Reversal of morphine-induced cell-type–specific synaptic plasticity in the nucleus accumbens shell blocks reinstatement

Significance

The recent rise in opioid addiction has made development of new treatments a public health priority. The effort has been impeded by a distinct lack of understanding how opioid-induced alterations in synaptic transmission and cellular plasticity within reward brain regions, such as the nucleus accumbens (NAc), drive addiction behavior. We examined whether repeated morphine induces differential alterations in synaptic strength and transmission in subpopulations of NAc neurons, those expressing dopamine D1 or D2 receptors, that play opposing roles in addiction behavior. Morphine enhanced synaptic strength and transmission at D1 medium spiny neuron (MSN) synapses and reduced signaling in D2-MSN. Reversal of this plasticity with in vivo optogenetics or the antibiotic ceftriaxone disrupted the rewarding properties of morphine, providing a targetable molecular mechanism for future pharmacotherapies.

Abstract

Drug-evoked plasticity at excitatory synapses on medium spiny neurons (MSNs) of the nucleus accumbens (NAc) drives behavioral adaptations in addiction. MSNs expressing dopamine D1 (D1R-MSN) vs. D2 receptors (D2R-MSN) can exert antagonistic effects in drug-related behaviors, and display distinct alterations in glutamate signaling following repeated exposure to psychostimulants; however, little is known of cell-type–specific plasticity induced by opiates. Here, we find that repeated morphine potentiates excitatory transmission and increases GluA2-lacking AMPA receptor expression in D1R-MSNs, while reducing signaling in D2-MSNs following 10–14 d of forced abstinence. In vivo reversal of this pathophysiology with optogenetic stimulation of infralimbic cortex-accumbens shell (ILC-NAc shell) inputs or treatment with the antibiotic, ceftriaxone, blocked reinstatement of morphine-evoked conditioned place preference. These findings confirm the presence of overlapping and distinct plasticity produced by classes of abused drugs within subpopulations of MSNs that may provide targetable molecular mechanisms for future pharmacotherapies.

Opioid-based drugs are mainstays for pain management (1). However, side effects such as euphoria and the development of tolerance and dependence contribute to an increasing diversion of these readily available compounds for nontherapeutic use (2). Opioid agonist-based treatments are known to reduce some aspects of opioid addiction. On the other hand, these therapies often lead to high relapse rates when discontinued because they fail to eliminate key aspects of addiction such as conditioned associations that can trigger intense drug craving (2). Currently, development of alternative treatments for opioid addiction is hampered by a distinct lack of knowledge of the cellular plasticity that underlies persistent opioid-induced changes in behavior.

The nucleus accumbens (NAc) region of the ventral striatum is involved in attribution of salience to drug-paired cues that can in turn motivate reward-related behavior (3, 4). Medium spiny neurons (MSNs), the principal cells of the NAc, are GABAergic projection neurons that receive coordinated glutamatergic afferents arising from several cortical and limbic brain regions (5, 6). MSNs are divided into two subpopulations based on expression of the dopamine receptor 1 (D1R-MSN) or dopamine receptor 2 (D2-MSN), with a small fraction (∼6–17%) expressing both receptors (7). Importantly, these subpopulations have divergent projection targets and exert antagonistic effects in reward-related behaviors (8).

Long-lasting alterations in excitatory synaptic strength and glutamate release at MSNs produced by repeated exposure to drugs of abuse is a driving factor behind drug seeking and relapse (9–11). Numerous studies have examined effects of repeated psychostimulant exposure on synaptic strength and AMPA receptor (AMPAR)-mediated transmission in MSN subpopulations, with a majority of adaptations observed in D1R-MSN (12–14). Although opiate-induced changes in extracellular glutamate and glutamate receptor subunit expression have been reported, very little is known about opiate-induced changes in the efficacy of synaptic strength and function in the NAc, the degree to which this plasticity is cell-type specific, and the potential role for this plasticity in opiate-induced changes in behavior (15). To address these questions, we measured effects of repeated morphine on glutamatergic synaptic transmission in the NAc MSN subpopulations and used optogenetic and pharmacological approaches to determine the role of this pathophysiology in reward-seeking behavior.

Results

Anatomic and Cell-Type Specificity of Morphine-Induced Adaptations in NAc MSNs.

To assess long-lasting effects of repeated morphine on glutamatergic synaptic transmission in NAc shell MSNs, we used BAC transgenic mice expressing tdTomato or enhanced green fluorescent protein (eGFP) in D1R- and D2R-MSNs, respectively. Electrophysiological recordings were primarily performed in the dorsal aspects of the rostral shell, a region considered to be a hedonic “hot spot” for opioid reward (15). D1R-MSNs were identified using a crossover strategy in which we recorded from red cells in drd1a-tdtomato mice and nongreen cells in drd2a-eGFP mice (and vice versa for D2R-MSNs). This approach yielded very similar results; thus, data were pooled. Mice received five daily injections of saline or morphine (10 mg/kg), as this dose produced a maximal increase in AMPAR/NMDA receptor (NMDAR) (A/N) ratios (Fig. S1B) and transmitter release probability (Fig. S1C) in wild-type mice following 10–14 d of forced abstinence. Baseline and morphine-induced motor activity did not differ across wild-type and transgenic mice (Fig. S2). Morphine increased A/N ratios in NAc shell D1R-MSNs (Fig. 1B) but not D2R-MSNs (Fig. 1C). To test for morphine-induced synaptic AMPAR-specific plasticity, analysis of miniature excitatory postsynaptic current (mEPSCs) showed that amplitude and frequency of these events were elevated in D1R-MSNs compared with saline controls (Fig. 1D), whereas frequency, but not amplitude, was reduced in D2R-MSNs (Fig. 1E). Consistent with previous work in the NAc core (16), D2R-MSNs exhibited enhanced mEPSC frequency under basal conditions (saline) compared with D1R-MSNs [D1R-MSN, 4.8 ± 0.54 Hz, n_(cells) = 11, N_(mice) = 6; D2R-MSN, 6.6 ± 0.42 Hz, n = 12, N = 6; t₍₂₁₎ = −2.501, P = 0.02]. These results indicate that morphine-induced increases in synaptic strength reflect, in large part, an up-regulation of postsynaptic AMPAR-signaling.

Fig. S1.

Morphine dose–response in wild-type NAc shell MSN. (A) Experimental timeline (Left) including two acclimation days (H₁–H₂), 5 d of saline or morphine (10 or 20 mg/kg) injection, and 10–14 d of forced abstinence in the home cage. Electrophysiological recordings were performed in sagittal slices (Right) from wild-type mice containing NAc shell. (B) Ratio of AMPAR and NMDAR EPSCs (A/N ratios) at +40 mV in wild-type NAc shell MSN following 10–14 d of abstinence from saline (black outlined), 10 (black filled), and 20 (gray filled) mg/kg morphine [gray; F_(2,22) = 5.863, P = 0.01; n = 6–9 per group, N = 5–6 per group]. (C) Ratio of paired-pulse EPSCs in wild-type mice at 20-, 50-, 100-, and 200-ms interstimulus intervals from saline, 10, or 20 mg/kg morphine-exposed mice. A main effect of treatment but not interstimulus interval was observed [F_(2,39) = 5.49, P = 0.019; n = 4–7 per group, N = 3–7 per group]. *P < 0.05 vs. Sal.

Fig. S2.

Basal and morphine-induced motor activity are not different across genotypes. (A) Initial quantification of total distance traveled by wild-type (wt, black), drd1a-tdtomato (red), and drd2a-egfp mice (green) during a 30-min habituation period before saline injection on day 1 of habituation [F_(2,92) = 2.551, P = 0.084; n = 13–65 per group]. (B) Total distance traveled by wild-type, drd1a-tdtomato, and drd2a-egfp mice during a 90-min period following injection of morphine (10 mg/kg; filled circles) or saline (outlined circles) on days 1, 3, and 5 of injections shows that morphine increases motor activity in all three genotypes similarly. Three-way ANOVA revealed a significant main effect of treatment [F_{(2, 318)} = 54.402, P < 0.001] but not genotype or injection day. ***P < 0.001 vs. Sal.

Fig. 1.

Morphine-induced changes in NAc shell MSN synaptic strength and AMPAR-signaling. (A) Experimental timeline (Left) including two acclimation days (H1–H2), 5 d of saline or morphine (10 mg/kg) injections, and 10–14 d of forced abstinence in the home cage. Electrophysiological recordings were performed in sagittal slices (Right) containing NAc shell D1-MSN (B and D) and D2-MSNs (C and E) following the 10- to 14-d abstinence period. (B and C) Representative AMPAR (red, green) and NMDAR (black) EPSCs at +40 mV (Left) and mean AMPAR/NMDAR (A/N) ratio values (Right) in D1R-MSN [B, t₍₁₄₎ = −5.072, P < 0.001, n = 8 per group, N = 6–7 per group] and D2R-MSN [C, t₍₁₄₎ = 1.618, P = 0.128, n = 8 per group, N = 5–7 per group] from saline (Sal, outlined) and morphine (Mor, filled) exposed mice. (Scale bar, 50 pA/50 ms.) (D and E) Representative miniature EPSCs (Upper) and mean mEPSC amplitude (Left) and frequency (Right) in D1R-MSN [D, amplitude: t₍₃₀₎ = −3.773, P < 0.001; frequency: t₍₃₀₎ = −3.384, P = 0.002; n = 12–20 per group, N = 6–10 per group] and D2R-MSN [amplitude: t₍₁₃₎ = 1.072, P = 0.303; frequency: t₍₁₃₎ = 2.721, P = 0.017; n = 7–8 per group, N = 4–5 per group] from saline and morphine exposed mice. (Scale bar, 20 pA/50 ms.) *P < 0.05, **P < 0.01, ***P < 0.001 vs. Sal.

Alterations of synaptic AMPAR subunits can contribute to the expression of long-term changes in synaptic strength. To test for morphine-induced changes in AMPAR subunit composition, we evaluated current–voltage relationships of evoked EPSCs. Morphine reduced the rectification index in D1R-MSNs (Fig. 2A), indicating an increased presence of synaptic GluA2-lacking AMPARs; however, no change was observed in D2R-MSNs (Fig. 2B). In addition to postsynaptic receptor adaptations, fluctuations in synaptic strength can be mediated presynaptically through changes in transmitter release. Given the morphine-induced changes in mEPSC frequency, paired-pulse (PPR) stimulation was used to assess changes in glutamate release probability. Morphine reduced ratios in D1R-MSNs (Fig. 2C) and increased ratios in D2R-MSNs (Fig. 2D), suggesting significant increases and decreases in release probability, respectively.

Fig. 2.

Morphine-induced changes NAc shell MSN AMPAR subunit composition and neurotransmitter release. (A and B) Examples of normalized EPSC amplitudes at +40, 0, and −80 mV (Upper), with mean I–V plots (Lower) and rectification index (RI) at +40 mV (Inset) in D1R-MSNs [A, t₍₁₁₎ = 3.460, P = 0.005; n = 6–7 per group, N = 5 per group] and D2R-MSNs [B, t₍₈₎ = −0.579, P = 0.578; n = 4–6 per group, N = 4–6 per group] from saline- and morphine-exposed mice. (Scale bar, 50 pA/25 ms.) (C and D) Representative paired-pulse evoked EPSCs (50 ms; Upper) and mean ratios at 20- to 200-ms interstimulus intervals (Lower) in D1R- (C) and D2R-MSNs (D). A significant main effect of treatment was observed in both D1R-MSNs [F_(1,47) = 15.64, P = 0.003; n = 5–6 per group, N = 5 per group] and D2R-MSNs [D, F_(1,55) = 5.89, P = 0.032; n = 6–8 per group, N = 5–8 per group]. (Scale bar, 60 pA/25 ms.) *P < 0.05, **P < 0.01 vs. Sal.

To examine potential region-specific differences in morphine-induced synaptic plasticity, we measured A/N ratios, mEPSCs, and PPR in D1R- and D2R-MSNs of the NAc core. No significant differences were found in A/N in D1R- or D2R-MSNs (Fig. S3 B and E); however, mEPSC frequency, but not amplitude, was selectively reduced in D2R-MSNs (Fig. S3 C and F), the latter of which coincided with a trend toward reduced transmitter release selectively within D2R-MSNs (Fig. S3G). Because morphine’s effects on NAc synaptic function were more robust and diverse in the NAc shell, we focused our analysis and manipulation in subsequent experiments on this subdivision.

Fig. S3.

Morphine-induced plasticity in NAc core D1R- and D2R-MSN. (A) Experimental timeline (Left). Electrophysiological recordings were performed in D1R- and D2R-MSN within NAc core-containing sagittal slices (Right). (B and E) Ratio of AMPAR and NMDAR EPSCs at +40 mV in NAc core D1R-MSN [B, t₍₁₂₎ = 0.571, P = 0.579; n = 5–9 per group, N = 3–5 per group] and D2R-MSN [E, t₍₆₎ = 0.571, P = 0.787; n = 4 per group, N = 3 per group] following 10–14 d of abstinence from saline (outlined) or morphine (filled). (C and F) mEPSC amplitude (Left) and frequency (Right) in D1R-MSN [C, amplitude: t₍₁₆₎ = −0.648, P = 0.526; frequency: t₍₁₆₎ = 0.336, P = 0.741; n = 8–10 per group, N = 6 per group] and D2R-MSN [F, amplitude: t₍₁₇₎ = −0.395, P = 0.698; frequency: t₍₁₇₎ = 3.399, P = 0.003; n = 8–11, N = 6–7] from saline- or morphine-exposed mice. (D and G) Ratio of paired-pulse EPSCs at 20- to 200-ms interstimulus intervals in D1R- (D) and D2R-MSN (G). A main effect of interstimulus interval was observed in both D1R-MSN [F_(3,33) = 40.52, P < 0.001; n = 4–5 per group, N = 3 per group] and D2R-MSN [F_(3,21) = 46.14, P < 0.001; n = 4–5 per group, N = 3 per group], with a main effect of treatment trending toward significance in D2R-MSN [F_{(1, 21)} = 2.767, P = 0.128]. **P < 0.01 vs. Sal.

Reversal of Morphine-Induced Plasticity by the Antibiotic Ceftriaxone.

Opioid self-administration down-regulates expression of NAc glial excitatory amino acid transporter 2 (GLT-1) expression, a transporter responsible for 90% of glutamate uptake, and this effect increases vulnerability to relapse. Conversely, compounds that increase GLT-1 expression, such as the antibiotic ceftriaxone, reduce cocaine and heroin seeking (17, 18). Although these latter effects have been attributed to normalization of extracellular glutamate levels following cocaine seeking, the effect of ceftriaxone on opioid-induced changes in NAc MSN synaptic strength and function remains unclear. To test the ability of ceftriaxone to normalize morphine-induced changes in excitatory signaling, we treated mice with saline or ceftriaxone for 7–10 d following five daily injections of saline or morphine exposure, and measured excitatory synaptic parameters in the NAc shell. Ceftriaxone restored morphine-induced increases in D1R-MSN A/N ratios (Fig. S4A) as well as mEPSC amplitude and frequency to control levels (Fig. 2A). Initial assessments of mEPSCs demonstrated that amplitude and frequency were not altered in D1R- or D2R-MSNs from saline mice receiving ceftriaxone vs. vehicle (Fig. 2 B and C); thus, data were pooled. Unexpectedly, ceftriaxone treatment also normalized morphine-induced reductions in D2R-MSN mEPSC frequency and potentiated mEPSC amplitude compared with vehicle-treated morphine mice and saline controls (Fig. 3B). Evaluation of paired-pulse EPSCs in D1R- and D2R-MSNs showed that glutamate release probability was restored to saline control levels in ceftriaxone-treated morphine mice, indicating that ceftriaxone normalizes morphine-induced alterations in glutamate release probability at both cell types (Fig. 2 C and D).

Fig. S4.

Effects of ceftriaxone on NAc shell MSN plasticity. (A) D1R-MSN AMPAR/NMDAR (A/N) ratios in saline mice treated with vehicle (Sal; red, outlined) and morphine mice treated with either vehicle (Mor; red, filled) or ceftriaxone (Cef; pink filled) during abstinence shows that ceftriaxone normalizes morphine-induced increases in synaptic strength [F_(2,23) = 8.66, P = 0.002; n = 6–10 per group, N = 6–8 per group]. **P < 0.01 vs. Sal; ^##P < 0.01 vs. Cef. (B) D1R-MSN mEPSC amplitude (Left) and frequency (Right) in NAc shell from saline mice receiving subsequent vehicle (Veh, red outlined) or ceftriaxone (Cef, black outlined) treatment shows no effects of ceftriaxone on AMPAR-mediated signaling [amplitude: t₍₁₃₎ = −1.058, P = 0.309; frequency: t₍₁₃₎ = 1.976, P = 0.070; n = 6–9 per group, N = 4–6 per group]. (C) D2R-MSN mEPSC amplitude and frequency in NAc shell D2R-MSNs from saline mice receiving subsequent vehicle or ceftriaxone treatment shows no effect of ceftriaxone on AMPAR-mediated signaling [amplitude: t₍₁₅₎ = −0.952, P = 0.356; n = 8–9 per group, N = 4–5 per group].

Fig. 3.

Ceftriaxone normalizes plasticity in NAc shell D1R- and D2R-MSNs. Electrophysiological recordings were performed in NAc shell D1R- and D2R-MSN 10–14 d following abstinence from five daily injections of saline or morphine (10 mg/kg), during which they received 7–10 daily injections of vehicle or ceftriaxone (400 mg/kg) beginning 72 h following the final drug exposure. (A and B) Mean mEPSC amplitude (Left) and frequency (Right) in D1R-MSN [A, amplitude: F_(2,32) = 11.368, P < 0.001; frequency: F_(2,32) = 11.747, P < 0.001; n = 10–15 per group, N = 5–9 per group] and D2R-MSN [B, amplitude: F_(2,36) = 10.325, P < 0.001); frequency: F_(2,36) = 6.997, P = 0.003; n = 8–17 per group, N = 6–9 per group] from saline mice treated with vehicle or ceftriaxone (Sal, data are pooled) and morphine mice treated with vehicle (Mor) or ceftriaxone (Cef). (C and D) Mean EPSC paired-pulse ratios in D1R- (Left) and D2R- (Right) MSNs at 20–200 ms interstimulus intervals from mice treated with saline + vehicle, morphine + vehicle, or morphine + ceftriaxone. A significant interaction of drug-treatment and time was observed in D1R-MSN [C, F_(6,91) = 2.881, P = 0.016; n = 6–8 per group, N = 4–6], whereas a significant effect of treatment [D, F_(2,63) = 6.037, P = 0.014; n = 6–7 per group, N = 3–4 per group] and time [F_(3,63) = 8.974, P < 0.001] was observed in D2R-MSNs. (E and F) I–V plots of mean normalized AMPAR-EPSCs at +40, 0, and −80 mV (E) and mean rectification indices at +40 mV (F) in D1R-MSN from mice treated with saline + vehicle, morphine + vehicle, or morphine + ceftriaxone [F_(2,18) = 6.058, P = 0.011; n = 5–8, N = 5–6]. (G) Time course of EPSCs (−80 mV) before and during perfusion of Naspm (100 µM). (Scale bar, 100 pA/25 ms.) (H) Relative amplitudes of individual EPSCs in D1R-MSN before (black outlined) and after Naspm (gray filled) bath application from mice treated with saline + vehicle, morphine + vehicle, and morphine + ceftriaxone [F_(2,15) = 6.641, P = 0.009; n = 4–7 per group, N = 4–6 per group]. *P < 0.05, **P < 0.01, ***P < 0.001 vs. Sal; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. Cef; ^^^P < 0.001 vs. Mor.

We next examined effects of ceftriaxone on the morphine-induced increase in synaptic GluA2-lacking AMPARs in D1R-MSNs. Both the morphine-induced inward rectification of synaptic AMPARs (Fig. 3 E and F) and the increased sensitivity of AMPAR EPSCs to bath application of 1-naphthylacetylsperimine (Naspm), a selective antagonist of GluA2-lacking AMPARs, were blocked (Fig. 3 F–H), indicating that ceftriaxone treatment during abstinence restores expression of synaptic GluA2-lacking AMPARs to control levels. To test whether normalizing plasticity following ceftriaxone treatment alters reward behavior, mice underwent morphine place preference conditioning (CPP) followed by vehicle or ceftriaxone treatment during extinction. In these experiments, all mice were conditioned with four daily injections of morphine (5 mg/kg). This treatment regimen produced robust place preference and increased AMPAR plasticity in D1R-MSNs following 10–14 d of abstinence only, similar to that previously observed with five daily injections of 10 mg/kg morphine (Fig. 4 A and B). Although ceftriaxone did not alter extinction of preference (Fig. S5), morphine-induced reinstatement of CPP was blocked in ceftriaxone-treated morphine animals compared with vehicle treated controls (Fig. 4D).

Fig. 4.

Ceftriaxone and reinstatement of morphine place preference. (A) Experimental time line of CPP to assess AMPAR-transmission following CPP. Electrophysiological recordings were performed following 10–14 d of abstinence from the final conditioning session. (B) Mean mEPSC amplitude (Left) and frequency (Right) in NAc shell D1R-MSN from mice conditioned with morphine (5 mg/kg × 4) or saline [amplitude: t₍₁₁₎ = −3.527, P = 0.005; frequency: t₍₁₁₎ = −3.132, P = 0.010; n = 8–9 per group, N = 6 per group]. (C) Experimental time line of place preference to assess ceftriaxone’s ability to disrupt reinstatement of place preference. (D) Preference scores during 20-min session during pretesting, preference testing, the final extinction test, and reinstatement testing in morphine conditioned mice treated with vehicle (Veh, outlined) or ceftriaxone (Cef, filled). A significant interaction of treatment and test day was observed [F_(6,79) = 4.126, P = 0.002; n = 6–8 per group]; **P < 0.01, ***P < 0.001 vs. Sal; ^@@@P < 0.001 vs. Mor/Cef/Sal; ⁺⁺⁺P < 0.001 vs. Mor/Cef/Mor.

Fig. S5.

Ceftriaxone does not significantly alter extinction of preference. Preference scores during 20 min test sessions across pretesting (Pretest), preference testing (Pref.), extinction (Extinct.), and reinstatement (Reinstat.). All mice were conditioned with morphine (5 mg/kg). Seventy-two hours following preference testing, mice began extinction. Mice received six daily injections of vehicle or ceftriaxone following extinction training/testing. A main effect of test day but not treatment was observed [F_(3,54) = 12.429, P < 0.001; n = 9–12 per group], indicating that ceftriaxone treatment did not significantly affect extinction of preference. ^^^P < 0.001 vs. Pref. (within group).

Optically Induced LTD of ILC-NAc Shell Synapses Blocks Reinstatement.

Ceftriaxone-dependent reversal of synaptic potentiation in D1R-MSNs and subsequent blockade of reinstatement identifies this plasticity as a potential mechanism underlying the expression of conditioned reward, a key component of opioid addiction. However, an essential step in determining the functional role of opiate plasticity is elucidating the selectivity of these adaptations for specific afferent populations. Pharmacological studies have implicated the infralimbic cortex-accumbens (ILC-NAc) shell pathway in the expression of opiate-induced reinstatement of drug seeking (19, 20). The extent to which this pathway is subject to morphine-induced plasticity, and whether reversal of this plasticity prior to morphine reexposure prevents reinstatement behavior, an important goal in addiction research, is unknown. Thus, to test whether morphine-evoked potentiation in D1R-MSNs occurs within ILC-NAc shell projections and whether this plasticity is causally involved in addiction behavior, we used a CPP model where mice were injected with a channelrhodopsin (ChR2)-expressing adeno-associated virus in the ILC and implanted with optical fibers into the NAc shell, to allow selective stimulation of this major excitatory input (Fig. 5). Following extinction, mice were stimulated with light or received no stimulation 4 h before reinstatement testing. In agreement with electrically evoked long-term depression (LTD) protocols (15), stimulation at 10 Hz for 10 min reduced optically evoked A/N ratios in D1-MSN from morphine-injected mice receiving saline during reinstatement testing vs. nonstimulated mice (Fig. 5D). Although morphine increased motor activity compared with saline controls, no effect of optical stimulation in either morphine or saline reinstated mice was observed (Fig. S6A). Furthermore, reductions in this synaptic strength correlated with blockade of morphine-induced reinstatement of place preference (Fig. 5E). Taken together, these data indicate that morphine potentiates synaptic strength at D1R-MSNs within the ILC-NAc shell pathway and that this plasticity is a central factor in driving reinstatement of reward behavior.

Fig. 5.

Optical stimulation of ILC-NAc shell afferents depotentiates D1R-MSN synaptic strength and blocks reinstatement of morphine place preference. (A) Schematic depicting intra-ILC infusion of channelrhodopsin-expressing adeno-associated viral infusion (ChR2-AAV), optical stimulation of ILC terminals in the NAc shell, and recording of optically evoked EPSCs following reinstatement. (B) Experimental timeline of CPP. All mice were conditioned with morphine. Following optical (or mock) stimulation, mice were reinstated with an injection of morphine or saline. Electrophysiology recordings were performed 30–45 min following testing in a subset of mice receiving saline injections during reinstatement. (C) Representative optic fiber track placement (yellow outline) targeting the NAc shell (Left) and ILC ChR2-AAV infusion (Right). (Inset) High-magnification image of drd1a-tomato transgene (red) and ChR2-AAV terminal expression in NAc shell. (Scale bars, 20 µm.) (D) Optically evoked EPSCs in D1-MSNs from mice receiving stimulation (black) or mock stimulation (gray) and reinstated with saline [t₍₆₎ = 3.858, P = 0.008; n = 4 per group, N = 4 per group]. (Scale bar, 50 pA/50 ms.) (E) Preference scores across test days in stimulated (gray) and nonstimulated (black) mice reinstated with saline (outlined) or morphine (filled). All groups displayed statistically significant preference and extinction across test days with no differences observed across groups within test day. A significant interaction of test day and treatment group was observed [F_(9,103) = 4.305, P < 0.001; n = 6–8]. **P < 0.01; ***P < 0.001 vs. Mor/No stim/Sal; ^###P < 0.001 vs. Mor/Stim/Sal; ^^^P < 0.001 vs. Mor/Stim/Mor. PrLC, prelimbic cortex; cc, corpus callosum.

Fig. S6.

Optical fiber histology and optic stimulation effects on motor activity. (A) Motor activity during a 20-min reinstatement test in saline-reinstated mice (outlined) and morphine-reinstated (filled) receiving no stimulation (No stim, black) or stimulation (Stim, gray). A significant main effect drug was observed [F_(1,19) = 36.55, P < 0.001; n = 6–8 per group]. (B) Schematic representing optic fiber placements. ***P < 0.001 vs. Sal-No Stim; ^###P < 0.001 vs. Sal-Stim.

Discussion

To our knowledge, our study is the first to identify region-specific bimodal changes in synaptic plasticity within discrete subpopulations of NAc MSNs that are induced by repeated morphine exposure and demonstrate a causal link between this plasticity and behavior that models a key aspect of opiate addiction. As data on experience-dependent synaptic plasticity induced by cocaine are much more abundant, they provide a useful basis for comparison. Similar to cocaine, repeated morphine augmented synaptic strength and AMPAR-mediated transmission exlcusively in D1R-MSNs of the NAc shell (14). On the other hand, morphine-dependent increases in mEPSC frequency and reductions in paired-pulse ratios suggest an increase in release probability at D1R-MSNs, contrasting with cocaine studies, which have directly observed increases in glutamate release probability in the NAc core, but not shell (21–24).

In addition to enhanced synaptic strength and transmission in D1R-MSNs, morphine also weakens excitatory input at D2R-MSNs, which has not been observed following cocaine exposure (25). Indeed, increased release probability has been reported at D1R-MSNs 24 h following remifentanil self-administration; however, no alterations in release probability were found at D2R-MSNs, indicating that, unlike D1R-MSNs, alterations in glutamate release at D2R-MSNs requires more prolonged abstinence (26). This divergent regulation of synaptic transmission in D1R- vs. D2R-MSNs is congruent with their antagonistic roles in addiction behavior and the notion that MSN subpopulations receive differential afferent innervation (27). The exact mechanism behind cell-type–specific plasticity in the current study is unclear; however, increases in transmitter release at D1R-MSNs may reflect a reduction in presynaptic mu opioid receptor (MOR) inhibition (26). Alternatively, it is possible that reductions in transmitter release at D2R-MSNs may be due in part to endocannabinoid-mediated LTD (eCB-LTD), which selectively occurs in striatal D2R-MSNs (16).

Morphine Increases Expression of Ca²⁺-Permeable AMPARs.

In cocaine studies, increases in Ca²⁺-permeable, GluA2-lacking AMPARs within NAc MSNs have emerged as a key mediator of increased drug seeking and “incubation of craving” during prolonged (∼30–45 d) abstinence (28). Whether other classes of abused drugs produce this neuroadaptation is unknown. To our knowledge, our data provide the first evidence that repeated opiate exposure increases synaptic GluA2-lacking AMPARs in NAc, and that this adaptation occurs exclusively in D1R-MSNs. Interestingly, incubation of drug seeking appears to manifest more rapidly for opiates (within 14 d) (29) than for cocaine, consistent with the time frame for the appearance of GluA2-lacking AMPARs shown here. In addition, only a brief regimen of experimenter-administered morphine was sufficient for this plasticity, in contrast to the extended-access self-administration procedure needed for GluA2-lacking AMPAR accumulation by cocaine (28). Thus, although we do not yet know the mechanisms by which (i) drug exposure increases GluA2-lacking AMPARs and (ii) the synaptic incorporation of these receptors drives the incubation effect, use of opiate exposure models should facilitate the study of these important questions.

Morphine-Induced Synaptic Plasticity in the NAc Core.

The two major subdivisions of the NAc, the shell and core, are distinguished based on anatomic connectivity and their role in reward-related behavior (6, 30–33). In the present study, no significant adaptations in synaptic strength (A/N ratios) or glutamate release probability were observed in either D1R- or D2R-MSNs of the NAc core. Although repeated exposure to cocaine produces enduring increases in synaptic strength in both NAc shell and core, we recently demonstrated NAc synaptic plasticity following repeated amphetamine (22, 24, 32–34), which raises a question of whether cocaine, rather than morphine, may be unusual in regards to plasticity in the NAc core. That said, given that the behavioral and neurochemical effects of opiates are dependent on activation of mu opioid receptors (2, 35), the regional differences in morphine-induced plasticity demonstrated here may be related to the higher prevalence of mu opioid receptors in the NAc shell compared with the core (36, 37).

Although a specific role glutamate signaling in the shell vs. core in opiate seeking remains debatable, the shell-specific plasticity shown here is consistent with previous reports implicating a role for this region in opiate reward and relapse to drug seeking (19, 38–43). It is worth noting, however, that reductions in NAc core A/N ratios due to up-regulation of NR2B-containing NMDAR currents have been reported following heroin self-administration in rats, and that this adaptation drives reinstatement of heroin seeking (24). The mechanisms underlying this apparent discrepancy in plasticity are unclear; however, morphine and heroin have been shown to differentially modulate synaptic strength (44). Furthermore, distinctions in learning and motivational processes associated with operant responding tasks vs. CPP, as well as incorporation of extinction vs. abstinence following drug exposure may be contributing factors to these differences.

Reversal of NAc Morphine-Induced Plasticity Blocks Reinstatement.

Many studies on the neurobiology of drug relapse have focused on measuring glutamate release and cellular adaptations at excitatory synapses in the NAc (45). A common finding across drug classes is that reinstated drug seeking in animal models is associated with increased extracellular glutamate in the NAc (17). These increases in extrasynaptic glutamate have been ascribed in part, to reduced uptake of synaptically released glutamate via GLT-1 in the NAc, as compounds that increase expression of GLT-1 (i.e., ceftriaxone) can inhibit reinstatement of drug seeking (17). Although repeated cocaine reduces GLT-1 expression in the NAc core and shell, the antirelapse effects of ceftriaxone have been attributed to restoration of GLT-1 function selectively in the core (46). Similarly, heroin self-administration increases extrasynaptic glutamate spillover and impairs GLT-1 function in the NAc core (40). Reversal of this adaptation with ceftriaxone was associated with attenuated reinstatement of cue-induced heroin-seeking behavior. Although these findings support the notion that ceftriaxone normalizes reuptake in the NAc core, they do not address potential alterations in NAc shell glutamate or synaptic plasticity in either core or shell. Here, we find that repeated treatment with ceftriaxone during abstinence normalizes morphine-induced synaptic plasticity in D1R- and D2R-MSNs, providing, to our knowledge, the first evidence that ceftriaxone normalizes opiate-induced synaptic strength and transmission, and that these effects are distinctly different across MSN subpopulations in the NAc shell. In congruence with previous opiate extracellular glutamate work, ceftriaxone-dependent normalization of synaptic plasticity also blocks reinstatement reward-related behavior (41, 47); however, future studies are needed to determine whether this blockade is due to ceftriaxone’s effects on D1R- or D2R-MSN plasticity or a combination of the two.

The ability of ceftriaxone to normalize synaptic strength and transmitter release probability in D1R-MSNs indicated that morphine-induced potentiation in D1R-MSN synapses is a key factor for morphine reward-related behavior. However, these cells receive prominent glutamate input from multiple brain regions, and pathway-specific activation of these fibers has been demonstrated to elicit distinct behavioral responses (5, 48–50). Here we show that prior optogenetic stimulation of ILC-NAc shell terminals using a protocol known to produce LTD in NAc MSNs produces a depotentiation of morphine synaptic strength in D1R-MSNs and prevents morphine-induced reinstatement of place preference. Previous work has shown that reversible inactivation of the NAc shell during testing can block heroin-primed reinstatement (18). Furthermore, inactivation of the ILC in conjunction with D1R antagonism in the NAc shell during testing attenuates context-induced reinstatement of heroin seeking (19). These studies indicate a role for the ILC and NAc shell in the expression of drug-seeking behavior, but do not directly assess the role of the ILC-NAc shell pathway, as these brain regions retain numerous efferent and afferent connections, nor do they address any role for plasticity in this pathway. Thus, to our knowledge, these data are the first to directly demonstrate that morphine potentiates ILC-NAc shell synaptic strength, and that this pathway-specific plasticity is a primary mechanism by which opiates exert their rewarding effects. The possibility of antidromic activation of cell bodies within the ILC and subsequent engagement of collaterals projecting to other brain regions should be considered when interpreting these results (51). Importantly, these data are consistent with recent data suggesting that the ventromedial prefrontal cortex acts as a neural “off” switch for cocaine seeking, but an “on” switch for opiates (52). Future experiments will be necessary to determine whether plasticity at ILC-NAc shell synapses is important for reinstatement by other modalities (i.e., cues, context).

The need for developing nonopioid based treatments that target neural processes corrupted by chronic opioid use is clear. However, a better understanding of molecular and cellular mechanisms underlying effects of opioids on synaptic transmission and plasticity within reward-circuits is required. Our study identifies cell-type and pathway-specific alterations in synaptic strength and transmission within the NAc and implicates the NAc shell as a key site for morphine plasticity involved in the conditioned rewarding effects of morphine. Furthermore, our data highlight the complexities of NAc microcircuits, and confirm the presence of distinct and overlapping plasticity produced by classes of abused drugs that may provide targetable molecular mechanisms for future pharmacotherapies.

Materials and Methods

Mice.

C57BL/J6, drd1a-tdtomato, or drd2-eGFP male mice (P48-60) were used and have been described (refs. 14 and 21; SI Materials and Methods). All experiments were approved by the University of Minnesota Institutional Animal Care and Use Committee.

Locomotor Sensitization.

Initial studies involving repeated morphine and abstinence (Figs. 1–3 and Figs. S1–S4) included habituation, five once-daily injections of saline or morphine (10 mg/kg, i.p.), and 10–14 d of home cage abstinence. For ceftriaxone experiments (Fig. 2 A–G and Fig. S2 A–C), mice received 7–10 daily injections of vehicle or ceftriaxone (400 mg/kg, i.p) beginning 72 h following the final drug exposure (SI Materials and Methods).

Conditioned Place Preference.

All conditioned place preference experiments used a two-chamber apparatus, and included eight daily alternating sessions of saline or morphine (5 mg/kg, s.c.) and six daily extinction sessions as described (ref. 53; SI Materials and Methods).

Virus Injection of ChR2-AAV, Optic Fiber Implantation, and in Vivo Stimulation Protocols.

AAV2-CaMKII-ChR2-eYFP or AAV2-CaMKII-ChR2-mcherry (University of North Carolina Vector Core Facility) was bilaterally injected into the infralimbic cortex. Approximately 4–6 wk following virus surgeries, mice were implanted with indwelling optic fibers targeting the NAc shell. Blue LEDs (Plexon, 465 nm, 15 mW) were used to deliver a 10-min train of 5-ms pulses at 10 Hz in the home cage 4 h before receiving an injection of saline or morphine for testing reinstatement of place preference (SI Materials and Methods).

Slice Electrophysiology.

Whole-cell patch clamp recordings from MSNs in acute NAc sagittal slices (250 µm) were performed as described (22, 54).

Data Analysis.

All data are expressed as a mean ± SEM. Data obtained from individual cells is represented alongside bar graphs. Sample size in experiments is presented as n and N, where n is the number of cells and N is the number of mice. Statistical significance was assessed using Student’s t tests, one-way, two-way, and repeated measures ANOVAs (Sigma Plot) with further comparisons made using a Student–Newman–Keuls post hoc test. Traces in figures have had stimulus artifacts removed and are averages of 20–25 consecutive responses.

SI Materials and Methods

Mice.

Male mice (P48-60) were C57BL/J6 (wild type) or heterozygous bacterial artificial chromosome (BAC) transgenic mice (Jackson Laboratories), in which tdtomato or eGFP expression was driven by either DR1 (drd1a-tdtomato) or DR2 (drd2-eGFP) dopamine receptors. Mice were single and group housed in a temperature- and humidity-controlled environment with a 12-h light/12-h dark cycle with food and water available ad libitum.

Locomotor Sensitization.

Locomotor chamber apparatus was placed under a video tracking system (Any-maze, Stoelting) and measurements were made automatically by the software. Mice were habituated to experimenter handling and injections (i.p.) over a 2-d period, followed by five consecutive once-daily injections of saline or morphine. Initial studies characterizing morphine effects on plasticity involved administration of 10 or 20 mg/kg (Fig. S1 B and C), with a dose of 10 mg/kg used for all subsequent sensitization experiments, as it produced a maximal effect on AMPAR/NMDAR ratios. For all test days, animals were habituated to testing chambers for 30 min and motor activity was monitored for 90–180 min following drug or saline administration. Following the final day of drug treatment (saline, morphine), animals were returned to their home cage for 10–14 d. For experiments involving ceftriaxone (Fig. 2 A–G and Fig. S2 A–C), mice received 7–10 daily injections (i.p.) of either vehicle (saline) or ceftriaxone (400 mg/kg, i.p) beginning ∼72 h following the final drug treatment. Mice were killed for electrophysiological recordings within 10–14 d of the final drug treatment and within 24–48 h of the final abstinence treatment (vehicle, ceftriaxone).

Previous studies demonstrated that administration of ceftriaxone (200 mg/kg) in rats is sufficient to block drug-seeking behavior (24, 39). In the present study, a dose of 400 mg/kg was chosen using an allometric scaling calculator to estimate interspecies dosage scaling between rats and mice using exponential allometry.

Conditioned Place Preference.

Conditioned place preference experiments involving ceftriaxone (Fig. 4), spanned ∼20 d and were conducted in an apparatus (22 × 45 × 20 cm) divided into two compartments by wall (black/opaque) with a guillotine-style door. Experiments included a combination of wild-type and transgenic mice for electrophysiological recordings. The floor of one compartment consisted of an array of 2.3-mm stainless steel rods, centered every 6.5 mm (grid), and the other consisted of thin chicken wire with openings of 2.5 × 2.5 mm (mesh). All data were acquired by video camera and movements analyzed with ANY-maze software. Mice were habituated to experimenter by 3 d of handling. During preconditioning tests (Pretest; one per day over 2 d), mice were placed in the apparatus and allowed to freely explore for 20 min; time spent in the grid and mesh chambers was recorded, and averaged over the test sessions. Any mice displaying a marked preference (>800 s) for any one side were excluded from further testing (1 of 22 mice). Vehicle (CS−) or morphine (CS+) was paired with either the grid or mesh compartment using a counterbalanced approach based upon preconditioning test performance. During the conditioning trials, subjects were injected with either morphine or vehicle, and after a 20-min delay were confined for 30 min in the corresponding CS+/CS− chamber. A total of four morphine (5 mg/kg) and four saline trials were performed in alternating fashion, with only one trial performed per day. A dose of 5 mg/kg was chosen because it produced optimal preference and induction of plasticity. Side preferences were evaluated 24 h following the eighth (final) conditioning session. Seventy-two hours following preference testing, animals underwent six daily extinction sessions. On days 1, 3, and 5, animals were confined to the CS+ and CS− compartment for 20 min each (extinction training). On days 2, 4, and 6, animals were allowed free exploration of both compartments for 20 min, and movement was recorded to track extinction (extinction testing). Data obtained on extinction test 3 was used in two-way repeated measures ANOVA analysis. Twenty-four hours following the final extinction test, animals underwent reinstatement testing, with injection of saline or morphine 20 min before being placed into the chamber. Place preference was determined by calculating the difference in time spent in the CS+ and CS− chambers during the test sessions; “preference” is defined as CS+ − CS− within a given test session.

Place preference experiments involving in vivo optical stimulation used similar conditions for preconditioning testing, conditioning, extinction, and reinstatement. Testing chambers for these experiments were identical, with the exception of using an open-chamber design without guillotine-style doors to allow animals connected to fiber optic patch cables unobstructed passage between grid and mesh compartments.

Virus Stereotaxic Injection of ChR2-AAV.

AAV1-CaMKII-ChR2-eYFP or AAV1-CaMKII-ChR2-mcherry produced at the University of North Carolina (Vector Core Facility) was injected into medial prefrontal cortex (focusing on the infralimbic region) of 7- to 8-wk-old mice. Animals were anesthetized with a ketamine and xylazine mixture (100 and 10 mg/kg; i.p.) and placed into the stereotaxic frame (Kopf Instruments). Craniotomies were performed using stereotaxic coordinates adapted from a mouse brain atlas (anterior-posterior: +1.65; medial-lateral: ±0.4; dorsal-ventral: −3.2 from bregma). Injections of virus (0.5 µL per injection site) were performed with a 5-µL Hamilton syringe using an UltraMicroPump with SYS-Micro4 controller (World Precision Instruments). After a 5-min delay to reduce solution backflow along the infusion track, the syringe was slowly removed over a 5-min period and incisions were closed using Vetbond (3M).

Optic Fiber Cannulation and in Vivo Optogenetic Stimulation Protocols.

Approximately 4–6 wk following virus surgeries, animals underwent a second craniotomy for bilateral implantation of chronically indwelling optic fiber implants (200/230 µm core/cladding, 0.66 NA) targeting the NAc shell (angled: 14 degrees; anterior-posterior: +1.5 mm; medial-lateral: ±1.63; dorsal-ventral: −4.1) using stereotaxic apparatus described above. Light guides were secured to the skull with a dual-cure resin-lonomer (Geristore) that was anchored with two 0.0625-in steel machine screws (amazon.com) with caps (ThorLabs) placed over the top.

Blue LEDs (Plexon, 465 nm, 15 mW) were used to deliver a 10-min train of 5-ms pulses at 10 Hz in the home cage 4 h before receiving an injection of saline or morphine for testing reinstatement of place preference. All mice were infected with ChR2, with animals in the no stimulation group undergoing the same procedures as animals receiving optical stimulation; however, no stimulation was applied. During initial place preference experiments, testing of various stimulation time points involved stimulation immediately before reinstatement injection. In these experiments, mice from saline + stimulation (n = 3 of 8), morphine + no stimulation (n = 3 of 9), and morphine + stimulation (n = 2 of 6) underwent a subsequent extinction training and extinction test session and were then reinstated using a stimulation time point of 4 h before injection with morphine. Only data from the subsequent extinction and reinstatement session were included in the data set and were not significantly different from animals only undergoing a single simulation session 4 h before reinstatement. Post hoc histological examination of virus expression was performed for each animal. Animals displaying significant viral expression outside of the ILC (e.g., prelimbic or cingulate cortices) or unilateral expression were excluded from data analyses.

Slice Electrophysiology.

Sagittal slices (250 µm) containing the NAc shell were sliced in ice-cold kynurenic acid (5–7 mM) as described (3). Slices recovered for at least 30 min in ACSF solution saturated with 95% O₂/5% CO₂ containing 119 mM NaCl, 2.5 mM KCl, 1.0 mM NaH₂PO₄, 1.3 mM MgSO₄, 2.5 mM CaCl₂, 26.2 mM NaHCO₃, and 11 mM glucose. Cells were visualized in sagittal slices using infrared-differential contrast (IR-DIC) microscopy, and MSNs were identified by their morphology and typical hyperpolarized resting potential (−70 to −80 mV). When using BAC transgenic mice, cell types were identified by a crossover strategy, recording from red cells in drd1a-tdTomato mice and nongreen cells in drd2-eGFP mice (and vice versa for D2-MSNs), which yielded similar results, thus data were pooled. Using an Axon Instruments Multiclamp 700B (Molecular Devices), MSNs were voltage clamped at −80 mV with electrodes (2.5–3.5 MΩ) containing a cesium-gluconate based internal solution described (24). Series resistance was monitored continuously during all recordings, and a change beyond 20% resulted in exclusion of the cell from data analyses. Data were filtered at 2 kHz and digitized at 5 kHz via custom Igor Pro software (Wavemetrics). At the beginning of each sweep, a depolarizing step (4 mV, 100 ms) was generated by a Master-8 stimulator (A.M.P.I.) to monitor series (10–40 MΩ) and input resistance (>400 MΩ).

For all electrophysiological recordings, picrotoxin (100 µM) was added to block GABAergic neurotransmission. Holding potentials were corrected for liquid junction potential (∼8 mV). Electrically evoked AMPAR/NMDAR ratios (A/N) were computed from EPSCs at +40 mV with and without 50 µM d-aminophosphonovaleric acid (d-APV). Optically evoked EPSCs were obtained every 10 s with pulses of 473-nm wavelength. Pulses were generated by a Master-8 stimulator, and light was delivered via full-field illumination (submerged 40× objective) using a SOLA SE II 354 Light Engine (Lumencor). Pulse duration (0.7–2 ms) and light intensity were adjusted to obtain EPSC amplitudes of ∼150–250 pA across cells. For paired-pulse ratio experiments, EPSCs were measured at 20-, 50-, 100-, and 200-ms interstimulus intervals. The ratio at each interval was calculated by dividing the amplitude of the second EPSC by the amplitude of the first. For mEPSCs, lidocaine (0.7 mM) was bath applied to prevent action potentials. Measurement, data collection, and analysis of mEPSC were performed as described (24).

To examine AMPAR subunit composition, a current–voltage (I–V) curve was plotted. Evoked AMPAR-mediated EPSCs were measured at the membrane potentials of −80, −40, 0, +20, and +40 mV following correction of holding potentials for liquid junction potential (∼8 mV). The rectification index was calculated by comparing the peak AMPAR-mediated EPSC amplitude at +40 mV to −80 mV. For studies involving Naspm, a baseline of 20–25 sweeps was collected in the presence of ACSF while holding the cell’s voltage at −80 mV. After Naspm (100 µM) had been bath applied to the slice for 8–10 min to allow sufficient time for blockade of GluA2-lakcing AMPARs, 20–25 additional traces were collected to assess the degree of change in AMPAR EPSC peak amplitude. Only recordings with <20% variability in EPSC amplitude were used. For I–V and Naspm experiments, 0.1 mM spermine was included in the internal solution. Note that, whenever possible, the electrophysiologist was blinded (∼60% of the time) to the drug-treatment condition.

Drugs.

d-AP-5 and Naspm were purchased from Tocris Bioscience, whereas picrotoxin, lidocaine, and kynurenic acid were purchased from Sigma Aldrich. Morphine (Medisca) and ceftriaxone (Lupin) were obtained from Boynton Pharmacy (University of Minnesota, Minneapolis).