Separating polycistronic pre-mRNAs into monocistronic mRNAs involves both cleavage and polyadenylation at 3' ends of upstream genes and SL2-specific trans-splicing 100 bp away at the 5' ends of downstream genes. To elucidate the mechanisms of these events, we are using a heat shock dependent transgenic system to study the
gpd-2/gpd-3 operon containing mutations in sequences predicted to affect 3' end formation or trans-splicing. Following heat shock, we observe several products: mature
gpd-2 and
gpd-3 mRNAs, a very small amount of polycistronic pre-mRNA, and a product extending from a position 22 bp downstream of the
gpd-2 poly(A) site (-76 from the
gpd-3 trans-splice site) and extending through the
gpd-3 gene. We hypothesized that this product may result from branching of SL2 RNA at the -76 A residue, and that utilization of this unusual site near the 3' end of the upstream gene may be the key to the mechanism of SL2-specific processing of operon RNAs. In support of this idea the sequence around the -76 A is capable of forming a short hybrid at the appropriate location on U2 snRNA; the sequence is conserved in C. briggsae, although most sequences in the intercistronic region are not; a poly(Y) tract is present just downstream of the -76 A; and a similar product has been observed previously in the case of
tra-2, another SL2 trans-spliced gene (Kuwabara and Kimble, personal communication). When we either deleted this A or mutated the poly(Y) tract,
gpd-2 3' end formation was largely unaffected, as expected. However, in both cases, the proposed branch product was lost and the
gpd-3 product severely reduced or lost. This suggests that trans-splicing failed, since mutation of the trans-splice site also gave loss of
gpd-3 product (but not loss of the proposed branched molecule), presumably because the downstream uncapped product resulting from 3' end formation is rapidly degraded unless it is given a cap by trans-splicing. Mutation of the
gpd-2 AAUAAA resulted in complete failure to process
gpd-2 mRNA at the 3' end, as expected, but it also reduced SL2 specificity of trans-splicing to
gpd-3. Thus, either formation of a cleavage and polyadenylation complex at the 3' end of the upstream gene or 3' end formation itself is required for SL2 specificity in downstream gene processing. Combining the mutations in the AAUAAA and the poly(Y) tract completely eliminated SL2 trans-splicing. This double mutation resulted in increased accumulation of polycistronic RNA and mature
gpd-3 mRNA trans-spliced to SL1. It appears that when a polycistronic pre-mRNA accumulates because the AAUAAA is deleted, it can serve as a substrate for trans-splicing, but the trans-splicing is aberrant. When the poly(Y) tract sequence is intact, both SL1 and SL2 are used, but when it is deleted, all trans-splicing is to SL1. This shows that the poly(Y) tract near -76 affects trans-splicing directly. We hypothesize that branching near the
gpd-2 3' end is a key element in SL2-specific polycistronic RNA processing, and that molecules involved in 3' end formation may also play a role in selection of this unusual branch point location. Although branch points generally occur within 20 bp upstream of a splice site, there are rare exceptions (usually involving alternative splicing) that occur this far upstream. What makes the branching occur at -76 in this case and why it is specific for the SL2 snRNP are subjects for future studies.