Genetic and Biochemical Evidence for a Novel Avermectin-sensitive Chloride Channel in Caenorhabditis elegans

Avermectins are a class of macrocyclic lactones that is widely used in crop protection and to treat helminth infections in man and animals. Two complementary DNAs (GluClα and GluClβ) encoding chloride channels that are gated by avermectin and glutamate, respectively, were isolated from Caenorhabditis elegans. To study the role of these subunits in conferring avermectin sensitivity we isolated a mutant C. elegansstrain with a Tc1 transposable element insertion that functionally inactivated the GluClα gene (GluClα::Tc1). GluClα::Tc1 animals exhibit a normal phenotype including typical avermectin sensitivity. Xenopus oocytes expressing GluClα::Tc1 strain mRNA elicited reduced amplitude avermectin and glutamate-dependent chloride currents. Avermectin binding assays in GluClα::Tc1 strain membranes showed the presence of high affinity binding sites, with a reducedB max. These experiments suggest that GluClα is a target for avermectin and that additional glutamate-gated and avermectin-sensitive chloride channel subunits exist in C. elegans. We isolated a cDNA (GluClα2) encoding a chloride channel that shares 75% amino acid identity with GluClα. This subunit forms homomeric channels that are gated irreversibly by avermectin and reversibly by glutamate. GluClα2 coassembles with GluClβ to form heteromeric channels that are gated by both ligands. The presence of subunits related to GluClα may explain the low level and rarity of target site involvement in resistance to the avermectin class of compounds.

Avermectins are a class of macrocyclic lactones that is widely used in crop protection and to treat helminth infections in man and animals. Two complementary DNAs (GluCl␣ and GluCl␤) encoding chloride channels that are gated by avermectin and glutamate, respectively, were isolated from Caenorhabditis elegans. To study the role of these subunits in conferring avermectin sensitivity we isolated a mutant C. elegans strain with a Tc1 transposable element insertion that functionally inactivated the GluCl␣ gene (GluCl␣::Tc1). GluCl␣::Tc1 animals exhibit a normal phenotype including typical avermectin sensitivity. Xenopus oocytes expressing GluCl␣::Tc1 strain mRNA elicited reduced amplitude avermectin and glutamate-dependent chloride currents. Avermectin binding assays in GluCl␣::Tc1 strain membranes showed the presence of high affinity binding sites, with a reduced B max . These experiments suggest that GluCl␣ is a target for avermectin and that additional glutamate-gated and avermectinsensitive chloride channel subunits exist in C. elegans. We isolated a cDNA (GluCl␣2) encoding a chloride channel that shares 75% amino acid identity with GluCl␣. This subunit forms homomeric channels that are gated irreversibly by avermectin and reversibly by glutamate. GluCl␣2 coassembles with GluCl␤ to form heteromeric channels that are gated by both ligands. The presence of subunits related to GluCl␣ may explain the low level and rarity of target site involvement in resistance to the avermectin class of compounds.
Extensive use of parasiticides and pesticides has resulted in the emergence of resistant strains of target organisms. Characterization of the molecular targets of such compounds would help elucidate their mode of action and mechanisms of resistance. Avermectins are a class of macrocyclic lactones with insecticidal and nematocidal activity. Abamectin (avermectin B 1 ) is a miticide and insecticide used in crop protection. Ivermectin (22,23-dihydroavermectin B 1 ) is an anthelmintic used to treat Dirofilaria immitis (heartworm) infections in dogs and onchocerciasis (river blindness) in humans (1). Avermectins have been shown to increase chloride permeability in insect and crustacean muscles (2-8). Incubation of Caenorhabditis elegans with ivermectin results in inhibition of pharyngeal pumping and egg laying as well as overall paralysis (9 -12). Xenopus laevis oocytes injected with C. elegans mRNA express avermectin-sensitive glutamate-gated chloride (GluCl) 1 channels (13)(14)(15). Using Xenopus oocytes as an expression system, two functional cDNAs (GluCl␣ and GluCl␤) encoding C. elegans glutamate-gated and avermectin-sensitive chloride channel subunits were isolated (16). The individual subunits expressed functional homomeric channels that were selectively responsive to ivermectin or glutamate, respectively. Coexpression of GluCl␣ and ␤ lead to formation of heteromeric channels that exhibited the rapidly activating reversible glutamate-and irreversible ivermectin-sensitive currents found in C. elegans mRNA injected oocytes (15,16). Phylogenetic analysis suggests that GluCl␣ and GluCl␤ represent an evolutionarily distinct class of ligand-gated ion channels that may be orthologous to the vertebrate glycine receptors (17).
Extensive use of avermectins has led to the appearance of resistance to this class of compounds. Low level (3-10-fold) ivermectin resistance has only been detected in parasitic nematodes of goats and sheep. In most cases multiple resistance to other groups of drugs is involved (18). These findings suggest that either ivermectin acts on multiple targets or that the target site is involved in essential processes.
The extensive genetic characterization of C. elegans makes it a useful system to study the function of GluCl channels in vivo and the mechanisms of resistance to ivermectin. In an attempt to understand the role of GluCl channels in conferring avermectin sensitivity and their physiological function, we determined the chromosomal localization of the GluCl␣ and GluCl␤ genes and searched for known mutations mapping to these loci. However, mutations attributed specifically to these genes could not be identified. We then identified a C. elegans strain with a Tc1 transposable element insertion in the GluCl␣ gene. Here, we report the biochemical characterization of this mutant strain and the isolation of a novel subunit (GluCl␣2) with properties similar to GluCl␣.

EXPERIMENTAL PROCEDURES
Chromosomal Location of the GluCl␣ and GluCl␤ Genes-Grids with ordered YAC clones were hybridized with 32 P-labeled probes using the * 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.
¶ To whom correspondence should be addressed: Identification of the GluCl␣::Tc1 Strain-Pools of DNA lysates corresponding to a mutant bank of frozen C. elegans strains with random Tc1 transposable element insertions were screened by PCR (19). Specific primers for the GluCl␣ gene included the nested primers DKV1.3 (5Ј-TAATGGAGGACCAGTTGTGG-3Ј) and DKV1.4 (5Ј-GTGATTCT-CACTGTTGGAC-3Ј) and for Tc1 RI (5Ј-GCTGATCGACTCGATGC-CACGTCG-3Ј) and RII (5Ј-GATTTTGTGAACACTGTGGTGAAG-3Ј) (Fig. 2). Three orthogonal matrices of 10 96-tube trays were screened with each set of the GluCl␣-specific nested primer pair one-dimension at a time in quadruplicate. A single lysate corresponding to a particular frozen stock in the mutant library was identified. This stock was thawed and a single animal with a Tc1 insertion in the GluCl␣ gene was identified by single worm PCR (19). A homozygote strain for the GluCl␣::Tc1 insertion was identified in the progeny of the original animal by Southern analysis. The GluCl␣::Tc1 strain was backcrossed five times to remove the mutator locus and other possible Tc1 insertions. Some of the strains used were obtained from the Caenorhabditis Genetics Center. GluCl␣::Tc1 hermaphrodites were backcrossed to dpy-5/ϩ males. dpy-5 balances the mut-2 locus. Homozygous Dpy F2 progeny that no longer carry the mut-2 locus were identified. These animals were used for further backcrosses to unc-76/ϩ males. The use of the unc-76 marker helped to balance the GluCl␣::Tc1 insertion. The presence of the GluCl␣::Tc1 mutation after each backcross was verified by Southern analysis and nested PCR reactions. The exact location of the Tc1 insertion was identified by nested PCR amplification of genomic DNA from the GluCl␣::Tc1 strain using the DKV1-3, DKV1-4, and RI and RII primers, and subcloning into the TA cloning vector (Promega). The sequence of the insertion point was determined using the dideoxy termination method (Amersham).
DNA and RNA Isolation-C. elegans strains N 2 and GluCl␣::Tc1 maintained on seeded agar Petri dishes were used to start 1-liter liquid cultures in S Basal media (20). The resuspended bacterial pellet of a 1-liter overnight culture of Escherichia coli (OP50) in LB was added to the S Basal and the cultures were shaken overnight at ambient temperature. The bacteria were replenished 24 h later and the incubation continued overnight. The worms were isolated by flotation on 35% sucrose as described and stored at Ϫ80°C (20). DNA and RNA were isolated following the Trireagent protocol (21). mRNA was isolated by two rounds of purification on oligo(dT)-cellulose columns (5Ј-3Ј). The mRNA was ethanol precipitated, resuspended in RNase-free water, and stored at Ϫ80°C.
Southern and Northern Analysis-Restriction endonuclease-digested C. elegans genomic DNA was fractionated on agarose gels. The genomic fragments were transferred to nitrocellulose filters and hybridized with 32 P-labeled probes prepared using the Random Priming Kit (Boehringer Mannheim). High stringency hybridizations were performed at 42°C in 50% formamide. Filters were washed to 0.1 ϫ SSC, 0.1% SDS at 65°C prior to exposure to Kodak x-ray film. Northern analysis of C. elegans mRNA was performed according to the procedure of Boedtker (22). Low stringency Northern hybridizations with [␥-32 P]ATP-labeled ANTI 1.1 oligonucleotide were performed at 35°C in 6 ϫ NET (1 ϫ NET: 150 mM NaCl, 1 mM EDTA, 15 mM Tris-HCl, pH 7.5, 1 ϫ Denhardt's solution, 2% dextran sulfate, 0.02% sodium pyrophosphate, 0.1% SDS) with tRNA (40 g/ml; Sigma) added as carrier. The filters were washed at 50°C in 6 ϫ SSC, 0.5% Triton X-100 with several buffer changes and exposed at Ϫ80°C on Kodak x-ray film. Several washes at higher stringencies (up to 62°C) were performed to improve signal to noise ratios.
RNase H Digestions-10 g of either wild type or GluCl␣::Tc1 mRNA was mixed with 200 ng of an antisense oligonucleotide (ANTI 1. Reverse Transcription and PCR-Reverse transcription reactions were performed using 200 ng of C. elegans mRNA. cDNA was synthesized with random primers using the reagents of the Superscript kit (Life Technologies, Inc.). PCR reactions were performed using the Superscript kit (Life Technologies, Inc.) adding 20 Ci of [␣-32 P]dCTP and the following primers: (base 565 of the GluCl␣ cDNA) 5Ј-GAATACA-CAATGATGGTACAG-3Ј and (base 680 of the GluCl␣ cDNA) 5Ј-TG-CAAGATCAATGGAACACTG-3Ј. PCR conditions were 95°C 1 min, 55°C 2 min, 72°C 3 min for 35 cycles. The reactions were phenol extracted, ethanol precipitated, resuspended in 2 l of H 2 O, mixed with stop solution of the Sequenase kit (U. S. Biochemical Corp.), heated at 65°C for 4 min, and electrophoresed in a 7% sequencing gel. The gel was fixed in 5% methanol, 5% acetic acid for 15 min, dried, and exposed to x-ray film.
Isolation of GluCl␣2 cDNA-Reverse transcription reactions were performed using 200 ng of C. elegans mRNA. First strand cDNA was synthesized with random or oligo(dT) primers using the reagents of the Superscript kit (Life Technologies, Inc.). PCR reactions were performed using the Superscript kit (Life Technologies, Inc.) and the following primers: 5Ј-TTCAAACGTCATTTCATCCAATG-3Ј and 5Ј-TCTCCATT-GAGGGCACCGGTATG-3Ј. PCR conditions were 95°C 1 min, 62°C 2 min, 72°C 3 min for 45 cycles. The GluCl␣2 cDNA PCR product was subcloned to the pGEM-T easy vector (Promega). Subsequently, the SacII/PstI fragment containing the GluCl␣2 cDNA was transferred to Bluescript SK vector. A poly(A) tail was added by annealing and ligation of two oligonucleotides containing 25 As to the NotI/PstI sites.
Electrophysiology-Xenopus oocytes from the same donor were used in every set of experiments. Oocytes were injected with 50 nl of mRNA (1 g/l) or cRNA and voltage-clamped at Ϫ80 mV at room temperature using a standard two-microelectrode voltage clamp as described (13,16). Oocytes injected with mRNA were analyzed 2 days after injection while those injected with cRNA 2-4 days after injection. The watersoluble avermectin derivative, IVMPO 4 , was used for these studies. GluCl␣2 cRNA was synthesized using the T3 promoter of the Bluescript vector as described (16). The reported current amplitudes are average responses of at least four oocytes. All data in text are mean Ϯ S.E.
Ivermectin Binding Assays and Pharmacology-Four different membrane preparations of wild-type and GluCl␣::Tc1 animals were tested for [ 3 H]IVM binding as described (23). Wild-type and GluCl␣::Tc1 animals were grown in liquid media, membranes were prepared and binding assays were done in parallel for each set (20). Motility assays in solution and plates as well as egg laying assays were done as described (9,24). Approximately 50 -100 animals of mixed developmental stages or single L4 hermaphrodites were used for each ivermectin concentration (0.1-500 nM).

Chromosomal Mapping of GluCl␣ and GluCl␤ Genes-The
GluCl␣ and GluCl␤ genes were assigned to yeast artificial chromosome clones and corresponding cosmids (Fig. 1). The GluCl␣ gene was found to be within cosmid clone C25D4 of contig 313 on chromosome V, while the GluCl␤ gene was present within cosmid clone C04E4 of contig 465 on chromosome I. These data indicate that the GluCl␣ gene is located between unc-76 and dpy-21 while the GluCl␤ gene is distal to unc-63 and next to unc-38. The unc-38 mutation has been assigned to cosmid clone ZZ#11 that overlaps with cosmid C04E4. The nucleotide sequence of unc-38 shows it to be related to the nicotinic acetylcholine receptors and therefore does not represent a mutation in the GluCl␤ gene (25). No other mutations have been assigned to the cosmids containing the GluCl␣ and GluCl␤ genes.
Isolation of the GluCl␣::Tc1 Strain-To study the functional role of the GluCl channels in C. elegans we sought to develop null mutants in the GluCl␣ and GluCl␤ genes using transposon insertional mutagenesis (19). The C. elegans strain, MT3126, which shows a high frequency of Tc1 transposable element insertions, was used to screen for mutations in the GluCl␣ and GluCl␤ genes. Tc1 insertions in the GluCl␣ gene (strain NL704, allele pk54::Tc1) and GluCl␤ gene (strain NL705, allele pk55::Tc1) were identified by screening a frozen transposon mutant bank arranged in an orthogonal matrix (19). A Tc1 insertion in the GluCl␤ gene that was identified did not result in a null mutation (data not shown). Our subsequent attempts to obtain a null mutant in the GluCl␤ gene by searching for a Tc1 excision/deletion event by PCR were unsuccessful (data not shown). A GluCl␣ gene Tc1 insertion event identified did result in gene inactivation. A homozygous GluCl␣::Tc1 strain with a Tc1 insertion in the GluCl␣ gene was isolated and the majority of the irrelevant Tc1 insertions were eliminated by backcrosses.
Sequencing of the Tc1 insertion region of the GluCl␣::Tc1 gene revealed that the Tc1 insertion (amino acid 255, Figs. 2 and 7) disrupted the putative extracellular N-terminal domain of the GluCl␣ protein. The Tc1 insertion interrupts a highly conserved, length invariant disulfide loop that is found in all ligand-gated anion channels (Fig. 2). Thus, it is unlikely that the GluCl␣::Tc1 gene encodes a functional channel subunit.
Analysis of GluCl␣::Tc1 Transcripts-To confirm that the Tc1 insertion resulted in a mutated GluCl␣ gene we compared the GluCl␣ gene products of wild-type N2 and GluCl␣::Tc1 animals (Fig. 3, lanes 1 and 2, respectively) by Northern blot analysis. A GluCl␣ cDNA probe detected 1.6-and 2.3-kilobase (kb) transcripts in the N 2 lane and 0.8-, 1.6-, 2.3-, and 3.2-kb transcripts in the GluCl␣::Tc1 lane (Fig. 3). The 0.8-kb mRNA that is specific for GluCl␣::Tc1 hybridized to a probe corresponding to the region 5Ј of the Tc1 insertion site but not to a probe derived from the region 3Ј of the Tc1 insertion (data not shown). These results indicate that the 0.8-kb mRNA (denoted with an asterisk in Fig. 3) is an N-terminal truncated transcript of the GluCl␣ gene. The 1.6-and 2.3-kb mRNAs that appear in the wild-type lane were reduced in intensity in the GluCl␣::Tc1 lane by 15-20 fold, as determined by PhosphorImager intensity quantitation. Expression of GluCl␤ mRNA in the GluCl␣::Tc1 strain was unaffected, showing that disruption of the GluCl␣ gene does not affect GluCl␤ transcription levels (data not shown).
The GluCl␣::Tc1 strain is homozygous for the Tc1 insertion and appears genetically stable. However, low frequency somatic excision of Tc1 transposable elements has been reported by several laboratories (26,27). Thus, it is likely that the faint transcripts at 1.6-and 2.3-kb that appear in the GluCl␣::Tc1 strain are the result of somatic excision of Tc1. This was confirmed by analysis of the Tc1 integration/excision site by reverse transcription of mRNA followed by PCR amplification of the resultant cDNA. Reverse transcriptase-PCR reactions of GluCl␣::Tc1 strain mRNA with primers flanking the integration site revealed multiple reverse transcriptase-dependent bands that were not present in the N 2 mRNA (see "Experimental Procedures" for details) (data not shown). These bands were up to 50 nt larger or up to 20 nt smaller than the wild-type GluCl␣ PCR-derived control fragment. Bands corresponding to the wild-type GluCl␣ PCR-derived control fragment were not detected in the GluCl␣::Tc1 strain mRNA. These results indicate that the 1.6-and 2.3-kb GluCl␣ transcripts arise from imperfect somatic Tc1 excision and are mutated at the site of excision. Due to small deletions or amino acid additions at this highly conserved region the majority of these (near) full-length GluCl␣ mRNAs are likely to be defective (28). However, it is possible that a small percentage of these transcripts is functional.
In summary, as a result of the Tc1 insertion a truncated GluCl␣ mRNA encoding a nonfunctional protein is generated. A low level of near full-length GluCl␣ mRNA may result in some somatic cells by Tc1 excision that could restore the reading frame in one-third of these rare mRNAs. We therefore conclude that the GluCl␣ gene is functionally inactivated in most cells of the GluCl␣::Tc1 animals and the amount of functional GluCl␣ mRNA in the GluCl␣::Tc1 strain is reduced at least 45-60-fold.
Phenotypic Analysis and Pharmacology-The GluCl␣::Tc1 strain was inspected for phenotypic abnormalities and found to lack any visible defects. Since the GluCl␣ channel is thought to be a target of avermectin, we tested the GluCl␣::Tc1 strain for sensitivity to avermectin and found it to be equal to the wildtype strain. Ivermectin sensitivity was tested in motility assays in liquid (9) plates and egg laying (24) assays (data not shown). This result may indicate that C. elegans expresses additional ivermectin-sensitive GluCl channel(s) that can functionally compensate for the elimination of GluCl␣.
Expression of GluCl␣::Tc1 Strain mRNA-Wild-type and GluCl␣::Tc1 animals were tested for the presence of ivermectin-sensitive GluCl channels by mRNA expression in Xenopus oocytes. The average response to 1 mM glutamate was 266 Ϯ 20 nA for oocytes injected with wild-type mRNA and 84 Ϯ 7 nA for those injected with GluCl␣::Tc1 strain RNA; the average IVMPO 4 response was 218 Ϯ 16 and 75 Ϯ 8 nA, respectively (Fig. 4). Perfusion of Xenopus oocytes with 100 M picrotoxin (PTX), can distinguish between glutamate elicited currents from homomeric GluCl␤ (PTX sensitive) and heteromeric GluCl␣ and ␤ (PTX insensitive) GluCl channels (13). 2 The glutamate responses in oocytes injected with either wild-type or GluCl␣::Tc1 mRNA were insensitive to 100 M PTX, indicating that the expressed channels were heteromeric (data not shown). These data indicate an approximately 3-fold difference in amplitude of the glutamate and IVMPO 4 responses between GluCl␣::Tc1 and wild-type samples. To exclude that this difference results from an overall decreased ability of the GluCl␣::Tc1-injected RNAs to be expressed in Xenopus oocytes, we measured the induced amplitude of the endogenous calcium-activated chloride channel currents in the same oocytes. 3 The amplitude of this current was comparable for oocytes injected with either wild-type or GluCl␣::Tc1 strain mRNA (812.5 Ϯ 77 and 886 Ϯ 83 nA, respectively; control oocytes showed responses of 250 Ϯ 35 nA). If the glutamate and IVMPO 4 responses of oocytes injected with wild-type mRNA were dependent on the contribution of GluCl␣ mRNA alone, an approximately 45-60-fold reduction in the current amplitudes would have been expected in oocytes injected with GluCl␣::Tc1 strain RNA.
In reconstitution experiments, synthetic GluCl␣ and GluCl␤ cRNAs were expressed in a 1:50 (␣:␤) ratio, which resulted in a 97% decrease in IVMPO 4 and glutamate responses as compared with a 1:1 (␣:␤) ratio (data not shown). The channels formed in these experiments were insensitive to 100 M PTX, 2 A. Etter, personal communication. 3  indicating that the expressed channels were heteromeric. Expression of GluCl␣::Tc1 mRNA results only in a 3-fold reduction in its response to IVMPO 4 and glutamate. Since the reduction in the current amplitudes of oocytes injected with GluCl␣::Tc1 strain RNA is only 3-fold, these experiments predict the presence of additional GluCl␣-like subunit(s) in C. elegans.
Specific Elimination of GluCl␣ Transcripts from Wild-type mRNA-Specific elimination of GluCl␣ transcripts by RNase H digestion was employed to confirm that other C. elegans genes exist that can encode avermectin-sensitive GluCl channels. mRNA from wild-type and GluCl␣::Tc1 animals were hybridized with an antisense oligonucleotide (ANTI 1.1) to the GluCl␣ coding sequence, spanning the translation initiation site, and then treated with RNase H. Specific digestion of the DNA-RNA hybrid of the GluCl␣ mRNA was confirmed by Northern analysis (Fig. 5A) (29). Labeled ANTI 1.1 oligonucleotide detected intact GluCl␣ transcripts only in the untreated samples (lane N2, band labeled with arrow in Fig. 5A). RNase H treatment, as expected, eliminated the hybridization to ANTI 1.1 oligonucleotide. The cross-hybridizing bands detected with this probe are unaffected and are of identical size in the RNase H-treated and untreated samples, indicating specific elimination of the GluCl␣ transcripts. The RNase H-treated GluCl␣ mRNA in the wild-type and the GluCl␣::Tc1 strain were reduced in size by 60 nt as determined by hybridization of the same membrane with a GluCl␣ cDNA probe (Fig. 5B). The remainder of the RNase H-treated RNA samples were expressed in Xenopus oocytes (Fig. 5C). The oocytes expressing either wild-type or GluCl␣::Tc1 strain mRNA treated with RNase H responded to both glutamate (1 mM) and IVMPO 4 (1 M): 18 Ϯ 3 nA glutamate, 24 Ϯ 4 nA IVMPO 4 ; 14 Ϯ 4 nA glutamate, 21 Ϯ 4 nA IVMPO 4 , respectively. The channels formed were insensitive to 100 M PTX, indicating that they were heteromeric. The magnitude of the currents to IVMPO 4 and glutamate were reduced by about 3-fold when compared with the expression of untreated GluCl␣::Tc1 mRNA, as were the responses of the endogenous Ca 2ϩ -activated Cl Ϫ channels indicating a nonspecific effect. These experiments indicate that additional chloride channel subunit genes with properties similar to GluCl␣ are expressed in C. elegans.
Equilibrium Binding Analysis of GluCl␣::Tc1 Membranebinding Sites-Membrane preparations of wild-type and GluCl␣::Tc1 strain animals were tested for [H 3 ]ivermectin binding (30). The B max of the wild-type membranes was 0.55 Ϯ 0.09 pmol/mg of protein (n ϭ 4) while for the GluCl␣::Tc1 membranes this level dropped ϳ40% to 0.32 Ϯ 0.02 pmol/mg of protein (n ϭ 4) (Fig. 6). These data indicate that high affinity ivermectin-binding sites exist in C. elegans membranes in the absence of functional GluCl␣ gene expression. Scatchard plot analysis of the wild-type membrane binding indicated a K d of 0.111 Ϯ 0.019 nM, while for the GluCl␣::Tc1 membrane binding of 0.153 Ϯ 0.018 nM (Fig. 6).
Isolation of a GluCl␣2 cDNA-The indications for the presence of additional GluCl␣-like channels in C. elegans prompted us to search the data bases for related genes. One such gene, GluCl␣2, that showed homology to GluCl␣ (from here on GluCl␣1) was identified within cosmid T10G3 of contig 313 on chromosome V. The introns of the GluCl␣2 gene appear in identical positions to the introns of the GluCl␣1 gene (17). Oligonucleotides corresponding to the 5Ј-and 3Ј-ends of the GluCl␣2 gene were used to isolate a GluCl␣2 cDNA by reverse transcriptase-PCR. The 5Јoligonucleotide included a 20-nt upstream sequence of the putative initiator Met codon, while the 3Ј-oligonucleotide was upstream of the putative poly(A) addition signal. Nucleotide sequencing of an isolated GluCl␣2 cDNA revealed no mismatches to the sequence of the predicted exons of cosmid T10G3. The GluCl␣2 cDNA revealed 80, 63, and 61% nucleotide identity with GluCl␣1, GluCl␤, and DrosGluCl, respectively (31); the corresponding predicted amino acid identities were 75, 49, and 49%. Alignment of GluCl␣2, GluCl␣1, GluCl␤, and DrosGluCl-predicted amino acid sequences is shown in Fig. 7. The putative GluCl␣2 protein has a 289-amino acid long N-terminal extracellular domain followed by three closely spaced putative transmembrane domains (M1-M3), an 83-amino acid cytoplasmic loop and a fourth transmembrane domain (M4). Two pairs of cysteine residues that are conserved in all GluCls and glycine receptors are found in the N-terminal extracellular domain of GluCl␣2 and a protein kinase C recognition sequence is conserved between GluCl␣1 and GluCl␣2.
GluCl␣2 Expression-Xenopus oocytes injected with GluCl␣2 cRNA (500 pg/oocyte) responded to both glutamate (1 mM) and IVMPO 4 (1 M) (Fig. 8A). The glutamate responses were rapid, desensitizing, and reversible while the IVMPO 4 responses were slow and irreversible. The average glutamate response was 5.7 Ϯ 0.5 A with an EC 50 of 208.3 M and a Hill coefficient of 2.1, indicating that more than 1 glutamate molecule is required to open the channel (Fig. 8A). IVMPO 4 responses of the same oocytes were 21.4 Ϯ 1.8 A with an EC 50 of 107.8 nM and a Hill coefficient of 1.6 (Fig. 8C). Coapplication Coinjection of Xenopus oocytes with GluCl␣2 and GluCl␤ (10 and 50 pg, respectively) resulted in heteromeric channel formation that responded to both glutamate (1 mM) and IVMPO 4 (1 M) (Fig. 8A). The kinetics of the glutamate response retained the kinetics attributed to GluCl␤, such as reversibility and slow desensitization, but were PTX (100 M) insensitive. The maximal glutamate response was 3356 Ϯ 341 nA, while the maximal IVMPO 4 response was 3574 Ϯ 273 nA. Doseresponse curves for ivermectin and glutamate of oocytes coexpressing GluCl␣2 and ␤ are shown in Fig. 8, B and C. The ivermectin EC 50 was calculated to be 103 nM with a Hill coefficient of 1.85 and 62 M glutamate and 2.4, respectively. The GluCl␣2 and ␤ heteromeric channels were insensitive to ␥-aminobutyric acid (1 mM), glycine (1 mM), aspartate (1 mM), and kainate (1 mM). These experiments indicate that the pharmacological properties of Xenopus oocytes injected with GluCl␣1::Tc1 mRNA could be accounted for by the formation of GluCl␣2 and ␤ heteromeric channels.
Since GluCl␣2 is homologous to GluCl␣1, we tested whether its expression is altered in the GluCl␣1::Tc1 strain. Northern analysis of mRNA from wild-type and GluCl␣1::Tc1 revealed two GluCl␣2 transcripts of approximately 1.7 and 2.3 kb (data not shown). Expression of the two GluCl␣2 transcripts in the GluCl␣::Tc1 strain was unaffected (data not shown). DISCUSSION A comparison of wild-type and mutant organisms can be instrumental in addressing questions about gene function. Despite detailed characterization of the GluCl␣1 and GluCl␤ proteins in vitro, their in vivo function remains unknown (16,32). We tried to identify previously characterized mutations that map to the GluCl␣ and GluCl␤ genes by determining their chromosomal localization (Fig. 1). Since such mutations were not identified, a C. elegans strain with an insertion of the Tc1 transposable element in the GluCl␣1 gene was identified in a frozen transposon mutant bank. The Tc1 insertion occurred in an exon encoding part of the extracellular region of the GluCl␣1 protein and resulted in truncated GluCl␣1 transcripts.
Correlation between biological activity, binding affinity to C. elegans membranes, and in vitro potency of a series of avermectin analogs strongly suggests that GluCl␣1 and GluCl␤ represent efficacious ivermectin targets that may have a physiologically essential function (16,33,34). However, inactivation of the GluCl␣1 gene resulted in animals with no obvious phenotype and typical ivermectin sensitivity, suggesting that there could be functional complementarity of GluCl channel subunit genes with similarity to GluCl␣1. It is also possible that the developmental and spatial expression of GluCl␣1 is such that interference of ivermectin with its function does not result in a detectable phenotype, or that GluCl␣1 is expendable. The electrophysiological and biochemical experiments presented demonstrate the existence of ivermectin-sensitive, GluCl channels in addition to GluCl␣1 in C. elegans.
As predicted by the above experiments, a gene (GluCl␣2) with high degree of homology to GluCl␣1 was identified in the data bases. Cloning and expression of a cDNA encoding GluCl␣2 revealed pharmacological properties similar to GluCl␣1. However, in contrast to GluCl␣1, GluCl␣2 is gated by glutamate. Co-injection of GluCl␣2 and ␤ in Xenopus oocytes resulted in heteromeric channel formation as indicated by the decreased rate of desensitization of the glutamate response; by the PTX insensitivity of the channels formed, and by the shift of the glutamate dose-response curve to lower values (Fig. 8). Interestingly, coexpression of GluCl␣2 and ␤ reduced the EC 50 for glutamate of GluCl␤ from 380 to 62 M while coexpression of GluCl␣1 and ␤ increased it to 1.3 mM (16). This may indicate that the GluCl␣2 and ␤ subunit combination may be the naturally occurring one. Co-immunoprecipitation and colocalization experiments could prove this association. Our experiments indicate that GluCl␣2 can account, at least partially, for the ivermectin responses of the GluCl␣1::Tc1 mRNA, but cannot exclude the existence of additional avermectinsensitive GluCl channels besides GluCl␣1 and GluCl␣2 or that ivermectin acts on other targets in addition to the GluCl channels.
The target site involvement in the development of resistance to avermectins is of primary importance. Significant resistance to ivermectin has emerged in the field but target site involvement has not been documented. Four-fold resistant strains of Haemoncus contortus, a nematode parasite of sheep, tested for the presence of high affinity ivermectin-binding sites did not indicate an alteration in the K d or B max over controls (18,35).
Our data indicate that ivermectin may be acting on multiple related targets. The experiments presented in this paper predict that high level resistance to ivermectin may be rare because of the presence of several avermectin-sensitive GluCl channel subunits, each of which may need to be mutated to confer resistance.
GluCl channels are proteins with interesting pharmacological properties (36). Understanding the physiology of the GluCls can provide insight into mechanisms of anthelmintic and insecticide resistance (37). Furthermore, their study could lead to rational approaches to the discovery of novel anthelmintics and other antiparasitic and insecticidal agents (38). The GluCl␣::Tc1 strain, although it does not exhibit an obvious phenotype, could be useful in understanding the physiological function of GluCl channels in C. elegans and other invertebrates. Possibly in other genetic backgrounds a phenotype can be detected. Mutations in other GluCl channel subunits such as GluCl␣2, Cegbr, or GluCl␤ should help in elucidating the physiological role of the GluCl channels and the action of the avermectin class of compounds.