Active and passive MDMA (‘ecstasy’) intake induces differential transcriptional changes in the mouse brain

Authors

  • N. Fernàndez-Castillo,

    1. Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    2. The Biomedical Network Research Centre on Rare Diseases (CIBERER), Barcelona, Spain
    3. Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain
    Search for more papers by this author
    • These authors contributed equally to this work.

  • M. J. Orejarena,

    1. Laboratori de Neurofarmacologia, Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, PRBB, Barcelona, Spain
    Search for more papers by this author
    • These authors contributed equally to this work.

  • M. Ribasés,

    1. Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain
    2. Psychiatric Genetics Unit, Hospital Universitari Vall d’Hebron, Barcelona, Spain
    Search for more papers by this author
  • E. Blanco,

    1. Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    2. Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain
    Search for more papers by this author
  • M. Casas,

    1. Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain
    2. Department of Psychiatry and Legal Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain
    Search for more papers by this author
  • P. Robledo,

    1. Laboratori de Neurofarmacologia, Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, PRBB, Barcelona, Spain
    2. Neuropsychopharmacology Program, IMIM (Hospital del Mar Research Institute), PRBB, Barcelona, Spain
    Search for more papers by this author
  • R. Maldonado,

    1. Laboratori de Neurofarmacologia, Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, PRBB, Barcelona, Spain
    Search for more papers by this author
    • These seniors authors contributed equally to this work.

  • B. Cormand

    Corresponding author
    1. Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    2. The Biomedical Network Research Centre on Rare Diseases (CIBERER), Barcelona, Spain
    3. Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain
    Search for more papers by this author
    • These seniors authors contributed equally to this work.


B. Cormand, PhD, Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Av. Diagonal 643, 08028 Barcelona, Spain. E-mail: bcormand@ub.edu

Abstract

3,4-Methylenedioxymethamphetamine (MDMA, ‘ecstasy’) is a recreational drug widely used by adolescents and young adults. Although its rewarding effects are well established, there is controversy on its addictive potential. We aimed to compare the consequences of active and passive MDMA administration on gene expression in the mouse brain since all previous studies were based on passive MDMA administration. We used a yoked-control operant intravenous self-administration paradigm combined with microarray technology. Transcriptomic profiles of ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus were analysed in mice divided in contingent MDMA, yoked MDMA and yoked saline groups, and several changes were validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The comparison of contingent MDMA and yoked MDMA vs. yoked saline mice allowed the identification of differential expression in several genes, most of them with immunological and inflammatory functions, but others being involved in neuroadaptation. In the comparison of contingent MDMA vs. yoked MDMA administration, hippocampus and the dorsal raphe nucleus showed statistically significant changes. The altered expression of several genes involved in neuroadaptative changes and synapse function, which may be related to learning self-administration behaviour, could be validated in these two brain structures. In conclusion, our study shows a strong effect of MDMA administration on the expression of immunological and inflammatory genes in all the four brain regions studied. In addition, experiments on MDMA self-administration suggest that the dorsal raphe nucleus and hippocampus may be involved in active MDMA-seeking behaviour, and show specific alterations on gene expression that support the addictive potential of this drug.

3,4-Methylenedioxymethamphetamine (MDMA, ‘ecstasy’) is a recreational drug used around the world by young adults. MDMA induces euphoria and a ‘feeling’ of well-being in humans (Parrott 2001), and its rewarding/reinforcing effects have been well established in animal models (Cole & Sumnall 2003). Although the addictive potential of this substance is still a matter of debate, there is evidence showing that a proportion of MDMA users meet the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria for dependence (Cottler et al. 2001; Leung & Cottler 2008; Stone et al. 2006). MDMA acutely increases brain levels of dopamine (DA), serotonin (5-HT) and noradrenalin in monkeys, rats and mice by potently inhibiting neurotransmitter reuptake mechanisms (Green et al. 2003). Repeated administration of MDMA in humans produces long-term psychiatric disorders, including anxiety and mood alterations, as well as cognitive deficits (Zakzanis et al. 2007), which may be associated with persistent neuroadaptations dependent on changes in gene expression.

Single or repeated administration of MDMA in animals induces changes in gene expression similar to what has been observed following treatment with other psychostimulants such as cocaine, amphetamine or methamphetamine (Hemby 2006; Yuferov et al. 2005; Zhang et al. 2005). Acute administration has been reported to dose-dependently increase the expression of several immediate early genes, such as c-fos and Egr1 in different brain structures (Shirayama et al. 2000; Stephenson et al. 1999). Similarly, the Rnd3 gene involved in actin cytoskeleton modulation and cell adhesion was upregulated in the striatum of mice after acute MDMA administration (Marie-Claire et al. 2007). Repeated treatment with MDMA increased DeltaFosB expression in mice (Olausson et al. 2006) and induced pronounced alterations in gene expression of glutamate transporters as well as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), N-methyl d-aspartate (NMDA) and metabotropic glutamate receptor subunits in different brain regions in rats (Kindlundh-Hogberg et al. 2008). In addition, changes in pro-dynorphin (Pdyn) and pro-enkephalin (Penk) gene expression have been observed in several brain areas of rats treated either acutely or repeatedly with MDMA (Adams et al. 2005; Di Benedetto et al. 2006). Using microarray technology, alterations in the expression of numerous genes involved in the modulation of signalling pathways, transcription regulators or xenobiotic metabolism have been shown in the frontal cortex of rats following a single MDMA administration (Thiriet et al. 2002). Although these data provide evidence for the effects of non-contingent administration of MDMA on gene expression in the brain, there are no studies available using models of MDMA operant self-administration, which are more relevant to the human pattern of drug consumption. In this sense, the use of a yoked-control operant intravenous self-administration paradigm coupled with microarray studies has shown different profiles of gene transcript alterations in the nucleus accumbens shell and core comparing contingent vs. non-contingent heroin and cocaine administration (Jacobs et al. 2004, 2005), which suggests that the learning component associated with active drug-taking is a critical factor affecting changes in gene transcription.

This study was designed to identify changes in gene expression in different brain structures (ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus) in mice receiving repeated contingent or yoked administration of MDMA in order to better understand the consequences of MDMA consumption and seeking behaviour. These structures are known to mediate different aspects of drug reward and instrumental conditioning (Belin et al. 2009; Ikemoto 2010), participate in the neurochemical and behavioural effects of MDMA (Cole & Sumnall 2003) and show gene expression changes following acute (Stephenson et al. 1999; Thiriet et al. 2002) and repeated (Kindlundh-Hogberg et al. 2008; Olausson et al. 2006) non-contingent MDMA administration.

Materials and methods

Animals

Male C57Bl/6J mice weighing 20–24 g at the beginning of the experiments were initially housed five per cage in a room with controlled temperature (21 ± 1°C) and humidity (65 ± 10%), with a reversed light/dark cycle (lights off from 08:00 to 20:00 h), and with ad libitum food and water. The experiments took place during the dark phase. Behavioural tests and animal care were conducted in accordance with the standard ethical guidelines (National Institutes of Health, 1995; European Communities Directive 86/609 EEC) and approved by the local ethical committee (CEEA-PRBB).

Drugs

MDMA hydrochloride was obtained from Lipomed, A.G. (Arlesheim, Switzerland) and dissolved in sterile 0.9% physiological saline solution.

Surgery and self-administration procedure

Mice were anaesthetized with an intraperitoneal (i.p.) injection of a ketamine/xylazine mixture (5:1; 0.10 ml/10 g) and then implanted with an indwelling intravenous (i.v.) silastic catheter in the right jugular vein, as previously described (Orejarena et al. 2009). The animals were pre-treated with ketoprofen 5 mg/kg subcutaneously (s.c.) for post-surgery analgesia. After surgery, the mice were housed individually for the remainder of the experiments. In order to avoid clots and infection, the animals were flushed through the catheter with 0.02 ml of a solution containing heparin (30 UI/ml), cefazoline (50 mg/ml) and sodium chloride (0.09%) for 5 days.

The patency of the catheters was evaluated once a week by the injection of 0.1 ml of thiopental (5 mg/ml). If prominent signs of anaesthesia were not apparent within 3 seconds of the infusion, the mouse and its corresponding data were removed from the experiment. Three days after surgery, the animals were randomly assigned to either contingent or yoked groups. Contingent mice were trained to self-administer MDMA (0.125 mg/kg per infusion delivered in a volume of 23.5 µl over 2 seconds) in single daily 3-h sessions. Acquisition of drug self-administration was performed using a fixed ratio 1 (FR1) schedule of reinforcement such that one nose poke in the active hole resulted in one MDMA infusion, while nose poking in the inactive hole had no programmed consequences. As previously reported (Orejarena et al. 2009), mice had to achieve all of the following conditions to be included in the analysis: (1) less than 20% deviation from the mean of the total number of infusions earned in three consecutive sessions (80% stability), (2) at least 65% responses at the active hole and (3) a minimum of five infusions earned per session. Each contingent mouse was connected to two yoked mice; one receiving an identical dose of MDMA (yoked MDMA) and the other a saline solution (yoked saline). When a contingent mouse had a failed catheter or did not meet the acquisition criteria, the corresponding yoked mice were discarded from the study. A light stimulus, located above the active hole, was paired with the delivery of the drug or saline according to the response of the contingent mouse. To avoid interference by acute transcriptional changes, animals were killed by cervical dislocation 8 h after the last exposure to the self-administration boxes. The brains were quickly removed, and the following brain areas were dissected according to Franklin and Paxinos (1997): ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus. Brain tissues were then frozen by immersion in 2-methylbutane surrounded by dry ice, and stored at −80°C for later quantification of gene expression.

RNA isolation and microarray hybridization

Twenty-seven mice (nine animals per group of contingent MDMA, yoked MDMA and yoked saline) and four brain areas (ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus) were used in the expression microarray study. Three pools consisting of three mice per pool were used for each experimental group. The pools were organized to homogenize the average number of nose pokes in the different pools. The pooled individuals were the same for all brain regions. The frontal cortex and dorsal raphe nuclei tissue samples belonging to the same pool were pooled before RNA extraction to optimize the yielding of the isolation, given the limited tissue size. In contrast, for the hippocampus and ventral striatum, which are larger brain structures, RNA was isolated separately from each animal and then pooled for the array hybridization. Samples from all tissues were homogenized using the TissueRuptor system (Qiagen, Düsseldorf, Germany), and total RNA was isolated using the RNeasy Lipid Tissue Mini Kit (Qiagen) according to the manufacturer's protocol. RNA concentration was determined using the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), and integrity was evaluated using the Bioanalyzer 2100 platform (Agilent Technologies, Santa Clara, CA, USA). RNA samples were stored at −80°C until analysed. For the microarray experiment, we used the GeneChip® Mouse Expression Set 430 array (Affymetrix, Santa Clara, CA, USA), which contains probes that cover more than 39 000 transcripts and variants from more than 34 000 genes. A total of 36 chips were used: three pools of three individuals per condition (contingent, MDMA, yoked MDMA and yoked saline) and four brain areas. Two micrograms of RNA from each pool were used to hybridize arrays at the Genomics Unit of Hospital Clínic-IDIBAPS (Barcelona, Spain). Chips were scanned using a GenePix4000B scanner (Molecular Devices, Inc, Sunnyvale, CA, USA), and raw data were obtained using the GenePix Pro 4.0 (Molecular Devices) and GCOS softwares (Affymetrix).

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

To confirm expression changes in genes selected on the basis of their function, total RNA from the four brain regions of contingent MDMA, yoked MDMA and yoked saline mice was reverse-transcribed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster city, CA, USA). For this experiment, we used the same pools that were previously subjected to hybridization with the arrays. The only exception was hippocampus in the comparison of contingent vs. yoked MDMA mice, where the validations were performed in samples from separate subjects (nine animals instead of three pools of three animals) in order to increase the chances of validation of small gene expression changes. This brain structure is large enough to apply this approach. The Mouse Endogenous Control Array (Applied Biosystems) was used to select endogenous controls. Real time-PCR experiments were performed for 22 genes using the LightCycler 480 II system and the Universal Probe Library (Roche Applied Science, Penzberg, Germany). Gene assays were designed using the Universal ProbeLibrary Assay Design Center software (Roche Applied Science, www.roche-applied-science.com). Sequence of the primers and probes used are available upon request. Beta-actin (Actb) and glyceraldehyde-3-phosphate dehydrogenase (Gapdh) were used to normalize the relative amounts of mRNA.

Statistical and bioinformatic analyses

The self-administration behavioural data were analysed using a three-way repeated measures analysis of variance (anova) with group as a between-subject factor and hole and day as within-subjects factors followed by post hoc tests for individual comparisons when appropriate. Statistical significance was set at P < 0.05.

For the microrray data, we used the Bioconductor software for R environment and the affy library (www.bioconductor.org) (Gentleman et al. 2004). The quality assessment of the chips was performed using the affyPLM library. Background correction, normalization and summarization were performed using the background method, Robust Multichip Average (RMA) (Irizarry et al. 2003) and the median-polish method, respectively. For gene filtering, we discarded those probes that did not correspond to known genes and considered a threshold of log2(60) for signal filtering and an interquartile range (IQR) >25% for variability filtering. The IQR method discards genes showing low expression variance among arrays without considering the comparisons performed, and is described to increase statistical power (Hackstadt & Hess 2009). The Linear Modeling for Microarray Analysis (LIMMA) package (Smyth 2004) was used for class comparison, by which we compared the expression patterns of the pairs contingent MDMA–yoked MDMA, contingent MDMA–yoked saline and yoked MDMA–yoked saline. Correction for multiple testing was achieved by adjusting the P-value with a false discovery rate (FDR) of 5%. To evaluate drug reinforced learning, FDR threshold was initially set at 5%, which allowed the identification of differentially expressed genes that showed low fold changes (absolute average fold change = 1.34), in contrast with the figures obtained in genes that were differentially expressed as a consequence of the direct effect of the drug (absolute average fold change = 2.51). Table S1 shows the proportion of differentially expressed genes identified across each range of fold change. Thus, we increased the FDR threshold up to 15% in the comparisons of contingent MDMA–yoked MDMA and contingent MDMA–yoked saline, identifying changes in an additional brain structure, dorsal raphe nucleus and obtaining a significant increase in the average fold change in this structure (1.66). Functional group over-representation analysis of genes with significant differential expression was performed using the DAVID Annotation Tool (david.abcc.ncifcrf.gov) (Dennis et al. 2003) considering GO biological processes (NFAT) and was supported by literature searches.

Gene expression networks were constructed using the Ingenuity Pathway Analysis 8.8 software (Ingenuity Systems, Redwood city, CA, USA). This software estimates a score, calculated with the right-tailed Fisher's exact test, based on the probability of finding the observed number of differentially expressed genes in a given network by chance [score = −log (Fisher exact test probability)].

The identification of over-represented transcription factor binding sites (TFBSs) in the different sets of differentially expressed genes was performed using Single Site Analysis with the oPOSSUM 2.0 software (www.cisreg.ca/cgi-bin/oPOSSUM/opossum) (Ho Sui et al. 2005), using the default parameters of the vertebrate Jaspar Core profile, and sorting the top 20 results by Z-score.

KEGG pathways enrichment analyses as well as Gene Ontology (GO), Cytogenetic band and microRNA targets enrichment analyses were performed using the WebGESTALT software (bioinfo.vanderbilt.edu/webgestalt).

In the quantitative reverse transcription polymerase chain reaction (qRT-PCR) experiments, gene expression changes for each comparison were evaluated using a U-Mann–Whitney non-parametric test, and statistical significance was set at P < 0.05.

Results

MDMA self-administration

The average number of active and inactive nose pokes carried out by contingent mice trained to self-administer MDMA (0.125 mg/kg per infusion) as well as for yoked MDMA and yoked saline mice is shown in Fig. 1. Seventy percent of the contingent mice met all the acquisition criteria within a short time period (8 ± 0.76 days), and showed a mean cumulative intake of 19.7 ± 1.62 mg/kg of MDMA during the entire training period. Saline- or MDMA-yoked animals did not discriminate between holes on any of the training sessions. Eleven training sessions were performed until all contingent mice reliably acquired MDMA self-administration behaviour. Three-way repeated measures anova comparing responses in the active and inactive holes for all groups during the entire testing period showed a significant main effect of group (F2,24 = 80.600,P < 0.001), a significant main effect of hole (F1,24 = 77.770,P < 0.001), a significant group × hole interaction (F2,24 = 77.498,P < 0.001) and a significant group × day interaction (F20,240 = 1.969,P < 0.01). Subsequent Bonferroni post hoc analysis showed significant differences between contingent mice vs. both yoked groups (P < 0.001). Discrimination between holes was significant only in the contingent MDMA group from day 1 through day 11 (Table S2).

Figure 1.

Operant yoked-control responding for intravenous infusions of MDMA (0.125 mg/kg per infusion). (a) The contingent group received an infusion of MDMA with every active nose-poke (n = 9), (b) the yoked MDMA group received an MDMA infusion everytime the contingent mouse made an active nose-poke (n = 9), (c) the yoked saline group received a saline infusion everytime the contingent mouse made an active nose-poke (n = 9). The data represent means + SEM active and inactive nose-pokes in 3-h sessions during the acquisition period. The asterisks denote significant differences between active and inactive nose-pokes for each training day. *P < 0.05; **P < 0.01; ***P < 0.001 (one-way anova).

MDMA-induced transcriptional changes

To assess possible transcriptional changes caused by active or passive MDMA administration, gene expression profiles in the four brain areas (ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus) from contingent MDMA, yoked MDMA and yoked saline mice were compared between the three possible pairs of experimental situations using microarray technology. This study design allowed to identify genes modulated by the direct pharmacological effect of MDMA on the brain (those differentially expressed in both the contingent MDMA–yoked saline and yoked MDMA–yoked saline comparisons) as well as genes involved in the cognitive processes related to active MDMA self-administration (those differentially expressed in both the contingent MDMA–yoked MDMA comparison and the contingent MDMA–yoked saline comparison).

Active and passive MDMA administration vs. saline: direct effect of the drug

In this part of the study, we focused on genes that were differentially expressed as a consequence of the MDMA effect. We considered only those genes displaying changes in the same direction in the two comparisons: yoked MDMA vs. yoked saline and contingent MDMA vs. yoked saline (Fig. 2a-I). Significant differences were observed in both comparisons in the four brain structures studied, ranging from 16 in dorsal raphe nucleus to 183 in hippocampus, most of them upregulated in contingent and yoked MDMA mice (Figs. 2a-II and S1a; Tables S3–S6). Those genes identified in the two comparisons were analysed for functional group over-representation using the DAVID database (except for raphe, because of the low number of common positive genes in the two comparisons), and similar clusters were obtained in all brain regions, the most significant ones being those involved in immune or inflammatory response, as well as in response to wounding or to stress (Fig. 2a-III). The pathway ‘Complement and coagulation cascades' was found to be enriched in the KEGG pathway analysis in frontal cortex, hippocampus and ventral striatum. Also, the pathways ‘B cell receptor signaling’ and ‘Natural killer cell mediated cytotoxicity’ were identified in frontal cortex and hippocampus, respectively.

Figure 2.

Gene expression changes caused by the direct effect of MDMA (a) or to MDMA-reinforced learning (b). (I) Venn diagrams, with the group of differentially expressed genes considered in each study shadowed in grey. On the left, expression changes in common in the comparisons of contingent MDMA–yoked saline and yoked MDMA–yoked saline. On the right, expression changes that are observed in the comparisons of contingent MDMA–yoked MDMA and contingent MDMA–yoked saline but not in yoked MDMA–yoked saline. CON: contingent. (II) Total number of genes that are upregulated or downregulated in the different comparisions performed. (III) Selection of over-represented biological categories that include differentially expressed genes. The number of positive genes included in each category is indicated on the right side of each bar. Biological categories correspond to gene ontology (GO) terms.

For the genes that were differentially expressed in the two comparisons, we obtained gene networks of inflammatory and immune response as the best scored in all four brain regions (Fig. S2). Interestingly, the NF–κB complex was a central node in three of the four identified networks. Remarkably, these gene networks comprise several of the upregulated genes in our microarray experiments, such as Lcn2 (lipocalin 2), with a dramatic upregulation in all regions, and Tgtp (T cell-specific GTPase) (frontal cortex and hippocampus; Fig. S2b,d, respectively).

As most of the genes were upregulated, we performed an analysis of over-representation of evolutionarily conserved TFBSs in upregulated genes in every structure, and the results pointed at a possible common regulation mechanism for all brain regions by the REL transcription factor class, specifically NF–κB and Rela transcription factors (Table 1a). As these two transcription factors can form homodimeric or heterodimeric complexes (containing the two proteins or combinations with other NF–κB complex proteins), we investigated whether the target sequences contained predicted binding sites for both and observed that NF–κB binding sites are also predicted in most of Rela targets (79.5%). Remarkably, separate KEGG analysis of these NF–κB and Rela targets identified two enriched pathways in common for the two transcription factors: ‘Jak-STAT signaling’ and ‘Chemokine signaling’.

Table 1.  Over-representation of transcription factor binding sites. (a) Predicted targets for transcription factors in common in all four brain regions in upregulated genes because of the direct effect of MDMA. b) Best predicted targets for transcription factors in upregulated and downregulated genes in hippocampus and dorsal raphe nucleus because of MDMA-reinforced learning in contingent mice
(a) MDMA direct effect
 Transcription factorTarget genes Z-score P-value
Ventral striatumNF–κB3310.91.1e-27
 Rela259.94.1e-23
Frontal cortexNF–κB2811.03.8e-28
 Rela2511.31.3e-29
Dorsal raphe nucleusNF–κB53.11.9e-03
 Rela30.30.76
HippocampusNF–κB325.05.7e-07
 Rela309.81.1e-22
(b) MDMA-reinforced learning
 UpregulatedDownregulated
Transcription factorTarget genes Z-score P-valueTranscription factorTarget genes Z-score P-value
Dorsal raphe nucleusNkx2-53116.82.4e-63Nhlh1419.51.1e-84
HippocampusNkx2-55729.77.7e-194Mzf1_1-441376.50

Finally, we performed cytogenetic bands enrichment analysis in all differentially expresed genes to identify clusters of co-regulated genes, and observed two interesting regions in common between hippocampus and frontal cortex: chromosome 11 B1 (ID:10742) and chromosome 17 B1 (ID:9061), including genes encoding interferon gamma-induced GTPases and histocompatibility molecules (class I MHC), respectively, which were confirmed only in these two brain regions in the GO enrichment analysis (data not shown). Finally, microRNA target enrichment analysis did not show significant results in upregulated genes in all structures (data not shown).

Based on their function, we chose 10 genes among those that were differentially expressed in the two comparisons (contingent MDMA–yoked saline and yoked MDMA–yoked saline) for further validation: seven genes related to immunological functions (Lcn2, which was differentially expressed in all four brain regions, and Ctla2a, Gbp2, Igtp, Iigp1, Iigp2 and Tgtp, which were identified in hippocampus and frontal cortex) and three genes involved in neurological processes (Sgk1 and Sgk3 in dorsal raphe nucleus and Slc17a7 in ventral striatum) (Table 2). QRT-PCR experiments validated the results of the microarray analysis, confirming the overexpression of these genes caused by active and passive MDMA intake in these particular brain structures, with the exception of differences in Sgk3 and Iigp2 that were not confirmed in the dorsal raphe nucleus and in hippocampus, respectively (Table 2).

Table 2.  Direct pharmacological effect of MDMA: qRT-PCR validation of microarray data of 10 genes
  Ave Expr*Contingent MDMA vs. yoked salineYoked MDMA vs. yoked saline
MicroarrayqRT-PCRMicroarrayqRT-PCR
Fold change P-value (Adj P-value)Fold changeFold change P-value (Adj P-value)Fold change
  1. NS, not significant.

  2. *Ave expr: average log2-expression for the gene probe over all arrays and channels.

  3. P-value < 0.05; normalized to Actb.

  4. Genes showing significant differential expression in two independent probe sets. The smallest absolute fold change is shown.

Frontal cortex
Lcn2 Lipocalin 28.8 52.8 5.5 e-8 (3e-4) 55.6 55.7 4.7e-8 (2.5e-4) 65.9
Ctla2a Cytotoxic T lymphocyte-associated protein 2 alpha8.8 3.0 8.5 e-4 (0.04) 4.8 4.9 3.7e-5 (8e-3) 7.3
Gbp2 Guanylate-binding protein 28.7 8.3 2.5 e-4 (0.03) 11.3 7.7 3.3e-4 (0.01) 12.9
Igtp Interferon gamma-induced GTPase8.4 4.2 3.7 e-4 (0.04) 6.1 4.0 5e-4 (0.02) 5.3
Iigp1 Interferon inducible GTPase 17.1 5.5 2.1 e-4 (0.03) 7.9 6.4 1e-4 (0.01) 9.4
Iigp2 Interferon inducible GTPase 27.4 3.2 1.9 e-4 (0.03) 3.7 3.0 3e-4 (0.02) 3.0
Tgtp T-cell-specific GTPase9.2 5.8 1.2 e-4 (0.03) 7.3 6.6 6.7e-5 (9.2e-3) 8.1
Hippocampus
Lcn2 Lipocalin 28.5 25.0 1.7e-6 (4e-3) 33.5 36.4 7e-7 (1.6e-3) 49.6
Ctla2a Cytotoxic T lymphocyte-associated protein 2 alpha8.2 3.0 1.5e-3 (0.01) 3.7 5.2 9.6e-5 (0.01) 8.0
Gbp2 Guanylate-binding protein 28.0 7.5 4e-3 (0.02) 12.0 6.9 5e-3 (0.03) 13.4
Igtp Interferon gamma-induced GTPase8.5 4.0 4.8e-3 (0.02) 7.0 4.1 4.3e-3 (0.03) 6.8
Iigp1 Interferon inducible GTPase 17.0 3.9 0.01 (0.04) 5.4 4.7 6e-3 (0.04) 7.2
Iigp2 Interferon inducible GTPase 27.3 2.9 3.7e-3 (0.02) NS 2.7 4.9e-3 (0.03) NS
Tgtp T-cell-specific GTPase9.2 4.4 1.6e-3 (0.01) 5.0 4.7 1.1e-3 (0.02) 5.5
Dorsal raphe nucleus
Lcn2 Lipocalin 28.7 26.1 2.7e-6 (8e-3) 37.7 52.4 5e-7 (7 e-4) 59.3
Sgk1 Serum/glucocorticoid regulated kinase 110.8 3.1 1.3e-5 (0.01) 2.8 4.3 1.6e-6 (1.4e-3) 4.1
Sgk3 Serum/glucocorticoid regulated kinase 38.6 2.1 0.01 (0.09) NS 4.2 5.3e-4 (9e-3) 2.8
Ventral striatum
Lcn2 Lipocalin 28.1 32.8 1.6e-7 (3.8e-4) 37.7 41.3 9.5e-8 (2.2e-4) 52.9
Slc17a7 Solute carrier family 17 (sodium-dependent inorganic phosphate cotransporter)8.1 3.6 6.9e-4 (0.04) 2.7 3.6 7.2e-4 (7.4e-3) 2.9

Contingent MDMA self-administration vs. yoked MDMA: drug reinforced learning

In the second part of the study, we focused on genes displaying differential expression as a consequence of active MDMA administation. We considered only those genes with changes in the same direction in the two comparisons that were performed: contingent MDMA vs. yoked MDMA and contingent MDMA vs. saline (Fig. 2b-I). Differences in gene expression were observed only in hippocampus (537 genes), with low fold changes (average fold change = 1.34). For a fair comparison, we changed the significance threshold to a less restrictive value (see Materials and methods) and observed differential expression of a number of additional genes in hippocampus (645, most of them downregulated in contingent mice) and also in dorsal raphe nucleus (61, most of them upregulated in contingent mice) (Fig. 2b-II; Tables S7 and S8). No differences were observed in frontal cortex or ventral striatum. In hippocampus, functional group over-representation performed using the DAVID database targeted some interesting neurological functions, such as synaptic transmission, regulation of synaptic plasticity, axononogenesis or learning and memory (Fig. 2b-III). In contrast, no specific neurological functions were identified in dorsal raphe nucleus in this step. Interestingly, KEGG pathway enrichment analyses showed three altered pathways in hippocampus and dorsal raphe nucleus: ‘long-term potentiation’ (LTP), ‘MAPK signaling’ and ‘Wnt signaling’, in which most of the genes were also downregulated in the contingent mice in hippocampus (Fig. 3).

Figure 3.

Selection of altered KEGG pathways in hippocampus after MDMA-reinforced learning. Heatmap showing the level of expression in the microarray of genes identified in two KEGG pathways. For each experimental group the three replicates are shown.

Gene network construction on all these genes showed that the best scored network in hippocampus (score = 47; Fig. 4a) involves cell-to-cell signalling and interaction/nervous system development and function. Remarkably, most of the genes present in this network are differentially expressed in hippocampus of contingent mice. This network includes genes involved in synaptic vesicle fusion, synapsis formation and neurotransmitter release, such as Cplx2, Vamp2, Ngln2, Nrxn1 and Nrxn2, consistently with the over-represented GO categories identified. Interestingly, the best scored network in dorsal raphe nucleus (score = 26; Fig. 4b) also includes genes involved in cell-to-cell signalling, interaction/nervous system development and function and behaviour, such as Camk2a and Kalrn.

Figure 4.

Gene network graphical representation of interaction between differentially expressed genes after MDMA reinforced learning (contingent MDMA vs. yoked MDMA and yoked saline mice). The best scored gene network in (a) hippocampus and in (b) dorsal raphe nucleus includes genes involved both in cell-to-cell signaling and interaction as well as in nervous system development functions. Genes differentially expressed in the two comparisons in the same direction are represented as nodes depicted in red (upregulated) or green (downregulated), and biological relationship between two nodes is represented with a solid or dashed line (indicating direct or indirect interaction, respectively). Each node is displayed with different shapes that indicate the functional class of the gene product shown on the right. Modulatory effects on expression are indicated by arrows. Gene expression changes validated by qRT-PCR are indicated by blue arrows.

Analysis of over-representation of TFBS in hippocampus and dorsal raphe predicted a possible common regulatory mechanism in upregulated genes, but not in the downregulated ones (Table 1b). Upregulation in dorsal raphe nucleus and hippocampus could be related to the action of the Nkx2-5 transcription factor, which is the best scored prediction in both structures. The subset of genes predicted to be regulated by Nkx2-5 showed enrichment of LTP in both hippocampus and dorsal raphe nucleus in the KEGG pathways analysis. Downregulation of hippocampal genes could be mediated mainly by the Mzf1_1-4 transcription factor (Table 1b). Cytogenetic bands enrichment analyses did not show any significantly enriched region (data not shown). MicroRNA target enrichment analysis performed in the two brain structures identified, among others, miR-96, with predicted targets in 21 genes downregulated in hippocampus (adjusted P-value = 4.3e-06), all of them showing interesting functions such as cell communication and synapse organization in the GO enrichment analyses.

Based on their function, we selected eight genes in hippocampus (Amigo1, Bzrap1, Gprin1, Mapk8ip1, Nlgn2, Vgf, Madd and Axin2) and four genes in dorsal raphe nucleus (Camk2a, Kalrn, Ddn and Egr3) for further validation. QRT-PCR results confirmed a downregulation of the Bzrap1, Nlgn2 and Axin2 genes in the contingent mice in hippocampus, and an upregulation of the Camk2a, Kalrn, Ddn and Egr3 genes in the contingent mice in dorsal raphe nucleus (Table 3). Camk2a is present in the ‘LTP pathway’ in dorsal raphe nucleus. It is also present in the ‘Wnt signaling pathway’ in this structure, like Axin2 in hippocampus. Nlgn2 gene together with Camk2a and Kalrn are present in hippocampus and dorsal raphe nucleus gene networks, respectively (Fig. 4). The genes whose upregulated expression was validated by qRT-PCR in dorsal raphe nucleus (Camk2a, Kalrn, Ddn and Egr3) have predicted binding sites for Nkx2-5.

Table 3.  Effect of active MDMA self-administration: qRT-PCR validation of microarray data of three genes in hippocampus and four genes in dorsal raphe nucleus
Gene symbolGene nameAve Expr*Contingent MDMA vs. yoked MDMAContingent MDMA vs. yoked saline
MicroarrayqRT-PCRMicroarrayqRT-PCR
Fold change P-value (Adj P-value)Fold changeFold change P-value (Adj P-value)Fold change
  1. NS, not significant.

  2. *Ave Expr: Average log2-expression for the gene probe over all arrays and channels.

  3. Genes showing significant differential expression in two independent probe sets. The smallest absolute fold change is shown.

  4. P < 0.05; normalized to Actb.

Hippocampus
Bzrap1 Benzodiazepine receptor-associated protein 110,3 −1.4 1e-4 (4e-3) −1.6 −1.2 0.01 (0.04) −1.3
Nlgn2 Neuroligin 29.9 −1.3 6e-4 (6e-3) −1.3 −1.2 3e-3 (0.02) −1.4
Axin2 Axin 27.4 −1.3 2e-3 (9e-3) −1.2 −1.3 6e-4 (0.01) −1.3
Dorsal raphe nucleus
Camk2a Calcium/calmodulin-dependent protein kinase II alpha7.1 2.8 0.01 (0.11) 2.3 1.9 0.02 (0.12) NS
Kalrn Kalirin, Rho GEF kinase7.8 2.9 4e-3 (0.10) 1.5 3.2 2.3e-3 (0.06) NS
Ddn Dendrin7.1 3.5 8e-3 (0.11) 5.5 3.3 0.01 (0.09) 5.3
Egr3 Early growth response 35.4 4.0 5e-4 (0.09) 3.0 3.3 1.5e-3 (0.06) 2.4

Discussion

The aim of this study was to identify alterations in brain gene expression as a result of the pharmacological effect of MDMA administration, as well as neuroadaptative changes underlying the learning process associated with operant MDMA self-administration. For that purpose, we have validated a new operant paradigm consisting in master mice that are trained to acquire a stable operant behaviour to self-administer a reinforcing dose of MDMA (Orejarena et al. 2009). Each master mouse is connected to an MDMA-yoked animal that passively receives an identical dose of MDMA and to another yoked mouse that receives saline infusions. This yoked-control operant intravenous self-administration paradigm was combined with microarray technology. The results of this experimental design suggested that (1) MDMA modulates the expression of genes involved in inflammatory and immune response in different brain areas; and (2) the hippocampus and the dorsal raphe nucleus may participate in the neuroadaptative changes leading to active MDMA seeking behaviour.

Changes in gene expression relevant to the direct effects of the drug were evaluated by comparing the contingent MDMA and the yoked MDMA mice to the yoked saline mice. In this case, most of the hits corresponded to genes involved in immunological or inflammatory response. Among them, we identified a strong overexpression of Lcn2 in all the brain regions, which was also present in all the best scored gene networks identified. The Lcn2 gene encodes lipocalin2 that mediates astrocytosis under inflammatory conditions and is induced after chronic or thermal stress in brain reward regions (Krishnan et al. 2007; Lee et al. 2009; Roudkenar et al. 2009). We also validated the overexpression of other genes (Ctla2a, Gbp2, Igtp, Iigp1, Iigp2 and Tgtp) both in hippocampus and in frontal cortex. All of them, except for Ctla2a, are genes coding for GTPases that are induced by type II interferon (INF-γ) (Carlow et al. 1998; Miyairi et al. 2007; Vestal et al. 1998; Yamada et al. 2009; Zerrahn et al. 2002; Zhang et al. 2003), and are involved in some cellular processes mediating interferon control of immune and inflammatory responses.

Our results showing that repeated MDMA administration produces generalized changes in the expression of genes related to inflammatory and immunological responses are in accordance with previous evidence showing that exposure to MDMA disrupts the immune system (Connor 2004), which may contribute to its neurotoxic effects (Torres et al. 2010). From our analyses, we could hypothesize that the expression of many of these genes is regulated by NF–κB and Rela, which is also supported by in silico networks pointing at the NF–κB complex as a central node. This finding is consistent with previous studies showing that MDMA may induce the activation of NF–κB (Montiel-Duarte et al. 2004; Orio et al. 2010; Tiangco et al. 2005). However, the expression of neither NF–κB nor Rela was altered in our animal model according to microarray data. Our results obtained from repeated MDMA administration differ from previous data obtained after acute MDMA exposure in murine models, where mainly serotonin receptors, several transcription factors, cytoskeletal, cell adhesion and metabolic genes were differentially expressed in cortical areas or in the striatum (Marie-Claire et al. 2007; Thiriet et al. 2002).

Another direct consequence of the exposure to MDMA was the upregulation of genes involved in neuroadaptations and synaptic plasticity, including the Sgk1, Sgk3 and Slc17a7 genes. Sgk1 and Sgk3, which encode the serum/glucocorticoid regulated kinase 1 and 3, respectively, were identified in the dorsal raphe nucleus, composed mainly by serotoninergic neurons. Both have been described to be involved in memory consolidation in hippocampus (Von Hertzen & Giese 2005) and regulate glutamatergic neurotransmission (Boehmer et al. 2003a,b, 2005, 2006; Strutz-Seebohm et al. 2005a,b). In addition, Sgk1 increases neurite formation and dendrite growth in spinal cord and hippocampal neurons (David et al. 2005; Yang et al. 2006). Although these functions have been described in hippocampus and are related to the glutamatergic neurotransmission, they may also occur in other cell types, such as serotoninergic neurons. Interestingly, Slc17a7, encoding the vesicular glutamate transporter 1 (Vglut1), was upregulated in ventral striatum. Vglut1 is involved in synaptic plasticity and plays an important role in excitatory transmission (Fremeau et al. 2004). In this regard, our findings are consistent with a recent study showing gene expression changes in several glutamine transporters and receptors after repeated MDMA administration (Kindlundh-Hogberg et al. 2008). Although the ventral striatum is mainly formed by GABAergic neurons, it contains glutamatergic afferences, and overexpression of Vglut1 may be localized in glutamatergic axons after axonal transport of the corresponding mRNA molecules followed by translation in situ as previously described (Donnelly et al. 2010; Wei 2011). Although not validated, enriched cytogenetic bands and GO analysis also pinpointed class I major histocompatibility complex (MHC) genes in hippocampus and frontal cortex, which, besides from their involvement in immune response, have been related to activity-dependent remodeling and plasticity of connections in the CNS, playing an important role in LTP and long-term depression (Huh et al. 2000).

In order to evaluate gene expression changes related to the learning component of the operant task to obtain MDMA infusions, we compared brain expression profiles of contingent MDMA vs. yoked MDMA mice. The highest number of statistically significant changes in gene expression was observed in hippocampus, supporting the crucial role of this brain structure in the control of memory and cognitive functions. These results are also in agreement with previous studies showing that cocaine-induced conditioned place preference (CPP) depends on molecular changes that occur in the hippocampus of trained rats (Krasnova et al. 2008; Tzschentke 1998). In this regard, several genes differentially expressed in hippocampus are involved in LTP, as well as in other important pathways for learning and memory processes, which are also altered in hippocampus of rats with cocaine-induced CPP (Krasnova et al. 2008). Surprisingly, several additional genes showed differential expression also in dorsal raphe nucleus in our study, some of them also related to these pathways. Interestingly, recent evidence suggests that dorsal raphe nucleus is involved in encoding reward-related aspects of motivated behaviour (Bromberg-Martin et al. 2010; Nakamura et al. 2008). Also, concerning active MDMA self-administration, we observed more similarities between hippocampus and dorsal raphe nucleus in our study: (1) we identified common gene networks involved in cell-to-cell signaling and nervous system development functions in the two structures and (2) analysis of over-representation of TFBS pointed at Nkx2-5 as a common modulator of genes that are upregulated in active MDMA self-administration in both brain regions. However, microarray data did not show altered expression of Nkx2-5. This transcription factor has been described to be involved in neuronal differentiation (Riazi et al. 2005).

Most genes showing differential expression in hippocampus, in contrast with dorsal raphe nucleus, are downregulated in contingent mice. Such downregulation may be related to the effect of transcription factor Mzf1_1-4, which has predicted binding sites in >75% of the genes that are downregulated in this brain structure (Table 1b). However, the expression of this transcription factor is not altered according to microarray data. Also, several downregulated genes in hippocampus are predicted targets for the microRNA miR-96, which is expressed in frontal cortex and seems to be involved in modulating gene expression in several neuronal processes such as LTP and depression (Juhila et al. 2011).

Seven of the 12 genes could be further validated by qRT-PCR, three in hippocampus (Bzrap1, Nlgn2 and Axin2) and four in dorsal raphe nucleus (Camk2a, Kalrn, Ddn and Egr3). The three genes validated by qRT-PCR in hippocampus are underexpressed in contingent mice and modulate synapse and neurogenesis. The benzodiazapine receptor-associated protein 1 (encoded by Bzrap1) is an adaptor molecule thought to regulate synaptic transmission by linking vesicular release machinery to voltage-gated Ca2+ channels (Wang et al. 2000). The Nlgn2 gene encodes neuroligin 2, which has an important role in organizing the functional properties of synapses, acting preferentially at inhibitory synapses (Gibson et al. 2009). Other genes encoding proteins involved in neurotransmission have also been described to be altered by cocaine self-administration in rats (Ahmed et al. 2005). Axin2 is a negative regulator of the Wnt signalling pathway, signalling that induces neurogenesis in hippocampal neurons (Jho et al. 2002; Lie et al. 2005).

The four genes validated by qRT-PCR in dorsal raphe nucleus are overexpressed in contingent mice and are related to synaptic plasticity and neuroadaptations. Camk2a encodes the Ca2+/calmodulin-dependent protein kinase II alpha (CaMKIIα), which mediates activity-dependent synaptic plasticity and has an essential role in dendritic spine enlargement, LTP and learning (Yamagata et al. 2009). Dendritic spine morphogenesis is also induced by kalirin-7, an isoform encoded by the Kalrn gene (Penzes & Jones 2008). In addition, another study links these two proteins with the same signalling pathway that controls functional and structural spine plasticity (Xie et al. 2007). Kalirin is also involved in neurite outgrowth through the nerve growth factor (NGF) signalling pathway (Chakrabarti et al. 2005). In addition, we identified these two genes in the best scored network in dorsal raphe nucleus, in which the ras-dependent protein kinase ERK is a central node (Fig. 4b). This is in agreement with a previous study supporting the role of the ERK pathway in the development of addiction-like properties of MDMA (Salzmann et al. 2003). On the other hand, the Ddn gene encodes a dendritically localized mRNA that is translated to the protein dendrin, potentially involved in neuroplasticity events and modulation of post-synaptic cytoskeleton (Kremerskothen et al. 2006). In addition, Egr3 is a member of the Egr gene family, a group of synaptic activity-inducible immediate early genes involved in neuroplasticity related to memory and learning (Guzowski 2002; Li et al. 2005, 2007). Interestingly, the best characterized gene of the family is Egr1, and its expression is increased by MDMA in rat prefrontal cortex, striatum and hippocampal dentate gyrus (Shirayama et al. 2000).

In our experimental conditions, qRT-PCR validation succeeded in 76% of the cases (considering all the brain structures and comparisons tested for each gene: 47 validated/62 tested). The limited number of replicas in our study may have prevented some validations, specially when differences in gene expression were low in the microarray (see Table S1).

Our results may suggest a role for dorsal raphe nucleus and hippocampus in the motivational and learning processes needed to actively self-administer MDMA because they are the only studied brain regions that show statistically significant changes in the comparison of contingent MDMA vs. yoked MDMA mice. In addition, the best scored gene network identified common functions in dorsal raphe nucleus and hippocampus, with genes involved in cell-to-cell signaling and nervous system development functions. And finally, validated genes that are upregulated in the dorsal raphe nucleus are involved in neuroplasticity and neuron remodelling, and validated genes downregulated in hippocampus are involved in synapse function and neurogenesis.

In conclusion, using the yoked-control operant intravenous self-administration paradigm, which is the most relevant animal model to study the addictive potential of drugs of abuse in humans, we showed that repeated exposure to MDMA induces the expression of genes related to inflammatory and immunological responses in several brain structures including the ventral striatum, frontal cortex, dorsal raphe nucleus and hippocampus. In addition, the gene expression changes identified in hippocampus and dorsal raphe nucleus following MDMA self-administration suggest that both brain regions may be involved in motivated learning associated with active MDMA seeking behaviour. However, because of the sample size limitation, as we evaluated nine mice per condition of contingent MDMA, yoked MDMA and yoked saline, further studies should be performed to confirm these results.

Acknowledgments

This work was supported by the Spanish ‘Ministerio de Ciencia e Innovación (MICINN)’ (SAF2007-64062), ‘Instituto de Salud Carlos III’ (RD06/001/001 and PI070709), the Catalan Government (SGR2009-00131 and SGR2009-00971), the ICREA Foundation (ICREA Academia-2008), ‘Plan Nacional Sobre Drogas' (PNSD#2009/022 and PNSD#2009/026), Fundació‘La Marató de TV3’ (2007) and the DG Research of the European Commission (GENADDICT, LSHM-CT-2004-05166; and PHECOMP, LSHM-CT-2007-037669). Partial support from FEDER funds is also acknowledged. MR is a recipient of a Miguel de Servet contract from ‘Instituto de Salud Carlos III-MICINN’ (Spain) and NF-C was supported by a fellowship from the Biomedical Network Research Centre on Rare Diseases (CIBERER). These institutions had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.

Financial disclosures

None of the authors reported any biomedical financial interests or potential conflicts of interest.

Ancillary