Dopaminergic regulation of nucleus accumbens cholinergic interneurons demarcates susceptibility to cocaine addiction

Abstract

Background:

Cholinergic interneurons (ChINs) in the nucleus accumbens (NAc) play critical roles in processing information related to reward. However, the contribution of ChINs to emergence of addiction-like behaviors and its underlying molecular mechanisms remain elusive.

Methods:

We employed cocaine self-administration to identify two mouse subpopulations: susceptible and resilient to cocaine seeking. We compared the subpopulations for physiological responses with single-unit recording of NAc ChINs, and for gene expression levels with RNA sequencing of ChINs sorted using fluorescence-activated cell sorting. To provide evidence for a causal relationship, we manipulated the expression level of dopamine D2 receptor (DRD2) of ChIN in a cell-type specific manner. Using optogenetic activation combined with a double whole-cell recording, the effect of ChIN-specific DRD2 manipulation on each synaptic inputs were assessed in NAc medium spiny neurons (MSNs) in a pathway-specific manner.

Results:

Susceptible mice showed higher levels of nosepoke responses under progressive ratio schedule, and impairment in extinction and punishment procedures. DRD2 was highly abundant in NAc ChIN of susceptible mice. Elevated abundance of DRD2 in NAc ChINs was sufficient and necessary to express high cocaine motivation, putatively through reduction of ChIN activity during cocaine exposure. DRD2 overexpression in ChINs mimicked cocaine-induced effects on the dendritic spine density and the ratios of excitatory inputs between two distinct MSN cell-types, while DRD2 depletion precluded cocaine-induced synaptic plasticity.

Conclusions:

These findings provide a molecular mechanism for dopaminergic control of NAc ChINs that can control the susceptibility to cocaine-seeking behavior.

Introduction

Drug addiction is a neuropsychiatric disorder characterized by compulsive and persistent drug seeking along with excessive motivation to consume substances. Epidemiologic studies revealed that only a small subset of the population exposed to cocaine develops addiction-related symptoms (1, 2). Due to the socioeconomic burdens of cocaine addiction, the behavioral and neurobiological changes after exposure to cocaine have been extensively studied in rodent models for the past several decades. However, the molecular profile of drug addiction and cellular mechanism underlying susceptibility to cocaine addiction remain largely unresolved in part due to the lack of reliable mouse models demonstrating individual differences.

The mesolimbic dopaminergic (DAergic) pathway, which includes the nucleus accumbens (NAc), is dysregulated in various states of drug addiction (3). The NAc predominately comprises medium spiny neurons (MSNs), which are divided into two subtypes based on expression of different DA receptor subtypes, DA D1 receptors (D1-MSNs) and DA D2 receptors (D2-MSNs). They produce two functionally distinct outputs of the NAc: D1-MSNs mainly form the direct pathway that promotes reward-related behaviors, while D2-MSNs form the indirect pathway that serves a critical role in aversion and negative reinforcement (4,5). In the NAc, both types of MSNs receive excitatory inputs from the basolateral amygdala (BLA), the ventral hippocampus (vHPC), and the medial prefrontal cortex (mPFC). All inputs are capable of inducing reinforcement (6), but each pathway is independently regulated after withdrawal from cocaine self-administration (7). However, the functional distinctions and regulatory mechanisms of these pathways for cocaine seeking remain elusive.

MSNs can be regulated by several types of striatal interneurons (8,9). A growing body of evidence suggests that cholinergic interneurons (ChINs) in the NAc might play critical roles in processing reward- and addiction-related information (10,11). Despite comprising a minute portion (1–2%) of the entire NAc neuronal population, ChIN axonal fields are widespread (12), leading to extensive modulation of a large number of NAc neurons including MSNs. In fact, cocaine exposure induces both ChIN activation and acetylcholine (ACh) release (13), and ablation of NAc ChINs results in an exaggerated locomotor response to non-contingent cocaine injections (14). Consistently, locomotor sensitization is prevented by an increase in NAc ACh levels (15). NAc ChINs also receive direct inputs from mesolimbic DAergic neurons, which activate ChINs primarily through glutamate co-release (16). Once activated by the firing of DAergic neurons, NAc ChIN firing is paused, which is in part mediated by activation of dopamine D2 receptor (DRD2) (17). Conversely, selective activation of NAc ChINs causes DA release through activation of nicotinic ACh receptors (nAChRs) expressed on the axonal terminals of DAergic neurons (18). Given these anatomical and physiological observations, modification of NAc ChINs would potentially control neuronal features and synaptic plasticity of MSNs, which could consequently contribute to addiction-like behaviors.

Here, we employed a behavioral paradigm to quantify motivation for cocaine after extended self-administration in mice, which allowed for a segregation of subject animals exhibiting susceptibility versus resilience to cocaine-seeking behaviors. We also conducted RNA-sequencing on isolated ChINs from the NAc of susceptible and resilient mice to obtain differential gene expression at the whole transcriptome level, which indicated that the downstream DRD2 signaling pathway was upregulated in susceptible animals. To examine the cellular mechanism and functional impact on cocaine-seeking, we exploited cell-type-specific manipulation of DRD2 and then monitored physiological and behavioral changes. Taken together, our findings indicate that DRD2 abundance in NAc ChINs is both necessary and sufficient for development of susceptibility to cocaine seeking, which is mediated by a pathway-specific modulation of structural and synaptic plasticity in distinct MSN subtypes.

Methods and Materials

For detailed methods, please see the Supplemental Methods and Materials. Briefly, wild-type male mice were subjected to cocaine self-administration (1.2 mg/kg/infusion), which was conducted for 23 h/day. After 10 daily progressive ratio (PR) sessions, the subject mice were divided into susceptible or resilient groups based on the average breakpoint of the last three days. Breakpoint was defined as the number of active nosepoke respondings to receive the last infusion. Subsets of mice identified as susceptible or resilient group were then subjected to extinction or punishment procedure. During 7-day extinction (1 h/day), nosepoke responding resulted in no consequence. Using distinct subsets of mice, baseline level of nosepoke response was assessed for 3 days with fixed ratio 5 (FR5) schedule without any presentation of punishment. For subsequent 3 days, electric foot-shocks were randomly presented as a punishment, by one-third chance upon each active nosepoke. Other subsets of susceptible and resilient mice were employed for in vivo extracellular recording, RNA sequencing of sorted ChINs, ex vivo slice recordings, or for single-cell RT-PCR experiment.

For cell-type specific manipulations, we used adult male Chat-Cre or Chat-Cre/Drd2-eGFP double transgenic mice. We bilaterally injected adeno-associated virus (AAV) into the NAc, which allowed for Cre-specific DRD2 overexpression, DRD2 knock-down, or expression of designer receptor exclusively activated by a designer drug (DREADD) derived from the kappa-opioid receptor (KORD) (19). For optogenetic experiments, a channelrhodopsin-2 (ChR2)–expressing AAV was additionally injected into the mPFC, the vHPC or the BLA. After injection, at least 5-week recovery period was allowed prior to ex vivo patch-clamp recordings, morphological analysis, or behavioral assessments. To assess postsynaptic current in DRD2 knock-downed mice, cocaine-HCl (15 mg/kg) was intraperitoneally (i.p.) injected for 5 consecutive days, which was followed by 1-d withdrawal.

Results

Segregation of susceptible and resilient mice by behavioral phenotypes

We sought to divide inbred C57BL/6 mice based on the level of motivation toward contingent consumption of cocaine. To accomplish this, we devised a behavioral paradigm modified from a self-administration procedure previously used for rapid increase in drug motivation in rats (20). The procedure consists of a fixed ratio 1 (FR1) schedule for cocaine infusion followed by a 10-day progressive ratio (PR) schedule (Figure 1A). The final breakpoints, which were measured as an index of motivation, were distributed over a wide range with a local minimum around breakpoint 32 (Figure 1B). We employed a k-means clustering analysis using final breakpoints and timeout responding (Supplemental Figure 1A), to systematically separate a mouse population into high- and low-motivation groups (division at breakpoints at 32.33; Figure 1B). As expected, mice displaying high motivation showed progressively increasing levels of responding toward cocaine infusions throughout the 10-day assessment, which was significantly higher than the low-motivation group (Figure 1C and Supplemental Figure 1B–E). Interestingly, the difference in cocaine intake between groups was undetectable during the initial acquisition (Supplemental Figure 1F), and the initial acquisition efficacy (as measured by the latency to accomplish 40 infusions) was not predictive of the final measures of motivation (Supplemental Figure 1G).

Figure 1. Segregation of mouse populations based on susceptibility to addiction-like behaviors after cocaine self-administration.

(A) A timeline of experimental procedure. After initial acquisition and breakpoint measurement using cocaine self-administration, a subset of mice were subjected to extinction or punishment procedure. (B) A summary histogram showing final breakpoints that individual mice displayed. The dotted line depicts the borderline breakpoint that divides susceptible and resilient animals. (C) Daily breakpoints that resilient (n = 102 mice) and susceptible (n = 56 mice) groups displayed throughout progressive ratio (PR) sessions (group × time, F_9,495 = 9.12, ***P < 0.001). (D) Relative numbers of active nosepokes despite random delivery of electric foot-shocks (group × time, F_5,40 = 4.11, **P < 0.01, Susceptible, n = 10 mice; Resilient, n = 7 mice). (E) Number of active nosepokes under cocaine unavailability (group × time, F_6,72 = 3.51, **P < 0.01, Susceptible, n = 11 mice; Resilient, n = 13 mice). (C–E) Two-way repeated-measures analysis of variance was used. Data are presented as mean ± s.e.m.

Consistent with our hypothesis that the motivation index could represent susceptibility to addiction-related phenotypes, mice with high motivation exhibited other core addiction-like behaviors. Compared to the low-motivation animals, the high-motivation group exhibited higher levels of nosepokes during random presentation of electric foot-shocks at the time of each active nosepoke (Figure 1D and Supplemental Figure 1H, I), which indicates drug seeking despite receiving contingent punishment. The high-motivation group also displayed higher level of seeking behaviors during extinction procedure, as they performed more nosepokes than low-motivation mice in the absence of cocaine infusion (Figure 1E). Finally, the high-motivation group exhibited more impulsive seeking behaviors than the low-motivation group, as assessed by the number of nosepoke during a post-infusion time-out period (Supplemental Figure 1J, K) (21). Collectively, these behavioral analyses supported our notion that the high- and low-motivation groups represented animals that were susceptible and resilient, respectively, to express addiction-like behaviors.

Activation of NAc ChINs by cocaine infusion is blunted in susceptible mice

We recorded NAc ChINs to examine whether and how ChINs of susceptible and resilient mice responded differently to systemic cocaine infusion. After catheter implantation and cocaine self-administration, single-unit recording was conducted from the NAc shell region in vivo (Figure 2A, B, and Supplemental Figure 2A, B). The unit activity of putative ChINs was identified by inter-spike interval and spike frequency (10). Consistent with previous reports (13,22), intravenous infusion of cocaine increased ChIN firing rates in drug-naïve control mice (Figure 2C). We also observed elevated ChIN activity in the resilient group (Figure 2D), but failed to detect any significant alterations of ChIN activity of the susceptible mice (Figure 2E). Hence, NAc ChIN activity would be affected in susceptible mice where it could potentially play instructive roles for progressive ratio responding.

Figure 2. In vivo single unit recording from the nucleus accumbens (NAc) cholinergic interneurons (ChINs) in each group of mice.

(A) An experimental timeline for identifying resilient/susceptible mice. Drug-naïve mice were implanted with intravenous catheter and kept in the home cage until recording. Recordings were conducted from resilient and susceptible mice after three-day withdrawal from the last PR test. (B) A schematic illustration of in vivo single unit recording. Cocaine (1.2 mg/kg, i.v.) was injected through a catheter connected to right jugular vein during recording. (C) Average firing rate of putative ChIN units in drug-naïve mice during the first exposure to cocaine (1.2 mg/kg, i.v.) are presented as Z-score (n = 9 units from 5 mice). (D, E) Unit recordings from resilient (D) (n = 12 units from 8 mice) or susceptible (E) (n = 23 units from 9 mice) mice. Recordings were conducted after three-day withdrawal from the final PR test. Dotted lines indicate the onset of intravenous cocaine injection. Data are presented as mean ± SEM. FR, fixed ratio; i.v., intravenous; NAc, nucleus accumbens; PR, progressive ratio.

Higher Drd2 expression in NAc ChINs of susceptible mice

Neural and behavioral plasticity underlying drug addiction are associated with altered expression of numerous genes in the NAc (23). To capture those changes on a transcriptome-wide level, we exploited fluorescence-activated cell sorting (FACS) to perform cell-type-specific RNA-sequencing from NAc ChINs after completion of the PR schedule (Figure 3A, Supplemental Figure 3A, B, and Supplemental Table 1). As expected, sorted ChINs expressed a significantly higher level of Chat, a molecular marker for ChINs, and contained only marginal levels of mRNAs encoding marker proteins for other striatal interneuron subtypes and MSNs (Figure 3B). Differential expression analysis (fold change > 2 and FDR < 0.01) identified 2,909 differentially expressed genes (DEGs) in NAc ChINs from susceptible mice compared to resilient mice. These DEGs were composed of 1,685 upregulated genes and 1,224 downregulated genes (Figure 3C, D, and Supplemental Table 2). Our DEG analysis also revealed that the most abundant readouts were transcripts of protein-coding genes (66.8%), followed by long noncoding RNAs (12.9%) and pseudogenes (8.3%) with remaining categories accounting for 12.1% (Supplemental Figure 3C), similar to those of NAc D1- and D2-MSNs (24). However, there was little overlap of responsive DEGs (only eleven DEGs in common) between SST-positive interneurons and NAc ChINs (Supplemental Figure 3D) (25).

Figure 3. Cell-type-specific RNA sequencing of susceptible and resilient mice shows upregulated DRD2 signaling in NAc ChINs of susceptible mice.

(A) Representative FACS gating for isolation of NAc ChINs. Isolated and suspended NAc cells from wild-type mice were incubated with or without an Alexa Fluor 488-conjugated antibody (Ab) against choline acetyltransferase (ChAT). (B) Expression levels of various neuronal markers in sorted ChAT+ cells, as assessed by RT-PCR (n = 4–5 experiments). Chat transcripts were highly enriched in the Ab+ fraction. Ppp1r1b (DARPP-32) as a marker for medium spiny neuron (MSN) and markers for individual GABAergic interneurons such as Sst, Vip, Calb2, Cck, and Pvalb were assessed (Student’s paired t-test, t = 6.02, **P < 0.01, relative to mRNA levels of individual genes in No Ab fraction). Data are presented as mean ± s.e.m. (C) A volcano plot based on p-values and fold changes (FC) of individual genes highlights DEGs in NAc ChINs of susceptible mice (n = 9 mice). Example genes implicated in the dopamine (DA) signaling pathway are specified. (D) Heat map illustration for DEGs in ChINs of susceptible mice, with a red-to-blue gradient depicting upregulation (red) and downregulation (blue). Each row represents an individual experiment (n = 3 mice per experiment). (E) Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis indicating responsive signaling pathways in NAc ChINs of the susceptible group compared to the resilient group. (F) Gene Ontology enrichment analysis of differentially expressed genes showing the altered expression of genes that are intimately involved in dopaminergic signaling, such as cytoskeleton organization and small GTPase-mediated signaling. **p < .01. cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; ChAT, choline acetyltransferase; ChINs, cholinergic interneurons; DRD2, dopamine D2 receptor; FC, fold change; FDR, false discovery rate; GABAergic, gamma-aminobutyric acidergic; KEGG, Kyoto Encyclopedia of Genes and Genomes; mRNA, messenger RNA; NAc, nucleus accumbens; ncRNA, noncoding RNA; PI, propidium iodide; PKG, protein kinase G.

Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis of upregulated DEGs in NAc ChINs from susceptible mice revealed that DAergic signaling was one of the top three upregulated pathways in susceptible animals (Figure 3E and Supplemental Table 3). Other highly affected pathways, including long-term potentiation, endocytosis, and cAMP/cGMP signaling, are also implicated in DAergic signaling (26,27). Gene ontology terms related to DAergic signaling, such as cytoskeleton organization and small GTPase-mediated signaling, were identified (Figure 3F) (28,29). Among genes involved in DAergic signaling, upregulation of Drd2 was of particular interest, as DRD2 in ChINs are proposed to regulate neuronal activity, synaptic plasticity, and behaviors (30,31). Also, increased DRD2 activity, due to its elevated abundance in ChINs of susceptible mice, could potentially counteract the cocaine-triggered activation of ChINs observed in control and resilient mice (Figure 2C).

One of physiological consequences that higher expression of DRD2 in ChINs can produce would be exaggerated DRD2-mediated reduction of neuronal activity (30). We perfused a DRD2 agonist quinpirole onto acute brain slices containing the NAc and measured spontaneous firing rates of NAc ChINs in a cell-attached configuration (Figure 4A–C). Quinpirole successfully decreased spontaneous activity of NAc ChINs (Figure 4D). The quinpirole-induced inhibition of ChIN activity was intact under treatment with a cocktail of synaptic blockers (APV, DNQX, and picrotoxin), suggesting that the DRD2-mediated reduction of neuronal activity was independent of synaptic inputs (Supplemental Figure 4A). To elucidate the possible correlation between drug motivation and DRD2 abundance in NAc ChINs, we assessed the final breakpoints from the PR test and the magnitude of DRD2-mediated inhibition by performing ex vivo cell-attached recordings (Figure 4A). Importantly, the magnitude of firing rate reduction by quinpirole positively correlated with measured breakpoints (Figure 4E). Furthermore, dispersed levels of DRD2 abundance appeared not to arise from an innately conferred difference, given that magnitudes of firing rate reduction were more widely distributed after completing the PR test, compared to drug-naïve controls (Supplemental Figure 4B, C). We also measured Drd2 mRNA levels by qRT-PCR with aspirated intracellular contents of ChINs from mice subjected to cocaine self-administration (Figure 4A, F, and Supplemental Figure 4D). Indeed, single-cell qRT-PCR corroborated higher levels of Drd2 mRNAs in NAc ChINs of susceptible mice than those of resilient mice (Figure 4G).

Figure 4. Elevated dopamine D2 receptor (DRD2) abundance in NAc ChINs of susceptible mice.

(A) Schematic workflow of measuring DRD2 abundance and testing physiological change in NAc ChINs between resilient and susceptible mice. (B) Confocal images of an example recorded NAc ChIN. Neurons were recorded with cell-attached recording, and neurobiotin was injected during subsequent whole-cell recording. Recorded cells were verified by co-labeling of neurobiotin (purple) and ChAT (green). Scale bar, 50 μm. (C) Representative traces of whole-cell recording in NAc ChIN. 1-s current steps (−200, −100, 0, and +50 mV) were applied. Note that the recording was performed after bath application of quinpirole during cell-attached recording. Scale bars, 20 mV and 200 ms. (D) Representative traces of a cell-attached recording in NAc ChINs before (baseline) and after (+quinpirole) bath application of quinpirole (1 μM). Scale bars = 20 pA and 6 seconds. (E) Correlation between the level of motivation toward cocaine infusion and magnitude of firing rate reduction by quinpirole application (Pearson’s correlation coefficient, r = 0.668, P < 0.01, n = 15 mice). Dotted lines represent 99% confidence interval. (F) DIC images of a putative ChIN in acute brain slices before (Whole-cell mode) and after aspiration of intracellular contents. Dotted lines delineate the ChIN soma, which is initially selected by its large soma size. (G) Relative amounts of Drd2 mRNA in ChINs of resilient (R, n = 18 cells from 4 mice) and susceptible (S, n = 10 cells from 4 mice) groups (Mann-Whitney U test, *P < 0.05). Data are presented as mean ± SEM. ChAT, choline acetyltransferase; ChINs, cholinergic interneurons; DIC, differential interference contrast; DRD2, dopamine D2 receptor; FR, fixed ratio; mRNA, messenger RNA; NAc, nucleus accumbens; PR, progressive ratio; R, resilient; RT-PCR, reverse transcriptase polymerase chain reaction; S, susceptible; WD, withdrawal.

DRD2 in NAc ChINs is sufficient and necessary for cocaine seeking

To examine whether increased abundance of DRD2 was sufficient to induce addiction-like behavior, we overexpressed DRD2 by microinjecting Cre-dependent AAV vectors into the NAc of Chat-Cre mice. We confirmed the selective overexpression by co-labeling of turboGFP and ChAT (Supplemental Figure 5A–D), and validated that firing rates were further reduced by quinpirole in DRD2-overexpressing ChINs compared to uninfected control ChINs without affecting the baseline firing rate (Figure 5A and Supplemental Figure 5E). Importantly, Chat-Cre mice with DRD2 overexpression in NAc ChINs exhibited higher motivation toward cocaine infusion than Cre-negative controls (Figure 5B–E and Supplemental Figure 5F, G). The increased motivation was unlikely due to reduced anxiety-like behavior, since novelty-induced suppression of feeding was comparable between the two groups (Supplemental Figure 5H).

Figure 5. DRD2 in NAc ChINs is necessary and sufficient for cocaine seeking.

(A) Electrophysiological verification of DRD2 overexpression (DRD2-OE) and knockdown (shDrd2) (Control, n = 17 cells; DRD2-OE, n = 6 cells; shDrd2, n = 6 cells) (One-way RM-ANOVA for shDrd2, F_17,85 = 0.56, P > 0.1 n.s., not significant; Two-way RM-ANOVA for DRD2-OE vs. control, group × time, F_17,85 = 4.99, ***P < 0.001). (B) A timeline depicting virus injection and breakpoint measurement. (C) Schematic illustration of stereotaxic surgery for AAV-induced overexpression of DRD2 in NAc ChINs. (D, E) Breakpoints of the first day (D) and the final 3 days (E) of PR tests after DRD2 overexpression (Control, n = 13 mice; DRD2-OE, n = 11 mice). (F) Viral infusion of Cre-dependent shDrd2 into the NAc of Chat-Cre mice. White and red triangles represent lox2722 and TATA-lox, respectively. (G, H) Breakpoints on the first day (G) and the final 3 days (H) of PR tests in Cre-negative control (n = 14 mice) and shDrd2 group (n = 16 mice). (I, J) Representative whole-cell recording traces from an uninfected ChIN (I) and designer receptor derived from the kappa-opioid receptor (KORD) -expressing NAc ChIN (J). Black lines depict bath-application of salvinorin B (SalB, 100 uM). Scale bars, 20 mV and 30 s. (K) Schematic illustration of breakpoint assessment after bilaterally injecting AAV encoding double-floxed KORD into the NAc of Chat-Cre or Cre-negative control mice. Cocaine and SalB were injected during PR schedule. (L, M) Breakpoints on the first day (L) and the final 7–10 days (M) of PR tests (Control, n = 10 mice; KORD, n = 9 mice). Mann-Whitney U test were used for comparisons. *P < 0.05, **P < 0.01. Data are presented as mean ± SEM. AAV, adeno-associated virus; Chat, choline acetyltransferase; ChINs, cholinergic interneurons; DIO, double-floxed inverted orientation; DRD2, dopamine D2 receptor; eYFP, enhanced yellow fluorescent protein; FR, fixed ratio; i.v., intravenous; KORD, kappa opioid receptor; NAc, nucleus accumbens; n.s., not significant; OE, overexpression; PR, progressive ratio; SalB, salvinorin B; tGFP, turbo green fluorescent protein.

To determine whether DRD2 in NAc ChINs was required for development of addiction-like behaviors, we next depleted DRD2 in NAc ChINs using a Cre-dependent knockdown (cKD) AAV (32). We microinjected Cre-dependent knockdown AAV expressing a short hairpin RNA against the Drd2 (shDrd2) into the NAc of Chat-Cre mice (Figure 5F), and confirmed that most of the ChAT-positive neurons expressed eYFP which was co-expressed in cells expressing shDrd2 (Supplemental Figure 5I–L) (32). Using cell-attached recordings, we validated that DRD2 knockdown abolished quinpirole-induced reduction of spontaneous firing rates in NAc ChINs (Figure 5A and Supplemental Figure 5E). Importantly, mice with depleted DRD2 had significantly lower motivation toward cocaine infusion than control animals (Figure 5G, H and Supplemental Figure 5M, N). These findings support the notion that DRD2 in NAc ChINs is sufficient and necessary for development and maintenance of susceptibility traits to cocaine addiction.

Inactivation of NAc ChINs leads to increased drug motivation

DRD2 activation in striatal ChINs induces various physiological changes, including depression of spontaneous neuronal activity, elongation of post-burst pause, and disinhibition of Ca²⁺ channels in downstream MSNs (16,30,31). We sought to determine whether the emergence of addiction-like behaviors could result from inhibition of NAc ChIN activity induced by activation of DRD2. Thus, we inhibited ChIN activity by cell-type-specific expression of KORD and its activation by salvinorin B (SalB) administration (19), which is known to exert minimal effects on endogenous DA receptors (33). We expressed KORD selectively in NAc ChINs by using a Cre-dependent AAV vector and Chat-Cre mice, and confirmed that SalB application suppressed spontaneous activity in AAV-infected ChINs, but had negligible effect on uninfected control ChINs (Figure 5I, J). Mice expressing KORD in NAc ChINs were intravenously infused with SalB together with cocaine during PR tests (Figure 5K). KORD-expressing mice exhibited higher motivation to cocaine compared to Cre-negative control mice which also received SalB during cocaine infusion (Figure 5L, M and Supplemental Figure 5O, P).

DRD2-induced behavioral changes are accompanied by cell-type-specific alteration of dendritic spine density and excitatory inputs of MSNs

Chronic cocaine exposure can cause morphological changes in NAc MSNs. Specifically, repeated non-contingent cocaine injection followed by short withdrawal increased dendritic spine density in D1-MSNs, but not in D2-MSNs (34). Motivation to cocaine, which was elevated by DRD2 overexpression in ChINs, would be associated with structural changes of neighboring NAc MSNs. Consistent with this notion, D1-MSNs, but not D2-MSNs, exhibited increased spine density after a cell-type specific overexpression of DRD2 in NAc ChINs despite the absence of cocaine injection (Figure 6A, B).

Figure 6. DRD2 in NAc ChINs is sufficient and necessary to induce cocaine-related cell type- and pathway-specific changes in NAc MSNs.

(A) Representative images of dendritic spines from D1- and D2-MSNs. A DRD2-encoding AAV was injected into the NAc of Chat-Cre mice (ChIN::DRD2-OE) or Cre-negative mice (Control). (B) Spine densities from D1- and D2-MSNs of control and DRD2-OE mice (Kruskal-Wallis H test with Dunn’s post hoc test for multiple comparison, **P < 0.01, n ≥ 8 cells for each group). (C) Schematic depiction of dual whole-cell recording from D1- and D2-MSNs. Channelrhodopsin2 (ChR2) is expressed in the ventral hippocampus (vHPC) or basolateral amygdala (BLA) of Chat-Cre × Drd2-eGFP double transgenic mice. (D) Example excitatory postsynaptic current (EPSC) traces simultaneously recorded from D1-MSNs (red) and D2-MSNs (black) upon stimulation of the vHPC-NAc pathway. (E, F) Summary of EPSC amplitudes in the vHPC-NAc pathway, recorded in slices from Cre-negative control (E) (n = 13 MSN pairs from 5 mice) or DRD2-overexpressing (F) (n = 9 MSN pairs from 3 mice) mice. (G) Example EPSC traces upon stimulation of the BLA-NAc pathway. Scale bars, 40 ms and 50 pA. (H, I) Summary of EPSC amplitudes in the BLA-NAc pathway, recorded in slices from Cre-negative control (H) (n = 12 MSN pairs from 5 mice) or DRD2-overexpressing (I) (n = 9 MSN pairs from 4 mice) mice. Wilcoxson signed-rank test was used for analysis of paired-recordings. **p < .01. AAV, adeno-associated virus; BLA, basolateral amygdala; Chat, choline acetyltransferase; ChINs, cholinergic interneurons; ChR2, channelrhodopsin-2; DIO, double-floxed inverted orientation; DRD2, dopamine D2 receptor; EPSC, excitatory postsynaptic current; eYFP, enhanced yellow fluorescent protein; i.v., intravenous; MSN, medium spiny neuron; NAc, nucleus accumbens; n.s., not significant; OE, overexpression; tGFP, turbo green fluorescent protein; vHPC, ventral hippocampus.

Chronic exposure to cocaine results in altered excitatory postsynaptic current (EPSC) ratios between D1- and D2-MSN in a pathway-specific manner (35). After cocaine injection, EPSC ratio of D1-/D2-MSN increased in the BLA-NAc pathway, but decreased in the vHPC-NAc pathway, while the mPFC-NAc pathway remained unaffected (35). Using channelrhodopsin 2 (ChR2)-encoding AAV and Chat-Cre/Drd2-eGFP double transgenic mice, we monitored impacts of ChIN-specific DRD2 overexpression on optically-driven EPSCs in distinct MSN subtypes (Figure 6C). In three distinct experiments, ChR2 was expressed in upstream brain regions of the NAc: the vHPC, the BLA, or the mPFC (Figure 6C and Supplemental Figure 6A). Simultaneous whole-cell recording from D1- and D2-MSNs revealed that the EPSC ratio between two MSN subtypes were shifted in a pathway-selective fashion after NAc ChIN-specific DRD2 overexpression in drug-naïve mice. EPSC ratios of D1- to D2-MSNs decreased in the vHPC-NAc pathway (Figure 6D–F), but increased in the BLA-NAc pathway (Figure 6G–I), whereas the mPFC-NAc pathway was unaffected (Supplemental Figure 6B, C). To further investigate which component of the EPSCs accounted for DRD2-induced plasticity, we next assessed quantal EPSCs (qEPSCs) in bath solutions containing Sr²⁺ (Supplemental Figure 6D). While both qEPSC amplitude and frequency remained unaltered in the mPFC-NAc pathway (Supplemental Figure 6E, F), qEPSC frequency, but not amplitude, was increased in the BLA-NAc pathway after ChIN-specific DRD2 overexpression (Supplemental Figure 6G, H).

We also examined whether DRD2 in NAc ChINs was required for cocaine-induced synaptic alterations in the BLA-NAc and vHPC-NAc pathway, which have been implicated in cocaine craving and sensitization (6,35,36). DRD2 was selectively depleted in NAc ChINs, and ChR2 was expressed in the vHPC or in the BLA (Figure 7A, E, and Supplemental Figure 7A, B). Consistent with a previous report (35), Cre-negative control group showed alteration of EPSC ratios after 5 daily cocaine i.p. injection: attenuated D1/D2 EPSC amplitude ratios in vHPC-NAc pathway but increased D1/D2 ratios in the BLA-NAc pathway (Figure 7B, C, F, and G). ESPC ratios in both pathways remained unaffected despite cocaine exposure when DRD2 was depleted (Figure 7D, H). These data highlight the necessity of DRD2 signaling in ChINs for cocaine-induced synaptic plasticity in both the vHPC-NAc and the BLA-NAc pathways. Interestingly, mice harboring DRD2 overexpression in NAc ChINs also displayed elevated locomotion to non-contingent cocaine administration, which was apparent from the first exposure to cocaine (Supplemental Figure 8A, B). Collectively, increased abundance of DRD2 induced synaptic plasticity in MSNs and subsequently triggered behavioral sensitization, further substantiating a causal relationship between synaptic plasticity and drug-related behaviors.

Figure 7. Cre-dependent depletion of ChIN DRD2 in the NAc interferes with cocaine-induced synaptic plasticity in a pathway-specific manner.

(A) Schematic illustration of Cre-dependent DRD2 knockdown experiment in the vHPC-NAc pathway. (B) Representative EPSC traces induced by optic stimulation of the vHPC-NAc pathway in Cre-negative (control) and DRD2-depleted (ChIN::shDrd2) mice. Recordings were conducted after 1–2 d withdrawal from five-day exposure to cocaine (15 mg/kg, i.p.). (C, D) Summary of EPSC amplitudes in D1- and D2-MSNs of cocaine-injected mice. EPSCs were recorded ex vivo in control (C) (n = 11 MSN pairs from 4 mice) or ChIN::shDrd2 (D) (n = 14 MSN pairs from 5 mice) mice. (E) Schematic illustration for knock-down experiments in the BLA-NAc pathway. (F) Representative EPSC traces induced by stimulating the BLA-NAc pathway. (G, H) Summary of EPSC amplitudes in D1- and D2-MSNs of (G) cocaine-injected control (n = 16 MSN pairs from 5 mice) and (H) cocaine-injected ChIN::shDrd2 (n = 13 MSN pairs from 5 mice) mice. Blue bars indicate timing of optic stimulation. Scale bars = 40 ms and 50 pA. Wilcoxon signed rank test was used for analysis of paired-recordings. *P < 0.05, **P < 0.01, and ***P < 0.001.

Discussion

Here, we demonstrated that DA signaling in NAc ChINs causally controls cocaine seeking behavior. Our genome-wide analysis of NAc ChINs from susceptible versus resilient animals identified numerous DEGs that could potentially contribute to the emergence of susceptibility traits after cocaine self-administration. Among these DEGs, Drd2 was upregulated in susceptible animals, and was sufficient and necessary for development of addiction-like behaviors. We also detected that activation of NAc ChINs during cocaine exposure was blunted in susceptible animals, consistent with the observation that chemogenetic suppression of ChIN activity resulted in elevated drug motivation. This DRD2/ChIN-mediated behavioral change appeared to arise primarily due to pathway-specific synaptic plasticity occurring at MSNs.

By adapting the behavioral paradigm in which rats developed rapid escalation of drug motivation (20), we could segregate susceptible and resilient mice within a relatively short period. The two distinct subpopulations showed a relatively small difference (<15%) of cocaine intake between subpopulations, together with limited lifetime intake (<250 mg/kg). Because exposure to high doses of cocaine could affect release and uptake of DA and consequently alter locomotor responses even after long-term withdrawal (37), identification of susceptible and resilient animals based on volitional seeking with lower cocaine dose may be helpful for investigation of precise mechanisms underlying susceptibility. We further validated that the high-motivation group exhibited core addiction-like behaviors (2). Measurement of each behavioral phenotype in distinct groups could also exclude possibilities of repeated extinction, reinstatement, and association of drug-predicting cues with aversive stimulus, which might affect NAc neuroplasticity (38).

The individual difference of DRD2 abundance is likely to be induced during cocaine self-administration, rather than to arise from innate heterogeneity, not only because we used inbred mice, but also because magnitudes of quinpirole-induced depression of ChIN activity were more widely distributed after completion of the PR schedule. The detailed mechanism by which DRD2 levels are elevated during cocaine self-administration still remains to be determined. Importantly, DRD2 downstream genes comprising both PKA (protein kinase A)-dependent (Atf4, Creb1, Crebbp, Prkaca) and PKA-independent (Arrb2, Akt1) signaling pathways were upregulated in NAc ChINs of susceptible mice (Table S2 in Supplement 2) (39). Moreover, our KEGG pathway analysis revealed that signaling pathways previously reported to be altered by exposure to addictive drugs, such as cAMP-PKA-CREB pathways, were also differentially regulated (40). Although cocaine-affected genes previously reported using whole-NAc assessments were conserved in ChINs, by contrast, further comparison with SST+ interneuron data showed only little overlapping of responsive DEGs (Supplemental Figure 3D). These results highlight potential roles of cell-type-specific gene regulation between distinct neuronal subtypes for addiction susceptibility.

We established a causal relationship between ChIN activity and motivation to cocaine using multiple genetic tools to control DRD2 abundance specifically in NAc ChINs; DRD2 overexpression in NAc ChINs increased motivation to cocaine and, conversely, DRD2 depletion reduced cocaine-seeking behavior. These results appear to be at odds considering a previous report suggesting reduced reward-like effect under cocaine conditioned place preference paired with and optogenetic silencing of ChINs (22). However, it is worth emphasizing that genetic tools that we used, compared to a transient inactivation during single non-contingent cocaine exposure, allows for DRD2 overexpression for at least 5 weeks, which by itself result in MSN synaptic plasticity which could support excessive cocaine seeking (36). Finally, our single unit recording data indicated that cocaine exposure increased ChIN activity in resilient animals, but not in susceptible animals in which the ChIN DRD2 abundance was upregulated. It is reasonable to speculate that ChIN activation during cocaine exposure, which was exhibited by resilient animals, exerts a defensive action and thereby deters the transition to addictive states after cocaine consumption.

Activation of ChINs can increase the release of DA in the NAc by direct activation of nAChRs localized on the axon terminals of DAergic neurons (18). Reduction of ChIN activity is likely to induce susceptibility to cocaine seeking, at least in part through decreased DA release in the NAc, which would occur due to elongated ChIN pausing (17). It was recently shown that the phasic DA release independent of DAergic neuron activity underlies motivation for reward seeking (41), suggesting the physiological significance of locally induced DA release in the NAc. Interestingly, ample evidence indicates that enhanced DA efflux plays a central role in development of reward-like behaviors elicited by cocaine (3), but it was also previously demonstrated that DA levels during cocaine intake diminished after prolonged long-term access to cocaine (42,43). Simultaneous monitoring of ACh and DA concentrations, along with MSN neuronal activity during drug-seeking behavior, will make it possible to delineate their subsecond dynamics, which would be a key regulating factor for synaptic plasticity and addiction-like behaviors. In fact, because both muscarinic ACh receptors and DA receptors are G protein coupled and share downstream signaling pathways, the crosstalk between these two neurotransmitters could yield synergetic or gating effects on signaling molecules critical for synaptic plasticity (44,45).

One of compelling features in this study is that ChIN-specific DRD2 overexpression was sufficient to induce pathway- and cell-type-specific synaptic plasticity in MSNs without cocaine exposure. Importantly, all types of synaptic plasticity we observed without cocaine treatment were reminiscent of the alterations observed in mice that received repeated cocaine (34,35). Furthermore, ChIN-specific depletion of DRD2 precluded cocaine-induced alteration of EPSC ratios, indicating a necessary role of ChIN DRD2 for cocaine-induced synaptic plasticity. In the dorsal striatum, DRD2 activity in ChINs was proposed to regulate cell type–specific synaptic plasticity via activation of muscarinic ACh receptors in the corticostriatal pathway (31,45). However, the effects of ChIN DRD2 and presynaptic muscarinic ACh receptors on MSN plasticity have not been fully addressed in the other pathways, especially in the NAc. It is possible that the BLA-NAc and vHPC-NAc circuits can be distinctly regulated by a single signaling pathway, as it has been shown for other types of presynaptic G protein-coupled receptors such as KORDs (46). Furthermore, while it was widely documented that GABAergic (gamma-aminobutyric acidergic) regulation of neuronal activity or spike timing can affect synaptic plasticity (47), effects of ChIN-induced multisynaptic inhibitory transmission (8) on MSN plasticity still remain elusive. Therefore, unraveling the molecular crosstalk and antagonistic interactions among distinct neuronal subtypes and differential cholinergic actions in the NAc will provide new insight into the mechanisms of cocaine addiction.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Antibody	Goat anti-Choline Acetyltransferase	Millipore	AB144P
Antibody	Rabbit anti-turboGFP	Evrogen	AB513
Antibody	Donkey anti-goat Alexa Fluor 488	ThermoFisher	A-11055
Antibody	Donkey anti-goat Cy5	Jackson ImmunoResearch	705-175-147
Antibody	Donkey anti-rabbit Alexa Fluor 488	ThermoFisher	A-21206
Antibody	Rabbit anti-Choline Acetyltransferase antibody, Alexa Fluor 488 conjugate	abcam	ab192465
Bacterial or Viral Strain	rAAV5-Ef1a-DIO-Drd2-turboGFP-WPRE	This paper	n/a
Bacterial or Viral Strain	rAAV1-Ef1a-DIO-TATAlox-eYFP-U6-shDrd2-WPRE	This paper	n/a
Bacterial or Viral Strain	rAAV5-hSyn-dF-HA-KORD-IRES-mCitrine	Addgene	#65417
Bacterial or Viral Strain	rAAV5-CamKIIa-hChR2-mCherry-WPRE	UNC vector core	n/a
Bacterial or Viral Strain	rAAV5-CamKIIa-hChR2-eYFP-WPRE	UNC vector core	n/a
Cell Line	HEK293T	ATCC	CRL-11268
Chemical Compound or Drug	Cocaine-HCl	Macfarlan Smith Ltd.	n/a
Chemical Compound or Drug	DL-APV	Sigma	A5282
Chemical Compound or Drug	DNase I	Sigma	AMPD1
Chemical Compound or Drug	Heparin	Sigma	H3393
Chemical Compound or Drug	NBQX	Tocris	'0373
Chemical Compound or Drug	Picrotoxin	Tocris	'1128
Chemical Compound or Drug	Salvinorin B	Tocris	'5611
Chemical Compound or Drug	(–)-Quinpirole hydrochloride	Tocris	'1061
Commercial Assay Or Kit	Directzol RNA-Mini prep	Zymo	R2051
Commercial Assay Or Kit	Single Cell-to-CT™ qRT-PCR Kit	ThermoFisher	'4458237
Commercial Assay Or Kit	SMART-Seq® v4 Ultra® Low Input RNA Kit for Sequencing	Clontech	'634894
Deposited Data; Public Database	Raw and analyzed RNA-sequencing data	This paper	GEO: GSE130376
Organism/Strain	Chat-Cre	MMRRC	#030869-UCD
Organism/Strain	Drd2-eGFP	MMRRC	#000230-UNC
Recombinant DNA	Drd2(m)-eGFP	Origene	MG226860
Sequence-Based Reagent	Taqman Gene expression assays: Drd2	Applied Biosystems	Mm00438545_m1
Sequence-Based Reagent	Taqman Gene expression assays: Rn18s	Applied Biosystems	Mm03928990_g1
Sequence-Based Reagent	Taqman Gene expression assays: Chat	Applied Biosystems	Mm01221882_m1
Sequence-Based Reagent	shDrd2	Genepharma	n/a	Targeting sequence:ACCTGGCATGCCAAGATTGTCAATCTTCAAGAGAGATTGACAATCTTGGCATGCCTTTTTTG
Software; Algorithm	Clampex 10.2	Molecular Devices
Software; Algorithm	Cluster 3.0 for Windows	Michael Eisen, Stanford University
Software; Algorithm	FV10-ASW	Olympus
Software; Algorithm	MATLAB	Mathworks
Software; Algorithm	MED-PC IV	Med associates Inc.	SOF-735
Software; Algorithm	Microsoft Excel 2016	Microsoft
Software; Algorithm	Summit software system	Cytomation
Transfected Construct	pXR5-bam	NGVB	n/a
Transfected Construct	pXR1	NGVB	n/a
Transfected Construct	pXR6–80	NGVB	n/a