LONG-RANGE BASE-PAIRING IN DROSOPHILA AND HUMAN MESSENGER-RNA SEQUENCES

The potential for long-range base pairing between the 5' and 3' ends o f mRNA molecules was examined for 134 Drosophila and 204 human sequenc es collected from the GenBank database. Each sequence was divided into two parts, a 5' sequence taken from the start of the protein-encoding region and a 3' sequence taken from the end of the transcript. The st rongest RNA pairing stem between each pair of 5' and 3' sequences was identified and scored using an alignment program modified to incorpora te RNA base pairing. The observed pairing scores were then compared wi th a random distribution of scores generated by aligning each 5' seque nce to random permutations of its corresponding 3' sequence. For both the Drosophila and the human mRNAs, the observed pairing scores were s ignificantly biased toward the upper tail of the random distributions, with 61% of the Drosophila sequences and 64% of the human sequences f alling within the upper half of the random distributions. This suggest s that a pattern of long-range base pairing may be a common feature of eukaryotic mRNAs. We have also analyzed a subset of Drosophila and hu man mRNAs which show the greatest potential for long-range pairing. Th e human pairings appear to be stronger and localized to more specific regions near the ends of the mRNA sequence than those of Drosophila.