The Allele-Specific Suppressor sup-39 Alters Use of Cryptic Splice Sites in Caenorhabditis elegans (original) (raw)

Abstract

Mutations in the Caenorhabditis elegans sup-39 gene cause allele-specific suppression of the uncoordination defect of unc-73(e936). e936 is a point mutation that changes the canonical G at the 5′ end of intron 16 to a U. This mutation activates three splice donors, two of which define introns beginning with the canonical GU. Use of these two cryptic splice sites causes loss of reading frame; interestingly these messages are not substrates for nonsense-mediated decay. The third splice donor, used in 10% of steady-state e936 messages, is the mutated splice donor at the wild-type position, which defines an intron beginning with UU. In the presence of a sup-39 mutation, these same three splice donors are used, but the ratio of messages produced by splicing at these sites changes. The percentage of unc-73(e936) messages containing the wild-type splice junction is increased to 33% with a corresponding increase in the level of UNC-73 protein. This sup-39-induced change was also observed when the e936 mutant intron region was inserted into a heterologous splicing reporter construct transfected into worms. Experiments with splicing reporter constructs showed that the degree of 5′ splice site match to the splicing consensus sequence can strongly influence cryptic splice site choice. We propose that mutant SUP-39 is a new type of informational suppressor that alters the use of weak splice donors.

ANALYSIS of heritable genetic defects in humans has shown that 15% of point mutations responsible for genetic disease disrupt signals involved in splicing (Krawczak et al. 1992). When splice site consensus sequences are mutated, changes in the way the splicing machinery identifies splice junctions occur. These changes usually involve skipping of the exon bounded by the mutation, retention of the intron containing the mutation, or activation of cryptic splice donor or acceptor sites in the region of the mutation (Krawczak et al. 1992). Because splicing to one in three cryptic sites will maintain the correct open reading frame, some cryptic splice sites activated by mutation are much less deleterious to gene function than others. Surprisingly, siblings with an identical mutation that activates cryptic splice sites in the cystic fibrosis gene show differences in the ratios of cryptic splice sites used, resulting in dramatic differences in the severity of disease display (Rave-Harel et al. 1997). In amyotrophic lateral sclerosis (ALS), processing of the EAAT2 glutamate transporter pre-mRNA becomes aberrant in the human brain even though the gene itself is not mutated (Lin et al. 1998). These studies imply that second site mutations, perhaps otherwise silent, can lead to differences in the ability of the splicing machinery to recognize weak or cryptic splice sites. We sought to use a genetically tractable system to identify genes involved in regulation of suboptimal splice site choice. A good candidate system is the nematode Caenorhabditis elegans, a metazoan with a well-developed genetic system, a completely sequenced genome, and intron-exon structure similar to higher eukaryotes with many regulated alternative splicing events (Blumenthal and Steward 1997).

The 32P-induced e936 mutant allele of the C. elegans unc-73 gene is a point mutation that changes the canonical G at the 5′ end of intron 16 to a U (Steven et al. 1998). Figure 1A, adapted from Steven et al. (1998) shows the genomic organization on chromosome I of the messages encoding the two isoforms of UNC-73 protein and the location of the e936 mutation. Injection transformation into unc-73 mutant animals of the region containing the B isoform is sufficient for rescue (Steven et al. 1998). unc-73 encodes a guanine nucleotide exchange factor that interacts with the Rac GTPase and is important for axon guidance during development (Steven et al. 1998). Animals homozygous for unc-73(e936) are uncoordinated with axon guidance and cell migration defects for many different classes of neurons (Steven et al. 1998 and references therein). Because the e936 mutation changes the canonical G at the 5′ end of an intron, it is a reasonable assumption that this mutation will produce a change in the splicing of the unc-73 message.

This e936 splice donor mutation is of interest because mutations in another gene, sup-39, cause dominant allele-specific extragenic suppression of the unc-73(e936) uncoordination phenotype (Run et al. 1996). Two alleles of sup-39, je5 and je6, were identified in a screen of 250,000 mutagenized genomes. These two sup-39 mutant alleles behave identically in allele-specific suppression of e936 and in addition display the same phenotypes in an unc-73(+) background. These phenotypes are a partly penetrant dominant maternal effect embryonic lethality and a variably penetrant morphological phenotype of two rows of oocytes in the gonad, instead of the single row found in wild-type animals (Run et al. 1996). In this article we investigate the intriguing possibility that sup-39 mutations suppress unc-73(e936) in an allele-specific manner by altering the splicing of the unc-73(e936) pre-mRNA. We begin by determining the effects of the e936 mutation on unc-73 splicing and follow with an analysis of the effects of sup-39 mutations on this splicing.

MATERIALS AND METHODS

Growth of C. elegans strains and isolation of RNA: C. elegans strains were grown on plates and in liquid culture using standard methods (Lewis and Fleming 1995). RNA was isolated from washed and packed worm pellets using a protocol developed by Rebecca Burdine and Michael Stern (personal communication). Worm pellets were resuspended in four-pellet volumes of TRIZOL (GIBCO BRL, Gaithersburg, MD) and vortexed vigorously to solubilize the worms. After 10 min of incubation at room temperature, 1.25 ml of solubilized extracts were aliquoted into 1.7-ml microcentrifuge tubes. The tubes were spun at 13,000 × g for 10 min at 4°. Supernatant fractions were removed to new microcentrifuge tubes and 200 μl of CHCl3 was added to each tube. Tubes were vortexed for 15 sec and incubated at room temperature for 3 min. Tubes were then spun at 4° for 15 min at 13,000 × g to separate the phases. The upper aqueous phase was transferred to a new microcentrifuge tube and mixed with 500 μl of isopropanol. After a 10-min incubation at room temperature, the RNA was pelleted by spinning the tubes for 10 min at 13,000 × g at 4°. The pellets were then rinsed with 70% ethanol and air dried. RNA was resuspended in 50 μl of deionized H2O.

Generation of complementary DNAs: Complementary DNAs (cDNAs) for the unc-73 gene were made in 25-μl reaction mixtures. Reaction mixtures contained 5 μg of total RNA, 25 pmol of oligodeoxynucleotide primer complementary to unc-73 exon 17 (5′ ACTTGTCCATCAAAATCTGC 3′), 1 mm each of dATP, dCTP, dGTP, and dTTP, 1 unit of RNA Guard RNase inhibitor (Pharmacia, Piscataway, NJ), 1X AMV reverse transcriptase buffer (Promega, Madison, WI), and 10 units of AMV reverse transcriptase (Promega). Reaction mixtures were incubated at 37° for 1.5 hr and stored at −20°. One microliter of this reaction mixture was added directly to 25 μl PCRs.

Polymerase chain reactions: Oligodeoxynucleotides corresponding to unc-73 exon 13 (5′ ATCAAAGATCTCGAGAGATG 3′), exon 15 (5′ AGAAGTTGTACGGATAAGAC 3′), complementary to exon 16 (5′ GAAACTTCAATGCGTTTAGC 3′), and complementary to exon 17 (5′ ACTTGTCCATCAAAATCTGC 3′) were used in PCR mixtures with Taq DNA polymerase (Fisher Scientific, Pittsburgh) and unc-73 cDNAs. Gel-purified PCR products were ligated into EcoRV-cut and phosphatased pBluescript II KS/+ and transformed into Escherichia coli. For 32P PCRs, the exon 15 primer was 5′ end-labeled at low specific activity with [γ-32P]ATP (New England Nuclear, Boston) by T4 polynucleotide kinase (Amersham, Arlington Heights, IL). PCR products were phenol:CHCl3 (1:1) extracted and ethanol precipitated. Product DNA was resuspended in formamide and loaded onto 0.4-mm-thick 6% poly-acrylamide urea gels in TBE buffer. After electrophoresis, the gels were dried onto filter paper and visualized with a Molecular Dynamics (Sunnyvale, CA) PhosphorImager. Quantitation of relative splice site usage was done using ImageQuant software (Molecular Dynamics).

RNase protection assays: RNase protection assays were performed using the Ambion (Austin, TX) RPA II kit according to the manufacturer's instructions. 32P antisense probes were generated by in vitro transcription from digested plasmids. The ribosomal protein L5 mRNA probe was transcribed by T3 RNA polymerase from BglII-digested plasmid pMK250, which contains the C. elegans ribosomal protein L5 cDNA in pBluescript II/SK vector (a gift from Michael Koelle). The unc-73 antisense probe was transcribed using T7 RNA polymerase from pBluescript II/KS+ containing exons 15 and 16 of the wild-type unc-73 cDNA (described above) cut with HindIII. Two different mixed-stage total RNA isolates from each strain were probed. Initially 10 μg of total RNA was tested and RNA concentrations were adjusted for each strain to normalize the L5 mRNA signal. After digestion and ethanol precipitation, protected RNAs were separated on a 1-mm-thick 15-cm-long 6% urea acrylamide denaturing gel, the gel was dried onto chromatography paper, and bands were visualized with a Molecular Dynamics PhosphorImager. The radioactivity in each band was quantitated and the background radiation for the same position in the yeast RNA control lanes was subtracted. The normalized unc-73 mRNA signal was divided by the normalized L5 mRNA signal to determine the relative level of unc-73 mRNA. Digests were done at low RNase concentrations (1:1000 dilution of RNase A/T1 stock supplied with kit). This enzyme concentration did not distinguish between the different unc-73 mutant cryptic splice forms. At high RNase A/T1 concentrations (1:50 dilution of RNaseA/T1 stock supplied with kit) the +23 splice form could be detected as two different bands of higher mobility indicating digestion at the point where 23 nucleotides of this mRNA form looped out from the probe, while the wt and −1 splice forms were not internally cleaved (data not shown).

unc-73(e936); smg-3(r867) strain construction: The following crosses were performed to generate unc-73(e936) I; smg-3(r867) IV double mutant animals. Males from a smg-3(r867) IV; him-5(e1490) V strain were crossed to unc-73(e936) hermaphrodites. Wild-type F1 progeny were selected. F2 hermaphrodite progeny showing the protruding vulva phenotype were then selected and F3 progeny showing both the unc-73 and smg-3 phenotypes were selected.

In addition to testing for the protruding vulva phenotype as a sign of the presence of smg-3(r867), a molecular test was done to confirm the presence of the smg-3 mutation in these strains. The C. elegans homolog of the pre-mRNA splicing factor SRp30b, EEED8.7, is alternatively spliced and the alternatively spliced form of the message containing a premature termination codon is a natural target for degradation by the smg pathway (Morrison et al. 1997). Mutations in the smg genes stabilize alternative forms of the message and can be detected by reverse transcriptase (RT)-PCR analysis (Morrison et al. 1997). A primer complementary to EEED8.7 exon 4 (5′ AGATCTTGAGCGAGAATATC 3′) was used to reverse transcribe EEED8.7 and served as a 3′ primer in PCRs. A primer consisting of exon 2 sequence (5′ ACGTCGTGATGCTGAGCACG 3′) was used as the 5′ primer in PCRs.

Preparation of protein extracts and immunoblotting: M9 buffer was used to rinse worms from three 150-mm plates into a centrifuge tube. Tubes were then centrifuged at 4000 × g to pellet the worms. The worm pellets were resuspended in an equal volume of SDS protein sample buffer [4% (w/v) SDS, 0.125 m Tris-Cl pH 6.8, 20% (v/v) glycerol, 10% (v/v) 2-mercaptoethanol] with 2.5 mm phenylmethylsulfonylfluoride added. The resuspended worms were immediately disrupted using a cuphorn sonicator probe for 1 min and then heated to 90° for 15 min. A total of 20 μl of each sample was then separated on a 1-mm-thick 10% SDS-PAGE gel and electroblotted onto nitrocellulose. The filter was probed with affinity-purified anti-UNC-73 rabbit polyclonal antibody (Steven et al. 1998), and the protein bands were visualized with a horse-radish peroxidase-tagged secondary antibody and chemiluminescence (Pierce, Rockford, IL).

Generation of GFP/lacZ/unc-73 splicing reporter constructs: The Fire vector pPD96.02 contains the unc-54 promoter driving expression of a green fluorescent protein (GFP)/β galactosidase (lacZ) fusion with a nuclear localization signal. The coding region is interrupted by several synthetic introns to improve gene expression. There is a unique EcoRI site in the plasmid in a lacZ exon 16 codons before the end of the open reading frame. The exon 15/16 region of unc-73 genomic DNA was cloned into this EcoRI site. This region contained the last 117 nucleotides of exon 15, all 203 nucleotides of intron 16, and the first 141 nucleotides of exon 16, maintaining the open reading frame with the terminal 16 codons of lacZ when the wild-type intron 16 splice donor is used. Constructs containing the wild-type, e936 point mutation or various other test sequences were generated. Injection/transformation procedures were used to generate N2 worms containing these constructs as extrachromosomal arrays (Mello and Fire 1995). Transformed animals were identified by GFP expression in body wall muscle cell nuclei.

Messages produced by the extrachromosomal arrays were tested for the presence of different splice sites in both a sup-39(+) and sup-39(je5) background. The following crosses were done to generate sup-39(je5) animals containing the extrachromosomal arrays. unc-73(e936) hermaphrodites were mated with N2 males. Male progeny from this cross were mated with N2 hermaphrodites containing the extrachromosomal array. GFP-positive uncoordinated F2 animals were selected. These e936; GFP hermaphrodites were mated with e936; je5 males. GFP-positive F2 progeny with wild-type movement were selected and those that did not segregate any uncoordinated progeny were judged to be e936 I; je5 II; Ex.

Detection of splice site usage in reporter constructs: Total RNA was isolated from the mixed-stage progeny of 20 GFP-expressing animals and reverse transcribed as described above. A reverse transcription primer specific for the extrachromosomal array (5′ GTTGAAGAGTAATTGGACTTA 3′) or for the native unc-73 gene was used. PCRs with Pfu polymerase were performed for both the native unc-73 gene and the extrachromosomal array using a 3′ primer for exon 16 (5′ CAGAACGTCAAAGTGTCTTGAG 3′). The 32P-labeled 5′ primers were specific for the lacZ gene (5′ CTGGAGCCCGTCAGTATCGGC 3′) or the native unc-73 gene in a region of exon 15 not included in the reporter construct (5′ CTATCGTCTGTTGATTGAGCA 3′). Both 5′ primers hybridized to the specific transcripts the same distance from the splice site at the end of exon 15.

RESULTS

The unc-73(e936) mutation results in the activation of cryptic 5′ splice sites: To test the effect of the e936 splice donor mutation on splicing, RT and PCRs were performed on the unc-73 message. RT-PCRs spanning exons 15 and 16 were performed on total RNA isolated from mixed stage unc-73(+); sup-39(+), unc-73(e936); sup-39(+), and unc-73(e936); sup-39(je5) worms. All three strains gave PCR products of similar size to the 1100-bp wild-type product (Figure 1C, lanes 1–3); no products indicative of skipping of exon 15 or inclusion of intron 16 were observed. This result suggests that cryptic splice sites are activated in unc-73(e936) worms.

To identify the cryptic splice sites used in the e936 mutant, the PCR products shown in Figure 1C were cloned, and multiple isolates from each strain were sequenced. For the wild-type unc-73 gene in the unc-73(+); sup-39(+) strain, the predicted exon 15 splice donor site was used. For the strains carrying the unc-73(e936) mutation alone or in the presence of sup-39(je5), three different 5′ splice sites were identified (Figure 1D). Two of these are cryptic splice sites that define introns beginning with the canonical GU. The first is located 1 nucleotide 5′ of the wild-type 5′ splice site and is referred to as the “−1” 5′ splice site. The second is located 23 nucleotides downstream of the wild-type 5′ splice site and is referred to as the “+23” 5′ splice site. Splicing to either of these sites changes the reading frame to the same nonsense reading frame, which prematurely terminates translation of the protein 12 amino acids into exon 16. The third splice site utilized in both e936 mutant strains was at the correct wild-type exon/intron boundary, demonstrating that an intron beginning with UU is spliced in vivo. This exceptional splice donor, which does not follow the convention of introns beginning with G, is referred to as “wt” because use of this site leads to production of wild-type unc-73 mRNA and protein.

sup-39 mutation alters the relative levels of messages spliced at the different cryptic splice sites: Having identified the cryptic splice sites used and noting that they were the same for unc-73(e936) in the presence or absence of the sup-39 mutation, we next asked whether the relative levels of messages arising from use of these different cryptic splice sites changed in the presence of the suppressor. Multiple RNA preparations from worms grown on plates or in liquid culture were tested for each strain. cDNAs were generated from total RNA from mixed-stage worms using a primer complementary to unc-73 exon 17. PCRs were then performed on these cDNAs using primer pairs in unc-73 exons 15 and 16. The exon 15 primer was 5′ 32P-labeled at a low specific activity. Figure 2A summarizes this approach. PCR products were separated on a 6% polyacrylamide urea sequencing gel.

RT-PCR products representing the three cryptic splice sites could be visualized on the sequencing gel (Figure 2B). For the unc-73(+); sup-39(+) strain, only a single band corresponding to the use of the wild-type exon 15 5′ splice site was observed (lanes 1 and 2). For the unc-73(+); sup-39(je5) strain, only the wild-type intron 16 5′ splice site was detected, indicating that the sup-39 mutation did not affect the splicing of wild-type unc-73 (lane 9). For the unc-73(e936); sup-39(+) (lanes 3–5) and unc-73(e936); sup-39(je5) (lanes 6–8) strains, messages resulting from splicing at the three cryptic splice sites identified by sequencing were detected. To quantitate the relative levels of the three different spliced messages, ImageQuant software was used to analyze the PhosphorImager image shown in Figure 2B. The amount of radioactivity in messages using the +23 5′ splice site was compared to the amount of radioactivity in the −1 and wt 5′ splice sites together (Figure 2C). The −1 and wt 5′ splice site bands were difficult to separate using quantitation rectangles so they were counted together. For other gels where separation was sufficient to allow the determination of the individual

Figure 1.

Determination of the splicing induced by the unc-73(e936) mutation. (A) Genomic organization of the unc-73 gene showing the two isoforms and the location of the e936 mutation. The top is adapted from Steven et al. (1998). The middle is an exploded view of the intron and exon structure of the region of unc-73 containing the e936 mutation. Exon numbers are indicated in the boxes. Exon size in nucleotides is indicated above the boxes and intron size below the lines. The bottom shows the sequence of the wild-type unc-73 gene and the e936 mutation at the 5′ end of intron 16. (B) Location of primers used in RT-PCR assays. Exon sizes and primer locations (arrows) are indicated. Numbers above the arrows correspond to the distance from the 5′ end of the primer to the end of the exon. (C) The unc-73(e936) mutation does not cause exon skipping or intron inclusion. RT-PCRs were performed on total RNA isolated from mixed-stage cultures of the indicated strains. The products from each strain were separated on a 1% agarose gel and visualized with ethidium bromide. (D) Location of cryptic splice donor sites. PCR products shown above were cloned and sequenced. The wild-type unc-73 gene gave a single type of insert corresponding to the wild-type splice junction between exons 15 and 16. The unc-73(e936); sup-39(+) and unc-73(e936); sup-39(je5) strains each yielded clones from three different splice donor sites. Capital letters are exon 15 sequence and lower case letters are the sequence of the 5′ end of intron 16. The three cryptic splice sites are designated −1, wt, and +23.

radioactivity in these two bands located one nucleotide apart, the amount of radioactivity in the −1 and wt splice site products in both the unc-73(e936); sup-39(+) and unc-73(e936); sup-39(je5) animals was roughly equal (data not shown). Because the usage of the wt and −1 splice sites is equal, the wt 5′ splice site represents 10% of the steady-state message in unc-73(e936); sup-39(+) animals and 33% of steady-state message in the unc-73(e936); sup-39(je5) animals.

sup-39 mutations cause 50% embryonic lethality and mixed-stage cultures contain many dead embryos (Run et al. 1996). The apparent alteration in steady-state messages arising from different cryptic splice site usage in the sup-39(je5) strain could reflect a different ratio of these spliced products in dying embryos. To test this possibility, the levels of these messages in mixed-stage populations were compared to those of L3 and L4 larvae (chosen because they were easy to pick and contain no developing embryos) from unc-73(e936) mutants with or without sup-39 mutations (Figure 2D, lanes 2 vs. 4 and lanes 3 vs. 5). The same alteration in the level of steady-state messages representing the different cryptic splice sites by sup-39 mutation in mixed-stage populations is observed for the L3/L4 animals.

Determination of the overall level of unc-73 mRNA between wild-type and e936 strains: The experiments shown in Figure 2 measure the relative amount of steady-state messages spliced at each of the three splice sites used in the presence of the e936 mutation, not actual splice site usage. Splicing at the two cryptic splice sites that change the reading frame and result in the premature termination of translation in exon 16 produces mRNAs that may be substrates for the nonsense-mediated decay pathway of C. elegans (Anderson and Kimble 1997). To determine whether there are differences in the overall level of unc-73 mRNA in the different strains tested, the steady-state level of total unc-73 mRNAs in mixed-stage cultures was determined for N2, e936, and e936; je5 animals by RNase protection assays (Figure 3A). PhosphorImager analysis was used to compare the level of unc-73 message in each sample as a percentage of the signal present for the ribosomal L5 protein mRNA. The averages of the levels of unc-73 message detected in the RNAs from the e936 and e936; je5 strains were both within 20% of the wild-type strain average. One model to account for the changes in the relative abundance of different unc-73 messages seen in the e936; je5 strain relative to the e936 strain (Figure 2C) is

Figure 2.

Quantitation of cryptic splice site usage. (A) Schematic representation of PCR primer locations. The exon 15 primer is 5′ end-labeled with 32P at low specific activity (indicated by asterisk on arrow). Numbers above the arrows show distance from 5′ end of the primer to the wild-type splice sites. (B) Visualization of different cryptic splice product ratios. PCRs with cDNAs generated from distinct reverse transcription reactions of different total RNA preparations were separated on a 6% polyacrylamide/7 m urea sequencing gel and visualized. Lanes 1 and 2, WT. Lanes 3–5, unc-73(e936); sup-39(+). Lanes 6–8, unc-73(e936); sup-39(je5). Lane 9, unc-73(+); sup-39(je5). Size markers and the identities of the different cryptic splice sites are indicated at the left. (C) Quantitation of relative cryptic splice site usage levels. ImageQuant software was used with the PhosphorImager image in B to determine the relative amounts of the different mRNAs. The radioactivities in the −1 and wt bands were measured together and compared to the radioactivity found in the +23 band for each of lanes 3–8. (D) Embryonic lethality caused by sup-39 mutations is not responsible for the differences in the ratios of cryptic splice sites used. Total RNA was isolated from mixed-stage populations or from 100 L3 and L4 stage animals of the strains indicated, RT-PCR reactions similar to those in B were performed, and the products were analyzed.

that je5 could promote selective degradation of the +23 message but not the −1 message. If this model were correct, the e936; je5 strain would possess 30% of the total unc-73 message in the RPA assay relative to the e936 strain. While some differences in the relative amount of unc-73 message between the e936 and e936; je5 strains were observed, these could not account for the differences in the relative amounts of the different messages seen in Figure 2. The fact that roughly equal levels of unc-73 messages were detected for the wild-type and e936 strains is surprising because 90% of the stable e936 messages in unc-73(e936); sup-39(+) animals contain a premature termination codon.

The different forms of the unc-73(e936) message are not substrates for smg regulation: In C. elegans, the seven smg genes (suppressors with morphogenetic effects on genitalia) are required in the pathway for nonsense-mediated decay (Anderson and Kimble 1997). Mutations in these genes stop nonsense-mediated decay of mRNA. unc-73(e936); smg-3(r867) double mutant strains were constructed and tested for the relative levels of stable messages representing usage of the different e936 cryptic splice sites (Figure 3B, lanes 4 and 5). The relative levels of stable e936 messages from the three cryptic splice sites are equivalent between the e936 single mutant and e936; r867 double mutant strains (lanes 2, 4, and 5). This indicates that the messages containing the −1 and +23 cryptic splice sites, which possess premature termination codons, are not preferentially degraded by the smg pathway. Confirmation of the presence of smg-3(r867) in these strains is shown in Figure 3C. This was done using an assay developed by Morrison et al. (1997) in which alternative forms of the C. elegans SRp30b gene, EEED8.7, natural targets for the smg pathway containing premature termination codons, are stabilized in the presence of smg-3 mutations.

sup-39 mutations increase the amount of UNC-73 protein in e936 mutant strains: Use of the wt site in the unc-73(e936); sup-39(+) worms allows a small amount of full-length mRNA to be made that encodes wild-type protein. This is consistent with the demonstration by Steven et al. (1998) that e936 animals produce a low level of full-length UNC-73 protein. unc-73(e936); sup-39(je5)

Figure 3.

Analysis of UNC-73 mRNA and protein levels in different strains. (A) Measurement of the steady-state unc-73 message levels in the wild type, e936; +, and e936; je5 strains. RNase protection assays were performed on four different total RNA isolates from each of the indicated strains and a yeast total RNA control. Approximately 10 μg of total RNA from each isolation was used in each experiment. RNAs were tested for the ribosomal protein L5 mRNA signal and the unc-73 mRNA signal. The signal for each band was measured with a PhosphorImager and the background signal from the region with the same mobility as the digested probe in the yeast control lane was subtracted to give the normalized signal. Then the ratio of the normalized signal for the unc-73 mRNA was divided by the normalized signal for the L5 mRNA for each sample. This is shown as a percentage below each lane. The averages of the four samples for each strain are also shown. (B) The cryptically spliced e936 messages are not substrates for the nonsense-mediated decay pathway. RT-PCR analyses of the cryptic splice sites were performed as in Figure 2B. Total RNAs tested were from N2 (lane 1), unc-73(e936) (lane 2), unc-73(e936); sup-39(je5) (lane 3), and two independent isolates of RNA from unc-73(e936); smg-3(r867) animals (lanes 4 and 5). (C) Confirmation of the presence of smg-3 mutations in the strains. It had previously been reported that smg-3 mutations stabilize alternatively spliced forms of the C. elegans SRp30b (EEED8.7) message containing premature stop codons (Morrison et al. 1997). RT-PCRs with primers specific for the SRp30b gene were done on the same RNA isolations used in Figure 3B. Reaction products were separated on a 1% agarose gel and visualized with ethidium bromide. The presence of two higher molecular weight alternatively spliced forms of SRp30b is found only in the smg-3 mutant strains, confirming the presence of the r867 mutation. These products contain exon 3 but appear to use two different predicted 5′ splice sites at the end of this exon, located 37 nucleotides apart. Arrows show the positions of the PCR primers and the asterisks indicate the location of the stop codons. (D) Mutations in sup-39 increase the amount of full-length UNC-73 protein for unc-73(e936). Equivalent amounts of total protein from the indicated strains were separated on a 10% SDS-PAGE gel and immunoblotted with anti-UNC-73 polyclonal antibody (Steven et al. 1998). The two isoforms of UNC-73 protein are indicated. This polyclonal antibody was also reported to recognize additional specific and nonspecific bands in C. elegans extracts (Steven et al. 1998). These bands, indicated by one and two asterisks, can be used as internal loading standards. The band indicated by three asterisks is a cross-reactive protein found in the OP50 strain of E. coli, which is used as a food source for the worms.

animals have equivalent levels of total unc-73 message to unc73(e936); sup-39(+) animals, yet contain a 3.3-fold increase in the fraction of stable unc-73 mRNA encoding full-length UNC-73 protein. This suggests the possibility that unc-73(e936); sup-39(je5) animals produce more UNC-73 protein than unc-73(e936); sup-39(+) animals and less than wild-type animals. Total protein was extracted from wild-type, unc-73(e936); sup-39(+), and unc-73(e936); sup-39(je5) animals and immunoblotted with a polyclonal antibody against UNC-73 (Steven et al. 1998; Figure 3D). Wild-type worms produced a large amount of UNC-73 proteins (lane 1) relative to the barely detectable level observed in unc-73(e936); sup-39(+) animals (lane 2). unc-73(e936); sup-39(je5) animals produced an amount of UNC-73 protein greater than unc-73(e936) animals and less than wild-type animals (lane 3).

sup-39 affects the relative level of messages produced by cryptic splicing of an in vivo splicing reporter: Because differences in message stability between the e936 spliced isoforms do not appear to account for differences in their relative levels caused by sup-39(je5), perhaps the sup-39 mutation affects the levels of these different spliced isoforms through changes in the splicing of the e936 message. If this is the case, a splicing effect should be seen when this intron is found in the context of another gene. To test this hypothesis, a splicing reporter construct was engineered (Figure 4A). This reporter

Figure 4.

sup-39 regulates usage of the e936 splice donor in a heterologous construct. (A) Schematic representation of the splicing reporter constructs. A total of 117 bases of exon 15, all of intron 16, and 141 bases of exon 16 were introduced into a GFP/lacZ reporter construct under the control of the unc-54 promoter. These were transformed into worms and maintained as extrachromosomal arrays selected for the ability of the animals to express GFP in body wall muscle cell nuclei. Six arrays were produced containing the various point mutations in the area around the exon 15 splice donor. The positions of the primers used for RT-PCR are indicated. (B) 32P RT-PCRs were performed to detect the relative steady-state levels of mRNAs representing exon 15/16 cryptic splice site usage from either the native unc-73 gene (lanes 1–5) or the reporter constructs (lanes 6–10). Total RNA used in RT-PCRs was obtained from the indicated strains as described in materials and methods. Multiple lanes for a strain represent independent RNA isolations from that strain. Bands corresponding to usage of the different cryptic splice sites are indicated. (C) Additional splicing reporters were tested to determine cis-acting sequences responsible for use of UU splice donor. 32P RT-PCRs were performed to detect the relative levels of stable RNAs representing exon 15/16 cryptic splice site usage from the indicated reporter constructs. Multiple lanes for a strain represent independent RNA isolations from that strain. Bands corresponding to usage of the different cryptic splice sites are indicated.

contained 117 nucleotides of unc-73 exon 15, all of intron 16, and the first 141 nucleotides of exon 16 inserted into the terminal exon of a GFP/lacZ fusion construct under control of the unc-54 promoter. Six constructs, summarized in Figure 4A, were transformed into worms and maintained as extrachromosomal arrays. Worms expressing GFP were selected and RT-PCR was performed on them with primer combinations specific for the native unc-73 gene exon 15/16 splice junction (Figure 4B, lanes 1–5) or for the extrachromosomal array splice junction (Figure 4B, lanes 6–12). As can be seen for azEx1, the e936 point mutation in the context of the extrachromosomal array in a sup-39(+) background (lanes 7 and 8) yielded similar ratios of messages containing the cryptic splice sites as the e936 mutation in its native context (lane 2). In a sup-39(je5) background, changes in the ratio of stable messages arising from the different e936 cryptic splice sites in the native gene context (lane 3) are observed in the context of the reporter construct as well (lanes 9 and 10).

Analysis of cis-flanking sequences necessary for use of UU splice donor: Because the use of an intron beginning with UU is highly unusual, we wanted to identify sequences flanking this splice donor involved in promoting usage of this site. Our in vivo splicing reporter fusion construct with the e936 mutation was modifed so that the GU defining the intron beginning at the −1 site was mutated to AU. This construct was transformed into worms, and the extrachromosomal array resulting from this transformation is azEx3. RT-PCR analysis of the stable RNAs arising from azEx3 shows that this mutation appears to eliminate splicing at both the −1 and wt sites (Figure 4C, lane 4). Therefore the G at the −1 position is essential for the use of both the −1 and wt splice sites in e936. As a control, we tested the effect of the −1 G → A mutation on the wild-type intron sequence by testing the splicing of azEx5. This mutation resulted in a slight weakening of use of the wild-type site indicated by a slight activation of splicing to the +23 site in this construct (lanes 5 and 6), while the majority of splicing still occurred at the wt site. This result suggests that, while the G at the −1 position acts to increase the efficiency of splicing to the wt site in the wild-type intron, it is required for splicing to both the −1 and wt splice donors in the e936 intron.

We next sought to determine what would happen if we strengthened the −1 and wt splice sites by improving their match to the 5′ splice site consensus sequence. U is found at the +6 position of 62% of C. elegans introns while G, present at this position in unc-73 exon 15, is found at this position in only 7% of C. elegans introns (Blumenthal and Steward 1997). When the −1 splice site is used, this position becomes the +7 position of the intron. U is found at the +7 position in 54% of C. elegans introns while G is found at this position in only 12% of introns. Therefore changing this G at +6 to a U in the e936 intron would increase the match of both the −1 and wt cryptic splice donor sequences to the C. elegans 5′ splice site consensus sequence (Blumenthal and Steward 1997). An e936 splicing reporter construct with a G to U mutation at the +6 position was transformed into C. elegans and the resulting extrachromosomal array is called azEx4. RT-PCR analysis of the stable RNAs arising from azEx4 shows that this mutation increased the usage of both the −1 and wt sites relative to the +23 site (Figure 4C, lane 7). This is striking because this cis mutation mimics the effects of sup-39 mutation in trans on the levels of stable messages arising from the different cryptic splice sites of the e936 construct (Figure 4C, lane 2 vs. lane 3). We next created a splicing reporter that strengthened the match to the 5′ splice site consensus sequence for the −1 splice site while decreasing the match for the wt site. G is found at the +5 position in 76% of C. elegans introns while U is found at this position only 10% of the time. We constructed a splicing reporter in which the +5 position of the e936 splice donor was mutated from G to U. This construct was transformed into C. elegans and the resulting extrachromosomal array is called azEx6. This mutation resulted in changes to the relative levels of messages arising from use of cryptic splice sites from the azEx6 reporter construct array. The use of the +23 splice site decreased, the use of the wt splice site was eliminated, and the use of the −1 splice site was dramatically increased (Figure 4C, lanes 8 and 9). These results are consistent with a model in which the usage frequency of different cryptic splice sites is determined by the degree to which surrounding sequences match splicing consensus sequences.

DISCUSSION

The e936 point mutation of unc-73 activates three different cryptic splice donors. Two of these cryptic splice sites define introns beginning with the canonical GU dinucleotide, and the third is the wild-type splice site that is used even though it defines an intron beginning with a highly unusual UU dinucleotide. The same three cryptic splice sites in the unc-73(e936) message are also used in the presence of sup-39(je5), although the ratio of steady-state unc-73 messages using these three different splice sites changes. The fraction of steady-state unc-73 mRNA produced by splicing at the correct wild-type splice donor site increases from 10 to 33%. This change cannot be accounted for by any decrease in the total level of unc-73 messages caused by je5, indicating that the total amount of mRNA encoding full-length UNC-73 protein increases in the presence of je5. The increase in the level of wild-type message leads to an increase in the UNC-73 protein level in these animals. sup-39 mutations appear to function in allele-specific suppression of the unc-73(e936) uncoordination phenotype by increasing the steady-state level of wild-type UNC-73 mRNA, allowing for a sufficient level of wild-type UNC-73 protein to be produced so that proper axon guidance and cell migrations can occur.

Our splicing assays using RT-PCR on total RNA tested only the relative amount of steady-state mRNA present for each of the cryptic splice forms of unc-73, not splice site usage directly. Because differential stability of the e936 mRNA isoforms was a possibility, it was important to determine whether any of these different splice products, especially the two with premature termination codons (−1 and +23), would be substrates for the nonsense-mediated decay pathway. smg genes regulate the steady-state ratios of mRNAs arising from alternative splicing of two genes encoding SR proteins (Morrison et al. 1997). We showed that the two messages arising from e936 containing premature stop codons are not substrates for the smg degradation pathway (Figure 3, A and B). smg mutations appear to have less of an effect on the stability of messages with premature termination codons when these are found closer to the end of the gene (Pulak and Anderson 1993). Perhaps the large size of the coding region upstream of the premature termination codon (1380 amino acids) plays a role in the smg independence of the e936 messages. sup-39 suppression of unc-73(e936) appears to be independent of the smg pathway based on the demonstration that smg mutations do not suppress unc-73(e936) uncoordination (Run et al. 1996) nor do they alter the relative steady-state levels of messages arising from e936 in the same way as sup-39 mutants (Figure 3B). This observation supports the model that sup-39 mutations act at the level of splice site choice of e936 pre-mRNA, not at the level of message stability.

Experiments with in vivo splicing reporter constructs demonstrate that sup-39 mutant function is specific for the exon 15/16 region of e936 (Figure 4B, lanes 7–10). sup-39 mutations alter the relative levels of steady-state messages transcribed from the chimeric reporter construct in the same way that they do for messages transcribed from the e936 mutation in the native unc-73 gene. Because the gene context surrounding this splicing region is completely different in the reporter construct and the unc-73 gene yet the usage of the different cryptic splice sites is altered identically by sup-39 mutations, we propose that mutant SUP-39 plays a role in the regulation of splice site choice. A cis mutation that increases the match of the wt and −1 cryptic splice donors of e936 intron 16 to the 5′ splice site consensus sequence increases the usage of these sites relative to the +23 site (Figure 4C, lane 7). This cis mutation, chosen for its ability to strengthen the wt and −1 5′ splice sites, mimics the effect of mutant SUP-39 in trans. This result also suggests that mutant SUP-39 plays a role in splice site selection.

Allele-specific extragenic suppression can generally occur by two main mechanisms. In the first, compensatory mutations in a protein that interacts with a mutant protein can act to reestablish or strengthen protein:protein interactions disrupted by the original mutation (Prelich 1999 and references therein). In the second, informational suppression occurs. Informational suppressors are allele-specific, gene nonspecific suppressors of gene processing defects. Run et al. (1996) originally suggested that sup-39 is an informational suppressor based on its dominant and allele-specific effects. Examples in C. elegans of informational suppressors include tRNA amber suppressors and the smg genes involved in nonsense-mediated decay of messages containing premature stop codons (Hodgkin et al. 1989; Anderson and Kimble 1997). sup-39 may represent a new type of informational suppressor that acts at the level of altering splice site choice.

The observation that a point mutation of the canonical G at the 5′ end of an intron leads to some level of proper splicing is unprecedented in C. elegans. Unusual splice donors, including introns beginning with UU, have been observed before in C. elegans but not as a result of point mutation. Insertion of the Tc1 and Tc4:rh1030 transposable elements into genes in C. elegans can affect splice site choice and lead to use of noncanonical splice donors (Li and Shaw 1993; Rushforth and Anderson 1996). It is unclear whether the mechanism by which these transposable element sequences interfere with splicing is similar to the splicing disruptions caused by point mutations, but these two transposable elements contain inverted repeats at each end, 54 and 297 nucleotides in length, respectively, thus having the potential to form large secondary structures in the introns (Li and Shaw 1993; Plasterk and van Luenen 1997). Work in S. cerevisiae has shown that mutation of the first G of an intron to a noncanonical A results in activation of cryptic splice sites (Newman and Norman 1991). However, intron lariats could be detected that showed correct use of the mutant splice donor in the first catalytic step of splicing (Newman and Norman 1991). These lariats represent dead-end intermediates of splicing and indicate that only the first of the two transesterification reactions catalyzed by the spliceosome had occurred. In C. elegans, mutation of the canonical G nucleotide at the 3′ end of the intron to an A can result in a low level of correct splice acceptor usage along with activation of cryptic splice acceptors (Aroian et al. 1993). The use of the mutant splice acceptor can be increased by removing nearby cryptic AG dinucleotides (Zhang and Blumenthal 1996).

We searched the C. elegans genome for the presence of splice donor sites that match the e936 intron 16 −1 splice donor sequence A/guuagg. Out of 28,637 C. elegans introns confirmed by cDNA sequence, only one 5′ splice site matching this sequence was identified (W. J. Kent and A. M. Zahler, unpublished data). The fact that this is such a weak splice site (leading to 90% splicing at cryptic sites in e936) and the fact that it is so rare in C. elegans suggest that this cryptic phenomenon is not a normal mechanism for regulation of splicing. We have shown that mutation of the +5 position of the e936 intron from G to U dramatically reduced usage of the UU splice donor by making the −1 splice site more closely resemble a consensus splice donor with a U at the +6 position (Figure 4C, lanes 8 and 9). Some combination of unusual or weak consensus sequences at this splice donor appears to be responsible for the equal use of the UU and GU splice donors separated by one nucleotide.

While sup-39 mutations do lead to an increase in the level of messages produced by splicing at an unusual UU splice donor, they do not specifically promote this event. Usage of the −1 and wt cryptic splice sites of e936 is roughly equal in the presence or absence of sup-39 mutations. The sup-39 mutant gene product may function to alter the recognition of the +23 cryptic splice donor relative to the −1 and wt sites, either by promoting assembly of the spliceosome at the wt and −1 splice sites or by inhibiting spliceosome assembly at the +23 splice site. At a subsequent step, the splicing machinery then must distinguish between the −1 and wt splice sites (see Figure 5 for a diagram of this model). The −1 and wt sites contain overlapping regions for interaction with the splicing machinery, specifically for base-pairing with U1 snRNA and U6 snRNA (Ares and Weiser 1994). Mutational analysis of our splicing reporter constructs has shown that recognition of the UU cryptic splice site requires the presence of the G at the −1 position (Figure 4C, lane 4). This indicates that the initial recognition of the overlapping regions for the −1

Figure 5.

Model for a role of sup-39 mutations in splice site selection. SUP-39 mutant gene products may affect the selection of cryptic splice sites by changing the preference of the spliceosomal machinery for assembly at the +23 splice site [preferred 80% of the time in a sup-39(+) background] relative to the −1/wt splice site region of unc-73(e936) intron 16. At a subsequent step, the choice between the −1 and wt splice sites is made and this appears to be independent of sup-39 mutations.

and wt sites requires the presence of a GU dinucleotide, which is present at the start of 98% of 28,637 confirmed C. elegans introns (W. J. Kent and A. M. Zahler, unpublished data). U6 snRNA interacts with the 5′ end of the intron at a step in spliceosome assembly after U1 snRNA base-pairing with this region has been disrupted. These U6 interactions assist in specifying the precise bond to be attacked in the first step of splicing (Kandels-Lewis and Seraphin 1993; Lesser and Guthrie 1993). Perhaps it is at this step that the choice between the −1 and wt splice sites is made.

sup-39 mutations could affect RNA processing at a number of different levels. sup-39 mutations may lead to an increase in the activity of a specific splicing factor, either through increases in production or changes in post-translational modification, which can then change the preference of the splicing machinery for particular splice sites. For example, when the splicing factors SF2/ASF and hnRNP A1 are overexpressed in vertebrate tissue culture cells, changes in alternative splice site usage for a number of genes are observed (Caceres et al. 1994). sup-39 could be a splicing factor, and the suppressor mutations might slightly alter its RNA binding specificity or its ability to interact with other splicing factors, causing differences in the way in which it controls weak splice site choice. An example of this type of mutation is seen in the Drosophila SR protein B52. A dominant point mutation in B52 near its RNA binding domain leads to changes in splicing of a weak splice donor (Peng and Mount 1995). sup-39 might also be an snRNA. Mutations in U1, U2, U5, and U6 snRNA have been shown to affect recognition of cryptic splice sites (reviewed in Ares and Weiser 1994). Attempts at cloning and identification of sup-39 are ongoing in our laboratory. Characterization of its mutant alleles will be helpful in understanding the mechanisms by which cryptic splice sites are chosen. By identifying genes involved in altering cryptic splice site choice in this isogenic system, we hope to begin to understand the types of silent mutations in humans that lead to differences in the use of weak and cryptic splice sites, which can have profound effects on genetic disease progression.

Acknowledgement

We are indebted to Yishi Jin for initially suggesting this project. We thank Yishi Jin and Andrew Chisholm for many helpful discussions, encouragement, technical training, and the use of their laboratory facilities without which this work would not have been possible. Thanks to Mark Roth, Jane Silverthorne, Massimo Caputi, John Tamkun, Ian Chin-Sang, and Mei Zhen for suggestions for working with RNA and protein extracts and for many helpful discussions. Thanks to Jim Kent and Leslie Grate for computer analysis of C. elegans splice sites and to Leslie Holeman for technical assistance. Many thanks to Terry Kubiseski, Rob Steven, T. Pawson, and J. Culotti for the generous gift of anti-UNC-73 antibody. Thanks to Rebecca Burdine and Michael Stern for developing the total RNA isolation protocol and to Michael Koelle for the gift of ribosomal protein L5 cDNA plasmid. Thank you to Yishi Jin and Jim Kent for comments on this manuscript. This work was supported by the following grants to A.M.Z.: grant 1R01GM52848 from the National Institutes of Health (NIH), a Junior Faculty Research Award from the American Cancer Society, and a grant from the University of California Cancer Research Coordinating Committee. The Caenorhabditis Genetic Center, which is funded by the NIH National Center for Research Resources, provided the strains used in this study. PhosphorImager analysis was supported by grant BIR-9318111 from the National Science Foundation.

Footnotes

Communicating editor: R. K. Herman

LITERATURE CITED

Anderson

Kimble

1997

mRNA and translation

, pp.

185

–

208

C. elegans II

, edited by

Riddle

D L

Blumenthal

Meyer

B J

Priess

J R

Cold Spring Harbor Laboratory Press

Plainview, NY

Ares

Jr.,

Weiser

1994

Rearrangement of snRNA structure during assembly and function of the spliceosome

Prog. Nucleic Acids Res. Mol. Biol.

131

–

159

Aroian

R V

Levy

A D

Koga

Ohshima

Kramer

J M

et al. .,

1993

Splicing in Caenorhabditis elegans does not require an AG at the 3′ splice acceptor site

Mol. Cell. Biol.

626

–

637

Blumenthal

Steward

1997

RNA processing and gene structure

, pp.

117

–

145

C. elegans II

, edited by

Riddle

D L

Blumenthal

Meyer

B J

Priess

J R

Cold Spring Harbor Laboratory Press

Plainview, NY

Caceres

J F

Stamm

Helfman

D M

Krainer

A R

1994

Regulation of alternative splicing in vivo by overexpression of antagonistic splicing factors

Science

265

1706

–

1709

Hodgkin

Papp

Pulak

Ambros

Anderson

1989

A new kind of informational suppression in the nematode Caenorhabditis elegans

Genetics

123

301

–

313

Kandels-Lewis

Seraphin

1993

Involvement of U6 snRNA in 5′ splice site selection

Science

262

2035

–

2039

Krawczak

Reiss

Cooper

D N

1992

The mutational spectrum of single base-pair substitutions in mRNA splice junctions of human genes: causes and consequences

Hum. Genet.

–

Lesser

C F

Guthrie

1993

Mutations in U6 snRNA that alter splice site specificity: implications for the active site

Science

262

1982

–

1988

Lewis

J A

Fleming

J T

1995

Basic culture methods

Methods Cell Biol.

–

Shaw

J E

1993

A variant Tc4 transposable element in the nematode C. elegans could encode a novel protein

Nucleic Acids Res.

–

Lin

C L

Bristol

L A

Jin

Dykes-Hoberg

Crawford

et al. .,

1998

Aberrant RNA processing in a neurodegenerative disease: the cause for absent EAAT2, a glutamate transporter, in amyotrophic lateral sclerosis

Neuron

589

–

602

Mello

Fire

1995

DNA transformation

Methods Cell Biol.

451

–

482

Morrison

Harris

K S

Roth

M B

1997

smg mutants affect the expression of alternatively spliced SR protein mRNAs in Caenorhabditis elegans

Proc. Natl. Acad. Sci. USA

9782

–

9785

Newman

Norman

1991

Mutations in yeast U5 snRNA alter the specificity of 5′ splice-site cleavage

Cell

115

–

123

Peng

Mount

S M

1995

Genetic enhancement of RNA-processing defects by a dominant mutation in B52, the Drosophila gene for an SR protein splicing factor

Mol. Cell. Biol.

6273

–

6282

Plasterk

R H A

van Luenen

H G A M

1997

Transposons

, pp.

–

116

C. elegans II

, edited by

Riddle

D L

Blumenthal

Meyer

B J

Priess

J R

Cold Spring Harbor Laboratory Press

Plainview, NY

Prelich

1999

Suppression mechanisms: themes from variations

Trends Genet.

261

–

266

Pulak

Anderson

1993

mRNA surveillance by the Caenorhabditis elegans smg genes

Genes Dev.

1885

–

1897

Rave-Harel

Kerem

Nissim-Rafinia

Madjar

Goshen

et al. .,

1997

The molecular basis of partial penetrance of splicing mutations in cystic fibrosis

Am. J. Hum. Genet.

–

Run

J Q

Steven

Hung

M S

van Weeghel

Culotti

J G

et al. .,

1996

Suppressors of the unc-73 gene of Caenorhabditis elegans

Genetics

143

225

–

236

Rushforth

A M

Anderson

1996

Splicing removes the Caenorhabditis elegans transposon Tc1 from most mutant pre-mRNAs

Mol. Cell. Biol.

422

–

429

Steven

Kubiseski

T J

Zheng

Kulkarni

Mancillas

et al. .,

1998

UNC-73 activates the Rac GTPase and is required for cell and growth cone migrations in C. elegans

Cell

785

–

795

Zhang

Blumenthal

1996

Functional analysis of an intron 3′ splice site in Caenorhabditis elegans

RNA

380

–

388