The RNA-binding protein SUP-12 controls muscle-specific splicing of the ADF/cofilin pre-mRNA in C. elegans

Pre-messenger RNA splicing is mediated by spliceosomes that contain small ribonucleoprotein particles and splicing factors (Jurica and Moore, 2003; Nilsen, 2003). Pre-mRNAs are often alternatively spliced to give rise to multiple mRNA species, which requires additional regulation of splice-site selection (Smith and Valcarcel, 2000; Maniatis and Tasic, 2002; Black, 2003). Tissue-specific alternative mRNA splicing is essential for expression of functionally different gene products from a single gene, and, thus, it is an important mechanism of cellular differentiation. Disruption of the splicing machinery is known to cause various human diseases (Nissim-Rafinia and Kerem, 2002; Stoilov et al., 2002; Faustino and Cooper, 2003). In particular, myotonic dystrophy has been linked to perturbation of the mechanism of alternative splicing in muscle (Kanadia et al., 2003; Ho et al., 2004). However, the molecular mechanism by which tissue-specific alternative splicing is regulated remains largely unknown, except for a few well-characterized examples (Smith and Valcarcel, 2000; Maniatis and Tasic, 2002; Black, 2003).

Actin depolymerizing factor (ADF)/cofilin proteins enhance actin filament dynamics by severing filaments and accelerating monomer dissociation from the pointed ends of the filaments (Bamburg, 1999; Bamburg et al., 1999; Maciver and Hussey, 2002; Ono, 2003). Essential functions of ADF/cofilin in in vivo actin dynamics and cell viability have been demonstrated in several organisms (McKim et al., 1994; Gunsalus et al., 1995; Lappalainen and Drubin, 1997). In mammals, three ADF/cofilin isoforms are encoded by separate genes and expressed in different patterns of tissue distribution (Matsuzaki et al., 1988; Moriyama et al., 1990; Ono et al., 1994; Gillett et al., 1996; Thirion et al., 2001; Vartiainen et al., 2002). However, in the nematode Caenorhabditis elegans, the unc-60 gene undergoes alternative splicing and expresses two ADF/cofilin isoforms, UNC-60A and UNC-60B (McKim et al., 1994). Our previous studies have indicated that the two ADF/cofilin isoforms have different activities: UNC-60A strongly depolymerizes filaments, whereas UNC-60B binds to filaments with only weak depolymerizing activity (Ono and Benian, 1998; Ono, 1999; Mohri and Ono, 2003). More importantly, they are expressed in different tissues and required for specific actin-dependent processes: UNC-60A is expressed in nonmuscle cells and is required for embryonic cytokinesis (Ono et al., 2003), whereas UNC-60B is specifically expressed in the body wall muscle and regulates myofibril assembly (Ono et al., 1999). The unc-60 gene has nine exons and only the first exon is shared by unc-60A and unc-60B (McKim et al., 1994). Therefore, the tissue-specific expression of unc-60A or unc-60B is proposed to be determined by selection of the first splice acceptor site at the 5′-end of either exon 2A or 2B (McKim et al., 1994). However, the regulatory mechanism of tissue-specific splicing of the unc-60 pre-mRNA is unknown.

In this work, we report identification and characterization of a putative splicing factor that regulates muscle-specific splicing of the unc-60 pre-mRNA in C. elegans. We cloned SUP-12, a conserved RNA-binding protein, as a suppressor of unc-60B. sup-12 mutations strongly suppress the muscle defects of unc-60B mutants. This suppression is likely due to alteration in expression of the unc-60 splice variants in the muscle cells. SUP-12 localizes to the nuclei in body wall muscle and its RNA-binding domain directly binds to the unc-60 pre-mRNA in vitro. Our data support that SUP-12 is a novel member of tissue-specific regulators of alternative splicing.

In a previous effort to identify genes that functionally interact with unc-60, sup-12 mutant alleles st89 and st203 were isolated from a screen for extragenic suppressors of unc-60 mutants (G.R. Francis and R.H. Waterston, personal communication; Waterston, 1988). sup-12 mutations alone have only minor effects on motility of the worms (Fig. 1 A). However, in the unc-60B mutant backgrounds, sup-12 strongly suppressed the motility defects (Fig. 1 A). The suppression by sup-12 was also equally strong for a strong loss of function allele unc-60B(e677) and a null allele unc-60B(su158) (Fig. 1 A), as well as other unc-60B loss of function alleles (McKim et al., 1988), and the motility of the unc-60B;sup-12 double mutants was restored nearly to the level of wild-type worms (Fig. 1 A). The motility defects of the unc-60B mutants are caused by disorganization of actin filaments in the body wall muscle (Ono et al., 1999; Ono et al., 2003). Actin was predominantly found in the large aggregates but not in the myofibrils (Fig. 1 B). However, in the unc-60B;sup-12 double mutants, actin was organized into the myofibrils as well as wild type (Fig. 1 B). The sup-12 single mutants had no detectable phenotype in the myofibril organization (Fig. 1 B; Francis and Waterston, 1985). sup-12(st203) exhibited slightly stronger suppressor effect than sup-12(st89) in worm motility (Fig. 1 A) and in actin organization in muscle (unpublished data). These results suggest that sup-12 genetically interacts with unc-60B and regulates actin organization in the body wall muscle.

By positional cloning and identification of mutation sites, we cloned the sup-12 gene and identified it as T22B2.4 (GenBank/EMBL/DDBJ accession no. NM_076273) that is mapped to the left arm of the X chromosome by the C. elegans Sequencing Consortium (1998). The SUP-12 protein (248 aa) is a putative RNA-binding protein that has a single RNA recognition motif (RRM) domain (Shamoo et al., 1995) in its NH₂ terminus (residues 35–117) and alanine- and glutamine-rich (A/Q-rich) sequence in its COOH terminus (26 A and 24 Q in residues 121–246; Fig. 2 A). Missense mutations G59R, G77E, and G113E were found within the RRM domain in the sup-12 mutant alleles s2900, st89, and s2896, respectively (Fig. 2 A). s2901 has the same mutation as st89, although they were isolated in different laboratories. A mutation (G to A) was found in st203 at the splice donor site in the second intron. RT-PCR analysis showed that aberrantly spliced sup-12 mRNA was predominantly expressed in the sup-12(st203) mutants (unpublished data). RNA interference of sup-12 phenocopied the suppressor phenotype of sup-12 mutants and suppressed the Unc-60B phenotype of multiple unc-60B alleles (unpublished data). Homology searches revealed that SUP-12 is an orthologue of a human protein SEB-4 (Fig. 2 B). Two human SEB-4 isoforms (GenBank/EMBL/DDBJ accession no. AK095016 and NM_153020) are found in the database, but their function is unknown. In Xenopus, two SEB-4 isoforms, XSEB-4 and XSEB-4R, are reported: XSEB-4 is expressed in muscle precursor cells (Fetka et al., 2000), whereas XSEB-4R is strongly expressed in neuronal cells and involved in neural differentiation (Boy et al., 2004). Interestingly, Arabidopsis has at least six SEB-4 orthologues (Fig. 2 B), but functional studies on these proteins are not reported. However, no SEB-4 orthologues were found in yeasts and Drosophila. Sequence alignment of these proteins showed highly conserved sequences in the RRM domain in the NH₂ termini (unpublished data). The COOH-terminal halves are not highly conserved, yet the A/Q-rich sequences are present in all these proteins. The similarity in the sequences suggests that SUP-12 and SEB-4 homologues belong to a new family of functionally conserved RNA-binding proteins.

To understand the mechanism of suppression of the Unc-60 mutant phenotype by sup-12, expression patterns of the splice variants of the unc-60 gene products were examined. Surprisingly, the protein level of UNC-60B, the muscle-specific isoform, was greatly reduced in the sup-12 mutants both in wild-type and unc-60B(e677) backgrounds (Fig. 3 B, lanes 1–6). In contrast, the total level of UNC-60A, the nonmuscle isoform, was not significantly altered by mutations in sup-12 or unc-60B (Fig. 3 B). Similar changes were detected at the mRNA levels: the unc-60B mRNA was reduced as compared with wild type, whereas the unc-60A mRNA was not significantly altered (Fig. 3 C). In wild-type background, the unc-60B mRNA was decreased in the sup-12 mutants (Fig. 3 C, lanes 1–3). unc-60B(e677) has a missense mutation (Ono et al., 1999) and had a greatly reduced level of the unc-60B mRNA (Fig. 3 C, lanes 4–6), which correlates with the reduced protein level (Fig. 3 B, lane 4). unc-60B(su158) has a 600-bp deletion in the unc-60B region without affecting unc-60A (Ono et al., 2003) and expressed a shorter unc-60B mRNA (Fig. 3 C, lane 7) that is not translated into a protein (Fig. 3 B, lane 7). Nonetheless, its mRNA level was still reduced in the sup-12 mutants (Fig. 3 C, lanes 8 and 9), strongly suggesting that SUP-12 affects the levels of the unc-60B mRNA, but not directly of the protein.

UNC-60A is expressed in a variety of nonmuscle cells (Ono et al., 2003). Therefore, changes in the level of UNC-60A in a subset of tissues may not be detected by the Northern blot analysis. Therefore, we examined the tissue distribution of UNC-60A by immunofluorescence microscopy and found that the level of UNC-60A in body wall muscle is altered by sup-12. In wild type and the unc-60B(e677) single mutant, UNC-60A was not detectable in body wall muscle (Fig. 4, A–F, arrows). However, in the unc-60B;sup-12 double mutant and the sup-12 single mutant, UNC-60A protein was detected in the diffuse cytoplasm (Fig. 4, G–L, arrows). These results indicate that the sup-12 mutations have opposite effects on the levels of the two splice variants, unc-60A and unc-60B, in muscle cells. Although UNC-60A and UNC-60B have quantitatively different biochemical activities, both isoforms can enhance actin filament dynamics by depolymerizing actin filaments (Ono and Benian, 1998; Ono, 1999; Ono et al., 1999). Thus, the suppression of the Unc-60 phenotype by sup-12 could be explained by the up-regulation of UNC-60A in muscle, which may compensate for the function of UNC-60B.

We next investigated expression and subcellular localization of SUP-12. An anti–SUP-12 antibody against its COOH terminus specifically recognized the SUP-12 protein by Western blot but did not detect the protein by immunofluorescence microscopy (unpublished data). The 3.1-kb promoter region of the sup-12 gene was able to drive expression of a reporter GFP in body wall muscle (Fig. 5 A, arrowheads) and pharynx (Fig. 5 A, asterisk). GFP-tagged SUP-12, which was expressed in the body wall muscle under the control of the myo-3 promoter (Okkema et al., 1993), predominantly localized to the nuclei (Fig. 5, B–D). Within the nuclei, GFP-SUP-12 was found in diffuse and often speckled patterns in the nucleoplasm but excluded from the globular region, possibly representing the nucleolus (Fig. 5 B). The speckled localization of SUP-12 in the nucleus is similar to patterns that are commonly observed for other splicing factors (Misteli, 2000; Dundr and Misteli, 2001; Lamond and Spector, 2003). Although the myo-3 promoter is active in the body wall muscle but not in the pharynx, myo-3–driven expression of GFP-SUP-12 was sufficient to rescue the Sup-12 mutant phenotype (Table I), indicating that GFP-SUP-12 is functional and expression of SUP-12 in the body wall muscle is functionally important for its interaction with unc-60. These results strongly suggest that SUP-12 is a muscle-specific regulator of pre-mRNA splicing in the nucleus.

Neither the NH₂-terminal RRM domain nor the COOH-terminal A/Q-rich domain alone was able to rescue the sup-12 mutant phenotype (Table I). The RRM domain of SUP-12 (residues 1–117) localized to both nuclei and cytoplasm in a diffuse pattern (Fig. 5 E), whereas the A/Q-rich domain of SUP-12 (residues 118–248) strongly localized to the nuclei in a similar speckled pattern to the full-length protein and weakly to the cytoplasm (Fig. 5 F). These results suggest that either the RRM or the A/Q-rich domain is sufficient for nuclear localization, but both domains are required for the function of SUP-12.

To determine whether the SUP-12 protein may be directly involved in pre-mRNA splicing of unc-60, we examined direct interaction between SUP-12 and unc-60 pre-mRNA in vitro by an electophoretic mobility shift assay (EMSA). Recombinant GST-tagged full-length SUP-12 protein or the COOH-terminal portion (residues 118–248) of SUP-12 were poorly soluble (unpublished data) and therefore were not examined. However, the NH₂-terminal portion (residues 1–117) of SUP-12 containing the RRM domain was stable and soluble as a GST-fusion protein. When purified GST-SUP-12 (RRM) was incubated with various portions of in vitro transcribed unc-60 pre-mRNA (Fig. 6 A), it caused a band shift of only the 978-nt RNA fragment that encompasses the sequence of exon 1 to a portion of exon 5A (Fig. 6, A and B). GST alone did not cause a band shift of the RNAs (Fig. 6 B). These results indicate that the RRM domain of SUP-12 is sufficient for direct RNA-binding and so could confer specificity for the 5′-region of the unc-60 pre-mRNA.

The SUP-12 binding region in the unc-60 pre-mRNA was further narrowed down by a pull-down assay with biotin-labeled RNA fragments (Fig. 6, C–G). We first tested interactions between GST-SUP-12 (RRM) and four RNA fragments, A1-1, A1-2, A1-3, and A1-4, within the unc-60A region (Fig. 6 C). GST-SUP-12 (RRM) bound to the magnetic beads in the presence of A1-1 or A1-2 (Fig. 6 E, lanes 2–5). However, in the presence of A1-3 or A1-4 (Fig. 6 E, lanes 6–9), the amounts of captured GST-SUP-12 (RRM) were not significantly different from that in the absence of RNA (Fig. 6 E, lane 1). GST alone did not bind to the beads in the presence or absence of RNA (Fig. 6 E), indicating that the RRM domain of SUP-12 mediated the interactions with RNAs.

We then used shorter RNA fragments of 10–152 bases in the pull-down assay with GST-SUP-12 (RRM) (Fig. 6, C and D) and characterized the interactions in a quantitative manner (Fig. 6, F and G). GST-SUP-12 (RRM) showed relatively strong binding with exon 1 (A1-1-1), the first intron (A1-1-2), and the second intron (A1-2-2) but did not significantly interact with exon 2A (A1-2-1; Fig. 6 F). Interestingly, an 18-nt truncation of A1-1-2 at the 3′-end (A1-1-2-ΔUG; Fig. 6, C and D) weakened the interaction with GST-SUP-12 (RRM; Fig. 6, F and G). The truncated region contains repeats of UG that have been reported to bind to several RNA-binding proteins (Mittag, 1996; Takahashi et al., 2000; Buratti et al., 2004). The RNA oligonucleotide UG (5′-UGUGUGCCUG-3′) strongly interacted with GST-SUP-12 (RRM; Fig. 6 E), whereas the oligonucleotide UC (5′-UCUCUCCCUC-3′) showed nearly insignificant interaction (Fig. 6 E). Densitometric quantification of the results in Fig. 6 E indicates that only UG exhibited strong saturable binding with GST-SUP-12 (RRM; Fig. 6 F). Binding of GST-SUP-12 (RRM) to UG was saturated at a molar ratio of 1.1:1.0 with a dissociation constant of 0.31 μM, suggesting that they form a stoichiometric 1:1 complex with physiologically strong affinity. Removal of the UG repeats from A1-1-2 did not completely abolish the interaction of GST-SUP-12 (RRM) with A1-1-2-ΔUG (Fig. 6, F and G), suggesting that the UG repeats are sufficient but not necessary for this interaction. Binding of GST-SUP-12 (RRM) to A1-1-1, A1-1-2, or A1-2-2 did not reach saturation within the conditions used in this work (Fig. 6 G), so we were not able to determine stoichiometry and affinity. These results demonstrate that the RRM domain of SUP-12 directly interacts with the unc-60 pre-mRNA at multiple sites within exon 1 and the first and second introns. In particular, strong interaction of the SUP-12 RRM domain with the UG repeats near the 3′-splice site in the first intron supports that SUP-12 may function as a regulator of pre-mRNA splicing.

In this work, we identified SUP-12 as a critical regulator of the muscle-specific pre-mRNA splicing of unc-60. sup-12 mutations suppressed muscle-specific defects in actin organization which was caused by unc-60B mutations. This suppression involved switching of the unc-60 isoforms in the muscle cells: sup-12 mutations enhanced muscle expression of UNC-60A, the nonmuscle ADF/cofilin isoforms, but reduced expression of UNC-60B, the muscle isoforms, thus allowing UNC-60A to compensate for the function of UNC-60B in the muscle cells. SUP-12 localized to the muscle nuclei and bound directly to exon 1 and the first and second introns of the unc-60 pre-mRNA in vitro, suggesting that SUP-12 is directly involved in muscle-specific splicing of the unc-60 pre-mRNA.

Our data indicate that, in muscle cells, SUP-12 normally inhibits production of the unc-60A mRNA, but it enhances expression of the unc-60B mRNA. Several possibilities for the mechanism by which SUP-12 regulates expression of unc-60A and unc-60B could be considered. The most probable model is that the general splicing machinery may preferentially induce the splicing between exons 1 and 2A to produce unc-60A, whereas SUP-12 likely acts as an inhibitor of this splicing event. This model is strongly supported by the presence of UG repeats near the 3′-end of the first intron, which strongly interacts with the RRM domain of SUP-12. In the human cystic fibrosis transmembrane conductance regulator pre-mRNA, the nuclear RRM protein TDP-43 binds to the UG repeats at the 3′-end of intron 8 and causes exon skipping (Niksic et al., 1999; Pagani et al., 2000; Buratti et al., 2004). Thus, SUP-12 and TDP-43 may negatively regulate splicing in a similar manner by binding to UG repeats and directly competing with the U2 auxiliary factor, which is an essential splicing factor that binds to 3′-splice sites (Merendino et al., 1999; Wu et al., 1999; Zorio and Blumenthal, 1999).

An alternative model is that SUP-12 may indirectly inhibit 3′-end processing of the unc-60A pre-mRNA. Because the SUP-12–binding sites on the unc-60 pre-mRNA are not close to the polyadenylation site of unc-60A, SUP-12 may have to interact with 3′-end processing factors to influence this process. The last exon (5A) for unc-60A that contains the 3′-untranslated region resides upstream of exon 2B (Fig. 3 A). In nonmuscle cells, the mRNA 3′-end processing factors may cleave the pre-mRNA and polyadenylate after exon 5A but before exon 2B is transcribed. Therefore, in muscle cells, SUP-12 may inhibit 3′-end processing at exon 5A and promote elongation of the pre-mRNA. Indeed, in case of polycistronic genes in C. elegans, an interaction between a 3′-end processing factor and a factor for trans-splicing is reported (Evans et al., 2001). Also, it is possible that SUP-12 affects RNA stability or interacts with transcription factors and regulates transcription and pre-mRNA processing because splicing factors and transcription factors functionally interact and regulate pre-mRNA splicing in many instances (Bentley, 2002). In addition, we cannot exclude the possibility that SUP-12 may regulate relative stability of the two mRNAs.

The RRM domain of SUP-12 had activity to bind to the 5′-region of the unc-60 pre-mRNA but was unable to rescue the sup-12 mutant phenotype. This suggests that the COOH-terminal A/Q-rich sequence plays an important function. Although the function of the A/Q-rich sequence is unknown, it is intriguing that MEC-8, which regulates alternative splicing of unc-52 in the hypodermis in C. elegans (Lundquist et al., 1996; Spike et al., 2002), also contains an A/Q-rich region in addition to two RRM domains. The A/Q-rich sequence might have a regulatory function for a splicing factor or mediate interactions with other splicing factors, transcription factors, or 3′-end processing factors. We showed that the A/Q-rich region of SUP-12 is necessary and sufficient for speckled localization in the nuclei, suggesting that this region is important for SUP-12 to localize to speckles. Also, we noted that the bacterially expressed full-length SUP-12 protein was not only insoluble but also very susceptible for proteolysis (unpublished data), suggesting that the A/Q-rich sequence may regulate protein stability.

Furthermore, this work suggests that functional redundancy of the two ADF/cofilin isoforms in muscle is normally masked by tight regulation of tissue-specific splicing by sup-12. Thus, sup-12 mutations unmask the redundancy and allow UNC-60A to compensate for mutated UNC-60B in unc-60B mutants. From our previous work, it seemed logical to hypothesize that UNC-60B with weaker depolymerizing activity might be more suitable in muscle cells than UNC-60A, where less dynamic actin reorganization is needed than nonmuscle cells. Therefore, it is somewhat surprising that UNC-60A can substitute for UNC-60B in muscle. However, it is possible that, although the sup-12 mutants have apparently normal myofibrils, their muscle may exhibit different physiological properties from that of wild type under specific conditions. A number of human diseases are caused by alterations in the pre-mRNA splicing (Nissim-Rafinia and Kerem, 2002; Stoilov et al., 2002; Faustino and Cooper, 2003). However, our results suggest that, the splicing machinery is a potential therapeutic target for certain genetic diseases in which manipulation of tissue-specific splicing machinery may reveal hidden functional redundancy among splice variants to compensate for a disease gene. We propose that SUP-12– and SEB-4–related proteins are a new family of tissue-specific splicing factors in multicellular organisms.

Wild-type N2 and unc-60B(e677) (Waterston et al., 1980) were obtained from the Caenorhabditis Genetics Center (Minneapolis, MN). sup-12(st89) and sup-12(st203) were obtained from G.R. Francis (Exelixis, Inc., South San Francisco, CA) and R. Waterston (University of Washington, Seattle, WA). sup-12(s2896), sup-12(s2900) and sup-12(s2901) were isolated by a screen for suppressors of unc-60B. In brief, unc-60B(e677) homozygotes were mutagenized by ethyl methanesulfonate and F2 worms with improved motility were isolated. unc-60B(su158) (Zengel and Epstein, 1980) was originally obtained from H.F. Epstein (University of Texas Medical Branch at Galveston, Galveston, TX) and characterized previously (Ono et al., 2003).

sup-12 was mapped to the left arm of the X chromosome that was included in the duplication mnDp33 (G.R. Francis and R.H. Waterston, personal communication). We further narrowed down sup-12 by the snip-SNP mapping with polymorphisms in CB4856 (Wicks et al., 2001) to an interval between cosmid clones ZC64 and T06F4 that contained ∼20 genes. We performed feeding RNA interference of 10 genes in wild type and unc-60B(su158) and examined for a suppressor phenotype for unc-60B(su158). We found that T22B2.4(RNAi) suppressed the motility defect of unc-60B(su158) but did not affect motility of wild type. To confirm that T22B2.4 is sup-12, we sequenced the T22B2.4 gene in the sup-12 mutants and identified mutations in multiple sup-12 alleles (Results and Fig. 2 A).

Actin filaments were visualized by staining adult worms with tetramethylrhodamine-phalloidin (Sigma-Aldrich) as described previously (Ono, 2001). Immunofluorescent staining was performed on adult worms that were permeabilized with a freeze-crack method (Epstein et al., 1993) and fixed with methanol for 5 min at −20°C. Primary antibodies used were anti–UNC-60A (Ono et al., 1999) and anti-myoA (mAb 5.6, obtained from H.F. Epstein; Miller et al., 1983). Secondary antibodies were Alexa488-labeled goat anti–rabbit IgG and Alexa647-labeled goat anti–mouse IgG (Molecular Probes). To visualize nuclei, worms were fixed with 4% formaldehyde in PBS for 30 min at RT, permeabilized with acetone at −20°C for 5 min, and stained with DAPI (Sigma-Aldrich) at 1 μg/ml in PBS containing 0.5% Triton X-100, 1 mM EDTA, and 0.05% sodium azide for 15 min.

Fluorescent samples were mounted with the ProLong antifading reagent (Molecular Probes) and viewed by epifluorescence using an inverted microscope (model Eclipse TE2000; Nikon) with a 40× CFI Plan Fluor objective (dry; NA 1.4). Images were captured by a SPOT RT Monochrome CCD camera (Diagnostic Instruments) and processed by the IPLab imaging software (Scanalytics, Inc.) and Adobe Photoshop 6.0.

Total nematode RNA was isolated using a TRI reagent (Sigma-Aldrich). RNA samples (10 μg) were subjected to formaldehyde-agarose gel electrophoresis, transferred to positively charged nylon membranes (Millipore), and fixed by ultraviolet irradiation. cDNAs for unc-60A (505 bp), unc-60B (460 bp), and actin (act-1) (1.1 kb) were amplified by PCR, labeled with digoxigenin with random priming, and used as probes for Northern blotting using the DIG-High Prime DNA Labeling and Detection Starter Kit II (Roche Applied Science). Western blot was performed as described previously (Ono and Ono, 2002) using the following primary antibodies: anti–UNC-60A (Ono et al., 1999), anti–UNC-60B (Ono et al., 1999), anti-actin monoclonal C4 (ICN Biomedicals), and anti–α-tubulin (Amersham Biosciences).

To determine the promoter activity of sup-12, a 3,418-bp genomic fragment containing the 3,087-bp upstream region, exon 1, intron 1, and 58-bp of exon 2, was amplified by PCR using ExTaq DNA polymerase (Takara) and cloned into the gfp expression vector pPD95.67 (obtained from A. Fire, Stanford University, Stanford, CA) at the 5′-end of the gfp coding region. For expression of GFP-SUP-12 in body wall muscle, the sup-12 cDNA (yk1125e08, obtained from Y. Kohara, National Institute of Genetics, Mishima, Japan) was ligated in-frame with the 3′-end of gfp in pPD118.20 (obtained from A. Fire) that has the myo-3 promoter. For expression of fragments of SUP-12 as GFP-fusion proteins, fragments of the sup-12 cDNA encoding residues 1–117 or 118–248 were amplified by PCR and ligated in-frame with the 3′-end of gfp in pPD118.20. A synthetic stop codon was added for expression of residues 1–117. The plasmids were injected into hermaphroditic gonads of wild type or unc-60B(su158);sup-12(st89) at 10 μg/ml together with a dominant marker pRF4(rol-6(su1006)) at 90 μg/ml, and transgenic worms were selected by their roller phenotype or expression of GFP.

A cDNA fragment encoding residues 1–117 of SUP-12 and a synthetic stop codon was amplified by PCR and cloned into the SmaI–EcoRI cloning site of a GST expression vector pGEX-2T (Amersham Biosciences). The insert was sequenced to confirm that no mutations were introduced by PCR. GST alone or GST-SUP-12 (RRM) was expressed in Escherichia coli BL21 (DE3) by induction with 0.1 mM IPTG for 3 h and purified with Glutathione-Uniflow (BD Biosciences Clontech) following manufacturer's instruction.

Template DNA fragments used in the in vitro transcription reactions were amplified from C. elegans genomic DNA using the primers listed in Tables S1 and S2 . The sense primers contain the SP6 promoter sequence. Each RNA was in vitro transcribed using the MAXIscript kit (Ambion) in the presence of 50 μCi of α-[³²P]CTP (Amersham Biosciences) and purified with a Sephadex G-25 spin column (Amersham Biosciences). RNAs (5 × 10⁵ cpm) were denatured in water at 95°C for 2 min, chilled on ice for 2 min, and incubated with either buffer, or buffer and 1 μM GST or buffer and 1 μM GST-SUP-12 (RRM) for 30 min at RT. The buffer contained 50 mM KCl, 5 mM MgCl₂, 10% glycerol, 40 mM DTT, 0.1 mg/ml yeast tRNA, and 10 mM Hepes, pH 7.4). The samples were resolved on a 1% agarose gel in 1 × TAE (40 mM Tris-acetate, 2 mM EDTA, pH 8.3). The gels were dried and analyzed by the PhosphorImager (Molecular Dynamics).

Biotin-RNA pull-down assay (Lee and Schedl, 2001) was performed with the following modifications. Template DNAs for in vitro transcription of RNAs (95–284 bases) were amplified from C. elegans genomic DNA using the primers listed in Tables S1 and S2. The sense primers contain the T7 promoter sequence. Biotin-labeled RNAs were in vitro transcribed with T7 RNA polymerase (Invitrogen) in the presence of Biotin RNA Labeling Mix (Roche Applied Science) at 37°C for 2 h. The template DNAs were digested by RNase-free DNase I (Roche Applied Science) and the labeled RNAs purified with SigmaSpin Post-Reaction Clean-Up Columns (Sigma-Aldrich). RNA oligonucleotides UG (5′-UGUGUGCCUG-3′) and UC (5′-UCUCUCCCUC-3′) were synthesized and chemically labeled by biotin at the 5′-ends by Integrated DNA Technologies.

Biotin-labeled RNA at 0.1 μM was incubated with GST or GST-SUP-12 (RRM) in RP buffer (50 mM KCl, 1 mM MgCl₂, 10 mM Hepes-NaOH, pH 7.5) containing 10 mM DTT and 100 μg/ml yeast tRNA (Ambion) in a final volume of 200 μl at RT for 30 min. The mixtures were incubated with 0.1 mg of streptavidin MagneSphere Paramagnetic Particle (Promega) at RT for 20 min. The magnetic particles were isolated with a magnetic separation stand, washed three times with RP buffer, suspended in 20 μl of SDS sample buffer (2% SDS, 80 mM Tris-HCl, 5% β-mercaptoethanol, 15% glycerol, 0.05% bromophenol blue, pH 6.8), and incubated at 97°C for 2 min. Bound proteins were analyzed by SDS-PAGE (12% acrylamide gel) and staining with Coomassie Brilliant blue R-250 (National Diagnostics). The Coomassie-stained gels were scanned by a UMAX PowerLook III scanner at 300 dots per inch and the band intensity was quantified with Scion Image Beta 4.02 (Scion Corporation) by comparing with the intensities of known amounts of GST-SUP-12 (RRM).

Table S1 contains a list of PCR primers for amplifying template DNAs for in vitro transcription. Table S2 contains a list of primer combinations for PCR amplification of the template DNAs.

We thank G.R. Francis and R.H. Waterston for sharing data on sup-12 and strains, A. Fire for expression vectors, H.F. Epstein for anti-myosin antibody and a strain, and K. Bhat and D. Kalman for critical comments.

Some preliminary data were obtained by S. Ono while he was a postdoctoral fellow in the laboratory of Guy Benian, which was supported by a grant from the National Science Foundation (MCB-9728762). Some worm strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources. This work was supported by grants from the NSERC Canada to R.C. Johnsen and D.L. Baillie, the American Heart Association Southeast Affiliate, and the National Institutes of Health (R01 AR048615) to S. Ono.