in the IL, and this effect is abolished by blocking TrkB receptors during extinction, highlighting a potential mechanism for the therapeutic effects of VNS in addiction.
Significance Statement
Extinction training can reverse maladaptive neuroplasticity induced by drugs of abuse, but adjunct treatments are sought that can facilitate the process and consolidate the newly formed memories. Pairing extinction training with vagus nerve stimulation (VNS) facilitates extinction and reduces drug-seeking behavior during reinstatement. Here, we show that rats receiving a single extinction session with VNS exhibit elevated brain-derived neurotrophic factor (BDNF) levels in the medial prefrontal cortex (mPFC). We also demonstrate that VNS consolidates the extinction of drug-seeking behavior by reversing cocaine-induced changes in synaptic AMPA receptors in the infralimbic cortex of the mPFC. This effect is blocked by the TrkB antagonist ANA-12, emphasizing the role of BDNF and TrkB receptors in the therapeutic effects of VNS in addiction.
Introduction
Exposure to drug-associated cues or stress induces craving and relapse in abstinent patients with substance use disorder (Ehrman et al., 1992). During extinction learning, new associations are formed that compete with these triggers to inhibit behavioral responses (Millan et al., 2011). However, extinction training alone is often insufficient to prevent relapse (Weiss et al., 2001; Conklin and Tiffany, 2002), potentially because the corticolimbic networks that regulate cue and reward processing, which include areas in the medial prefrontal cortex (mPFC), the amygdala, and the ventral striatum, become dysregulated by drug use (Fowler et al., 2007; Liu et al., 2009; Nic Dhonnchadha and Kantak, 2011). Modulating extinction processes and consolidating the newly formed memories are therefore clinically relevant to reshape maladaptive behavior and prevent relapse (Taylor et al., 2009).
Vagus nerve stimulation (VNS) is clinically used to treat epilepsy (Ben-Menachem et al., 1994) and depression (Rush et al., 2005) but is considered for the treatment of an expanding range of neurologic conditions (Tyler et al., 2017; Dawson et al., 2021). VNS causes an increase in the levels and/or release of several neuromodulators, including brain-derived neurotrophic factor (BDNF; Dorr and Debonnel, 2006; Roosevelt et al., 2006; Follesa et al., 2007; Nichols et al., 2011; Manta et al., 2013), which modulate cortical plasticity (Cao et al., 2016; Collins et al., 2021), and this plasticity can facilitate extinction training and reduce cue-induced reinstatement of drug-seeking (Liu et al., 2011; Childs et al., 2017, 2019). However, the networks and neurotransmitters that support these effects are still poorly understood.
Cocaine self-administration induces aberrant plasticity at glutamatergic synapses in the mPFC, leading to altered salience attribution and cue-induced drug-seeking (Otis and Mueller, 2017; Caffino et al., 2018; Sasase et al., 2019). Repeated exposure to cocaine and withdrawal also lower the level of BDNF in the mPFC (Fumagalli et al., 2007; McGinty et al., 2010; Pitts et al., 2016). BDNF modulates cocaine-induced changes in glutamatergic transmission in the mPFC (Berglind et al., 2007; Otis et al., 2014; Go et al., 2016; Barry and McGinty, 2017; Otis and Mueller, 2017) and supplementation of BDNF via a single infusion into the mPFC directly after a final cocaine self-administration session reduces cue- and cocaine prime-induced reinstatement following extinction or abstinence (Berglind et al., 2007; Whitfield et al., 2011; Go et al., 2016; Barry and McGinty, 2017). We hypothesized that cocaine self-administration alters glutamatergic signaling in the infralimbic cortex (IL), an area in the mPFC that is important for the expression of extinction memories (Peters et al., 2009; Augur et al., 2016; Gutman et al., 2017; Muller Ewald et al., 2019). We also hypothesized that pairing the extinction training with VNS leads to BDNF release in the mPFC which can reverse these changes and reduce reinstatement. Our data show that VNS-evoked release of BDNF consolidates extinction of drug-seeking behavior by reversing drug-induced changes in synaptic AMPA receptors in the IL, and this effect is abolished by blocking tropomyosin receptor kinase B (TrkB) receptors during extinction.
Materials and Methods
Subjects
We used male Sprague Dawley rats (Taconic Biosciences) that were at least 80 d old (250–300 g) at the time of surgery. Rats were individually housed and kept on a 12 h. reverse light/dark cycle, with ad libitum access to food and water until surgery, when food was restricted to 25 g/d standard rat chow. All protocols were approved by the International Animal Care and Use Committee of The University of Texas at Dallas and were conducted in compliance with the NIH Guide for the Care and Use of Laboratory Animals.
Drug self-administration and extinction training
Drug self-administration and extinction training were performed as previously described (Childs et al., 2017, 2019): male Sprague Dawley rats (>80 d) were anesthetized and implanted with a catheter in the right external jugular vein for drug administration. During the same surgery, a custom-made cuff electrode was placed around the left vagus nerve for the delivery of VNS (Childs et al., 2015). Five to seven days following surgery, rats were trained in a single overnight session to self-administer food pellets (45 mg, Bio-Serv) in an operant conditioning chamber (Med Associates). Drug self-administration training took place in the same chamber, which was equipped with two levers, a house light, a cue light, and a tone. Each active lever press produced a 0.05 ml infusion of 2.0 mg/ml cocaine (Sigma-Aldrich) in saline, and the presentation of drug-paired cues (illumination of the light over the active lever and the presentation of a 2,900 Hz tone), followed by a 20 s time-out. Self-administration sessions ended after 2 h. Both right and left levers were available for the duration of the session, and drug-seeking behavior was quantified as active lever presses. Rats self-administered cocaine for 15–18 d, during which they had to achieve at least 20 infusions per session. Subjects in the extinction groups underwent 10 d of extinction training in which lever presses on the previously active lever no longer produced cocaine or presentation of drug-paired cues. During extinction training, rats received either sham stimulation or VNS. Noncontingent VNS stimulation occurred every 5 min for 30 s at 0.4 mA for the duration of the training session. These stimulation parameters are similar to those previously used to enhance retention performance in rats (Clark et al., 1995) and humans (Clark et al., 1999). Previous work has shown that pairing extinction training with either brief contingent VNS (i.e., 500 ms VNS pulses delivered with each nonreinforced lever press) or noncontingent VNS (as described above) can reduce cue-induced reinstatement (Childs et al., 2017). Here we used noncontingent VNS to ensure that all subjects received the same number of VNS deliveries. After 10 d of extinction training, drug-seeking behavior was reinstated by the presentation of the drug-associated cues in the operant conditioning chambers. During the reinstatement session, presses on the “active” lever led to the presentation of the previously drug-associated tone and light, but did not result in drug delivery or VNS. Additional groups of rats receive yoked-saline injections during the self-administration sessions, to serve as controls in electrophysiological experiments examining the effects of drug self-administration on glutamatergic synaptic transmission.
TrkB receptor antagonism
Rats received intraperitoneal injections of the TrkB receptor antagonist ANA-12 (0.5 mg/kg in DMSO) or DMSO vehicle at a volume of 0.5 ml/kg 1 h prior to the start of the session during the 10 d of extinction learning (but not during the reinstatement session).
Electrophysiology
Rats were anesthetized with an overdose of urethane (3 g/kg body weight; Thermo Fisher Scientific) and transcardially perfused for 1 min with ice-cold oxygenated (95% O2, 5% CO2) cutting ACSF, consisting of the following (in mM): 110 choline (Sigma-Aldrich), 25 NaHCO3 (Thermo Fisher Scientific), 1.25 NaH2PO4 (Thermo Fisher Scientific), 2.5 KCl (Sigma-Aldrich), 7 MgCl2 (Sigma-Aldrich), 0.5 CaCl2 (Sigma-Aldrich), 10 dextrose (Thermo Fisher Scientific), 1.3 L-ascorbic acid (Thermo Fisher Scientific), and 2.4 Na+- pyruvate (Sigma-Aldrich). Brains were extracted and coronal sections (350 μm) of the frontal cortex were cut on a vibratome (VT1000S, Leica Biosystems) in cutting ACSF. Slices were transferred to a holding chamber containing warmed (35°C) recording ACSF and cooled to room temperature over a 1 h period. The recording ACSF consisted of the following (in mM): 126 NaCl (Thermo Fisher Scientific), 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 2 MgCl2, 2 CaCl2, 10 dextrose, 2.4 Na+-pyruvate, and 1.3 L-ascorbic acid. For data collection, slices were transferred to a recording chamber affixed to an Olympus BX61WI microscope (Olympus) with continuous perfusion of oxygenated recording ACSF at room temperature. Whole-cell voltage–clamp recordings were obtained from Layer 5 pyramidal cells in the IL using Axon MultiClamp 700B amplifiers (Molecular Devices). Data were acquired and analyzed using AxoGraph X (AxoGraph Scientific). Recording electrodes (World Precision Instruments; 4–6 MΩ open-tip resistance) were filled with an internal solution consisting of the following(in mM): 130 CsCl (Sigma-Aldrich), 20 tetraethylammonium chloride (Sigma-Aldrich), 10 HEPES (Sigma-Aldrich), 2 MgCl2, 0.5 EGTA (Sigma-Aldrich), 4 Mg2+-ATP (Sigma-Aldrich), 0.3 lithium-GTP (Sigma-Aldrich), 14 phosphocreatine (Sigma-Aldrich), and 2 QX-314 bromide (Tocris Bioscience). Theta-glass pipettes (Warner Instruments) connected to a stimulus isolator (World Precision Instruments) were used for focal stimulation of synaptic potentials. Access resistance was monitored throughout the recording, and a < 20% change was deemed acceptable. EPSCs were isolated by blocking chloride channels with the addition of picrotoxin (75 μM; Sigma-Aldrich) into the recording ACSF. The ratio of currents through AMPA or NMDA receptors (AMPAR:NMDAR) was obtained by clamping cells at +40 mv holding potential and applying local electrical stimulation. A compound evoked EPSC (eEPSC) was first recorded, and then the AMPA component was isolated by washing (±)-3-(2-carboxypiperazin-4-yl)propyl-1-phosphonic acid (10 μM; Sigma-Aldrich) into the bath. A minimum of 20 sweeps each were average for the compound and AMPA-only eEPSCs. The NMDA component was then obtained by digital subtraction of the AMPA component from the compound trace. The peak amplitude of the NMDA and AMPA traces were used to calculate the NMDAR:AMPAR ratio. For all recordings a minimum of four rats were used per treatment group.
BDNF ELISA
One cohort of rats was killed immediately after a single day of extinction training with VNS or sham stimulation to determine VNS-induced levels of BDNF in the mPFC via an enzyme-linked immunosorbent assay (ELISA). Brains were extracted and frozen in 2,3-methylbutane on dry ice and stored at −80°C. They were transferred to a −20°C freezer 24 h prior to tissue collection. Using chilled single-edged blades, the cerebellum was removed prior to placing the brain cortex-side up in a coronal rat brain matrix (Harvard Apparatus). The 2-mm-thick cortical tissue sections (∼4.5–2.5 mm anterior to the bregma) were cut with razor blades, and 2.0-mm-diameter medial prefrontal cortical tissue punches were collected on dry ice using Integra Miltex Biopsy Punches with Plunger System (Thermo Fisher Scientific). The punches were placed in labeled, prechilled 2 ml Pre-Filled Bead Tubes with 2.8 mm Ceramic Beads (VWR International), and weighed. Samples were homogenized in 250 µl of 1× PBS using a Bead Mill Homogenizer (VWR International) at a speed of 4.85 m/s set for two 20 s cycles at an interval of 11 s. An additional 250 µl of Lysis Buffer 17 (R&D Systems) and 2.5 µl of 100× Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific) were added, followed by 30 min of gentle agitation and 10 min of centrifugation (500 × g, 40°C) before the supernatant was collected. Total BDNF Standard was reconstituted in 1 ml milliQ water and Calibrator Diluent RD5K (450 μl). This stock solution was used to produce a dilution series with the following concentrations (in pg/ml): 1,000, 500, 250, 125, 62.5, 31.3, 15.6, 0. BDNF standards and tissue protein levels were run in triplicate per manufacturer instructions using Total BDNF Quantikine ELISA Kit (catalog #DBNT00, R&D Systems), a sandwich enzyme immunoassay technique. Optical density of each sample was read at 450 nm with a wavelength correction set to 540 nm using a BioTek Gen5 Microplate Reader and Imager Software (Agilent Technologies). The triplicate OD measurements for each standard and tissue sample were averaged, and the BDNF concentration in picograms per milliliter was calculated and converted to picograms per milligram tissue weight.
Experimental design and statistical analyses
All statistical analyses were performed in GraphPad Prism 7.0.5 (GraphPad Software). A Shapiro–Wilk test was used to assess the normal distribution of data.
Experiment 1
We compared active and inactive lever presses, respectively, during both the 10 d self-administration period and the 10 d extinction period in sham-stimulated (n = 10) and VNS-treated rats (n = 11) using separate two-way ANOVAs with the factors group and time. A one-way ANOVA was performed on the cue-induced reinstatement session. Post hoc analyses of main effects and interactions used Tukey's multiple-comparison tests.
Experiment 2
We compared active lever presses during the self-administration period using a two-way ANOVA with the factors time (sessions) and group (sham + vehicle, n = 10; sham + ANA12, n = 11; VNS + vehicle, n = 13; VNS + ANA-12, n = 9). Lever presses during the extinction period were analyzed using a separate three-way ANOVA with the factors stimulation (VNS vs sham), drug (vehicle vs ANA-12), and time. A two-way ANOVA with the factors stimulation and drug was performed on the cue-induced reinstatement session. Post hoc analyses of main effects and interactions were performed using Tukey's multiple-comparison tests.
Experiment 3
Single-session extinction data and BDNF ELISA data for sham-stimulated (n = 11) VNS–treated rats (n = 12) were compared using unpaired two-tailed t tests.
Experiment 4
We compared AMPAR:NMDAR ratios in three groups (naive, 8 recordings from four rats; yoked-saline, 9 recordings from seven rats; cocaine, 10 recordings from eight rats) using a one-way ANOVA. Post hoc analysis was performed using Tukey's multiple-comparison test.
Experiment 5
Data obtained in this experiment stemmed from a subset of the rats used in Experiment 2. We compared AMPAR:NMDAR ratios (as well as the changes in the individual currents) in four groups (sham + vehicle, eight cells from four rats; sham + ANA12, six cells from four rats; VNS + vehicle, eight cells from six rats; VNS + ANA-12, eight cells from seven rats) using separate two-way ANOVAs with the factors stimulation (sham or VNS) and drug (vehicle or ANA-12). Post hoc analyses of main effects and interactions were performed using Tukey's multiple-comparison tests.
An α level of p < 0.05 was considered significant in all experiments. Data are expressed as mean ± standard error of the mean. Individual responses are plotted over averaged responses. Asterisks on all figures represent differences between groups as indicated in the figure legends. Experimental designs and sample sizes are aimed at minimizing the usage of animals and are sufficient for detecting robust effect sizes. While no statistical methods were used to predetermine sample sizes, our sample sizes are similar to those reported in previous publications.
Results
Experiment 1: VNS facilitates extinction learning and reduces cue-induced reinstatement
In order to determine the effects of VNS on drug-seeking behavior, rats were trained to self-administer cocaine for 15–18 d, followed by 10 d of extinction paired with VNS (n = 11) or sham stimulation (n = 10). After 10 d, reinstatement to drug-seeking was measured in a cued reinstatement session by presenting the conditioned drug cues (Fig. 1). A two-way ANOVA with the factors group and time (session) showed that there were no differences in the rates of responding on the active lever between groups over the last 10 d of drug self-administration (group, F(1,19) = 0.021, p = 0.886; time, F(9,171) = 0.49, p = 0.88), and responses on the active lever exceeded responses on the inactive lever in both groups (two-way ANOVAs; sham, F(1,18) = 37.92, p < 0.0001; VNS, F(1,20) = 37.05, p < 0.0001). We used responses at the previously active lever as an indicator of extinction learning. A repeated-measure two-way ANOVA with the factors time (sessions) and treatment (sham or VNS) showed a significant degree of extinction (main effect of time; F(9,170) = 12.91; p < 0.0001), a significant difference between treatment groups (main effect of treatment; F(1,19) = 17.48; p = 0.0005; Fig. 1A), and an interaction between the factors (F(9,171) = 3.26; p = 0.0011). Post hoc analysis using Tukey's multiple-comparison test showed that sham-simulated rats pressed the active lever significantly more on the first 5 d of extinction (Day 1, p < 0.0001, Fig. 1B; Day 2, p = 0.0002; Day 3, p = 0.0056; Day 4, p = 0.0040; Day 5, p = 0.0099). A separate two-way ANOVA for responses at the previously inactive lever showed significant treatment differences (F(1,19) = 5.507; p = 0.03; Fig. 1A). Post hoc analysis with Tukey's multiple-comparison test showed that sham-stimulated rats responded more often at the previously inactive lever on Day 1 of extinction (p = 0.044, Fig. 1B). Twenty-four hours after the last extinction session, drug-seeking was reinstated in a cued reinstatement session. Pairing extinction training with VNS significantly reduced responding at the active lever during cue-primed reinstatement (one-way ANOVA with active and inactive levers in sham- and VNS-treated rats; F(3,38) = 13.16; p < 0.0001; Fig. 1C). Post hoc analysis with Tukey's multiple-comparison test showed a significant decrease in active lever presses in VNS-treated rats (p = 0.0019), but no differences in responses at the inactive lever (p = 0.74).
VNS facilitates extinction from cocaine-seeking and reduces cue-induced reinstatement. Rats were trained to self-administer cocaine and then underwent extinction training for 10 d while receiving either VNS (circles) or sham stimulation (squares) before drug-seeking was reinstated by presentation of the previously drug-paired cues during a reinstatement session (R). A, Responses of VNS- or sham-stimulated rats at the active (solid symbols) or inactive (open symbols) lever during cocaine self-administration (last 10 d are shown), extinction, and cue-induced reinstatement. VNS was delivered during the 2 h extinction sessions on Days 11–20. B, Two-way ANOVAs over the extinction period show that rats which received VNS showed a reduction in both active and inactive lever presses during the first day of extinction. C, A one-way ANOVA for the responses during the reinstatement session shows that VNS-treated rats also respond less at the active lever during cued reinstatement (see text for details). P values are (*) <0.05, (**) <0.01, and (****) <0.0001.
Experiment 2: systemic blockade of TrkB receptors alters VNS’ effects on cue-induced reinstatement
VNS alters the levels of several neuromodulators in the CNS, including BDNF (Dorr and Debonnel, 2006; Roosevelt et al., 2006; Follesa et al., 2007; Nichols et al., 2011; Manta et al., 2013). In order to determine if VNS-induced BDNF release modulates extinction learning and cue-induced reinstatement, we gave VNS-treated or sham-stimulated rats systemic intraperitoneal injections of the TrkB receptor antagonist ANA-12 (0.5 mg/kg in DMSO or DMSO vehicle at a volume of 0.5 ml/kg) during the 10 d of extinction learning (but not during the reinstatement session) and examined their drug-seeking behavior (Fig. 2). The combination of VNS or sham stimulation and ANA-12 or vehicle infusion, respectively, resulted in four treatment groups (sham + vehicle, n = 10; sham + ANA12, n = 11; VNS + vehicle, n = 13; VNS + ANA-12, n = 9). A two-way ANOVA with the factors group and time showed no main effect of either factor and no interaction between the factors on active lever presses during the last 10 d of self-administration (group, F(3,39) = 0.195, p = 0.899; time, F(9,351) = 0.775, p = 0.640; interaction, F(27,351) = 0.805, p = 0.747), indicating that self-administration rates did not differ between groups prior to the start of extinction. For responses at the active lever during the extinction period, a three-way ANOVA with the factors stimulation (sham or VNS), drug (vehicle or ANA-12), and time revealed main effects of stimulation (F(1,39) = 6.91; p = 0.012) and time (F(9,351) = 37.72; p < 0.0001), but not drug (F(1,39) = 0.057; p = 0.81), as well as interactions between stimulation and drug (F(1,39) = 6.35; p = 0.016) and between all three factors (F(9,351) = 6.15; p < 0.0001). Post hoc analysis showed that on the first day of extinction, sham + vehicle-treated rats responded significantly more on the previously active lever than VNS + vehicle- (p < 0.0001), sham + ANA-12- (p < 0.0001), or VNS + ANA-12-treated rats (p = 0.025; Fig. 2B). A separate two-way ANOVA with the factors stimulation and drug was performed on the active lever responses during the cue-induced reinstatement session indicating a significant effect of stimulation (F(1,39) = 4.99; p = 0.031) and an interaction of the factors (F(1,39) = 13.55; p = 0.0007), but no effect of the drug (F(1,39) = 2.44; p = 0.126). Post hoc analysis with Tukey's multiple-comparison test showed that VNS vehicle rats performed significantly fewer active lever presses than rats in all other groups (sham + vehicle, p = 0.0006; sham + ANA-12, p = 0.034; VNS + ANA-12, p = 0.0034; Fig. 2C). Taken together, these results show that TrkB receptor blockade alters the effects of VNS reinstatement, causing VNS + ANA-12-treated rats to respond like sham-stimulated rats.
The effects of VNS on cue-induced reinstatement depend on the activation of TrkB receptors. A, Active lever presses of VNS- or sham-stimulated rats that received intraperitoneal infusions of either vehicle solution (sham, black squares; VNS, red circles) or the TrkB antagonist ANA-12 (sham, teal triangles; VNS, olive triangles) during extinction training on Days 11–20. Shown are the last 10 d of cocaine self-administration, extinction training, and cue-induced reinstatement (R). B, Results of a three-way ANOVA show that rats which received VNS and vehicle (red bar) showed the largest reduction in active lever presses during the first day of extinction. Rats in the sham + ANA-12 and VNS + ANA-12 groups also responded less than the sham vehicle-treated rats. C, Results of a separate two-way ANOVA on the responses during the reinstatement session show that VNS-treated rats also show fewer active lever presses during cue-induced reinstatement. Blocking TrkB receptors with ANA-12 abolished this effect, resulting in drug-seeking behavior similar to that seen in sham vehicle-treated rats. P values are (*) <0.05, (**) <0.01, (***) <0.001, and (****) <0.0001.
Experiment 3: VNS during extinction training leads to elevated BDNF levels in the mPFC
Infusion of BDNF in the mPFC following cocaine self-administration reduces relapse to cocaine-seeking after abstinence and reinstatement following extinction (Berglind et al., 2007; Whitfield et al., 2011; Go et al., 2016; Barry and McGinty, 2017). VNS can increase BDNF levels in the brain (Follesa et al., 2007; Olsen et al., 2022), and thus we sought to determine if our conditions of VNS also lead to significant changes in BDNF levels that could explain its effects on extinction (Fig. 3). Rats self-administered cocaine followed by 1 d of extinction training paired with sham stimulation (n = 11) or VNS (n = 12) (Fig. 3A). An unpaired t test showed VNS-treated animals responded significantly less on the previously active lever during the extinction session (t(21) = 3.583; p = 0.0018; Fig. 3B). Immediately following the extinction session, rats were killed for quantification of BDNF levels in mPFC via ELISA. Assays were performed in two separate cohorts, and therefore measures are expressed as the percent change in BDNF levels compared with the average of the sham brains within each cohort. An unpaired t test showed VNS-treated rats had significantly higher levels of BDNF in the mPFC compared with sham rats (t(21) = 2.452; p = 0.023; Fig. 3C).
VNS increases the levels of BDNF in the mPFC. A, Active lever presses during the last 10 d of cocaine self-administration and a single day of extinction training during which rats received either sham stimulation (black squares) or VNS (red circles). B, VNS reduced active lever presses during the extinction session. C, Pairing extinction training with VNS increased BDNF content in mPFC tissue punches as measured by ELISA. Results are expressed as percent change from the average of the sham-stimulated group across two different cohorts (see text for detail). Unpaired t tests in B and C. P values are (*) <0.05 and (**) <0.01.
Experiment 4: cocaine self-administration leads to alterations in the AMPAR:NMDAR current ratio in the mPFC
Cocaine alters glutamatergic transmission in the mPFC (Kasanetz et al., 2013; Otis and Mueller, 2017; Caffino et al., 2018; Sasase et al., 2019). Alterations in the AMPAR:NMDAR current ratio provide sensitive assays for changes in postsynaptic receptor function. We first examined the effects of cocaine self-administration on glutamate receptors in the IL. Figure 4A shows active lever presses in a cohort of rats self-administering cocaine or receiving yoked-saline infusions. Figure 4B shows representative voltage-clamp recordings of AMPA and NMDA currents from IL Layer 5 pyramidal neurons from cocaine-administering or yoked-saline rats, killed immediately after their last self-administration session, as well as a group of drug-naive rats for comparison. A one-way ANOVA found a significant group effect in AMPAR:NMDAR ratios (F(2,24) = 4.811; p = 0.017; Fig. 4C), and post hoc testing with Tukey's multiple-comparison test indicated that AMPAR:NMDAR ratios in cocaine self-administering rats were significantly smaller than in both naive rats (p = 0.03) and yoked-saline rats (p = 0.044). Ratios of AMPA and NMDA currents were similar in yoked-saline and naive rats (p = 0.966).
Cocaine self-administration alters the ratio of currents through AMPA and NMDA receptors. A, Active lever presses in rats self-administering cocaine (blue triangles) and their yoked-saline controls (diamonds) during the last 10 d of self-administration. Brain slices for whole-cell patch–clamp recordings were prepared immediately after the last self-administration session (Day 10). B, Representative currents through AMPA- (black lines) and NMDA receptors (red lines) from Layer 5 pyramidal cells in the IL of naive rats or rats that received yoked-saline or cocaine infusions as illustrated in A. C, Results of a one-way ANOVA show that cocaine self-administration lowers the AMPAR:NMDAR ratio in IL pyramidal neurons compared with cells from naive or yoked-saline–treated rats. P values are (*) <0.05.
Experiment 5: pairing extinction with VNS reverses drug-induced changes in AMPAR:NMDAR ratios and is blocked by ANA-12
The effects of BDNF infusions into the mPFC on reinstatement are mediated by glutamatergic mechanisms (Berglind et al., 2007, 2009; Whitfield et al., 2011; Go et al., 2016; Barry and McGinty, 2017). Therefore, we next determined whether pairing extinction with VNS (1) can reverse cocaine-induced alterations in glutamatergic signaling and (2) whether TrkB receptor signaling is required for this effect. We obtained whole-cell recordings of AMPA and NMDA currents in IL Layer 5 pyramidal neurons immediately following the reinstatement session in a subset of rats that received VNS or sham stimulation with ANA-12 or vehicle during extinction in Experiment 2 (Fig. 5A). A two-way ANOVA with the factors stimulation (sham or VNS) and drug (vehicle or ANA-12) found significant main effects for both factors and an interaction of the effects on AMPAR:NMDAR ratios (stimulation, F(1,26) = 4.689, p = 0.040; drug, F(1,26) = 5.718, p = 0.024; interaction, F(1,26) = 22.14, p < 0.0001). Post hoc testing with Tukey's multiple-comparison test showed that AMPAR:NMDAR ratios in recordings from VNS vehicle rats were significantly larger than in recordings from sham + vehicle rats (p = 0.0001), sham + ANA-12 rats (p = 0.022), or VNS + ANA-12 rats (p = 0.0002). AMPAR:NMDAR ratios in sham vehicle-treated rats were not different from those in sham + ANA-12 rats (p = 0.32) or rats in the VNS + ANA-12 group (p > 0.998). Similarly, sham + ANA-12 rats and VNS + ANA-12 rats showed comparable AMPAR:NMDAR ratios (p = 0.41; Fig. 5B) These changes in synaptic plasticity were mostly driven by changes in AMPA currents: a two-way ANOVA comparing raw AMPA currents found a significant main effect of stimulation (F(1,26) = 7.151; p < 0.0128) and a drug × stimulation interaction (F(1,26) = 6.578; p = 0.016), but no main effect of the drug (F(1,26) = 2.433; p = 0.13). Post hoc analysis with Tukey's multiple-comparison test indicated that AMPA currents in VNS + vehicle-treated rats were significantly larger than in sham + vehicle rats (p = 0.0036), sham + ANA-12 rats (p = 0.037), and VNS + ANA-12 rats (p = 0.026; Fig. 5C). A similar comparison for NMDA currents in the same cells found no difference for either factor (two-way ANOVA; stimulation, F(1,26) = 1.272, p = 0.2693; drug, F(1,26) = 0.0000349, p = 0.995; interaction, F(1,26) = 0.131, p = 0.720; Fig. 5D). Taken together, our results suggest that pairing extinction with VNS reverses changes in AMPAR:NMDAR ratios in IL Layer 5 pyramidal neurons caused by cocaine self-administration and that this reversal is abolished by TrkB receptor blockade.
VNS reverses cocaine-induced changes in glutamatergic plasticity, which is dependent on TrkB receptor activation. A, Representative currents through AMPA (black lines) and NMDA receptors (red lines) from Layer 5 pyramidal cells in the IL of rats that received intraperitoneal infusions of either vehicle solution or the TrkB antagonist ANA-12 during extinction training paired with VNS (VNS + vehicle, VNS + ANA-12) or sham stimulation (sham + vehicle, sham + ANA-12). B, A two-way ANOVA showed that IL pyramidal neurons from rats in the VNS vehicle group (red bar) have significant larger AMPAR:NMDAR ratios following cue-induced reinstatement than cells from sham vehicle- (gray bar), sham + ANA-12- (teal bar), or VNS + ANA-12-treated rats (olive bar). C,D, Two-way ANOVAs indicate that the shift in AMPAR:NMDAR ratios is driven by changes in AMPA currents, (C), while the amplitudes of NMDAR currents did not significantly differ between the four treatment groups, (D). P values are (*) <0.05, (**) <0.01, and (***) <0.001.
Discussion
VNS provides a means to modulate extinction circuits to improve treatment retention and outcome. Here we replicate previous findings demonstrating that VNS facilitates extinction learning and reduces cue-induced reinstatement of cocaine-seeking (Childs et al., 2017, 2019), and we determine the importance of BDNF in these effects. VNS increased BDNF levels in the mPFC, and the systemic blockade of TrkB receptors diminished VNS's impact on reinstatement. Cocaine self-administration altered glutamatergic synaptic plasticity in the IL, and pairing extinction with VNS reversed these effects in a TrkB-dependent manner.
Facilitation of extinction learning by VNS
Stimulation of ascending fibers of the vagus nerve leads to the release of several neuromodulators, including norepinephrine, acetylcholine, and BDNF in the CNS, causing widespread cortical and subcortical activation (Cao et al., 2016; Collins et al., 2021). The release of neuromodulators alters cognitive and motivational states, thereby influencing how contexts are perceived and remembered. Consequently, VNS-induced changes may act as a primer, modulating synaptic plasticity in response to specific inputs that occur during sensory stimulation (Borland et al., 2018) or learning and memory (Porter et al., 2012; Pena et al., 2014). First, we replicated previous findings (Childs et al., 2017, 2019) demonstrating that pairing extinction training with VNS enhances both the speed of extinction learning during the initial session and the reduction of cue-induced reinstatement, indicating improved consolidation of extinction memories. Previous work has shown that VNS does not affect behavior during the extinction training period by interrupting ongoing behavior or acting as punishment (Childs et al., 2017). Additionally, the stimulation of the left vagus nerve does not induce conditioned place preference or avoidance, suggesting that it is neither rewarding nor aversive on its own (Childs et al., 2019; Noble et al., 2019a).
The importance of BDNF for the effects of VNS on extinction learning
Dysregulation of BDNF affects the formation and persistence of memories induced by psychostimulants, as well as the mechanisms underlying extinction learning. Prolonged cocaine exposure results in reduced BDNF levels across several brain regions, compromising neuroplasticity and stress resilience, both contributing factors to addiction (McGinty et al., 2010; Pitts et al., 2016; Verheij et al., 2016). The vagus nerve regulates constitutive BDNF expression in the brain (Amagase et al., 2023), and VNS has been shown to increase BDNF in various brain areas (Follesa et al., 2007; Olsen et al., 2022). To determine the contribution of BDNF to VNS effects on extinction and reinstatement, we used the TrkB receptor antagonist ANA-12. Systemic application of ANA-12 diminished VNS's ability to reduce cue-induced reinstatement. Surprisingly, rats in the sham + ANA-12 group also exhibited significantly reduced responding at the previously active lever during the first day of extinction, similar to VNS-treated rats. This effect may be attributed to the rapid anxiolytic-like effects of ANA-12 observed in mice (Cazorla et al., 2011), potentially counteracting cocaine withdrawal-induced anxiety or uncertainty resulting from changing contingencies. However, rats in the sham + ANA-12 group showed no reduction in responding during reinstatement, which suggests that the initial facilitation of extinction is insufficient to reduce reinstatement 11 d later. The disparate response rates during reinstatement also imply that the effect of VNS during prolonged extinction is not solely due to its potential anxiolytic effects (Noble et al., 2019b; Mathew et al., 2020) but rather reflects changes in extinction learning.
VNS modulation of BDNF levels in the mPFC
Because TrkB receptors are widely distributed throughout the brain (Yan et al., 1997), the changes in behavior that we observed following systemic ANA-12 application likely reflect the combined effects of TrkB antagonism across a number of brain areas involved in extinction learning and drug-seeking. In a first attempt to understand how VNS and BDNF may modulate extinction and reinstatement, we focused on the mPFC and specifically the IL, due to its importance in the acquisition and expression of extinction memories (Peters et al., 2009; Augur et al., 2016). Repeated exposure to cocaine and withdrawal decrease BDNF levels in the mPFC (Fumagalli et al., 2007, 2013; McGinty et al., 2010). Extinction by itself leads to a small increase in BDNF in the PFC (Li et al., 2018; Hastings et al., 2020). Our ELISA data show that pairing extinction with VNS elevates BDNF levels in the mPFC significantly beyond those induced by extinction alone. Thus, our data suggest that pairing extinction with VNS increases the release of BDNF in the mPFC and that this release is crucial for the modulation of both extinction and reinstatement behavior. These results mimic previous findings showing that a single infusion of BDNF into the prelimbic cortex (PL) post-cocaine self-administration session reduces relapse to cocaine-seeking after abstinence, as well as cue- and cocaine prime-induced reinstatement following extinction (Berglind et al., 2007; Whitfield et al., 2011; Barry and McGinty, 2017), while infralimbic BDNF can facilitate the extinction of cocaine-seeking (Otis et al., 2014; Yousuf et al., 2019).
Cocaine and VNS-induced alterations of synaptic plasticity in the IL
Cocaine induces aberrant plasticity at glutamatergic synapses in the mPFC, leading to altered salience attribution and cue-induced drug-seeking (Kasanetz et al., 2013; Otis and Mueller, 2017; Caffino et al., 2018). The direction and magnitude of changes depend on the brain region examined and the timepoint of testing (acute vs prolonged withdrawal, extinction, or after reinstatement). Cocaine self-administration decreases NMDA receptor expression and GluN2B dephosphorylation in the mPFC during withdrawal (Ben-Shahar et al., 2007; Sun et al., 2013). However, previous work in the PL has also shown increases in AMPA receptor function following cocaine regimen that lead to locomotor sensitization (Ruan and Yao, 2021) or conditioned place preference (Otis and Mueller, 2017). Here we observe reduced AMPAR:NMDAR ratios in IL pyramidal neurons during acute cocaine withdrawal. Similarly, sham-stimulated rats showed reduced AMPAR:NMDAR ratios after relapse, which were inversely correlated with the rate of responding at the active lever during the cue-induced reinstatement session. These shifts in synaptic plasticity were due to smaller AMPA currents in sham-stimulated rats. Taken together, these findings suggest that cocaine self-administration and reinstatement alter synaptic plasticity in the IL by reducing AMPA currents. In contrast, in rats that received VNS during extinction training, the AMPAR:NMDAR current ratios following cued reinstatement closely resembled those of naive or yoked-saline rats. Therefore, VNS modulates glutamatergic plasticity in the IL to reduce reinstatement, presumably by strengthening the effects that occur during extinction learning, and these results depend on stimulation of TrkB receptors. Inhibiting TrkB receptors, ERK/MAP-kinase activation, or NMDA receptors blocks the effects of BDNF infusions into the PFC on cocaine-seeking (Berglind et al., 2007; Whitfield et al., 2011; Go et al., 2016; Barry and McGinty, 2017), indicating an important interaction between TrkB signaling and NMDARs in BDNF's suppressive effects on drug-seeking (Otis et al., 2014; Otis and Mueller, 2017). However, BDNF can also mediate activity-dependent changes in synaptic strength by regulating AMPAR trafficking (Li and Wolf, 2011, 2015; Reimers et al., 2014; Fukumoto et al., 2020). In our experiments, the TrkB receptor blockade during extinction training primarily affected the VNS modulation of AMPA currents, resulting in reduced AMPAR:NMDAR ratios in the ANA-12-treated groups during reinstatement that paralleled those in the sham-treated group.
Additional limitations and considerations
Interpretation of our results is limited by our exclusion of females. There are known sex differences regarding both drug-seeking and drug-taking. Additionally, BDNF levels differ between sexes throughout development (Becker and Chartoff, 2019) and following cocaine self-administration (Towers et al., 2022), suggesting that VNS-induced elevations in BDNF could affect males and females differently.
VNS influences various neuromodulators involved in regulating reward-seeking behavior and extinction, offering multiple avenues to curb drug-seeking. Recent studies have shown that vagal sensory neurons play a crucial role in a circuit connecting gut nutrient information to nigrostriatal dopamine release (Han et al., 2018; McDougle et al., 2024), providing a means for natural reinforcers to regulate the reward system. However, current evidence suggests that only the stimulation of the right vagus nerve leads to a significant dopamine release and self-stimulation (Han et al., 2018; Brougher et al., 2021; McDougle et al., 2024).
Finally, our data do not address how often VNS-induced BDNF release must be paired with extinction learning to effectively diminish reinstatement, nor do they indicate the optimal timing for VNS delivery. Evidence from sensory and motor learning suggests that the temporally precise modulation of networks engaged in task performance significantly enhances VNS efficacy (Hays et al., 2014; Borland et al., 2018; Ruiz et al., 2023). However, in the context of extinction of drug self-administration, both contingent VNS (paired with nonreinforced lever presses) and noncontingent VNS (administered at fixed intervals) effectively reduce reinstatement (Childs et al., 2017). This discrepancy may reflect the complex nature of drug self-administration extinction, involving memories of both instrumental responses and drug-associated contexts, each with distinct consolidation trajectories. Further experiments altering VNS delivery during extinction or abstinence are warranted to address these questions.
Conclusion
The extinction of reactivity to drug-associated environments and the instrumental response for drugs are crucial for substance use disorder treatment (Childress et al., 1993). The progress made during rehabilitation is often undermined when patients return to environments formerly associated with drug-taking (Hammond and Wagner, 2013). Our data further support the notion that VNS can promote synaptic plasticity to enhance extinction from drug-seeking and reduce relapse, thereby increasing the efficacy of behavioral therapies where pharmacological approaches have yielded mixed results (Kantak and Nic Dhonnchadha, 2011; Nic Dhonnchadha and Kantak, 2011). We provide evidence that, at least in the mPFC, these changes are mediated by the activation of TrkB receptors, which may further aid the development of adjunct treatments.
Footnotes
This work was supported by National Institute on Drug Abuse (NIDA) Grants 1R01 DA055008-01 to S.K. and DA 049711 to J.F.M.
The authors declare no competing financial interests.
- Correspondence should be addressed to Sven Kroener at kroener{at}utdallas.edu.
