Abstract
The acetylcholine receptor (AChR), an oligomeric protein composed of five subunits, is a component of the postsynaptic membrane at the vertebrate neuromuscular junction that plays a central role in synaptic transmission. The zebrafish mutation nic1 blocks the expression of functional and clustered nicotinic muscle AChRs. To understand the mechanisms underlying this lack of AChRs, we characterized the molecular defect in nic1 mutants. Our results suggest that the mutation affects the gene coding for the α-subunit of the AChR. Southern blot hybridization and DNA sequence analyses showed that the nic1 AChR α-subunit gene lacks part of intron 6 where the splicing branchpoint normally forms. Several lines of evidence suggest that this deletion blocks normal splicing; most nic1 α-subunit mRNAs retain intron 6 and are larger and less abundant than wild-type, some nic1 α-subunit mRNAs are internally deleted, and wild-type α-subunit mRNA rescues nic1 mutant cells. The nic1 mutation reduces the size of an intron, which prevents efficient splicing of the pre-mRNA, thus blocking synthesis of the α-subunit and assembly of AChRs. By this route, the nic1 mutation leads to paralysis.
ACETYLCHOLINE receptors (AChRs) are oligomeric proteins composed of four types of polypeptide subunits assembled in the ratio of α2βγδ. The AChR α-subunit (AChRα) is expected to play a critical role in normal assembly, clustering, and function of the AChR. Expression of the subunit proteins in Xenopus oocytes has shown that all four subunits are required to form receptors that respond normally to ACh (Mishinaet al. 1984). Although only the α-subunit is indispensable for α-bungarotoxin binding, the absence of any of the other subunits reduces binding. Mutational analysis of the subunit genes expressed in transfected COS cells demonstrated that the N-terminal domains of the α and δ subunits are required for surface expression of AChRs and for the formation of αδ heterodimers, an early step in the assembly of AChRs (Verrall and Hall 1992). Further, a cytoplasmic loop on the α subunit is required either in the early association of α and γ subunits or in the final assembly of αβδ and αγ subunits (Yu and Hall 1994).
Nicotinic AChRs cluster in the muscle cell membrane at high density at the neuromuscular junction. This clustering is regulated by the motor nerve terminal; in the living animal, large clusters of AChRs appear at the junction when the muscle is innervated (reviewed in Hall and Sanes 1993), and such clusters fail to form in the absence of innervation (Dahm and Landmesser 1991; Liu and Westerfield 1992). Normal clustering also requires the presence of all four subunits to form assembled receptors, although single subunits have clustered in a highly expressing cell line (Maimone and Merlie 1993).
To examine further the roles of the α-subunit during the formation of neuromuscular junctions in vivo, we previously screened for nonmotile mutants and isolated the γ-ray-induced nic1 zebrafish mutation (Westerfieldet al. 1990). The nic1 mutant embryos are paralyzed but otherwise develop normally. We showed that although nic1 mutant muscle cells contract in response to direct electrical stimulation, they lack functional AChRs as indicated by (1) an absence of AChR clusters identified by α-bungarotoxin binding or binding of antibodies that recognize the β, γ, or δ subunits and (2) their lack of response to the acetylcholine agonist, carbachol (Westerfieldet al. 1990). From analysis of genetic mosaic embryos, we showed that the mutation acts autonomously in muscle cells (Sepichet al. 1994); mutant muscle cells fail to cluster AChRs even when innervated by wild-type motoneurons, whereas wild-type muscles cluster AChRs when innervated by nic1 mutant motoneurons. Based on these results, we reasoned that the nic1 mutation may directly affect expression of one or more of the AChR subunit genes.
Here we provide evidence that the nic1 mutant phenotype is the result of a mutation in the gene coding for the α subunit of the AChR. We demonstrate that the phenotype is linked to a restriction fragment length polymorphism (RFLP) in the α-subunit gene and that both the RFLP and mutant phenotype map to the same location on the genetic map. We then show through sequence analysis that the nic1 mutant α-subunit gene lacks a specific part of intron 6 where the branchpoint may normally form during splicing. The deletion may block splicing of this intron. Consistent with this interpretation, we find that the α-subunit AChR mRNA is larger in mutants than in wild-type. Further, nic1 mutant muscle cells express intron-containing AChRα mRNA and other incorrectly spliced transcripts with partial deletions of the coding sequence. We find that providing nic1 mutant cells with wild-type AChRα mRNA rescues the mutant phenotype.
Our results suggest that a zebrafish mutation that prevents expression of functional α-subunits blocks expression of AChRs, thus providing strong support for the central role of the α-subunit. Furthermore, our analysis of the nic1 mutation suggests that a reduction in the size of an intron can decrease the efficiency of splicing of the α-subunit gene to a level that blocks assembly of the oligomeric AChR protein, leading to paralysis of the embryo.
MATERIALS AND METHODS
Fish stocks: We obtained diploid homozygous mutant embryos by crossing fish heterozygous for the nic1b107 (Westerfieldet al. 1990) mutation. Haploid mutant embryos were obtained by previously described methods (Streisingeret al. 1981). Embryos were staged according to hours (h) or days (d) postfertilization at 28.5° or by number of somites. Embryos and adults were maintained with standard conditions (Westerfield 1995).
Identification of fish heterozygous for nic1: Fish heterozygous for nic1 were identified either by examination of the progeny from individual crosses or by PCR amplification of intron 6 from the individual's genomic DNA.
DNA was extracted from caudal fins clipped from anesthetized adult fish. Fins were placed in 50 μl of digestion buffer (1.5 mm MgCl2, 50 mm KCl, 0.3% Tween 20, 0.3% NP-40, 10 mm TRIS, pH 8.3) heated at 98° for 10 min and then cooled to 55°. We added 150 μg of proteinase K (Boehringer Mannheim, Indianapolis, IN) and digested the fins at 55° for 60 min and then heated them at 98° for 10 min to inactivate the protease. The resulting DNA solution was diluted approximately 25-fold in water and stored at −20°. The DNA (1–5 μl) was denatured in the presence of PCR primers (40 ng) in a total volume of 10 μl for 5 min at 98° (primers sequence 5′-CGG GCT TGT GTT TTA CCT GC, corresponding to sequence GLVFYLI at amino acid 229 in the protein, and 3′-GAG GTG GAC GGG ATC AAC TCG, corresponding to sequence ELIPSTS at amino acid 262). PCR contained 10 μl of DNA primer, 1× Vent buffer (New England Biolabs, Beverly, MA), 6 μg of bovine serum albumin (New England Biolabs), 0.25 μM dNTPs (Boehringer Mannheim), and 0.2 μl of Amplitaq polymerase (Perkin-Elmer, Norwalk, CT) in a total volume of 30 μl. Reactions were amplified in an MJ Research (Waltham, MA) thermocycler, with program 3step (94° for 45 sec, 36 cycles of 92° for 45 sec, 60° for 45 sec, 72° for 1.5 min, 72° for 5 min) and held at 4° until collected.
Isolation of zebrafish AChRα cDNA: We screened 1.3 × 106 clones from a 32h embryonic zebrafish cDNA expression library (λgt11, a gift from Kai Zinn, Stanford University) with 32P-labeled random hexamer-labeled probes (Feinberg and Vogelstein 1984). The probes were made from two HinfI fragments from the Torpedo AChRα cDNA (a gift from Toni Claudio, Yale University). The coding sequence corresponds to the ACh-binding site and transmembrane region III. Filters were hybridized in 5× SSC, 50 mm sodium pyrophosphate, 5× Denhardt's, 100 μg/ml calf thymus DNA, 100 μg/ml yeast tRNA, with 2.2 × 105 cpm/μl of probe at 54° for 2 days, washed twice in 5× SSC with 0.1% SDS at 54°, twice in 2× SSC with 0.1% SDS at 54°, and exposed to film for 2–5 days at −70° with intensifying screens. Three overlapping clones were recovered and subcloned into the EcoRI site of pBlueScript II KS(−) (Stratagene, La Jolla, CA).
The sequence of the longest clone (pAChRa or p3a1) was determined. The sequence was assembled and the encoded protein sequence deduced (Genetics Computer Group, University of Wisconsin). Alignments with sequences from other species were made using Lasergene (DNAstar, Madison, WI).
Isolation of wild-type AChRα gene: We screened 4.6 × 105 clones from a zebrafish genomic library (λDASH II, a gift from Barbara Jones and Martin Petkovich, Queen's University, Kingston, Ontario) with a 32P-labeled random hexamer-primed zebrafish AChRα cDNA (pAChRa). Filters were hybridized with 5 × 105 cpm/μl probe in 5× SSC, 50 mm sodium pyrophosphate, 5× Denhardt's, 100 μg/ml calf thymus DNA, 100 μg/ml yeast tRNA at 62° for 2 days, then washed to 0.25× SSC, 0.1% SDS at 65°. Filters were exposed for 6 days at −70° with an intensifying screen. Three clones were isolated. We determined the DNA sequences for exon/intron junctions surrounding introns 5 and 6 (numbered by analogy to the human gene; Nodaet al. 1983).
Isolation of nic1 mutant AChRα gene: Oligonucleotide primers were synthesized (Biotech Services, University of Oregon; primers α5′-GCT GCA GAT CGT GGT GCT TCA C, and α3′-GCT GCA CCA CAG AGA TGC TGA) to regions of the AChRα cDNA flanking the RFLP. An approximately 2.0-kb product was amplified from genomic DNA isolated from nic1 mutant haploid embryos or from DNA isolated from adult fish heterozygous for the RFLP in a 100 μl reaction mix of 100 pmol of each primer, 200 μM dNTPs, 1× Replinase buffer (New England Nuclear, Dupont) and 1 unit of Replinase. PCR amplifications were obtained in a MJ Research thermocycler. The resulting PCR products were size fractionated on an agarose gel. Ends of the DNA products were repaired with T4 DNA kinase and DNA fragments were self-ligated (Kaufman and Evans 1990) and digested with PstI. PCR products were cloned into the PstI site of pBlueScript II KS(+) (Stratagene). Colonies were screened by hybridization with an AChRα cDNA. Subclones (p14.2ampmut and p15ampmut) were recovered from material amplified from the DNA of two heterozygous fish and were partially sequenced.
Sequencing: DNA was sequenced by a modification of the method of Chen and Seeburg (1985); double-strand sequencing mixes were prepared with Sequentide (New England Nuclear, Dupont). Some of the sequence was determined using automated DNA sequencers at the Central Services Laboratory of the Center for Gene Research and Biotechnology at Oregon State University at Corvallis and at the Research Services of the Molecular Genetics Facility at the University of Georgia (GenBank accession numbers: U70435, U70436, U70437, U70438) or at the University of Oregon Biotech Lab.
In situ hybridization: AChRα antisense riboprobe was labeled with digoxygenin-UTP and transcribed by T7 RNA polymerase from an XhoI-digested plasmid p3a1 using the Genius kit 1 (Boehringer Mannheim). In situ hybridizations were performed as previously described (Oxtoby and Jowett 1993). Embryos were frozen and sectioned. Nuclei were labeled with 1% neutral red after in situ hybridization.
Mapping: The nic1 mutant phenotype was mapped with the random amplified polymorphic DNA (RAPD) method used to establish the zebrafish linkage map (Postlethwaitet al. 1994). In brief, fish heterozygous for the nic1 mutation were crossed to Darjeeling zebrafish to give nic1/Dar F1 heterozygotes. Haploid progeny were obtained from the F1 female fish and were sorted into motile and nonmotile (wild-type and nic1 mutant) phenotypes. DNA was isolated and pooled from a batch of about 50 wild-type or nic1 mutant haploid embryos. Approximately 150 RAPD primer sets that were used to establish the map were tested against the pooled wild-type or nic1 mutant DNA. Primer sets that appeared to be linked to the mutation were tested against a second batch of DNA. Primer sets that still appeared to be linked were then tested against about 50 individual wild-type and about 50 nic1 mutant haploid embryos. Once a linkage group was found, all primer sets within that group were tested. The number of recombinant embryos was used to calculate recombination distances (LINKER program, Postlethwaitet al. 1994). Mapmaker (Landeret al. 1987; Postlethwaitet al. 1994) was used to prepare a linkage map.
RNA isolation, Northern blot analysis, and RNase protection: RNA was isolated from zebrafish embryos and adults by an acid–phenol method (Chomczynski and Sacchi 1987). Poly(A)+ RNA for Northern blots was isolated using the Microfast Tract kit (Invitrogen, San Diego, CA). Blots were hybridized overnight in 5× SSC, 5× Denhardt's, 50% formamide, 0.1% sodium dodecyl sulfate (SDS) and 100 μg/ml salmon sperm DNA (Sambrooket al. 1989), washed to 0.5× SSC at room temperature and exposed to film at −70°, with an intensifying screen for 49 hr or washed to 0.1× SSC at room temperature and exposed for 7 days.
RNase protection experiments were performed with the RPA II kit (Ambion, Austin, TX). A 320 bp RsaI fragment from the nic1 mutant gene (p14.2ampmut) was cloned into the SmaI site of pBlueScript II KS(+). Antisense riboprobe was transcribed from the T7 promoter of XhoI-digested plasmid. Antisense zebrafish MyoD1 (Weinberget al. 1996) was transcribed with T7 polymerase from XbaI-digested plasmid. Riboprobes were synthesized with the Maxiscript kit (Ambion) and purified by gel electrophoresis. Total RNA (20 μg) was hybridized with about 50 kcpm at 42° overnight. Gels were dried onto paper and exposed to film on an intensifying screen at −70° for 12 hr to 10 days. Relative abundance of mRNAs was estimated by comparison to the MyoD1 control.
RT-PCR: RNA was prepared from 70 wild-type and 70 nic1 mutant embryos at 50h using Tripure isolation reagent (Boehringer Mannheim). RNA (1 μg) was reverse transcribed (RT) and subsequently amplified using a CapFinder PCR cDNA synthesis kit (CLONTECH). The following primers were used for amplification of α-subunit cDNAs from wild-type and nic1 mutant cDNA samples: (forward, exon 5 primer A) GAG GTG GAC GGG ATC AAC TCG, (forward, exon 6 primer B) GGG CTT GTG TTT TAC CTG CCC; (reverse, exon 7 primer C) CGA GTT GAT CCC GTC CAC CTC, (reverse, exon 8 primer D) TGG TTA TGG GAG AGT GGT AAG. PCR conditions were 94° for 1 min (first cycle only), 92° for 30 sec, 60° for 30 sec, 75° for 90 sec, cycle 30 times, 75° for 5 min. DNA fragments were subcloned into the pCR2.1 vector using the TA cloning kit (Invitrogen). Samples were sequenced by the University of Oregon sequencing facility.
Southern blot analysis: DNA was extracted from zebrafish by a modification of the method of Blin and Stafford (1976). Adults were anesthetized by chilling with ice. Tissues were quick frozen in liquid nitrogen and ground with pestles (Kontes, pellet pestle) in extraction buffer (embryos) or in a frozen mortar and pestle (adults) and dispersed in extraction buffer (50 mm EDTA pH 8.0, 0.5% sarcosyl, and 0.2 mg/ml proteinase K). The tissue was incubated at 50° for 1–3 hr, extracted three times with phenol/chloroform/isoamyl alcohol, ethanol precipitated, and resuspended in sterile H2O. Blots were prepared according to Sambrook et al. (1989). Gels were blotted onto positively charged nylon membranes (Boehringer Mannheim) or Hybond-N membranes (Amersham, Arlington Heights, IL).
Blots were hybridized with either 32P-labeled random hexamer-labeled probes or digoxygenin-labeled (Boehringer Mannheim) probes made from all or part of the AChRα cDNA. Radioisotope-labeled probes were hybridized in 5× SSC, 50 mm sodium pyrophosphate, 5× Denhardt's, 100 μg/ml calf thymus DNA, 100 μg/ml yeast tRNA, and 0.2% SDS buffer at 65° for 16–48 hr, washed in 0.5× SSC at 62°, and exposed to film on an intensifying screen at −70° for 3–7 days. Digoxygenin-labeled probes were hybridized overnight at 65° in 5× SSC, 0.1% N-lauroylsarcosine (Sigma), 0.02% sodium lauryl sulfate, 1% blocking reagent (Boehringer Mannheim), with the addition of 50 μg/ml heparin (Sigma), and subsequently washed in 0.1× SSC, 0.1% SDS at 65°. Bands were revealed with anti-digoxygenin–alkaline phosphatase antibodies and Lumiphos (Boehringer Mannheim) according to the manufacturer's instructions and exposed to film for 3 hr to overnight. Seven probes were used for Southern blot analysis: a full-length cDNA, a partial cDNA whose 3′ extent is coincident with the most 3′ probe of Figure 2D, three fragments shown in Figure 2D, and two BamHI fragments from the 5′ end of the cDNA.
Western blot analysis: Western blot analysis was conducted essentially as described previously (Hattaet al. 1991). Wild-type or nic1 mutant embryos at 3–4d were solubilized in 3.5% SDS. Proteins were separated by electrophoresis on a 10% polyacrylamide gel and blotted to polyvinylidene difluoride sheets (Towbinet al. 1979). After blotting, the sheets were cut into strips that contained the equivalent of approximately 100 embryos each. The strips were probed with a primary antibody (MAB 149; Sargentet al. 1984) specific for the AChRα protein.
Expression of wild-type AChRα cDNA in nic1 mutant embryos: The AChR cDNA was subcloned into a plasmid (Klein-Hitpasset al. 1986) containing estrogen-responsive elements (ERE) from the frog vitellogenin A2 gene to form plasmid pEREaAChR. Embryos at the two- to four-cell stage were coinjected with the pEREaAChR (30 ng/μl) and pKCR2-ER (15 ng/μl), which contains the human estrogen receptor cDNA driven by an SV40 promoter (Breathnach and Harris 1983; Greenet al. 1986). To activate AChR expression, embryos were transferred at 6 or 24h to medium containing 10−7 m β-estradiol.
After the embryos grew 28h to 3d, their clustered AChRs were labeled as previously described (Westerfieldet al. 1990). In brief, they were soaked in 40 μM tetramethylrhodamine α-bungarotoxin (R-BTX; Molecular Probes, Eugene, OR) in 15% dimethyl sulfoxide in L-15 medium at 2–12° for 15–30 min. Embryos were rinsed three to five times in L-15 medium. Rhodamine-labeled AChR clusters were visualized with a Zeiss UEM microscope, Videoscope KS 1381 intensifying tube, and MTI series 68 video camera. Images were averaged (8–16 individual images) using a Macintosh IIci computer (Apple) and Axovideo software (Myers and Bastiani 1991). Embryos were subsequently fixed and processed as whole mounts for in situ hybridization to reveal cells expressing AChRα mRNA at high levels. Whole mounts were photographed from the microscope. Composite images were made with Photoshop software (Adobe). Digitized fluorescence images were processed only by subtracting the background below a minimum intensity (threshold pixel value). This minimum intensity level was selected by eye. Composite images were made by adding fluorescent images to Nomarski images.
RESULTS
The paralyzed phenotype of nic1 mutation is linked to an RFLP in the AChRα gene: The zebrafish mutation nic1 blocks AChR function (Westerfieldet al. 1990) and acts autonomously in muscle cells (Sepichet al. 1994). Because the mutation produces no other obvious phenotypes and lacks embryonic lethality, we suspected the mutation might specifically block the synthesis of AChRs rather than block a general cellular function that would indirectly cause paralysis. To examine AChR synthesis, we cloned the zebrafish AChRα cDNA. The longest clone of 2.2 kb spans the coding sequence and includes approximately 150 base pairs (bp) of untranslated 5′ leader and approximately 700 bp of untranslated 3′ region, ending in a 26 bp poly(A) tail.
The conceptually translated protein sequence (Figure 1) is similar at the amino acid level to the AChRα proteins of other vertebrates: 77% identical to Xenopus 1A (Hartman and Claudio 1990), 78% to Xenopus 1B (Baldwinet al. 1988), 77% to Torpedo californica (Nodaet al. 1982), 77% to mouse (Boulteret al. 1985), and 79% to chicken (Conti-Tronconi et al. 1988).
We compared the Southern blot patterns of the AChRα gene from haploid nic1 mutant and wild-type sibling embryos. We sorted haploid progeny from individual heterozygous female fish (nic1/+) according to phenotype, extracted DNA from pools of phenotypic nic1 mutant or wild-type embryos, and hybridized the resulting Southern blots with probes from the zebrafish AChRα cDNA. Mutant embryos had RFLPs in the AChRα gene for restriction enzymes HindIII, BglII (Figure 2A), and PstI (not shown). As expected, the Southern blot pattern in wild-type sibling embryos matches that of the clonal line, C29 (Figure 2) and C32 (data not shown), the backgrounds on which the nic1 mutation was generated.
The AChRα gene and the nic1 mutation map together: We mapped (Postlethwaitet al. 1994) both the nic1 mutant phenotype and the AChRα gene. The nic1 mutant phenotype is linked to four markers on linkage group VI of the zebrafish map (Figure 2C); nic1 is 39.8 cM from SSR.26, 8.5 cM from M12.350, 22.4 cM from N5.675, and 29.6 cM from U6.850. The map distances we report differ slightly from the published linkage group VI. Presumably this reflects the effect of genetic background on recombination frequency or statistical variation owing to the small number of embryos used to map the nic1 locus.
We mapped the AChRα gene using a polymorphism present in nic1 mutants. PCR primers to sequences flanking intron 6 amplified an approximately 220-bp product from the DNA of haploid nic1 mutant embryos, but not from their haploid wild-type siblings. Using the same embryos from the previous mapping cross, we amplified the 220-bp product from all 46 haploid phenotypically nic1 mutant embryos, but not from their 48 haploid wild-type siblings. Thus, the AChRα gene is located within 0–2 cM of nic1.
nic1 mutation is a deletion: The differences in the Southern blot pattern between haploid mutants and their wild-type siblings can be explained by deletion of a region containing a HindIII site (Figure 2D). The nic1 mutant Southern blot pattern shows an 8.1-kb HindIII fragment, whereas two fragments of 6.2 and 3.5 kb appear in wild-type (Figure 2B). Digestion with restriction enzymes PstI (not shown) or BglII (Figure 2A) also revealed a single fragment in nic1 mutants that is smaller than in wild-types. These differences are consistent with a deletion.
We identified the region of the gene that corresponds to the RFLP. Most of the nic1 mutant AChRα gene is indistinguishable from wild-type; probes from the 5′ or 3′ ends of the cDNA showed identical patterns in the genomic DNA from nic1 mutants and wild-type siblings (see Figure 2B, most bands are identical). The differing polymorphic fragments correspond to a single region of the AChRα gene. Probes hybridizing to the coding region between the ACh-binding site and transmembrane domain I detect the 6.2-kb HindIII fragment in wild-type DNA (Figure 2D, the two more 5′ probes). Probes spanning transmembrane domains II and III detect the 3.5-kb fragment in wild-type (Figure 2D, most 3′ probe), whereas both sets of probes detect the nic1 mutant 8.1-kb fragment. The smaller DNA fragments in nic1 mutants suggest that the nic1 mutation resulted from the loss of approximately 1–3 kb of the AChRα gene, where it codes for transmembrane domains I to domain II.
To learn whether the AChRα gene had suffered a deletion and to learn how the mutation might act, we compared the suspected regions from nic1 mutants and wild-type zebrafish. We isolated the wild-type AChRα gene by screening a genomic library made from wild-type (outbred) zebrafish DNA. To obtain the nic1 mutant AChRα gene, we amplified by PCR the sequence spanning transmembrane domains I and II from the genomic DNA of nic1/+ heterozygotes. An approximately 2.0-kb product was amplified and cloned from genomic DNA isolated from two different adult fish heterozygous for the RFLP. To learn whether coding sequence or splice junctions had been deleted, we determined the sequence. We found that the nic1 mutant gene has all the coding sequences and splice junctions within the region of the suspected deletion (from the 3′ end of intron 5 to exon 7, which encodes transmembrane domain II). However, intron 6 is only 110 bp in the nic1 mutant compared to more than 1 kb in wild-type.
Interspecies comparisons of the conceptual translation of the zebrafish AChRα cDNA. Comparison with AChRα proteins from frog 1A and frog 1B expressed in oocyte, ray, chick, and mouse. Spaces that were introduced to improve the alignment are shown as asterisks (*) and amino acids that are identical to the zebrafish sequence are shown as dashes (−). The locations of structural motifs are indicated above the sequences; the location of the last amino acid before intron 6 is indicated by the vertical arrow (↓).
The differences we observed between the sequences of the wild-type and nic1 mutant AChRα genes were confined to intron 6. As expected from the Southern blot analysis (Figure 2B), the nic1 mutant gene lacked approximately 1 kb of intron 6 present in the wild-type gene (Figure 3), reducing the intron to 110 bp. Wild-type and nic1 mutant sequences are identical for the first 39 bp and almost identical for the last 71 bp of intron 6. We found single basepair mismatches at nucleotide (nt) 42 (C→G) and at nt 45 (G→A) of the nic1 mutant intron 6.
Neither the wild-type nor the nic1 mutant intron contains a polypyrimidine tract or potential branchpoint in the typical location, adjacent to the 3′ splice junction. Both introns have long (60–120 bp) runs of poly(GT) adjacent to the 3′ splice site. The wild-type intron has several polypyrimidine tracts (Figure 3, bars above sequences) near potential branchpoints (Figure 3, arrows) at 150–250 bp upstream of the 3′ splice site. Although there are several potential branchpoint adenosines in the nic1 mutant intron, all are within 44 bp of the 5′ splice site and thus may be too close for efficient splicing (Fuet al. 1988; Smith and Nadal-Ginard 1989; reviewed in Smithet al. 1989). Northern blot analysis (data not shown) supported this interpretation; AChRα mRNA in nic1 mutants is larger than in wild-type siblings.
The nic1 mutation and the AChRα gene are linked. (A) The paralyzed nic1 phenotype is linked to an RFLP in the AChRα gene. Southern blot of haploid nic1 mutant embryos, their haploid wild-type siblings, and C29, one of the clonal lines on which the mutation was made (C32, the other parental line, has a Southern pattern identical to C29). Haploid wild-type siblings are indistinguishable from parental lines. DNA was digested with restriction enzymes HindIII (H) or Bgl II (B). Blot was probed with a DNA fragment corresponding to the 5′ probe in (D). In 12 out of 14 cases the haploid embryos from single female fish (nic1/+) showed linkage of the mutant phenotype to the nonclonal RFLP pattern, whereas the wild-type phenotype was found in embryos with a clonal RFLP pattern as illustrated here. The other two cases had opposite or mixed results probably owing to errors in sorting the embryos. (B) The nic1 mutant AChRα gene lacks a HindIII site. Genomic DNA from nic1 mutant embryos and adult C29 fish were digested with HindIII. Southern blot was probed with the entire zebrafish AChRα cDNA. The nic1 mutants differ from C29; clonal wild-types have 3.5 and 6.2 kb HindIII fragments (right arrows), whereas nic1 mutants have an 8.1 kb fragment (left arrow). The AChRα genes in both lines have 2.0, 5.2, and 10 kb HindIII fragments. (C) The nic1 mutant phenotype (nic1) and a PCR marker of the nic1 mutant AChRα gene (AChRα) map to linkage group VI on the zebrafish map. (D) The nic1 mutant Southern blot pattern predicts a small deletion in the AChRα gene. (Top) Three probes are represented as boxes below their locations in the cDNA. The coding region of the cDNA is shown as a black box; noncoding regions are shown as white boxes. Hnf I sites (Hf) in the cDNA are shown. (Bottom) Probes are shown adjacent to the genomic HindIII fragments they detect. Fragments that differ between mutants and wild types are shown as solid white boxes. Fragments that are the same in nic1 mutants and wild types are shown as dashed line boxes. Fragment sizes are indicated in kilobases.
nic1 mutants make mutant AChRα mRNA: The differences between the nic1 mutant and wild-type AChRα genes suggest that the nic1 mutant gene makes an AChRα mRNA that retains intron 6. We tested this prediction using an RNase protection assay. We designed a riboprobe (Figure 4) that would protect intron 5 (predicted to be spliced out in all cases), exon 6 (predicted to be retained in all cases), and intron 6 (predicted to be retained only in mutants). Wild-type AChRα mRNA protected a fragment of about 239 nt, corresponding to exon 6. RNA from embryos carrying the nic1 mutant gene protected two fragments: the 239-nt exon 6 fragment and an approximately 290-nt fragment corresponding to exon 6 and intron 6. Both RNAs in nic1 mutants are reduced in abundance relative to myoD mRNA, which was used to indicate the total quantity of muscle mRNA (data not shown). The smaller fragment protected in nic1 mutants is present at approximately 25% of the level observed in wild-type embryos. The intron 6-containing fragment is present at about 15% of the amount observed for the wild-type protected fragment in wild-type embryos.
To obtain further proof of the mutant nature of the nic1 AChRα mRNA, we amplified RNA from mutant and wild-type embryos by RT-PCR using four sets of primers that span intron 6 (Figure 5). The largest band amplified from mutant nic1 RNA by each primer pair was 110 bp longer than the corresponding product from wild-type RNA (Figure 5, asterisk). Sequence analyses of these products showed that they all contained the 110 bp nic1 mutant intron 6 (Figure 3), demonstrating that these nic1 transcripts fail to splice the mutant intron. Amplification of nic1 mutant RNA with two of the primer pairs, AD and BD, revealed additional products (Figure 5, solid line) that appeared consistently in each RT-PCR. Sequence analyses of these products showed that they represented incorrectly spliced transcripts which lacked parts of the coding sequence. Out of a total of 15 sequenced clones, none corresponded to the wild-type sequence; all were incorrectly spliced. Clones were of six types that retained intron 6 or spliced exon 6 to exon 8 and four types that spliced out parts of the sequence from exons 6 to 8.
nic1 mutants express reduced amounts of AChRα protein: The splicing defects demonstrated by our RNase protection (Figure 4) and RT-PCR (Figure 5) analyses predicted that the nic1 mutant gene produces truncated α-subunit proteins at reduced levels. To examine this possibility, we compared AChRα proteins from wild-type and nic1 mutant embryos by Western blot analysis (Figure 6). We probed blots of proteins isolated from whole embryos with an antibody (MAB 149) that has been shown to recognize an epitope in the intracellular domain near the C-terminal of the AChRα protein (Sargentet al. 1984). The antibody revealed a band at about 50 kD in wild-type, but not in nic1 mutant protein, consistent with the mutant producing reduced amounts of AChRα.
The nic1 mutant AChRα gene lacks intron sequence and makes an mRNA that is larger than in wild types. Wild-type gene sequence appears in upper line and nic1 mutant sequence below. Splice junctions are denoted by boldface GT at 5′ junction (consensus, (C/A)AG GTRAGT) and AG at the 3′ junction (YnNYAG G). Arrows point to A's at potential splicing branchpoints (consensus, YNY-TRAY). Polypyrimidine tracts are indicated by bars above the sequence. Arrowheads point to single nucleotide differences between nic1 mutant and wild-type AChRα genes. An in-reading-frame stop codon is indicated.
Early expression of the AChRα gene differs between mutants and wild types: We used RNA in situ hybridization to examine levels and patterns of expression in vivo. The same tissues express AChRα mRNA in nic1 mutant embryos and their wild-type siblings; however, there are distinct differences in abundance and cellular localization of the message.
At 30h, muscle cells in both mutant and wild-type embryos express high levels of the AChRα mRNA. Transcripts are most abundant in muscle cells near the lateral surface of the myotome (Figure 7, A and B), which may include the precursors of the adult slow muscles (Devotoet al. 1996). Abundance of the mRNA in lateral nic1 mutant muscle cells is comparable to or exceeds that in wild-type cells and appears to be more closely associated with nuclei than it is in wild-type muscle cells (Figure 7, C and D). Wild-type embryos also express AChRα mRNA in the cytoplasm of more medial muscle cells, whereas, consistent with the RNase protection studies (Figure 4), cytoplasmic expression of the mRNA is reduced in nic1 mutants (Figure 7, B and D).
Wild-type AChRα mRNA rescues the nic1 mutant phenotype: If the mutant lacks AChRs because of an absence of the α-subunit mRNA, then supplying the wild-type mRNA should restore AChRs. We tested this prediction by injecting nic1 mutant and wild-type embryos, at the two- to four-cell stage, with a plasmid expressing the wild-type AChRα cDNA. At 28h, 2d, or 3d, we examined the surviving embryos for the presence of clustered AChRs by labeling with rhodamine–α-bungarotoxin. We then fixed the embryos and examined them for AChRα RNA by in situ hybridization. Out of 68 injected embryos, 75% were phenotypically wild type, as expected; they could swim and contained many labeled AChR clusters in a wild-type pattern. Seventeen embryos were identified as nic1 mutants because they didn't swim. These embryos had rhodamine–bungarotoxin labeling on some of their muscle cells, unlike uninjected nic1 mutants, which never show detectable levels of bungarotoxin labeling (Westerfieldet al. 1990). Fifteen of these embryos had both rhodamine–bungarotoxin labeling on muscle cells and muscle cells that contained high levels of AChRα mRNA, and we could identify individual muscle cells that contained both the wild-type AChRα mRNA (Figure 8, top) and AChR clusters (Figure 8, bottom). The patchy appearance of these rescued cells is probably the result of the mosaic distribution of the injected cDNA (Westerfieldet al. 1992). In two 48h embryos, the injected AChRα cDNA rescued nerve–muscle communication; each embryo had a muscle cell that contained AChRα mRNA, that clustered AChRs, and that contracted, whereas neighboring cells did not. Contractions are never observed in uninjected nic1 mutants at this stage (Westerfieldet al. 1990). The rescue of AChR clustering and neuromuscular function by supplying a source of wild-type mRNA supports the interpretation that the paralysis in nic1 mutants results from an absence of wild-type α-subunit mRNA.
Embryos that carry the nic1 mutant gene make an AChRα RNA that retains intron 6. (Left top) Schematic representation of riboprobe. Exon is represented by open box, introns by thin lines, vector sequences by thick lines. (Left bottom) Predicted protected fragments. (Right) RNase protection analysis. Mutant nic1 and wild-type siblings (wtsib) make an RNA that retains intron 6, whereas unrelated wild-types (wt) make only fully spliced AChRα mRNA.
The nic1 mutant AChRα gene is incorrectly spliced. (Top) Schematic diagram of part of the AChRα pre-mRNA in wild types (wt) and nic1 mutants (nic1). Solid lines indicate coding sequences of exons 5–8 and the intron 6 sequence from the nic1 mutant (exons 5–8 are truncated). Introns (not drawn to scale) are indicated by dashed lines. Primers used for PCR amplification of RT RNA are indicated by half arrows and letters. Scale bar, 50 bp. (Bottom) RT-PCR amplification products from wild-type (WT) and nic1 mutant RNA using primer pairs AD, AC, BD, and BC, whose binding sites are indicated by the schematic above. Amplification products from the bands indicated with asterisks (*) were sequenced and correspond to the nic1 mutant transcripts with unspliced intron 6. Amplification products from the bands indicated with dashes (-) were sequenced and correspond to incorrectly spliced transcripts with internal deletions of the coding sequence. Size markers in kilobases.
Expression of AChRα protein is reduced in nic1 mutants. Western blot of proteins extracted from wild-type (wt) and nic1 mutant (nic1) embryos probed with mab 149 (Sargentet al. 1984), which is specific for the intracellular domain of the AChR α-subunit. Control lane (con) is protein from wild-type embryos processed without the primary antibody. All three lanes were from the same gel. Molecular weight marker in kilodaltons.
Feeble movements may indicate leakiness of the mutation: Some mutant embryos recover slightly from paralysis, beginning at 2–3 days of development. Compared to movements in wild-type siblings, the movements in nic1 mutants are fewer and slower. Because the feeble movements resemble movements in wild-type siblings and are provoked by touching the embryos, we speculate that nic1 mutant embryos may have a few functional AChRs. We have been unable to detect AChR clusters with rhodamine–α-bungarotoxin labeling in weakly motile mutant embryos, suggesting they are too few in number to be detected by this method. Alternatively, the feeble movements could result from depolarization of dying muscle cells. This seems unlikely, however, because weakly motile nic1 mutant embryos can live for up to 12 days and we observe no obviously abnormal cell death.
The expression pattern of the AChRα differs between wild-type sibling and nic1 mutant embryos. Transverse sections at trunk levels of 30-hr wild-type (wt) (A) and nic1 mutant (B) embryos. AChRα expression is revealed by the blue reaction product. Nuclei are stained with neutral red. AChRα mRNA is less abundant in the nic1 mutant medial myotome. Side view of whole-mount wild-type (C) and nic1 mutant (D) embryos. AChRα mRNA is more closely associated with nuclei in nic1 mutant embryos. Scale bar, 50 μm.
DISCUSSION
The nic1 mutant phenotype is linked to a deletion in the AChRα gene: Our results provide several lines of evidence that implicate the AChRα gene as the locus of the nic1 mutation. First, the nic1 paralyzed phenotype is linked to the AChRα gene. The nic1 paralyzed phenotype maps to the same location as the PCR marker of the small nic1 mutant intron 6, placing the mutation within 0–2 cM of the AChRα gene (Figure 2C). Second, our Southern blot analyses (Figure 2, A and B) demonstrate that the paralyzed nic1 mutant phenotype is linked to an alteration in the AChRα gene and suggest that approximately 1.5 kb of the AChRα gene containing a HindIII site is absent from nic1 mutants. Third, direct comparison of the corresponding regions of the wildtype and mutant genes (Figure 3A) shows that a sequence is missing that reduces the size of intron 6 in nic1 mutants without disturbing coding sequence or splice junctions.
Clustered AChRs can be rescued in nic1 mutants with a plasmid encoding wild-type AChRα cDNA. (Top) Photomicrograph of whole-mount in situ hybridization of an injected nic1 mutant embryo shows that a muscle cell expressed AChRα mRNA (arrow) above the endogenous levels of expression. (Bottom) Composite photomicrograph combining bright-field and fluorescence views of live embryo. The same cell as shown above formed clusters of AChRs (arrow) as revealed by rhodamine–α-bungarotoxin labeling. Similar results were obtained in other experiments; in a set of 10 nic1 mutant embryos, we counted 18 cells that expressed the injected AChRα cDNA. Seventeen of these cells also had AChR clusters. Arrowhead indicates a neighboring pigment cell. Scale bar, 50 μm.
The nic1 mutation makes unspliced AChRα transcripts: The structure of intron 6 of the AChRα in the nic1 mutant suggests how the mutation may act; an essential interaction between the splice junction and splicing branchpoint may be sterically inhibited. Splicing requires the presence of conserved sequence elements at the exon/intron borders (5′ and 3′ splice junctions), an element within the intron where the branchpoint forms, and an adjacent polypyrimidine tract (Smithet al. 1989). The branchpoint and polypyrimidine tract are usually near (18–37 nt upstream) the 3′ splice junction but can be much farther upstream. In the first step of splicing, a guanosine at the 5′ splice junction makes a covalent bond with an adenosine in the splicing branchpoint. Efficient splicing requires a minimum separation of 50 or more nucleotides between the 5′ splice site and the branchpoint. In the splicing of SV40 small-t antigen pre-mRNA, deletions that reduced the separation to less than 46-nt reduced or blocked splicing (Fuet al. 1988). In α-tropomyosin pre-mRNA, the normal inhibition of splicing of exon 2 is relieved by increasing the separation from 42 nt to 51–59 nt (Smith and Nadal-Ginard 1989). The nic1 mutant intron is small (110 nt) with several potential splice branchpoints near the 5′ splice junction. The potential branchpoint farthest downstream from the 5′ splice junction is 44 nt away, which is potentially too close for efficient splicing. The potential branchpoint that most resembles the canonical sequence is located 24 nt downstream of the 5′ splice junction, too close to support splicing. Splicing efficiency and the use of a particular branchpoint are also improved by an adjacent polypyrimidine tract (Reed 1990; Roscignoet al. 1993). In the wild-type intron, there are polypyrimidine tracts mixed with potential branchpoints at 140–240 nt upstream of the 3′ splice site. In the nic1 mutation, candidate branchpoints lack these downstream polypyrimidine tracts. Consequently, the potential branchpoints in nic1 mutant AChRα may be unable to engage in splicing.
In our RNase protection experiments, we found that most of the RNA in nic1 mutants retains intron 6 (Figure 4). A smaller number of transcripts retain exon 6 and splice out intron 6. Sequence analysis of RT-PCR products (Figure 5) demonstrated that nic1 mutant embryos can remove intron 6 only by splicing out flanking coding sequence and consequently make no detectable wild-type mRNA. We also found the amount of AChRα message reduced in nic1 mutants (Figures 4 and 7). The amount of intron 6-containing mRNA in nic1 mutants and the messages with internal deletions are much reduced from wild-type mRNA levels, suggesting that splicing efficiency is lower or that the unspliced and incorrectly spliced RNAs are unstable.
There are several possible consequences of an unspliced pre-mRNA. First, if the remaining intron is unrecognized, the RNA may leave the nucleus unspliced. The translation of a nic1 mutant mRNA that contains intron 6 would end at an in-frame stop codon at the intron/exon junction, producing a truncated protein. The predicted truncated protein includes the extracellular domain, transmembrane domain I, and 35 novel amino acids encoded by the intron. Such a truncated protein might block assembly of functional receptors as a dominant negative mutant. A mutant α-subunit protein that was engineered with a truncation after transmembrane domain I was shown to bind to the δ-subunit and block the assembly of the receptor (Vernall and Hall 1992). Because the nic1 heterozygote is without apparent phenotype, such as weakness or fatiguability, we have no evidence for the translation of an unspliced RNA that acts as a dominant negative.
Second, if the intron sequence remains unspliced, the RNA may be retained within the nucleus until it is degraded. We see a close association of AChRα mRNA with nuclei in nic1 mutant embryos (Figure 7). The reduced amount of AChRα RNA in mutants (Figures 4 and 7) is consistent with efficient breakdown of the unspliced mRNA, either in the nucleus or the cytoplasm. The mislocalization and reduced levels of AChRα mRNA in nic1 mutants are presumably responsible for the reduced expression of AChRα protein (Figure 6).
Third, the RNA may be spliced inefficiently to yield low levels of functional mRNA. Our RNase protection experiments showed both spliced and unspliced AChRα RNA (Figure 4), and our sequence analysis of the RT-PCR products (Figure 5) showed that all spliced transcripts include deletions of coding sequence. Thus, cryptic splicing, forced by the small nic1 mutant intron 6, forms internally deleted products. Some of these products, however, may code for functional or partially functional receptor subunits. This interpretation may explain the partial recovery of the paralyzed phenotype we see in older embryos.
Interpretation of the nic1 mutant phenotype: The nic1 mutant is a paralyzed embryo that later often shows weak contractions of a few muscles. The behavioral recovery in nic1 mutants may indicate expression of a few transcripts from the mutant locus that code for functional AChR α-subunits. An alternative explanation for the functional recovery is that muscles may be activated by nonmuscle AChRs. Corriveau et al. (1995) have shown that embryonic chick muscles express substantial amounts of neuronal AChRs containing α7-subunits, as well as other neuronal AChRs. If zebrafish muscles also synthesize neural α7-subunit-containing receptors, then these receptors could theoretically mediate movement in nic1 mutant embryos.
Our analysis of the nic1 mutant AChRα gene provides a plausible explanation for the paralyzed phenotype. The mutation reduces the splicing efficiency of the AChRα mRNA, leading to a reduced amount of functional mRNA and a virtually receptorless, paralyzed embryo. This model is supported by several observations: (1) The mutant phenotype is linked to an RFLP in the AChRα gene. (2) The RFLP results from a deletion of sequences within intron 6. (3) AChRα mRNA in nic1 mutant embryos is longer than in wild-types, consistent with a retained intron. (4) The nic1 mutant embryos express α-subunit mRNAs that retain intron 6 or are internally deleted and express reduced amounts of α-subunit protein. (5) Our in situ hybridization experiments showed that α-subunit mRNA is much reduced in abundance in nic1 mutant embryos, consistent with an unstable mRNA. (6) The absence of clustered and functional AChRs is rescued in nic1 mutant embryos by supplying wild-type AChRα mRNA. Consequently, partial loss of intron sequences leads to paralysis of the nic1 mutant embryo.
Acknowledgments
We thank Jon Lindstrom for the generous gift of AChR antibodies, Ruth Bremiller and Michelle McDowell for help with histology, Frankie Kimm and Ellen Johnson for help with fish, and Charles Kimmel for comments on the manuscript. The ERE construct was developed with help from Uwe Strähle. This work was supported by National Institutes of Health grants NS21132, HD22-486, and GM7257.
Footnotes
-
Communicating editor: N. A. Jenkins
- Received July 22, 1997.
- Accepted September 30, 1997.
- Copyright © 1998 by the Genetics Society of America
