Novel Putative Nicotinic Acetylcholine Receptor Subunit Genes, D5, D6 and D7, in Drosophila melanogaster Identify a New and Highly Conserved Target of Adenosine Deaminase Acting on RNA-Mediated A-to-I Pre-mRNA Editing

Corresponding author: D. B. Sattelle, Department of Human Anatomy and Genetics, University of Oxford, S. Parks Rd., Oxford OX1 3QX, United Kingdom., david.sattelle{at}anat.ox.ac.uk (E-mail)

Communicating editor: A. J. LOPEZ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genome analysis of the fruit fly Drosophila melanogaster reveals three new ligand-gated ion channel subunits with the characteristic YXCC motif found only in -type nicotinic acetylcholine receptor subunits. The subunits are designated D5, D6, and D7. Cloning of the D5 embryonic cDNAs reveals an atypically large N terminus, part of which is without identifiable sequence motifs and is specified by two polymorphic alleles. Embryonic clones from D6 contain multiple variant transcripts arising from alternative splicing as well as A-to-I pre-mRNA editing. Alternative splicing in D6 involves exons encoding nAChR functional domains. The D6 transcript is a target of the Drosophila adenosine deaminase acting on RNA (dADAR). This is the first case for any organism where a nAChR gene is the target of mRNA editing. Seven adenosines could be modified in the extracellular ligand-binding region of D6, four of which are also edited in the D6 ortholog in the tobacco budworm Heliothis virescens. The conservation of an editing site between the insect orders Diptera and Lepidoptera makes nAChR editing the most evolutionarily conserved invertebrate RNA editing site so far described. These findings add to our understanding of nAChR subunit diversity, which is increased and regulated by mechanisms acting at the genomic and mRNA levels.

NICOTINIC acetylcholine receptors (nAChR) mediate fast excitatory transmission at cholinergic synapses (CHANGEUX and EDELSTEIN 1998 ). They are pentameric membrane proteins that incorporate an integral ion channel (UNWIN 1996 ). Typically, they become transiently permeable to cations (Na⁺, K⁺, Ca²⁺) following the binding of the natural ligand, the neurotransmitter acetylcholine (ACh; KARLIN and AKABAS 1995 ). In vertebrates (MCGEHEE and ROLE 1996 ) and in the nematode Caenorhabditis elegans (FLEMING et al. 1997 ; RICHMOND and JORGENSEN 1999 ), nAChRs are present both at the neuromuscular junction and in the nervous system. Large families of nAChR subunit genes have been identified: 17 in vertebrates (LE NOVERE and CHANGEUX 1999 ; ELGOYHEN et al. 2001 ) and 27 in C. elegans (MONGAN et al. 1998 ; SATTELLE et al. 2002 ). Each nAChR subunit typically possesses an extracellular N-terminal domain containing residues responsible for agonist binding, four transmembrane regions (TM1–4), the second of which (TM2) lines the channel, and a long, poorly conserved intracellular loop between TM3 and TM4 that possesses multiple regulatory phosphorylation sites (SWOPE et al. 1999 ). In vertebrates, nAChR subunits are classified as either or non- types (ß, , , ), the subunits being characterized by the presence in the extracellular N-terminal domain of two adjacent cysteines in loop C of the ACh-binding site. Other specific residues in loops A and B of subunits contribute to the ACh binding. Non- subunits carry complementary ligand-binding domains (loops D–F) and they assemble with subunits, typically with an ₍₂₎, non-₍₃₎ stoichiometry (although other combinations are possible) in native neuronal heteropentameric receptors (GALZI and CHANGEUX 1995 ; ARIAS 1997 ; KURYATOV et al. 2000 ). In the vertebrate central nervous system, homopentameric receptors [e.g., 7₍₅₎] have also been detected (DRISDEL and GREEN 2000 ).

In insects, cholinergic synaptic transmission is restricted to the nervous system (BREER and SATTELLE 1987 ; GUNDELFINGER and HESS 1992 ; LEE and O'DOWD 1999 ). Both and non- nAChR subunit genes have been cloned from a variety of insect species and both homooligomeric and heterooligomeric receptors containing insect nAChR subunits have been (i) postulated to exist in situ (SCHLOS et al. 1991 ) and (ii) shown to assemble as functional recombinant receptors in Xenopus laevis oocytes (MARSHALL et al. 1990 ; AMAR et al. 1994 ; SGARD et al. 1998 ). Insect nAChR subunits more closely resemble in sequence vertebrate neuronal subunits than vertebrate muscle subunits. Nevertheless, insect and vertebrate neuronal nAChRs show some distinct pharmacological properties; e.g., the insecticide imidacloprid appears to be more effective on insect nAChRs (LIU and CASIDA 1993 ; MATSUDA et al. 1998 , MATSUDA et al. 2001 ; LANSDELL and MILLAR 2000A ).

The model organism Drosophila melanogaster is well suited to genetic, molecular, and physiological approaches to nAChR function. Patch-clamp electrophysiological recordings from native Drosophila neurons (ALBERT and LINGLE 1993 ; GUNDELFINGER and SCHULZ 2000 ) point to a diversity of nAChR subtypes, as is the case in vertebrates (CORDERO-ERAUSQUIN et al. 2000 ).

At present in D. melanogaster only six nAChR subunits have been cloned. Of these, ALS, SAD, D3, and D4 are subunits (BOSSY et al. 1988 ; JONAS et al. 1990 ; SAWRUK et al. 1990A ; SCHULZ et al. 1998 ; LANSDELL and MILLAR 2000B ), whereas ARD and SBD are referred to as non- or ß subunits (HERMANS-BORGMEYER et al. 1986 ; SAWRUK et al. 1990B ).

Immunoprecipitation experiments show that ALS and SAD nAChR subunits may be part of the same receptor complex at synapses in the nervous system (SCHULZ et al. 2000 ). On the other hand, both immunoprecipitation and binding experiments suggest that D3 and the ß subunit ARD may coassemble in a receptor complex (SCHLOS et al. 1988 ; CHAMAON et al. 2000 ). However, to date no fully functional recombinant nAChR has been generated in a heterologous system using only Drosophila nAChR subunits, although hybrid recombinant receptors containing a vertebrate neuronal ß subunit have been employed to investigate, in vitro, the roles of the subunits (BERTRAND et al. 1994 ; MATSUDA et al. 1998 ; SCHULZ et al. 2000 ).

The recently sequenced genome of D. melanogaster (ADAMS et al. 2000 ) has renewed interest in molecular and functional diversity in the Drosophila nAChR gene family. This has led to the cloning of three new, 7-like, putative nAChR subunit genes (D5, D6, and D7) and has provided evidence of genetic variation, alternative splicing, and site-specific mRNA editing events. Our results describe the first example of A-to-I pre-mRNA editing in a nAChR subunit.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Animal strain and nucleic acids:
For RACE experiments, cDNA cloning and developmental expression studies, D. melanogaster Canton-S strain poly(A)⁺ RNA and genomic DNA were provided by CLONTECH (Palo Alto, CA). D. melanogaster flies for D5 polymorphism analysis were of Canton-S strain (provided by Dr. J.-M. Dura, Institut de Génétique Humaine, Montpellier, France) and single-fly genomic DNA for PCR amplification was quickly extracted by incubating ground fly for 30 min at 37° in 50 µl of 10 mM Tris pH 8, 1 mM EDTA, 25 mM NaCl, and 200 µg/ml Proteinase K. Drosophila lines for D6 editing studies were as follows: Drosophila adenosine deaminase acting on RNA (dADAR)^- animals were females doubly heterozygous for the dADAR^6C1 and dADAR^5G1 alleles, both of which remove the dADAR locus (PALLADINO et al. 2000A ). Hemizygous dADAR^5G1 males were also used. dADAR⁺ controls were females doubly heterozygous for dADAR^5G1 and the balancer chromosome FM7a (dADAR⁺) or FM7a males. RNA was extracted by a modified LiCl/Urea method (AUFFRAY and ROUGEON 1980 ).

RACE and cloning of cDNAs:
A double-strand cDNA library was prepared using the Marathon cDNA amplification kit (CLONTECH) on embryonic D. melanogaster poly(A)⁺ RNA following the manufacturer's protocol. Rapid amplification of cDNA ends (RACE) was performed using the Advantage cDNA PCR kit (CLONTECH) with 5 µl of the Marathon products in a final volume of 50 µl. Gene-specific primers (Table 1) were used in combination with the Marathon adaptor primer AP1. For each gene, forward and reverse primers were chosen to give overlapping 5' and 3' RACE amplifications to confirm that they belong to the same gene. Thermal cycler parameters for D5 and D6 genes were as follows: five cycles at 94° for 30 sec and at 72° for 4 min, five cycles at 94° for 30 sec and at 70° for 4 min, and 20 cycles at 94° for 30 sec and at 68° for 4 min. For the D7 gene, we used 30 cycles at 94° for 5 sec and at 68° for 4 min. RACE products were checked by nested PCR or Southern blotting and then cloned in PCR-Script vector (Stratagene, La Jolla, CA). Automated sequencing was performed by the Biochemistry Department Sequencing Facility (University of Oxford, UK).

PCR amplifications of full-length cDNAs were performed with the proofreading Pfu Turbo DNA polymerase (Stratagene) on 5 µl of the Marathon cDNA products in a final reaction volume of 100 µl. Specific 5' and 3' end primers were deduced from RACE product sequences (Table 1). To amplify D5 full-length cDNA, we carried out 30 cycles of denaturation at 94° for 1 min, annealing at 63° for 1 min, with extension at 72° for 4 min. For D6 cDNA, thermal cycling conditions were as follows: 30 cycles of denaturation at 94° for 1 min, annealing at 60° for 1 min, and extension at 72° for 4 min. Blunt-end PCR products were submitted to A-tailing at 70° and then cloned in pGEM-T vector. As described above, a large number of transforming colonies were screened by PCR using at least one gene-specific nested primer and positive clones were sequenced on both strands. The cDNA nucleotide differences with respect to the genomic sequence reported in databases have been confirmed with at least one independent amplification experiment to exclude the possibility of polymerase errors. Then, regions corresponding to nucleotide differences in cloned cDNAs were also amplified on genomic DNA using the proofreading polymerase, and products were sequenced to check for polymorphisms.

D6 RNA editing analysis:
RNA editing was assayed using a bulk sequencing method as described in PALLADINO et al. 2000A . This method has been reported to be accurate to as low as 5% editing. Reverse transcriptase (RT)-PCR was performed on whole adult D. melanogaster RNA using the primer DART (see Table 1) to specifically prime the reverse transcription reaction and primers Dare-1 and Dare-2 were used to amplify the D6 transcript spanning all suspected editing sites (exons 4–9). Similar reactions were performed on whole RNA from dADAR^5G1 males, which lack dADAR, the only adenosine deaminase acting on RNA in Drosophila. Reaction products were subjected to automated sequence analysis.

Analysis of RNA editing in Heliothis virescens was performed using whole adult RNA as the template for RT-PCR. The primers AHVRT-1 and AHVRT-2 were used to specifically prime the reverse transcription reactions for Heliothis 7-1 and 7-2 transcripts, respectively. PCR was performed with the primer pairs Darv-1 and Dave-2 or Darv-2 and Dare-2 for the 7-1 or 7-2 genes, respectively. Amplification products were subjected to automated sequence analysis. Primers for the D6 ortholog exon 5 in Heliothis 7-2 (Hvex5.1 and Hvex5.2) were used to amplify H. virescens genomic DNA from the same animals used for RNA isolation. The genomic PCR amplification product was subjected to direct sequencing to demonstrate that genomic products give a pure A signal at editing sites, ruling out a polymorphism.

Quantitative RT-PCR:
Quantitative RT-PCR was performed using the Roche (Indianapolis) LightCycler apparatus and the Roche Expand reverse transcriptase for reverse transcription. D. melanogaster poly(A)⁺ RNA (50 ng) from each of the three major developmental stages was incubated at 65° for 10 min in the presence of 50 pmol random hexanucleotides. Then 1x RT buffer, 10 mM DTT, 1 mM of dNTP mix, 0.5 µl of Promega (Madison, WI) rRnasin, and 50 units of reverse transcriptase were added in a 20-µl final reaction volume. Reverse transcription conditions were as follows: 10 min at 30° and 45 min at 42°. A 2-µl sample of each RT product was used in the LightCycler-DNA Master SYBR Green I PCR reaction. The anti-Taq DNA polymerase antibody (CLONTECH) was used for hot-start reactions.

Gene-specific primers for PCR were designed in the low-homology intracellular TM3-TM4 loop of the nAChR subunits, to amplify a 300-bp segment (Table 1). Specificity of PCR amplifications was checked by Southern blot. The Drosophila RpL17A housekeeping gene (NOSELLI and VINCENT 1992 ) was also amplified as a control for the amount of RNA used. A standard amplification curve for each gene was constructed using three or four serial dilutions of plasmids containing these genes and covering the expected range of their expression level. Thus, we used the following quantities to establish the reference curve: 3 x 10^-3, 3 x 10^-4, 3 x 10^-5, and 3 x 10^-6 ng for D5 and D6 genes and 3, 3 x 10^-1, 3 x 10^-2, and 3 x 10^-3 ng for the Rpl17A gene. The second derivative maximum method was used to estimate DNA concentration. Concentration values for each stage, expressed as ratio with the RpL17A value, were calculated with regard to the embryonic concentration (that was taken as 100%).

Computer analysis:
Sequence similarity searches were performed using the BLAST method (ALTSCHUL et al. 1990 ). The Genetics Computer Group (GCG) package was used for general DNA sequence analysis (WOMBLE 2000 ). We used several proteomic tools for protein analysis. Multiple protein sequence alignment was performed with the GeneDoc application (http://www.psc.edu/biomed/genedoc). The neighbor-joining method and bootstrap resampling, available in the ClustalX program (THOMPSON et al. 1997 ), were deployed in a phylogenetic study of nAChR subunits. The phylogenetic tree was then drawn using the TreeView application (PAGE 1996 ). Analysis of RNA secondary structure was conducted using the M. Zucher mfold program (mfold server at http://bioinfo.math.rpi.edu/~mfold/rna/form1.cgi), setting a folding temperature of 25°.

Nomenclature:
During the preliminary stage of this work, four unknown nAChR subunit candidate genes had been annotated in the Genome Annotation Database (GadFly, http://flybase.bio.indiana.edu/annot/; annotation of March 2000): CG4498, CG4128, CG8109, and CG11822. The first three corresponded to the putative novel nAChR subunits we studied. The formal nomenclature for genes encoding Drosophila nAChR subunits includes information derived from the gene's chromosomal location but we formally named them D5, D6, and D7, respectively, for convenience (Table 2).

View this table: In this window In a new window   Table 2. Novel 7-like D. melanogaster nAChR subunit genes

nAChR subunit sequences:
GenBank accession numbers of the three new nAChR subunits genes are as follows: D5 full-length cDNAs are AF272778 (from allele A) and AY036613 (from allele B); D6 cDNA variants from type I to V are from AF321445, AF321446, AF321447, AF321448, AF321449; D7 full-length cDNA (a misprocessed messenger) is AY036614. GenBank accession numbers and sequences of all the other nAChR subunits used in this work are available at the Ligand-Gated Ion Channel database (LE NOVERE and CHANGEUX 1999 , http://www.pasteur.fr/recherche/banques/LGIC/LGIC.html).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of three new Drosophila genes with structural properties of nAChR subunits:
From a BLAST analysis against the D. melanogaster genome sequence at the Berkeley Drosophila Genome Project (BDGP, http://www.fruitfly.org), using previously cloned Drosophila nAChR subunit genes, high-score alignments were obtained for three different Drosophila genomic clones (see Table 2 for their accession numbers and the corresponding gene names). We recovered the three genomic sequences and reconstructed partial putative open reading frames (ORFs) using the BLAST alignments. Deduced partial ORFs revealed recognizable regions characteristic of a ligand-gated ion channel subunit: the "dicysteine loop" and the conserved ligand-binding loops (A–F), as well as the putative membrane-spanning (TM1–4) segments (Fig 1). The presence of a YXCC motif in loop C of the binding site confirmed that they represent nAChR subunits of the type (KAO et al. 1984 ). As noted by LITTLETON and GANETZKY 2000 , these subunits are closest to the vertebrate and C. elegans neuronal 7-like subunits (Fig 2). All three new genes are more similar to one another than to any previously described D. melanogaster nAChR subunit, displaying between 53 and 63% identity and between 63 and 71% similarity at the amino acid level to each other (Table 3). Moreover, they present the highest identity and closest phylogenetic relation with the H. virescens 7-like nAChR subunits, 7-1 and 7-2, identifying them as the budworm orthologs of D5 and D6, respectively.

View larger version (32K): In this window In a new window Download PPT slide   Figure 1. Genomic organization and structure of partial and full-length cDNAs of the new 7-like Drosophila nAChR subunit genes: (A) D5, (B) D6, and (C) D7. The genomic organization was deduced by BLAST alignment and RACE experiments (see text). Half arrows indicate the positions of primers used in RACE experiments. Exons are drawn to scale whereas the nucleotide intron lengths are just shown. Colored dots show ligand-binding domains of the extracellular N-terminal moiety. Alternative coding exons 3 and 8 of the D6 gene are red and green, respectively. In cDNAs, the gray regions correspond to the 5' and 3' untranslated regions. Plain lines indicate constitutive splicing whereas alternative splicing leading to different cDNA variants is shown by dotted lines. Lengths of cloned cDNAs and of the corresponding encoded polypeptides are also indicated. Both skipping of exon 5 (in D5 cDNA) and unsplicing of intron 5 (in D7 RACE clones) lead to truncated proteins (see text). , Coding exons; , transmembrane domains.

View larger version (25K): In this window In a new window Download PPT slide   Figure 2. Drosophila nAChR subunits phylogenetic relationships. The neighbor-joining protein distance analysis method was used on a ClustalX alignment of the nAChR subunit peptide sequences containing only the extracellular N-terminal ligand-binding region and the four transmembrane segments. Bootstrap values are indicated as percentage at each node when >50%. Subunit abbreviations are as follows: Hs, Homo sapiens; Ce, Caenorhabditis elegans; Gg, Gallus gallus; Hv, Heliothis virescens. The Drosophila RDL GABA-A receptor was chosen as the outgroup.

In the D6 genomic sequence, two alternative versions of coding exon 3 (3a and 3b) and three repeats of coding exon 8 (8a, 8b, and 8c) were found (Fig 1 and Table 4). Exon 3 specifies 15 amino acids around loop D of the ligand-binding region and exons 3a and 3b differ in only four residues. Exon 8 encodes 29 amino acids that include the channel-lining region, TM2, as well as the extracellular region linking transmembrane regions TM2 and TM3. Exons 8a and 8b differ in six residues, which are located in the TM2-TM3 linker region, whereas 11 amino acid differences are encoded by 8c, some of which are part of TM2, and the others define a TM2-TM3 extracellular linker sequence completely different from that of exon 8a. This suggests the potential existence of six alternative transcripts encoding polypeptide variants of D6 with potentially different physiological/pharmacological properties.

View this table: In this window In a new window   Table 4. D6 alternative exons

Determination of 5' and 3' ends and genomic organization:
BLAST alignments did not allow the precise identification of the N-terminal exons and of the stop codon. Thus, 5' and 3' RACE experiments were performed in an attempt to clone full-length cDNAs for D5, D6, and D7. The deduced genomic structures and open reading frames are shown in Fig 1.

Using 5' RACE experiments with two D5 gene-specific primers we obtained and sequenced 12 different partial D5 cDNA clones. They identified the same putative start codon and an N terminus containing two large coding exons, each almost 450 bp long, that precede the ligand-binding region and that share no similarity with any other known nAChR subunit domain. The RACE clones extended for a variable distance upstream of the ATG up to 290 bp. A 3' RACE experiment resulted in a partial cDNA identifying a stop codon just downstream of the transmembrane segment TM4, a 181-bp 3' untranslated region (UTR) portion and a polyadenylated tail. Thus, the D5 genomic sequence spans a distance of nearly 38 kb and contains 12 coding exons (Fig 1A).

A 5' RACE experiment for the D6 gene yielded three partial cDNAs, defining an identical putative start codon and the first coding exon, but extending for a different distance up to 380 bp upstream of the ATG. On the genomic D6 sequence, the first identified coding exon is separated from the second by a large intron of 43.7 kb that does not seem to contain in its sequence any other predicted gene. For D6, 3' RACE experiments were unsuccessful, so a putative stop codon and 139-bp 3' UTR were deduced by sequencing a D6 expressed sequence tag (EST) clone (GH15518) provided by the BDGP resources. Its sequence carried the alternatively spliced exons 3b and 8a and confirmed the N-terminal exons we found in 5' RACE experiments. The complete D6 genomic sequence covers a distance of almost 68 kb and contains 12 coding exons (Fig 1B).

For the D7 gene, we obtained two 5' RACE clones covering an untranslated upstream region from putative ATG up to 305 bp. Nevertheless, these partial cDNAs carried in their sequence the small (60 bp) unspliced intron 5, probably representing unprocessed messengers since its 5' and 3' splicing consensuses are correct. In these clones, lack of splicing introduces a loss of the reading frame and a premature stop codon. A 3' RACE clone identified the stop codon and a 341-bp 3' UTR preceding a polyadenylated tail. The D7 genomic sequence spans only 14 kb (Fig 1C). Surprisingly, the 3' RACE cDNA did not include the putative, BLAST-identified exons 7 and 9 that specify sequences for the ligand-binding loop C and TM3 domains, respectively.

Cloning D5 and D6 full-length cDNAs from a D. melanogaster embryo:
For the D5 gene, a 2447-bp full-length cDNA, encoding a putative polypeptide of 807 amino acids, was cloned (Fig 1A). In its first two large N-terminal exons we found regions differing in the nucleotide sequence from the AE003642 genomic sequence. Nucleotide substitutions as well as an insertion/deletion of a short stretch of nucleotides were detected (Table 5). The resulting changes introduce amino acid transitions in the primary sequence and include modification of putative N-linked glycosylation sites. Amplification by PCR on genomic DNA, using a proofreading polymerase, cloning, and sequence analysis, confirmed the existence of polymorphic nucleotide sites in this region. We arbitrarily assigned the name allele A to the allele encoding this full-length D5 cDNA. Almost one-half of the 5' RACE clones mentioned above show the allele A sequence. The same PCR experiment allowed the cloning of a 2336-bp D5 transcript that did not show any nucleotide differences in the N-terminal exons with respect to the BDGP genomic sequence. This cDNA could arise from the allele B of the D5 gene (Table 5). The existence of allele A was confirmed by also genotyping single flies from a laboratory Drosophila population of Canton-S strain. PCR primers with the 3' end complementary to either of the nucleotides in the insertion/deletion region in exon 2 (AS1 and BS1 for alleles A and B, respectively; Table 1) have been used in combination with a common backward primer in the downstream intron, to selectively amplify the polymorphic alleles (data not shown).

View this table: In this window In a new window   Table 5. D5 polymorphic alleles

The shorter cDNA from D5 allele B was lacking the fifth exon and the incorrect splicing results in loss of the open reading frame with a premature stop codon being created in exon 6. This cDNA is predicted to encode a truncated 472-amino-acid polypeptide, consisting of only a soluble partial ligand-binding domain.

Two independent D6 full-length cDNA amplification experiments yielded, as expected, several transcript variants, as exons 3 and 8 could occur in alternative versions (Fig 1B). However, cDNA containing exon 8c was found to be transcribed neither in the 25 full-length cDNAs we cloned nor in the 5' RACE clones. We classified the D6 transcripts into five main variant categories, depending on the alternatively spliced exons 3 and 8 versions: type I (3a-8b), type II (3b-8a), type III (3b-8b), type IV (3ab-8a), and type V (3b-8ab). Types IV and V are unusual because they contain, without loss of the open reading frame, a duplicate of either exon 3 or exon 8, respectively, thereby including certain motifs twice in the encoded polypeptide subunit, notably the ligand-binding loop D region (exon 3), in the case of type IV, and the channel-lining transmembrane segment TM2 (exon 8) for type V. Because such structural organization for an ion channel subunit was never described before, the novel receptor properties introduced by exon repeating and their significance remain to be explored.

On the other hand, we were unable to obtain a full-length D7 cDNA clone, and considering the exon skipping features of the RACE clones, we suspect D7 could not give a correctly processed nAChR subunit but, as in the case of the short version of D5 from allele B, a putative truncated soluble polypeptide consisting of only part of the ligand-binding domain.

For comparison purposes, a putative D7 full-length cDNA, based on both the high similarity with D5 and the overlapping 5' and 3' RACE sequences, is nevertheless described. Such a putative cDNA would be 2338 bp long and specify a 563-amino-acid polypeptide (Fig 1C).

D6 gene RNA editing:
Interestingly, many D6 variant cDNAs showed some nucleotide differences compared to the genomic clone AE003626. Some of these consist of site-specific adenosine-to-guanosine A-to-I pre-mRNA editing-like transitions that lead to amino acid changes in the ligand-binding domain of the encoded protein (Table 6). In fact, A-to-I conversions in pre-mRNA result in the appearance of a guanosine in double-stranded cDNA (EMESON and SINGH 2001 ). Sequence analysis of 25 embryonic transcripts, derived from two independent amplification experiments, showed that three amino acids could be changed by four possible adenosine-to-guanosine transitions in exon 5 (designated as sites 3–6) and one amino acid in exon 6 (site 7), thereby defining five specific candidate-editing positions. The frequency of such editing is shown in Table 7. We also checked for the existence of genetic variation in these D6 regions. PCR amplification on D. melanogaster genomic DNA, using primers surrounding exons 5 and 6 and direct sequencing, revealed that the nucleotide in genomic DNA was adenosine at all these five positions, confirming that post-transcriptional modifications, such as editing, occur in this case.

View this table: In this window In a new window   Table 6. A-to-I D6 editing sites

View this table: In this window In a new window   Table 7. D6 edited sites frequency in embryo

Thus, addition of A-to-I RNA editing modifications to alternative splicing reveals the existence of at least nine D6 isoforms in the embryo (Table 6; type I is an unmodified sequence with respect to the genomic one). Among these isoforms, type Iiii is similar to type Iii but is missing the first three nucleotides of exon 11 (CAG_1605-7). Because noncoding regions can also be targets of editing (RUETER et al. 1999 ), skipping of this triplet could be explained if we suppose the AG 3' splicing consensus of intron 10 is subject to editing, becoming GG. In this case, the splicing machinery will recognize the next AG, in the downstream CAG_1605-7, as splicing consensus, resulting in the elimination of the amino acid Ser410 in the translated protein. Interestingly, the D6 EST (GH15518) has edited sites 4–7 but also revealed two additional adenosine-to-guanosine transitions and amino acid changes in exon 5: site 1-A₇₇₆G (Asn133 Ser) and site 2-A₇₇₈G (Ile134 Val). Thus, we have evidence for seven A-to-I editing positions in the D6 gene, five of which could be present in a small region of 19 bp around loop E of the ACh-binding site. Editing could result in eight potential nontemplated amino acid changes and, amazingly, >30,000 D6 nAChR isoforms are theoretically possible through RNA editing and alternative splicing, without considering any linkage between these events.

Moreover, in some of the 25 D6 cloned cDNAs, we found that other silent, not A-to-I editing-like nucleotide modifications could also be present at three other positions. The existence of these modifications in the embryonic transcripts was confirmed by independent RT-PCR experiments. Referring to the D6 type I cDNA numbering, nucleotide modifications are C₁₃₁₄A in exon 9 encoding the TM3 segment and C₁₅₆₀T in exon 10 and C₁₆₈₃T in exon 11, both belonging to the long intracellular TM3-TM4 loop. Direct sequencing of a PCR amplification product covering these regions on genomic DNA did not reveal mixed signal at the three positions, suggesting an absence of polymorphism in these regions.

D6 gene A-to-I pre-mRNA editing via dADAR:
It was previously shown that an adenosine deaminase acting on RNA (ADAR) is responsible for A-to-I pre-mRNA editing (reviewed by RUETER and EMESON 1998 ). Drosophila possesses one adenosine deaminase enzyme acting on RNA, dADAR. Mutants in dADAR have been generated and shown to lack A-to-I editing of known target ion channels and receptors (PALLADINO et al. 2000A , PALLADINO et al. 2000B ). We made use of dADAR-deficient animals to prove that the D6 gene underwent editing and that the editing proceeded through hydrolytic deamination of A-to-I via dADAR. D6 transcripts were amplified by RT-PCR from wild-type, whole adult RNA. The resulting product, which was a mixture of the population of expressed/modified transcripts, was subjected to direct sequence analysis. All sites, with the exception of site 3, were detected as a mixed sequence signal (Fig 3A). Sites 1 and 2 were detected at a low level that disappeared in dADAR mutants (data not shown). The ratios of A/G signal were reproducible from site to site among different RT-PCR reactions and when using different sequencing primers. Thus, we believe that the levels of editing observed using this technique are a reasonable measure of the levels of conversion at a given site. We repeated the experiment on embryonic RNA and confirmed that the frequency of editing in each site seen in sequencing cDNA products was consistent with the A/G signal levels seen by this method (Table 7). Then, we analyzed D6 transcripts from a dADAR-deficient mutant background. As illustrated in Fig 3A, all detectable mixed sequence signals were reduced to pure A signals at all sites of editing in D6. Thus, dADAR, the identified RNA editase of several other ion channels and receptors, is also the enzyme performing A-to-I pre-mRNA editing of D6 transcripts in Drosophila.

View larger version (30K): In this window In a new window Download PPT slide   Figure 3. Pre-mRNA editing sites. Sequencing traces of edited regions (sites 3–7) have been analyzed for the presence of a mixed signal. (A) Edited sites in D. melanogaster wild-type (dADAR+) and dADAR mutant (dADAR-) D6 transcript. (B) Editing sites 3–6 are conserved in H. virescens 7-2 transcript. Direct sequencing of Hv7-2 RT-PCR product is compared with PCR-amplified Hv7-2 genomic DNA. (C) Comparison of the editing levels (A/G signal) of sites 3–5 between the fly and moth orthologs.

Evolutionary conservation of nAChR editing:
The new Drosophila nAChR subunit nucleotide sequences are homologous to nAChR genes in the Lepidoptera, H. virescens, named 7-1 and 7-2. We decided to determine whether transcripts of both genes were targets for dADAR. Using RT-PCR and PCR on genomic DNA, followed by direct sequencing, we demonstrated that the 7-2 site 2 was a genomically encoded G (Fig 4). All other homologous editing site positions in Heliothis 7-2 contained A in the genomic DNA (Fig 3B and Fig 4). The RT-PCR data revealed that editing was not detectable at sites 1 and 7 (not shown). This might have been expected for site 1, which is edited at a low level in D6 transcripts. However, site 7, which is edited to a high extent in D6, is apparently not edited in Heliothis 7-2 transcripts. Surprisingly, sites 3–6, all of which are in exon 5 of D6, are edited in H. virescens to a great extent. Site 3, which is barely detectable in D6 (occurrence is <5%), is edited to a high degree in 7-2. Sites 4–6 are highly edited, with sites 5 and 6 in 7-2 being almost completely edited. This pattern of editing of the D6 homolog in Heliothis is quite different from that seen in Drosophila D6 (Fig 3C). Nevertheless, RNA editing of exon 5 of this nAChR subunit has been conserved between the orders Diptera and Lepidoptera. This is the most highly conserved RNA editing event yet reported in invertebrates.

View larger version (37K): In this window In a new window Download PPT slide   Figure 4. Alignment of the homologous D6 and H. virescens 7-2 exon 5 genomic sequences. The editing sites from positions 1–6 are in boldface type. Conserved nucleotides are highlighted in gray. Note that site 2 is constitutively G in Heliothis 7-2 sequence.

Using this same approach, we examined whether the other Heliothis 7-like gene, 7-1, exhibits editing corresponding to that in D6/Heliothis 7-2 exon 5. We found no evidence of editing in 7-1 or its homolog, the Drosophila D5 gene (data not shown).

Primary sequence analysis:
Fig 5 shows an alignment of the proteins encoded by D5 (allele A), D6 (type I), and D7 (putative full-length cDNA) genes with the C. elegans ACR-16 subunit (BALLIVET et al. 1996 ), the chicken neuronal 7 nAChR subunit (COUTURIER et al. 1990 ), the Torpedo californica muscular 1 subunit (NODA et al. 1982 ), and the two H. virescens 7-like nAChR subunits.

View larger version (103K): In this window In a new window Download PPT slide   Figure 5. Novel D. melanogaster nAChR 7-like subunit sequences. Protein alignment was performed/displayed with the Genedoc program. The conserved residues are highlighted in gray. The D5 protein sequence is from allele A. D6 is from the unedited type I variant cDNA carrying alternative exons 3a and 8b (sequences of the other alternative exons are shown on the top of the D6 sequence and dots indicate conserved residues with type I version). D6 residues, which can be changed by pre-mRNA editing, are boxed. The putative D7 protein sequence is deduced by BLAST alignment with D5 protein (italic type) and from the 5' and 3' RACE clones. Hv7-1 and Hv7-2 are the H. virescens 7-like subunits. CeACR-16 is the C. elegans ACR-16 subunit. Gg7 is the chicken 7 subunit. Tc1 is the T. californica 1 subunit. Ligand-binding consensus residues are in boldface type. Thick solid lines indicate the TM1–4 putative membrane-spanning segments. Putative peptide signals are underlined. The amino acid numbering starts after the peptide signal for Hv7-1, Hv7-2, CeACR-16, Gg7, and Tc1. Solid diamonds indicate conserved N-glycosylation sites with the chicken 7 subunit. Additional putative N-glycosylation sites in the large N terminus of D5 are doubly underlined. Putative phosphorylation sites in the TM3-TM4 loop are underlined.

The D5-encoded protein exhibits an unusually large N terminus of 304 amino acids specified by the first two coding exons. Analysis of D5 primary sequence for the presence of a signal peptide consensus and other protein domains did not show any homology with known proteins or domains, other than arginine-rich (amino acids 129–144) and serine-rich (amino acids 266–280) regions. The SignalP prediction server detected an internal signal peptide (underlined in Fig 5). We can speculate that a post-translational maturation of D5 polypeptide could occur by protease cleavage to yield a mature nAChR subunit able to insert into the plasma membrane in the correct topology. The putative protein encoded by D7 also seems to have a candidate internal signal peptide whereas the D6 deduced protein sequence shows a putative signal peptide in the N terminus of the protein (both underlined in Fig 5).

The three new D. melanogaster subunits possess the two 13 residue-spaced cysteines involved in the "cys-loop" disulphide bridge that determines the correct folded status of the ligand-binding, extracellular domain. All the T. californica nAChR subunit structural canonical residues of the main ligand-binding loops are present in the novel Drosophila subunits (ARIAS 1997 ; CORRINGER et al. 2000 ): W⁸⁶ and Y⁹³ of loop A; W¹⁴⁹ and Y¹⁵¹ of loop B; and Y¹⁹⁰, CC^192-3, and Y¹⁹⁸ of loop C are conserved. Moreover, D5, D6, and D7 show some vertebrate non- muscle nAChR subunit residues that belong to the complementary ligand-binding region loops D (W⁵⁷) and F (D¹⁸⁰; T. californica subunit numbering), which are present in the homooligomeric-forming chicken 7 subunit. The more variable loop E is also present in D5, D7 (R¹¹³; T. californica subunit numbering), and D6 (K³⁴; T. californica subunit numbering). The N-terminal extracellular region of the novel Drosophila subunits shows conservation of the three putative chicken 7 N-linked glycosylation sites (N²³, N⁶⁷, and N¹¹⁰). Additional N-linked glycosylation sites are present in the large N terminus of D5 protein and some reside in the region of the insertion/deletion polymorphism (Table 5). Multiple putative serine/threonine kinase phosphorylation sites were identified in the large cytoplasmic loop between TM3 and TM4, where phosphorylation is known to act, regulating the desensitization rate (SWOPE et al. 1999 ) and/or receptor expression and localization (WILLIAMS et al. 1998 ).

We noted the chicken 7G¹⁶⁷ homologous position in the D6 subunit is an aspartic acid residue. The mutation G D, reported as G189D in MATSUDA et al. 2000 , was previously generated in chicken 7 subunit and was shown to reduce receptor sensitivity to the neonicotinoid insecticides. We note that an aspartic acid in the homologous position is also present in the H. virescens 7-2 subunits as well as in the insect Locusta migratoria Loc3 subunit (HERMSEN et al. 1998 ).

Due to alternative splicing and RNA editing, the D6 proteins display variability in their amino acid composition (Table 6). For example, editing at site 1 in adult EST cDNA disrupts an asparagine residue (N S) that is highly conserved in the other homologous subunits. On the other hand, the I V transition, in edited site 2, could reintroduce a conserved valine residue. In the embryonic D6 type IV variant, editing at site 3 introduces the transition H R in loop E, thus adding another positively charged residue. The coding triplet corresponding to the conserved N-linked glycosylation site chicken 7N¹¹⁰ belongs to editing sites 4 and 5. In some variant transcripts this glycosylation site can be eliminated by the N S/G transition.

The alternative splicing of exon 8a or 8b versions will introduce in the D6 subunit the asparagine residue at the chicken 7L²⁵⁴ homologous position in TM2 (N is also present in the homologous position of D5 and D7 subunits). This Leucine residue has been shown to be involved in the cation selectivity of the chicken 7 homomeric receptor since the transition L254Q decreases the Ca²⁺ affinity and permeability of this receptor (BERTRAND et al. 1993 ). Due to the similar polarity of Q and N amino acids, we predict that subunits encoded by D5, D7, and D6 containing either exon 8a or 8b should have a different permeability to Ca²⁺ than the nontemplated subunits carrying exon 8c. Thus, alternative splicing of exon 8 could have important consequences on cation permeability of D6-containing nAChRs. More interestingly, this position corresponds to the so-called Q/R site in glutamate receptors, which in the GluR-B AMPA receptor is a target of A-to-I mRNA editing with the consequent Q R amino acid transition, resulting also in cation permeability modification (SOMMER et al. 1991 ). It is noteworthy that for both receptor molecules two different post-transcriptional mechanisms have been used to modify one key amino acid position.

D5 and D6 developmental expression:
A developmental study of the expression of D5 and D6 genes was conducted by quantitative RT-PCR, using staged RNA samples, with reference to the expression of the housekeeping Rpl17A gene. On the basis of the plasmid dilutions used to construct the standard curve, we deduce that D5 and D6 are expressed at very low levels during development, on the order of 1000-fold lower than the level of ribosomal protein. However, even when MgCl₂ concentration and annealing temperature were optimized for each gene PCR amplification, it was difficult to obtain comparable amplification efficiencies, thereby precluding any comparison of the expression between these genes at the same developmental stage. Consequently, the results were derived from RT-PCR replicates and D5 and D6 transcript levels were expressed as a percentage of the levels in the embryonic stage (Table 8). The findings indicate a major expression of D5 and D6 in the embryo that declines, severely in the case of D5, in larvae and adults.

View this table: In this window In a new window   Table 8. D5 and D6 expression level

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Three novel 7-like nAChR subunit genes have been identified in D. melanogaster, D5, D6, and D7. All three genes are transcribed and quantitative RT-PCR indicates that D5 and D6 are developmentally regulated (the D7 expression level was not studied). The high degree of identity among the D5, D6, and D7 genes and the conservation of seven intron positions (data not shown) lead to the hypothesis that they are the product of duplication events.

We obtained from embryonic RNA full-length cDNAs for D5 and D6 genes, showing the existence of several expressed variants of D6 resulting from alternative splicing of coding exon 3 and exon 8 repeats and multisite pre-mRNA editing. Moreover, we have evidence for the existence of two polymorphic alleles of the D5 gene leading to transcripts with amino acid changes in an atypically large N terminus. The cDNA derived from allele B could encode for a soluble protein whereas a classical nAChR subunit from allele A cDNA could be generated only if it is subject to post-translational maturation, which remains to be demonstrated. In fact, it has been recently shown that a soluble ACh-binding protein (AChBP) secreted by glial cells is involved in cholinergic synapse modulation in an invertebrate (SMIT et al. 2001 ). AChBP may decrease or even terminate the postsynaptic ACh response. One may imagine that D5 (allele B) and D7 truncated polypeptides could take part in analogous mechanisms in Drosophila.

Genomic variation and alternative splicing have been reported also in human nAChR subunits (VILLARROEL 1999 ; WEILAND et al. 2000 ). For some vertebrate ligand-gated and voltage-gated ion channels, it has been shown that alternative splicing or pre-mRNA editing of subunit transcripts can greatly increase structural and functional diversity (SOMMER et al. 1991 ; KOHLER et al. 1993 ; NISWENDER et al. 1999 ). In the insect D. melanogaster, both alternative splicing and mRNA editing contribute to chloride channel subunit diversity (FFRENCH-CONSTANT and ROCHELEAU 1993 ; SEMENOV and PAK 1999 ). The Drosophila nAChR subunit D4 has been reported to be subject to alternative splicing of an encoding exon, generating subunits with different assembly efficiency (LANSDELL and MILLAR 2000B ). Editing has been shown in sodium channel and calcium channel genes (SMITH et al. 1998 ; HANRAHAN et al. 2000 ). Our work provides additional evidence for an increased diversity in nAChR subunits arising by alternative splicing of exons encoding functional domains and, moreover, the first evidence for pre-mRNA editing in any nAChR subunit. In fact, A-to-I conversions occur in several specific positions in D6 embryonic transcripts that are restricted to the ligand-binding regions.

D6 editing is dADAR dependent:
In the mutant Drosophila dADAR^- that completely lacks ADAR activity, site-specific A-to-I editing of all known pre-mRNA targets in Drosophila is abolished. RT-PCR on dADAR mutant RNA for the D6 gene showed only adenosine in all the seven editing sites we identified, thus demonstrating that D6 editing is dADAR dependent and is abolished in the ADAR mutant fly. In mammals editing by ADAR has been show to occur within the context of predicted RNA secondary structure formed through interactions between exon and intron sequences (HIGUCHI et al. 1993 ). We searched for RNA secondary structure leading to base pairing between the main group of editing sites in D6 exon 5 and its downstream or upstream intron. In both cases, the edited region seems to form base pairing within exon 5 itself (data not shown). A similar result was obtained for the Fsp site in the para channel (HANRAHAN et al. 2000 ).

The presence of two additional putative sites of editing, found only in the adult D6 EST clone (sites 1 and 2), suggests that some sites could be edited in a stage-specific manner. The existence of such developmentally regulated editing has been also demonstrated at two of the four editing sites in the D. melanogaster para channel transcript, the Ssp and Sfc sites (HANRAHAN et al. 2000 ). We speculate that D6 alternative splicing of multiple exons could also be developmentally regulated, as recently found for the exon 4 region of the Drosophila Dscam pre-mRNA (CELOTTO and GRAVELEY 2001 ).

Functional significance of D6 variants:
Most of the D6 residues changed by both alternative splicing and pre-mRNA editing are localized to key functional domains like the ligand-binding loops and TM2. We know that the dADAR mutant fly, where editing is missing in a number of ligand and voltage-gated ion channels, exhibits various age-dependent behavioral deficits accompanied by neurodegeneration (PALLADINO et al. 2000B ). The dADAR mutant has also been independently discovered in a genetic screen for mutants sensitive to hypoxia conditions (MA et al. 2001 ). Electrophysiological recordings on primary culture of dADAR mutant neurons show that the para channel conductivity is altered in the mutant fly especially in oxygen deprivation conditions (MA et al. 2001 ). This result clearly indicates that misediting of channels could result in neuronal activity defects. Thus the not-edited/edited D6 subunit-containing receptors could play a critical role in nervous system function and integrity. Because editing at some D6-specific sites occurs also in the homologous positions of the 7-2 H. virescens gene (sites 3–6), we demonstrate for the first time a high degree of evolutionary conservation of the pre-mRNA editing between two distantly related insect groups (moths and flies). This implies modifications introduced by editing are of functional relevance.

Residue changes introduced by such post-transcriptional modifications suggest novel target residues for further investigation, which could be initially studied by site-directed mutagenesis of homologous positions in the well-characterized homomeric-forming chicken 7 subunit. Furthermore, as nAChRs are known targets of insect control chemicals, such as imidacloprid, a detailed study on subunit variants in a simple model such as the fruit fly could contribute to new approaches for controlling the pest species H. virescens.

With the three novel nAChR subunits, the Drosophila nAChR family now stands at nine members. A tenth gene (CG11822) is designated as nicotinic-like in the annotated Drosophila genome. On the basis of the 5' and 3' ends of the corresponding EST clone (SD09326) we cloned and sequenced a putative full-length cDNA (GenBank accession no. AY005148) and compared its protein sequence to the other nAChR genes. We found that it is only weakly related to any other existing group, but may be of interest (Fig 2).

Finally, our findings provide an important additional example that post-transcriptional modifications play a role in expanding protein diversity in an organism with a small genome such as Drosophila and that the gene number does not correspond to animal complexity. The 10 members of the D. melanogaster nAChR subunit family identified so far and the diversity resulting from their expression may help account for the physiological and pharmacological diversity observed in native Drosophila nAChRs.

	FOOTNOTES

¹ Present address: Institut de Génétique Humaine, UPR1142-CNRS, 141 rue de la Cardonille, 34396 Montpellier cedex 5, France.
² Present address: Institut de Génétique et Microbiologie, UMR8621, Université Paris Sud, Bat. 400, 91405 Orsay cedex, France.

	ACKNOWLEDGMENTS

The authors thank Nicolas Buisine (Department of Biochemistry, University of Oxford, United Kingdom) and Arnaud Chatonnet (Institute National de la Recherche Agronomique, Montpellier, France) for their useful suggestions regarding phylogenesis and RNA secondary structure analysis. This work was supported by the Medical Research Council to M.G., E.C., and D.B.S.; a research grant from Merck & Co. to D.B.S. and M.G.; and by funding from the National Science Foundation (0091142) and National Insitutes of Health (1R01GM062291-01) to R.R.

Manuscript received June 4, 2001; Accepted for publication January 29, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAMS, M. D., S. E. CELNIKER, R. A. HOLT, C. A. EVANS, and J. D. GOCAYNE et al., 2000  The genome sequence of Drosophila melanogaster.. Science 287:2185-2195[Abstract/Free Full Text].

ALBERT, J. L. and C. J. LINGLE, 1993  Activation of nicotinic acetylcholine receptors on cultured Drosophila and other insect neurones. J. Physiol. 463:605-630[Abstract/Free Full Text].

ALTSCHUL, S. F., W. GISH, W. MILLER, E. W. MYERS, and D. J. LIPMAN, 1990  BLAST: a basic local alignment search tool. J. Mol. Biol. 215:403-410[Medline].

AMAR, M., P. THOMAS, S. WONNACOTT, and G. G. LUNT, 1994  A nicotinic acetylcholine receptor subunit from insect brain forms a non-desensitising homo-oligomeric nicotinic acetylcholine receptor when expressed in Xenopus oocytes. Neurosci. Lett. 199:107-110.

ARIAS, H. R., 1997  Topology of ligand-binding sites on the nicotinic acetylcholine receptor. Brain Res. 25:133-191.

AUFFRAY, C. and F. ROUGEON, 1980  Purification of mouse immunoglobulin heavy-chain messanger RNAs from total myeloma tumor RNA. Eur. J. Biochem. 107:303-314[Medline].

BALLIVET, M., C. ALLIOD, S. BERTRAND, and D. BERTRAND, 1996  Nicotinic acetylcholine receptors in the nematode Caenorhabditis elegans. J. Mol. Biol. 258:261-269[Medline].

BERTRAND, D., J. L. GALZI, A. DEVILLERS-THIERY, S. BERTRAND, and J.-P. CHANGEUX, 1993  Mutations at two distinct sites within the channel domain M2 alter calcium permeability of neuronal 7 nicotinic receptor. Proc. Natl. Acad. Sci. USA 90:6971-6975[Abstract/Free Full Text].

BERTRAND, D., M. BALLIVET, M. GOMEZ, S. BERTRAND, and B. PHANNAVONG et al., 1994  Physiological properties of neuronal nicotinic receptors reconstituted from the vertebrate beta 2 subunit and Drosophila alpha subunits. Eur. J. Neurosci. 6:869-875[Medline].

BOSSY, B., M. BALLIVET, and P. SPIERER, 1988  Conservation of neural nicotinic acetylcholine receptors from Drosophila to vertebrate central nervous systems. EMBO J. 7:611-618[Medline].

BREER, H. and D. B. SATTELLE, 1987  Molecular properties and functions of insect acetylcholine receptors. J. Insect Physiol. 33:771-790.

CELOTTO, A. M. and B. R. GRAVELEY, 2001  Alternative splicing of the Drosophila Dscam pre-mRNA is both temporally and spatially regulated. Genetics 159:599-608[Abstract/Free Full Text].

CHAMAON, K., R. SCHULZ, K. SMALLA, B. SEIDEL, and E. D. GUNDELFINGER, 2000  Neuronal nicotinic acetylcholine receptors of Drosophila melanogaster: the alpha subunit D3 and the ß-type subunit ARD co-assemble within the same receptor complex. FEBS Lett. 482:189-192[Medline].

CHAN