Identification of a Drosophila melanogaster glutamate-gated chloride channel sensitive to the antiparasitic agent avermectin.

Glutamate-gated chloride channels, members of the ligand-gated ion channel superfamily, have been shown in nematodes and in insects to be a target of the antiparasitic agent avermectin. Two subunits of the Caenorhabditis elegans glutamate-gated chloride channel have been cloned: GluCl-α and GluCl-β. We report the cloning of a Drosophila melanogaster glutamate-gated chloride channel, DrosGluCl-α, which shares 48% amino acid and 60% nucleotide identity with the C. elegans GluCl channels. Expression of DrosGluCl-α in Xenopus oocytes produces a homomeric chloride channel that is gated by both glutamate and avermectin. The DrosGluCl-α channel has several unique characteristics not observed in C. elegans GluCl: dual gating by avermectin and glutamate, a rapidly desensitizing glutamate response, and a lack of potentiation of the glutamate response by avermectin. The pharmacological data support the hypothesis that the DrosGluCl-α channel represents the arthropod H-receptor and an important target for the avermectin class of insecticides.

Glutamate-gated chloride channels, members of the ligand-gated ion channel superfamily, have been shown in nematodes and in insects to be a target of the antiparasitic agent avermectin. Two subunits of the Caenorhabditis elegans glutamate-gated chloride channel have been cloned: GluCl-␣ and GluCl-␤. We report the cloning of a Drosophila melanogaster glutamate-gated chloride channel, DrosGluCl-␣, which shares 48% amino acid and 60% nucleotide identity with the C. elegans GluCl channels. Expression of DrosGluCl-␣ in Xenopus oocytes produces a homomeric chloride channel that is gated by both glutamate and avermectin. The DrosGluCl-␣ channel has several unique characteristics not observed in C. elegans GluCl: dual gating by avermectin and glutamate, a rapidly desensitizing glutamate response, and a lack of potentiation of the glutamate response by avermectin. The pharmacological data support the hypothesis that the DrosGluCl-␣ channel represents the arthropod H-receptor and an important target for the avermectin class of insecticides.
Glutamate-gated chloride channels were first identified in arthropods as extrajunctional glutamate receptors (H-receptors) that hyperpolarized the membrane potential of locust (Schistocerca gregaria) leg muscle (1)(2)(3) and later cloned from the soil nematode Caenorhabditis elegans (4). Glutamate-gated chloride channels are activated by the glutamate analog ibotenic acid and are inhibited weakly by the ligand-gated chloride channel blocker picrotoxin (1, 4 -9). No glutamate-gated chloride channels have been identified in vertebrate species.
Glutamate-gated chloride channels are important targets for the widely used avermectin class of anthelmintic and insecticidal compounds (10,11). The avermectins are a family of macrocyclic lactones used throughout the world to treat parasitic helminths and insect pests of man and animals (12). Avermectins directly activate C. elegans glutamate-gated chloride channels when expressed in Xenopus oocytes (4,9,13) and activate a glutamate-sensitive chloride channel in locust muscle (10).
The C. elegans avermectin-sensitive glutamate-gated chloride channel (GluCl) 1 when expressed in Xenopus oocytes, ex-hibits a pharmacology that is distinct from all other known ligand-gated chloride channels (4,9,13). The activity of the avermectins on the C. elegans GluCl channels shows a good correlation with their in vivo activity, suggesting that the GluCl channels are the targets for avermectin action (11). After many years of use in the field, there remains little resistance to avermectin in the insect populations (14). The combination of a good therapeutic index and low resistance strongly suggests that the GluCl channels remain good targets for insecticide development. To understand further the mode of action of the avermectins in insects we have cloned a GluCl channel subunit from Drosophila melanogaster.

RNA Isolation and Northern
Analysis-Poly(A) ϩ RNA was prepared from heads of the Oregon-R strain of adult D. melanogaster, heads of the fifth instar larvae of the Helicoverpa zea (tomato fruitworm) and Spodoptera fugiperda (fall army worm), Ctenocephalides felis (flea), Ornithodoros moubata (tick), and Tetranychus urticae (two spotted spider mite), which were frozen and ground in liquid N 2 . RNA was extracted with TriReagent (Molecular Research) and poly(A) ϩ RNA was isolated by two rounds of purification on oligo(dT)-cellulose columns (Pharmacia Biotech Inc.). Northern blots were hybridized with a probe containing the DrosGluCl-␣ cDNA corresponding to the largest open reading frame.
Isolation of a Drosophila GluCl Polymerase Chain Reaction (PCR) Product-Two degenerate oligonucleotides were synthesized based upon the predicted amino acid sequences of the conserved M1 and M3 transmembrane domains of C. elegans GluCl-␣ (4): a 5Ј primer, 5Ј-TGGGT(AGCT)(TA)(CG)(AGCT)TT(CT)TGGTT-3Ј; and a 3Ј primer, 5Ј-GC(TGCA)CC(TGA)ATCCA(TGCA)AC(GA)TC(TGA)AT-3Ј. The 3Ј primer was used to synthesize first strand cDNA from Drosophila poly(A) ϩ RNA using Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.). The cDNA was used as the template for a PCR containing 1.2 M each of the 5Ј and 3Ј primers with AmpiTaq DNA polymerase (Perkin-Elmer) in the presence of 20 Ci of [ 32 P]dCTP. The thermocycler was programmed for 25 cycles at 94°C for 1 min, 37°C for 2 min, and 72°C for 3 min. The 32 P-labeled PCR products were electrophoresed on a 6% acrylamide-urea sequencing gel (15). The DNA from the region of the gel corresponding to approximately 152 base pairs (the distance between the M1 and M3 region of C. elegans GluCl-␣) was eluted and subjected to two more rounds of PCR amplification using the same primers and incubation conditions. The amplified PCR products were ligated into the vector pCRII (Invitrogen). One clone, pPCR-2, whose DNA sequence was 70% identical to the corresponding region of C. elegans GluCl-␣, was selected as a hybridization probe.
Identification and Analysis of the DrosGluCl cDNA Clone-A random primed cDNA library in the phagemid cloning vector ZAPII (Stratagene) was made from RNA isolated from heads of the Oregon-R strain of Drosophila (kindly provided by Dr. T. Schwartz, Stanford University). This library was screened with the insert from pPCR-2 using established procedures (15). Thirty-six positive phage clones were identified and converted into plasmids by in vivo excision following the manufacturer's instructions (Stratagene). One clone, DrosGluCl-␣, was * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U58776.
Electrophysiological Recordings-Xenopus oocytes were prepared, injected (in vitro RNA 0.1-10 ng), and membrane currents recorded as described previously (9,13). Recordings were made in standard frog saline consisting of (in mM): NaCl (115), KCl (2), MgCl 2 (1), CaCl 2 (1.8), HEPES (10) adjusted to pH 7.5 with NaOH. Pipettes were filled with 3 M KCl and had resistances between 1.0 and 3.0 megohm. Avermectin analogs were prepared as 1 mM stock solutions in dimethyl sulfoxide and diluted into frog saline. A water-soluble avermectin derivative, ivermectin phosphate (22,23-dihydroavermectin B1 a -4ЈЈ-O-phosphate, monosodium salt, IVMPO 4 ), was used for most of the experiments because it is easily washed out of the perfusion system (11,13). Picrotoxin (Sigma) was made as a 50 mM stock solution in dimethyl sulfoxide. Concentration-response curves were fit to a modified Michaelis-Menten equation: where I max is the maximal response, [D] is the drug concentration, EC 50 is the drug concentration for half-maximal effect, and H is the Hill coefficient. All data in text and figures are means Ϯ S.E. Methods for obtaining concentration effect curves for IVMPO 4 can be found in Ref.

11.
Current voltage (I/V) relationships were determined using 1-s voltage ramps from Ϫ110 to 80 mV in the presence and absence of IVMPO 4 . Drug-sensitive currents were generated by subtracting drug-free from drug-containing data. Ion substitutions were accomplished by replacing NaCl in standard frog saline with KCl, choline chloride, or sodium isethionate as indicated. The bath was connected to the ground through a 3 M KCl agar bridge for experiments in which external chloride was replaced. To limit contributions from calcium-activated chloride currents to the I/V relationship, oocytes were preincubated in frog saline containing 10 M BAPTA-AM (11).

RESULTS
Isolation of a GluCl cDNA from D. melanogaster-Alignment of the C. elegans GluCl proteins with the GABA A and glycine receptor subunits highlighted areas of conservation as well as regions of diversity (4). Degenerate oligonucleotide primers were designed which, in theory, would generate a PCR product related to the GluCl family and not the GABA A or glycine gene families. Several PCR clones were obtained which contained inserts representing the region spanning the M1-M3 domains of the GluCl proteins. A Drosophila cDNA clone (DrosGluCl-␣) was obtained by hybridization screening of a cDNA library with one of the M1-M3 PCR products. The nucleotide sequence of DrosGluCl-␣ (3,958 base pairs) revealed an open reading frame that predicted a protein of 456 amino acids which contained features common to ligand-gated anion channels, such as a large amino-terminal putative extracellular domain and four putative transmembrane domains (Fig. 1).
The DrosGluCl-␣ cDNA contains a 5Ј-and 3Ј-untranslated region of 404 and 2186 base pairs, respectively. It is missing a 3Ј-poly(A) tract and is presumed to be less than full length. Alignment of the DrosGluCl-␣ protein with the C. elegans GluCl-␣ and -␤ and Drosophila rdl proteins revealed 48, 43, and 37% identity and 67, 62, and 58% similarity, respectively (Fig. 1). Drosophila rdl represents the closest related example of a Drosophila ligand-gated chloride channel and was therefore included for comparison. Four cysteines present in the amino-terminal domain of DrosGluCl-␣ are also found in the C. elegans GluCl-␣ and -␤, the Lymnae stagnalis protein (16), and the glycine receptors (17). Two of these cysteines are absent in the Drosophila rdl GABA-sensitive receptor and in all other GABA receptors subunits. DrosGluCl-␣ is most closely related to the ␣ subunit of the glycine receptor family, showing 57% identity. Phylogenetic analysis of all ligand-gated channels indicates that DrosGluCl-␣ is present on a separate branch along with the C. elegans GluCl-␣ and -␤ (data not shown).
Functional Expression of DrosGluCl-␣--Glutamate evoked a rapidly activating inward membrane current in Xenopus oocytes injected with in vitro RNA transcribed from Dros-GluCl-␣ ( Fig. 2A). The EC 50 for glutamate activation of current was 23 M, and the Hill coefficient was 2.0, suggesting that binding of more than one glutamate was necessary to gate the channel. Glutamate-sensitive current desensitized rapidly in the continued presence of glutamate ( Fig. 2A). The rate and magnitude of desensitization were concentration-dependent. In the presence of 100 M glutamate, the time constant for desensitization was 2.5 Ϯ 0.3 s, and current desensitized 87 Ϯ 2% of the peak value in 15 s (n ϭ 5). With 10 M glutamate the time constant for desensitization was 6.5 Ϯ 1.2 s, and the current desensitized 24 Ϯ 7% of the peak current (n ϭ 5). We did not systematically measure the time course for recovery from desensitization, but applications of 300 M glutamate separated by 1 min had identical amplitudes, demonstrating that recovery from desensitization was complete within 1 min.
Ibotenic acid, a structural analog of glutamate known to activate glutamate-gated chloride channels (1, 5-9), maximally activated glutamate-sensitive current with desensitization kinetics similar to glutamate (Table I). L-Aspartate (1 mM) acti-vated 5% of the maximal current, whereas other glutamate analogs such as kainic acid, quisqualic acid, and N-methyl-Daspartic acid were inactive. Agonists known to activate vertebrate and invertebrate chloride channels such as glycine, GABA, and histamine were also inactive. IVMPO 4 slowly and irreversibly activated current in oocytes injected with DrosGluCl-␣ (Fig. 2B). Washing with drug-free frog saline for 10 min after application of IVMPO 4 failed to return the current to base line. The EC 50 for IVMPO 4 was 41  nM, and the Hill coefficient was 1.2 (Fig. 2B). Octahydroavermectin, a biologically inactive avermectin analog with poor receptor binding affinity, activated 11% of the current activated with IVMPO 4 (Table I). Ivermectin, the commercially used insecticide abamectin, and the avermectin analog L-648,548, all maximally activated current at 500 nM (Table I). The maximal current elicited with IVMPO 4 was 35% of the maximal current elicited with 300 M glutamate.
Avermectins potentiate the glutamate response of C. elegans GluCl channels (4,9,11). Pretreatment of oocytes expressing C. elegans GluCl channels with concentrations of avermectin below the threshold for direct activation increases glutamatesensitive currents 300 -800% (4,9,11). In contrast, the glutamate (10 M) response of DrosGluCl-␣-injected oocytes was potentiated only 24 Ϯ 5% (n ϭ 7), after a 2-min pretreatment with 1 nM IVMPO 4 (Fig. 3). The peak glutamate-sensitive current elicited with 300 M glutamate was not altered significantly. When the IVMPO 4 concentration during the pretreatment was increased to 10 nM, we failed to observe any significant potentiation of the glutamate response and found a 49 Ϯ 5% (n ϭ 3) reduction in the maximal glutamate response. In addition, 10 nM IVMPO 4 slowed the rate of desensitization of the glutamate response (data not shown).
The current voltage relationship for the IVMPO 4 -sensitive current was slightly inwardly rectifying (Fig. 4A). The reversal potentials of Ϫ28 Ϯ 1 mV (n ϭ 3) and Ϫ27 Ϯ 1 mV (n ϭ 4) for glutamate and IVMPO 4 -sensitive currents, respectively, were close to the Nernst potential for chloride (Ϫ28 mV, assuming 40 mM intracellular chloride) (18). The reversal potential of the IVMPO 4 -sensitive current showed a strong dependence on ex-ternal chloride with a 57-mV change for a 10-fold change in external chloride (Fig. 4B). Replacement of external NaCl with KCl or choline chloride did not shift the reversal potential or the shape of the current voltage relationship. Picrotoxin is an inhibitor of ligand-gated chloride channels which inhibits glutamate-gated chloride channels weakly (1, 4 -9). The IVMPO 4sensitive DrosGluCl-␣ current was resistant to picrotoxin at concentrations up to 100 M and was blocked only 14% with 500 M.
Analysis of DrosGluCl Transcripts-Several different size classes of RNA transcripts were identified, by Northern blot analysis, from three different developmental stages of Drosophila: embryo, larvae, and adult (Fig. 5A). Transcripts of 4.4 and 9.0 kb were found exclusively in embryo and larvae, and those , and mite (lane 8) was separated (3 g each except 10 g for rat) and transferred to a Nytran membrane. The filter was hybridized at low stringency (50% formamide at 30°C) with the Dros-GluCl-␣ probe and washed in 6 ϫ SSC; 0.1% sodium dodecyl sulfate at 52°C. of 9.7 and 6.0 kb were found solely in adult. The predominant 3.0-kb transcript, found in all stages, appeared to be underrepresented in larvae. Since the 4.0-kb DrosGluCl-␣ cDNA was isolated from an adult cDNA library it may be derived from any one of the larger transcripts detected by Northern analysis. The multiple transcripts detected in this high stringency hybridization analysis of Drosophila RNA may indicate the presence of other GluCl subunits, as has been observed in C. elegans (4). Multiple related subunits have been identified in other ligand-gated chloride channels (19).
Transcripts related to DrosGluCl-␣ were found in other insects: C. felis (flea), O. moubata (tick), S. fugiperda (fall army worm), and H. zea (tomato fruitworm), but not in the arthropod T. urticae (two spotted spider mite). High stringency washes of this Northern blot indicated that there was a high degree of identity between DrosGluCl-␣ and the major transcripts found in flea, fall army worm, and tomato fruitworm (data not shown). No C. elegans transcripts were detected, even though the DrosGluCl-␣ cDNA is 60% identical to the C. elegans GluCl-␣ and -␤ cDNAs. Several rat transcripts hybridized at low stringency to DrosGluCl-␣ but were only faintly detected following a high stringency wash (data not shown).

DISCUSSION
DrosGluCl-␣ encodes a glutamate-gated chloride channel with characteristics similar to those first reported for the Hreceptor glutamate-gated chloride channels on locust leg muscle (1)(2)(3)8). Insect glutamate-gated chloride channels have since been reported in Drosophila muscle (20), in several insect neuronal preparations (6,7,21), and have been expressed from S. gregaria leg muscle mRNA (22). DrosGluCl-␣ channels expressed in Xenopus oocytes are pharmacologically and kinetically similar to insect glutamate-gated chloride channels. Both the expressed and native channels desensitize rapidly in the continued presence of glutamate and are activated by the structurally constrained glutamate analog ibotenate (1,2,8,23,24). Similarly, native and expressed channels are activated weakly with aspartate (23) and inhibited with high concentrations of picrotoxin. Finally, avermectins directly activate glutamategated chloride channel conductance in S. gregaria leg muscle (10,25), as they do for the DrosGluCl-␣ subunit expressed in Xenopus oocytes.
The physiological relevance of the glutamate-gated chloride channels remains unknown. Excitatory glutamate-gated cation channels are located at insect neuromuscular junctions, the site of glutamate release (26). Glutamate-gated chloride channels, on the other hand, are located extrajunctionally on the muscle and on neuronal cell bodies, both of which are distal to the glutamate synapses (1-3, 6, 8, 21, 23). It is possible that the glutamate-gated chloride channels are activated by "spillover" of glutamate from hyperactive synapses to limit muscle excitability. Alternatively, the inhibitory action of the GluCl channel would act to limit calcium entry following muscle or nerve damage. Other roles could involve developmental regulation and feeding behavior. Genetic analysis with Drosophila mutants defective in the DrosGluCl-␣ gene may resolve these issues.
Several properties of the DrosGluCl-␣ channel differ from the nematode C. elegans GluCl-␣ and -␤ channels. A single insect subunit is sufficient to enable avermectin and glutamate gating, whereas two nematode subunits are required (4). The DrosGluCl-␣ channel desensitizes rapidly and completely in the continued presence of glutamate compared with a slow and incomplete desensitization of the nematode channel (4). Finally, the DrosGluCl-␣ channel is less sensitive to avermectin potentiation of the glutamate response. The presence of other related genes in both C. elegans and Drosophila is evident both by Northern analysis and by PCR amplification of related genes (data not shown). It remains to be seen whether there are additional subunits in Drosophila which, like the nematode GluCl-␣ and -␤, are uniquely sensitive to either avermectin or glutamate.
The results presented here clearly show that the Dros-GluCl-␣ channel is a target for the avermectin class of insecticides. Avermectins also interact with insect GABA channels (25,27,28), therefore, multiple targets for the avermectins may be present in insects. However, since no GluCl channels have been identified in mammals, where GABA channels are ubiquitous, the GluCl channels may be the safer target for future insecticide development.