Long-distance splicing.

W ith the draft sequence of the human genome came the surprise that there were fewer genes than imagined. From where does complexity spring if not from the number of genes in an organism? RNA splicing provides at least part of the answer. Pre-mRNA splicing by alternative pathways is well known to expand an organism’s protein diversity by generating distinct protein isoforms. Beyond cis-splicing at a single locus (Fig. 1a), there is evidence for specialized cis-splicing that results from read-through transcription of adjacent loci followed by splicing to generate transcription-induced chimeras (TICs) from two genes (Fig. 1b) (1, 2), as in the TNSF12/TNSF13 chimera expressed in human T cells (3). In contrast to these cis-splicing events, trans-splicing joins exons from separate pre-mRNA transcripts. These transcripts can be encoded by different DNA strands at the same locus, as in trans-splicing of the mod(mdg4) gene in Drosophila, or by different alleles at the same locus, as for the lola gene, also in Drosophila (Fig. 1c) (4). All of these RNA splicing events involve transcripts from the same general region of the genome. In the work by Di Segni et al. in this issue of PNAS (5), the authors provide evidence suggesting yet another pathway to increase protein diversity, a pathway that involves cis-splicing of a single mRNA or trans-splicing of distinct mRNAs from distant genes by the tRNA-splicing machinery (Fig. 1 d and e). Whereas splicing of nuclear pre-mRNA generally requires an RNA–protein machine, termed the spliceosome, splicing of eukaryotic tRNAs requires only proteins—an endonuclease, a ligase, and a phosphotransferase. These enzymes catalyze tRNA splicing in four steps (6, 7). First, the tRNA endonuclease cleaves both exon–intron junctions. This cleavage produces two half-tRNAs as well as a linear intron and at the cleavage sites, free hydroxyls at the cleaved 5 ends and 2 ,3 cyclic phosphates at the cleaved 3 ends. The next three steps are catalyzed by the tRNA ligase and involve the polynucleotide kinase, cyclic phosphodiesterase, adenylate synthetase, and ligase activities of this enzyme (7). First, the 5 -OH is phosphorylated and the 3 -cyclic phosphate is opened to form a 3 -OH and 2 phosphate. Then, the two tRNA halves are ligated and, at least in vitro, the excised tRNA intron is also ligated, circularizing the intron (8). Finally, the 2 phosphate is cleaved from the ribose. Just as pre-mRNA splicing generally occurs in the nucleus, pre-tRNA splicing in higher eukaryotes occurs in the nucleus, but pretRNA splicing in budding yeast occurs in the cytoplasm (9). How does the pre-tRNA splicing pathway recognize a substrate? In general, eukaryotic pre-tRNA is recognized for splicing, not by its intron sequences, as is largely the case in pre-mRNA splicing, but by the structure of the mature tRNA domain itself (10). In contrast, archaeal pre-tRNAs form a structure at the splice sites called a bulge-helix-bulge (BHB), which is required for recognition by the tRNAsplicing machinery. In either case, it is the tRNA-splicing endonuclease that recognizes the splice sites (11). Previous work from Tocchini-Valentini and colleagues (12, 13) has demonstrated that mature domain-independent splicing can occur in yeast or Xenopus if a sequence that will form a BHB structure is inserted into a model pre-tRNA or mRNA. Moreover, an exogenous archaeal endonuclease expressed endogenously in mammalian cells can splice a BHB structure inserted into a reporter mRNA (14). In this issue, Di Segni et al. (5) designed an elegant system to ask whether tRNAs can be used to recruit the endogenous tRNA-splicing machinery in yeast to mediate mRNA splicing (5). The authors engineered constructs that would produce hybrid pre-tRNA/pre-mRNAs. The constructs contain SUP4 tRNATyr, a suppressor of nonsense ochre (UAA) mutants, within STE2 or STE3, two genes involved in the mating signal transduction pathway in Saccharomyces cerevisiae. The authors tested for cisor trans-splicing by using constructs either with the entire tRNA inserted into a single mRNA or with half-tRNAs fused to two mRNA fragments (Fig. 1 d and e). Functionality of the tRNA after splicing was assayed by the ability of the hybrid construct to confer prototrophy for methionine, lysine, or adenine in a strain harboring nonsense ochre mutations in genes involved in the biosynthesis of these nutrients. Additionally, a reporter for a functional mating signal transduction pathway indicated whether splicing of STE2 or STE3 into mature mRNAs was successful. First, the authors tested whether the yeast cells could excise a pre-tRNA from the middle of a pre-mRNA. In this cis-splicing system, the mRNA essentially corresponds to the intron of a permuted pre-tRNA gene in which the 3 half lies upstream of the 5 half (15). Remarkably, the yeast cells not only process functional tRNA from the hybrid construct but also ligate the flank-