Alcohol-responsive genes in the frontal cortex and nucleus accumbens of human alcoholics

Authors

  • Traute Flatscher-Bader,

    1. Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, The University of Queensland, Queensland, Australia
    Search for more papers by this author
    • Traute Flatscher-Bader and Marcel van der Brug contributed equally to this research.

  • Marcel van der Brug,

    1. Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, The University of Queensland, Queensland, Australia
    Search for more papers by this author
    • Traute Flatscher-Bader and Marcel van der Brug contributed equally to this research.

    • The present address of Marcel van der Brug is Laboratory of Neurogenetics, National Institute on Ageing, NIH, 9000 Rockville Pike, Bethesda, MD 20892, USA.

  • John W Hwang,

    1. Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, The University of Queensland, Queensland, Australia
    Search for more papers by this author
  • Peter A Gochee,

    1. Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, The University of Queensland, Queensland, Australia
    Search for more papers by this author
  • Izuru Matsumoto,

    1. Department of Neuropsychiatry, School of Medicine, Fukushima Medical University, Fukushima, Japan
    Search for more papers by this author
    • The present address of Izuru Matsumoto is Department of Pathology, University of Sydney, NSW 2006, Australia.

  • Shin-ichi Niwa,

    1. Department of Neuropsychiatry, School of Medicine, Fukushima Medical University, Fukushima, Japan
    Search for more papers by this author
  • Peter A Wilce

    1. Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, The University of Queensland, Queensland, Australia
    Search for more papers by this author

Address correspondence and reprint requests to Traute Flatscher-Bader, Department of Biochemistry and Molecular Biology, School of Molecular and Microbial Sciences, University of Queensland, QLD 4072, Australia.

Abstract

The molecular processes underlying alcohol dependence are not fully understood. Many characteristic behaviours result from neuroadaptations in the mesocorticolimbic system. In addition, alcoholism is associated with a distinct neuropathology. To elucidate the molecular basis of these features, we compared the RNA expression profile of the nucleus accumbens and prefrontal cortex of human brain from matched individual alcoholic and control cases using cDNA microarrays. Approximately 6% of genes with a marked alcohol response were common to the two brain regions. Alcohol-responsive genes were grouped into 11 functional categories. Predominant alcohol-responsive genes in the prefrontal cortex were those encoding DNA-binding proteins including transcription factors and repair proteins. There was also a down-regulation of genes encoding mitochondrial proteins, which could result in disrupted mitochondrial function and energy production leading to oxidative stress. Other alcohol-responsive genes in the prefrontal cortex were associated with neuroprotection/apoptosis. In contrast, in the nucleus accumbens, alcohol-responsive genes were associated with vesicle formation and regulation of cell architecture, which suggests a neuroadaptation to chronic alcohol exposure at the level of synaptic structure and function. Our data are in keeping with the previously reported alcoholism-related pathology characteristic of the prefrontal cortex, but suggest a persistent decrease in neurotransmission and changes in plasticity in the nucleus accumbens of the alcoholic.

Abbreviations used
APOD

apolipoprotein D

APP

amyloid beta A4 precursor protein

BA9

Brodmann area 9

GAPD

glyceraldehyde 3-phosphate dehydrogenase (EC 1.2.1.13)

IKKB

inhibitor of nuclear factor kappa B kinase beta subunit

ITPKA

inositol-triphosphate 3-kinase A (EC 2.7.1.127)

LPIN1

lipin1

MAPT

microtubule-associated protein tau

MDK

midkine

NA

nucleus accumbens

PFC

prefrontal cortex

PI3 kinase

phosphatidylinositol-3 kinase (EC 2.7.1.153)

PMP22

peripheral myelin protein 22

RT-PCR

real-time PCR

SLC1A3

excitatory aminoacid transporter 1

SYT1

synaptotagmin 1

TIMP3

metalloproteinase inhibitor 3

WASF1

Wiskott–Aldrich syndrome protein family member 1

The mechanisms by which alcohol exerts its effects in the brain have only been partially defined. At least some of the long-term effects are a result of neuroadaptive changes at the cellular level, probably involving an enduring change in gene expression. Although studies of individual candidate genes have linked aspects of alcohol's action to changes in expression of these genes, they have failed to fully elucidate the scope of the genetic cascade following ethanol exposure. Application of cDNA microarrays in which the expression of thousands of genes can be studied simultaneously can expand our understanding of the molecular effects of chronic alcohol abuse. Micorarrays provide an overview of the genes that are involved in the processes of tolerance, dependence, alcohol-induced cellular injury and recovery.

The mesocorticolimbic dopamine system, consisting of the ventral tegmental area, the nucleus accumbens (NA), the amygdala, the septum and the prefrontal cortex (PFC), is considered the prime target for drugs of abuse and the origin of many addictive behaviours (Bassareo et al. 2003; Ericson et al. 2003). The NA is thought to mediate the rewarding effects of addictive drugs, including the development and maintenance of positive reinforcement (Czachowski et al. 2001; Besheer et al. 2003). Adaptive changes in this region may underlie psychological and behavioural changes generated by chronic alcohol use such as craving, but the exact molecular targets in the human alcoholic brain remain unclear. There is evidence that polymorphisms in selected dopamine receptors and transporters may contribute to alcohol susceptibility and withdrawal severity (Hietala et al. 1997; Heinz and Goldman 2000; Hutchison et al. 2002; Wernicke et al. 2002). However, the development of alcohol tolerance and dependence is likely to involve more than just the dopaminergic receptor/transporter system.

Alcoholics often show impaired cognitive functions such as planning ability (Ratti et al. 2002), especially in speed and flexibility (Pfefferbaum et al. 2001). These deficits may reflect damage to the PFC, the brain region responsible for many of these higher brain functions. Loss of white matter volume in the frontal regions of older alcoholics (Harper et al. 1985), along with significant neuronal loss in grey matter, has been observed (Kril and Harper 1989; Kril et al. 1997). In contrast, there have been no reports of alcohol-related morphological abnormalities in the NA.

In this study, we have identified alcohol-responsive genes in the NA and the PFC, key areas of the mesocorticolimbic system. Our aim was to identify alcohol-responsive gene sets common to the two brain regions and those that demonstrate a region-selective response. In this way, delineation of the region-selective, alcohol-responsive cellular systems may provide an insight into the basis of alcohol's action in the two tissues.

cDNA microarrays were used to compare the expression profiles of the two regions in individual alcoholic and control cases. The changes in expression of selected genes were confirmed by real-time PCR (RT-PCR). We identified approximately 230 genes with a marked change in expression within these two brain regions. However, 6% of these genes were common to both regions. When the alcohol-responsive genes were assigned to functional groups, a clear region-specific pattern was apparent. This suggests that alcoholism may be associated with adaptive changes specific to each brain region.

Materials and methods

Case selection and classification

Cases were classified into two groups: alcoholics (A: > 100 g of ethanol/day) and controls (C: low alcohol consumption < 20 g of ethanol/day). All cases had full autopsy with diagnostic neuropathological examination. All of the alcoholic patients fulfilled the criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM IV-R) of the American Psychiatric Association for alcohol dependence. A full medical history, including duration and quantity of alcohol consumption, was accessible. None of the alcoholic patients used in this study showed any confounding neurological disorders. The non-alcoholic control groups had no episodes of neurological or psychiatric disorder. A summary of case information for alcoholics and controls is presented in Table 1. Numbers indicate pairings for microarray analysis. Brains were frozen and stored at − 80°C. Grey matter from Brodmann area (BA9) was dissected following criteria previously described (Rajkowska and Goldman-Rakic 1995). Approximately 50 mg of tissue were taken at random from a 500 mg sample of grey matter. The NA was located in a brain slice containing the caudate putamen (DeArmond et al. 1989) and a 100 mg punch biopsy of the NA was randomly dissected. Approval from the Central Sydney Area Health Service (Protocol no. X03-0117) and the Ethics Committee of the University of Sydney (Ref. no. 95/2/7) was obtained for the use of post-mortem brain samples in this study.

Table 1. Case information. Gender, age, post-mortem interval (PMI) and the cause of death of alcoholic and control cases. Numbers indicate cases paired for microarray analysis
DiagnosisBrain regionSexAge (years)PMI (h)Cause of death
PFCNA
Controls11M6013Acute myocardial infarction
22F5430Sudden death
33M5029Ischaemic heart disease
4 M7047Myocardial infarct
5 F7515Cardiac arrest
6 M4318Cardiac respiratory arrest
4M3711Pulmonary embolism
5M6811Suicide
6M4917Cardiac failure
Alcoholics11M6913Sudden death
22F4627Epilepsy
3 M5024Upper gastrointestinal haemorrhage
4 M4235Carbon monoxide poisoning
53M5619Cardiac failure
64M5767Alcoholic pancreatitis
5M7015Myocardial infarct
6M4624Alcohol toxicity

RNA extraction and aRNA synthesis

Total RNA was extracted from 50 mg frozen tissue with 1 mL TRIzol (Invitrogen, Melbourne, Victoria, Australia) following the manufacturer's instructions. Samples were suspended in RNAse/DNAse-free water, maintaining concentrations above 1 µg/µL, then frozen at – 80°C. Anti-sense RNA (aRNA) was synthesised by linear amplification (Eberwine et al. 1992). Briefly, 1 µg total RNA with 1 µg T7-oligo dT21 primer (5′-TCTAGTCGACGGCCAGTGAATTGTAATACGACTCACTATAGGGCGT21-3′) in 5 µL was incubated at 70°C for 4 min then chilled on ice. Next, 1 µL SuperScript II (200 U, GibcoBRL, Melbourne, Australia), 2 µL 100 mm dithiothreitol (DTT), 4 µL SuperScript II 1st strand buffer, 1 µL RNase inhibitor (40 U, GibcoBRL), 1 µL 10 mm dNTPs (Sigma, Sydney, Australia) and 1 µL H2O were added. Samples were incubated at 42°C for 1 h then at 65°C for 15 min. To this, 1 µL ribonuclease H (2 U, Promega, Sydney, New South Wales, Australia), 1 µL Escherichia coli DNA ligase (5 U, New England Biolab, Brisbane, Queensland, Australia), 5 µL E. coli DNA polymerase I (40 U, GeneWorks, Adelaide, South Australia, Australia), 3 µL 10 mm dNTPs, 15 µL 2nd strand buffer [200 mm Tris pH 6.9, 46 mm MgCl2, 900 mm KCl, 100 mm (NH4)2SO4, 1.5 mm NAD] and 105 µL H2O were added. After incubation at 16°C for 2 h, 2 µL T4 DNA polymerase (10 U, New England Biolab) were added; samples were incubated at 16°C for 15 min and heat-inactivated at 70°C for 10 min. Double-stranded cDNA was extracted with 150 µL phenol : chloroform (1 : 1) and washed three times with 500 µL Rnase-free H2O in a Microcon YM-30 (Millipore, Sydney, New South Wales, Australia) filter. The cDNA was recovered and the volume adjusted to 8 µL. To the concentrated cDNA, 2 µL each of ATP, GTP, CTP, UTP (75 mm), 10× transcription buffer and MEGAscript T7 enzyme mix (Ambion, Adelaide, South Australia, Australia) were added and the solution incubated at 37°C for 4–16 h to allow in vitro transcription. The aRNA was extracted with TRIzol according to the manufacturer's instructions.

cDNA microarray hybridization

To 6 µg aRNA, 5 µg random primers were added in 20 µL with Rnase-free H2O. Samples were incubated at 70°C for 10 min, then chilled on ice. To each sample, 1.5 µL (300 U) Superscript II, 8 µL 5 × 1st strand buffer, 4 µL 100 mm DTT, 1 µL RNase inhibitor, 1 µL 25 mm d(GAT)TP, 2 µL 2.5 mm dCTP, and either 2 µL FluoroLink Cy3-dCTP or 2 µL FluoroLink Cy5-dCTP (Amersham Biosciences, Sydney, New South Wales, Australia), were added. Samples were incubated at 42°C for 1 h. RNA was hydrolysed by the addition of 1 µL 0.5 m EDTA (pH 8) and 2 µL 2 m NaOH, followed by incubation at 65°C for 10 min. Solutions were neutralized by the addition of 4 µL 1 m HCl, 4 µL 1 m Tris/HCl (pH 8) and 100 mm sodium acetate. cDNA samples were combined and purified though a QIAQuick PCR purification column (Qiagen, Melbourne, Victoria, Australia) according to the manufacturer's instructions. To the eluent, 10 µL (10 µg) human Cot-1 DNA (GibcoBRL) were added and the volume reduced to 22 µL with a Microcon YM-30 filter (Millipore). To the concentrated cDNA, 16 µL 20× saline sodium citrate (SSC), 40 µL deionized formamide and 2 µL 20% (w/v) sodium dodecyl sulfate (SDS) were added. Samples were incubated at 95°C for 5 min, then at 45°C for 60 min, then chilled on ice. The cDNA microarrays were manufactured by the University Health Network (UHN) Microarray Centre at the Ontario Cancer Institute, Canada. Arrays were robotically printed on CMT-GAPS slides (Corning Inc., Acton, MA, USA). The Human 19K version 2.2 array contained 19200 I.M.A.G.E. Consortium clones spotted in duplicate over two slides. Clones were sequenced and verified by the UHN Microarray Centre. Published sequences are available at http://www.microarrays.ca/. aRNA from the PFC and NA of alcoholics (six cases) were paired with controls (six cases), best-matched for age and post-mortem delay and hybridized on individual arrays (Table 1). The experiment was repeated and the dye channels were exchanged to minimize dye bias. The microarray slides were incubated overnight for 18 h at 42°C. Slides were washed in 1× SSC/0.1% (w/v) SDS, 0.5× SSC, 0.2× SSC, sequentially, at 50°C for 10 min, with gentle agitation and then centrifuged at 400 g for 5 min to remove residual wash solution. Slides were immediately scanned at 10 µm resolution on a Genetic Microsystems scanner using GMS Scan Array software (Genetic Microsystems, Inc., Woburn, MA, USA). Each pixel was assigned a value between 0 and 65 535 for each channel based on signal fluorescence.

Spot identification and quantification

Spots were identified using the ImaGene 4.1 (BioDiscovery, Inc., El Segundo, CA, USA) software platform. Data were interpreted using GeneSpring 4.1 software (Silicon Genetics, Redwood City, CA, USA). Using the median fluorescence signal for each spot, a LOWESS curve was fitted to the log-intensity versus log-ratio plot. Twenty percent of the data were used to calculate the LOWESS fit at each point. This curve was applied to adjust the control value for each measurement. GeneSpring's Cross-Gene Error Model was used to assess variability across replicate hybridizations.

Transcripts were accepted for analysis only if the median signal intensities were at least two standard deviations above the experimental or control background signals and, in addition, the median control signal was also above the control strength cut-off as determined using the GeneSpring Cross-Gene Error Model. Data points were selected for further analysis if the overall difference was statistically significant (a one sample t-test p-value of less than 0.05 with the ‘Benjamini and Hochberg false discovery rate’ multiple testing correction applied) when the variance between individual cases was considered. For subsequent analysis, genes with a mean fold change > 1.5 or < 0.7 were selected and annotated.

Real-time PCR validation of selected gene transcripts

Differential expression of selected transcripts was assessed by RT-PCR, using glyceraldehyde 3-phosphate dehydrogenase (GAPD) as a reference. Primers were designed on Primer3 software (http://www-genome.wi.mit.educgi-binprimerprimer3_http://www.cgi). The genes selected for RT-PCR and their corresponding sequences are presented in Table 2. Primers for GAPD were: forward CTCTCTGCTCCTCCTGTTCGAC and reverse CAATACGACCAAATCCGTTGACT. All reactions were performed using the QuantiTech SYBR Green kit (Qiagen) on the ABI Model 7700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) and the following cycling conditions: step 1, 95°C for 15 min; step 2, 95°C for 15 s, 60°C for 30 s, 72°C for 30 s. Step 2 was repeated for 45 cycles. Each sample was assayed in triplicate. The production of the specific amplicon was demonstrated for each primer set by including a no template control containing only enzymes, dye and primer. Moreover, a melt curve analysis was performed following amplification. The formula 2–ΔΔCT was used for relative quantification, where ΔCT represents the difference in threshold cycles between the target and reference gene in any one sample and ΔΔCT, the difference in ΔCT in alcoholic and control samples (ABI User Bulletin 1997). To assess the validity of changes in gene expression patterns observed in the microarrays by RT-PCR, pooled samples of the aRNA were used. For selected genes, individual samples were assayed and additional cases were introduced (n = 9–12).

Table 2. Sequences of primer pairs used for RT-PCR. Abbreviation of gene names: APP, amyloid beta A4 precursor protein; ITPKA, inositol-triphosphate3-kinase A (EC 2.7.1.127); SLC1A3, excitatory aminoacid transporter 1; LPIN1, lipin1; MAPT, microtubule-associated protein tau; MDK, midkine; PMP22, peripheral myelin protein 22; SYT1, synaptotagmin 1; TIMP3, metalloproteinase inhibitor 3; WASF1: Wiskott–Aldrich syndrome protein family member 1
RegionForward primerReverse primer
SLC1A3ATGGGAATGCGAGCTGTAGTGCCTTCTCTGTGCATGTTTTC
MDKCCAAGAAAGGGAAGGGGAAAAGAGCTAACGAGCACAGAAGG
TIMP3GGATGTGACAACAACTTCCAAACTTCTGGGTTTCAGGACAGC
MAPTGCAGACCTGGGACTTTAGGGACGGCAGACAACAGCACA
ITPKACCCAGAGCCAAATGACACTATGAGACGTTAGACCGGAAGG
WASF1TAAATGACCCTTTTCCTCCATAACCTCTAACAACAGCTCCATGC
LPIN1TCCACAGTCTTCCCTGTTCCTCAATGGGCTGGACTCTTTC
PMP22CCTCCCAGTCCACCTCATTTGGGCATTTTGTCCGTGTG
APPTGGAAGAGGTGGTTCGAGAGACATCCGCCGTAAAAGAATG
SYT1AACATGGGGTTGGCTGTTTCGGCAGACGGTTATTTTCCT

Results

Amplification of 2 μg total RNA typically produced 50–80 μg aRNA. When this aRNA was separated on a denaturing formaldehyde gel and stained with ethidium bromide, all amplified samples demonstrated a diffuse band of high molecular weight RNA. Samples were then transferred to a nylon membrane and probed for GAPD. Samples that exhibited a single non-diffused GAPD signal of the same molecular weight as evident in total RNA control samples were used for array analysis (data not shown).

The reliability of each array experiment was judged by the normal distribution of the signals from all genes on the array. The initial comparison between the NA and PFC revealed that, of the 19 200 expressed sequence tags (ESTs) on the array, expression of approximately 80% was detected in one or both brain regions. In the NA, 1649 clones were identified as differentially expressed, 2246 in the PFC (p > 0.05). Of all clones, 67% mapped to characterized genes and 33% represented uncharacterized transcripts or hypothetical proteins. These gene lists are available at http://www.smms.uq.edu.au/dLoads/wilce/microarray.xls. Gene lists were filtered by applying a cut-off point of either > 1.5 or < 0.7 mean fold change for subsequent analysis. All genes identified as alcohol-responsive were annotated, based upon sequence alignments to publicly accessible sequence data. Within the PFC, 68 alcohol-responsive genes were known while in the NA, 51 were known genes. The remaining annotated transcripts (111 total in both regions) mapped to hypothetical proteins, ESTs and uncharacterized cDNA clones. In both regions, the majority of alcohol-responsive genes were down-regulated (36 in the NA and 47 in PFC).

The expression of 13 genes was altered in both the NA and the PFC (Table 3). Several genes that were marginally outside the arbitrary selection criterion (> 1.5-fold change) have been included in this list. The alcohol-responsive genes identified as common to both areas are representative of a variety of different functional groups. Expression of peripheral myelin protein 22 mRNA, a major component of myelin, is down-regulated in both tissues whereas expression of the cytoskeletal proteins, tubulin beta 5 and tau, is up-regulated. Levels of mRNA for the glycoprotein, clusterin, were increased in both brain areas. There were also six genes showing differential expression in both regions that could not be assigned a clear function.

Table 3. Differentially expressed genes common to both the NA and PFC. Values are taken from microarray analysis and represent mean ratios of alcoholic cases compared with matched control cases (n =6). The common name for the gene has been included along with the GenBank accession number for the sequenced clone spotted on the microarray
Brain regionDescriptionGenBank
NAPFC
1.5691.708Microtubule-associated protein tau BM725941
1.4761.553Hematopoietic PBX-interacting protein W07077
1.4671.486Clusterin BQ045101
1.4661.431Oligonucleotide/oligosaccharide-binding fold containing 1 W38731
1.4481.734SAM domain and HD domain, 1 BG536068
1.4321.428Tubulin, beta, 5 BG753663
1.4151.418CGI 07 protein N29213
0.6540.680Peripheral myelin protein 22 AA128253
0.5890.347Similar to cytochrome c oxidase III BE874451
0.5480.667Zinc finger protein 434 (cervical cancer suppressor 5) H93281
1.4331.511Hypothetical protein AL133206 N49623
1.4331.433EST N76969
1.4501.512EST, Weakly similar to putative p150 N95721

Ten up- or down-regulated genes with a range of alcohol response and representative of a variety of functional groups were chosen for verification using RT-PCR (Table 4). In each case there was a tight correlation between data obtained from the two methods of analysis. Excitatory aminoacid transporter 1 (SLC1A3), midkine (MDK), synaptotagmin 1 (SYT1) and metalloproteinase inhibitor 3 (TIMP3) were selected for more detailed examination using additional alcoholic cases (Fig. 2). These analyses confirmed the array data.

Table 4. Comparison of gene expression of selected genes using microarray and RT-PCR. Abbreviation of gene names: APP, amyloid beta A4 precursor protein; ITPKA, inositol-triphosphate3-kinase A (EC 2.7.1.127); SLC1A3, excitatory aminoacid transporter 1; LPIN1, lipin1; MAPT, microtubule-associated protein tau; MDK, midkine; PMP22, peripheral myelin protein 22; SYT1, synaptotagmin 1; TIMP3, metalloproteinase inhibitor 3; WASF1, Wiskott–Aldrich syndrome protein family member 1
Gene nameGenBankaMicroarraybRT-PCRcRegion
Ratio p-value
  1. a GenBank refers to the GenBank accession number for the sequenced clone spotted on the microarray. b The mean expression level in alcoholic cases relative to appropriate controls (n = 6) was determined from microarray analysis. c For Real-time PCR the mean relative expression levels of pooled alcoholic cases and pooled control cases are given, each using GAPD as a reference gene. The p-value represents the significance between the mean replicate determinations using the formula 2 ΔΔ CT.

SLC1A3 R59684 2.3102.4430.0052PFC
MDK W19684 1.9352.4080.0055PFC
TIMP3 BQ044940 1.6431.9850.00013PFC
MAPT BM725941 1.5691.4840.046NA
ITPKA N48505 1.5881.9380.024PFC
WASF1 N95702 1.5871.5340.015PFC
LPIN1 N45520 1.5641.5870.0013NA
PMP22 AA128253 0.6800.7040.012PFC
PMP22 AA128253 0.6540.7140.0047NA
APP BM841483 0.5720.4990.0043NA
SYT1 N92197 0.5620.6060.006NA
Figure 2.

RT-PCR on individual alcoholic and control cases. Cases were assayed individually by RT-PCR. Additional alcoholic cases were included (controls n = 6; alcoholics: TIMP3, SLC1A3 MDK n = 12, SYT: n = 9). The data are expressed as mean 2–ΔΔCT± SEM, where 2–ΔΔCT represents the difference in threshold cycles between target and reference gene (GAPD) in any one sample (*p > 0.05; **p > 0.01, Student's t-test). Abbreviations of gene names: TIMP3: metalloproteinase, SLC1A3: excitatory aminoacid transporter 1, member 3, MDK: midkine 1, SYT1: synaptotagmin 1.

Annotated genes with a clearly defined function were arranged into functional categories reflecting 11 major cellular processes (Fig. 1, Tables 5 and 6). The largest group of alcohol-responsive genes in the PFC comprised DNA-binding proteins. Predominant in this group were seven zinc finger proteins that included several with unknown function. Interestingly, four of the five genes with a putative transcription repressor function were down-regulated.

Figure 1.

Functional grouping of genes. Microarray analysis was performed, filtering out characterized genes that show a significantly differential expression of greater than 1.5 or less than 0.7 in the alcoholic PFC and NA, respectively. These genes were placed into 11 functional groupings based on the most prominent function for the gene product. Gene grouping in the NA (a) and PFC (b).

Table 5. Differentially expressed genes in the PFC in major functional groups. The common name of the gene has been given together with the GenBank accession number for the sequenced clone spotted on the microarray. Values are taken from microarray analysis and represent mean ratios of alcoholic cases compared with matched control cases (n = 6)
Functional GroupDescriptionAccessiont-test p-valueRatio
DNA bindingRPA-binding trans-activator N32669 8E-031.72
MUS81 endonuclease homologue N99229 7E-031.69
Zinc finger protein, subfamily 1 A, 1 (Ikaros) W01226 0.021.67
Promyelocytic leukaemia W38850 0.011.59
Polymerase (DNA directed), delta 2, regulatory subunit (50 kDa) W01850 0.041.52
Zinc finger and BTB domain containing 11 (ZBTB11) N53706 2E-030.70
Upstream binding protein 1 (LBP-1a) H63585 0.050.68
Zinc finger protein 24 (KOX 17) AA134770 1E-030.67
Moderately similar to Zinc finger protein 91 N53915 2E-030.66
CCAAT/enhancer binding protein (C/EBP), gamma AA055333 1E-030.65
Cervical cancer suppressor gene 5 (ZNF 434) H93281 4E-040.67
Translational/protein traffickingRibosomal protein L22 W15194 1E-031.77
Peter Pan homologue (Drosophila) BQ051603 3E-031.63
Ribosomal protein S29 BQ028504 0.051.51
FUS interacting protein (serine-arginine rich) 1 BM014323 6E-030.69
Proteasome (prosome, macropain) subunit, beta type, 2 BQ050102 5E-030.68
MetabolismUridine phosphorylase BQ017463 9E-031.68
Uroporphyrinogen decarboxylase BQ064110 0.030.70
Alcohol dehydrogenase 6 (class V) W92014 7E-030.69
Arylacetamide deacetylase (esterase) N99791 0.020.69
Phosphomannosemutase domain   
containing: BM32A mRNA for PMMLP N79332 0.020.69
Thioredoxin domain-containing N78351 0.020.66
Pipecolic acid oxidase H66019 0.020.70
Aconitase 1, soluble AA114871 5E-030.67
Acetyl-CoA hydrolase (CACH-1), cytostolic AA002136 0.010.66
Lysosomal-associated membrane protein 1 N42285 1E-030.63
Ribulose-5-phosphate-3-epimerase (RPE) AL711113 0.020.70
Mitochondria/energyNADH dehydrogenase (ubiquinone) 1 BM468661 0.010.68
Electron-transfer-flavoprotein, alpha polypeptide (glutaric aciduria II) AA130253 9E-030.66
Mitochondrial import inner membrane translocase subunit TIM 9B R92073 0.030.65
Cytochrome c oxidase polypeptide I BM888296 2E-030.60
cytochrome c oxidase polypeptide II BE873873 3E-050.54
NADH-ubiquinone oxidoreductase chain 2 AV751900 3E-050.53
ATP synthase 6 BM794983 1E-050.40
Cytochrome B AV711588 3E-070.38
Similar to cytochrome c oxidase III BE874451 9E-050.35
NADH-ubiquinone oxidoreductase chain 4 BM782955 7E-060.28
SignallingSAM domain and HD domain, 1 BG536068 7E-051.73
Inositol-triphosphate3-kinase A N48505 3E-031.59
Haematopoietic PBX-interacting W07077 2E-031.55
Regulator of G-protein signalling 8 R85247 0.030.69
Channels/ Neuro-transmissionExcitatory aminoacid transporter 1 R59684 0.052.31
Gamma-aminobutyric acid (GABA) B receptor, 1 W07715 2E-031.55
Chloride channel 4 AA071201 9E-050.69
Zinc transporter ZTL1 like AA134752 2E-030.66
Potassium voltage-gated channel, Isk-related family, member 3 AA127801 3E-030.65
MyelinationPeripheral myelin protein 22 AA128253 1E-030.68
Secreted phosphoprotein 1 R97904 8E-030.65
Apolipoprotein D AA131720 4E-030.61
Structural, adhesion, migrationMicrotubule-associated protein tau (MAPT) BM725941 0.041.71
Wiskott–Aldrich syndrome protein family member 1 N95702 5E-031.59
Syndecan-2 BM971009 0.031.65
Vesicle related, transportAmphiphysin 1 W01422 9E-040.69
Growth arrest, stress, apoptosisMidkine W19684 6E-031.94
Metalloproteinase inhibitor 3 BQ044940 1E-041.64
BCL2/adenovirus E1B 19 kDa interacting protein 3 BQ002402 0.020.70
Nuclear mitogen- and stress-activated protein kinase-1 (MSK1) AW968068 7E-030.69
Sestrin 2 H51912 0.010.68
CDC14 cell division cycle 14 homologue A (S. cerevisiae) AA002016 3E-030.68
CD33 antigen (gp67) BI910568 2E-030.63
CA11(gastrokine 1) AA099387 3E-030.63
Immune response, protectionInhibitor of nuclear factor kappa B kinase beta subunit BQ014475 0.021.51
Thioredoxin AA150817 3E-051.50
Stromal cell-derived factor 1 W02528 2E-040.70
Leupaxin W00660 1E-040.67
Major histocompatibility complex, class II, DR beta 4 H50622 2E-030.64
Progesterone-induced blocking factor 1 AA099685 8E-030.64
Major histocompatibility complex, class II, DR alpha AA026536 5E-030.60
Table 6. Genes differentially expressed in the alcoholic NA in major functional groups. Differentially expressed genes of the PFC tied into major functional groups. The common name for the gene has been given together with the GenBank accession number for the sequenced clone spotted on the microarray. Values are taken from microarray analysis and represent mean ratios of alcoholic cases compared with matched control cases (n = 6)
Functional groupDescriptionAccession t-test p-valueRatio
DNA bindingRAD51 A homologue (RecA homologue, E. coli) (S. cerevisiae) N72527 0.042.04
Putative homeodomain transcription factor 1 N92222 2E-030.68
Modulator recognition factor protein 2, MRF1-like BM968725 5E-030.67
Zinc finger protein 387 N47518 6E-050.60
Translational/ protein traffickingProteasome (prosome, macropain) subunit, beta type, 6 BQ052212 9E-040.67
SECIS binding protein 2 W58640 5E-040.66
Ubiquitin-conjugating enzyme E2 variant 2 BI856490 8E-030.66
MetabolismAcyl-Coenzyme A oxidase 1, palmitoyl H63016 0.051.92
Phosphoenolpyruvate carboxykinase 1 (soluble) R86197 0.051.60
Lipin 1 N45520 2E-031.56
Pyruvate dehydrogenase phosphatase N47861 6E-030.70
Fatty acid-Coenzyme A ligase, long-chain 3 N76698 0.010.68
Mannosidase, alpha, class 2 A, member 1 AA115560 9E-040.67
UDP-Gal:betaGlcNAc beta 1,4-   
galactosyltransferase, polypeptide 1 W58426 0.020.64
Carbonic anhydrase II H23302 0.020.62
Mitochondria/energySimilar to cytochrome c oxidase III BE874451 3E-030.59
SignallingRho-related BTB domain containing 3 BM907252 0.032.34
Epithelial membrane protein 1 T75112 6E-031.71
Calcium/calmodulin-dependent serine protein kinase (MAGUK family) N92643 0.051.51
Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase   
activation protein, eta polypeptide AA127798 0.020.68
Protein kinase, lysine deficient 1 H89618 0.040.63
Highly similar to HUMB2CHIM Homo sapiens beta2-chimaerin BG563170 0.020.61
Channels/ neuro-transmissionSolute carrier family 3 member 2 BM771973 0.011.69
NMDA receptor-associated protein 1 GRINA R73412 0.021.51
Contactin associated protein-like 2 N53741 2E-030.69
MyelinationPeripheral myelin protein 22 AA128253 5E-030.65
Structural, adhesion, migrationESTs, weakly similar to I37356 epithelial microtubule-associated protein N40823 0.021.65
Microtubule-associated protein tau (MAPT) BM725941 0.051.57
Collagen, type IV, alpha 1 AA150402 0.020.68
Integrin, alpha 6 H16046 0.030.70
Integrin beta 1 binding protein (melusin) 2 AA134808 0.030.69
cerebral cell adhesion molecule 1 AA137135 0.010.68
metalloproteinase inhibitor 2 BG542006 3E-030.67
Ninjurin 2 H91351 0.040.66
KIAA0923 (afadin- and alpha-actinin-binding protein, APID) H20898 7E-0.30.65
Tubulin, alpha 1 (testis specific) W33064 3E-040.63
Sparc/osteonectin, cwcv and kazal-like domains proteoglycan (testican) AL566877 8E-030.63
Junctional adhesion molecule 3 H23302 0.010.63
Fasciculation and elongation protein zeta 1 (zygin I): AA131602 6E-030.63
ESTs, Weakly similar to S41161 keratin 9, cytoskeletal – human [H.sapiens] BI916651 4E-030.61
Schwannomin interacting protein 1 R35596 1E-020.59
Amyloid beta (A4) precursor protein BM841483 5E-030.57
Vesicle related, transportVacuolar protein sorting 26 (yeast) T94596 1E-030.65
Lysosomal-associated membrane protein 2 R13247 0.040.58
Synaptotagmin 1 N92197 6E-030.56
Immune response, protectionImmunoglobulin lambda locus N92095 0.011.60
Properdin p factor, complement H52352 0.031.59
Lymphocyte adaptor protein H77460 6E-031.67
Interleukin 1 receptor, type I W03241 0.051.50
Similar to T-lymphoma invasion and metastasis inducing protein 1 AL533992 9E-30.66
Complement component 4 binding protein, beta H78141 0.050.48

Genes grouped under metabolism are representative of a variety of metabolic pathways and show no consistent theme. These are predominantly down-regulated in the PFC. Similarly, mitochondrial gene expression was down-regulated in the PFC of the alcoholic cases.

Unique to the PFC are changes in genes associated with growth arrest, stress, apoptosis and immune response/protection. The majority of these genes were down-regulated, although there was a marked increase in expression of three important genes: inhibitor of nuclear factor kappa B kinase beta subunit (IKKB), a key activator of necrosis factor kappa B (NF-κB), the neurotrophic factor midkine (MDK) and metalloproteinase inhibitor 3 (TIMP3). The latter two genes encode proteins involved in neuroprotection.

Expression of genes associated with myelination, apolipoprotein D, peripheral myelin protein 22 and the secreted phosphoprotein 1, was decreased in the PFC. Although markedly down-regulated in the PFC in three of the six alcoholic cases, the response of genes encoding other critical constituents of myelin, e.g. myelin basic protein and myelin-associated oligodendrocyte basic protein, was variable and did not reach significance when all replicates were considered.

In the NA, the largest group of alcohol-responsive genes comprised genes involved in maintenance of cytoarchitecture, cell adhesion and migration. The mRNA level of microtuble-associated protein tau was up-regulated while levels of the remaining 12 genes were significantly decreased. Amongst these down-regulated genes were integrin α6, the integrin binding protein β1, collagen, tubulin and schwannomin interacting protein, which are involved in regulation of cell architecture. In addition, expression of the putative modulators of axonal outgrowth, zygin 1 and testican, were down-regulated. A significant decrease in the mRNA level of synaptotagmin 1, a gene crucial for neurotransmitter release, was noted in the NA.

Discussion

Human alcoholics have an extended history of alcohol abuse spanning decades and incorporate periods of abuse and abstinence. Alcoholics may be at increased risk of relapse even after extended periods of abstinence. One challenge is to identify the stable changes in brain structures and molecular mechanisms that underlie these behaviours. In this study, we have compared the expression profile of two distinct brain regions that play key roles in the development and maintenance of dependence. The PFC plays an important role in motivation, planning, attention and the general temporal organization of behaviours. The NA is involved in reaction to natural reinforcement such as food and drink but also plays a crucial role in drug responses.

When the alcohol-responsive genes were assigned functional groups, there was a distinct regional difference. This immediately suggests that the brain regions responded to chronic alcohol exposure very differently. However, alcohol-responsive genes common to both regions included peripheral myelination protein 22, a constituent of myelin, which was down-regulated, and the glycoprotein clusterin, which was elevated. Clusterin is increased after brain injury (Schauwecker et al. 1998), in senile plaques of Alzheimer patients (Duguid et al. 1989) and in the brain of Down syndrome cases (Kida et al. 1995). It remains unclear whether clusterin's role in these pathologies is neurotrophic or toxic (Jones and Jomary 2002).

Six genes encoding transcription factors were down-regulated in the PFC, four of which were identified as suppressors of transcriptional activity. This may be indicative of a broader reprogramming of transcription associated with prolonged alcohol exposure. Increased expression of the DNA-directed polymerase δ, the RPA-binding trans-activator and the DNA repair gene, MUS81 endonuclease homologue is a strong indication of an enhanced response to DNA damage in the PFC. The damage may be a result of oxidative stress which is known to induce a suite of repair processes (Gonthier et al. 2004; Iida et al. 2004). There is over expression of a homologue to Rad51 A, a protein essential for DNA repair (Lio et al. 2003), in the NA of the alcoholics. This observation, coupled with the changes in clusterin and peripheral myelin protein 22, suggests that this brain region, although spared from major pathology, is also affected by the neurotoxic effects of alcohol.

We noted a down-regulation in nuclear genes encoding mitochondrial proteins in the PFC consistent with the results of recent expression profiling of the temporal cortex of the alcoholic (Sokolov et al. 2003). Several studies have reported ethanol-related alterations in mitochondrial structure and protein expression in both brain and liver, which have been interpreted as an indication of an underlying alcohol-associated defect in mitochondrial function (Schilling and Reitz 1980; Tavares and Paula-Barbosa 1983; Morvai and Ungvary 1987; Thayer and Rottenberg 1992; Cunningham and Bailey 2001). Since the brain is highly energy-dependent, any mitochondrial deficit may have profound effects on neuronal function in the PFC of the alcoholic and contribute to oxidative damage and the cellular pathology in this brain region.

It has been suggested that apoptosis may play a role in alcoholism-related cell loss in the human PFC (Freund 1994). A variety of proteins involved in the apoptotic cascade were expressed in the cortex of alcohol-treated rats (Rajgopal et al. 2003). In the human PFC, we noted the over expression of TIMP3, which has been associated with angiogenesis and neurogliomas (Huang et al. 2000; Qi et al. 2003). TIMP3 may also function to induce apoptosis since in neuronal cell culture, it facilitates activation of the apoptotic pathway (Wetzel et al. 2003). In contrast, there is elevation of several genes, which may function to inhibit apoptosis and thus represent a protective reaction. A key mediator of apoptosis is NF-κB, which is modulated by IKKB and thioredoxin (Hayashi et al. 1993; Liou and Baltimore 1993; Hirota et al. 1999). Expression of both these modulating genes is elevated in the PFC and suggests activation of protective mechanisms. The elevated levels of thioredoxin may also be indicative of a protective response to oxidative stress (Andoh et al. 2002). The neurotrophic factor, MDK, is expressed by reactive astrocytes around the site of nerve damage in the ischemic brain of rat and human (Satoh et al. 1993; Yoshida et al. 1995; Wada et al. 2002). It promotes the outgrowth of neurones in development (Li 1990) and inhibits apoptosis via activation of the PI3 kinase/ERK pathway in neuronal cell culture (Owada et al. 1999). The highly induced expression of MDK in the PFC may be an additional neuroprotective response.

Our data revealed a consistent down-regulation of the myelination genes apolipoprotein D, peripheral myelin protein 22 and secreted phosphoprotein 1 in the PFC of each alcoholic case. Levels of mRNA for myelin basic protein and myelin-associated oligodendrocyte basic protein were highly variable. In a previous study, there was a down-regulation in major myelin-associated proteins in two pooled samples of PFC from alcoholic cases (Lewohl et al. 2000). However, in a later study, up-regulation of myelin-associated oligodendrocyte basic protein was reported and the changes of other major myelin components were inconsistent between the pools (Mayfield et al. 2002). It has been argued that perturbation of myelination-associated gene expression is not unique to alcoholism but is common to other psychiatric disorders, including schizophrenia (Hakak et al. 2001) and mood disorders (Sokolov et al. 2003) that have a high rate of co-morbidity with alcoholism. In addition, smoking, which also has a high coincidence with alcoholism (Burling and Ziff 1988; DiFranza and Guerrera 1990; Batel et al. 1995), may effect expression and contribute to the variation.

Dysregulation of myelin-related gene expression accompanied by a loss of myelin-producing oligodendrocytes is an effect on the NA associated with cocaine abuse (Albertson et al. 2004). Although peripheral myelin protein 22 expression was reduced in the NA of the alcoholics, down-regulation of myelin-related genes was not a prominent feature. It would appear that demyelination is not a feature common to drug action on the NA.

Synaptotagmin 1, which acts as a calcium sensor and plays a critical role in neurotransmitter release (Bommert et al. 1993; Greengard et al. 1993), is down-regulated in the NA. Ethanol acutely increases extracellular levels of dopamine in the rat NA (Di Chiara and Imperato 1988) and dopamine release in the human NA (Boileau et al. 2003). In animals, these acute effects result in neuroadaptation and markedly reduced activity of the mesolimbic system, persisting for several days (Thielen et al. 2004). Down-regulation of vesicle transport and release would be consistent with a similar neuroadaptation in the human dopaminergic system to chronic alcohol consumption.

Strikingly, the expression of a number of genes involved in the organization of cellular architecture was altered in the NA. These included the up-regulation of microtubule-associated protein tau and the down-regulation of tubulin α1, integrin β1 binding protein and collagen, type IV, α1. Neuronal outgrowth and adhesion is mediated by integrin α6β1 (Emsley and Hagg 2003) and testican. A down-regulation of integrin α6β1 inhibits neuronal outgrowth (Ivins et al. 2004) while a decrease in the expression level of the chondroitin sulfate proteoglycan testican leads to neurone adhesion and extension (Marr and Edgell 2003). In addition, we noted down-regulation of a group of hypothetical and novel proteins with putative structural function. Schwannomin interacting protein 1 was down-regulated on three different cDNA clones on the microarray. As a binding partner of merlin, it may be involved in reorganization of the actin skeleton (Gronholm et al. 2003). The afadin- and α-actinin-binding protein has a putative function in the formation of cell–cell junctions (Asada et al. 2003) while zygin 1, a protein kinase ξ substrate, modulates axonal outgrowth in vitro (Kuroda et al. 1999). Structural changes in neuronal morphology, as the end point of neuroadaptation, have been reported in animal models after chronic drug exposure. Repeated cocaine treatment changes dendritic morphology by increasing branch points and spine density in the NA and neocortex (Robinson and Kolb 1999; Robinson et al. 2001), and opiate exposure reduces the number and area of dopaminergic neurones in the ventral tegmental area (Sklair-Tavron et al. 1996). Our data point to a change in neuronal plasticity, which may reflect persistent neuroadaption in the NA of the human alcoholic.

This current comparative study reveals distinct regional effects of alcohol. In the NA, a striking observation was the down-regulation of genes encoding key proteins involved in vesicular transport and cellular architecture, suggesting that an important component of alcoholism might be a persistent deficit in synaptic transmission and changes in plasticity. As the NA is a key mediator in the action of other drugs of abuse, it will be important to determine if the changes seen in the alcohol-exposed brain are common to other addictive substances. Another important question to be addressed is the drug-related changes in other regions of the mesocorticolimbic system, particularly the ventral tegmental area. Clearly, documenting the adaptive changes occurring in the human mesocorticolimbic system in the alcoholic brain may provide important insights into the molecular mechanism underlying the development of dependence.

Acknowledgements

The work was partially funded by a grant from The Philip Morris External Grant Program to Fukushima Medical University. TF-B is a recipient of an Australian Postgraduate Award from the Commonwealth Government of Australia.

Ancillary