The Caenorhabditis elegans Gene mfap-1 Encodes a Nuclear Protein That Affects Alternative Splicing

RNA splicing is a major regulatory mechanism for controlling eukaryotic gene expression. By generating various splice isoforms from a single pre–mRNA, alternative splicing plays a key role in promoting the evolving complexity of metazoans. Numerous splicing factors have been identified. However, the in vivo functions of many splicing factors remain to be understood. In vivo studies are essential for understanding the molecular mechanisms of RNA splicing and the biology of numerous RNA splicing-related diseases. We previously isolated a Caenorhabditis elegans mutant defective in an essential gene from a genetic screen for suppressors of the rubberband Unc phenotype of unc-93(e1500) animals. This mutant contains missense mutations in two adjacent codons of the C. elegans microfibrillar-associated protein 1 gene mfap-1. mfap-1(n4564 n5214) suppresses the Unc phenotypes of different rubberband Unc mutants in a pattern similar to that of mutations in the splicing factor genes uaf-1 (the C. elegans U2AF large subunit gene) and sfa-1 (the C. elegans SF1/BBP gene). We used the endogenous gene tos-1 as a reporter for splicing and detected increased intron 1 retention and exon 3 skipping of tos-1 transcripts in mfap-1(n4564 n5214) animals. Using a yeast two-hybrid screen, we isolated splicing factors as potential MFAP-1 interactors. Our studies indicate that C. elegans mfap-1 encodes a splicing factor that can affect alternative splicing.


Introduction
RNA splicing removes non-coding introns and joins adjacent coding exons from pre-mRNAs to generate functional coding mRNAs. Alternative splicing can generate numerous splice isoforms from the same pre-mRNA [1]. The complex proteome encoded by mRNA splice isoforms is believed to be a major driving force for the evolving complexity of metazoans [1,2]. Numerous proteins and non-coding RNAs regulate RNA splicing [3]. The U1 snRNP complex and the SF1/U2AF65/U2AF35 protein complex recognize the 59 and 39 splice sites of an intron, respectively [4][5][6][7][8][9][10][11], and the U2 and U4/U5/U6 snRNP complexes assemble in a step-wise manner and undergo compositional and conformational rearrangements to drive the two steps of the trans-esterification reaction in RNA splicing [7,12]. Mutations in trans-splicing factors or cis-regulatory splicing elements cause numerous diseases [13,14].
Over 200 protein factors have been shown to regulate splicing or associate with the splicing machinery or other splicing factors [3]. Splicing factors have been identified mostly using biochemical approaches, and the in vivo functions of many splicing factors remain largely unknown. In vivo analysis of these factors remains a challenge, since many are essential for viability and the analyses of their in vivo functions can be limited by the lethality caused by mutations in these factors.
Recent studies indicate that conclusions concerning splicing factors derived from in vitro analyses should be complemented by in vivo analyses. For example, in vivo studies suggest that the splicing factor SF1/BBP, once thought to be ubiquitously required for splicing, might be required for the splicing of only a subset of genes [15][16][17]. Some Saccharomyces cerevisiae core splicing factors, once thought to be essential for the splicing of all introns, were found to affect splicing of only a subgroup of introns when the effects of loss-of-function mutations in these factors were examined [18]. Also, the functions of some splicing factors extend beyond regulating RNA splicing. For example, the U2AF large subunit is required for the efficient export of intronless mRNAs in Drosophila in addition to its essential role in regulating 39 splice site recognition [19]. Thus, in vivo functional analyses are essential for an accurate understanding of the biological functions of splicing factors.
In Caenorhabditis elegans, the genes unc-93, sup-9 and sup-10 encode components of a presumptive two-pore domain K+ channel complex that affects muscle activity [20][21][22][23]. Animals carrying rare gain-of-function (gf) mutations in any of these three genes are sluggish, defective in egg laying and exhibit a rubberband phenotype: when touched on the head, the animal contracts and relaxes along its entire body without moving backwards. Complete loss-of-function (lf) mutations of unc-93, sup-9 or sup-10 do not cause obvious abnormalities [21,22]. The SUP-9 protein is similar to the mammalian Two-pore Acid Sensitive K+ channels TASK-1 and TASK-3 [20]. sup-10 encodes a novel single-transmembrane domain protein without identified mammalian homologs [20], and unc-93 encodes a multiple transmembrane-domain protein that defines a novel family of proteins conserved from C. elegans to mammals [20,23]. A mammalian UNC-93 homolog, UNC-93b, interacts with Toll-like receptors and regulates innate immune responses [24][25][26][27][28].
We previously screened for new suppressors of the ''rubberband'' Unc phenotype of unc-93(e1500) animals and isolated mutations affecting the splicing factors U2AF large subunit (UAF-1) and SF1/BBP (SFA-1) [29]. Our analysis suggested that mutations in uaf-1 and sfa-1 result in the suppression of the rubberband Unc phenotype of unc-93(e1500) animals by altering the splicing of the pre-mRNA of an unknown gene [29]. We identified the pre-mRNA of the gene tos-1 as abnormally spliced in uaf-1 and sfa-1 mutants and determined that tos-1 is a sensitive endogenous reporter for analyzing in vivo functions of splicing factors [30]. In this study, we describe a third isolate, n4564 n5214, from the genetic screen in which we identified the uaf-1 and sfa-1 mutations [29]. We found that the gene affected by n4564 n5214 encodes a novel splicing factor that affects alternative splicing in C. elegans.

Results
n4564 n5214 suppresses the rubberband Unc phenotype of unc-93(e1500) animals Previously we performed a genetic screen for mutations that caused sterility and/or lethality and concurrently suppressed the rubberband Unc phenotype caused by the unc-93(e1500) mutation [29]. Besides uaf-1(n4588) and sfa-1(n4562), we also isolated the mutation n4564 [29], which we renamed n4564 n5214 in this study (see below). Although the precise mechanisms underlying the suppression of the rubberband Unc phenotypes by the uaf-1, sfa-1 [29] and n4564 n5214 (see below) mutations remain to be determined, the uaf-1 and sfa-1 mutations we isolated and the splicing of the uaf-1 target tos-1 can provide tools for studying in vivo functions of splicing factor genes [29,30], which we now report include the gene affected by n4564 n5214. n4564 n5214 mutants exhibited temperature-sensitive lethality: at 15uC, n4564 n5214 homozygous animals grew and behaved similarly to the wild type; at 20uC, mutant animals grew more slowly, had few progeny and were hyperactive (Table 1); at 25uC, the mutant strain was embryonically lethal. n4564 n5214 weakly suppressed the Unc phenotype of unc-93(e1500) animals at 15uC (L. Ma and H. R. Horvitz, unpublished observations) and was a stronger suppressor at 20uC (Table 1). n4564 n5214/+; unc-93(e1500) animals were as Unc as unc-93(e1500) animals (L. Ma and H. R. Horvitz, unpublished observations), indicating that n4564 n5214 is a recessive suppressor of the Unc phenotype of unc-93(e1500) animals and suggesting that n4564 n5214 causes a reduction or loss of mfap-1 function. We tested whether n4564 n5214 also suppressed the Unc phenotype caused by the rubberband mutants unc-93(n200), sup-9(n1550) and sup-10(n983). n200 is a weakly semi-dominant unc-93 allele that causes a weak rubberband Unc phenotype [22]. The sup-9(n1550) and sup-10(n983) gain-of-function mutations cause strong and moderate rubberband Unc phenotypes, respectively [20,21,23,31]. n4564 n5214 strongly suppressed sup-10(n983) but did not suppress unc-93(n200) or sup-9(n1550) ( Table 1). Thus, n4564 n5214 suppressed the same rubberband Unc mutants as do the uaf-1(n4588) and sfa-1(n4562) mutations, both of which suppress the rubberband Unc phenotypes of unc-93(e1500) and sup-10(n983) animals but not the rubberband Unc phenotypes of unc-93(n200) or sup-9(n1550) animals [29]. n4564 n5214 contains two missense mutations in the gene mfap-1, which encodes a highly conserved nuclear protein We cloned the gene affected by n4564 n5214 by genetic mapping and transgene rescue experiments ( Figure 1A, 1B; see Materials and Methods). The n4564 n5214 strain contains two missense mutations in the gene F43G9.10, changing the adjacent conserved amino acids Asp 426 and Thr 427 to Val (n5214) and Ala (n4564) in the predicted protein, respectively ( Figure 1C). F43G9.10 encodes the C. elegans ortholog of a highly conserved mammalian protein that has been called ''microfibrillar-associated protein 1'' (41% identity between the C. elegans and human orthologs) ( Figure 1D). We named Author Summary RNA splicing removes intervening intronic sequences from pre-mRNA transcripts and joins adjacent exonic sequences to generate functional messenger RNAs. The in vivo functions of numerous factors that regulate splicing remain to be understood. From a genetic screen for suppressors of the rubberband Unc phenotype caused by the Caenorhabditis elegans unc-93(e1500) mutation, we isolated a mutation that affects a highly conserved essential gene, mfap-1. MFAP-1 is a nuclear protein that is broadly expressed. MFAP-1 can affect the alternative splicing of tos-1, an endogenous reporter gene for splicing, and is required for the altered splicing at a cryptic 39 splice site of tos-1. mfap-1 enhances the effects of the gene uaf-1 (splicing factor U2AF large subunit) in suppressing the rubberband Unc phenotype of unc-93(e1500) animals. Our studies provide in vivo evidence that MFAP-1 functions as a splicing factor. F43G9.10 mfap-1 (microfibrillar-associated protein 1). The chicken ortholog of MFAP-1 was suggested to be an extracellular matrix protein [32]. However, the Drosophila ortholog of MFAP-1 (dMFAP1) interacts with the splicing factor dPrp38, is required for normal expression of c-tubulin and stg/cdc25 mRNAs and has been proposed to act as a splicing factor [33]. We found that the expression of a human MFAP1::GFP fusion protein (Hsmfap-1) in C. elegans body-wall muscles rescued the suppression of unc-93(e1500) by mfap-1(n4564 n5214) ( Figure 1B), indicating that the function of MFAP-1 is conserved from nematodes to humans. We examined the expression pattern of mfap-1 in transgenic animals using a transcriptional fusion reporter construct that drives GFP expression under the control of an approximately 2.5 kb mfap-1 promoter (Figure 2A). We observed strong GFP expression in the intestine, pharynx and vulval muscles (Figure 2A). We also observed GFP expression in the body-wall muscles (Figure 2A), consistent with the finding that body-wall muscle-specific expression of mfap-1 rescued the suppression of the unc-93(e1500) Unc phenotype by mfap-1(n4564 n5214) (Figure 1). Both C. elegans MFAP-1::GFP and human MFAP-1::GFP, when expressed in body-wall muscles, were exclusively localized in nuclei ( Figure 2B and unpublished observations), indicating that MFAP-1 is a nuclear protein.
We obtained a 746 bp mfap-1 deletion allele, tm3456D (kindly provided by S. Mitani, personal communication), which removes the entire first intron and most of the second exon of mfap-1 ( Figure 1C). tm3456D is predicted to encode a truncated protein with a frameshift after amino acid 53, suggesting that tm3456D is likely a null allele of mfap-1. mfap-1(tm3456D)/+animals grew and behaved like the wild type, and mfap-1(tm3456D) homozygous animals arrested developmentally at the L1 or L2 larval stages. tm3456D/n4564 n5214 animals similarly arrested at the L1/L2 larval stages, suggesting that the lethal phenotype of tm3456D homozygotes is caused by the mfap-1(tm3456D) mutation and consistent with the hypothesis that mfap-1(n4564 n5214) causes a reduction or loss of mfap-1 function.
The D426V(n5214) and T427A(n4564) mutations act together to suppress unc-93(e1500) The temperature-sensitive lethal phenotype of mfap-1(n4564 n5214) at 25uC provided an approach to the identification of genes that interact with mfap-1 by seeking suppressors of the temperature-sensitive lethality. We screened about 50,000 haploid genomes and identified two independent suppressors, which we later found to cause the same intragenic change in mfap-1 (see Materials and Methods). Specifically, both suppressors caused a reversion of the T427A (n4564) mutation (see Materials and Methods) to the wild-type codon (ACA) and amino acid (A427T) and retained the D426V (n5214) mutation. mfap-1(n5214) did not obviously suppress the rubberband Unc phenotype of unc-93(e1500) animals at 20uC but did do so at 25uC (Table 1), indicating that mfap-1(n5214) is a temperature-sensitive allele.
We next examined the splicing of tos-1 in animals fed dsRNA targeting mfap-1. We observed an apparent increase of tos-1 isoforms 1 and 2 ( Figure 4A). Increased expression of isoforms 1 and 2 is caused by increased intron 1 retention and exon 3 skipping in tos-1 splicing, as was seen for mfap-1(n4564 n5214) mutated in mfap-1(n4564 n5214) mutants. Amino acids conserved in at least three orthologs are darkly shaded, while amino acids with similar physical properties or conserved in two orthologs are lightly shaded. doi:10.1371/journal.pgen.1002827.g001  (Figure 3). The similarity of mfap-1(n4564 n5214) and mfap-1(RNAi) in affecting tos-1 splicing further suggests that n4564 n5214 causes a reduction or loss of mfap-1 function. Reducing mfap-1 expression by RNAi feeding did not cause the recognition of the cryptic 39 splice site in intron 1 ( Figure 4B). Similarly, that cryptic splice site was not recognized in mfap-1(n4564 n5214) animals ( Figure 3D).

MFAP-1 interacts physically with known splicing factors in a yeast two-hybrid experiment
To further understand the function of MFAP-1, we sought to identify interacting proteins of MFAP-1 using a yeast two-hybrid screen (see Materials and Methods). We isolated and retested nine genes encoding proteins that might potentially interact with MFAP-1 (Table S1). Two of the genes, K04G7.11 and D1054.14, encode the C. elegans orthologs of the presumptive splicing factors SYF2 [34] and Prp38 [35]. Drosophila MFAP-1 (dMFAP1) was found to be in a protein complex with dPrp38 in co-immunoprecipitation experiments and to interact with dPrp38 physically [33]. SYF2 mutations were found to cause synthetic lethality with mutations in splicing factors clf1Delta2 and prp17/cdc40Delta in S. cerevisiae [34], and mammalian orthologs of SYF2 were identified in various spliceosomal complexes by proteomic approaches [3], suggesting that SYF2 acts as a splicing factor. We also identified MFAP-1 as an interactor in the two-hybrid screen, indicating that MFAP-1 might form homodimers. The other potential interactors are two proteins involved in rRNA processing (C05C8.2 and T22H9.1), a ribosomal protein (RPS-6), a Ras-associated PH-domain containing protein (MIG-10), a Rab11 familyinteracting protein 2 (Y39F10B.1) and a protein of unknown function (F43C11.9). Although the interactions between MFAP-1 and the proteins from the yeast two-hybrid screen remain to be verified by other approaches, the observation that two presumptive splicing factors (Prp38 and SYF2) are among the candidates is consistent with our findings and also those of Andersen and Tapon [33] indicating that MFAP-1 might function as a splicing factor.
We next examined the splicing of tos-1 in animals fed dsRNAs targeting mfap-1, D1054.14 and K07G7.11 ( Figure S4). We found that D1054.14(RNAi) caused an increase of tos-1 splice isoform 2 similar to that seen in mfap-1(RNAi) animals ( Figure S4). No apparent alteration in tos-1 splicing was detected in K07G7. 11(RNAi) animals. This result is consistent with previous studies indicating that homologs of D1054.14 can function as splicing factors [33,35] and identified tos-1 as a target of D1054.14 in C. elegans. Table 2. mfap-1(n4564) and mfap-1(n5214) are weak mutations that act together to suppress the rubberband Unc phenotype of unc-93(e1500) animals.  Discussion MFAP-1 is essential for animal development MFAP1 was initially identified as a putative extracellular matrix protein based on a screen of an embryonic chicken cDNA expression library using antibodies against elastic fiber microfibrils-enriched bovine ocular zonule proteins [32]. Our analysis of C. elegans mfap-1 and the study of the Drosophila ortholog dMFAP1 [33] indicate that MFAP-1 orthologs in these two species are nuclear proteins that affect RNA splicing. In addition, the human ortholog of MFAP-1 was found to be associated with spliceosomes by mass spectrometry analysis [36][37][38]. Since the human ortholog of mfap-1 can rescue the activity of C. elegans mfap-1(n4564 n5214) for suppressing the Unc phenotype of unc-93(e1500) animals, the function of MFAP-1 is likely conserved from nematodes to humans.
Drosophila dMFAP1 acts with several other splicing factors in G2/M progression during mitosis and affects the ratio of pre-mRNA to mature mRNA of the c-tubulin gene and the mRNA level of the stg/cdc25 gene [33]. It is not known if dMFAP1 affects alternative splicing [33]. C. elegans mfap-1(n4564 n5214) mutants exhibit a temperature-sensitive phenotype with normal growth at 15uC and embryonic lethality at 25uC; mfap-1(tm3456D) homozygous animals arrest at L1 to L2 larval stages at 20uC, and mfap-1 RNAi-treated animals arrest at variable larval stages at 20uC (L. Ma and H. R. Horvitz, unpublished observations). These observations indicate that C. elegans mfap-1 is essential for animal development. Whether the developmental defect of mfap-1deficient animals is caused by mitotic defects remains to be determined.

MFAP-1 is probably required for the recognition of weak 39 splice sites
We isolated mfap-1(n4564 n5214) from the same genetic screen in which we identified mutations affecting uaf-1 and sfa-1 [29]. mfap-1(n4564 n5214) suppressed the phenotypes of different rubberband Unc mutants in patterns similar to those of uaf-1(n4588) and sfa-1(n4562) mutations. The findings led us to hypothesize that MFAP-1 might function as a splicing factor. Our analysis of tos-1 splicing in mfap-1 mutant animals and in animals treated with mfap-1 RNAi provided molecular evidence that mfap-1 can affect alternative splicing. mfap-1(n4564 n5214) and mfap-1(RNAi) both increased tos-1 intron 1 retention (with a weak 39 splice site) and tos-1 exon 3 skipping (with a different weak 39 splice site), and neither affected the splicing of intron 3 (with a strong consensus 39 splice site TTTTCAG). These observations suggest that MFAP-1 might be required for the recognition of weak 39 splice sites and not for the recognition of strong 39 splice sites. We propose that MFAP-1 can act much like UAF-1 and SFA-1, mutations in which cause reduced recognition of the weak 39 splice sites in introns 1 and 2 and do not affect the recognition of the consensus 39 splice site in intron 3 of tos-1 [30]. Similar to sfa-1(n4562), mfap-1(n4564 n5214) or mfap-1(RNAi) did not cause recognition of the cryptic 39 splice site in tos-1 intron 1, suggesting that MFAP-1 probably is not essential for determining the specificity of 39 splice sites.
The effects of mfap-1(n4564 n5214) and mfap-1(RNAi) on tos-1 splicing are qualitatively similar but quantitatively different. We also observed differences in the effects on tos-1 splicing by genetic mutations and RNAi treatments in our previous studies of uaf-1 and sfa-1, in which we found that uaf-1(n4588) or sfa-1(n4562) caused more dramatic alterations of tos-1 splicing than did either uaf-1(RNAi) or sfa-1(RNAi). This difference was at least partially caused by an altered function of UAF-1 in uaf-1(n4588) animals [30]. Thus, differing effects on target gene splicing by RNAitreatment and genetic mutation of a splicing factor gene could be caused by several factors, including the differing level of the splicing factor in RNAi-treated and mutant animals and the effects of mutations on the function of the splicing factor.

MFAP-1 might interact with multiple partners to affect RNA splicing
From a yeast two-hybrid screen, we identified nine candidate MFAP-1-interacting proteins. One, D1054.14, is homologous to the S. cerevisiae splicing factor Prp38 [35]. Another, K04G7.11, is orthologous to the S. cerevisiae candidate splicing factor SYF2 [34]. Drosophila splicing factors dPrp38 and dSYF1 were found in a protein complex with dMFAP1, and dPrp38 interacts with dMFAP1 in GST-pulldown experiments [33]. We did not identify orthologs of SYF1 or several other splicing factors found in the Drosophila dMFAP1 complex, possibly because the co-immunoprecipitation experiments could have identified Drosophila proteins that did not directly interact with dMFAP1, whereas our yeast two-hybrid screen could identify only proteins that directly interact with MFAP-1. Both studies suggest that MFAP-1 interacts with multiple splicing factors directly or indirectly and hence that MFAP-1 might play multiple roles in affecting RNA splicing.
We found that reducing the expression of D1054.14 by RNAi caused alterations in tos-1 splicing similar to those caused by mfap-1(RNAi) (Figure S4), suggesting that MFAP-1 and D1054.14/ PRP38 might interact to regulate the splicing of the same set of genes. We did not identify alterations in tos-1 splicing in animals fed dsRNA targeting K04G7.11, indicating that K04G7.11 might not be required for the splicing of tos-1.
The uaf-1(n4588) mutation likely causes the recognition of the cryptic 39 splice site of tos-1 intron 1 by altering rather than decreasing the function of the UAF-1 protein [30]. uaf-1(n4588 n5127) is a weaker mutation than uaf-1(n4588), and the recognition of the cryptic 39 splice site in uaf-1(n4588 n5127) animals but not in uaf-1(n4588) animals was reduced by the presence of the mfap-1(n4564 n5214) and mfap-1(n5214) mutations ( Figure S3C). These results suggest that the recognition of the cryptic 39 splice site of tos-1 requires MFAP-1 only if that recognition is already weak as in uaf-1(n4588 n5127) mutants. These results further support our conclusion that MFAP-1 affects alternative splicing.
In short, we propose that MFAP-1 can act as a splicing factor. Future studies should reveal how MFAP-1 interacts with other splicing factors to affect C. elegans development by regulating splicing of its target genes.
Screen for suppressors of mfap-1(n4564 n5214) Synchronized L4 mfap-1(n4564 n5214) animals (P 0 ) grown at 15uC were mutagenized with ethyl methanesulfonate (EMS) as described [39] and grown to gravid adults at 15uC. The P 0 adults were bleached, and their F 1 progeny were grown to young adults at 15uC and moved to 25uC. After three days at 25uC, the plates were examined for surviving fertile animals. From about 50,000 haploid genomes (F 2 ) screened, two independent suppressed strains were isolated. As animals with a mfap-1(n4564 n5214) phenotype (partial sterility at 20uC and temperature-sensitive lethality at 25uC) were not identified from about 1800 progeny of mfap-1(n4564 n5214) sup/ ++animals, we concluded that the two suppressors were either intragenic or extragenic and very closely linked to mfap-1. DNA sequence determination identified in both suppressors the same nucleotide change (GCA-to-ACA at codon 427), which converted the mutated amino acid alanine in the mfap-1(n4564 n5214) mutant to the wild-type threonine (A427T) while retaining the D426V mutation. One of the two isolates was used for subsequent studies, and the single mutation it carried was designated mfap-1(n5214).

Yeast two-hybrid screen
We used a yeast two-hybrid screen [43,44] to identify proteins that might interact with MFAP-1. A pACT2.2 C. elegans yeast twohybrid library (Addgene plasmid 11523, provided by Dr. Guy Caldwell) was used to screen for MFAP-1 interactors. For the bait, a full-length mfap-1 cDNA was subcloned into the pGBK T7 plasmid (Clontech) and then transfected into the S. cerevisiae strain PJ69-4A. pGBK T7-mfap-1-positive yeast cells were transfected with 2 mg of DNA from the pACT2.2 yeast two-hybrid library, and cells were grown on a SD-Leu-Trp-His medium with 5 mM 3-AT. Positive clones were picked and grown in separate cultures using the same SD medium. Plasmids were obtained from each clone by the smash-and-grab method [45]. Purified plasmids were transfected into the bacterial strain DH5a and grown on LBampicillin plates. Plasmids were purified from the bacterial cultures, and sequences of the inserts were determined. Using yeast two-hybrid assays, we confirmed that these clones did not cause survival of yeast cells transfected with the pGBK T7 empty vector on SD-Leu-Trp-His medium, suggesting that proteins encoded by these clones potentially interact with MFAP-1 but not with the GAL4 DBD domain expressed by the pGBK T7 vector.

RNA interference
Young adult animals were fed HT115 (DE3) bacteria containing plasmids directing the expression of dsRNAs targeting uaf-1, sfa-1, mfap-1, D1054.14 or K04G7.11 on NGM plates with 1 mM IPTG and 0.1 mg/ml Ampicillin [46]. F 1 progeny of these animals, which arrested at various developmental stages, were washed from plates, rinsed with H 2 O and resuspended in Trizol (Invitrogen) for preparation of total RNA. The DNA construct that expresses dsRNA targeting uaf-1 and the bacterial strain that expresses dsRNA targeting sfa-1 were described previously [29]. We obtained a bacterial strain expressing dsRNA targeting mfap-1 from a whole-genome RNAi library [47] and bacterial strains expressing dsRNAs targeting mfap-1, D1054.14 or K04G7.11 from an ORFeome-based RNAi library [48]. The sequences of plasmids from single colonies were determined to confirm the presence of coding sequences for each gene.

Body-bend assay
L4 animals were picked 16-24 hrs before being tested. One day later, young adults were individually picked to plates with OP50 bacteria, and body-bends were counted for 30 sec using a dissecting microscope as described [49].

Molecular biology
Total RNA was purified with Trizol (Invitrogen), and cDNA was generated following the protocol provided with the Superscript II kit (Invitrogen). PCR was performed with Eppendorf Cyclers, and DNA products were resolved using agarose gels. NIH ImageJ software was used to quantify tos-1 splice isoforms [30]. The percentages of tos-1 intron 1 retention, exon 3 skipping and the recognition of the intron 1 cryptic 39 splice were analyzed as described [30]. We performed RT-PCR experiments with animals at different developmental stages and found no indication that tos-1 splicing is regulated developmentally ( Figure S5). DNA sequence analysis was performed using an ABI Prism 3100 Genetic Analyzer and an ABI 3730XL DNA Analyzer.

Transgene experiments
Microinjection of DNA into the syncytial gonad and the generation of animals with germline transmission of transgenes were performed as described [50] with mfap-1(n4564 n5214) animals grown at 15uC. DNA injection mixtures generally contained 20 mg/ml 1 kb DNA ladder, 20 mg/ml Arabidopsis genomic DNA and 20 mg/ml of the transgene of interest. When the transgene did not express a GFP fusion protein, 20 mg/ml pPD95.86-GFP plasmid (which expresses GFP in body-wall muscles) was added to the injection mixture as a visible fluorescence marker. Figure S1 mfap-1(n4564 n5214) does not obviously affect the altered splicing of unc-93(e1500) exon 9. Real-time RT-PCR was performed as described [29] to quantify the recognition of the cryptic 39 splice site of unc-93(e1500) exon 9. No apparent difference between unc-93(e1500) and mfap-1(n4564 n5214); unc-93(e1500) animals was detected. *: results reported previously by Ma and Horvitz (2009) [29]. (TIF) compared to all isoforms spliced at either the endogenous 39 splice site or the cryptic 39 splice site. For all analyses, isoform intensities were obtained by analyzing biological duplicates or triplicates using NIH ImageJ software. N.S., no significant difference. (TIF) Figure S4 Reducing the expression of mfap-1 or D1054.14 by RNA interference altered the splicing of tos-1. RT-PCR experiments showing the effects of reducing the expression of mfap-1, D1054.14 or K04G7.11 by RNAi feeding on tos-1 alternative splicing. mfap-1(RNAi) and D1054.14(RNAi) caused similar alterations in tos-1 splicing, while K04G7.11(RNAi) did not obviously affect the splicing of tos-1. RNAi bacterial strains were obtained from an ORFeome-based RNAi library [48]. (TIF) Figure S5 The splicing of tos-1 is not regulated developmentally. RT-PCR experiments examining the splicing of tos-1 in animals at different synchronized developmental stages. tos-1 splicing was similar in all developmental stages examined. YA: young adult animals 24 hours after the L4 larval stage. (TIF)