WWW.JNEUROSCI.ORG
-
The Journal of Neuroscience Advertisement
 QUICK SEARCH:   [advanced]


     
-


HOME
  |  
SEARCH  |   ARCHIVE  |   SUBSCRIBE  |   CONTACT  |   HELP

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit an eLetter
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (103)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fleming, J. T.
Right arrow Articles by Lewis, J. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fleming, J. T.
Right arrow Articles by Lewis, J. A.
Right arrowPubmed/NCBI databases
*Nucleotide*Protein
*UniGene
Medline Plus Health Information
*Genes and Gene Therapy

 Previous Article  |  Next Article 

Volume 17, Number 15, Issue of August 1, 1997 pp. 5843-5857
Copyright ©1997 Society for Neuroscience

Caenorhabditis elegans Levamisole Resistance Genes lev-1, unc-29, and unc-38 Encode Functional Nicotinic Acetylcholine Receptor Subunits

John T. Fleming1, 6, Michael D. Squire2, 3, Thomas M. Barnes1, Camilla Tornoe3, Kazuhiko Matsuda3, Joohong Ahnn4, Andrew Fire4, John E. Sulston1, Eric A. Barnard2, David B. Sattelle3, and James A. Lewis5

1 Laboratory of Molecular Biology and 2 Molecular Neurobiology Unit, Medical Research Council Centre, Cambridge CB2 2QH, United Kingdom, 3 The Babraham Institute Laboratory of Molecular Signalling, Department of Zoology, Cambridge CB2 3EJ, United Kingdom, 4 Department of Embryology, Carnegie Institute of Washington, Baltimore, Maryland 21210, 5 Division of Life Sciences, University of Texas at San Antonio, San Antonio, Texas 78249, and 6 Department of Pediatric Hematology/Oncology, Massachusetts General Hospital, Boston, Massachusetts 02114

ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
REFERENCES


ABSTRACT

We show that three of the eleven genes of the nematode Caenorhabditis elegans that mediate resistance to the nematocide levamisole and to other cholinergic agonists encode nicotinic acetylcholine receptor (nAChR) subunits. unc-38 encodes an alpha  subunit while lev-1 and unc-29 encode non-alpha subunits. The nematode nAChR subunits show conservation of many mammalian nAChR sequence features, implying an ancient evolutionary origin of nAChR proteins. Expression in Xenopus oocytes of combinations of these subunits that include the unc-38 alpha  subunit results in levamisole-induced currents that are suppressed by the nAChR antagonists mecamylamine, neosurugatoxin, and d-tubocurarine but not alpha -bungarotoxin. The mutant phenotypes reveal that unc-38 and unc-29 subunits are necessary for nAChR function, whereas the lev-1 subunit is not. An UNC-29-GFP fusion shows that UNC-29 is expressed in body and head muscles. Two dominant mutations of lev-1 result in a single amino acid substitution or addition in or near transmembrane domain 2, a region important to ion channel conductance and desensitization. The identification of viable nAChR mutants in C. elegans provides an advantageous system in which receptor expression and synaptic targeting can be manipulated and studied in vivo.

Key words: acetylcholine receptor; levamisole resistance genes; receptor mutations; Caenorhabditis elegans; evolution; nematode, unc-29; unc-38; lev-1; transmembrane domain mutation; Xenopus oocyte expression; GFP; confocal microscopy; receptor localization


INTRODUCTION

Nicotinic acetylcholine receptors (nAChRs) are the most thoroughly characterized receptors in the family of ligand-gated ion channels (Karlin, 1993; Unwin, 1993a,b, 1995). The vertebrate muscle nAChR is a pentamer of four subunit types in the stoichiometry alpha 2, beta , gamma , and delta . Each subunit contains four hydrophobic, putative transmembrane regions (TM1-TM4), with TM2 in each subunit contributing to the lining of the channel (Devillers-Thiery et al., 1993). A main part of the binding sites for ACh has been localized to each of the two alpha  subunits, close to a pair of vicinal cysteines (positions 192 and 193 in the Torpedo alpha  subunit) that define alpha subunits in all nAChRs (Kao and Karlin, 1986). Residues of the gamma  and delta  subunits also contribute to ACh binding (Karlin, 1993). The subunits of vertebrate neuronal nAChRs are encoded by a separate set of genes and fall into two major classes, either alpha  or non-alpha , each with several isoforms (alpha 2-alpha 9 and beta 2-beta 5) (Sargent, 1993; Ortells and Lunt, 1995).

Beyond cloning the subunits of nAChRs, considerable effort has been expended both in vitro and in vivo toward understanding the precise contribution of each subunit and the identities and functions of proteins that interact with the nAChR (Gautam et al., 1995; Karlin and Akabas, 1995; Gautam et al., 1996). To aid in this endeavor, Caenorhabditis elegans offers a genetic facility for defining the molecules associated with nAChR function that would be difficult to achieve in more complex animals.

Nematodes possess nAChRs (Johnson and Stretton, 1980; Fleming et al., 1993; Ajuh and Egwang, 1994; Squire et al., 1995; Treinin and Chalfie, 1995; Ballivet et al., 1996; Wiley et al., 1996). Although >20 nAChR sequences have been uncovered in the C. elegans genome sequencing project, our attention has been focused on the levamisole receptor. Levamisole is a more potent agonist than acetylcholine at nematode muscle nAChRs (Lewis et al., 1980b; Harrow and Gration, 1985; Martin et al., 1991; Robertson and Martin, 1993). Mutants resistant to levamisole define 11 genes (Brenner, 1974; Lewis et al., 1980a). These mutants fall into three classes: uncs, pseudo-wild types, and twitchers (Table 1). Mutants in six genes with the unc phenotype exhibit extreme levamisole resistance, uncoordinated motor behavior, and resistance to other cholinergic agonists. Weakly resistant mutants (pseudo-wild types) move normally but have severalfold higher resistance to levamisole and other cholinergic agonists. All genes mutable to extreme resistance (except unc-50) also produce weaker mutant alleles having a partially resistant phenotype (Table 1). lev-1 is the only locus for which the predominant mutant phenotype is that of partial resistance but for which two rare unc extreme resistance alleles, x21 and x61, have also been found. These two alleles are the only extreme levamisole resistance mutations that show any dominance.

Table 1. Major levamisole resistance loci of Caenorhabditis elegans


Levamisole resistance phenotype Gene Phenotypesa Number of alleles [3H]MAL bindingb Gene product

Extremely resistant unc lev-1 s + w  2 + 13 Variable Non-alpha -nAChR subunitc
unc-29 s + w 74 + 2  - Non-alpha -nAChR subunitc
unc-38 s + w 44 + 3 +/-  alpha -nAChR subunitc
unc-63 s + w 58 + 1 +/- Unknown
unc-74 s + w 27 + 1  - Unknown
unc-50 s  5  - Not an nAChR subunitd
Partially resistant pseudo-wild type lev-8 w  1 + Unknown
lev-9 w  3 + Unknown
lev-10 w  1 + Unknown
Twitcher unc-22 w 16 + Muscle-specifice
lev-11 w  1 + Muscle-specificf

a Phenotypes: s, strong; w, weak. Only a limited effort was made to isolate and characterize weak, partially levamisole-resistant alleles. The number of weak alleles isolated thus mainly reflects their relative ease of isolation. This table summarizes discussion of previously reported work described in the text (Lewis et al., 1980a, 1987b; work cited in other footnotes). b The [3H]MAL binding characteristic of the most levamisole-resistant alleles of each gene is indicated. +, Substantial specific [3H]MAL binding; -, little or no specific [3H]MAL binding. c This work. d M. O. Hengartner, N. Tsung, J. A. Lewis, and H. R. Horvitz, unpublished data. e Moerman et al. (1988). f Williams and Waterston (1994).

Using [3H]meta-aminolevamisole (MAL), nAChR binding has been detected in nematode extracts (Lewis et al., 1987a,b). Mutants of the six loci that give rise to strong resistance have deficient or altered receptor binding. Such receptor mutants are only moderately impaired in motor behavior as adults but are severely incapacitated at early larval stages.

Here we show that three genes associated with levamisole resistance encode nAChR subunits, and that alpha  and non-alpha subunit combinations of these genes generate functional nAChRs when co-expressed in Xenopus oocytes. The eight other levamisole resistance genes are candidates for components of nAChR-mediated synaptic signaling. The ability to exploit C. elegans genetics offers the prospect of identifying additional novel molecular components of nicotinic cholinergic synapses.


MATERIALS AND METHODS

Nematode strains. The wild-type C. elegans used was Bristol strain N2 (Brenner, 1974). The Bergerac BO transposon mutator strain (Moerman and Waterston, 1984) was obtained from the Caenorhabditis Genetics Center, and the TR679 mutator strain was kindly provided by P. Anderson (University of Wisconsin) (Collins et al., 1987)

Mutant isolation. Mutants containing restriction fragment polymorphisms were obtained in the following ways. For the isolation of spontaneous transposon-induced mutations, 30 (BO strain) or 40 (TR679 strain) adult hermaphrodites were placed on a 100 mm diameter Petri dish spread with bacteria. After 4-5 d at 20°C, progeny worms were washed off the plate and placed to one side of a separate plate containing 1 mM levamisole, as described previously (Lewis et al., 1980a). Drug plates were then screened at daily intervals for extreme resistance to levamisole. To isolate gamma -ray-induced mutants, worms were first irradiated for 6 min with 1500 rads from a 60Co source, and 20 mutagenized adults were transferred to each Petri plate. Mutants were tested for genetic identity as described previously (Lewis et al., 1980a).

Identification of restriction fragment length polymorphisms in mutants. The mutator strains used contain high copy numbers of the transposon Tc1. Genomic probing with Tc1 DNA was used to identify a novel candidate Tc1 element generating a levamisole resistance mutation. Extraneous background Tc1 elements were eliminated by balancing the mutation against a Bristol chromosome containing left- and right-flanking markers and then back-crossing the mutation 12 times into a strain homozygous for the same genetic markers. The candidate mutation was then recombined with the left- and right-flanking markers, and, if it proved necessary to achieve adequate viability, the mutation was separated by further recombination from one or both flanking markers before isolating a strain homozygous for the back-crossed mutation. Control constructs homozygous for approximately the same high-copy Tc1 chromosomal region, but otherwise having a Bristol low-copy Tc1 genetic background, were generated by back-crossing the wild-type allele present in the parent mutator strain into a Bristol strain having an ethylmethane sulfonate (EMS)-induced mutation in the gene of interest, usually flanked by the same left and right genetic markers used in constructing the mutants. For producing unc-38 mutant constructs, unc-57(e406) and dpy-5(e61) marker mutations (0.5 map units to either side of unc-38) were used. For control constructs, unc-11(e47) was used instead of unc-57. In the case of unc-29 constructs, unc-13(e450) and lin-11(n389), 1.1 and 1.6 map units to either side of unc-29, were used. For lev-1, unc-30(e191) and dpy-4(e1166), 0.3 and 4.5 map units to either side of lev-1, were used.

lev-1 mapping. Several mistakes in the chromosomal placement of lev-1 were discovered by T.M.B. The needed corrections in its map position were the placement of lev-1 close to unc-30, the inversion of the unc-26, lev-1 gene order, and the finding that the deficiency sDf23 complements lev-1(x38) and therefore does not delete the gene. The correct position of lev-1 on the genetic map was determined as follows. First, the two-factor distance between lev-1 and dpy-4 was measured. From a strain that was lev-1(x21)tra-3(e1767)dpy-4(e1166)/+++, 74 Lev non-Dpy and 61 Dpy non-Lev recombinants were found among the 3014 progeny constituting 10 complete broods, giving a distance of 4.5 map units. For deficiency mapping, the recessive allele x38 was used. x38/+ males were mated into Df/nT1 strains, and outcross males were scored. The deficiencies sDf21 and nDf27 failed to complement x38, whereas sDf22, sDf23, and sDf60 did complement x38. For three-factor mapping, Unc non-Dpy and Dpy non-Unc recombinants were picked from a strain that was dpy-20(e1282) + unc-26(e205)/+ lev-1(x21) +. Twenty of 22 Dpy recombinants and 2 of 21 Unc recombinants contained x21. In a further three-factor mapping experiment, Unc non-Dpy and Dpy non-Unc recombinants were selected from a strain that was unc-30(e191) + dpy-4(e1166)/+ lev-1 +, using mutator- or gamma -ray-generated lev-1 alleles (x505, x508, x548, and x566) isolated in this study (see above). The cumulative data were that 4 of 64 Unc and 24 of 27 Dpy recombinants contained the lev-1 marker. In summary these data give a gene order of unc-30-lev-1-unc-26. unc-26 is genetically inseparable from the Bergerac polymorphism nP33 both from the left (Yuan et al., 1993) and to the right (T. M. Barnes and J. Hodgkin, unpublished data). Thus lev-1 was inferred to lie physically between the unc-30-rescuing cosmid (Jin et al., 1994) and the nP33-detecting cosmid, a distance of about 120 kb.

Recombinant DNA techniques. Standard recombinant DNA techniques were used (Sambrook et al., 1989). The Tc1 transposable element used as a genomic probe was an EcoRV fragment prepared by D. Bird (North Carolina State University) from an isolate supplied by S. Emmons (Albert Einstein College of Medicine). The flanking genomic DNA associated with a novel Tc1-induced mutation was separated from the Tc1 DNA in any genomic subclone by excision with EcoRV and purification by gel electrophoresis. Most hybridizations and washes performed in the course of screening C. elegans genomic libraries were performed under high stringency conditions (washing at 65°C with 0.2× SSC).

Reduced stringency conditions were used for screening with an 800 bp ard probe (from position 500 to position 1350) (Hermans-Borgmeyer et al., 1986). The probe was labeled to a specific radioactivity of about 1 × 109 cpm/µg by random priming. Suitable hybridization conditions were chosen by probing a Southern blot of C. elegans genomic DNA with the ard cDNA. Hybridization was performed at 30°C in 43% (w/v) formamide and 5× SSC for 48 hr. Filters were washed for 30 min each, twice at room temperature, twice at 37°C in 2× SSC and 0.1% SDS, and then for at least 45 min at 45°C in 1× SSC and 0.1% SDS before autoradiography with intensification at -70°C. Fifty thousand phage from a C. elegans genomic library in lambda 2001 (approximately eight genome equivalents) were screened.

Phage clones were fingerprinted by A. Coulson [Medical Research Council (MRC), Cambridge, UK] (Coulson et al., 1986). DNA sequencing followed the method of McCombie et al. (1991). Cosmids spanning the unc-38 and unc-29 loci were provided by A. Coulson. Cosmid DNA was microinjected after removal of RNA by LiCl precipitation and purification by repeated ethanol precipitation. For reverse transcription (RT)-PCR, except as noted below, poly(A+) RNA (Jacobson, 1987) was prepared from total nematode RNA or was obtained as a gift from D. Zarkower (MRC); 5'- or 3'-anchored PCR was performed using standard procedures (Frohman et al., 1988).

lev-1 cDNA. A full-length cDNA of lev-1 was obtained as follows. An oligonucleotide complementary to bp 2208-2237 of the genomic sequence was used to isolate a partial cDNA clone from a mixed stage C. elegans library (provided by I. Maruyama, MRC). The cDNA contained 232 bp of sequence upstream of the predicted start of translation but lacked the predicted 3' end. The 3'-anchored PCR (Frohman et al., 1988) using an internal primer (position 3181-3200) upstream of the EcoRI site at 3306 was used to generate a fragment containing the 3' end of the message. The available partial cDNA clone contained what appeared to be an unspliced 48 bp intron, but because there was no termination codon in this intron and the reading frame was maintained, it was not clear whether this sequence was present in the mature mRNA. Therefore we analyzed the splicing pattern across this region using RT-PCR. The data (not shown) were consistent with the quantitative removal of this intron. Therefore, a partial cDNA lacking the 48 bp intron was cleaved with EcoRI (which cleaves in the 3' primer as well), and the resulting fragment was cloned into the EcoRI site of the original cDNA, thus regenerating a complete cDNA.

Cloning unc-29. The mutator-induced unc-29 mutation x513 was back-crossed into the wild-type strain and recombined with the closely linked left and right flanking markers unc-13(e450) and lin-11(n389) to reduce the background of unrelated Tc1 elements. Hybridization of a Tc1 probe to EcoRI-digested DNA from the back-crossed strain showed a novel 3.7 kb EcoRI fragment not present in the wild-type strain. The Tc1 element was not separable from the x513 mutation in 12 recombination events with either the unc-13 or the lin-11 flanking markers. The x513 Tc1 element was cloned, and the genomic DNA flanking the insert was used as a probe to identify nine wild-type genomic clones in an EMBL4 phage library (supplied by C. Link and W. Wood, University of Colorado). The physical map position of these phage was the same as that of the putative nAChR homolog JF#WA33 isolated by cross-hybridization with ard and was consistent with the genetic map location of unc-29 on chromosome I (Brenner, 1974; Lewis et al., 1980a) (Fig. 1B). Southern blots of a gamma -ray-induced mutant and five additional mutator-induced mutants all revealed chromosomal rearrangements when DNA from one of the EMBL4 phage (ZZ#1) was used as a probe (data not shown). Of the putative transposon-induced unc-29 mutants, four had insertions in either a 1.4 or 5.0 kb EcoRI fragment. These EcoRI fragments lie immediately downstream from the 2.0 kb EcoRI fragment in which both x513 and another insertion occurred (Fig. 1B). The insertions in two of these six mutants were lost in spontaneously occurring revertants to the wild-type phenotype, further indicating association of the restriction fragment length polymorphisms (RFLPs) with the unc-29 gene.


Fig. 1. Structures of the lev-1, unc-29, and unc-38 nAChR subunit genes. A, lev-1 structure. The position of lev-1 on the genetic map of chromosome IV is shown in relation to nearby genetic markers. The relative contig positions on the physical map of cosmids W07H6, C43C9, and B0564 and lambda  phage clones JF#WA10 and JF#WA18 derived from the lev-1 region are indicated. The genomic organization of the lev-1 gene is shown. Restriction enzyme sites are indicated to define possible limits of mutations: R, EcoRI; S, SalI; H, HindIII; X, XhoI; and B, BamHI. Mutant alleles found to have substantial DNA sequence rearrangements or Tc1 insertions are diagrammed. The open bar for x548 represents a deletion with a range of end points indicated. x416, x427, x438, and x566 represent complex rearrangements, mostly insertions of the sizes indicated, affecting the restriction fragments spanned by the bar shown for each mutation. For x504, x508, and x562, the positions of 1.6 kb Tc1 insertions are indicated by arrows. For e211, e289, x21, x38, x61, x400, and x505, no DNA differences were detected. GenBank entry X98601 gives the DNA sequence of lev-1. B, unc-29 structure. The position of unc-29 on the genetic map of chromosome I is shown in relation to nearby genetic markers. The relative contig positions on the physical map of cosmids C45D10, C11C3, and C34D2 and lambda  phage clones ZZ#1, ZZ#2, and JF#WA33 derived from the unc-29 region are indicated. The genomic organization of the unc-29 gene is shown. Restriction enzyme sites: R, EcoRI; S, SalI; and H, HindIII. The positions of a mutation caused by DNA rearrangement and of six mutations caused by apparent transposon insertion are shown relative to the EcoRI fragments that span the unc-29 gene. The sizes of the apparent inserts found were 1.6 kb each for x513, x522, x544, x545, and x554 and 2.5 kb for x520. The inserts of x520, x544, and x554 were lost in revertants, and the insert of x513 hybridized to Tc1. The exact nature and extent of the x415 mutation are unknown, but at least several hundred bases near the 3' end of the coding seem to be involved (see Materials and Methods). For x401, x417, x429, and x433, noDNA differences were detected. The DNA sequence of unc-29 is given by GenBank entry U81144. C, Structure of the unc-38 nAChR alpha  subunit gene. The position of unc-38 on the genetic map of chromosome I is shown in relation to nearby genetic markers. The relative contig positions on the physical map of cosmids C09C3, B0241, and C04E4 and lambda  phage clones ZZ#11 and ZZ#15 are indicated. The genomic organization of the unc-38 gene is shown. Restriction enzyme sites: P, PstI; and H, HindIII. Mutations associated with Tc1 insertion or DNA rearrangements within the HindIII fragment spanning the unc-38 gene are indicated. For x402, x404, x414, and x511, no DNA differences were found in this fragment. Other than causing a size alteration of the 3.2 kb HindIII fragment, the exact nature and extent of the x411 and x419 mutations are unknown. The DNA sequence of unc-38 is given by GenBank entry X98599.
[View Larger Version of this Image (25K GIF file)]

A 6.0 kb HindIII fragment contained within the 8.7 kb of DNA spanned by the three neighboring EcoRI fragments associated with the unc-29 mutation was sequenced (GenBank accession number U81144). The subcloned 1.4 kb EcoRI fragment (Fig. 1B) was used as a probe to isolate a full-length cDNA. The cDNA had an SL1 trans-spliced leader sequence at its 5' end, as commonly found on many nematode mRNAs (Krause and Hirsh, 1987; Zorio et al., 1994), eight nucleotides upstream of the ATG start.

Cloning unc-38. A Southern blot containing HindIII-digested genomic DNA from three back-crossed unc-38 mutator-induced mutants, five nonmutant back-crossed control strains, and the Bristol wild-type strain was probed with Tc1. All three unc-38 mutants contained a 4.8 kb HindIII band not present in the wild-type strain and all nonmutant controls. The 4.8 kb fragment from unc-38(x525) was subcloned, and the genomic DNA flanking the Tc1 site was used as a probe on Southern blots of five mutator-induced and five gamma -ray-induced unc-38 mutants. Four of the mutator-induced (including the Tc1 mutants probed above) and two of the gamma -ray mutants showed a size distinct from that seen in the wild-type strain (3.2 kb). The subcloned DNA flanking the x525 Tc1 element was used as a hybridization probe to identify corresponding lambda  phage from a wild-type EMBL4 genomic library. The physical map position of the phage was consistent with the unc-38 genetic map position on chromosome I (Brenner, 1974; Lewis et al., 1980a) (Fig. 1C). The sequence of the 3' end of the unc-38 gene, which was missing from the 3.2 kb HindIII fragment, was obtained directly from cosmid B0241 (Fig. 1C) by linear amplification sequencing (Craxton, 1991) using oligonucleotide primers derived from the cDNA sequence that was obtained.

The HindIII fragment was used to screen two C. elegans cDNA libraries (provided by S. Kim, Stanford University; and R. Barstead, Oklahoma Medical Research Foundation). Three cDNAs were isolated from 240,000 clones screened. One cDNA was sequenced and appeared to have the complete 3' end, because it contained a polyA tail 65 bases downstream of the predicted stop codon. Based on the genomic sequence, the 5' coding region was not present on any of the cDNA clones. To recover the missing 5' end, a forward primer was made corresponding to the 5' untranslated leader predicted from the genomic sequence (nucleotides 273-309, but containing an XbaI site: ATTCTC<UNL>TCTAGA</UNL>ACACTTTCTTTCAAGGCTTTTCATA). This forward primer and a reverse primer (nucleotides 854-882) corresponding to the known partial cDNA sequence were used to amplify total first strand cDNA from a mixed stage population of C. elegans. The resulting PCR product was cleaved with XbaI and BamHI (unique BamHI site at position 837) and ligated to the partial, but mature, cDNA that itself was cleaved with BamHI. The XbaI site was ligated to the vector polylinker cleaved with XbaI. The resulting hybrid PCR cDNA sequence corresponded to the predicted mature full-length cDNA sequence.

To search for null mutations of unc-38 and unc-29, cDNA clones were generated by RT-PCR using total RNA prepared according to the method of Chomczynski and Sacchi (1987) with Trizol (Life Technologies, Gaithersburg, MD) phenol-guanidine isothiocyanate reagent. The RNA (20 µg) was treated with 10 U of RQ1 RNase-free DNase and 56 U of RNasin (both from Promega, Madison, WI) in 85 µl of buffer for 20 min at 37°C. After phenol/CHCl3 extraction and ethanol precipitation, 1 µg of the resuspended RNA was used in RT-PCR according to the protocols of the Perkin-Elmer (Norwalk, CT) GeneAmp RNA PCR kit in a Perkin-Elmer Thermocycler 480, using thick-walled GeneAmp tubes. RT-PCR reactions, after a hot start, were conducted for 45 cycles of 1 min denaturation at 94°, 1 min annealing at 59°, and 2 min extension at 72°C with a final extension of 5 min. To produce cDNA products from the 5' ends of unc-38 transcripts, a forward primer derived from positions 339-364 in the genomic sequence was used with a reverse primer from positions 2184-2209 antisense. To obtain an overlapping cDNA from the 3'-end of unc-38 transcripts, forward and reverse primers from positions 1669-1691 and 3517-3541 antisense were used, respectively. The 5' and 3' RT-PCR products obtained for unc-38(x411) and unc-38(x419) were the same sizes as for wild type, 915 and 847 bp, respectively, as was the 3' PCR product of unc-38(x20). The 5' RT-PCR product from x20 RNA was only about 760 bp in length in two independent amplifications done from the same RNA preparation. After purification with the QIAquick PCR Purification System (Qiagen, Chatsworth, CA), the 5' x20 products were ligated to pT7Blue (Novagen, Madison, WI) and electroporated into DH12S. Sequencing of the two independent x20 5' clones showed that the base sequence of the third exon was precisely missing in both clones, accounting for the smaller size of the PCR products. Two independent genomic clones were generated using the Boehringer-Mannheim (Indianapolis, IN) EXPAND PCR kit with forward and reverse primers derived from positions 638-663 and 1120-1143 antisense, respectively. Each PCR amplification was done using the genomic DNA from two adult worms digested with protease K (Williams et al., 1992) using the shortened cycle times recommended (10 cycles without autoextension followed by 20 cycles with an additional 20 sec each). Unlike the cDNA products, the genomic products seemed to be of wild-type length (506 bp). In both clones the G found in the universal 3' splice acceptor AG consensus sequence was found mutated to an A, accounting for the loss of the third exon in the RNA transcript (genomic position 764). The finding of apparently normal sizes found for 5' and 3' RT-PCR products prepared from unc-38(x411) and unc-38(x419) was surprising, because these mutants showed significant RFLP differences from the wild type on Southern blots (Fig. 1C). To obtain 5' and 3' RT-PCR products of unc-29 transcripts, sets of forward and reverse primers were used from positions 494 to 521 and 2996 to 2972 antisense and from positions 2632 to 2655 and 4147 to 4170 antisense, respectively. The 5' and 3' RT-PCR products produced from unc-29(x29) and unc-29(e1072) total RNA were wild type in size (835 and 915 bp, respectively), as was the 5' product of unc-29(x415). No 3' product was obtained for x415 RNA even when a primer 214 bases downstream from the initial reverse primer position was used (positions 4361-4381). Southern blots showed that the 5.0 kb EcoRI fragment containing the 3' coding region of unc-29 was about 200 bases shorter than the wild-type fragment, consistent with the PCR results and indicating that x415 contains a rearrangement involving the 3' end of the unc-29 coding region (Fig. 1B).

Mutant rescue. Germ line transformation of unc-29 and unc-38 was accomplished using the methods developed by Fire (1986) and Mello et al. (1991).

Construction of UNC-29:: GFP fusions. Two UNC-29:: GFP fusions were made for this work, and each was found to rescue the unc-29 mutant phenotypes (both uncoordinated movement and levamisole insensitivity). Rescue indicates that transgene expression is physiologically relevant. Details of transgene construction are as follows. For LJH5, a genomic clone carrying the unc-29 coding region and 1.5 kb upstream was fused after the C terminus to the coding region for Aequora victoria gfp (Chalfie et al., 1994). For these constructs, the Ser65right-arrowThr variant of GFP (Heim et al., 1994) was used for its improved fluorescent properties. The gfp coding region in these constructs also incorporates several consensus nematode introns to facilitate nuclear export of RNA products (A. Fire, G. Seydoux, J. Ahnn, and S. Xu, unpublished observations). The UNC-29-GFP fusion junction is just at the C terminus of UNC-29, so that no UNC-29 amino acids are removed from the construct. The 3' nontranslated region is from unc-54. Construct LJH9 is similar to LJH5, except that the unc-29 upstream regions have been replaced by the body muscle-specific myo-3 promoter (Okkema et al., 1993) to generate a myo-3:: unc-29:: gfp fusion. The entire coding sequence from unc-29 is in this construct. There are six introns from unc-29 (these are introns 6-11 from the 3' end of the endogenous primary transcript), and one upstream intron is provided by the expression vector, whereas three artificial introns are present in gfp. The 3' untranslated region is from unc-54. The unc-29 introns are not necessary for rescue by the myo-3:: unc-29 constructs; a simple fusion of the myo-3 promoter to unc-29 cDNA was constructed and found to rescue the unc-29 movement defect. Injection of these constructs, derivation, and maintenance of transgenic lines were by standard protocols (Mello and Fire, 1995), using the selectable marker rol-6 (Mello et al., 1991). Strain PD9253 carries the gfp-tagged unc-29 clone LJH5, whereas strain PD9254 carries the myo-3:: unc-29:: gfp fusion construct LJH9, each as an unstable, extrachromosomal array. A control transgenic PD9258 strain carries just the rol-6 marker. Each of these transgenics was constructed by injection into a unc-29(e193) genetic background. To examine the unc-29:: gfp fusions in other genetic backgrounds, hermaphrodites of strains PD9253 and PD9254 were first crossed with N2 males to obtain the arrays in a wild-type background. The arrays were then crossed into the background of other unc-29 mutants and were found to rescue all other unc-29 mutants tested (mutant alleles x29, x513, x520, x522, x545, and e1072).

Confocal microscopy on strains carrying gfp fusions. Worms were grown at 20°C on NGM plates. Young adult worms that rolled strongly were picked to a 5% agarose pad containing 1× M9 and 10 mM sodium azide (Bargmann and Avery, 1995) and covered with a number 11/2 coverslip. Confocal microscopy was done with a Bio-Rad (Richmond, CA) MRC-1024 microscope equipped with a 60×, 1.4 numerical aperture oil immersion objective. Specimens were viewed with 488 nm excitation from a krypton-argon laser at 10% transmission. Emitted light passing through a 522DF32 filter was collected at a normal scan speed with an iris setting of 3.4, gain of 1113, and a black level of -2, using Lasersharp software. Each optical section shown represents the accumulation over a 3 µm vertical distance of images scanned every 0.5 µm.

Transient heterologous expression in Xenopus oocytes. In vitro RNA transcription was performed using a Riboprobe kit obtained from Promega. Oocytes were obtained from female Xenopus laevis from Blades Biological (Kent, UK). The oocytes were kept in standard oocyte saline (SOS) medium [containing (in mM) 100 NaCl, 5 HEPES, 1.8 CaCl2, 1 MgCl2, and 2 KCl, pH 7.6]. Under the dissecting microscope, individual oocytes were defolliculated manually (i.e., the outer thecal layer and the follicle cell layers were removed with forceps), leaving the innermost vitelline membrane intact. The oocytes were then transferred to 1 mg/ml collagenase (type 1A; Sigma, St. Louis, MO) in SOS and incubated for 20 min to ensure that any remains of the follicle layer were digested. After a short recovery period in SOS (10-30 min), the oocytes were ready for injection. Test oocytes were injected with 50 nl of RNA solution (1 ng/nl for each subunit transcript) in the vegetal hemisphere. Control oocytes were either left uninjected or were injected with sterile distilled water.

Recordings were performed using a standard two-electrode voltage clamp (Dascal et al., 1984). A single oocyte was placed on the Sylgard base in a 1 ml experimental chamber. The oocyte was impaled with two glass microelectrodes (Clarke Electromedical glass GC 150 GF-150), fabricated using an electrode puller (Scientific and Research Instruments Ltd; catalog number 2001), and filled with 1 M KCl. Electrodes with a resistance of 5 MOmega were used for the voltage electrode, and current electrodes of 1-2 MOmega were used. Oocytes were voltage-clamped using a GeneClamp 500 amplifier. Current responses were monitored on an oscilloscope (Nicolet 3091) and recorded on a Gould BS-272 pen recorder. In some experiments, responses were recorded and stored using pClamp5 software installed on an IBM Personal Computer. Medium (SOS) and drugs were applied by a perfusion pump (Pharmacia LKB pump P-1) at a rate of 4 ml/min. Levamisole, mecamylamine, d-tubocurarine, neosurugatoxin, and alpha -bungarotoxin were each dissolved in the standard SOS medium. Levamisole was bath-applied. In experiments with nAChR antagonists, oocytes were preexposed to either mecamylamine (at the concentration indicated), d-tubocurarine (1 × 10-5 M), or neosurugatoxin (5 × 10-7 M) for 5 min, and then mecamylamine, d-tubocurarine, or neosurugatoxin was co-applied with levamisole for 30 sec. For alpha -bungarotoxin (Mr ~7800), preexposure was done with a 5.0 × 10-6 M concentration for 30 min.

Construction of an evolutionary tree. An alignment of 23 receptor sequences was generated using the computer program CLUSTAL V (Higgins et al., 1992) and edited using Genetic Data Environment. The resulting data set was analyzed under maximum parsimony conditions using the program Phylogenetic Analysis Using Parsimony (PAUP) (Swofford, 1991). Briefly, a 1000 replicate heuristic search was performed using independent random number seeds and tree bisection reconnection. The resulting tree of length 2384 changes was evaluated using a 1000 replicate bootstrap analysis in PAUP (Swofford and Olsen, 1990; Felsenstein, 1992). The resulting tree was rooted using the rat GABAA alpha 1 receptor subunit. The data set contains phylogenetic data as determined using the gl statistic bracket (gl = -0.869766; p < 0.01) (Huelsenbeck and Hillis, 1993). An essentially identical tree was generated from the same starting alignment by nearest neighbor-joining analysis using the program CLUSTAL W (Thompson et al., 1994), which uses the neighbor-joining algorithm of Saitou and Nei (1987). The alignment and the PAUP data file are available from J.T.F., D.B.S., or J.A.L.

The following protein sequences, available from the SWISS-PROT protein sequence database, were used in constructing Figures 2 and 4. All except GAA1_RAT are nAChR subunits. Rat alpha  subunits: ACHA_RAT, ACH2_RAT, ACH3_RAT, ACH4_RAT, ACH5_RAT, ACH6_RAT, and ACH7_RAT (muscle alpha  and neuronal alpha 2-7). Rat neuronal beta  subunits: ACHN_RAT, ACHO_RAT, and ACHP_RAT (beta 2-beta 4). Rat muscle non-alpha : ACHB_RAT, ACHD_RAT, ACHE_RAT, ACHG_RAT (beta , delta , epsilon , and gamma ). Drosophila subunits: ACH1_DROME, ACH2_DROME, ACH3_DROME, ACH4_DROME (ALS, SAD, ARD, and SBD). Locust Schistocerca gregaria alpha L1: ACH1_SCHGR. Rat GABA alpha  subunit: GAA1_RAT (GABAA alpha 1).


Fig. 2. Amino acid sequence alignment of the Caenorhabditis elegans presumptive mature nAChR subunit sequences UNC-38, UNC-29, and LEV-1 with locust (Schistocerca gregaria) alpha L1, Drosophila (Dros.) non-alpha ARD, and rat alpha 2 and beta 2 nAChR subunit sequences. The alignment was constructed using MACAW, version 2.0.5 (Schuler et al., 1991). Regions of sequence identity or high similarity within blocks of homology are indicated by dark coloring. Regions of moderate similarity or regions at boundaries of homology blocks are indicated by light coloring. Sequences in regions with no significant similarities between subunits are given in lower case letters, and no effort was made to align the amino acids in these regions. The positions of the four transmembrane domains (TM1-TM4) and the extracellular dicysteine loop (CL) characteristic of nAChR subunits are indicated. Asterisks indicate the positions of the vicinal cysteines characteristic of alpha -acetylcholine-binding subunits. The percent identity and similarity of UNC-38 to the locust and rat alpha  sequences are 48 and 58% and 42 and 55%, respectively, as determined by individual pairwise comparisons. The percent identity and similarity of UNC-29 to the LEV-1, ARD, and rat beta  sequences are 66 and 77%, 50 and 65%, and 39 and 56%, respectively, determined by pairwise comparisons.
[View Larger Version of this Image (127K GIF file)]


Fig. 4. Maximum parsimony phylogenetic tree showing the relationship between the nAChR subunits. The tree is shown rooted using the rat GABAA receptor alpha 1 sequence. LEV-1 and UNC-29 are shown in a class of polypeptides that include mammalian muscle nAChR non-alpha subunits. UNC-38 is shown clearly related to other invertebrate alpha -like AChR subunits. The values over the branches represent the minimum number of times from 1000 random seeds in the bootstrap analysis that a particular branch is expected to appear (p < 0.01). Branches without numbers do not have significant probability of appearing at that exact point in the tree. Regions with little homology, such as the intracellular cytoplasmic loop, were not used in the comparison. A tree of the same shape was generated by nearest neighbor-joining analysis, and the bootstrap values for 1000 random seeds are shown for comparison below the branches.
[View Larger Version of this Image (31K GIF file)]


RESULTS

Deletion and transposon insertion alleles of the levamisole resistance genes

The isolation and characterization of nAChR subunit genes is an essential beginning to understanding the effects of levamisole resistance mutations on synaptic signaling. To clone the subunit genes, two different experimental approaches were adopted to maximize the chances of success. One approach was to search for C. elegans nAChR subunit homologs by cross-hybridization and placement of the clones on the physical map of the nematode genome (Coulson et al., 1986). The other approach was to generate levamisole-resistant mutants by transposon insertion followed by cloning of the insertion site (Greenwald, 1985; Moerman et al., 1986), allowing genes to be cloned regardless of their actual functions or homology to known nAChR subunit genes.

Putative transposon insertion mutants were obtained by screening the progeny of either the Bergerac BO or the TR679 mutator strains (Moerman and Waterston, 1984; Collins et al., 1987) for spontaneous resistance to 1 mM levamisole. Isolates showing strong levamisole resistance were tested by complementation against the six known extreme levamisole resistance loci (Table 1). To confirm the identity of any clone found by transposon tagging or by cross-hybridization, mutants that might represent DNA rearrangements in the same resistance genes were also isolated by a similar selection scheme using the progeny of gamma -irradiated worms. New alleles of unc-29, unc-38, unc-63, unc-74, unc-50, and lev-1 were isolated. All showed extreme levamisole resistance, except for the lev-1 isolates, which produced only partial resistance, consistent with partial resistance being the null phenotype for this gene (Lewis et al., 1980a).

Isolation and identification of lambda  phage containing nAChR subunit gene homologs

To isolate lambda  phage carrying nAChR subunit genes, a library of C. elegans genomic DNA was screened as described in Materials and Methods with an ard cDNA probe, which encodes a Drosophila non-alpha subunit (Hermans-Borgmeyer et al., 1986). Fifty-eight positive hybridizing phage were isolated. DNA from the 14 strongest positives was isolated and placed on the physical map by fingerprinting (Coulson et al., 1986). Several of these nAChR homologs mapped to positions on the physical map that were very close to the known genetic locations of levamisole resistance genes or corresponded to clones simultaneously identified by transposon tagging (unc-38 and unc-29). After revision of the genetic map position of lev-1 by T.M.B., the map position of this gene on chromosome IV was also consistent with the physical map position of two phage clones (JF#WA10 and JF#WA18; Fig. 1A and Materials and Methods).

Characterization of the lev-1 locus

The identity of lev-1 as a gene encoding a non-alpha -nAChR was established as follows. First, sequence analysis of the phage occupying the expected physical map position of lev-1 identified a 4.8 kb HindIII fragment that contained an almost intact nAChR subunit gene, organized into eight exons (GenBank accession number X98601). Second, a complete cDNA sequence recovered from this region was found to encode an open reading frame for a non-alpha -nAChR subunit of 507 amino acids clearly homologous to vertebrate and insect nAChR subunits (Fig. 2). Third, when the genomic phage clones were used to probe Southern blots of genomic DNA from 11 mutator- and gamma -ray-induced lev-1 mutants and five EMS-induced mutants, allele-specific rearrangements were found by T.M.B. that defined the lev-1 locus (Fig. 1A). Three of four gamma -ray-induced alleles (x416, x427, and x438) and two of seven mutator-induced alleles (x548 and x566) were found to have rearrangements that included at least part of the 4.8 kb HindIII fragment identified by ard hybridization or the adjacent 1.4 kb HindIII fragment (x416). For three other mutator-induced alleles, the 4.8 kb HindIII fragment was replaced by a 6.4 kb fragment, consistent with the insertion of a typical 1.6 kb Tc1 transposon element. By using two PCR primers, one an oligonucleotide corresponding to the ends of the Tc1 element together and the other an oligonucleotide corresponding to the sequence of either strand at various positions within the 4.8 kb HindIII fragment (Barnes, 1990), it was possible to locate precisely the transposon within the 4.8 kb HindIII fragment (Fig. 1A). These results show that the lev-1 gene and a nAChR homolog co-localize to the same 4.8 kb HindIII fragment. The null phenotype of lev-1 was defined by the finding that most, if not all, of the coding region of the gene is deleted by the x548 mutation and the homozygous x548 null mutants are viable, partially levamisole-resistant, pseudo-wild-type mutants.

Characterization of the unc-29 locus

unc-29 was identified as a non-alpha -nAChR subunit as follows. Genomic phage fingerprinting to the region of unc-29 on chromosome I were isolated based on the identification of a novel Tc1 insertion associated with the unc-29(x513) mutation. These phage overlapped the phage JF#WA33 identified by cross-hybridization with the Drosophila ard probe (Fig. 1B). Hybridization of the phage to Southern blots of DNA from mutator- and gamma -ray-induced unc-29 mutant identified RFLPs through a region spanned by a 6.0 kb HindIII fragment (Fig. 1B). The sequence the 6.0 kb HindIII fragment contained the entire coding region, divided into 12 exons, of a non-alpha -nAChR subunit highly homologous to lev-1 (GenBank accession number X98601). A full-length 1.7 kb cDNA that contained an open reading frame of 493 amino acids (Fig. 2) was obtained from a C. elegans mixed stage library (provided by R. Barstead).

Mutant rescue experiments confirmed the cloning of unc-29. Three cosmids that encompass the lambda  clone JF#WA33, and thus include all the regions affected by the unc-29 transpositions described above (Fig. 1B), could rescue the unc-29 mutant phenotype in transgenic animals. The cosmids were able to rescue completely the unc phenotype and to restore levamisole sensitivity, as were the smaller lambda  phage clones ZZ#1 and JF#WA33. Even the 6.0 kb HindIII fragment that spans the unc-29 coding region and contains only 490 bp of DNA upstream from the translational start site (Fig. 1B) is capable of rescuing unc-29 (S. Kim, personal communication) (this work).

Because the mutant phenotype can be rescued by a transgene sequence, it can be inferred that the transgene expression pattern includes those tissues in which the endogenous gene is required. To monitor transgene expression, we constructed a set of plasmids in which the coding region for GFP (A. victoria green fluorescent protein) (Chalfie et al., 1994) was fused to the C-terminal coding region for UNC-29. An activated form of GFP (S65T) (Heim et al., 1994) was used to maximize fluorescent signal. The resulting constructs were capable of rescuing the mutant phenotypes of unc-29 alleles e193, x29, x513, x520, x522, x545, and e1072. A major focus of unc-29 promoter activity was seen in body muscles. To test the hypothesis that UNC-29 expression in body muscles was sufficient for its function, we produced a construct in which unc-29 upstream sequences ("promoter") were removed and UNC-29/GFP expression was driven by the body muscle-specific myo-3 promoter. The myo-3:: unc-29:: gfp fusion was indeed capable of rescuing the mutant phenotypes of unc-29 mutant animals. This was consistent with the hypothesis that the primary focus of unc-29 was in body muscle. The muscle staining is consistent with previous pharmacological data from studying C. elegans (Lewis et al., 1980b) and physiological data from Ascaris that levamisole-sensitive nAChRs are present on muscle (Harrow and Gration, 1985). Recent mosaic studies using a unc-29 clone provided by us also show a major focus of unc-29 expression to be muscle (Miller et al., 1996b). We observed a low level of neuronal fluorescence, both with the unc-29:: gfp fusion and with the myo-3:: unc-29:: gfp fusion. It is not clear whether this fluorescence represents bona fide activity in these cells.

The ability of GFP fusions to rescue the mutant phenotype allowed us to examine the intracellular localization of a biologically functional unc-29 derivative in living cells. Intracellular localization in a wild-type background (i.e., in the presence of a wild-type chromosomal copy of the gene) was somewhat surprising, with activity predominantly internal to muscle cells. Because the presence of the natural UNC-29 product might be expected to affect assembly of the recombinant protein, we examined localization of the protein produced from an LJH5-derived extrachromosomal array in a variety of unc-29 mutant genetic backgrounds. Several unc-29 mutant backgrounds behaved similarly to wild type in these assays (Fig. 3A-E), with the exogenous GFP fusion present primarily in the cell body (e193, e1072, x520, x522, and x545). However, when the fusions were crossed into the homozygous mutant background provided by unc-29(x29) (and to a lesser extent, x513), relative localization was closer to that expected for a neurotransmitter receptor (head region for x29 homozygote shown in Fig. 3F-J) In addition, the overall amount of staining within muscle appeared to decrease. Internal staining of body muscles almost disappeared with punctate staining found along the nerve cords. Because of the background of neuronal staining, it was difficult to say what fraction of the staining in the nerve cords arose from body muscles. Some staining in the unc-29(x29) mutant background was still found internally in head muscles, but the relative amount of staining localized to synaptic regions of the ring (the central neuropil) greatly increased. These observations are consistent with the idea that the UNC-29 protein produced from the chromosomal copy of the unc-29 gene competes in the assembly and localization of nAChR molecules with the UNC-29-GFP fusion protein. Paradoxically, from the decrease in overall staining observed in a unc-29(x29) background, the endogenous gene product may also help stabilize the GFP fusion protein in assembly intermediates, possibly suggesting that more than one UNC-29 subunit can be assembled into an nAChR molecule. The null phenotype of unc-29 has yet to be defined, but both homozygotes of the unc-29(x29) and unc-29(e1072) mutations have been shown previously to contain little or no detectable specific high-affinity [3H]MAL binding (Lewis et al., 1987b). Although the focus of unc-38 and lev-1 expression has not been defined in this study, the great similarity in mutant phenotypes of unc-38 mutants and lev-1 semidominants to unc-29 mutants makes it likely that muscle is also the major focus for the expression of these genes as well.


Fig. 3. Serial optical sections showing expression of the unc-29 promoter-driven unc-29:: gfp fusion LJH5 in N2 wild type and in the unc-29(x29) mutant strain. Confocal microscopic sections through the head region of a wild-type and an x29 mutant animal were accumulated as described in Materials and Methods. Each picture represents a succesive 3 µm thickness of the head. A-E, Wild type. Arrows in A point to fluorescence in head muscles. Fluorescence is seen more intensely inside the same muscles in B and also within a body muscle (arrow). In E, a muscle process entering the central neuropil next to the isthmus of the pharynx is stained (arrow). Stain is also accumulated in a neuronal cell body (arrowhead). F-J, unc-29(×29) mutant. Less staining is seen overall, and the stain is relatively concentrated in head muscle processes and nerve cords with continued neuronal staining. In F, process from anterior head muscles stain (arrowhead) and punctate staining is seen in a nerve cord (arrow). In H, the arrow points to a brightly staining neuronal cell body with a process running to the central neuropil, appearing as a hazy band in the center. In I, arrows point to brightly stained areas in the central neuropil immediately adjacent to the isthmus of the pharynx that appear to be associated with muscle processes running into the neuropil at this point. The same region is stained in the center in J, with two neuronal cell bodies below the center.
[View Larger Version of this Image (28K GIF file)]

Characterization of the unc-38 locus

The unc-38 gene was identified by a novel Tc1-containing RFLP found in four of five spontaneously induced unc-38 mutants (see Fig. 1C and Materials and Methods). Two of five gamma -ray-induced unc-38 mutants also showed a restriction size difference from wild type when the subcloned genomic DNA flanking one of the Tc1 inserts was used as a hybridization probe on genomic Southern blots. DNA extending through and beyond the 3.2 kb HindIII fragment that was the site of six unc-38 mutations was sequenced and found to encode a complete alpha  subunit of an nAChR (GenBank accession X98599). A complete cDNA sequence reconstructed from this region encoded an open reading frame of 511 amino acids with strong homology to known alpha  subunit sequences of vertebrates and insects (Fig. 2). The fingerprint of genomic phage containing this alpha  subunit gene was consistent with the genetic map position of unc-38 to the left of dpy-5 on chromosome I (Fig. 1C).

The cloning of unc-38 was confirmed by mutant rescue. The injection of either cosmid B0241 or C04E4 into the germ line of unc-38 mutants completely restored normal movement in the L1 stage offspring, and the transgenic worms could now be killed by exposure to 1 mM levamisole. The B0241 and C04E4 cosmids completely encompass the lambda  phage ZZ#11 and ZZ#15 shown to contain the sites of the unc-38 mutations (Fig. 1C).

The null phenotype of unc-38 was defined to be that of an extremely levamisole-resistant unc. This finding was originally suggested by sequence analysis that showed that the Tc1 insertion site in unc-38(x525) disrupts a reading frame 15 amino acids upstream from the dicysteine loop (amino acids 128-142, Torpedo alpha  numbering) that is characteristic of all nAChR subunits. Further mutant analysis showed that the EMS-induced mutation unc-38(x20) is an absolute splicing defect in which the third exon is skipped because the universal AG intron consensus sequence at the third exon splice acceptor site is mutated to AA. The mutation causes a transcript to be produced that is 155 bases shorter than the wild type when detected by RT-PCR. Homozygous unc-38(x20) mutants grow well and are among the most resistant of all levamisole-resistant mutants to the effects of levamisole and other cholinergic agonists (Lewis et al., 1980b). Because body muscles of unc-38 mutants are more resistant than head muscles to agonists, it is likely that there is at least one other alpha -nAChR subunit participating in the formation of other nAChR isotypes, and these isotypes differ in tissue distribution from the receptor formed with the unc-38 alpha  subunit.

Sequence comparison of LEV-1, UNC-29, and UNC-38

In Figure 2, the deduced amino acid sequences of the LEV-1, UNC-29, and UNC-38 proteins are aligned and compared with vertebrate and insect nAChR subunit sequences. Comparison of the C. elegans sequences with database sequences shows that the nematode sequences are most similar to the Drosophila melanogaster nAChR subunit sequences ARD and ALS (Hermans-Borgmeyer et al., 1986; Bossy et al., 1988) and to a partially sequenced putative nAChR subunit of the parasitic nematode Onchocerca volvulus (Ajuh and Egwang, 1994). The nematode non-alpha subunits UNC-29 and LEV-1 exhibit the highest amino acid sequence identity to ARD (50 and 46%), whereas the alpha  subunit UNC-38 shows 49% identity to the ALS subunit. The predicted mature lengths of the three C. elegans nAChR subunits are 488 amino acids for UNC-38, 467 for UNC-29, and 473 for LEV-1. Whereas the alpha  subunit of vertebrate muscle is the smallest muscle nAChR subunit, UNC-38 is the longest of the known nematode nAChR subunits, a relative size difference that holds for other invertebrate and for vertebrate neuronal alpha  subunits compared with their non-alpha counterparts. The main structural features of the nAChR subunits of higher organisms are strikingly conserved in phylogeny down to the nematode, including the positions of the four transmembrane domains (TM1-4), the high sequence similarity found in each TM region, the variable long loop between TM3 and TM4, and the dicysteine loop, which is a hallmark of this superfamily and invariably found at the equivalent to positions 128-142 of the Torpedo alpha  subunit (Barnard, 1992). There seem to be additional regions in all alpha subunits of invertebrates that have no counterpart in vertebrates. These insertions occur 25 amino acids N-terminal to the vicinal cysteines (equivalent to positions 192 and 193 of the Torpedo alpha  subunit) and at the C terminus of the polypeptide. The functional significance of these domains is unknown. The insertion N-terminal to the vicinal cysteines is 13 amino acids longer for the nematode UNC-38 alpha  subunit than for locust and Drosophila alpha  subunits and is absent from invertebrate non-alpha subunits.

UNC-29 and LEV-1 are highly homologous: 66% amino acid identity or, with conservative substitutions, 77% similarity, a resemblance found for few other nAChR subunit pairs from the same species. They are categorized as non-alpha subunits by the absence of vicinal cysteines (at positions 192 and 193, Torpedo alpha  numbering) and when compared with vertebrate nAChRs show the closest homology to neuronal subunits. Both UNC-29 and LEV-1 are about equally similar to neuronal alpha  and non-alpha of the rat and chick (rat alpha 2-alpha 4 and beta 2-beta 4, chick alpha 2 and alpha 4 and beta 2 and beta 4) with UNC-29 showing slightly greater over all sequence similarity, ~55 versus ~50% for LEV-1. Consistent with these results, many aspects of the in situ pharmacology of muscle nAChRs of C. elegans and Ascaris resemble those of vertebrate neuronal nAChRs, including insensitivity to alpha -bungarotoxin, block by mecamylamine (Lewis et al., 1987a, 1980b; Fleming et al., 1993), and, in the case of Ascaris suum muscle nAChRs, block by kappa -bungarotoxin (Colquhoun et al., 1993).

Intron-exon structure comparisons and phylogenetic analysis using accepted mutation parsimony trees

Between unc-29 and the gamma , delta , and epsilon  subunits of vertebrate muscle (Nef et al., 1984; Buonanno et al., 1989), the positions of 5 of the 11 introns are completely conserved, and an additional 3 intron positions are within four amino acids of the equivalent sites in the vertebrate muscle subunits. The inexact conservation of some intron-exon boundaries may be attributable to splice junction drift or de novo creation of introns at "permissible" sites. The locations of four of the seven lev-1 introns and three of the seven unc-38 introns are identical in unc-29. However, only one intron location, between exons 5 and 6 in unc-29, is shared between unc-38 and lev-1, and this site is present in all nAChR subunit genes. The flanking exons seem to form a calcium-binding domain (Godzik and Sander, 1989), although its significance has not been well defined. A unc-29-unc-38 common splice site occurring between exons 3 and 4 of unc-29 is conserved in all nAChR subunit genes examined except lev-1. These findings suggest that the conserved splice sites predate the divergence of nematodes, insects, and vertebrates, which occurred about 600 million years ago. Unique to the nematode receptors is an intron that interrupts the coding region between TM4 and the C terminus in each of the three nematode subunits. Although the gene structures of unc-29 and lev-1 show considerable similarity, unc-38 is no more homologous to the other C. elegans subunits than any other insect or vertebrate nAChR subunit genes, suggesting that the divergence between alpha  and non-alpha subunits is very ancient. The intron-exon structure of all three nematode genes is also more typical of the highly interrupted vertebrate muscle nAChR genes than it is of the vertebrate neuronal nAChR genes (Nef et al., 1988), in which there is generally only a single intron between the conserved splice sites in the extracellular domain and at the end of the protein.

A maximum parsimony analysis for nAChR polypeptides is shown in Figure 4. The polypeptide sequences of 22 nAChRs and the rat GABAA alpha 1 receptor subunit were aligned. Using PAUP, a single tree of length 2384 changes was found (Swofford, 1991). The significance of this tree was evaluated using bootstrap analysis (Felsenstein, 1992). The resulting tree shows that nAChRs fall into several distinct classes. The C. elegans subunits LEV-1 and UNC-29 are very closely related and, of the other sequences analyzed, are most similar to the Drosophila neuronal subunit ARD. UNC-38 represents an alpha  subunit type closely related to the Drosophila alpha -like neuronal subunits. Although the distinctiveness of UNC-38 on the tree is not strongly supported by the bootstrap analysis, it is consistent with the unique pharmacological properties of nematode nAChRs as observed for C. elegans and Ascaris.

Potential glycosylation and phosphorylation signals in UNC-29, UNC-38, and LEV-1

Sequence prediction of glycosylation sites for the vertebrate muscle nAChR subunits, now confirmed biochemically (Claudio et al., 1989), indicated a single glycosylation site for the alpha and beta  subunits, two for the gamma , and three for the delta  muscle subunits. The vertebrate neuronal sequences usually contain two potential glycosylation signals. The sequences of UNC-29, UNC-38, and LEV-1 predict two, three, and five asparagine-linked glycosylation sites, respectively. There is no glycosylation site common to all three subunits, and only one shared between UNC-29 and LEV-1 at asparagine 50 (UNC-29 numbering), a site that is also present in nearly all invertebrate and vertebrate neuronal sequences (but lacking in human beta 4 and goldfish non-alpha 3). UNC-38 is the only invertebrate sequence to lack this site, and the three glycosylation signals UNC-38 contains are not in positions common to any other nAChR. An asparagine 109-linked glycosylation site in LEV-1 is conserved in alpha 2 and alpha 4 neuronal sequences and in the chick muscle delta  subunit.

The Torpedo nAChR can be phosphorylated by at least three different protein kinases: cAMP-dependent protein kinase (PKA), protein kinase C (PKC), and a tyrosine kinase (TK) (Huganir and Greengard, 1983; Huganir et al., 1984; Huganir, 1988). The region of the Torpedo receptor that was shown to be phosphorylated is the intracellular loop between TM3 and TM4. In the equivalent regions, the C. elegans UNC-29 subunit has three PKC sites, one PKA site, and one TK site; UNC-38 has single PKC and TK sites; LEV-1 has only a single PKC site. Studies with vertebrate nAChR subunits have shown phosphorylation to be involved in desensitization, receptor turnover, and receptor assembly. The in vivo functions of the glycosylation and phosphorylation sites can be investigated in C. elegans by mutagenizing a given site and then returning the altered gene to a null mutant strain that cannot otherwise produce the targeted subunit.

Expression of unc-29, unc-38, and lev-1 in Xenopus oocytes

In earlier studies, RNA isolated from mixed stage wild-type C. elegans was injected into Xenopus oocytes with the result that a dose-dependent depolarization was detected in response to bath-applied levamisole (Fleming et al., 1991, 1993). The levamisole response was not blocked by alpha -bungarotoxin (Tornoe et al., 1995), in agreement with [3H]meta-aminolevamisole-binding studies using membrane fragments (Lewis et al., 1987a) and cut worm muscle contraction assays (Lewis et al., 1980b). Cytoplasmic co-injection of cRNAs encoding UNC-29, LEV-1, and UNC-38 resulted in inwardly directed currents (holding potential, -60 mV) in response to levamisole (1 × 10-4 M), whereas oocytes injected separately with message encoding a single subunit or the equivalent volume of distilled water showed no such responses (Fig. 5A-E). Pairwise injection of all possible combinations yielded either no responses or inconsistent responses. In oocytes inje