In general, poly(A)mRNA appears to be derived from larger nuclear RNA precursors. The maturation of these precursors involves excision of sequences of variable length from within the molecule and splicing of the remaining structural and coding sequences. The mechanism by which this process occurs is not known. It does not appear to operate solely through the recognition of a defined primary sequence or through the formation of a consistent secondary structure. We propose an alternative model in which poly(A) facilitates the splicing event by promoting the formation of triple-stranded structures within the mRNA precursor.The structural genes of eukaryotes and viruses that specify polyadenylated § mRNA species are interrupted by nucleotide sequences not represented in the mature mRNA transcripts. Such sequences have been termed "intervening sequences" or "introns" (for review see refs. 1-3). The primary products of such transcription units are large mRNA precursors that include intervening sequences (4, 5). Present evidence indicates that, after the addition of a poly(A) tract to its 3' end, the mRNA precursor is processed within the nucleus or during transport to the cytoplasm to produce a mature mRNA molecule (5-8). Processing is accomplished by the successive excision of introns and splicing of conserved sequences. The only examples of eukaryotic mRNAs that are known not to follow such a processing pathway are those specifying histones. Those histone genes that have been examined do not contain intervening sequences (9), and histone mRNAs are predominantly not polyadenylated (10)(11)(12) The conservation of sequences spanning the splicing sites in RNA species from evolutionarily distant organisms points both to a common general mechanism for processing of polyadenylated RNA and to the possibility that, in a given organism, processing may be catalyzed by a single enzyme or enzyme complex. At the moment, a mechanism that allows the accurate removal of intervening sequences from a mRNA precursor and the precise ligation of the remaining structural sequences is unknown. It seems likely that if the order of structural sequences is to be conserved, the ends of an intervening sequence must be brought close to each other prior to endonucleolytic cleavage, so that ligation can take place without physical separation of the 3' and 5' ends of the structural sequences that are to be spliced. However, intervening sequences are variable in length and, with the exception of the border regions already mentioned, possess no marked homologies or symmetries of sequence. Attempts to draw the ends of an intervening sequence together by maximizing Watson-Crick base pairing do not generate structures in which donor and acceptor sites exhibit a reasonably constant configuration. A splicing mechanism based solely on the maintenance of a fixed secondary structure also seems unlikely for other reasons. For example, it has been shown that intervening sequences are free to change much more rapidly than structural sequences (16)(17)(...