Transcriptional response to alcohol exposure in Drosophila melanogaster.

BACKGROUND: Alcoholism presents widespread social and human health problems. Alcohol sensitivity, the development of tolerance to alcohol and susceptibility to addiction vary in the population. Genetic factors that predispose to alcoholism remain largely unknown due to extensive genetic and environmental variation in human populations. Drosophila, however, allows studies on genetically identical individuals in controlled environments. Although addiction to alcohol has not been demonstrated in Drosophila, flies show responses to alcohol exposure that resemble human intoxication, including hyperactivity, loss of postural control, sedation, and exposure-dependent development of tolerance.
RESULTS: We assessed whole-genome transcriptional responses following alcohol exposure and demonstrate immediate down-regulation of genes affecting olfaction, rapid upregulation of biotransformation enzymes and, concomitant with development of tolerance, altered transcription of transcriptional regulators, proteases and metabolic enzymes, including biotransformation enzymes and enzymes associated with fatty acid biosynthesis. Functional tests of P-element disrupted alleles corresponding to genes with altered transcription implicated 75% of these in the response to alcohol, two-thirds of which have human orthologues.
CONCLUSION: Expression microarray analysis is an efficient method for identifying candidate genes affecting complex behavioral and physiological traits, including alcohol abuse. Drosophila provides a valuable genetic model for comparative genomic analysis, which can inform subsequent studies in human populations. Transcriptional analyses following alcohol exposure in Drosophila implicate biotransformation pathways, transcriptional regulators, proteolysis and enzymes that act as metabolic switches in the regulation of fatty acid metabolism as important targets for future studies of the physiological consequences of human alcohol abuse.

Background

The National Institute on Alcohol Abuse and Alcoholism has estimated that approximately 14 million people in the United States suffer from alcoholism [1]. Genetic studies have demonstrated that the propensity for alcohol abuse is determined by multiple genes. Thus, vulnerability to alcohol abuse shows all the hallmarks of a quantitative trait, as it is polygenic and subject to genotype by environment interactions. Efforts to map chromosomal regions (quantitative trait loci (QTL)) that harbor genes responsible for alcohol-related traits have identified at least 24 QTL regions in the mouse genome [2]. Although candidate genes have been identified within such QTL regions, including the multiple PDZ domain protein Mpdz on chromosome 4 [3], conclusive evidence that links candidate genes within QTL regions directly to the phenotype is often difficult to obtain. Quantitatively measuring alcohol dependence without confounding contributions of other psychiatric or social disorders is challenging with human subjects, although studies in ethnically defined populations have implicated alcohol dehydrogenase [4], the GABAA receptor complex [5], and the serotonin 1B receptor in alcohol sensitivity [6,7].

Flies are naturally exposed to ethanol, as they feed on fermented food. Exposing flies to low concentrations of ethanol stimulates locomotor activity, whereas high concentrations of ethanol induce an intoxicated phenotype that shows marked similarities to human alcohol intoxication, characterized by locomotor impairments, loss of postural control, sedation [8,9] and exposure-dependent development of tolerance [10]. Alcohol sensitivity in Drosophila melanogaster can be quantified in an 'inebriometer', a 122 cm long vertical glass column, which contains a series of slanted mesh partitions to which flies can attach [11]. Flies are introduced in the top of the column and exposed to ethanol vapors. As they lose postural control they fall through the column. The elution time from the column is used as a measure of sensitivity to alcohol intoxication. Following an initial exposure to alcohol, flies develop tolerance, manifested upon a second exposure by a shift in the elution profile [10]. Tolerance peaks within hours after exposure and persists up to 24 hours in some individuals [10].

Here, we exploited the power of the Drosophila system to evaluate whole-genome transcriptional profiles that reflect alterations in gene expression upon exposure to alcohol and during development of tolerance, to gain insights into the underlying cellular and physiological mechanisms that respond to alcohol exposure. We then evaluated alcohol sensitivity and the development of tolerance in flies with mutations in 20 genes that were implicated by the analysis of differential transcript abundance, and identified 15 novel candidate genes affecting alcohol related phenotypes, 10 of which have human orthologues. Thus, whole genome transcriptional analysis following alcohol exposure in Drosophila can identify candidate genes and metabolic pathways that are relevant to alcohol abuse and its physiological consequences in people.

Results and discussion

We hypothesized that analysis of gene expression after one and two exposures to ethanol would reveal novel genes and pathways underpinning the genomic response to ethanol, and that subsequent functional tests with mutations in such genes would confirm them as candidate genes for future detailed investigation. This strategy requires that mutations are generated in a common homozygous (isogenic) genetic background, since effects of segregating variants in an outbred strain will be of the same magnitude as the mutational effects we wish to detect. Fortuitously, a panel of single P-transposable element insertional mutations has been generated in each of five Canton-S derived isogenic strains (Canton S A, B, C, E, F) as part of the Drosophila Gene Disruption Project [12]. We took advantage of this resource and characterized for each of the 5 isogenic strains the responseto ethanol after an initial exposure and the development of tolerance after a second exposure 2 hours later, at 2 ages (3 to 5 days and 12 to 15 days post eclosion). We observed that sensitivity and development of tolerance are highly sensitive to genetic background (Additional data file 5). Canton-S B flies show the most pronounced development of tolerance, which is maximal two hours after exposure to undiluted ethanol vapors. Tolerance is least apparent in flies in the Canton-S E genetic background (Figure 1a). Furthermore, three- to five-day old flies are more sensitive than older flies to the initial alcohol exposure, and consequently show greater tolerance (Figure 2). Therefore, we used three- to five-day old Canton-S B flies for our transcriptional analysis.

Figure 1

Sensitivity and development of tolerance to alcohol in Drosophila melanogaster. (a) Inebriometer elution profiles of Canton S B and Canton S E flies after exposure to concentrated ethanol vapor. Note the shifts in peak elution time between the first (solid symbols) and second (open symbols) ethanol exposure, and the difference in this peak shift between Canton S B and Canton S E. (b) Schematic diagram of the experimental design used to assess whole genome transcriptional changes after ethanol exposure. The numbers indicate samples collected for microarray analysis, as follows: 1, control flies; 2, flies after a single exposure to ethanol; 3, flies that develop tolerance during a 2 h interval prior to a second exposure to ethanol.

Figure 2

Inebriometer elution profiles of 3- to 5-day old (top) and 12- to 15-day old (bottom) flies after the initial (closed symbols, blue arrows) and a second, 2 h later (open symbols; red arrows) exposure to ethanol.

We collected flies either without exposure to ethanol, immediately after exposure to ethanol and passage through the inebriometer, or from a population that had developed tolerance two hours after the initial ethanol exposure (Figure 1b). We used 5 independent replicates of 30 3- to 5-day old males for each time point and generated cRNA for hybridization to high density oligonucleotide microarrays. Raw microarray data are given in Additional data file 1.

We performed analysis of variance to identify probe sets with significant differences in expression between the three treatments, and a false positive discovery rate of q < 0.05 to account for multiple tests [13]. Significant probe sets were further analyzed by post hoc Tukey tests to identify which samples accounted for transcriptional differences between the treatments (Additional data file 2). We identified 582 probe sets with changes in gene expression between the treatment groups at q < 0.05 (Additional data file 2). Among these genes 112 encode predicted transcripts of unknown function, 173 have both mouse and human orthologues, 13 genes have orthologues only in the mouse, and 21 have only human orthologues. A total of 34 probe sets were significant at a false positive discovery rate of q < 0.001; of these, 8 have mammalian orthologues (Table 1).

Table 1

Probe sets with altered transcriptional regulation after ethanol exposure with q < 0.001

Probe set ID	p value	Gene name	Biological process	Changes in expression
1623332_at	1.91E-05	CG31106	Transmission of nerve impulse; carbohydrate metabolism	Transient up-regulation
1623605_a_at	2.20E-05	Serpin-27A (M,H)*	Toll signaling pathway; defense response	Transient up-regulation
1624662_at	2.20E-05	CG11414	Protein ubiquitination	Acute up-regulation
1624736_a_at	2.20E-05	CG11878	Unknown	Slow up-regulation
1625193_at	2.20E-05	CG31272	Lipid metabolism; neurotransmitter secretion; transport	Slow up-regulation
1626882_at	2.35E-05	Pinocchio	Olfactory behavior;response to chemical stimulus	Transient up-regulation
1627080_at	5.07E-05	CG14893	Unknown	Slow up-regulation
1627385_at	6.86E-05	minidiscs	Amino acid transport, metabolism	Transient down-regulation
1628229_at	6.86E-05	Trypsin 29F	Proteolysis	Slow up-regulation
1628494_a_at	6.86E-05	CG33525	Negative regulation of transcription, DNA-dependent	Slow up-regulation
1628647_at	1.06E-04	CG33110	Fatty acid biosynthesis; very-long-chain fatty acid metabolism	Slow up-regulation
1628947_s_at	1.06E-04	Ecdysone-inducible gene L2 (M,H)*	Cell adhesion	Acute up-regulation
1629012_at	1.06E-04	period (M,H)*	Long-term memory; circadian rhythm; signal transduction; negative regulation of transcription from RNA polymerase II promoter	Transient down-regulation
1629592_at	1.32E-04	ηTrypsin (M,H)*	Proteolysis	Transient down-regulation
1629617_at	1.44E-04	cabut	Regulation of transcription from RNA polymerase II promoter; JNK cascade; autophagic cell death	Acute up-regulation
1630302_at	1.66E-04	Odorant-binding protein 99d	Transport; autophagic cell death	Slow up-regulation
1631406_at	1.66E-04	CG3106 (M,H)*	Unknown	Transient down-regulation
1632222_a_at	2.25E-04	CG17752	Extracellular transport; cation transport	Slow up-regulation
1632228_at	2.58E-04	CG10505	Response to toxin; extracellular transport; defense response	Slow up-regulation
1633700_at	2.78E-04	antennal protein 10 (M,H)*	Sensory perception of chemical stimulus	Acute down-regulation
1633794_a_at	3.22E-04	antennal protein 5	Signal transduction	Acute down-regulation
1633913_at	3.80E-04	Cytochrome P450-6a8	Electron transport; steroid metabolism; lauric acid metabolism	Slow up-regulation
1634318_at	3.98E-04	lush	Response to ethanol; sensory perception of smell; olfactory behavior	Acute down-regulation
1634591_at	4.38E-04	Rpd3	Regulation of transcription from RNA polymerase II promoter; determination of adult life span; gene silencing	Slow up-regulation
1635815_at	5.47E-04	CG5819 (M,H)*	Transmission of nerve impulse; cell adhesion	Acute up-regulation
1636202_s_at	6.40E-04	CG8931	Transport	Transient up-regulation
1636611_at	6.40E-04	Pheromone-binding protein-related protein 1 (M)*	Signal transduction; sensory perception; response to pheromone	Acute down-regulation
1636771_at	6.40E-04	CG8562	Proteolysis	Slow down-regulation
1638066_at	6.40E-04	CG9238	Glycogen metabolism	Slow up-regulation
1638246_at	6.52E-04	Olfactory-specific E	Signal transduction; response to pheromone; transport	Acute down-regulation
1638452_at	7.01E-04	CG6733	Proteolysis	Slow down-regulation
1639368_at	7.73E-04	CG5804	Lipid transport; cell acyl-CoA homeostasis	Transient down-regulation
1639934_at	7.77E-04	capulet	Actin filament organization; bristle morphogenesis	Slow up-regulation
1640755_at	8.83E-04	Picot	Phosphate metabolism; extracellular transport; cation transport	Slow up-regulation

Transcripts that show sustained altered expression after the first exposure to ethanol are indicated as 'acutely up- or down-regulated genes'. Transcripts that show altered expression only after the second exposure are designated as 'slowly up- or down-regulated genes'. The designation 'transient up-regulation' refers to genes that show increased expression after the first exposure to ethanol, but a return to control levels 2 h later after the second exposure. *Genes that have human (H) and/or mouse (M) orthologues.

We categorized the 582 significant probe sets as 'acutely' up- or down-regulated if transcript levels show sustained altered expression after the first exposure to ethanol; 'slowly' up-or down-regulated if transcript abundance only changes after the second exposure to alcohol; and 'transiently' up- or down-regulated if transcript levels are altered following the first exposure to ethanol, but are not sustained during the development of tolerance. We found 78 genes that show acute down-regulation and 77 genes that show acute up-regulation, In addition, 104 genes are slowly down-regulated, 258 genes are slowly up-regulated, 33 genes are transiently down-regulated, and 32 genes are transiently up-regulated.

Probe sets were annotated and compiled in Gene Ontology categories using Affymetrix software [14] and the FlyBase data base [15,16] (Additional data files 3 and 4). Gene Ontology categories that are significantly more abundant among the probe sets than expected by chance include genes associated with olfactory function, which are acutely down-regulated, regulators of signal transduction, which are acutely up-regulated, metabolic enzymes, including peptidases, hydrolases and pigmentation genes, which are slowly down-regulated, and transcriptional regulators and circadian genes, which are slowly up-regulated (Additional data files 3 and 4). It is possible that the slow up-regulation of genes affecting circadian rhythm could be attributable to the two hour difference in sampling the control animals and those tested for tolerance.

Expression of a suite of genes associated with odor recognition is acutely down-regulated, including genes that encode the odorant binding proteins lush, Obp19a, Pbprp1-5, the odorant receptor Or67d, and the ubiquitous odorant receptor Or83b, which is necessary for transport and insertion of odorant receptors in the chemosensory dendritic membranes of olfactory neurons [17] (Table 1 and Additional data file 2). Olfactory specific protein E and antennal proteins 5 and 10 are also acutely down-regulated (Table 1). This pattern of altered transcript abundance reflects acute down-regulation of olfactory function upon exposure to undiluted ethanol vapor. Expression of two olfactory proteins is up-regulated: Obp99d is slowly up-regulated and Pinocchio, a gene that encodes a protein implicated in removal of xenobiotics from the sensillar perilymph [18], is transiently up-regulated upon exposure to ethanol (Table 1). In addition, biotransformation enzymes, including those encoded by Cyp6a2 and Cyp6a13, and glutathione-S-transferase D5, are acutely up-regulated. Since Pinocchio is extensively expressed throughout the third antennal segment [18], it is unlikely that the observed reduction in expression of the subset of odorant binding proteins and odorant receptors is due to ethanol induced tissue damage.

Immediate up-regulation is also observed for transcripts involved in response to stress, including l(2)efl, Mpk2 and Hsp70Ab. Several transcriptional regulators are rapidly up-regulated, including cabut, Aly and Drop along with transcripts involved in protein ubiquitination (mib1, CG11414, CG40045 and TSG101). CG1516, which is orthologous to human pyruvate carboxylase is also acutely up-regulated after ethanol exposure (Additional data file 2).

During the two hour period in which tolerance culminates we observe modulation of transcription of a different set of genes. Sixteen genes that encode proteases are down-regulated along with six genes associated with lipid transport and fatty acid metabolism, including bubblegum, NLaz and Hmgs. Two biotransformation enzymes, encoded by Cyp12e1 and Cyp4ac1, are also down-regulated, as is the transcriptional regulator Sir2.

At the same time there is an extensive increase in expression of transcriptional regulators, proteases and metabolic enzymes. Nineteen transcription factors, eleven proteases and thirty enzymes associated with metabolism, including signal transduction components (protein kinase C δ, rdgA, G protein subunit γ1, fz2), enzymes associated with fatty acid biosynthesis and intermediary metabolism (for example, malic enzyme), and biotransformation enzymes (encoded by Cyp6a8, Cyp4e3, Cyp309a1, GstD1, GstE5, GstE7) are up-regulated (Additional data file 2). Thus, a single exposure to ethanol has the potential to elicit profound changes in the protein composition of the cell through altered transcriptional regulation and proteolysis aimed at rapidly adapting cellular metabolism to the consequences of alcohol intoxication.

Forty-six genes with altered transcript abundance on our microarrays correspond to murine orthologues implicated in altered transcriptional regulation in a recent meta-analysis study of alcohol drinking preference in mice [19] (Additional data file 2). Furthermore, several of the genes with changes in transcript levels have been associated previously with responses to alcohol exposure, including lush, which encodes a Drosophila odorant binding protein that interacts with short chain alcohols [20] and the contact pheromone cis-vaccenyl acetate [21], NLaz, which encodes an orthologue of human apolipoprotein D, which is preferentially expressed in prefrontal cortex of alcoholics [22], and period, a regulator of circadian activity that has been associated with alcohol consumption in mice and humans [23]. In addition, Sorbitol dehydrogenase 2, CG1600 and v(2)k05816, which harbor alcohol dehydrogenase activity (Additional data file 2), also show altered transcript levels.

Mutant analysis has implicated a handful of other genes in sensitivity to ethanol. These include the cheapdate allele of amnesiac [24], which encodes a neuropeptide thought to activate the cyclic AMP signaling pathway [25]; the calcium/calmodulin-dependent adenylate cyclase encoded by the rutabaga (rut) gene [24]; the axonal migration and cell adhesion receptor, fasciclin II (Fas2) [26]; PkaR2, which encodes a cyclic AMP-dependent protein kinase [27]; the gene encoding the GABA-B receptor 1 [28]; and the genes encoding the Drosophila neuropeptide F (npf, a homolog of the mammalian neuropeptide Y) and its receptor [29]. In addition, heat stress has been found to induce tolerance to a subsequent exposure to ethanol and implicated a nucleic acid binding zinc finger protein encoded by the hangover (hang) gene in both the response to heat stress and the induction of ethanol tolerance [30]. Induction of ethanol tolerance was completely abolished in flies carrying both null mutations in hang and in the gene encoding tyramine β-hydroxylase (Tbh), which suggested that hang and the neurotransmitter octopamine mediate separate pathways involved in the induction of ethanol tolerance [30]. Further, mutations in slowpoke, which encodes a large-conductance calcium-activated potassium channel, eliminate the capacity for rapid tolerance, defined as reduction of the duration of sedation on a second exposure to ethanol [31,32].

However, none of these genes exhibited significant alterations in transcript abundance following one or two exposures to ethanol in this study, at our false discovery rate threshold of q < 0.05. Several factors could account for this discrepancy. First, the genes may have low levels of expression in adult flies, such as amnesiac and GABA-B-1, which had absent calls and were consequently not included in the analysis. Second, the magnitude of the differences in expression may have been too small to be detected with our experimental design. Fas2, Pka-R2, rut, npf, hang and Tbh all exhibited changes in transcript abundance on exposure to ethanol, but the mean differences between treatments were not large enough to be significant. Third, differences in the definition of tolerance may be relevant. slowpoke expression in the nervous system is required for tolerance defined as reduced sedation on a second exposure to ethanol four hours after the initial exposure [31,32] but the effects of reduced slowpoke expression have not been tested in our paradigm. Fourth, the mutational effects may be due to post-transcriptional regulation. Nevertheless, our results point at modulation of a complex genetic network upon alcohol exposure, in contrast to a simple linear model of two distinct and independent cellular pathways [30].

Next, we asked to what extent genes with altered expression on the microarrays (Additional data file 2) can be causally implicated in determining alcohol sensitivity or the tolerance response. We identified 20 genes with altered transcription following alcohol exposure for which co-isogenic P-element insert lines were available with a transposon insertion in or near the candidate gene [12], and determined elution profiles during the initial alcohol exposure and two hours later. Five replicate assays were performed for each line and mean elution times were compared to the appropriate co-isogenic P-element free control strains in either the Canton-S A, B, C, E or F genetic backgrounds (Table 2). One line shows greater sensitivity and seven lines are more resistant to a single alcohol exposure than the controls. Four lines show a statistically significant reduction in the peak shift of the modal elution time two hours after the initial exposure, whereas seven showed a greater shift in modal elution time after developing alcohol tolerance. It is of interest to note that three P-element insertion lines in the Canton-S E genetic background, which itself shows only a small tolerance response (Figure 1a), show substantially increased tolerance responses when transposons are inserted in or near the Malic enzyme, Thor or fz2 genes. Overall, there is a significant positive correlation (r = 0.61 ± 0.187, 0.001