Between 15% and 25% of Caenorhabditis elegans genes are organized in polycistronic transcription units, or operons (Zorio et al. 1994; T. Blumenthal, D. Evans, C. Link, A. Guffanti, D. Lawson, J. Thierry-Mieg, D. Thierry-Mieg, K. Duke, and S. Kim, in prep.). Polycistronic pre-mRNAs from these operons are processed into monocistronic mRNAs by cleavage and polyadenylation at the 3′ ends of upstream gene mRNAs accompanied by trans-splicing at the 5′ ends of downstream gene mRNAs. In general, these two processes occur within a 100-nucleotide region (Blumenthal and Steward 1997) and are mechanistically coupled (Kuersten et al. 1997). In C. elegans, 3′ end formation is dependent on an AAUAAA signal (Kuersten et al. 1997; Liu et al. 2001). It is expected that this sequence is bound by cleavage and polyadenylation specificity factor (CPSF), as it is in mammalian cells (for reviews, see Colgan and Manley 1997; Keller and Minvielle-Sebastia 1997; Zhao et al. 1999). Presumably, C. elegans 3′ end formation also requires cleavage stimulation factor (CstF), which binds a U-rich or GU-rich sequence downstream of the cleavage site. Homologs of each of the subunits of both mammalian CstF and CPSF are present in the C. elegans genome (C.J. Wilusz and T. Blumenthal, unpubl.). Trans-splicing generates 5′ ends of mRNAs in trypanosomes and many animals (Murphy et al. 1986; Sutton and Boothroyd 1986; Krause and Hirsh 1987; Rajkovic et al. 1990; Tessier et al. 1991; Stover and Steele 2001; Vandenberghe et al. 2001). A spliced leader (SL) exon is donated to the 5′ ends of mRNAs by a short RNA donor called SL RNA. The SL RNA exists as a ribonucleoprotein (RNP) particle (Thomas et al. 1988; Van Doren and Hirsh 1988; Maroney et al. 1990; Goncharov et al. 1999) that includes the Sm core proteins (Lerner and Steitz 1979). Unlike the other U snRNPs, which are capable of catalyzing repeated splicing reactions, the SL snRNP is consumed during the trans-splicing reaction. C. elegans possesses two distinct SL RNAs, SL1 RNA (Krause and Hirsh 1987) and SL2 RNA (Huang and Hirsh 1989). SL1 RNA is present at ∼7–10 times the level of SL2 RNA (S. Kuersten, R. Conrad, and T. Blumenthal, unpubl.). Nevertheless, mRNAs from most genes located in downstream positions in operons receive almost exclusively SL2 (Spieth et al. 1993), suggesting that a mechanism exists to regulate which spliced leader a particular pre-mRNA receives (Huang and Hirsh 1989). SL1 and SL2 have different sequences, but sequence complementarity between the SL RNAs and the pre-mRNA is not a factor in trans-splice site choice (Blumenthal and Steward 1997). SL1 and SL2 RNAs have many similar features—both are small (∼100 and 110 nucleotides, respectively), possess a trimethylguanosine (TMG) cap, associate with Sm proteins (Krause and Hirsh 1987; Thomas et al. 1988; Van Doren and Hirsh 1988; Huang and Hirsh 1989), and can be folded into a characteristic secondary structure (Bruzik et al. 1988). This structure consists of three stem/loops, with the trans-splice site located on the 3′ side of stem I, and the Sm-binding site located between stems II and III (see Fig. Fig.2A,2A, below). These similarities between the SL1 and SL2 RNAs make the question of substrate specificity even more intriguing. Figure 2 SL2 RNA mutations still allow core snRNP formation. (A) The mutations studied are shown (boxed and in boldface type) adjacent to the predicted SL2α RNA structure. Nucleotide positions that are conserved in 100% of known SL2 RNA genes (boxed ... Sequence analysis of SL2 RNA genes from the nematodes Dolichorhabditis and C. elegans identified several conserved features, including the 5′ end of the spliced leader, the trans-splice site, part of stem II, the Sm-binding site, and the top of stem loop III (Evans et al. 1997). Whereas the sequence of the spliced leader is dispensable, the primary sequence of stem II and loop III are required for SL2 trans-splicing in vivo (Evans and Blumenthal 2000). Several observations have suggested that SL2 trans-splicing specificity is coupled to 3′ end processing upstream. Mutation of the AAUAAA polyadenylation signal reduced the SL2 specificity of trans-splicing to the downstream gene mRNA (Kuersten et al. 1997). Conversely, strengthening the poly(A) signal, by changing AGUAAA to AAUAAA, resulted in increased SL2 trans-splicing (Liu et al. 2001). Furthermore, a mutational analysis of an operon intercistronic region revealed a second sequence required for SL2 trans-splicing: a U-rich region, located downstream of the poly(A) site (Huang et al. 2001). The location and sequence of this site suggest it may serve as a CstF-binding site. Although mutation of this sequence had only a small effect on 3′ end formation, it did prevent SL2 trans-splicing to the downstream gene mRNA (Huang et al. 2001). Therefore, we hypothesized that a physical interaction exists between the 3′ end formation machinery and the SL2 trans-splicing machinery. In an attempt to detect such an interaction, we performed immunoprecipitations using an antibody to C. elegans CstF-64, the homolog to the RNA-binding subunit of mammalian CstF (MacDonald et al. 1994). We report here that SL2 RNA, but not other snRNAs, is precipitated with αCstF-64. A mutational analysis indicates that association of CstF-64 correlates with SL2 snRNP specificity. Changes in the third stem/loop of SL2 RNA both prevent SL2 trans-splicing in vivo and immunoprecipitation with αCstF-64. Interestingly, a complete replacement of the third stem/loop interferes with SL2 identity, but it does not prevent snRNP function. Therefore, a molecular framework for SL2 trans-splicing is suggested in which association with CstF allows the less abundant SL2 snRNP to be preferentially used at trans-splice sites located downstream of a poly(A) site.