Regulation of Ethanol-Related Behavior and Ethanol Metabolism by the Corazonin Neurons and Corazonin Receptor in Drosophila melanogaster

Impaired ethanol metabolism can lead to various alcohol-related health problems. Key enzymes in ethanol metabolism are alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH); however, neuroendocrine pathways that regulate the activities of these enzymes are largely unexplored. Here we identified a neuroendocrine system involving Corazonin (Crz) neuropeptide and its receptor (CrzR) as important physiological regulators of ethanol metabolism in Drosophila. Crz-cell deficient (Crz-CD) flies displayed significantly delayed recovery from ethanol-induced sedation that we refer to as hangover-like phenotype. Newly generated mutant lacking Crz Receptor (CrzR01) and CrzR-knockdown flies showed even more severe hangover-like phenotype, which is causally associated with fast accumulation of acetaldehyde in the CrzR01 mutant following ethanol exposure. Higher levels of acetaldehyde are likely due to 30% reduced ALDH activity in the mutants. Moreover, increased ADH activity was found in the CrzR01 mutant, but not in the Crz-CD flies. Quantitative RT-PCR revealed transcriptional upregulation of Adh gene in the CrzR01. Transgenic inhibition of cyclic AMP-dependent protein kinase (PKA) also results in significantly increased ADH activity and Adh mRNA levels, indicating PKA-dependent transcriptional regulation of Adh by CrzR. Furthermore, inhibition of PKA or cAMP response element binding protein (CREB) in CrzR cells leads to comparable hangover-like phenotype to the CrzR01 mutant. These findings suggest that CrzR-associated signaling pathway is critical for ethanol detoxification via Crz-dependent regulation of ALDH activity and Crz-independent transcriptional regulation of ADH. Our study provides new insights into the neuroendocrine-associated ethanol-related behavior and metabolism.


Introduction
Chronic ethanol consumption causes serious health problems such as liver cirrhosis and various types of cancer [1]. Detoxification of the ethanol involves a sequence of reactions, in which ethanol is first oxidized to acetaldehyde by ADH, and then further oxidized to acetate by mitochondrial ALDH. Acetaldehyde promotes adduct formation, leading to the dysfunction of various key proteins and DNA damage [2]. Thus, the accumulation of acetaldehyde introduces significant toxic effects, causing the ethanol-associated health problems. Therefore, the regulatory mechanisms of ethanol metabolism are important to understand pathophysiological effects of ethanol.
The rate of ethanol metabolism in individuals is greatly influenced by genetic polymorphisms, which gives rise to enzyme variants with different catalytic properties [3,4]. In addition to the genetic factors, transcriptional regulation and post-translational modifications affect ADH/ALDH activity [5][6][7][8]. Interestingly, hormones, such as thyroid and growth hormone, modulate ADH activity and its expression, indicating that hormonal factors could play a part in the ethanol detoxification [9][10][11][12][13]. However, little is known about neuroendocrine regulation of ethanol metabolism.
The genetic basis of ethanol-induced behaviors has been investigated in the fruit fly, Drosophila melanogaster, since its progressive behavioral patterns in response to acute ethanol exposure are quite similar to those of humans [14]. Moreover, genetic and transgenic toolkits available for this species make it an attractive model system to investigate the molecular mechanisms underlying pathophysiological and behavioral outcome of the ethanol consumption in vertebrates. Various studies have reported that several neuropeptides are involved in the regulation of ethanol-related behavior in Drosophila [15][16][17][18]. These studies suggest that peptidergic networks play an important role in modulating sensitivity to ethanol.
The neuropeptide Corazonin (Crz) was first found in the American cockroach [19]. Although the sequence and structure of Crz is highly conserved among different insect species [20], it has been shown to affect diverse physiological functions in a species-specific manner; cardio-acceleration in the cockroach [19], induction of cuticular pigmentation in the migratory locust [21], reduction of the spinning rate and pupal development in the silkworm [22], and induction of ecdysis in a moth [23]. In Drosophila adult, Crz is produced by a major group of neurosecretory cells in the brain and abdominal ganglion [24,25]. Given the complexity of the Crz neuronal architecture, Crz is predicted to deliver multiple biological functions: Crz neurons are shown to be associated with sperm transfer and copulation duration [26], regulation of trehalose levels [25], and responses to physiological and nutritional stresses [27,28].
Drosophila CrzR is a member of G-protein coupled receptor (GPCR) family and structurally homologous to the mammalian GnRH (Gonadotropin releasing hormone) receptor [29]. Physiological roles of the CrzR have not been explored. Using mutant flies lacking Crz neurons or CrzR, we investigated biological functions of the Crz signaling system in D. melanogaster. Both types of mutant flies displayed significantly delayed recovery from ethanol-induced sedation. We further show that such 'hangoverlike' phenotype of the CrzR mutant likely results from fast acetaldehyde accumulation due to higher ADH production as a result of transcriptional up-regulation and lower ALDH activity. Only the latter event was observed in the Crz-CD mutant, suggesting a complicated signaling mechanism associated with the CrzR.

Hangover-like phenotype mediated by Crz neurons
To explore the role of the Crz signaling pathway in ethanolinduced behavior, we employed Crz-cell deficient (Crz-CD) D. melanogaster that was produced by transgenic expression of a cell death gene hid. When flies were exposed to the vapor derived from 30-70% ethanol, we did not see a significant difference in the rates of sedation between controls and Crz-CD [30]. This is in contrast to a recent report that showed that Crz-CD is resistant to ethanolinduced sedation [16]. We do not know the cause of such discrepancy, except for the experimental setting.
During this study, however, we noticed that Crz-CD flies showed significantly delayed recovery from ethanol-induced sedation. To characterize this phenotype further, flies that were completely sedated with vapor from 100% ethanol for 17,18 min were allowed to recover in an ethanol-free environment. Control flies began to assume normal standing posture and mobile around 40 min and most of them recovered after 2 h. In contrast, the recovery of Crz-CD flies (Crz::hid) was significantly delayed; it was first observed at 80 min and maximum 80% of flies recovered by 4 h while the remaining flies did not survive (Figure 1, A and C). Crz-CD flies generated by ectopic expression of a different cell death gene reaper (rpr) produced results similar to the Crz::hid, whereas expression of a mutant rpr (nrpr, an rpr lacking IBM death domain) showed no difference from controls ( Figure 1B). These results suggest that Crz neurons are required for the recovery from ethanol-induced sedation. We refer to such delayed recovery from ethanol-induced sedation as ''hangover-like'' phenotype.
Severe hangover-like phenotype of a CrzR null mutant To investigate whether the foregoing result is due to the lack of Crz signaling, we generated a null mutant lacking CrzR (a.k.a. GRHRII). A putative hypomorphic mutant allele, Mi{ET1}GRH-RII MB00838 (for short, MB00838), carries a Minos transposable element bearing eye-specific GFP (green fluorescence protein) marker inserted in the fifth intron (Flybase; Figure 2A). To generate a CrzR-null mutant, we mobilized MB00838 using heat-shock-induced expression of Minos transposase [31]. The mobilization was confirmed by a mosaic pattern of GFP expression in the eyes. Subsequently, we established one hundred GFP-negative lines, one of which was identified to be a deletion mutant lacking CrzR by PCR. Further refined PCR confirmed an 8-kb deletion including exon 2-5 ( Figure 2). In addition to this deletion, PCR revealed a footprint of the gal4-coding region derived from the MB00838 element at its original insertion site. RT-PCR targeting the 5' region of CrzR (-59,+328) confirmed the absence of CrzR transcription, suggesting that this is a bona fide null mutant ( Figure  2C). Thus we designate this mutant as CrzR 01 for the first amorphic allele of this gene.
Homozygous CrzR 01 mutants were viable and fertile and had no discernible morphological or developmental aberrations. Mutant females are reproductively normal as their fecundity is comparable to wild-type ( Figure S1). In addition, circadian locomotor activity rhythms are indistinguishable between wild-type and CrzR 01 mutant flies, indicating that CrzR 01 mutants are as healthy as wild-type ( Figure S2). Next we measured hangover-like phenotype of the CrzR 01 mutant, as done for Crz-CD. When flies were exposed to ethanol for 17.5 min, the recovery of CrzR 01 flies was severely retarded; only 20% recovered fully after 4 h, whereas 70% of control flies did ( Figure 3A). To confirm whether this is caused by the deletion of the CrzR locus, we performed similar assays using hemizygous CrzR 01 (CrzR 01 /Df) or control (+/Df) with w 1118 background. After 17.5-min exposure, most Df/+ flies recovered after 2 h, while CrzR 01 /Df did after 4 h ( Figure 3B). The recovery of Df/+ flies was largely unaffected when the exposure duration was increased to 18 min, while only 70% of CrzR 01 /Df recovered by 4 h ( Figure 3C). Similar hangover-like phenotype was observed with RNA interference (RNAi)-mediated CrzR knock-down flies (CrzR-KD). CrzR RNAi was constructed as described in the supporting protocol (Protocol S1). In response to an actin-gal4, two independent lines, UAS-CrzR RNAiS3S and UAS-CrzR RNAiT17 , showed severe reduction of CrzR mRNA ( Figure S3A). These CrzR-KD flies displayed hangover-like phenotype, with around 20% and 35% of the flies recovered by 4 h, which is in contrast to 80-90% recovery of the heterozygous transgenic controls ( Figure S3B). Taking together with the data obtained with Crz-CD flies suggests that Crz signaling is causally associated with the hangover-like phenotype. This phenotype is unlikely to represent a general response to sedative agents, as neither Crz-CD nor a CrzR 01 mutation affected the recovery rates from ethyl ether-induced sedation ( Figure S4).

Defective acetaldehyde metabolism in CrzR 01 mutant
Ethanol is first oxidized to acetaldehyde, which has been considered a major cause of hangover symptoms in humans [32]. To test this possibility for CrzR 01 flies, we used gas chromatography to monitor the contents of acetaldehyde following exposure to ethanol. Remarkably acetaldehyde contents in both CrzR 01 and CrzR 01 /Df flies were up to 100% higher than those of control flies 45 min after ethanol exposure ( Figure 4A).
The accumulation of acetaldehyde in CrzR 01 mutants is possibly due to subnormal acetaldehyde oxidation as a result from lower ALDH activity. Indeed, ALDH activity in whole fly extract of CrzR 01 was only 63% of wild-type ( Figure 4B). Since acetaldehyde oxidation occurs mainly inside the mitochondria, the ALDH activity was measured with isolated mitochondria. Succinate dehydrogenase activity, a mitochondrial marker, was observed almost exclusively in the mitochondrial fraction in both genotypes, verifying successful separation of the mitochondria from the cytosol ( Figure 4C). Around 30% reduced ALDH activity was observed in the CrzR 01 mutant, while succinate dehydrogenase activity was indistinguishable between wild-type and CrzR 01 ( Figure 4D). Reduction of ALDH activity was also observed in

Loss of CrzR results in enhanced ADH activity
Since CrzR 01 mutants have lower ALDH activity, they could be more sensitive to exogenously provided acetaldehyde than wildtype. To test this, flies were fed low concentrations of acetaldehyde without causing acute toxicity [33]. Surprisingly, the CrzR 01 flies were more resistant to acetaldehyde-mediated intoxication than the wild-type. With 0.5% (v/v) acetaldehyde, , 50% of wild-type flies became immobilized after 60 h, while only 10% of CrzR 01 did so ( Figure 5A). Significant differences between the two genotypes were also apparent in response to 1% acetaldehyde.
Since D. melanogaster ADH is known to mediate the detoxification of the acetaldehyde by converting acetaldehyde to ethanol [33], we wondered if greater resistance to acetaldehyde by CrzR 01 is due to higher ADH activity in the mutant. Indeed, around 2.4fold increased ADH activity was measured in both homozygous CrzR 01 and CrzR 01 /Df as compared to wild-type ( Figure 5B). Df/+ flies showed a 1.5-fold increase of ADH activity, suggesting a dosage-dependent regulation of the ADH by CrzR.
In contrast to CrzR 01 , Crz-CD did not elevate ADH activity ( Figure 5, C and D), indicating that Crz neurons are required only for the regulation of ALDH activity. The reasons for such difference are currently unclear but may be related to an unknown ligand that controls CrzR activity or intrinsic activity of CrzR (see discussion).  To further delve into the mechanisms as to how Crz/CrzR regulates ADH and ALDH activities, mRNA levels of Adh and Aldh were analyzed by real-time RT-PCR. An average of 2.7-fold increase of Adh transcript levels were observed from the CrzR 01 using two different primer sets ( Figure 6A) while no difference was found for Aldh ( Figure 6B). These results indicate that CrzR is involved in the transcriptional regulation of Adh, but not of Aldh.
To elucidate signaling mechanisms underlying CrzR-regulated Adh transcription, transgenic manipulations were performed in the CrzR cells using a CrzR-gal4 line. Although we have been unsuccessful to obtain endogenous CrzR expression patterns using in situ hybridization, reporter gene expression patterns driven by the CrzR-gal4 lines were consistent with high throughput expression data from FlyAtlas (Flybase: flybase.org; see also ref. 28). Simultaneous expression of membrane-bound GFP (mCD8GFP) and nuclear RFP (Redstinger) showed respective florescent signals in the larval central nerve system (CNS) and salivary glands, but not in the larval fat body (Figure 7 A-C). Interestingly, unlike in the larval fat body, strong fluorescent signals were detected in the adult fat body, indicating developmental regulation of CrzR expression in this metabolic tissue ( Figure 7D). Such spatial and developmental expression patterns agree with the FlyAtlas data. Moreover, since the fat body is the major target tissue for the neuroendocrine regulation of hemolymph trehalose levels [34] and Crz is likely to be involved in such physiological event [25], CrzR expression in the adult fat body was expected.
CrzR is a member of Class-A GPCR family [20,29,35]. A signaling pathway typical of these receptors involves PKA, which regulates activity of a transcription factor CREB via phosphorylation. Thus we explored whether this is true for the CrzRregulated expression of Adh. To do this, PKA signaling was disrupted by ectopic expression of a PKA inhibitor (PKA in ), a dominant-negative form of the regulatory subunit of a Drosophila PKA [36], or Cbz, a dominant negative CREB [37] using CrzR-gal4. Remarkably, hangover-like phenotypes of these flies were comparable to that of the CrzR 01 mutant ( Figure 6, C and D). In addition, the inhibition of PKA also increased ADH activity and Adh mRNA levels significantly ( Figure 6, E and F), suggesting that CrzR-regulated Adh expression requires PKA-dependent pathway. In contrast, ectopic expression of the protein kinase C (PKC) inhibitor [38] showed a wild-type pattern of the hangover-like phenotype and ADH activity ( Figure 6, G and H), indicating no positive role for PKC.

Discussion
Drinking alcohol is one of the oldest habits of humankind and has some positive effects on human society, but uncontrolled consumption causes serious psychopathic symptoms and other health problems. Excessive consumption of alcoholic beverage and associated hangovers cost $223.5 billion per year in the U.S. alone, due to losses in workplace productivity, health care expenses, costs associated with law enforcement and criminal justice, and accidents (http://www.cdc.gov/Features/AlcoholConsumption/).
Depending on the concentrations, effects of ethanol consumption in humans include euphoria, impaired motor function and speech, followed by vomiting, coma and even death in certain cases. Upon cessation of drinking, hangover is characterized by unpleasant physical pains such as headache, sensory problems such as vertigo, gastrointestinal symptoms such as nausea and vomiting, and disruption of sleep and biological rhythms [39].
These symptoms are mainly from acetaldehyde accumulation, as supported by high incidence of ethanol intoxication in Eastern Asian populations due to the polymorphic deficiency of functional ALDH2 allele [40,41] and ALDH2-knockout mice [42]. In line with this, drugs inhibiting ALDH, such as disulfiram, have been used to treat chronic alcoholism by causing adverse symptoms from ethanol intake [43,44]. Despite these reports, relatively little is known about the regulation of ethanol metabolism by physiological factors.
Like mammals, D. melanogaster metabolizes ethanol to acetaldehyde and acetate catalyzed by ADH and ALDH, respectively [45], and Adh and Aldh mutant flies showed dramatically reduced tolerance to ethanol [46]. Intriguingly, our present data suggest that a neuroendocrine system involving Crz plays an important role in regulating ADH and ALDH activities. CrzR activation leads to the PKA-dependent transcriptional repression of Adh and to post-transcriptional activation of ALDH via unknown pathways, which together suppress acetaldehyde accumulation.
In contrast to CrzR 01 , Crz-CD did not affect ADH activity levels, but reduced ALDH activity. Such a difference could explain that the Crz-CD flies displayed milder hangover-like phenotype than CrzR 01 did. One might argue that Crz-CD still has residual Crz function. Although this is a possibility, undetectable Crz neurosecretory cells in the Crz-CD brains indicate near lack of Crz function. This is much more severe than RNAi-induced Crz knockdown with respect to the level of Crz expression [16]. Another caveat is that Crz-CD phenotype is due to an elimination of other co-existing transmitters. In fact, a subset of Crz neurons was shown to co-express small neuropeptide F (sNPF) [47]. However, since sNPF-producing neurons are so widely distributed in the brain, ablation of a few neurons co-expressing Crz and sNPF is unlikely to affect sNPF functions. Although more definitive evidence for the Crz await Crz-null mutant, our data with CrzRnull mutant support a role for Crz in the hangover-like phenotype.
Assuming that Crz-CD eliminates Crz function nearly entirely, we propose that CrzR mediates two separate signaling pathways; Crz-dependent up-regulation of ALDH activity and Crz-independent down-regulation of Adh transcription. The latter pathway might involve a distinct ligand that activates CrzR. One possible candidate is adipokinetic hormone (AKH), as the AKH receptor is structurally related to the CrzR with 56% amino acid sequence similarity [29]. However, CrzR showed very little affinity for AKH [35], indicating that AKH is unlikely to be a natural ligand for CrzR. Alternatively, CrzR might have an intrinsic activity that is not required for ligand binding. It is not uncommon that some GPCRs have intrinsic (or spontaneous) activities in the absence of ligand biding [48].
Although the identity of the second ligand for CrzR is speculative, a growing body of evidence suggests that a GPCR couples to different G-proteins in response to different agonists; such molecular flexibility is often referred to as ''functional selectivity'' of GPCR [49,50]. According to this hypothesis, binding properties of different ligands induce and stabilize a unique conformational status of GPCR, which in turn shifts coupling preference to different G-proteins. Recently, CrzR isolated from the silkworm, Bombyx mori, was shown to couple dually to the Gq and Gs proteins in cell-based assays [51]. However, our results indicate that Gq-associated PKC activation is unlikely to be involved in the regulation of Adh transcription in Drosophila. Thus we propose that Gs-led PKA activation is the major in vivo signaling pathway of the CrzR at least for the Adh regulation. Nevertheless, it seems that CrzR is an excellent model system to unravel the physiological dynamics of the GPCR.
The International Agency for Research on Cancer (IARC) classified ethanol as a Group 1 carcinogen (http://monographs. iarc.fr/ENG/Classification/ClassificationsAlphaOrder.pdf). Accumulation of acetaldehyde as a result of defective ALDH2 is a major culprit of the carcinogenesis [52,53]. In line with this, many cancer cells have disproportional activities of ADH and ALDH, making them less efficient in removing acetaldehyde compared to normal tissues [54]. In this regard, CrzR 01 flies might be prone to cancer development, and perhaps can serve as an interesting
CrzR null mutant and CrzR-gal4 line Mobilization of the y w; Mi{ET1}GRHRII MB00838 (Bloomington stock no. 22910) was used to induce CrzR mutation. Homozygous GRHRII MB00838 females were crossed to males carrying a genomic source of heat-shock inducible Minos transposase (hsLMit 2.4 ). Progenies were incubated at 37uC water bath for 1 h per day until pupariation. Fifty single male progeny with eyes showing mosaic pattern of GFP were crossed individually to a balancer y w;; Ly/TM6C, Sb, Tb females. Two GFP-negative progeny derived from each cross were randomly selected and crossed individually to the balancer to establish a total of 100 excision lines. Genomic DNA from homozygous flies was used for PCR-based screening to detect the deletion of the CrzR locus. For CrzR-gal4, 3.5-kb sequence upstream of the CrzR (-3130 to +331, +1 is transcription start site) was PCR amplified (primer sequences are shown in Table S1). This fragment was cloned into the pPTGAL vector at Xba I/Not I sites for germline transformation.

Behavior assays
Groups of 25 males (1-3 days old) were maintained in a food vial for 1-2 days and used for the following behavior assays.
Recovery from ethanol sedation. Flies were transferred into an empty plastic vial (O.D.6H: 25695 mm) and then sealed with cotton plug. One ml of 100% ethanol was applied onto the cotton plug. Following 17-18 min of exposure, the cottons were replaced with fresh buzz plugs and then the vials were placed upside down. Flies capable of climbing up were considered ''recovered'' and the numbers of recovered flies were recorded every 10 min.
Toxicity assay. Assay was performed according to Barbancho et al [56], with slight changes. Groups of 20 males were placed in hermetically sealed plastic vials (O.D.6H: 25695 mm) containing 3% sucrose medium supplemented with acetaldehyde. The toxicity of acetaldehyde was represented by the number of knocked-down flies as a function of exposure time.
Procedures of fecundity, circadian rhythm and ethyl ether recovery assays are provided in supporting protocol (Protocol S1).

Activity assays of ADH and ALDH
A hundred flies were homogenized in 1 ml of isolation buffer [50 mM sodium phosphate (pH 7.4), 0.24 M sucrose, 0.5 mM EDTA, 0.5 mM dithiothreitol, 0.001% (w/v) phenylthiourea], as described [57]. The homogenate was centrifuged for 10 min at 1000 rpm and then the supernatant was saved. The pellet was resuspended in 0.5 ml of the isolation buffer. Following centrifugation as before, the supernatant was combined with the first one. The pooled supernatant was centrifuged for 45 min at 14,000 rpm to separate cytosolic and mitochondrial fractions [58]. The pellet was washed with isolation buffer twice and resuspended in 250 ml of the same buffer supplemented with 1% (v/v) Triton-X-100. The resuspension was allowed to stand for 15 min and then centrifuged at 14,000 rpm for 20 min. The supernatant was used for mitochondrial fraction. All operations were performed on ice except centrifugation at 4uC. The protein content was determined by the BCA (Bicinchoninic acid) method. Three independent extracts per genotype were prepared. ALDH and ADH activities were assayed spectrophotometrically by measuring reduced b-NAD + at 340 nm. ALDH activity assay was done following the method of Moxon et al. [59]. Briefly, the assay was started by adding 25 ml fly extract into 500 ml of 50 mM Tris/HCl (pH 8.6) including 3.6 mM acetaldehyde, 1 mM b-NAD + , and 20 mM pyrazole. The absorbance was recorded every min for up to 20 min by using Smartspect TM 3000 kinetics program (BioRad). ADH activity was assayed, as described in Barbancho et al. [56], by adding 20 ml extract into 500 ml of 50 mM Tris/HCl (pH 8.6) buffer containing 100 mM 2-propanol, 1 mM b-NAD + and 1.6 mM cyanamide. The absorbance was recorded every 10 sec for up to 3 min. Succinate dehydrogenase activity was assayed by following the reduction of INT [2-(p-indophenyl)-3-(p-nitrophenyl)-5-phenyl tetrazolium] [60]. The reaction was initiated by adding 2 ml of mitochondrial or 20 ml of cytosolic fraction into 200 ml substrate solution (0.11% w/v INT, 55 mM sodium succinate, 25 mM sucrose in 100 mM sodium phosphate, PH 7.4), followed by incubation in 37uC water bath for 15 min. The reaction was stopped by adding 200 ml of 3% ice-cold HCl. The chromogen was extracted with 600 ml ethyl acetate by centrifugation at 14,000 rpm for 5 min, and then its absorption was determined at 490 nm.

Measurement of acetaldehyde content
Forty flies were homogenized in 1 ml sterilized ice-cold water with 20 mM pyrazole and 1.6 mM cyanamide to inhibit ADH and ALDH activities. The homogenates were centrifuged at 14,000 rpm for 20 min at 4uC and the supernatant was kept in ice. Standard solutions were prepared from pure acetaldehyde in the same solution. Acetaldehyde content was measured using an Agilent 7890 gas chromatograph equipped with Agilent model G1888 headspace auto sampler, a flame ionization detector and a DB-624 capillary column (60 m x 0.32 mm x 1.8 mm). Separation of acetaldehyde was complete under the following conditions; injection volume 0.2 ml; split ratio 50:1; injection port temperature 200uC; oven temperature program: 60uC for 2 min, 25uC min 21 to 200uC for 1 min, FID temperature 300uC; carrier gas N 2 ; and gas flow 3.0 ml/min (helium). For the auto sampler: auto sampler oven temperature 70uC; transfer line temperature 125uC; loop temperature 125uC; vial equilibration time 15 min; high shaking (mixing) speed; loop fill time 0.03 min; inject time 0.50 min; vial pressure 10 psig; pressurization time 0.5 min. All determinations were carried out in triplicates.

Quantitative RT-PCR
Total RNA was purified from 20 adult males using TRIzol reagent (Invitrogen) according to the manufacture's protocol. About 5-10 mg of total RNA was added to 40 ml of Go Superscript reverse transcription mix (Promega) to generate cDNA using the oligo-dT-Ad primer (GACTCGAGTCGACATCGAT 20 ). cDNA was then purified and eluted with 40-ml water using a PCR purification kit (Qiagen). SYBR Green reaction kit (Bio-Rad) was used for real-time PCR, with 0.5 ml cDNA included in 25 ml reaction volume. The PCR was carried out using an iQ TM 5 Multicolor Real-Time PCR Detection System (Bio-Rad). Amplifications were initiated with a 10-min denaturation at 95uC, followed by 40 cycles of 95uC, 15sec 55uC, 30 sec 72uC, 30 sec. Reactions were run in triplicates. Primers (Table S1) were designed to flank at least one intron to ascertain that the PCR products were derived from the cDNA. b-tubulin was used as an internal control to normalize expression levels of the target genes. Expression data were obtained using the 2 2DDCT method, as described by Livak and Schmittgen [61] where DDCT equals the normalized cycle threshold (DCT) of Aldh or Adh in test genotypes minus the DCT of the same gene in control flies.

Histology
CrzR-gal4 line was crossed to UAS-mCD8GFP;; UAS-Redstinger. The resulting male offspring were dissected and incubated in fixative (4% paraformaldehyde in PBS) for 30 min at room temperature. Tissues were washed in PBS (365 min) and mounted in a quenching medium (0.5% n-propyl gallate in 90% glycerol and 10 mM phosphate buffer, pH 7.4). Differential fluorescent signals were captured by Olympus BX-61 microscope equipped with a CCD camera and then merged.

Statistics
Statistical analyses were done using Instat 2.0 (GraphPad Software). Student's unpaired t-test was used to determine significant differences between two means. Difference among multiple means was tested by one-way ANOVA followed by Student-Newman-Keuls (SNK) multiple comparisons test.