Genes involved in pre-mRNA 3′-end formation and transcription termination revealed by a lin-15 operon Muv suppressor screen

Abstract

RNA polymerase II (Pol II) transcription termination involves two linked processes: mRNA 3′-end formation and release of Pol II from DNA. Signals for 3′ processing are recognized by a protein complex that includes cleavage polyadenylation specificity factor (CPSF) and cleavage stimulation factor (CstF). Here we identify suppressors encoding proteins that play roles in processes at the 3′ ends of genes by exploiting a mutation in which the 3′ end of another gene is transposed into the first gene of the Caenorhabditis elegans lin-15 operon. As expected, genes encoding CPSF and CstF were identified in the screen. We also report three suppressors encoding proteins containing a domain that interacts with the C-terminal domain of Pol II (CID). We show that two of the CID proteins are needed for efficient 3′ cleavage and thus may connect transcription termination with RNA cleavage. Furthermore, our results implicate a serine/arginine-rich (SR) protein, SRp20, in events following 3′-end cleavage, leading to termination of transcription.

Termination of RNA polymerase II (Pol II)-mediated transcription plays an important role in gene regulation and involves two linked processes: 3′-end formation and release of the Pol II from the DNA (1). Accordingly, termination of Pol II requires both the presence of an intact 3′-processing signal and several 3′-end processing factors including the cleavage and polyadenylation specificity factor (CPSF) and the cleavage stimulation factor (CstF) (2–6). Several of the polyadenylation factors are associated with the C-terminal domain (CTD) of the largest subunit of Pol II, which is known to be required for efficient 3′ processing (7–10). Much is known about what is required for 3′-end processing, but which factors specifically act in transcription termination and how these factors cause the Pol II complex to terminate are not entirely clear. The known link between 3′-end formation and transcription termination has led to multiple models to explain transcription termination.

Two such models have been proposed to explain the connection between 3′-end processing and transcription termination. One model, known as the “allosteric” model, proposes that termination is triggered by a conformational change of the Pol II complex that occurs on the emergence of the polyadenylation sequences (3). Another model, known as the “torpedo” model, proposes that termination is triggered subsequent to the cleavage event by the exonuclease XRN-2. When cleavage occurs at the poly(A) polymerase site, Pol II continues to transcribe a now-uncapped downstream RNA. This RNA is subject to degradation by the 5′-3′ exonuclease XRN-2; according to the torpedo model, termination occurs when the exonuclease collides with Pol II (11–13). Recently, a unified allosteric–torpedo model has been proposed to explain new experimental evidence in support of both the earlier models (14, 15).

One commonality between the three models of termination is the obligatory link between 3′-end formation signals and release of Pol II from the DNA. However, in Caenorhabditis elegans the presence of 3′-end formation signals does not always cause transcription termination, because ∼15% of C. elegans genes are organized in operons (16). C. elegans operons contain a promoter upstream of a tight cluster of two to eight genes. The genes in a single operon are transcribed into a polycistronic precursor mRNA (pre-mRNA) that is processed in two ways to create functional mRNAs of each gene: (i) the pre-mRNA of the upstream gene is cleaved and polyadenylated at its 3′ end; and (ii) the pre-mRNA of the downstream gene receives a 5′ cap by trans-splicing to a SL2 leader RNA. Genes in operons form normal 3′ ends and appear to have all of the functional sequences required for 3′-end formation. Nonetheless, unlike other Pol II genes, the 3′ end of an upstream gene in an operon forms without termination of transcription. The lack of termination after cleavage is intriguing; it is not yet known which features of operon 3′ ends allow apparently normal 3′-end formation without termination.

We used a C. elegans operon to screen for factors involved in transcription termination. The lin-15 operon contains two genes, lin-15B and lin-l5A. These genes encode structurally unrelated proteins that are involved in vulval development (17, 18). lin-15B and lin-15A belong to the class B and class A synthetic Multivulva (SynMuv) genes, respectively. Reduced function in both gene classes is required for the Multivulva (Muv) phenotype (19). To search for the potential targets or regulators of the SynMuv genes, we carried out a genome-wide RNAi screen on the lin-15AB(n765) mutant strain and identified 75 genes that suppressed the Muv phenotype (20). Fourteen of the identified genes were allele-specific suppressors of the SynMuv phenotype of lin-15AB(n765).

In this study, we characterized the lin-15AB(n765) molecular lesion and examined the group of 14 genes whose RNAi suppressed only this allele. This group includes factors with known functions in termination such as CstF and CPSF, as well as factors that previously were not known to function in termination. We analyzed the genes identified to determine the step of termination in which they functioned: pre-mRNA 3′-end formation, degradation of the RNA downstream of the cleavage site, or release of the Pol II complex from the DNA. In particular, we identified a well-known serine/arginine-rich (SR) protein, SRp20, as being involved in the process of termination at this 3′ end. We also identified two factors that have a role in enhancing cleavage at this 3′ end, and both these proteins contain a domain that can interact with Pol II CTD, suggesting they may serve to link transcription with RNA processing at the 3′ end of the gene.

Results

The lin-15AB(n765) Mutation Is a Unique Transposition That Causes a Dramatic Reduction of lin-15A mRNA Levels.

The lin-15AB(n765) mutation is 100% SynMuv at 19°C and originally was characterized as an ∼200-bp deletion in lin-15B (17). However, we have discovered that the n765 phenotype is caused by both a 171-bp out-of-frame deletion and a 982-bp transposition into the third exon of lin-15B (Fig. 1A), presumably fully inactivating this gene. The transposed fragment contains the 3′-end cleavage and polyadenylation signal of another gene (H18N23.2). The rest of the lin-15 operon, including the lin-15A gene, contains only the wild-type sequence. Because the SynMuv phenotype requires reduced expression of two genes and because the n765 allele contains only one molecular lesion, the mutation in n765 must affect a SynMuv A gene in addition to lin-15B. We thus hypothesized that Pol II transcription terminates at the transposed 3′ end within lin-15B, resulting in failure to transcribe the downstream gene, lin-15A. By RT-PCR we showed the 3′ end inserted into lin-15B is functional. By using qRT-PCR, we demonstrated that in n765 animals the lin-15A transcripts were reduced >10-fold (data not shown). This lin-15A mRNA was trans-spliced as in wild-type worms (data not shown). Thus, the SynMuv phenotype provides a functional readout of the amount of transcription termination occurring at the transposed 3′ end.

Fig. 1.

Suppression of the Muv phenotype of lin-15AB(n765). (A) (Upper) Schematic diagram of the exons (boxes)/introns (lines) of the lin-15 operon. The insertion in the lin-15AB(n765) allele is indicated. (Lower) Schematic diagrams of the exons (boxes)/introns (lines) of three suppressor genes for which deletion alleles were analyzed are shown above the data. Allele names are listed above lines that depict the extent of each deletion. (B) The Muv phenotype of lin-15AB(n765) is suppressed by feeding RNAi of the 14 allele-specific suppressors, all genes potentially involved in pre-mRNA 3′ processing and transcription termination. lin-15AB(n765) worms were fed with bacteria expressing dsRNA of either the GFP gene or each of the genes indicated. (C) Suppression of the Muv phenotype of lin-15AB(n765) by deletion strains compared with RNAi-treated worms. (D) Reversal of the n765 Muv suppression by lin-15A RNAi. lin-15AB(n765) young adult hermaphrodites were injected (inj) with 500 ng/μl of dsRNA of each of the five suppressors (rsp-6, cids-1, cids-2, nrd-1, and zfp-3) or a control gene (lin-39) and then fed (f) with bacteria expressing either lin-15A dsRNA or GFP dsRNA. For the two suppressors (cpsf-4 and cpf-1) that produce an embryonic lethal RNAi phenotype, we injected (inj) n765 animals with 500 ng/μl of either lin-15A dsRNA or GFP dsRNA, and fed (f) them with bacteria expressing dsRNA of each of the two suppressors. Error bars represent standard errors among the data collected from at least three independent experiments.

Disruption of Genes Involved in pre-mRNA 3′-End Formation and Transcription Termination Results in Allele-Specific Suppression of lin-15AB(n765).

In a genome-wide RNAi screen, we identified genes in which RNAi suppressed the SynMuv phenotype of lin-15AB(n765). The genes fall into several functional groups (20), but one group of RNAi suppressors showed a striking allele specificity [Fig. 1B, supporting information (SI) Table S1]. RNAi of these 14 genes did not suppress the SynMuv phenotype of three other SynMuv mutant combinations tested: (i) lin-15B(n309); (ii) lin-8(n111), lin-15AB(n765); and (iii) lin-8(n111), lin-15B(n744) (data not shown). Interestingly, this group of suppressors includes genes that encode CstF and CPSF, which are known to be required for 3′-end formation and transcription termination. Presumably RNAi of these genes results in partial inhibition of cleavage at the transposed 3′ end, thereby allowing continued transcription of the operon. Because the synMuv phenotype requires loss of activity of both lin-15B and lin-15A, an increase in lin-15A expression alone is sufficient to eliminate the SynMuv phenotype. As expected, these suppressors are allele specific, because they are capable only of restoring lin-15A expression after 3′-end formation at the introduced site.

The group of allele-specific suppressors (Table S1) also includes several genes whose protein products have a more tenuous connection to events at the 3′ end of the gene. Three of these genes (cids-1, cids-2, and nrd-1) encode proteins containing a CTD- interacting domain (CID). The presence of this domain indicates that these three proteins are likely to interact directly with the CTD on the large subunit of RNA Pol II (21, 22). The yeast orthologue of CIDS-1, RTT103, has been shown to interact with Ser2-phosphorylated CTD and to co-immunoprecipitate with Pol II and with RAT-1, the yeast homologue of XRN-2 (22). The yeast orthologue of another CID-containing protein, NRD-1, is required for 3′-end processing of small nucleolar RNAs and interacts with phosphorylated CTD (21). NRD-1 also is a key player in transcription termination of cryptic unstable transcripts (CUTs) (23, 24). The third suppressor with a CID, cids-2, contains a CID most closely related to that of cids-1, but otherwise these two proteins seem to be unrelated. cids-2 apparently has no orthologues outside of nematodes. Finally, another allele-specific suppressor, rsp-6, encodes SRp20, an SR protein that affects alternative splicing (25). This protein has no known role in transcription termination, although it has been implicated in polyadenylation of Rous sarcoma virus (26), an alternative polyadenylation site choice (27).

The levels of Muv suppression by RNAi of the different genes vary (Fig. 1B). Some possible reasons for the variability are that (i) RNAi may be more effective against some genes; (ii) genes may have various roles in different steps of termination; and (iii) some genes may play more significant roles in events at the 3′ end. In any case, because all the allele-specific suppressors form a coherent group of genes, many of which encode known RNA-interacting proteins, it is reasonable to propose that all are needed for some aspect of transcription termination at the transposed site.

We also tested deletion mutants for some of the suppressor genes (Fig. 1A). Two deletion alleles of rsp-6, which encodes SRp20, strongly suppressed the Muv phenotype of lin-15AB(n765) at levels similar to RNAi. Deletions in cids-1(tm2715) and cids-2(tm2802) also were effective in suppression and suppressed to very similar levels (Fig. 1C). The similarity in the levels of Muv suppression seen in the cids-1 and cids-2 deletions suggests they may function in a similar process, whereas the stronger suppression seen in the rsp-6 deletions suggests SRp20 may have a different function in termination.

Suppression Requires and Is Accompanied by lin-15A Expression.

If the suppressor genes are involved in termination, we would expect that restoration of lin-15A mRNA level would be responsible for the suppression of the SynMuv phenotype. In addition, the level of lin-15A transcript would increase when the suppressor genes are disrupted. To test whether restoration of lin-15A activity is necessary for suppression, we simultaneously treated lin-15AB(n765) animals with RNAi of both a suppressor gene and lin-15A. As a control, we first showed that RNAi of lin-39, a gene required for the SynMuv phenotype (28), resulted in suppression of the SynMuv phenotype of lin-15AB(n765), even when lin-15A was RNAi-treated simultaneously (Fig. 1D). In contrast, lin-15A(RNAi) strongly attenuated the Muv suppression by RNAi of the allele-specific suppressors (Fig. 1D). Clearly, expression of lin-15A is essential for the SynMuv suppression in the allele-specific group.

We next determined whether lin-15A is up-regulated on RNAi of the suppressor genes, by using qRT-PCR with lin-3 as a negative control (29). The data in Fig. 2 show that whereas lin-3(RNAi) did not alter lin-15A mRNA levels, RNAi of the allele-specific suppressors increased them. Clearly, disruption of a small group of allele-specific suppressor genes leads to an increase of the lin-15A transcript. This group of genes includes several that encode proteins known to be involved in 3′-end formation and transcription termination as well as several not previously shown to be required for pre-mRNA 3′-end formation or transcription termination. The rest of our analysis focuses on the three CID proteins and SRp20, potential players in 3′-end formation and/or transcription termination.

Fig. 2.

Abundance of lin-15A mRNA is increased in lin-15AB(n765) on RNAi of suppressor genes. qRT-PCR was performed on RNA samples from RNAi-treated mixed-staged n765 animals. rpl-26 expression was used for an internal reference. act-4, F41H10.11, and lin-15B mRNA levels from the same samples were measured as negative controls. Mean values of the tested genes are plotted as a ratio over rpl-26 mRNA. Error bars are based on at least three independent experiments.

Roles of cids-1, cids-2, and nrd-1 in 3′-End Formation and Transcription Termination.

Because of the nature of the lin-15AB(n765) allele and because many of the genes discovered among the allele-specific suppressors were factors known to be involved in 3′-end formation and/or transcription termination, we hypothesized that rsp-6, cids-1, cids-2, and nrd-1 also may be involved in 3′-end cleavage, degradation of the RNA downstream of the cleavage site, or release of Pol II from the DNA. We used qRT-PCR assays to distinguish between these possibilities with several amplicons and RT primers, measuring effects of suppressor-gene mutations on operon expression (Fig. 3). RT-PCR A reveals defects in 3′ cleavage, because it amplifies only from uncleaved RNA (Fig. 3A). In this case, the RT reaction was primed with an oligonucleotide in the transposed sequence downstream of the cleavage site, whereas the PCR was done with primers near the 5′ end of lin-15B. RT-PCR B reveals the level of lin-15B RNA downstream of the transposed cleavage site (Fig. 3B). This PCR product could increase via inhibition of 3′-end cleavage. Alternatively, inhibition of termination accompanied by inhibition of degradation of the RNA downstream of the cleavage site also would be expected to result in accumulation of this product. RT-PCR C measures levels of lin-15A mRNA, which should increase no matter which step of 3′-end formation or transcription termination was abrogated (Fig. 3C). To validate the assay, we first investigated worms RNAi-treated for CstF or CPSF. As expected, levels of all three PCR products increased (Fig. 3 A–C), consistent with the functions of CstF and CPSF in 3′-end formation.

Fig. 3.

Prevention of 3′-end cleavage or transcription termination by the SynMuv suppressors. Levels of three qRT-PCR products of RNAs from the mutant and RNAi-treated strains indicated in A–C were determined as described in Methods. In the diagrams above the data, the green arrow shows the site of the RT primer on the lin-15 operon, and the colored rectangle shows the PCR amplicon. In C, the RT was random-primed. Each RT-PCR was performed multiple times. For each experiment, levels were determined for nonsuppressed lin-15AB(n765) worms (circles) and for RNAi-treated or deletion strains (squares). The scales of the three panels are arbitrary, with the median control value set to 10 (gray circles). For SRp20, the three experiments controlled by white circles were independently set to 10. For results outside the range plotted in the three panels, the squares are shown above the graph, and the levels are given with their standard deviations. (A) The RT primer is downstream of the cleavage site, so only uncleaved lin-15B is detected. (B) This qRT-PCR depends on accumulation of lin-15B RNA downstream of the cleavage site. The RT-PCR B product should accumulate whether 3′-end cleavage is prevented, as shown in A, or whether transcription fails to terminate and/or exonucleolytic digestion fails to occur following 3′-end cleavage. (C) This qRT-PCR measures lin-15A RNA levels. All suppressors show increased lin-15A mRNA. (D) lin-15 RNA levels measured on a whole-genome tiling array. The green line shows hybridization to oligonucleotides spanning the lin-15 locus with RNA isolated from the lin-15AB(n765) strain. Signal essentially disappears at the site of the insertion. Small peaks are likely to be cross-hybridization with another gene. The blue line shows the recovery of signal 3′ of the insertion site in the lin-15AB(n765) strain with the rsp-6(tm367) allele.

The yeast homologue of cids-1, RTT103, previously had been reported to crosslink at the 3′ end of genes and to be associated with XRN-2. We therefore expected that cids-1 would be involved in degradation of the RNA downstream from the cleavage. Thus, it was somewhat unexpected when we found that, in the strain with cids-1–deletion, cleavage at the introduced 3′-end site is significantly reduced (Fig. 3A). Similarly, cleavage was reduced, but to a lesser extent, in the strain with cids-2 deletion. This finding suggests that both these presumed Pol II-interacting proteins play roles in 3′-end cleavage.

nrd-1 RNAi resulted in a small increase in lin-15A RT-PCR product (Fig. 3C), and it did so without increasing the levels of either lin-15B product (Fig. 3 A and B). One possibility is that there is a CUT somewhere 5′ to lin-15A, and when nrd-1 is knocked down by RNAi termination of the low-level CUT is prevented, resulting in a read-through transcript of lin-15A. Presumably this transcript would be trans-spliced to give functional lin-15A mRNA.

SRp20 Increases RNA Downstream of the Cleavage Site Without Affecting Cleavage.

In the SRp20-deletion strain (tm367), RT-PCR A did not increase, indicating that the pre-mRNA is cleaved normally (Fig. 3A). Nonetheless, RT-PCR B levels increased 4- to 14-fold (Fig. 3B). Therefore, SRp20 seems to function in transcription termination and not in cleavage. In the SRp20 mutant the product that would be expected to be degraded by XRN-2 exonuclease accumulates, and transcription fails to terminate at the site of the transposition. The data from the qRT-PCR was confirmed by comparing RNAs from the two strains, lin-15AB(n765) and lin-15AB(n765);rsp-6(tm367), on tiling microarrays (Fig. 3D). Whole-genome tiling arrays were hybridized with cDNA from the two strains to search for global differences in termination when the rsp-6/SRp20 gene was mutated. Although no such differences were obvious, we did find several genes in which alternative splicing was altered (data not shown), indicating the importance of SRp20 in splice-site choice, as in other organisms (25). Furthermore, we saw a striking difference at the lin-15 locus. In n765, RNA levels drop dramatically at the site of the insertion. In contrast, in the rsp-6 mutant strain, RNA levels recover at the site of the insertion and remain higher throughout the operon. This result confirms and extends results seen in the qRT-PCR assay, because it also demonstrates that the RNA downstream of the cleavage event accumulates in the absence of SRp20.

We considered the possibility that the SRp20 mutation could suppress the phenotype by affecting the splicing of lin-15B. For example, the rsp-6 mutation could cause removal of the lin-15B exon containing the insertion. However, RT-PCR from exon 2 to exon 4 in lin-15(n765); rsp-6(tm367) revealed no product missing this exon (data not shown). Alternatively, a cap supplied by a trans-spliced leader would stabilize the lin-15B mRNA downstream of the insert. However, we found no trans-splicing to downstream lin-15B exons in either the n765 or the rsp-6 mutant strain (data not shown). Thus, the increase in the levels of lin-15B RNA downstream of the cleavage site probably implicates SRp20 in some aspect of transcription termination. Knockdown of SRp20 could prevent degradation of the RNA following cleavage. Alternatively, 3′-end cleavage could result in proximal transcription termination, in which case knockdown of SRp20 could act by preventing release of Pol II from the DNA.

Identification of Additional 3′-End Formation Proteins That Can Suppress lin-15AB(n765).

Our results indicate that allele-specific suppression of lin-15AB(n765) can be used to identify and analyze genes involved in 3′-end formation or transcription termination. Some genes may have eluded the genome-wide RNAi screen because of RNAi-caused lethality or because of the absence of some genes from the RNAi library. We thus tested a number of C. elegans genes whose homologues in human and/or yeast have been implicated in pre-mRNA 3′-end events (Tables S1 and S2). Several were potent allele-specific suppressors of the Muv phenotype (Fig. 1B, Table S1). Both CPF-2/CstF-64 and Polypyrimidine tract binding protein-associated Splicing Factor (PSF-1) (a homologue of mammalian PSF and p54nrb) were strong suppressors. Mammalian PSF recently has been shown to interact with 3′-end formation factors and to have a role in recruiting the exonuclease XRN2 to facilitate pre-mRNA 3′ processing and transcription termination (30). Interestingly, proteins encoded by psf-1 and another gene zfp-3 (revealed in the screen as an allele-specific suppressor) were implicated as candidate binding partners of C. elegans CPF-2/CstF-64 in a two-hybrid screen (31). When future potential players in 3′-end formation are identified, as in the case of psf-1, the lin-15AB(n765) strain could be used to examine the function of the gene.

Discussion

The SynMuv phenotype requires loss of function of one A-class and one B-class gene. Uniquely, the lin-15 operon contains one member of each class transcribed from a single promoter upstream of the first gene, lin-15B. Nonetheless, a single mutation was able to reduce or eliminate expression of both genes. We show here that this mutation contains the 3′-end formation region from another gene transposed into lin-15B, which results in a premature 3′-end formation signal in lin-15B. Not surprisingly, this region drastically reduces lin-15A expression by terminating transcription before RNA polymerase reaches this gene. Among the many SynMuv suppressors obtained, only a small group was specific to this allele alone. We show here that many genes in this group are known 3′-end formation factors. Thus the remainder of this class of suppressors is of special interest, because they could play previously unknown roles in 3′-end formation or transcription termination (see the model presented in Fig. 4). In this paper we have performed analysis on four of the less characterized suppressors, three containing CIDs and one SR protein.

Fig. 4.

Model for 3′-end events. RNA Pol II with its C-terminal domain (CTD) is shown transcribing just beyond the 3′ end of a gene. The pre-mRNA is emerging before cleavage, with the cleavage site marked by an arrow. The cleavage and polyadenylation specificity factor (CPSF) is bound to the AAUAAA signal on the pre-mRNA, with the cleavage stimulatory factor (CstF) bound to a U-rich region just 3′ to the cleavage site. The three proteins with the CTD-interaction domain are shown bound to the CTD. This paper implicates two of these proteins, CIDS-1 and CIDS-2, in cleavage. SRp20 plays a role in termination of transcription, perhaps by binding to the RNA downstream of the cleavage site.

The screen identified CIDS-1, a C. elegans orthologue of yeast RTT103, and CIDS-2, a paralogue of CIDS-1. Yeast RTT103 has been shown to co-immunoprecipitate with Pol II and XRN-2 (22). Analysis of mutants of cids-1 and cids-2 genes suggests that CIDS-1 and CIDS-2 are involved in the 3′-end cleavage. However, the limitations of our assay prevent us from excluding the possibility that they also could function directly in terminating Pol II transcription. Both effects on 3′-end cleavage and on transcription termination are consistent with the fact that both CIDS-1 and CIDS-2 contain a CID domain that could interact with the phosphorylated CTD of Pol II.

We also found that another CID-containing protein, NRD-1, was one of the best suppressors of the n765 Muv phenotype (Fig. 1B). The yeast homologue of NRD-1, Nrd1p, is required for termination of nonpolyadenylated transcripts from Pol II-transcribed small nuclear RNA and small nucleolar RNA genes (32). However, it is unclear whether Nrd1 also is involved in terminating polyadenylated transcripts, other than its own (33). Read-through transcription from pre-mRNA has been observed previously in Nrd1 mutants in yeast (34, 35). Therefore, one could hypothesize that reduction of nrd-1function may result in read-through of the transcription of the lin-15B gene in the n765 mutant. Because the lin-15A and lin-15B genes are in an operon with the lin-15A gene downstream, read-through of the short lin-15B transcript could increase the level of polycistronic lin-15B and lin-15A pre-mRNA, resulting in a higher level of the lin-15A transcript. However, the nrd-1 mutant displayed no or little increase in the levels of RT-PCR A or B in the qRT-PCR assay (Fig. 3), and an increase in those products would be expected if the knockdown of nrd-1 was causing read-through of the inserted 3′ end. An alternative hypothesis is that a CUT is present upstream of lin-15A, and knockdown of nrd-1 allows read-through of the CUT.

Our analysis indicates that SRp20 is likely to function either in the degradation of the RNA downstream of the cleavage site or in release of the polymerase from the DNA. We can envision two possibilities. In the first, we propose that transcription normally does not terminate following cleavage at the introduced 3′-end formation site of n765, but the RNA synthesis downstream of the cleavage site is accompanied by degradation of the RNA. In this case, we propose that SRp20 may function through interactions with the exonuclease, XRN-2, to facilitate the RNA degradation. When SRp20 is missing, degradation is abrogated, so RT-PCR B product and lin-15A mRNA accumulate.

The other possibility is that transcription terminates soon after the cleavage event at the introduced 3′-end formation site and that SRp20 plays a role in the termination. In this case, the absence of SRp20 would prevent Pol II release from the DNA. Theoretically, these possibilities could be distinguished by measuring Pol II occupancy along the lin-15 operon in n765 and n765;tm367 strains. Functions for SRp20 in termination have not been reported previously. However, there are two reports of SRp20 at the 3′ ends of genes. SRp20 has been shown to co-immunoprecipitate with pre-messenger RNA cleavage factor 1 (CFIm), a factor involved in 3′-end formation in mammals (36), and to affect polyadenylation in Rous sarcoma virus (26). RNAi against the C. elegans homologue to CFIm did not suppress the Muv phenotype, so interaction with CFlm is unlikely to be the mechanism by which SRp20 functions. However, RNAi against poly(A) polymerase did suppress the phenotype weakly, so the function of SRp20 may be through the polyadenylation process. Because splicing of the last exon encourages 3′-end formation in mammals, another possibility is that splicing of the last exon may affect release of Pol II from the DNA. In an SRp20 mutant slower splicing of the last exon might lead to slower release of Pol II from the DNA.

The precise roles of the three CID proteins in 3′ cleavage, as well as how the CID interaction with Pol II contributes to that role, await further experimentation. Recently, it was reported that SRp20 interacts with Pol II to regulate exon inclusion. It is interesting that SRp20 also contributes in some way to events following mRNA 3′ cleavage that lead transcription to terminate.

Materials and Methods

RNAi Screen.

The RNAi screen was performed as reported in (20) except for the following changes. Many C. elegans homologues of CPSF and CstF subunits are essential for C. elegans development. RNAi of these genes produced severe embryonic lethality (Table S1 and Table S2), which makes the investigation of postembryonic development (vulval development in our study) impossible. To overcome this problem, RNAi feeding cultures of genes that had lethal phenotypes were diluted with a second bacterial culture expressing dsRNA of the control GFP gene, thereby reducing the effectiveness of the RNAi. Multiple dilutions were created, and the dilution with greatest Muv suppression was used in each SynMuv count.

Double RNAi.

Two genes (lin-15A and a suppressor gene) were subjected to RNAi simultaneously to test the influence of lin-15A levels in the rescue of the Muv phenotype by the suppressor genes. To perform RNAi in the suppressor genes for which RNAi is not lethal, DNA templates containing the T7 promoter were PCR amplified from plasmids corresponding to each RNAi clone. By using T7 polymerase-based transcription kit (Ambion Megascript Kit), dsRNAs were synthesized in a single reaction. Concentrations of dsRNAs were determined by using a spectrophotometer. The quality and size of the dsRNAs were assessed by gel electrophoresis. lin-15AB(n765) young hermaphrodites were injected with 500 ng/μl of the dsRNA. Worms were allowed to recover for 24 h before a single worm per plate was placed on either an lin-15A(feeding RNAi) or GFP(feeding RNAi) plate for consecutive 48-h egg laying. Five days later, the percentage of Muv in the F1 was scored. For lethal genes, the experiment was reversed: lin-15AB(n765) young hermaphrodites were injected with 500 ng/μl of the dsRNA of either lin-15A or GFP. Twenty-four hours later single worms were put on plates for feeding RNAi of the suppressor genes.

Real-Time RT-PCR Analysis.

The data in Fig. 2 were obtained as described in (29). The rpl-26 gene encoding a ribosomal protein was used as an internal control for data normalization. The data shown are representative of at least three independent worm cultures and RNA isolations.

The data in Fig. 3 were obtained as follows. Worms were grown for 4 to 5 days, washed with M9 buffer, and embryos or adults were harvested and broken by freezing and thawing in TRIzol reagent (Gibco BRL). Total RNA was prepared by TRIzol extraction. First-strand cDNA was prepared by reverse transcription by using SSII (Invitrogen) and random primers or a specific oligonucleotide primer (Table S3). The 20-μl reverse transcription reaction was diluted to 300 μl with water, and 3 μl of diluted reverse transcription reaction was used for each qRT reaction. Each 10-μl qRT-PCR mixture contained 0.25 μM of each primer (Table S3) and 5 μl of 1X SYBR Green Mix (Applied Biosystems). qRT-PCR was performed in triplicate on an ABI 7900HT Fast Real-Time PCR System (Applied Biosystems). Standard curves using 1 pg to 250 ng of cDNA pools were used. Relative fold changes were calculated by using the standard curve method for relative quantization, and numbers were corrected by removing outlier points that had a coefficient of variation greater than 17%. The rpl-26 gene was used as an internal control for data normalization.

Tiling Array Data.

Total RNA was prepared from two C. elegans strains, lin-15B(n765) and lin-15B(n765);rsp-6(tm367), as above. RNA was reverse-transcribed with random primers and then DNA Polymerase 1 was used to make double-stranded cDNA. The ds cDNA was digested with the endonuclease APE1 (Affymetrix) and fluorescently labeled. The cDNA then was hybridized to a GeneChip C. elegans Tiling 1.0R Array (Affymetrix) and scanned by a GeneChip Instrument System. The Affymetrix IGB program was used to visualize results.

Additional Materials and Methods are included in the SI Text.