INTRODUCTION

A hallmark feature of drugs of abuse, such as the psychostimulant cocaine, is to alter the expression of a wide array of behavioral, neurochemical, and cellular parameters.^1,2,3,4 It appears that most of the changes seen in mRNA and protein expression as well as biochemical parameters disappear rapidly when drug is discontinued,^5,6,7,8,9 and may even be blunted by chronic drug exposure.¹⁰ By effecting downstream events, even short-lived changes in expression patterns may lead to more enduring modifications, and thereby provide the neurobiological substrates for sensitization, tolerance, craving, and addiction.^3,11

While cocaine has a distinct effect in different anatomical regions,^12,13,14 it appears to rely heavily on the prefrontal cortex (PFC) both for the induction and long-term sensitization to cocaine.^1,15,16 Imaging studies have demonstrated altered neuronal activity in the PFC of cocaine abusers responding to cocaine or cocaine-related cues.^{17,18,19,20,21,22} Neuropsychological testing has indicated that executive planning and consequence evaluation, measures of the functionality of the ventromedial aspects of the PFC, are adversely affected, while explicit memory and learning, involving the dorsolateral PFC (dlPFC) circuitry, appears functional.^23,24,25 Consequently, the PFC appears to be critically involved in both experimental animal models and in human cocaine abuse.^1,26,27,28

Profiling the expression of large sets of transcripts has been greatly facilitated by the development and miniaturization of DNA array technology. This technique enables the concurrent screening of large sets of known and novel targets of drug-induced change. Equally important, by profiling a large number of transcripts in minute amounts of human brain tissue, the microarray technology allows for a judicious use of a limited and unique resource. The present study employed complementary DNA (cDNA) microarrays to analyze differences in transcription patterns in the dlPFC of seven individual sets of gender-, age- and post-mortem interval-matched cocaine abusers and controls (Table 1).

Table 1 - Demographic data for control- and cocaine-abusing subjects.

Full table

RESULTS

Selection Criteria

Transcripts for which the expression levels differed across all or most of the seven individual cocaine–control comparisons were identified by sorting the individual transcript expression data by the mean of the P-values for the seven matched comparisons. Selecting transcripts with a mean P-value <0.05 and a minimum of three individual matched comparisons with a P-value 0.001 yielded 65 clones Table 2. These represent 64 different gene transcripts, and correspond to 5.6% of the total number of transcripts present on the Neuroarray (Figure 1). Two of the 65 different clones overlapped within the same region of a tubulin transcript (TUBA2), and those two sets of data corresponded closely (Table 2).

Figure 1.

Example of a neuroarray used in the present study. Note that each clone is represented twice, once in Grid A, and once in Grid B.

Full figure and legend (97K)

Table 2 - Mean p-value and z-ratio scores for 65 selected transcripts.

Full table

Main Gene Expression Pattern

Subsequent inspection of the z-ratios revealed two distinct expression patterns, subdividing the seven individual comparisons into two subgroups (Table 2, Figures 2, 3 and 4). Pattern A included comparisons #1, #2, #4, and #6, while comparisons #3, #5, and #7 followed pattern B. As can be seen from Table 2, most of the 65 transcripts that were differentially expressed in the cocaine-control subject comparisons were changed in opposite directions in the two subgroups. Of the 65 transcripts selected, 17 (25%) did not adhere strictly to this pattern. Re-examining the entire set of expression data, there was a similar Group A vs Group B subdivision for approximately 10–15% of all transcripts. These included many transcripts with weaker differences between the subject and controls that did not meet the applied criteria for differential expression. The pattern was, however, more prominent for those transcripts with the greatest differences between the cocaine and control subjects.

Figure 2.

Comparison of individual QPCR ratios and microarray z-ratios for GNAO1, MDH1, MAP1B, and PLP1. For each transcript, the upper graph represents the QPCR results, expressed as numerical ratios between the cocaine subject and the respective matched control. Ordinate values above 1 indicate higher expression, while values below 1 indicate lower expression, with respect to the cocaine subject. The lower graph illustrates the corresponding microarray z-ratios. Positive values on the ordinate indicate higher, and negative values indicate lower, levels of expression with respect to the cocaine subject. For both types of experiments, the seven matched cocaine–control pairs are indicated on the abscissa. For each cocaine subject and matched control, the QPCR assay was run for eight paired replicated samples. For each transcript, the significance of the difference between each cocaine subject and the corresponding control was tested using a two-tailed Wilcoxon matched-pairs signed-ranks test.⁷³ The significance of the difference in MDH1 expression could not be tested because of a smaller number of replicates.

(=P<0.05, =P<0.01).

Full figure and legend (63K)

Figure 3.

Divergence of transcript expression levels in cocaine–control subject comparisons. A divergent pattern emerged when the overall mean t-test was used to select the genes. This is illustrated by plotting the average t-test value against the average z-ratio for the pattern. For clarity, only the selected transcripts are included. Blue squares indicate average z-ratios for the four pattern A subjects, while pink asterisks indicate average z-ratios for the three pattern B subjects. This illustrates that increased differential expression between controls and cocaine subjects, as indicated by t-test P-value, is also reflected in a greater divergence between the transcriptional profile of the patterns.

Full figure and legend (63K)

Figure 4.

Relation between overall differential gene expression (cocaine vs controls) and the divergence between patterns A and B expression profiles for all 1152 transcripts on the neuroarray. (a) The absolute z-ratio differences for patterns A and B expression levels as a function of the mean absolute z-ratio difference between patterns A and B. This illustrates that differentially expressed transcripts tend to show larger differences in expression between the two patterns. The selected 65 differentially expressed transcripts are indicated by blue squares, while all other transcripts are indicated by pink asterisks. (b) The relation between expression profiles for the two patterns. The z-ratios for all 1152 transcripts for pattern A are indicated on the ordinate, while the corresponding z-ratios for each transcript in pattern B comparisons are indicated on the abscissa. There is a tendency for transcripts with decreased expression for pattern B cocaine subjects to show increased expression in pattern A cocaine subjects and vice versa.

Full figure and legend (102K)

In other experiments (data not shown), microarray studies of pooled samples containing equal amounts of RNA from either cocaine abusers (cocaine pool) or matched controls (control pool) resulted in z-ratios for the transcripts selected in the present study that were intermediate in magnitude between those seen for either pattern A or pattern B, while the direction of change more closely resembled that of pattern B (data not shown).

Cross-referencing Expression Profiles

While a radioactive detection format dictates the use of separate arrays for individual samples, it also provides the opportunity to compare each individual array to any other individual array processed similarly. The normalized ratio provides an estimate of the relative changes in transcript expression for the matched comparisons, and concurred with the z-ratio data in both the general ranking of transcripts as well as the direction of change (data not shown). Normalized ratio comparisons among the cocaine-abusing individuals in pattern B showed that overall expression levels for COC#3 and COC#7 were similar, while those of COC#5 appeared generally depressed compared to either. Only a few differences in transcript expression were present between the cocaine-abusing individuals in pattern A, and none were transcripts selected by the z-ratio criteria.

GC-MS of cocaine, opiates, and related analytes (Table 3 and below), revealed the presence of low levels of cocaine and benzoylecgonine (BE) in the dlPFC of one control subject, CTR#5. By comparing the normalized ratio of two pattern B subjects, COC#5 and COC#7, to the A pattern CTRs that best approximated their age and PMI, CTR#4 and CTR#6, we found that the normalizing individual had some effect on both magnitude and direction of transcript expression for COC#7. In contrast, when comparing COC#5 to CTR#4 and to CTR#6, the magnitudes and directions of changes were nearly identical to those obtained using CTR#5, indicating that the presence of cocaine in this control had not affected the outcome of the comparison. Therefore, the results obtained for selected transcripts in the A and B groups were retained across a number of different analysis.

Table 3 - GC-MS of dorsolateral prefrontal cortex (dlPFC) tissue.

Full table

Cluster Analysis

Clustering of the data using the z-ratio comparisons for matched comparisons (Figure 5a), or a z-ratio based on comparing individual cocaine abusers vs pooled control values (Figure 5b), both produced two discrete clusters of cocaine abusers, corresponding to the Group A/Group B patterning (Table 2, Figures 2 and 3).

Figure 5.

Cluster analysis. Cluster analysis was performed using average-linkage hierarchical cluster analysis, described by Eisen et al,⁷² of the selected set of 65 transcripts or the entire set of 1152 transcripts (not shown). This analysis demonstrated the clustering of the A pattern cocaine-abusing individuals, differing from the B pattern cocaine-abusing individuals. This is illustrated by the clustering of the z-ratio for the individual cocaine–control comparisons (a) and for the comparison of individual cocaine abusers to a control pool, formed by combining all z-score data from all seven controls (b). The color and intensity indicate direction and level of change with respect to the cocaine-abusing individual: green spectrum colors indicate decreased expression, while red spectrum colors indicate upregulated transcript expression.

Full figure and legend (250K)

Quantitative polymerase chain reaction (PCR) (QPCR)

The unexpected finding of a divergent pattern of expression between individual cocaine–control comparisons was probed using an alternate RNA-based quantitative methodology. Real-time QPCR for select transcripts (guanine nucleotide binding protein (G protein) alpha activating activity polypeptide O (GNAO1), soluble malate dehydrogenase (MDH1), microtubule-associated protein-1B (MAP1B), and proteolipid protein 1 (PLP1) confirmed the observation that individual cocaine–control comparisons exhibited opposite changes in expression level (Figure 2). QPCR for comparison #2 and, in particular, comparison #5 did not confirm the z-ratio data from the microarray; nevertheless, the divergent pattern was confirmed. The Wilcoxon matched-pairs signed-rank test confirmed the statistical significance of most of the changes demonstrated by quantitative polymerase chain reaction (QPCR) (Figure 2). Multiple rounds of RNA extraction and subsequent cDNA transcription were performed on aliquots of the brain samples to guard against accidental loss or contamination of the sample, and to allow for replication. In the last round of extractions, each individual dlPFC tissue aliquot was subdivided prior to RNA extraction, yielding both QPCR data and GC-MS information on the presence of cocaine, opiates, and related analytes from the same dlPFC aliquot (Table 3). These results conclusively demonstrated the existence of two different patterns of regulation of the same set of transcripts in the cocaine-abusing cohort.

Toxicology

An individual aliquot of dlPFC brain powder was subdivided for use in GC-MS and QPCR. GC-MS demonstrated the presence of cocaine and two major cocaine metabolites, BE and ecgonine methyl ester (EME), in the dlPFC for all individuals in the cocaine-abusing group (Table 3). The presence of anhydroecgonine methyl ester (AEME) for three pattern A individuals (COC#1–2, COC#4) suggested that cocaine in those individuals had been smoked ('crack' cocaine). The only other drugs of abuse detected were alcohol for COC#2, indicated by metabolites cocaethylene (CE), ecgonine ethyl ester (EEE), and norcocaethylene (NCE), while morphine was detected for COC#6. One control subject, CTR#5, tested positive for cocaine and BE in the dlPFC, even though no drugs were detected at the post-mortem toxicology screen.

All samples were negative for methamphetamine, amphetamine, PCP, norbenzoylecgonine, m- and p-hydroxycocaine, m- and p-hydroxybenzoylecgonine, codeine, norcodeine, normorphine, and 6-acetylmorphine. As the GC-MS data were obtained from a tissue aliquot that was used for QPCR, the presence of two different transcriptional patterns could not have been caused by the inadvertent misidentification of samples at some point during dissection and other preparative routines.

Genotyping

There have been reports that polymorphisms in the catechol-O-methyltransferase (COMT) gene influences prefrontal functioning²⁹ and that dopamine transporter (DAT) function is altered in cocaine abusers.^30,31 Genotyping was performed to determine if DAT or COMT polymorphisms could account for the observed divergence in transcriptional patterns. There were no obvious differences in DAT or COMT genotypes between the cocaine abusers and their respective controls, or between Group A and Group B cocaine subjects (Table 4).

Table 4 - Individual DAT and COMT genotypes.

Full table

Functional Groupings

The 65 selected transcripts fell into several distinct functional categories. Seven transcripts (CYC1, MDH1, SOD1, COX6C, CASP4, ENO2, ATP5G3) were associated with mitochondrial function and energy metabolism. Of the selected transcripts 24 related to the neuronal cytoskeleton (NEFL, NEF3, MAP1B, TUBA1, TUBA2, CAPZB, NP25), extracellular matrix (CDH1, CDH17, CTNND2, PNUTL2), growth cones (GAP43, DNCl1, PTN), and G-protein-related signaling and neurotransmission (GNAO1, GNB1, KCNJ6/GIRK2, GAD1, NSF, CLTB). Five transcripts were related to oligodendrocyte function and glycoproteins (PLP1, GPM6A, GPM6B, GSN, ST3GALV1). Six transcription factors were selected, including the immediate early gene fra-1 (FOSL1). Nine transcripts related to protein processing (UCHL1, UBE2I), growth factors (FGF2), and various other functions did not fall clearly into larger categories. Three transcripts specifically related to drug metabolism were altered (CES2, CYP51, SLC10A1). Phosphatases and kinases were highly represented with 11 different transcripts. The mitogen-activated protein kinase (MAPK)/extracellular-regulated kinase (ERK) signaling cascade was represented by MAPK10 (JNK3A/SAPK), MAPK3 (ERK1), and MAP2K1 (MEK1), while other selected transcripts interacting with the MAPK/ERK pathway are exemplified by PTK2 (FAK), PARG1, and PRKCB1.

Association With 'crack' cocaine Use

GC-MS detected the cocaine metabolite AEME, indicative of smoked or 'crack' cocaine, in the dlPFC of cocaine subjects COC#1, COC#2, and COC#4. Consequently, we examined the individual transcripts (Table 2) post hoc for changes potentially associated with recent crack cocaine use. Each of these three cocaine–control comparisons were in the Group A pattern. Approximately 45% (29 of 64) of the selected transcripts showed a similar pattern of consistently larger increases in COC#1, COC#2, and COC#4, and a smaller increase in COC#6 (Table 2). These transcripts included CYC1, SOD1, CASP4, TUBA1, TUBA2, GAP43, GPM6A, GPM6B, Go, KCNJ6, FOSL1, PARG1, PTK2, PRKCB1, MAP2K1, UCHL1, and YWHAH.

DISCUSSION

Main Findings

A consistent subset of 65 transcripts met our selection criteria for altered expression across the seven cocaine–control comparisons. The cocaine-abusing cohort was, however, divided into two subgroups, distinguished by opposite directions of change. Pattern A was associated with comparisons #1, #2, #4, and #6, while pattern B was associated with comparisons #3, #5, and #7. The pattern of bidirectional change was identifiable for 10–15% of all transcripts, including some transcripts that failed the selection criteria, but was most pronounced for transcripts exhibiting low mean P-values and large z-ratios. The pattern was replicated by QPCR for select transcripts (GNAO1, MDH1, MAP1B, PLP1). GC-MS confirmed the presence of cocaine and cocaine metabolites in samples originating from cocaine abusers. Differences in COMT and DAT genotypes did not account for differences between pattern A and pattern B. There was, however, an association of 'crack' cocaine use with the pattern A change in gene expression.

Divergent Expression Pattern

There is precedence for divergent expression of mRNA, protein or receptor function within subgroups of substance abusers. Decreased DAT mRNA was demonstrated in the substantia nigra in a subgroup of cocaine abusers exhibiting preterminal cocaine-induced excited delirium,³² while cocaine, ethanol, or co-abuse differentially affected serotonin transporter binding sites in the dorsal raphe.³³ In addition, three sets of pooled samples of frontal or motor cortex from chronic alcoholics differed in the direction of change in select transcripts,³⁴ among which were two oligodendrocyte-related genes selected presently (PLP1, GPM6B). In schizophrenia, it was noted that individual subjects differ, such that group means are not representative of the pathophysiology of the disorder.³⁵ Individual differences in schizophrenia were observed in an in situ hybridization study in support of microarray data.³⁶ The dynamic regulation of the soluble MDH1 transcript was illustrated by a downregulation in the dlPFC of schizophrenics in contrast to an upregulation in long-term haldol-treated monkeys.³⁷

Implications of the Two Patterns

Pattern A could represent a state of general activation, as exemplified by increases in transcripts associated with metabolic or mitochondrial activity (MDH1, CYC1), neurite growth and cytoskeletal remodeling (NEFL, NEF3, MAP1B, GAP43), protein metabolism (UCHL1, UBE2I), and the transcription factor fra-1 (FOSL1). The recent crack use evidenced in three of four pattern A subjects, and in none of the pattern B subjects, potentially indicates a different, or more acute, state of cocaine stimulation in the pattern A subjects. It is tempting, therefore, to speculate that cocaine abusers are either in an acute state of hyperstimulation with increased demand on gene expression, or in a refractory, destimulated state with decreased gene expression. Such surges or peaks in transcriptional activity might occur during the experiences of COC-related craving or euphoria reported by COC abusers.^17,22

Cytoskeleton and Signal Transduction

Virtually every aspect of cellular proliferation, differentiation, and cytoskeletal-linked signal transduction is regulated by changes in tyrosine phosphorylation. Tyrosine phosphorylation, in turn, is controlled by the opposing activities of protein tyrosine kinases (PTKs) and protein tyrosine phosphatases (PTPs, PTPRA). The mitogen-activated protein kinase (MAPK) family is thought to mediate pivotal intracellular signaling cascades implicated in neuronal plasticity, and includes the c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK, JNK3A1), the focal adhesion kinase (FAK/PTK2), and the extracellular signal-regulated protein kinases (ERKs, MEK1, ERK1). A different FAK transcript (PYK2) was reported among highly cocaine-responsive transcripts in the rat hippocampus,³⁸ while PYK2 and MEK1 expression were increased in the nucleus accumbens of chronic cocaine-treated macacaques.³⁹ Activation of the MAPK/MEK/ERK signaling pathway occurs via soluble growth factors like basic fibroblast growth factor (bFGF/FGF2) or cell-adhesion molecules (CAMs), such as catenin (CTNND2) and cadherins (CDH1, CDH17). Downstream genes activated include the ERK-dependent Fos-related antigen-1 (Fra-1, FOSL1), which is central to the mitogenic activation by activator protein-1 (AP-1).⁴⁰ The rewarding effects of cocaine were demonstrated to be abolished by selective systemic inhibition of MEK,⁴¹ further underscoring the central role of the MAPK signal transduction cascade in mediating cocaine-associated cellular communication.

Similar to tyrosine phosphorylation, phosphoprotein phosphorylation appears tightly regulated by the opposing action of kinases and phosphatases. Cocaine, morphine, opioids, and nicotine have all been demonstrated to regulate both the levels and phosphorylation state of neurofilaments.^{42,43,44,45,46} The major dephosphorylating agents, protein phosphatase 1 (PP1CA) and protein phosphatases 2A and 2B (PP2CA, PP2CB),^47,48 were prominently regulated along with their target proteins (NEFL, NEF3, MAP1B, TUBA1, TUBA2).

Synaptic Transmission and Neuroplasticity

Like memory and learning processes that activate specific pathways directing cytoskeletal organization and axonal growth, so has cocaine been demonstrated to produce distinct regional alterations in dendritic arborizations and spine densities.^4,49 Neuronal plasticity is a dynamic process requiring a coordinated restructuring of neuronal connectivity, and transcripts encoding growth cone-associated proteins, such as the 43 kDa growth-associated protein GAP43, dynein and pleiotropin (GAP43, DNCl1, PTN), were significantly altered.

G-protein-mediated signaling features prominently in the pathways thought to mediate cocaine-induced alterations.^50,51,52 Protein kinase C (PKC)-₁ (PRKCB1), a target of both G-protein-mediated signaling and diacylglycerol, has been associated with the PFC and sensitization in rats.⁸ The heterotrimeric G₀ protein is the major noncytoskeletal protein in the growth cone membrane, where it is colocalized with and appears to be regulated by GAP43.^51,53 Disrupting the G₀ gene (GNA01) abolishes G₀ function and uncouples dopamine D2 receptors from their effectors,^54,55 while G₀₁ (GNB1) has been associated with the establishment of cocaine sensitization⁵². The activation/deactivation of G-proteins is mediated via the subunit by activating guanine nucleotide exchange factors (GEFs, GAP43) and inactivating GTPase-activating proteins (GAPs). PTPL1-associated RhoGAP 1 (PARG1)⁵⁶ is a strong negative regulator of the active form of the small Ras-related monomeric GTPase Rho, relaying signals from cell-surface receptors to, among others, the actin cytoskeleton. PARG1 shifts the balance towards an inactive Rho-GDP form, which increases axonal branching and dendritic stability. The active modulation of actin-cytoskeletal modulators was further noted by the strong regulation of transcripts encoding the actin-severing protein gelsolin (GSN) and actin-capping proteins beta (CAPZB) and profilin 1 (PFN1).

None of the nine named transcripts associated with dopamine neurotransmission were selected presently. Changes in parameters associated with an altered inhibitory modulation in response to cocaine^57,58 were, however, suggested by the differential expression of transcripts encoding the 67 kDa isoform of the GABA biosynthetic enzyme glutamate decarboxylase (GAD1), and a G-protein activated inwardly rectifying potassium channel (GIRK) protein (GIRK2/KCNJ6) thought to regulate inhibitory responses by decreasing neuronal excitability.^59,60

Energy Metabolism and Mitochondrial Function

A number of differentially regulated transcripts were indicative of changes in energy metabolism and mitochondrial function. The mitochondrion is the converging point of all cellular catabolic processes and the locus of the tricarboxylic acid (TCA) cycle, fueled by factors such as -enolase (ENO2) and MDH1. Oxidative phosphorylation and ATP generation were represented by the inner mitochondrial membrane cytochrome b–c₁ complex (cytochrome c-1, CYC1), the cytochrome oxidase complex (COX6C), and a mitochondrial ATP synthase (ATP5G3). The cytosolic antioxidant superoxide dismutase 1 (SOD1) additionally guards against mitochondrial oxidative damage and may mitigate activation of procaspases, such as caspase 4 (CASP4).

Oligodendrocyte-associated Transcripts

A number of oligodendrocyte-associated transcripts implicated changes relating to the function of this cell type. The major myelin constituent, PLP1, was regulated, as also found in human brain tissue from chronic alcohol abusers.^34,61 Notably, the two-fold downregulation of PLP1 in the frontal cortex of chronic alcohol abusers⁶¹ is reminiscent of the downregulation present for the two individuals with an alcohol coabuse history (COC#3, COC#7). Phosphatidate phosphohydrolase (PAPase, PPAP2c) converts phosphatidic acid to diacylglycerol, regulating the de novo synthesis of glycerolipids and phospholipase-mediated signal transduction.⁶²

Conclusion

The existence of at least two different, but consistent, gene expression patterns associated with cocaine abuse in the present study suggests that cocaine targets a distinct subset of genes. While we cannot conclusively attribute these differences to the mode, duration or recency of abuse, the two patterns may represent acute stimulation vs a refractory, destimulated state. Three of the four pattern A cocaine subjects, all of whom showed generally increased gene expression, had recently abused 'crack' cocaine, as indicated by the presence of AEME in the dlPFC. The increases in gene expression in the pattern A subjects may therefore represent a general state of activation, in contrast to the generally decreased expression of the same genes in pattern B. Experimental models that induce similar differential transcriptional changes may aid our understanding of the human condition of drug abuse. Emerging from this study is the cocaine-induced transcriptional regulation of cytoskeletal elements and signal transduction, synaptic transmission and neuroplasticity, energy metabolism and mitochondrial function, and oligodendrocytic function/myelination. Changes converging on the MEK/ERK pathway were especially noted, and it is conceivable that a host of up- and downstream modulatory inputs onto this pathway produce both acute and enduring changes in response to cocaine. By combining post-mortem studies with regional and temporal maps of gene expression from experimental studies, it may be feasible to associate cellular transcriptional signatures of activation, and deactivation, with different abuse patterns.

MATERIALS AND METHODS

Experimental Brain Samples

DlPFC from cocaine abusers (N=7) and controls with no history of drug abuse or mental illness (N=7) was obtained from the brain repository of the Clinical Brain Disorders Branch, NIMH. Anatomically, the dlPFC was defined as Area 46 rostral to the most anterior aspect of the corpus callosum. In an attempt to circumvent a number of ante- and post-mortem factors affecting brain RNA quantity and quality, we matched as closely as possible for age, gender, and post-mortem interval (PMI) (Table 1). A short agonal state was suggested by the cause of death and brain pH,⁶³ indicating that the samples were appropriate for use in microarray experiments.⁶⁴ Determination of brain pH was performed as described previously.⁶⁵ All control subjects tested negative for drugs of abuse and alcohol by post-mortem toxicology screens.

Microarray experiments

RNA extraction

As previously described,⁶⁶ aliquots of dlPFC brain powder were kept at -80°C until extraction of total RNA by a modified guanidinium thiocyanate method using the TRIzol extraction protocol (Life Technologies, Inc., Gaithersburg, MD, USA). Individual RNA extractions utilized in the study had a minimum OD_260/280 ratio >1.6. The RNA was further checked by formaldehyde agarose gel electrophoresis for evidence of degraded RNA and for the integrity of the 28S and 18S rRNA bands. No evidence of high molecular weight bands, representing potential DNA contamination, was found.

Neuroarray

The neuroarray (Figure 1) was produced by sorting a set of 15 000 human cDNA clones (http://image.llnl.gov/, IMAGE Consortium clones, Research Genetics, Huntsville, AL, USA) for CNS relevant genes, and including additional candidate genes identified in mRNA or protein screenings of patients with neuropsychiatric disorders.⁶⁷ The resulting gene list contained 1152 clones, representing functional families such as synaptic, neuronal, neuroglial and cell adhesion molecules, transcription factors, kinases, phosphatases, proteases, oncogenes, and structural genes.

RNA labeling and hybridization

Total RNA (5 g) from each individual was reverse transcribed in the presence of ³³P-dCTP to produce ³³P-labeled cDNA following a previously published protocol.⁶⁶ The reaction product was purified by size separation through a spin column (BioSpin, BioRad, CA, USA). Heat-denatured probe (5 10⁶ c.p.m.) was diluted in 4 ml Microhyb solution (Research Genetics), and hybridized to the arrays for 16–18 h at 50°C with rotation. Two posthybridization washes were performed with 2 SSC/0.1% SDS at room temperature to remove nonspecific hybridization, and the array was placed under Saran wrap and exposed to a low-energy phosphor screen (Molecular Dynamics, Sunnyvale, CA, USA). After a 7-day exposure, the phosphor screen was scanned in a Phosphoimager 860 (Molecular Dynamics) at 50 m resolution, and the pixel intensities quantified using ImageQuant (Molecular Dynamics).

QPCR analysis of gene expression

Reverse transcription of sample RNA for PCR

All cDNA and PCR-related reagents were purchased from Roche Molecular Biochemicals (Indianapolis, IN, USA) if not otherwise indicated. Reverse transcription (RT) of the sample RNA into cDNA was performed by incubating 24 l RNA sample with 32 l AMV buffer, 1 mM dNTP, 1 g random hexamers (Promega, Madison, WI, USA), 160 U RNAse inhibitor, 20 M DTT, and 100 U AMV RT for 1 h at 42°C. The reaction mix was subsequently purified using the HighPure PCR product purification kit.

Primer design

MacVector software (Oxford Molecular Group, Madison, WI, USA) was used to design custom primers targeting the region on genes-of-interest corresponding to that of the microarray clone. The primers were purchased from Gene Probe Technology (Gaithersburg, MD, USA). As the regulation of most, if not all, genes appear context dependent,⁶⁸ no gene can be universally ascribed the role of a standardizing, 'housekeeping' gene. Literature searches suggested 18S and cyclophilin as the best candidates for standardization.^32,69 Both transcripts concurred in experimental/control cDNA ratios across a number of samples and conditions in pilot experiments, verifying them as suitable internal standards for standardization of cDNA and, by inference, of starting RNA concentration. For this study, we used an 18S primer set (489 nt amplicon, QuantumRNA 18S Classic; Ambion, Austin, TX, USA) and cyclophilin (206 nt amplicon; atg gtc aac ccc acc gtg ttc ttc g (F); cgt gtg aag tca cca ccc tga cac (R)). The following primers (5'3') were used for selected genes-of-interest: GNAO1 (244 nt): aca gaa ata cac agc cgt cag tca c (F), ccc cag cag tag tca aaa tgt tcc (R); MDH1 (339 nt): tcc atc cat ccc caa gga gaa c (F), gag ttg cca tca gag ata aca ccc (R); MAP1B (151 nt): cag tca aga gca gca agg aaa tg (F), gga tgc cac aca atc aaa gat gag (R); PLP1 (263 nt): tgg att gtg ttt ctt tgg ggt g (F); cag atg gtg gtc ttg taa gtc gc (R).

Real-time quantitative PCR (QPCR)

QPCR was performed using the Roche LightCycler (software version 3.5) and the FastStart DNA Master SYBR Green I kit. Correct primer and MgCl₂-concentrations were determined in pilot experiments, and primer concentrations ranged from 0.5 to 1.0 M with a standard Mg²⁺-concentration of 4 M. Preparation of a general PCR master mix (MM) for all samples ensured testing under near-identical conditions. This consisted of 11.6 l DEPC-treated water, 2.4 l MgCl₂, and 2.0 l FastStart DNA master SYBR Green I per sample. To further ensure that comparable cDNA content was probed for both the 18S standard and the gene-of-interest, cDNA-specific MM were prepared for each individual cDNA sampled. Aliquots of 18 l cDNA MM were mixed with 2 l of the appropriate primer to a final reaction volume of 20 l per test, and a final 1 : 100 cDNA dilution. All PCR products were checked by melting curve analysis and by agarose gel electrophoresis to exclude the possibility of multiple products or incorrect product size.

GC-MS Analysis of Cocaine, Opiates, and Related Analytes in dlPFC Brain Tissue

Lyophilized dlPFC specimens were weighed (30–52 mg), suspended in sodium acetate buffer (2.0 M, pH 4.0), and internal standards (100 ng of [²H₁₁]-methamphetamine, [³H₁₀]-amphetamine, [²H₃]-COC, [²H₃]-EME, [²H₃]-CE, [²H₃]-BE, [²H₃]-norcocaine, [²H₃]-codeine, [²H₃]-morphine, and [²H₃]-6 acetyl morphine in 100 l of H₂O) added. Samples were vortex mixed, allowed to stand at room temperature for 30 min, and centrifuged. Analytes were isolated by solid phase extraction (Clean Screen DAU solid phase extraction columns, United Chemical Technologies, Bristol, PA, USA) and eluted with methylene chloride, 2-propanol, and ammonium hydroxide (80 : 20 : 2, v : v : v). A 20 l aliquot of N-methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide + 1% tert-butyldimethyl-chlorosilane (MTBSTFA+TBDMCS, Pierce Chemical, Rockford, IL, USA) was added, followed by evaporation under nitrogen at 40°C until dry. Residues were reconstituted in acetonitrile, transferred to autosampler vials, 20 l MTBSTFA+TBDMCS was added and extracts were heated at 80°C for 15–20 min. Then, 20 l of N,O-bis(trimethyl)trifluoroacetamide with 1% trimethylchlorosilane (BSTFA+TMCS, Pierce Chemical, Rockford, IL, USA) was added, and samples were heated at 80°C for 45 min. Analyzes were performed on a Hewlett-Packard 6890 gas chromatograph with a Hewlett-Packard 5973 mass selective detector (GC-MS). A capillary inlet system, operated in splitless mode, and an HP-1 fused-silica capillary column (15 0.2 mm² ID, 0.10 m film thickness) were employed for chromatographic separation. The mass selective detector was operated in a selected ion monitoring mode with positive chemical ionization. A 1 l aliquot was injected into the GC-MS. Analytes were quantified by comparing respective analyte-internal standard target ion response ratios to calibration curves prepared in drug-free brain homogenate. Limits of quantitation (LOQ) were 125 pg/mg. At the LOQ, signal-to-noise ratios for all ions were 3 and analytes quantified within 20% of target. Correlation coefficients were 0.985 from the LOQ (31–125 pg/mg) to 3125 pg/mg. Duplicate controls at 312.5 and 3125 pg/mg were required to quantify within 20% of target values.

Individual DAT and COMT Genotypes

Genotyping for a DAT VNTR polymorphism was performed according to a previously published protocol.⁷⁰ The Val158Met COMT polymorphism was determined by a 5' exonuclease allelic discrimination Taqman assay.⁷¹ The following primers and probes were custom designed using Primer Express (Applied Biosystems) software: forward primer: tcg aga tca acc ccg act gt; reverse primer: aac ggg tca ggc atg ca; Fam-labeled probe (Val allele): cct tgt cct tca cgc cag cga, VIC-labeled probe (Met allele): acc ttg tcc ttc atg cca gcg aaa t.

Data Analysis

Ratio-based comparisons

Z-ratio. The z-ratio was in essence obtained as described previously.⁶⁶ Pixel intensities were log₁₀-transformed to reduce the variance introduced by extreme values. All 1152 transcripts were included in the distribution. The normalized z-score for each individual membrane was calculated by subtracting the average array log intensity from the gene log intensity, and dividing with the standard deviation (SD) of the array grid. This yielded an z-score distribution with a mean value of 0 and an SD of 1. The z-difference, with respect to the cocaine-abusing individual, was obtained for each gene by subtracting the mean matched control gene z-score from the mean gene z-score for the cocaine-abusing individual. The z-ratio, expressed with respect to the cocaine-abusing individual, was then calculated by dividing the z-difference for each gene by the SD of the z-differences distribution. The z-ratio represents a normalized, log-transformed fold-like change for each gene.

Normalized ratios. Normalized ratios were used to evaluate the absolute change in expression level of individual genes between any two individuals (control and/or cocaine). This enabled additional information on specific individual aberrations and the effect of the normalizing individual. Normalized ratios were calculated as a ratio of the scaled mean raw intensity values.

Selection criteria for transcripts associated with differential expression in cocaine abusers. A two-tailed Student's t-test assuming equal variance of the two samples was performed for each cocaine–control comparison to identify statistically significant gene expression changes. Our null hypothesis was that, for each transcript, the z-scores for the cocaine subject and the respective matched control would not differ. Two criteria determined whether the null hypothesis was accepted (transcript not differentially expressed) or rejected (transcript differentially expressed) for each individual transcript. The null hypothesis was rejected for transcripts where the average P-value was less than 0.05. The second criteria demanded that minimally three of the seven cocaine–control comparisons displayed individual P-values <0.001. No correction was made for the use of multiple t-tests. It is argued, that for a rejection criteria of P>0.001, we would have expected significant differences in 1–2 genes (0.001 1152) to occur at random per comparison. The probability that a P<0.001 or less would occur for the same gene for three or more subject comparisons would consequently be small (P0.01 to find one gene with P<0.001 for any three comparisons). The t-tests examined the four repeated measurements (z-scores) from each individual as representative of a larger population of measurements of the same individuals. The results of the t-tests therefore apply to the generality of the current arrays as extrapolated to a larger population of arrays, rather than to the more usual circumstance wherein a t-test examines the generality of a set of results if extrapolated to a larger population of subjects.

Cluster analysis. Average-linkage hierarchical clustering (Cluster software;72) was used to organize the relative expression patterns of the transcripts into dendrograms, which were visualized in Treeview (Cluster/Treeview software: http://rana.lbl.gov/EisenSoftware.htm). These indicate both the degree of similarity (clustering) of transcriptional expression patterns in the individual cocaine–control comparisons as well as the clustering of individual transcripts.

Analysis of the QPCR Data

A standard curve for each primer was used to calculate the respective relative concentrations. Using the same cDNA mastermix for both the primers for the gene-of-interest and the standardizing gene, the median value of minimally three replicate samples for each primer was obtained. To correct for potentially different amounts of cDNA subject to amplification, the ratio of the mean concentrations of standardizing gene for the cocaine and the matched controls was calculated. This ratio was applied to correct the concentration of the gene-of-interest for the COC individual. The QPCR ratio was then calculated between the standardized values for the cocaine and matched control individual. A ratio below 1 reflects a downregulation, a ratio of 1 indicates no change, while a ratio above 1 indicates upregulation with respect to the cocaine-abusing individual.

A statistical approach was included to evaluate the statistical reliability of the standardized (cDNA-content corrected) raw expression values. The QPCR analyses were performed multiple times (usually eight replicates). It was therefore possible to test the statistical reliability of each cocaine–control matched comparison, with the exception of MDH1 for which there were only four replicates. A nonparametric test had to be employed, because absolute values differed between runs. Differences between each cocaine subject and the corresponding matched control were tested for each transcript using Wilcoxon matched-pairs signed-ranks tests.⁷³

Notes

Note added in proof: Levels of expression of selected genes, as measured by z-scores for the individual subjects, were examined for a possible correlation with brain pH. No relationship was found, confirming that brain pH differences were not a factor in the present study.

DUALITY OF INTEREST

None declared.

References

1. Vanderschuren LJMJ, Kalivas PW. Alterations in dopaminergic and glutamatergic transmission in the induction and expression of behavioral sensitization: a critical review of preclinical studies. Psychopharmacology (Berl) 2000; 151: 99–120. | Article | PubMed | ChemPort |
2. Wolf ME. The role of excitatory amino acids in behavioral sensitization to psychomotor stimulants. Prog Neurobiol 1998; 54: 679–720. | Article | PubMed | ISI | ChemPort |
3. Nestler EJ. Molecular basis of long-term plasticity underlying addiction. Nat Rev Neurosci 2001; 2: 119–128. | Article | PubMed | ISI | ChemPort |
4. Robinson T, Gorny G, Mitton E, Kolb B. Cocaine self-administration alters the morphology of dendrites and dendritic spines in the nucleus accumbens and neocortex. Synapse 2001; 39: 257–266. | Article | PubMed | ISI | ChemPort |
5. Striplin CD, Kalivas PW. Correlation between behavioral sensitization to cocaine and G protein ADP-ribosylation in the ventral tegmental area. Brain Res 1992; 579: 181–186. | Article | PubMed | ChemPort |
6. Kalivas PW, Duffy P. Time course of extracellular dopamine and behavioral sensitization to Cocaine. II. Dopamine perikarya. J Neurosci 1993; 13: 276–284.
7. Hope BT, Nye HE, Kelz MB, Self DW, Iadarola MJ, Nakabeppu Y et al. Induction of a long-lasting AP-1 complex composed of altered Fos-like proteins in brain by chronic cocaine and other chronic treatments. Neuron 1994; 13: 1235–1244. | Article | PubMed | ISI | ChemPort |
8. Steketee JD, Rowe LA, Chandler LJ. The effects of acute and repeated cocaine injections on protein kinase C activity and isoform levels in dopaminergic brain regions. Neuropharmacology 1998; 37: 339–347. | Article |
9. Ghasemzadeh MB, Nelson LC, Lu X-L, Kalivas PW. Neuroadaptations in ionotropic and metabotropic glutamate receptor mRNA produced by cocaine treatment. J Neurochem 1999; 72: 157–165. | Article | PubMed | ISI | ChemPort |
10. Persico AM, Schindler CW, O'Hara BF, Brannock MT, Uhl GR. Brain transcription factor expression: effects of acute and chronic amphetamine and injection stress. Mol Brain Res 1993; 20: 91–100.
11. Koob GF, Sanna PP, Bloom FE. Neuroscience of addiction. Neuron 1998; 21: 467–476. | Article | PubMed | ISI | ChemPort |
12. Stein EA, Fuller SA. Cocaine's time action profile on regional cerebral blood flow in the rat. Brain Res 1993; 626: 117–126.
13. Hammer Jr RP, Cooke ES. Gradual tolerance of metabolic activity is produced in mesolimbic regions by chronic cocaine treatment, while subsequent cocaine challenge activates extrapyramidal regions of rat brain. J Neurosci 1994; 14: 4289–4298. | PubMed | ChemPort |
14. Hadfield MG. Cocaine. Selective regional effects on central monoamines. Mol Neurobiol 1995; 11: 47–53.
15. Beyer CE, Steketee JD. Dopamine depletion in the medial prefrontal cortex induces sensitized-like behavioral and neurochemical responses to cocaine. Brain Res 1999; 833: 133–141.  | Article | PubMed |
16. Jentsch JD, Taylor JR. Impulsivity resulting from frontostriatal dysfunction in drug abuse: implications for the control of behavior by reward-related stimuli. Psychopharmacology (Berl) 1999; 146: 373–390.  | PubMed | ChemPort |
17. Breiter HC, Gollub RL, Weisskoff RM, Kennedy DN, Makris N, Berke JD et al. Acute effects of cocaine on human brain activity and emotion. Neuron 1997; 19: 591–611. | Article | PubMed | ISI | ChemPort |
18. Maas LC, Lukas SE, Kaufman MJ, Weiss RD, Daniels SL, Rogers VW et al. Functional magnetic resonance imaging of human brain activation during cue-induced cocaine craving. Am J Psychiatry 1998; 155: 124–126.  | PubMed | ISI | ChemPort |
19. Childress AR, Mozley PD, McElgin W, Fitzgerald J, Reivich M, O'Brien CP. Limbic activation during cue-induced cocaine craving. Am J Psychiatry 1999; 156: 11–18. | PubMed | ISI | ChemPort |
20. Garavan H, Pankiewicz J, Bloom A, Cho J-K, Sperry L, Ross TJ et al. Cue-induced cocaine craving: neuroanatomical specificity for drug users and drug stimuli. Am J Psychiatry 2000; 157: 1789–1798. | Article | PubMed | ISI | ChemPort |
21. Volkow ND, Fowler JS. Addiction, a disease of compulsion and drive: involvement of the orbitofrontal cortex. Cereb Cortex 2000; 10: 318–325. | Article | PubMed | ISI | ChemPort |
22. Bonson KR, Grant SJ, Contoreggi CS, Links JM, Metcalfe J, Weyl HL et al. Neural systems and cue-induced cocaine craving. Neuropsychopharmacology 2002; 26: 376–386.  | Article | PubMed | ISI | ChemPort |
23. Grant S, London ED, Newlin DB, Villemagne VL, Liu X, Contoreggi C et al.. Activation of memory circuits during cue-elicited cocaine craving. Proc Natl Acad Sci USA 1996; 93: 12040–12045. | Article | PubMed | ChemPort |
24. Grant S, Contoreggi C, London ED. Drug abusers show impaired performance in a laboratory test of decision making. Neuropsychologia 2000; 38: 1180–1187. | Article | PubMed | ISI | ChemPort |
25. Bechara A, Dolan S, Denburg N, Hindes A, Anderson SW, Nathan PE. Decision-making deficits, linked to a dysfunctional ventromedial prefrontal cortex, revealed in alcohol and stimulant abusers. Neuropsychologia 2001; 39: 376–389. | Article | PubMed | ISI | ChemPort |
26. Pierce RC, Reeder DC, Hicks J, Morgan ZR, Kalivas PW. Ibotenic acid lesions of the dorsal prefrontal cortex disrupt the expression of behavioral sensitization to cocaine. Neuroscience 1998; 82: 1103–1114. | Article | PubMed | ISI | ChemPort |
27. Everitt BJ, Parkinson JA, Olmstead MC, Arroyo M, Robledo P, Robbins TW. Associative processes in addiction and reward. The role of amygdala-ventral striatal subsystems. Ann NY Acad Sci 1999; 877: 412–438.  | Article | PubMed | ChemPort |
28. McFarland K, Kalivas PW. The circuitry mediating cocaine-induced reinstatement of drug-seeking behavior. J Neurosci 2001; 21: 8655–8663. | PubMed | ISI | ChemPort |
29. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Massanti CM, Straub RE et al. Effect of COMT Val108/158Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci USA 2001; 98: 6917–6922. | Article | PubMed | ChemPort |
30. Hitri A, Casanova MF, Kleinman JE, Wyatt RJ. Fewer dopamine transporter receptors in the prefrontal cortex of cocaine users. Am J Psychiatry 1994; 151: 1074–1076.
31. Mash DC, Pablo J, Ouyang Q, Hearn WL, Izenwasser S. Dopamine transport function is elevated in cocaine users. J Neurochem 2002; 81: 292–300. | Article |
32. Chen L, Segal DM, Moraes CT, Mash DC. Dopamine transporter mRNA in autopsy studies of chronic cocaine users. Mol Brain Res 1999; 73: 181–185.  | Article |
33. Little KY, McLaughlin DP, Zhang L, Livermore CS, Dalack GW, McFinton PR et al. Cocaine, ethanol, and genotype effects on human midbrain serotonin transporter binding sites and mRNA levels. Am J Psychiatry 1998; 155: 207–213. | PubMed | ISI | ChemPort |
34. Mayfield RD, Lewohl JM, Dodd PR, Herlihy A, Liu J, Harris RA. Patterns of gene expression are altered in the frontal and motor cortices of human alcoholics. J Neurochem 2002; 81: 802–813. | Article | PubMed | ISI | ChemPort |
35. Stevens JR. Enough of pooled averages: been there, done that. Biol Psychiatry 1997; 41: 633–635. | Article |
36. Mirnics K, Middleton FA, Marquez A, Lewis DA, Levitt P. Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex. Neuron 2000; 28: 53–67. | Article | PubMed | ISI | ChemPort |
37. Middleton FA, Mirnics K, Pierri JN, Lewis DA, Levitt P. Gene expression profiling reveals alterations of specific metabolic pathways in schizophrenia. J Neurosci 2002; 22: 2718–2729.  | PubMed | ISI | ChemPort |
38. Freeman WM, Brebner K, Lynch WJ, Robertson DJ, Roberts DCS, Vrana KE. Cocaine-responsive gene expression changes in rat hippocampus. Neuroscience 2000; 108: 371–380. | Article |
39. Freeman WM, Nader MA, Nader SH, Robertson DJ, Gioia L, Mitchell SM et al. Chronic cocaine-mediated changes in non-human primate nucleus accumbens gene expression. J Neurochem 2001; 77: 542–549. | Article | PubMed | ISI | ChemPort |
40. Young MR, Nair R, Bucheimer N, Tulsian P, Brown N, Chapp C et al. Transactivation of Fra-1 and consequent activation of AP-1 occur extracellular signal-regulated kinase dependently. Mol Cell Biol 2002; 22: 587–598. | Article | PubMed | ISI | ChemPort |
41. Valjent E, Corvol JC, Pages C, Besson MJ, Maldonado R, Caboche J. Involvement of the extrallular signal-regulated kinase cascade for cocaine-rewarding properties. J Neurosci 2000; 20: 8701–8709. | PubMed | ISI | ChemPort |
42. Beitner-Johnson D, Guitart X, Nestler EJ. Neurofilament proteins and the mesolimbic dopamine system: common regulation by chronic morphine and chronic cocaine in the rat ventral tegmental area. J Neurosci 1992; 12: 2165–2176. | PubMed | ChemPort |
43. Guitart X, Beitner-Johnson D, Marby DW, Kosten TA, Nestler EJ. Fischer and Lewis rat strains differ in basal levels of neurofilament proteins and their regulation by chronic morphine in the mesolimbic dopamine system. Synapse 1992; 12: 242–253.
44. Garcia-Sevilla JA, Ventayol P, Busquets X, La Harpe R, Walzer C, Guimon J. Marked decrease of immunolabelled 68 kDa neurofilament (NF-L) proteins in brains of opiate addicts. Neuroreport 1997; 8: 1561–1565. | PubMed |
45. Bunnemann B, Terron A, Zantedeschi V, Merlo Pich E, Chiamulera C. Chronic nicotine treatment decreases neurofilament immunoreactivity in the rat ventral tegmental area. Eur J Pharmacol 2000; 393: 249–253. | Article |
46. Ferrer-Alcon M, Garcia-Sevilla JA, Jaquet PE, La Harpe R, Riederer BM, Walzer C et al. Regulation of nonphosphorylated and phosphorylated forms of neurofilament proteins in the prefrontal cortex of human opioid addicts. J Neurosci Res 2000; 61: 338–349.  | Article | PubMed |
47. Ulloa L, Dombradi V, Diaz-Nido J, Szucs K, Gergely P, Friedrich P et al. Dephosphorylation of distinct sites on microtubule-associated protein MAP1B by protein phosphatases 1, 2A and 2B. FEBS Lett 1993; 330: 85–89.  | Article |
48. Strack S, Westphal RS, Colbran RJ, Ebner FF, Wadzinski BE. Protein serine/threonine phosphatase 1 and 2A associate with and dephosphorylate neurofilaments. Brain Res Mol Brain Res 1997; 49: 15–28.  | Article |
49. Robinson TE, Kolb B. Alterations in the morphology of dendrites and dendritic spines in the nucleus accumbens and prefrontal cortex following repeated treatment with amphetamine or cocaine. Eur J Neurosci 1999; 11: 1598–1604.  | Article | PubMed | ISI | ChemPort |
50. Nestler EJ, Terwilliger RZ, Walker JR, Sevarino KA, Duman RS. Chronic cocaine treatment decreases levels of the G protein subunits G_i alpha and G_o in discrete regions of rat brain. J Neurochem 1990; 55: 1079–1082. | Article | PubMed | ChemPort |
51. Strittmatter SM, Valenzuela D, Sudo Y, Linder ME, Fishman MC. An intracellular guanine nucleotide release protein for G0. GAP-43 stimulates isolated alpha subunits by a novel mechanism. J Biol Chem 1991; 266: 22465–22471. | PubMed | ISI | ChemPort |
52. Wang X-B, Funada M, Imai Y, Revay RS, Ujike H, Vandenbergh DJ et al. rG₁: a psychostimulant-regulated gene essential for establishing cocaine sensitization. J Neurosci 1997; 17: 5993–6000. | PubMed | ISI | ChemPort |
53. Strittmatter SM, Valenzuela D, Kennedy TE, Neer EJ, Fishman MC. G₀ is a major growth cone protein subject to regulation by GAP-43. Nature 1990; 344: 836–841.  | Article | PubMed | ISI | ChemPort |
54. Jiang M, Gold MS, Boulay G, Spicher K, Peyton M, Brabet P et al. Multiple neurological abnormalities in mice deficient in the G protein Go. Proc Natl Acad Sci USA 1998; 95: 3269–3274. | Article | PubMed | ChemPort |
55. Jiang M, Spicher K, Boulay G, Wang Y, Birnbaumer L. Most central nervous system D₂ dopamine receptors are coupled to their effectors by G₀. Proc Natl Acad Sci USA 2001; 98: 3577–3582.  | Article | PubMed | ChemPort |
56. Saras J, Franzen P, Aspenstrom P, Hellman U, Gonez LJ, Heldin CH. A novel GTPase-activating protein for Rho interacts with a PDZ domain of the protein-tyrosine phosphatase PTPL1. J Biol Chem 1997; 272: 24333–24338. | Article | PubMed | ChemPort |
57. Ikeda M, Dohi T, Tsujimoto A. Inhibition of gamma-aminobutyric acid release from synaptosomes by local anesthetics. Anesthesiology 1983; 58: 495–509.
58. Karler R, Calder LD, Thai DK, Bedingfield JB. The role of dopamine and GABA in the frontal cortex of mice in modulating a motor-stimulant effect of amphetamine and cocaine. Pharmacol Biochem Behav 1998; 60: 237–244. | Article | PubMed | ChemPort |
59. Kofuji P, Davidson N, Lester HA. Evidence that neuronal G-protein-gated inwardly rectifying K+ channels are activated by G beta gamma subunits and function as heteromultimers. Proc Natl Acad Sci USA 1995; 92: 6542–6546. | PubMed | ChemPort |
60. Slesinger PA, Stoffel M, Jan YN, Jan LY. Defective gamma-aminobutyric acid type B receptor-activated inwardly rectifying K+ currents in cerebellar granule cells isolated from weaver and Girk2 null mutant mice. Proc Natl Acad Sci USA 1997; 94: 12210–12217. | Article | PubMed | ChemPort |
61. Lewohl JM, Wang L, Miles MF, Zhang L, Dodd PR, Harris RA. Gene expression in human alcoholism: microarray analysis of frontal cortex. Alcohol Clin Exp Res 2000; 24: 1873–1882. | Article | PubMed | ISI | ChemPort |
62. Kanoh H, Kai M, Wada I. Molecular characterization of the type 2 phosphatidic acid phosphatase. Chem Phys Lipids 1999; 98: 119–126. | Article |
63. Harrison PJ, Kleinman JE. Methodological issues. In: Harrison PJ, Roberts GW (eds). The Neuropathology of Schizophrenia. Oxford University Press: New York, 2000, pp 339–350.
64. Barton AJ, Pearson RC, Najlerahim A, Harrison PJ. Pre- and postmortem influences on brain RNA. J Neurochem 1993; 61: 1–11. | Article | PubMed | ChemPort |
65. Romanczyk TB, Shannon Weickert C, Webster MJ, Herman MM, Akil M, Kleinman JE. Alterations in trkB mRNA in the human prefrontal cortex throughout the lifespan. Eur J Neurosci 2002; 15: 269–280. | Article | PubMed | ChemPort |
66. Vawter MP, Barrett T, Cheadle C, Sokolov BP, Wood WH, Donovan DM et al. Application of cDNA microarrays to examine gene expression differences. Brain Res Bull 2001; 55: 641–650. | Article | PubMed | ISI | ChemPort |
67. Barrett T, Cheadle C, Wood WB, Teichberg D, Donovan DM, Freed WJ et al. Assembly and use of a broadly applicable neural cDNA microarray. Restor Neurol Neurosci 2001; 18: 127–135.
68. Lee PD, Sladek R, Greenwood CMT, Hudson TJ. Control genes and variability: absence of ubiquitous reference transcripts in diverse mammalian expression systems. Genome Res 2001; 12: 292–297. | Article |
69. Goidin D, Mamessier A, Staquet M-J, Schmitt D, Berthier-Vergnes O. Ribosomal 18S RNA prevails over glyceraldehyde-3-phosphate dehydrogenase and -actin genes as internal standard for quantitative comparison of mRNA levels in invasive and noninvasive human melanoma cell subpopulations. Anal Biochem 2001; 295: 17–21. | Article | PubMed |
70. Vandenbergh DJ, Persico AM, Hawkins AL, Griffin CA, Li X, Jabs EW et al. Human dopamine transporter gene (DAT1) maps to chromosome 5p15.3 and displays a VNTR. Genomics 1992; 14: 1104–1106. | Article | PubMed | ISI | ChemPort |
71. Livak KJ, Flood SJ, Marmaro J, Giusti W, Deetz K. Allelic discrimination using fluorogenic probes and 5' nuclease assay. Genet Anal 1999; 14: 143–149. | Article | PubMed | ISI | ChemPort |
72. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998; 95: 14863–14868. | Article | PubMed | ChemPort |
73. Siegel S, Castellan Jr NJ. The Wilcoxon signed ranks test. Nonparametric Statistics or the Behavioral Sciences, 2nd edn. McGraw-Hill: New York 1998, pp 87–95.

Acknowledgements

The authors gratefully acknowledge the expert assistance of Diane Teichberg and Chris Cheadle (DNA Array Unit, NIA IRP), and by Aaron Russell (BIS, NIDA IRP).