Effects of cocaine and withdrawal on the mouse nucleus accumbens transcriptome


Corresponding author: R. E. Mains, Department of Neuroscience, University of Connecticut Health Center, 263 Farmington Ave, MC3401, Farmington, CT 06030–3401, USA. E-mail: mains@uchc.edu


Genetic association studies, pharmacological investigations and analysis of mice-lacking individual genes have made it clear that Cocaine administration and Withdrawal have a profound impact on multiple neurotransmitter systems. The GABAergic medium spiny neurons of the nucleus accumbens (NAc) exhibit changes in the expression of genes encoding receptors for glutamate and in the signaling pathways triggered by dopamine binding to G-protein-coupled dopamine receptors. Deep sequence analysis provides a sensitive, quantitative and global analysis of the effects of Cocaine on the NAc transcriptome. RNA prepared from the NAc of adult male mice receiving daily injections of Saline or Cocaine, or Cocaine followed by a period of Withdrawal, was used for high-throughput sequence analysis. Changes were validated by quantitative polymerase chain reaction or Western blot. On the basis of pathway analysis, a preponderance of the genes affected by Cocaine and Withdrawal was involved in the cadherin, heterotrimeric G-protein and Wnt signaling pathways. Distinct subsets of cadherins and protocadherins exhibited a sustained increase or decrease in expression. Sustained down-regulation of several heterotrimeric G-protein β- and γ-subunits was observed. In addition to altered expression of receptors for small molecule neurotransmitters, neuropeptides and endocannabinoids, changes in the expression of plasma membrane transporters and vesicular neurotransmitter transporters were also observed. The effects of chronic Cocaine and Withdrawal on the expression of genes essential to cholinergic, glutamatergic, GABAergic, peptidergic and endocannabinoid signaling are as profound as their effects on dopaminergic transmission. Simultaneous targeting of multiple Withdrawal-specific changes in gene expression may facilitate development of new therapeutic approaches that are better able to prevent relapse.

Chronic drug exposure induces persistent changes in the brain that underlie the addiction-associated behavioral abnormalities seen in human addicts and rodent models of addiction. Data from broad expression studies using microarray technology and from single-gene approaches suggest that Cocaine exposure elicits widespread modifications to the transcriptional landscape, including drug-induced changes in epigenetics, RNA processing, microRNAs and gene transcription (Eipper-Mains et al. 2011; McClung & Nestler 2008). These alterations are believed to be integral to the neural plasticity seen in addiction.

Dopaminergic, glutamatergic and peptidergic signaling in the mesolimbic system are altered in response to Cocaine. Repeated exposure to Cocaine leads to sensitization, the progressive and persistent amplification of behavioral and motivational responses to a fixed dose of the same drug (Hyman et al. 2006). Sensitization may persist for weeks to years after cessation of drug taking, presumably contributing to a reformed addict's risk for relapse (Saunders & Robinson 2011), which occurs in up to 90% of addicted individuals (Kalivas & Volkow 2011; Wee & Koob 2010).

A majority of current research focuses on the effects of Cocaine on the mesolimbic dopaminergic system, identified in prior studies as the reward center of the brain (Nestler 2005a,b). Dopaminergic projections from the ventral tegmental area as well as glutamatergic projections from the prefrontal cortex synapse on inhibitory GABAergic medium spiny neurons (MSNs) in the ventral striatum (nucleus accumbens; NAc) (Kauer & Malenka 2007). This convergence of dopaminergic and glutamatergic projections, as well as accompanying biochemical and morphological changes, identified the NAc as a key integration point in the rewarding and addictive effects of drugs of abuse (Hyman et al. 2006; Kalivas & Volkow 2011; Nestler 2005a,b; Saunders & Robinson 2011).

High-throughput sequencing has a low background signal, facilitates identification and absolute quantification of low-abundance and novel transcripts (Eipper-Mains et al. 2011; Metzker 2010; Mortazavi et al. 2008; Wahlstedt et al. 2009; Wang et al. 2008a,b; 2008bWang et al. 2009). We used next-generation sequencing of mRNAs purified from the NAc to obtain a more complete picture of the molecular response to chronic Cocaine exposure and Withdrawal. Bioinformatic analysis of Cocaine-regulated transcripts revealed substantial involvement of the Wnt/cadherin and heterotrimeric G-protein signaling pathways. Evaluation of the synthetic enzymes, transporters and receptors involved in catecholaminergic, glutamatergic, GABAergic, cholinergic and lipid signaling identified complex responses unique to Cocaine and Withdrawal.

Materials and methods

Cocaine treatment of mice

Adult male C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME; 2–5 months old) were group-housed in the animal facility at the University of Connecticut Health Center on a 12 h light/dark cycle (lights, 0700–1900 h) and handled in accordance with Institutional Animal Care and Use Committee guidelines. After acclimation for 1 week, animals were handled for 1 min/day for 2–3 days prior to injections. On each treatment day, animals were allowed to acclimate to the behavior room for 45–60 min before beginning injections.

Cocaine treatment was performed as described (Eipper-Mains et al. 2011). Animals were given Saline (10 ml/kg/day) or Cocaine (10 or 20 mg/kg/day) intraperitoneally for 7 days and locomotor activity was monitored daily using a 15″ × 15″ Plexiglas chamber (PAS Open Field System; San Diego Instruments, San Diego, CA); after the seventh injection, locomotor activity was threefold higher in mice receiving Cocaine than in mice receiving Saline (Fig. S1) (Eipper-Mains et al. 2012; Kiraly et al. 2010). Administering 10 mg/kg Cocaine on the first and last days produces more reliable locomotor sensitization (locomotor ambulations on day 7 divided by locomotor ambulations on day 1), as originally demonstrated in rats (Pierce et al. 1996), and was used in later experimental sets in this work. Animals were sacrificed by decapitation 24 h after the final injection with the exception of mice in the Withdrawal groups, which were kept in the home cage for 8 or 28 days after the final Cocaine dose prior to tissue harvest; elevated locomotor responding was maintained for the 28 day Withdrawal period (Fig. S1). Quantitative polymerase chain reaction (qPCR) validation was performed on RNA samples used for library preparation (eight animals analyzed separately in each treatment group) and on two biological replicate groups of mice sensitized in a similar manner (Kiraly et al. 2010): eight each Saline and Cocaine; four each Saline, Cocaine and 28 days Withdrawal. Pairwise comparisons were performed and subjected to t-test (P < 0.05 taken as significant).

RNA and library preparation and sequencing

Nucleus accumbens punches were taken from individual mice (at least eight animals/treatment group). Total RNA was prepared using TRIzol (Invitrogen, Grand Island, NY, USA) (Eipper-Mains et al. 2011). Average yield of total RNA per NAc punch was 6.3 ± 0.6 µg. Total RNA used for NAc library preparation consisted of equal amounts of RNA from eight individual mice for each treatment. One Saline, one Cocaine and one 7-day Withdrawal mRNA-Seq library was prepared from each pooled RNA sample according to manufacturer's specifications (Illumina, San Diego, CA, USA). Briefly, total RNA was treated with DNase I; poly(A)+ RNA purified using Dynal magnetic beads (Invitrogen) was fragmented by partial alkaline hydrolysis (Ambion, Grand Island, CA, USA) and reverse-transcribed using random primers and SuperScript II (Invitrogen). cDNA was size-selected (∼200 bp insert) on 2% agarose gels. Libraries were prepared for sequencing using the Paired-End DNA Sample Prep Kit (Illumina). Nucleus accumbens libraries were sequenced in nine lanes (three technical replicates per sample) on an Illumina GAIIx using a 37-cycle paired-end sequencing protocol. Replicates were analyzed for intra-sample disparity and read data from all three lanes were then merged into one composite data file per sample; intra-sample coefficient of determination, R2 ≥ 0.98. The composite file was used for subsequent analyses.

mRNA-Seq data snalysis

Sequences and quality scores were extracted from image files using GAPipeline software v1.0 or v1.3.2 (Illumina). Sequences were aligned to the Mus musculus genome (2007, mm9; NCBI Build 37) using Bowtie (http://bowtie-bio.sourceforge.net; v0.12.7) (Langmead et al. 2009) permitting multi-alignment and allowing up to two mismatches. Ambiguous alignments were resolved using the Spliced Paired-end Aligner Perl script (Brooks et al. 2011). Robust assembly of novel transcripts and a complete analysis of Cocaine-induced alternative splicing requires greater transcript coverage and read depth than we obtained (Trapnell et al. 2010).

Normalized mRNA expression was calculated as RPKM (reads per kilobase gene model per million mapped reads) (Mortazavi et al. 2008) from aligned sequence reads; Z-scores were computed from frequency data using the following equation: (xμ)/ σ; x, sample frequency; μ, mean across all samples; σ, standard deviation across all samples. Gene expression data (Z-scores) were hierarchically clustered using Gene Cluster v3.0 (http://rana.lbl.gov/EisenSoftware.htm); heat maps were generated using Java Treeview v1.1.6r2 (http://www.sourceforge.net/projects/jtreeview/files). Gene lists were submitted to the Panther website (http://www.pantherdb.org) for analysis of Pathway and Protein Class enrichment compared to the Mus musculus reference gene list. Western and qPCR analyses were performed as described (Eipper-Mains et al. 2011); primer sequences are in Table S1.


Genome coverage

To characterize the effects of chronic Cocaine and Withdrawal on mRNA expression in the NAc, we prepared mRNA-Seq libraries from RNA pooled from mice subjected to daily Saline injections (Saline), daily Cocaine injections (Cocaine) or daily Cocaine injections followed by a week with no injections (Withdrawal). Paired-end sequencing yielded approximately 20 million uniquely mapped reads for each library (Table 1A). The current annotation of the Mus musculus genome (2007, mm9) spans ∼1.6 billion bases; 3.2% of the genome encodes mature mRNAs and 40.6% encodes primary transcripts. We detected 5–6% of the genome in the poly(A)+ RNA-Seq data from each NAc library. Approximately 80% of the reads in each library mapped to annotated exons; 15–16% of reads mapped to introns and 5% of reads aligned to intergenic regions (Table 1B). Mitochondrial reads accounted for roughly 2% of all mapped sequences in these libraries (Fig. S2). Scatter plots of mitochondrial (Fig. S3) and ribosomal (Fig. S4) gene expression showed no evidence for Cocaine regulation, so subsequent analyses were performed on a filtered gene list that excluded both ribosomal and mitochondrial genes.

Table 1. Read mapping and base coverage for RNA-Seq data
A. High-throughput sequencing alignment information
SampleTreatmentNumber aligning readsTotal genome coverage (%) 
  1. Sequences from NAc libraries were aligned to the Mus musculus reference genome (2007, mm9; NCBI Build 37) using Bowtie (Langmead et al. 2009) and allowing up to two mismatches. (A) Aligning read numbers and percent coverage of the genome. (B) Reads were characterized as aligning to exonic, intronic or intergenic regions based on the existing annotation.

NAc lysateSaline26 563 4285.8 
 Cocaine24 696 9815.8 
 Withdrawal19 767 3825.1 
B. Read mapping
SampleTreatmentExonic (%)Intronic (%)Intergenic (%)
NAc lysateSaline79.415.55.1

Genome-wide assessment of the effect of Cocaine on gene expression

Gene expression over a broad (106) dynamic range was evaluated in each library. Expression was quantified in reads per kilobase of exon per million mapped sequence reads (RPKM), a normalized measure of exonic read density (Mortazavi et al. 2008). For the three NAc libraries, pairwise scatter plots revealed similarly varied expression between the Saline/Cocaine, Saline/Withdrawal and Cocaine/Withdrawal libraries (Fig. 1a–c). The variability of gene expression between samples was highest for genes expressed at lower levels. For this reason, a lower limit of expression was applied before examining regulation. Since biogenic amine transporter (Slc6a2, NET; Slc6a3, DAT; Slc6a4, SERT) transcripts are not expressed in the NAc (http://mouse.brain-map.org/), we used their RPKM values to set this lower limit; for inclusion in subsequent analyses, a gene had to have an RPKM of >1.0 in at least one of the three treatment groups. We used qPCR to validate the RNA-Seq data (Fig. 1d,e). The RNA-Seq and qPCR expression levels of 47 transcripts from five neurotransmitter systems from the Saline sample showed strong agreement across the two methods (Pearson's r = 0.75) over a greater than 1000-fold range of expression (Fig. 1d). A comparison of the ratios of Cocaine/Saline measured by RNA-Seq and qPCR was also carried out (Fig. 1e). Cocaine-mediated changes in gene expression estimated from the RNA-Seq data were consistently larger than the changes observed with qPCR; nevertheless, the ratios show good qualitative agreement (Pearson's r = 0.60). Differences presumably reflect the much larger dynamic range of RNA-Seq and quantification across entire transcripts rather than between single primer pairs.

Figure 1.

High-throughput sequence analysis of NAc mRNA. (a–c) Pairwise scatter plots of mRNA expression data for NAc libraries; shown as log10(RPKM) per sample. Ribosomal and mitochondrial genes are not included in the scatter plots. R2 values were calculated by least squares best fit. Sal, Saline; Coc, Cocaine; W/D, Withdrawal; RPKM, reads per kilobase gene model per million mapped reads (Mortazavi et al. 2008). (d) RPKM data for 47 transcripts in the Saline library are compared to qPCR data (normalized to glyceraldehyde 3-phosphate dehydrogenase, GAPDH) for the same transcripts; Pearson's Coefficient represents the linear correlation coefficient, r. (e) Cocaine/Saline ratios for the same 47 transcripts were calculated using RPKM data and qPCR data. Pearson's Coefficient represents the linear correlation coefficient, r, between two samples, X and Y. Pearson's r = covariance(X,Y)/σXσY, where σ is the standard deviation.

Regulation of NAc transcript levels by Cocaine and Withdrawal

To evaluate enrichment and depletion of transcripts in NAc lysates prepared from mice in the Saline, Cocaine and Withdrawal groups, we calculated the relative abundance of each mRNA in the three treatment groups, computed Z-scores across samples, and hierarchically clustered genes for visualization in a heat map. Of the 19 890 detected genes, 13 184 were expressed in at least one library at an RPKM > 1.0 and were used in downstream analyses. Several patterns of expression emerged from hierarchical clustering, including transcripts showing (i) sustained increases in both Cocaine and Withdrawal, (ii) increases only in Withdrawal, (iii) transient increases only in Cocaine, (iv) transient decreases only in Cocaine and (v) sustained decreases in Cocaine and Withdrawal (Fig. 2a). The full gene list for each cluster is available in Table S2.

Figure 2.

Differential regulation of clusters of genes in the NAc following Cocaine and Withdrawal. (a) Normalized mRNA frequency as a function of total reads for NAc lysate libraries (Saline, Cocaine and Withdrawal) was calculated for all genes in which one sample had an RPKM ≥ 1.0 (n = 13 184). Hierarchical gene expression clustering using Gene Cluster 3.0 and Java Treeview generated the dendrogram and heat map. Z-score was computed on normalized mRNA frequency across all samples. Blue, low expression; yellow, high expression; gray, no expression. Gene lists from the indicated clusters (i–v) were used in subsequent protein class and pathway analyses. (b) The effects of Cocaine on one of the protocadherins identified as a target were verified. Subcellular fractionation of striatal tissue revealed enrichment of PCDH10 at the PSD. PSDs purified from the striata of Saline- and Cocaine-treated mice were analyzed for PCDH10; after normalization to βIII tubulin, the Coc/Sal ratio for PCDH10 rose 24% (P < 0.01; t-test).

Since transcripts exhibiting a sustained change in response to Cocaine are included in Clusters i (sustained increase), ii (increase, Withdrawal) and v (sustained decrease), we used Panther Pathway (Tables 2A and S3) analysis to identify common pathways. Three pathways were enriched (P < 0.005) in these three Clusters: cadherin signaling, heterotrimeric G-protein signaling (Gαi and Gαs-mediated) and Wnt signaling; since there was a great deal of overlap in the cadherin and Wnt signaling pathways, they are discussed together. Other pathways were unique to specific clusters (Table S3). For example, components of the ionotropic glutamate receptor, metabotropic glutamate receptor group III and nicotinic acetylcholine receptor pathways were enriched only in cluster i (sustained increase) while components of the muscarinic acetylcholine receptor 1 and 3 pathway were enriched in cluster ii (increase, Withdrawal).

Table 2. Panther Pathways from Cocaine- and Withdrawal-regulated genes in the NAc
Panther PathwayMus musculus genes (26 185)(i) Sustained Increase, Coc & W/D (1387)(i) P-value(ii) Increase, W/D only (1001)(ii) P-value(v) Sustained decrease, Coc & W/D (2670)(v) P-value
  1. Gene lists from clusters i–v (Fig. 2a) were used in Panther Pathway and Protein Class analysis (http://www.pantherdb.org); pathways enriched (P < 0.005) in all three lists relative to the Mus musculus reference genome are indicated. Complete pathway lists are provided in Table S3.

  2. Coc, Cocaine; W/D, Withdrawal.

Cadherin signaling167191.94E−03169.25E−04302.70E−03
Heterotrimeric G-protein signaling – Gαi and Gαs169192.21E−03181.32E−04328.89E−04
Wnt signaling348463.67E−08275.77E−04649.02E−06

The pathways exhibiting sustained changes were examined in more detail. Unique sets of cadherins and protocadherins appeared in each Cluster (Table 3). While many cadherins and protocadherins showed a sustained increase in expression (cluster i), others increased only in Withdrawal (cluster ii) or showed a transient response to Cocaine (clusters iii and iv). The clustered families of protocadherins, Pcdhα, -β and γ, are important in neuronal development and are synaptically enriched in mature myelinated neurons (Morishita & Yagi 2007). The Pcdhβ gene cluster includes 22 genes; 3 showed a sustained increase (cluster i), 5 were increased only after Withdrawal (cluster ii), 5 exhibited an increase only after Cocaine (cluster iii) and 1 showed a sustained decrease in expression (cluster v). In contrast, the Pcdhγ cluster of 23 genes exhibited a more homogeneous expression pattern in which expression of all 21 detected isoforms showed a sustained decrease (cluster v).

Table 3. Clustered genes from the cadherin/Wnt and heterotrimeric G-protein signaling pathways
 (i) Sustained increase, Coc and W/D(ii) Increase, W/D only(iii) Transient increase, Coc only(iv) Transient decrease, Coc only(v) Sustained decrease, Coc and W/D
  1. The cadherin and protocadherin genes appearing in clusters i–v are identified. Genes in the heterotrimeric G-protein signaling pathway (Gαi and Gαs) that fell into clusters i–v are grouped by function; GPCRs for dopamine, GABA, glutamate and acetylcholine are discussed separately and GPCRs without known ligands are not listed. Intracellular signaling proteins are grouped by function.

  2. Coc, Cocaine; W/D, Withdrawal.

Cadherin/Wnt signaling
CadherinsCdh4, Cdh5, Cdh6, Cdh8, Cdh11Cdh7, Cdh19, Cdh22Cdh24Cdh13 
ProtocadherinsPcdh1, Pcdh10, Pcdh18, Pcdhb13, Pcdhb14, Pcdhb16Pcdhb2, Pcdhb6, Pcdhb7, Pcdhb8, Pcdhb9Pcdh20, Pcdhb5, Pcdhb10, Pcdhb11, Pcdhb19, Pcdhb20 Pcdha4, Pcdhb12, Pcdhga1, Pcdhga2, Pcdhga3, Pcdhga4, Pcdhga5, Pcdhga6, Pcdhga7, Pcdhga8, Pcdhga9, Pcdhga10, Pcdhga11, Pcdhga12, Pcdhgb1, Pcdhgb2, Pcdhgb4, Pcdhgb5, Pcdhgb6, Pcdhgb7, Pcdhgb8, Pcdhgc3, Pcdhgc4, Pcdhgc5
Heterotrimeric G-protein signaling (Gαi and Gαs)
Type of GPCR
Neuropeptide bindingFzd5, Fzd8, Hcrtr2, Npffr1, Oprk1, Oprm1, Prlhr, Tacr1Fzd6, Fzd7, Mc3r, Oprd1, Pth1r, Smo, Trhr2Hctr1, Npbwr1, Npy2r, Oxtr, Sstr3, Tacr3Fzd9, Gipr, Nmbr, Npy5rMc4r, Mchr1, Mtnr1a, Sstr2
Lipid binding S1pr3 S1pr4Lpar1
Cadherin typeCelsr2, Celsr3Celsr1   
Intracellular signaling proteins
GPCR kinases   Grk6Grk1
ArrestinsArrb1   Arrb2
Gβ subunits Gnb4  Gnb1, Gnb1l, Gnb2, Gnb2l1, Gnb5
Gγ subunits Gng7Gng2 Gng3, Gng8, Gng10, Gng11, Gng13
RGSsRgs6Rgs8, Rgs11Rgs17Rgs9, Rgs19Rgs2, Rgs4, Rgs10, Rgs16, Rgs20
Adenylate cycasesAdcy1, Adcy4, Adcy6, Adcy9Adcy2, Adcy5Adcy8  
Phosphorylase b kinasesPhka2   Phkg1, Phkg2
CREBsCreb3l2, Crebbp   Creb3

We used an antibody to protocadherin 10 (PCDH10), which is known to play a role in the development of striatal axons and thalamocortical projections (Uemura et al. 2007), to evaluate its subcellular localization and quantify its levels in postsynaptic densities (PSDs) purified from the striata of Saline- and Cocaine-treated mice. PCDH10 was enriched in PSD preparations (Fig. 2b); furthermore, PSDs purified from striatal extracts of Cocaine-treated mice contained more PCDH10 than striatal PSDs from Saline controls. A role for extracellular matrix proteins, including cadherins and protocadherins, is consistent with previous studies of Cocaine-regulated gene expression in mice (Eipper-Mains et al. 2011) and postmortem human brain (Lull et al. 2008; Mash et al. 2007).

Consistent with the literature, heterotrimeric G-protein signaling (Gαi and Gαs-mediated) was identified as a major pathway targeted by Cocaine. In addition to G-protein-coupled receptors (GPCRs) for the major neurotransmitters (discussed below), neuropeptide (opioid, galanin and somatostatin), lipid and cadherin-type GPCRs showed Cocaine-related changes in expression (Table 3).

Panther Pathway analysis identified Cocaine-regulated changes in the repertoire of intracellular signaling proteins downstream of the regulated Gαi and Gαs-coupled GPCRs (Table 3 and Fig. 3). The diagram in Fig. 3 depicts the complex interplay of intracellular signaling proteins. For example, only two members of the Mus arrestin gene family are expressed outside of the retina (Kovacs et al. 2009); Arrb1 showed a sustained increase in expression (cluster i) while Arrb2 exhibited a sustained decrease (cluster v). Transcripts encoding the β- and γ-subunits of heterotrimeric G-proteins were primarily identified in cluster v (sustained decrease). In contrast, Regulators of G-protein signaling (RGS genes) appeared in all clusters. Transcripts encoding four adenylate cyclase (Adcy) genes showed a sustained increase in expression (cluster i) while the transcripts encoding two additional Adcy genes showed increased expression only in the Withdrawal group (cluster ii).

Figure 3.

i and Gαs heterotrimeric G-protein signaling schematic. Model based on Panther Pathway diagram (http://www.pantherdb.org). Venn diagram depicts color key for signaling components. Genes labeled yellow showed a sustained increase in Coc and W/D (cluster i), red genes increased in W/D only (cluster ii) and blue genes had decreased expression in both Coc and W/D (cluster v); overlap colors indicate contribution of >1 cluster to signaling group. For example, GPCRs, which appear in all three clusters, are shown in magenta. Genes from clusters i, ii and v were used for this analysis. Coc, Cocaine; W/D, Withdrawal.

Panther Protein Class analysis (Table S4) identified Cytoskeletal Proteins, Transcription Factors and Transferases as the three Protein Classes enriched in genes exhibiting a sustained response (clusters i, ii and v; P < 0.0001). Sustained up-regulation of G-protein modulators and guanyl-nucleotide exchange factors (cluster i) was also observed. Several Protein Classes involved in interactions with Ca2+ (calcium-binding proteins, calmodulin and intracellular calcium-sensing proteins) were enriched only in cluster v (sustained down-regulation). GPCRs, ligand-gated ion channels, membrane trafficking proteins and neuropeptides were also enriched in cluster v.

Widespread effects of Cocaine and Withdrawal on catecholaminergic, GABAergic, cholinergic, glutamatergic and endocannabinoid pathways

Dopamine, norepinephrine and serotonin receptors expressed in the NAc play a critical role in addiction. Transcripts encoding the dopamine receptors, Drd1a, Drd2, Drd3 and Drd5, exhibited different patterns of response to Cocaine and Withdrawal (Fig. 4). Drd1a was known to decrease in response to Cocaine treatment, while Drd2 was known to increase (Hoffmann et al. 2012). Adra1a, Adra2a and Adrb1 adrenergic receptor expression rose in response to Cocaine treatment, while expression of Adra1b, Adra2b and Adra2c fell. Six of the 12 serotonin receptors (Htr1b, Htr2a, Htr2c, Htr4, Htr5a and Htr6) exhibited increased expression in response to Cocaine and/or Withdrawal. The long-term changes during Withdrawal were verified by qPCR for Drd1a, Drd2, Adra1a, Adra2b, Htr1d, Htr2a, Htr2c and Ppp1r1b (DARPP32).

Figure 4.

Regulation of receptors for biogenic amines. RPKM data are shown for dopamine, adrenergic and serotonin receptor transcripts (RPKM > 1) in the Saline library. Z-scores were calculated as described in Materials and methods. Transcripts included in this group but expressed at an RPKM < 1 are listed to the side. Ppp1r1b (DARPP32), a multiply phosphorylated phosphatase inhibitor highly expressed in MSNs, is included for comparison.

Over 90% of the neurons in the NAc are GABAergic medium spiny interneurons (Kemp & Powell 1971); the genes involved in synthesizing, storing and retrieving GABA (γ-aminobutyric acid) are expressed in the NAc, along with ionotropic and metabotropic GABA receptors (Fig. 5). Expression of the GABA-synthesizing enzymes (Gad65 and Gad67), the vesicular inhibitory amino acid transmitter transporter (Viatt) and the plasma membrane transporters (Gat1, Gat3 and Gat4) rose in response to Cocaine and Withdrawal. The three Na+/Cl-dependent GABA transporters are expressed by astrocytes and retrieve GABA after secretion (Beenhakker & Huguenard 2010; Bolteus & Bordey 2004; Schousboe et al. 2004); neurons also express Gat1 (Schousboe et al. 2011). Levels of Gat4 were highest after Cocaine while levels of Gat1 were highest in Withdrawal. Multiple GABAA receptor subunits are expressed at similar levels; while some showed a sustained increase in expression (Gabrb2, Gabrb3, Gabrg3 and Gabrq), Gabrb1 expression decreased in Withdrawal and Gabrd expression decreased after Cocaine but recovered in Withdrawal. Expression of GABAB receptor subunits (Gabbr1 and Gabbr2) increased after Cocaine and increased further during Withdrawal.

Figure 5.

Regulation of genes affecting GABAergic transmission. Data were analyzed as described for Fig. 4. The transcripts encoding glutamic acid decarboxylase 65 and 67 (Gad1 and Gad2) are expressed at similar levels. GABA transport into synaptic vesicles requires Slc36a1; retrieval of GABA from extracellular space requires Slc6a1, Slc6a13 or Slca11. Multiple ionotropic GABAA receptor subunits are expressed in the NAc; expression of metabotropic GABAB receptor Gabbr1 greatly exceeds expression of Gabbr2.

Glutamatergic signaling plays an essential role in drug abuse, with major glutamatergic inputs onto NAc MSNs coming from amygdala, hippocampus and prefrontal cortex (Kalivas 2009; Wolf 2010). The effects of Cocaine on expression of ionotropic and metabotropic glutamate receptors have been studied in detail (Kalivas 2009; Kalivas & Volkow 2011; Kalivas et al. 2009) and were largely confirmed in our analysis (Fig. 6). Glutamatergic inputs acutely excite NAc MSNs primarily via activation of 2-amino-3-(3-hydroxy-5-methyl-isoxazol-4-yl)propanoic acid (AMPA) receptors, a step considered necessary for initiation of drug-seeking behavior (Wolf & Ferrario 2010). AMPA receptor expression showed only small changes during Cocaine administration; Gria1 increased substantially while Gria3 decreased during Withdrawal (confirmed by qPCR). Expression of several N-methyl-D----aspartate (NMDA) receptor subunits increased during Cocaine administration (Grin1, Grin2a, Grin2b, Grin2d and Grin3a) (several confirmed by qPCR). Expression of kainate receptors Grik1, Grik2 and Grik3 showed a sustained increase while expression of Grik4 and Grik5 showed a sustained decrease. Expression of the five most highly expressed metabotropic receptors for glutamate (Grm1, Grm3, Grm4, Grm5 and Grm7) showed a sustained increase. Vesicular glutamate transporters (Vglut) can be expressed in GABAergic and cholinergic neurons as well as in glial cells (El Mestikawy et al. 2011). Transcripts encoding all three Vgluts were detectable in the NAc, which contains no primarily glutamatergic neurons; expression of Vglut1 mRNA exceeded that of Vglut2. Na+-dependent plasma membrane glutamate transporters (Eaat1 through Eaat5), which remove glutamate from the synaptic cleft and perisynaptic area, also play a critical role in glutamatergic transmission. Three of the four Eaats expressed in the NAc (Eaat1, Eaat2 and Eaat3) exhibited a sustained increase in expression while Eaat4 exhibited reduced expression in Withdrawal. Eaat1 and Eaat2 are primarily expressed in glia, while Eaat3, Eaat4 and Eaat5 show widespread neuronal expression throughout the brain (Tzingounis & Wadiche 2007).

Figure 6.

Regulation of genes affecting glutamatergic transmission. Data were analyzed as described for Fig. 4. Expression of all three vesicular glutamate transporters (Vglut) and four of the plasma membrane glutamate transporters (Slc1a1, Slc1a2, Slc1a3 and Slc1a6) was above the cutoff level. Multiple subunits of the ionotropic glutamate receptors (AMPA, NMDA and Kainate) and multiple metabotropic glutamate receptors are expressed at similar levels.

A small population of giant aspiny cholinergic interneurons is found in the NAc (Threlfell & Cragg 2011); expression of both the plasma membrane choline transporter (Cht1), which provides the rate-limiting substrate, and choline acetyltransferase (ChAT) was increased by Cocaine and further increased by Withdrawal (Fig. 7). Expression of each of the major nicotinic ACh receptor subunits found in the NAc (Chrna4, Chrna7 and Chrnb2) and three muscarinic ACh receptors (Chrm2, Chrm3 and Chrm5) showed sustained increases, many verified by qPCR.

Figure 7.

Regulation of genes affecting cholinergic transmission. Data were analyzed as described for Fig. 4. Although only a small percentage of the neurons in the NAc are cholinergic, choline acetyltransferase (ChAT), high affinity plasma membrane choline transporter (Slc5a7) and vesicular acetylcholine transporter (Slc18a3) transcript levels all fall above the 1 RPKM cutoff. Only three of the many nicotinic acetylcholine receptor subunits are expressed in the NAc while all five muscarinic receptors are expressed. Expression of each of the genes involved in cholinergic transmission is elevated in Withdrawal.

The endocannabinoid system is heavily involved in many reward-seeking responses to drugs of abuse (Orio et al. 2009). Levels of transcripts encoding Cnr1 (the CB1 endocannabinoid receptor) and the cannabinoid synthetic enzymes Dagla and Napepld showed sustained increases in Cocaine and Withdrawal, while Daglb showed a sustained decrease (Fig. 8). The effect of Cocaine expression on each of these genes was confirmed by qPCR analysis of two separate sets of animals.

Figure 8.

Regulation of genes affecting endocannabinoid signaling. RPKM data were analyzed as described for Fig. 4. While transcripts encoding the two enzymes that synthesize the endocannabinoid 2-arachidonoylglycerol (diacylglycerol lipase α and β; Dagla and Daglb) are expressed at similar levels in the NAc, they respond to Cocaine in an opposite manner. Expression of transcripts encoding cannabinoid receptor 1 (Cnr1) greatly exceeds that of cannabinoid receptor 2 (Cnr2) and is responsive to Cocaine.


Single-gene, proteomic and microarray analyses showed that Cocaine produces widespread, multi-faceted responses in the NAc (Eipper-Mains et al. 2011; McClung & Nestler 2008). Next-generation sequencing identified pathways not previously thought to play any role in addiction and facilitated systematic cataloging of changes in all components of known signaling pathways. With greater sequencing depth, changes in editing, splicing, imprinting and allele-specific expression can be quantified (Eipper-Mains et al. 2011; Metzker 2010; Mortazavi et al. 2008; Wahlstedt et al. 2009; Wang et al. 2008a,b; 2008bWang et al. 2009). In situ hybridization and immunostaining will be required to localize changes in gene expression to specific cell types in the core and shell regions of the NAc. Components of the Wnt, cadherin and heterotrimeric G-protein signaling pathways were over-represented among Cocaine-responsive transcripts; cross-talk between these signaling cascades is extensive (Force et al. 2007).

Global comparisons of gene expression identify signaling pathways targeted by Cocaine

Cadherin/Wnt pathways

By sequencing PSD-localized, Cocaine-regulated striatal microRNAs, we previously identified miR-8 family members as key players in the response to Cocaine (Eipper-Mains et al. 2011). Although the miR-8 family has primarily been studied for its role in cancer progression and metastasis, we verified Cocaine regulation of two bioinformatically predicted cadherin family targets (Eipper-Mains et al. 2011). We identified over 40 Cocaine-regulated cadherin and protocadherin transcripts; genes from the Pcdhβ cluster were primarily up-regulated following Cocaine in both a transient and sustained fashion (clusters i–iii), whereas genes from the Pcdhγ cluster were exclusively found to undergo a sustained down-regulation following Cocaine exposure (cluster v).

A few studies have suggested a role for cell adhesion and extracellular matrix proteins in the response to Cocaine. Disturbances in cortical cytoarchitecture in animal models of prenatal Cocaine exposure resembled changes observed when Wnt/cadherin pathway function was altered; this lead to the prediction that these pathways play a role in the response to Cocaine (Novikova et al. 2005). As observed here (cluster i), Cdh6 and Cdh11 expression increased in the frontal cortex of E18 pups taken from dams treated with Cocaine for 10 days (Novikova et al. 2005). Also contributing to alterations in the extracellular milieu is Cocaine-mediated down-regulation of glycosyltransferase expression; these enzymes are crucial for the synthesis of cell surface glycoproteins. In a compilation documenting changes in gene expression in postmortem tissue from Cocaine addicts, cell adhesion molecules were identified as major targets (Lehrmann et al. 2003; Mash et al. 2007).

Heterotrimeric G-protein pathways

Transcripts encoding GPCRs, heterotrimeric G-protein subunits, regulators of G-protein signaling (RGS), GPCR kinases (GRK) and adenylate cyclases (Adcy) responded differently to Cocaine and Withdrawal. Genes encoding GPCRs and RGSs were identified in all five clusters. Particularly striking was the prevalence of GPCRs for neuropeptides, with different receptor subtypes in different clusters (e.g. Wnt, galanin, somatostatin and opioid receptors). The μ-opioid (Oprm1) receptor, associated with increased sensations of reward and pleasure, showed a sustained increase in expression in response to Cocaine, as expected (Di Chiara & Imperato 1988; Wee & Koob 2010), as did the endocannabinoid (Cnr1) receptor (Adamczyk et al. 2012a,b). The δ-opioid receptor (Oprd1), also involved in positive reinforcement of drugs of abuse (Le Merrer et al. 2009), increased expression only in Withdrawal. The κ-opioid receptor (Oprk1), crucial to the negative emotional state of Withdrawal (Wee & Koob 2010), showed a sustained increase, as also observed in postmortem tissue from Cocaine overdose victims (Mash & Staley 1999). Transcripts encoding several neuropeptides were substantially down-regulated by Cocaine: Pomc (proopiomelanocortin), Tac1 (tachykinin), Oxt (oxytocin) and Cartpt (CART; Cocaine and amphetamine regulated peptide).

Consistent with a major role for cadherins in the response to Cocaine, seven transmembrane cadherins (Celsr) were identified in clusters i and ii. Genes encoding Adcy appeared in clusters i, ii and iii. On the other hand, the majority of genes encoding the β- and γ-subunits of the heterotrimeric G-proteins fell into cluster v (sustained decrease). The few genes whose expression is uniquely regulated in Withdrawal are of special interest; these transcripts may provide insight into the biochemical and cell biological alterations underlying reinstatement of drug-seeking behavior, ultimately helping identify potential therapeutic approaches to prevent relapse (Wolf 2010).

Neurotransmitter-specific effects

Expression of the Drd3 increased in response to Cocaine and remained elevated after Withdrawal (cluster i); in contrast, D2 dopamine receptor (Drd2) expression increased only after Withdrawal. An increase in D3 receptor expression was observed in the NAc of Cocaine overdose victims (Mash & Staley 1999) and in rats 45 days after Cocaine self-administration ceased (Conrad et al. 2010). Recent work suggests that dopamine receptor dysregulation, mediated by decreased D2 and increased D3 receptor expression, contributes to increased Cocaine-seeking behavior after prolonged Withdrawal (Conrad et al. 2010). Alterations in serotonin signaling are also recognized as important in drug addiction (Dreyer 2010), and we observed Cocaine-responsive changes in gene expression across many of the serotonin receptors.

Expression of many of the genes responsible for GABA synthesis, transport into synaptic vesicles, and retrieval from the synaptic cleft rose following Cocaine treatment and increased further during Withdrawal. Daily Cocaine injections decreased GABAA receptor-mediated maximal evoked currents and miniature inhibitory postsynaptic currents on dopamine neurons in the ventral tegmental area (VTA) (Liu et al. 2005); furthermore, GABAA receptor function has been implicated in dopamine-mediated alcohol reward (Heilig et al. 2011). GABAB receptor agonists attenuated Cocaine reinforcement and reduced Cocaine craving in a study of human Cocaine users (Brebner et al. 2002); while the precise mechanism of these behavioral effects is unknown, activation of GABAB receptors in the NAc reduces dopamine release from dopaminergic projections from the VTA (Erhardt et al. 2002). Changes in GABAA receptor subunit expression, which fell during Withdrawal, and in GABAB receptor expression, which rose, could contribute significantly to altered synaptic transmission.

A growing body of literature has focused on the role of AMPA, NMDA and kainate receptor-mediated glutamatergic transmission into the NAc in the development of addiction and addictive-like behaviors (Kalivas et al. 2009). Increased glutamate receptor mRNA expression parallels increases in cell surface expression of GluR1-containing AMPA receptors following Cocaine Withdrawal (Boudreau et al. 2007), along with a specific population of GluR2-lacking calcium-permeable AMPA receptors (Wolf 2010). We saw sustained up-regulation of Gria1 (GluR1) transcripts, a transient increase in Gria4 (GluR4) transcripts, and no change in Gria2 (GluR2) transcripts. While expression of kainate receptor 3 (Grik3) rose substantially, expression of Grik4 and Grik5 fell. NMDA receptors are critical for the behavioral response to drugs of abuse (Lee et al. 2010; Pascoli et al. 2011); new spines appearing in the NAc following Cocaine are enriched in NMDA receptors and are not formed if NMDA receptor currents are blocked (Huang et al. 2009; Ren et al. 2010). Correspondingly, transcripts encoding several NMDA receptor subunits rose in response to Cocaine. Expression of the more prevalent metabotropic glutamate receptors also increased. Perhaps the most significant finding is that expression of Slc17a7 and Slc17a6 (Vglut1 and Vglut2) exhibited sustained down-regulation after Cocaine treatment and Withdrawal.

Cholinergic transmission in the NAc contributes to addiction through the modulation of dopamine signaling and reward processing (Mark et al. 2011). Increased expression of both choline acetyltransferase and the plasma membrane choline transporter in the giant aspiny cholinergic interneurons could increase cholinergic signaling at both nicotinic and muscarinic ACh receptors expressed on the GABAergic output neurons.


We used next-generation sequencing to generate a comprehensive catalog of transcriptional changes that occur in response to Cocaine treatment and Withdrawal in the primary reward center of the brain. Pathway analysis enabled the complete categorization of transcriptional changes in signaling pathways known to be important in addiction along with the identification of pathways not previously considered important. These data should serve as a resource for development of new targets for Cocaine therapy and facilitate selection of protein interaction and pathway targets for future study. With more information about their normal physiological roles, the extracellular domains of the many cadherins and protocadherins that respond to Cocaine may be useful therapeutic targets. Pharmaceutical cocktails targeted to a subset of the specific GPCRs whose expression is altered during Withdrawal may prove to be more effective than treatments targeting a single pathway.


We thank Darlene D'Amato for indefatigable laboratory support. This work was supported by National Institutes of Health Grants DA-15464 and DA-23082 (to B. A. E. and R. E. M.), DA-26706 (to D. D. K.) and GM-62516 (to B. R. G.). None of the authors have any conflicts of interest.