Advertisement
Research Article

A Factor Analysis of Global GABAergic Gene Expression in Human Brain Identifies Specificity in Response to Chronic Alcohol and Cocaine Exposure

  • Mary-Anne Enoch equal contributor mail,

    equal contributor Contributed equally to this work with: Mary-Anne Enoch, Basel Baghal

    maenoch@niaaa.nih.gov

    Affiliation: Laboratory of Neurogenetics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland, United States of America

    X
  • Basel Baghal equal contributor,

    equal contributor Contributed equally to this work with: Mary-Anne Enoch, Basel Baghal

    Affiliation: Laboratory of Neurogenetics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland, United States of America

    X
  • Qiaoping Yuan,

    Affiliation: Laboratory of Neurogenetics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland, United States of America

    X
  • David Goldman

    Affiliation: Laboratory of Neurogenetics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland, United States of America

    X
  • Published: May 22, 2013
  • DOI: 10.1371/journal.pone.0064014

Abstract

Although expression patterns of GABAergic genes in rodent brain have largely been elucidated, no comprehensive studies have been performed in human brain. The purpose of this study was to identify global patterns of GABAergic gene expression in healthy adults, including trans and cis effects in the GABAA gene clusters, before determining the effects of chronic alcohol and cocaine exposure on gene expression in the hippocampus. RNA-Seq data from ‘BrainSpan’ was obtained across 16 brain regions from postmortem samples from nine adults. A factor analysis was performed on global expression of 21 GABAergic pathway genes. Factor specificity for response to chronic alcohol/cocaine exposure was subsequently determined from the analysis of RNA-Seq data from postmortem hippocampus of eight alcoholics, eight cocaine addicts and eight controls. Six gene expression factors were identified. Most genes loaded (≥0.5) onto one factor; six genes loaded onto two. The largest factor (0.30 variance) included the chromosome 5 gene cluster that encodes the most common GABAA receptor, α1β2γ2, and genes encoding the α3β3γ2 receptor. Genes within this factor were largely unresponsive to chronic alcohol/cocaine exposure. In contrast, the chromosome 4 gene cluster factor (0.14 variance) encoding the α2β1γ1 receptor was influenced by chronic alcohol/cocaine exposure. Two other factors (0.17 and 0.06 variance) showed expression changes in alcoholics/cocaine addicts; these factors included genes involved in GABA synthesis and synaptic transport. Finally there were two factors that included genes with exceptionally low (0.10 variance) and high (0.09 variance) expression in the cerebellum; the former factor was unaffected by alcohol/cocaine exposure. This study has shown that there appears to be specificity of GABAergic gene groups, defined by covariation in expression, for response to chronic alcohol/cocaine exposure. These findings might have implications for combating stress-related craving and relapse.

Introduction

Gamma-aminobutyric acid (GABA) is the principal inhibitory neurotransmitter in the vertebrate central nervous system (CNS). It is estimated that approximately 40% of mammalian CNS synapses have GABA receptors of which there are two classes: GABAA receptors that are ligand-gated, chloride ion channels that confer fast synaptic inhibition, and metabotropic GABAB receptors that are coupled to downstream calcium and potassium channels by G proteins. GABA is involved in CNS development and organization, learning and memory, anxiety and addiction [1]. GABAA receptors undergo allosteric modulation by several drugs including ethanol, benzodiazepines (BZs), barbiturates, anesthetics and endogenous neurosteroids [1].

GABAA receptors are composed of five subunits, each of which has several isoforms (α1–6, β1–3, γ1–3, δ, ε, θ, π, ρ1–3) [2]. Most receptors consist of two α, two β and one γ subunits [3]. Preclinical studies have identified three main subunit combinations: α1β2γ2 (60%), α2β3γ2 (15%), and α3β3γ2 (10%) [4][5]. Studies in rodents and non-human primates have shown that the expression of the various subunit isoforms varies across brain locations and during development. The receptor subunit composition determines distinct pharmacological and electrophysiological properties [6]. It has been shown that in non-human primates, the anxiolytic and muscle relaxant effects of BZ-like drugs appear to involve receptors including α2, α3 and α5 subunits whereas the ataxic and sedative properties of BZ-like drugs are mediated by α1 subunits [7][8]. Receptors including α4 and α6 subunits are BZ insensitive.

Genes for 14 of the 19 GABAA receptor subunit isoforms are clustered in four chromosomal regions: 4p13–q11 (α2, α4, β1, γ1), 5q34–q35 (α1, α6, β2, γ2), 15q11–q13 (α5, β3, γ3), and Xq28 (α3, θ,ε) [2]. Alternative splicing has been demonstrated for most of the GABAA receptor subunit isoforms indicating complexity in gene expression [9][10]. There is considerable conservation and linkage disequilibrium extending across long stretches of sequence in the chromosome (chr) 4 and 5 clusters [1]. This extensive and conserved pattern of gene clustering was established early in vertebrate evolution, after divergence from the invertebrates [11][12]. It is thought that the four gene clusters originated from duplications of (and within) an ancestral gene cluster containing single genes encoding one α, one β and one γ subunit [13].

A key question is whether the expression of genes within each chromosomal cluster are correlated (cis effects) and whether there is correlation of gene expression between GABAergic genes that are located in different regions (trans effects). In order to address this issue we obtained RNA-Seq data from postmortem samples of 16 brain regions from nine healthy men and women aged 18–40 who died suddenly. The RNA-Seq data was obtained from BrainSpan, a publicly available resource (http://www.brainspan.org/). In addition to the GABAA subunit genes found in several gene clusters, we included genes encoding the two GABAB receptor subunits (GABBR1, GABBR2), genes encoding proteins involved in GABA synthesis (GAD1, GAD2) and genes encoding proteins involved in GABA regulation (GPHN, SLC6A1 and PRAF2). A factor analysis was performed to identify patterns of cis and trans correlations in gene expression.

The second aim of the study was to determine whether some of the factors (groups of genes defined by covariation in expression in healthy individuals) identified from the BrainSpan RNA-Seq data, might show changes in expression in response to chronic exposure to alcohol and cocaine whereas other factors might be unaffected by exposure to alcohol or drug. In order to do this we looked at gene expression data from an earlier RNA-Seq study in which we had identified changes in GABAergic gene expression in human postmortem samples of hippocampus from alcoholics, cocaine addicts and controls [14][15].

Materials and Methods

Databases

There were two sources of RNA-Seq data used in this study; a public resource for factor analyses based on cross-regional influences on GABAergic gene expression and a publicly accessible resource from our lab on effects of alcohol and cocaine exposure on hippocampal gene expression.

BrainSpan brain samples.

The publicly available data used in this study was derived from BrainSpan, an atlas of the developing human brain, accessed via the Allen Institute for Brain Science’s hosting website http://www.brainspan.org/. Using RNA sequencing, BrainSpan provides unbiased coverage of the complete transcriptome in 16 targeted cortical and subcortical structures (Table 1) in 42 brain specimens to date, spanning pre-natal development to adulthood in both males and females. Details pertaining to sample and tissue acquisition and qualification, processing and dissection, RNA analysis, sequencing and alignment, quality control as well as expression quantification can be found in the BrainSpan Technical White Paper 2011 [16].

thumbnail

Table 1. Distinct cortical and subcortical substructures of the adult brain available in BrainSpan for gene expression studies.

doi:10.1371/journal.pone.0064014.t001
BrainSpan ethics statement.

All work was performed according to guidelines for the research use of human brain tissue and with approval by the Human Investigation Committees and Institutional Ethics Committees of the participating institutes.

Only brains with no evidence of malformations or lesions at the macroscopic or microscopic levels were included. All brain tissue samples in BrainSpan met specific selection criteria to ensure consistency and limit potential variation due to disease conditions. In particular, specimens were excluded if excessive drug or alcohol abuse was reported, if the individual had any known neurological or psychiatric disorders or if any prolonged agonal conditions were reported. Other excluding factors included ingestion of neurotoxic substances at the time of death, suicide, severe head injury, significant hemorrhages, prominent vascular abnormalities, tumors, prominent brain lesions, stroke, congenital neural abnormalities, and signs of neurodegeneration. For further details see BrainSpan Technical White Paper 2011 [16].

Table 2 lists details of the nine adult (aged 18–40 years) postmortem brain specimens (samples A – J) that are the focus of this study. Information about ethnicity was not available.

thumbnail

Table 2. The nine adult postmortem brain specimens utilized in this study.

doi:10.1371/journal.pone.0064014.t002
Hippocampal samples from alcoholics, cocaine addicts and controls.

Postmortem brain tissue was provided by the University of Miami Brain Bank. Since we used publicly available pathological specimens, our study was exempt from NIH Institutional Review Board (IRB) review. RNA-Seq data from postmortem hippocampus of eight alcoholics, eight cocaine addicts and eight controls were previously generated by our lab [14][15] demonstrating neuroadaptive changes. This RNA-Seq data has been deposited in the GenBank database (accession nos. SRA029279 and SRA029275). The GABAergic gene expression data [15] was used to determine whether there might be specificity in the association of any of the six factors, derived from the BrainSpan dataset, with changes in expression in response to chronic alcohol and cocaine exposure.

GABAergic Genes

The study focused on investigating the expression profiles of genes involved in the GABAergic pathway, predominantly genes encoding the GABAA receptor subunits (Table 3). With the exception of the δ and ρ3 subunits, genes encoding the GABAA receptor subunits are clustered in groups of two to five on chromosomes 4, 5, 6, 15, and X. In addition, we also investigated the expression profiles of genes encoding proteins involved in GABA synthesis (GAD1, GAD2), neuronal re-uptake of GABA at the synaptic cleft (SLC6A1), GPHN that encodes a key protein involved in the regulation of synaptic localization of GABAA receptors [17] and PRAF2, encoding the PRAF2 protein that is enriched in synaptic vesicles and is thought to be involved in the regulation of vesicular trafficking and neurotransmitter release [18][19].

thumbnail

Table 3. GABAergic genes included in this study.

doi:10.1371/journal.pone.0064014.t003

Analyses

This study utilized the RNA-Seq data obtained via the BrainSpan “RNA-Seq summarized to genes” downloadable archive file which contains normalized expression values and meta-data. The archive consists of RPKM (Reads Per Kilobase of exon model per Million mapped reads) values for each gene measured in each of the collected brain structures from each sample. After the archive was downloaded and uncompressed, the relevant information (genes and samples of interest) was extracted and prepared/organized using simple Perl commands. Next, the data was imported into the R package for statistical computing which was used for all subsequent analysis. Box plots were used to visualize expression profiles both sample by sample and gene by gene. Scatter plots and linear regressions were used to visualize correlations in expression which was quantified using the correlation coefficient R2. With the exception of the box plots which consistently show Log2-transformed RPKM values, no data manipulation was performed.

Validation analysis in hippocampal samples from two sources.

We have previously used RNA-Seq to quantify mRNA transcripts in postmortem total hippocampus from eight male controls, mean age = 37.5 (SD = 6.1) years obtained from the University of Miami Brain Bank. Full details are available elsewhere [14][15]. We compared the expression of the GABAergic genes in our eight hippocampal samples with gene expression in the nine BrainSpan hippocampal samples.

Factor analysis.

An exploratory factor analysis was performed using the original BrainSpan gene expression values for the 21 GABAergic genes that were expressed in the 16 brain regions. The factor analysis was executed with R version 2.12.1 using the psych (Procedures for Psychological, Psychometric, and Personality Research) package version 1.2.4. We made the decision to include factors with ≥0.05 of the variance. Therefore six factors were extracted that accounted for 0.86 of the variance in total GABAergic gene expression. The fitting method was principal axis factoring (PAF) and the rotation method was set to varimax (orthogonal rotation) since we did not expect the factors to be correlated.

Results

Our analyses were performed on high quality RNA-Seq data acquired from BrainSpan via the Allen Institute for Brain Science’s hosting website http://www.brainspan.org/together with high quality RNA-Seq data from our earlier study on postmortem hippocampal samples from alcoholics, cocaine addicts and controls [14][15]. Figure S1 compares mean (S.E.) expression of the 21 GABAergic genes in hippocampal samples from the controls in our earlier study [15] with the BrainSpan hippocampal samples. It should be noted that the read counts were independently normalized across the two datasets and therefore gene expression levels can only be compared within each dataset and not between the two datasets. Nevertheless Figure S1 shows that the pattern of relative gene expression was virtually the same in the two datasets, thus validating the methodology.

GABRR1, GABRR2, GABRR3, GABRP and GABRE respectively encoding the GABAA receptor subunits ρ1, ρ2, ρ3, π and ε were barely expressed, if at all, across all 16 brain regions and were therefore excluded from further study.

As shown in Figure 1, the global RPKM expression levels across all the GABAergic genes varied across individuals: sample A (male, 18yrs) had the lowest and sample C (female, 21 years) had the highest global expression values. Since sample A had data available for only 13/16 brain regions, box plots excluding these 3 regions were generated for all 9 samples and yielded the same order of expression level as seen in Figure 1.

thumbnail

Figure 1. Global expression values across all 21 GABAergic genes in each of the nine BrainSpan individuals.

The Y axis indicates log2 RPKM (Reads Per Kilobase of exon model per Million mapped reads) expression values for all 21 genes across all 16 brain regions for each BrainSpan sample. The age and sex of each individual can be found in Table 2.

doi:10.1371/journal.pone.0064014.g001

Figure S2 displays the expression levels of the 21 GABAergic genes averaged across the 16 brain regions. GABRA6 and GABRQ had the lowest global expression levels and GABBR1 and GABRD had the highest. Figures S3, S4, S5, S6, S7, S8 show the expression levels for each gene in all 16 brain regions.

Correlation of Expression of 21 GABAergic Genes Across Whole Brain (BrainSpan Samples)

Figure 2 presents the correlation in expression of the 21 GABAergic genes. The scatter plot matrices allow for the identification of genes that show coordinate expression.

thumbnail

Figure 2. Correlation matrix of gene-gene expression for each GABA receptor subunit and associated GABAergic genes.

The heat map is shaded according to the R2 value (see color key). However, the values printed in each cell are the R values so that information about directionality is preserved.

doi:10.1371/journal.pone.0064014.g002

Factor Analysis of 21 GABAergic Genes Across Whole Brain (BrainSpan Samples)

The factor analysis identified six factors that together accounted for 0.86 of the variance in total GABAergic gene expression (Table 4, Figure 3). The results presented below are for genes that loaded ≥0.5 on a factor. Out of the total of 21 genes, 15 loaded ≥0.5 onto only one factor. The remaining 6 genes loaded ≥0.5 onto two factors.

thumbnail

Figure 3. Factor analysis: 6 factor solution.

The 6 factors are shown as boxes. Shaded boxes indicate factors influenced by chronic alcohol or cocaine exposure. Dashed outlined boxes indicate the two cerebellar factors. The proportion of the total variance accounted for by each factor is: factor 1:0.31; factor 2:0.17; factor 3:0.14; factor 4:0.10; factor 5:0.09; factor 6:0.06.

doi:10.1371/journal.pone.0064014.g003
thumbnail

Table 4. Factor analysis.

doi:10.1371/journal.pone.0064014.t004
Factor 1: GABAA subunit genes, including the chr 5 cluster.

Ten of the 21 genes loaded onto this factor that accounted for 0.31 of the total variance in GABAergic gene expression. The strongest loading (0.9) was for the chr 5 cluster of genes (GABRA1, GABRB2, GABRG2) that together encode the most abundant GABAA receptor subunit group: α1β2γ2. Two chr 15 genes: GABRB3 and GABRG3 also had high loadings (0.8) on this factor. The chr 4 gene GABRA4 and the chr X gene GABRA3 loaded more modestly (0.7–0.6). In addition, there were loadings for GABBR2, the GABAB receptor subunit gene (0.7), GAD1 (0.5), implicated in GABA synthesis and GPHN (0.6), implicated in synaptic localization of GABAA receptors. However it should be noted that GAD1 and GPHN had higher loadings on Factor 2.

Factor 2: genes involved in synaptic transport.

Seven genes loaded onto this factor that accounted for 0.17 of the total variance in GABAergic gene expression. Six of these genes are implicated in GABA transport across the synaptic cleft. The strongest loading was for SLC6A1 (0.8) that encodes the neuronal GABA transporter GAT-1. There were also loadings for the two GABAB receptor genes GABBR1 (0.7) and GABBR2 (0.6) and for PRAF2 (0.7), that encodes PRAF2, a protein that is enriched in synaptic vesicles. GPHN (0.6) and GABRD (0.5) that encodes the extrasynaptic δ GABAA subunit also loaded onto this factor. Finally, GAD1, implicated in GABA synthesis, loaded more strongly (0.7) on this factor than on factor 1.

Factor 3: chr 4 gene cluster.

All four GABAA subunit genes in the chr 4 cluster loaded onto this factor that accounted for 0.14 of the total variance in GABAergic gene expression: GABRG1 (1.0); GABRA2 (0.8); GABRB1 (0.8) and GABRA4 (0.5). Three of these genes encode the GABAA receptor subunit combination α2β1γ1 that is found almost exclusively in brain regions that form the reward pathway. It is noteworthy that GABRA4 loaded more strongly on factor 1 (0.7).

Factor 4: genes with low cerebellar expression.

Three GABAA receptor subunit genes loaded onto this factor that accounted for 0.10 of the total variance in GABAergic gene expression: chr 15 GABRA5 (1.0); chr X GABRA3 (0.6) and GABRQ (0.5). All these genes have very low expression in the cerebellar cortex (Figure 4).

thumbnail

Figure 4. High (Panel A) and low (Panel B) expression cerebellar factors.

Panel A: Expression of GABRA6 is exceptionally high in the cerebellar cortex (CBC) and very low in all other brain regions. GABRD is globally highly expressed with the highest expression of all in the CBC. Panel B: GABRA5 has moderately high global expression but exceptionally low expression in the CBC. The expression of GABRQ varies throughout the brain with lowest expression in the CBC. GABRA3 is moderately well expressed globally with lowest expression in the CBC and striatum. Y axis: gene expression levels are provided as Log2 transformed RPKM (Reads Per Kilobase of exon model per Million mapped reads) values. X axis: the 16 brain regions left to right are the brain regions 1–16 as listed in Table 1.

doi:10.1371/journal.pone.0064014.g004
Factor 5: genes with high cerebellar expression.

Two GABAA receptor subunit genes loaded onto this factor and accounted for 0.09 of the total variance in GABAergic gene expression: GABRA6 (1.0) and GABRD (0.7). Both genes are highly expressed in the cerebellar cortex (Figure 4); GABRA6 is not expressed in any other brain region. These genes encode the extrasynaptic α6 and δ subunits.

Factor 6: GAD2, implicated in GABA synthesis.

Only GAD2 loaded onto this factor (0.7) that accounted for 0.06 of the total variance in GABAergic gene expression. Interestingly, there was also a sub-threshold loading for GAD1 (0.4).

Expression of 21 GABAergic Genes in Hippocampal Samples from Alcoholics, Cocaine Addicts and Controls

Using RNA-Seq, expression levels of 17 of the GABAergic genes listed in Table 4 were previously measured in the hippocampus of alcoholics, cocaine addicts and controls from the University of Miami Brain Bank [15]. A pattern of altered expression in cocaine addicts and alcoholics, largely down-regulation, was previously detected [15]. We have now also extracted the GABRD, PRAF2, GABRQ and GABRA3 expression values from this previously described dataset [14]. There were significant differences in gene expression for GABRD and PRAF2:

GABRD: F(21,2) = 4.1, p = 0.032; increased expression in alcoholics relative to cocaine addicts, intermediate expression in controls. PRAF2: F(21,2) = 3.6, p = 0.044; decreased expression in alcoholics and cocaine addicts relative to controls.

There were no significant differences in gene expression by diagnosis for GABRQ and GABRA3: p = 0.63–0.25 respectively.

Here we have utilized these data to search for a relationship with the six GABAergic gene expression factors derived from the BrainSpan whole brain data. Our aim was to determine whether some of the factors might show collective changes in gene expression in response to chronic alcohol and cocaine exposure whereas other factors might be unaffected.

Factors Unresponsive to Chronic Alcohol/Cocaine Exposure

Factor 1: GABAA subunit genes, including the chr 5 cluster: of the 10 genes included in this factor, only GABRG2, GAD1 and GPHN showed expression changes in alcoholics/cocaine addicts. It should be noted that GAD1 and to some extent GPHN load more strongly on factor 2.

Factor 4: genes with low cerebellar expression: none of the three genes with low expression in the cerebellar cortex were affected by chronic exposure to alcohol or cocaine.

Factors Responsive to Chronic Alcohol/Cocaine Exposure

Factor 2: genes involved in synaptic transport: with the exception of GABBR2 that loads more strongly on factor 1, the remaining six genes all showed altered expression in alcoholics/cocaine addicts.

Factor 3: chr 4 gene cluster: with the exception of GABRA4 that loads more strongly on factor 1, the other three chr 4 complex genes had altered expression in alcoholics/cocaine addicts.

Factor 5: high cerebellar expression: of the two genes that have exceptionally high expression in the cerebellar cortex, GABRD had altered expression in the hippocampus of alcoholics and cocaine addicts. We had no data on GABRA6 since it is only expressed in the cerebellar cortex.

Factor 6: GABA synthesis; GAD2: GAD2 showed expression changes in alcoholics/cocaine addicts.

Discussion

It has been known for approximately 20 years that there are four clusters of genes encoding 14 of the 19 GABAA receptor subunits that are located on chromosomes 4, 5, 15 and X [20][24]. These subunits comprising the pentameric GABAA receptors are expressed in different combinations in different cell populations and brain regions and differentially during development. There are many examples of clustered genes, for example the hemoglobin and immunoglobulin genes, the major histocompatibility complex and genes relevant to neurobiology such as MAOA-MAOB on chr X, HTR3A-HTR3B on chr 11 and genes encoding the class 1 alcohol dehydrogenases on chr 4. Gene clusters can arise due to random duplications that offer the possibility of functional diversification. It is thought that the four GABAA gene clusters originated from duplications of (and within) an ancestral gene cluster containing single genes encoding one α, one β and one γ subunit [13]. It is possible that the clustering of genes could enable coordinate transcription resulting in specific receptor subunit combinations. For example, three of the chr 5 cluster genes encode the ubiquitous GABAA α1β2γ2 receptor and three of the chr 4 cluster genes encode the GABAA α2β1γ1 receptor that is found in the mesolimbic reward pathway. Coordinate expression may vary through development, something that we were not able to explore in this study of adults. Linkage disequilibrium provides a mechanism for facilitating coordinate gene expression by maintenance of gene clusters. Although the GABAA gene duplication event is thought to be ancient, yet there is strong linkage disequilibrium across the chr 4 and chr 5 regions, implying selective pressure [1]. Nevertheless, the mechanisms underlying the GABAA gene clustering are currently unknown.

Up to this point information about relative GABAergic gene expression (predominantly of GABAA receptors) has largely derived from preclinical studies, and usually from only a few brain regions. Furthermore, it is important to understand global patterns of GABAergic gene expression in healthy individuals before considering the effects of alcohol and drugs of abuse on GABA neurotransmission. Therefore in this study we accessed data from the BrainSpan Atlas that has been designed as a public resource for studying transcriptional mechanisms involved in human brain development. We were able to perform a comprehensive correlational analysis of the expression of 21 GABAergic genes across 16 brain regions in healthy adults. In order to identify distinct groupings of genes in the correlation analyses we performed a factor analysis which revealed six factors, each of which contributed to >0.05 of the variance and together accounted for 0.86 of the total variance in gene expression. The factor that accounted for one third of the total variance largely comprised genes encoding GABAA receptor subunits. Notably, the highest loading on this factor was for the chr 5 cluster genes that encode the group of subunits comprising the most abundant (60%) α1β2γ2 GABAA receptor. There were also trans correlations of chr X, chr 15 and chr 5 genes that together encode the third most common subunit group: α3β3γ2. Considering only the major loadings on this factor, none of the genes, with the exception of GABRG2, showed altered hippocampal expression in alcoholics and cocaine addicts. In line with this finding, it is noteworthy that there have been few if any positive association studies between variants in this group of genes and alcoholism (reviewed in Enoch, 2008 [1]).

Three of the four chr 4 cluster genes had major loadings on factor 3. These genes encode the GABAA receptor subunit group α2β1γ1 that is found almost exclusively in the reward pathway. Many human studies have implicated the chr 4 gene cluster in alcoholism and drug addiction (reviewed in Enoch, 2008 [1]). It has been shown that the mRNAs from the chr 4 cluster genes predominate in rat embryo but these genes are generally down regulated in the adult rat except in the hippocampus, the majority of DA neurons in the substantia nigra and the VTA where they are highly expressed [25][27]. Indeed, our results showed generally reduced expression of the chr 4 cluster genes relative to the chr 5 cluster genes (Figures S3, S4). In contrast to the factor 1 GABAA receptor subunit genes, all three of the factor 3 chr 4 cluster genes showed expression changes after chronic exposure to alcohol/cocaine [15].

Factor 2 can be regarded as a synaptic transport factor that is strongly influenced by chronic alcohol and cocaine exposure. The five genes that had major loadings on this factor are implicated in GABA synthesis and GABA transport across the synaptic cleft. GAD1 encodes GAD67, the rate-limiting enzyme for the production of GABA [28]. SLC6A1 encodes the neuronal GABA transporter GAT-1. GABBR1 encodes the GABAB1 receptor subunit. Presynaptic GABAB receptors suppress GABA release by inhibiting the activity of voltage-gated calcium channels. PRAF2 encodes the PRAF2 protein that has been implicated in the control and feedback regulation of synaptic transmission [19]. The fifth gene loading onto this factor, GPHN, encodes gephyrin, a scaffolding protein that anchors GABAA receptors to the postsynaptic skeleton. In our earlier study [15] we found that GAD1, GPHN, SLC6A1, and GABBR1 were down-regulated in alcoholics and cocaine addicts and we have now shown the same result for PRAF2 in a re-analysis of the data.

Factor 6 is also implicated in GABA synthesis and is also influenced by chronic alcohol and cocaine exposure. Only GAD2 loaded onto this factor although GAD1 had a sub-threshold loading. In mammals, glutamic acid decarboxylase (GAD), the enzyme responsible for metabolizing glutamate to GABA, exists in two isoforms: GAD67 and GAD65 respectively encoded by GAD1 and GAD2. Both isoforms are present in most classes of GABA neurons but GAD67 is present in many cell bodies whereas GAD65 is prominent in axon terminals [29]. It has been shown that GAD67 is the rate-limiting enzyme for the production of intracellular GABA for phasic inhibition [28] and that GAD65 (encoded by GAD2) plays a significant role in the synthesis of GABA for tonic inhibition [30].

There were distinct factors for low and high expression of GABAA receptor subunit genes in the cerebellar cortex. Factor 4 included GABRA5, GABRA3 and GABRQ, all of which had low expression. The strongest loading was for GABRA5 (1.0) and as can be seen from Figure 4, GABRA5 is expressed at moderately high levels in all brain regions except for the cerebellum. Only a small proportion of receptors include α5 subunits and these are largely expressed extrasynaptically. Factor 5 included GABRA6 and GABRD, both of which have substantially high expression in the cerebellum. Animal studies have shown that GABRA6 is only expressed in cerebellum and our study confirms that this is also the case in humans. It has been shown that the α6 and the δ subunits are included in extrasynaptic receptors that respond to tonic GABA currents. The cerebellar granule cells appear to have large tonic GABAA extrasynaptic receptor-mediated conductance [31]. Our earlier study showed that the low expression genes loading onto factor 4 appeared to be unaffected by exposure to chronic alcohol and cocaine. In contrast, we have shown that the factor 5 GABRD is sensitive to the effects of alcohol/drug exposure. Although we could not show effects on GABRA6 since it is only expressed in cerebellum, it is perhaps likely to follow the same direction as GABRD since both extrasynaptic receptors are potentiated by low doses of ethanol.

One limitation to this study is the fact that we extrapolated the results of the BrainSpan factor analysis derived from the whole brain to the hippocampus (samples from alcoholics/cocaine addicts/controls). This is not ideal however we did not have sufficient data points to perform a factor analysis in hippocampus alone. Nevertheless, it can be seen from figures S3, S4, S5, S6, S7, S8 that for each gene, expression in the hippocampus is similar to expression in nearly all other brain regions. It is only the cerebellum (CBC), and to some extent the mediodorsal nucleus of thalamus (MD) and the striatum (STR) in which some genes show very different expression patterns relative to other brain regions. Whereas it would be questionable to extrapolate the results from the whole brain to the cerebellum, we feel that the evidence provided in figures S3, S4, S5, S6, S7, S8 justify our extrapolation of the factor analysis from the whole brain to the hippocampus.

In conclusion, we have been able to elucidate patterns of GABAergic gene expression across numerous brain regions in healthy adult humans. Although there is a comprehensive understanding of GABAergic gene expression in rodent brain, no such comprehensive studies have previously been performed in humans. Our study has shown that the GABAA gene clusters show both cis and trans correlations of gene expression. Moreover, the factor analysis revealed that distinct groups of genes, notably those involved in GABA synthesis and synaptic transport and the chr 4 cluster, previously associated with alcohol and drug dependence in humans, are sensitive to the effects of alcohol/cocaine. On the other hand the two factors on which the chr 15, chr X and most of the chr 5 subunit genes loaded were impervious to alcohol/cocaine effects. Although our study was limited to mRNA expression data and cannot necessarily be extrapolated to protein levels, nevertheless our findings might have implications for therapeutic targets to combat stress-related craving and relapse.

Supporting Information

Figure S1.

Relative GABAergic gene expression in human postmortem hippocampal samples from two datasets: BrainSpan and the Miami Brain Bank. The graphs shows mean (S.E.) expression levels for each gene.

doi:10.1371/journal.pone.0064014.s001

(TIF)

Figure S2.

Expression of GABAergic genes averaged across all 16 brain regions.

doi:10.1371/journal.pone.0064014.s002

(TIF)

Figure S3.

Expression of chromosome 4 GABAA receptor subunit genes in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s003

(TIF)

Figure S4.

Expression of chromosome 5 GABAA receptor subunit genes in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s004

(TIF)

Figure S5.

Expression of chromosome 1 and chromosome 15 GABAA receptor subunit genes in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s005

(TIF)

Figure S6.

Expression of chromosome X GABAA receptor subunit genes and GABAB receptor subunit genes in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s006

(TIF)

Figure S7.

Expression of PRAF2, GPHN, GAD1 and GAD2 in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s007

(TIF)

Figure S8.

Expression of SLC6A1 in 16 brain regions. The brain regions are identified in Table 1.

doi:10.1371/journal.pone.0064014.s008

(TIF)

Author Contributions

Conceived and designed the experiments: MAE BB QY DG. Performed the experiments: BB QY. Analyzed the data: MAE BB. Contributed reagents/materials/analysis tools: QY. Wrote the paper: MAE BB DG.

References

  1. 1. Enoch M-A (2008) The role of GABAA Receptors in the Development of Alcoholism. Pharmacol Biochem Behav 90: 95–104. doi: 10.1016/j.pbb.2008.03.007
  2. 2. Barnard EA, Skolnick P, Olsen RW, Mohler H, Sieghart, et al (1998) International Union of Pharmacology. XV. Subtypes of gamma-aminobutyric acidA receptors: classification on the basis of subunit structure and receptor function. Pharmacol Rev 50: 291–313.
  3. 3. Sigel E, Baur R, Boulineau N, Minier F (2006) Impact of subunit positioning on GABAA receptor function. Biochem Soc Trans 34: 868–871. doi: 10.1042/bst0340868
  4. 4. Benke D, Fritschy JM, Trzeciak A, Bannwarth W, Mohler H (1994) Distribution, prevalence, and drug binding profile of gamma-aminobutyric acid type A receptor subtypes differing in the beta-subunit variant. J Biol Chem 269: 27100–27107.
  5. 5. Michels G, Moss SJ (2007) GABAA receptors: properties and trafficking. Crit Rev Biochem Mol Biol 42: 3–14.
  6. 6. Minier F, Sigel E (2004) Positioning of the alpha-subunit isoforms confers a functional signature to gamma-aminobutyric acid type A receptors. Proc Natl Acad Sci USA 101: 7769–7774. doi: 10.1073/pnas.0400220101
  7. 7. Rowlett JK, Platt DM, Lelas S, Atack JR, Dawson GR (2005) Different GABAA receptor subtypes mediate the anxiolytic, abuse-related, and motor effects of benzodiazepine-like drugs in primates. Proc Natl Acad Sci USA 102: 915–920. doi: 10.1073/pnas.0405621102
  8. 8. Rudolph U, Crestani F, Benke D, Brünig I, Benson JA, et al. (1999) Benzodiazepine actions mediated by specific gamma-aminobutyric acid(A) receptor subtypes. Nature 401: 796–800.
  9. 9. Jin P, Fu GK, Wilson AD, Yang J, Chien D, et al. (2004) PCR isolation and cloning of novel splice variant mRNAs from known drug target genes. Genomics 83: 566–571. doi: 10.1016/j.ygeno.2003.09.023
  10. 10. Tian H, Chen H-J, Cross TH, Edenberg HJ (2005) Alternative splicing and promoter use in the human GABRA2 gene. Brain Res Mol Brain Res 137: 174–183. doi: 10.1016/j.molbrainres.2005.03.001
  11. 11. Martyniuk CJ, Aris-Brosou S, Drouin G, Cahn J, Trudeau VL (2007) Early evolution of ionotropic GABA receptors and selective regimes acting on the mammalian-specific theta and epsilon subunits. PLoS ONE 2(9): e894. doi: 10.1371/journal.pone.0000894
  12. 12. Tsang SY, Ng SK, Xu Z, Xue H (2007) The evolution of GABAA receptor-like genes. Mol Biol Evol 24: 599–610. doi: 10.1093/molbev/msl188
  13. 13. Russek SJ (1999) Evolution of GABA(A) receptor diversity in the human genome. Gene 227: 213–222. doi: 10.1016/s0378-1119(98)00594-0
  14. 14. Zhou Z, Yuan Q, Mash DC, Goldman D (2011) Substance-specific and shared transcription and epigenetic changes in the human hippocampus chronically exposed to cocaine and alcohol. Proc Natl Acad Sci USA 108: 6626–6631. doi: 10.1073/pnas.1018514108
  15. 15. Enoch M-A, Zhou Z, Kimura M, Mash DC, Yuan Q, et al. (2012) Postmortem Hippocampal GABAergic Gene Expression in Alcoholics and Cocaine Addicts; Corresponding Findings in Alcohol Naïve P and NP Rats. PLoS ONE 7(1): e29369. doi: 10.1371/journal.pone.0029369
  16. 16. BrainSpan Technical White Paper: Transcriptome Profiling by RNA Sequencing and Exon Microarray (June 2011, version 3). Available: http://help.brainmap.org/display/devhuma​nbrain/Documentation. Document: developmental transcriptome.
  17. 17. Jacob TC, Bogdanov YD, Magnus C, Saliba RS, Kittler JT, et al. (2005) Gephyrin regulates the cell surface dynamics of synaptic GABAA receptors. J Neurosci 25: 10469–10478. doi: 10.1523/jneurosci.2267-05.2005
  18. 18. Fo CS, Coleman CS, Wallick CJ, Vine AL, Bachmann AS (2006) Genomic organization, expression profile, and characterization of the new protein PRA1 domain family, member 2 (PRAF2). Gene 371: 154–165. doi: 10.1016/j.gene.2005.12.009
  19. 19. Koomoa DL, Go RC, Wester K, Bachmann AS (2008) Expression profile of PRAF2 in the human brain and enrichment in synaptic vesicles. Neurosci Lett 436: 171–176. doi: 10.1016/j.neulet.2008.03.030
  20. 20. Greger V, Knoll JH, Woolf E, Glatt K, Tyndale RF, et al. (1995) The gamma-aminobutyric acid receptor gamma 3 subunit gene (GABRG3) is tightly linked to the alpha 5 subunit gene (GABRA5) on human chromosome 15q11–q13 and is transcribed in the same orientation. Genomics 26: 258–264. doi: 10.1016/0888-7543(95)80209-5
  21. 21. McLean PJ, Farb DH, Russek SJ (1995) Mapping of the alpha 4 subunit gene (GABRA4) to human chromosome 4 defines an alpha 2-alpha 4-beta 1-gamma 1 gene cluster: further evidence that modern GABAA receptor gene clusters are derived from an ancestral cluster. Genomics 26: 580–586. doi: 10.1016/0888-7543(95)80178-o
  22. 22. Sinnett D, Wagstaff J, Glatt K, Woolf E, Kirkness EJ, et al. (1993) High-resolution mapping of the gamma-aminobutyric acid receptor subunit beta 3 and alpha 5 gene cluster on chromosome 15q11–q13, and localization of breakpoints in two Angelman syndrome patients. Am J Hum Genet 52: 1216–1229.
  23. 23. Russek SJ, Farb DH (1994) Mapping of the beta 2 subunit gene (GABRB2) to microdissected human chromosome 5q34–q35 defines a gene cluster for the most abundant GABAA receptor isoform. Genomics 23: 528–533. doi: 10.1006/geno.1994.1539
  24. 24. Wilke K, Gaul R, Klauck SM, Poustka A (1997) A gene in human chromosome band Xq28 (GABRE) defines a putative new subunit class of the GABAA neurotransmitter receptor. Genomics 45: 1–10. doi: 10.1006/geno.1997.4885
  25. 25. Okada H, Matsushita N, Kobayashi K, Kobayashi K (2004) Identification of GABAA receptor subunit variants in midbrain dopaminergic neurons. J Neurochem 89: 7–14. doi: 10.1111/j.1471-4159.2004.02271.x
  26. 26. Steiger JL, Russek SJ (2004) GABAA receptors: building the bridge between subunit mRNAs, their promoters, and cognate transcription factors. Pharmacol Ther 101: 259–281. doi: 10.1016/j.pharmthera.2004.07.003
  27. 27. Wisden W, Laurie DJ, Monyer H, Seeburg PH (1992) The distribution of 13 GABAA receptor subunit mRNAs in the rat brain. I. Telencephalon, diencephalon, mesencephalon. J Neurosci 12: 1040–1062.
  28. 28. Asada H, Kawamura Y, Maruyama K, Kume H, Ding R, et al. (1996) Mice lacking the 65 kDa isoform of glutamic acid decarboxylase (GAD65) maintain normal levels of GAD67 and GABA in their brains but are susceptible to seizures. Biochem Biophys Res Commun 229: 891–895. doi: 10.1006/bbrc.1996.1898
  29. 29. Esclapez M, Tillakaratne NJ, Kaufman DL, Tobin AJ, Houser CR (1994) Comparative localization of two forms of glutamic acid decarboxylase and their mRNAs in rat brain supports the concept of functional differences between the forms. J Neurosci 14: 1834–1855.
  30. 30. Walls AB, Nilsen LH, Eyjolfsson EM, Vestergaard HT, Hansen SL, et al. (2010) GAD65 is essential for synthesis of GABA destined for tonic inhibition regulating epileptiform activity. J Neurochem 115: 1398–1408. doi: 10.1111/j.1471-4159.2010.07043.x
  31. 31. Kullmann DM, Ruiz A, Rusakov DM, Scott R, Semyanov A, et al. (2005) Presynaptic, extrasynaptic and axonal GABAA receptors in the CNS: where and why? Prog Biophys Mol Biol 87: 33–46. doi: 10.1016/j.pbiomolbio.2004.06.003