Pseudoknots in RNA: A Novel Folding Principle

Abstract

Models of the secondary structure of RNA usually contain a number of characteristic structural elements like base paired stem regions and various kinds of single stranded regions like hairpin, bulge, interior and bifurcation loops (Zuker and Stiegler, 1981). In such models mostly about one third or more of the nucleotide residues remains unpaired, which in some cases is confirmed by experimental data or computer-aided predictions. Although stem regions unquestionably are an important feature in the structure of RNA molecules, the final three-dimensional structure will be determined mainly by the interactions of the residues left in the single stranded regions. This is clearly illustrated in the case of tRNA, where the T-, D- and variable loop are largely responsible for maintaining the typical L shape of the molecule (Kim et al., 1974; Robertus et al., 1974). The tertiary interactions in the native conformation of tRNA often involve non standard base pairs or base triplets, while only in a few cases normal Watson-Crick base pairs are found. In fact Watson-Crick base pairing between complementary sequences might be considered an obvious possibility for tertiary interactions. Such interactions have been proposed indeed for the ribosomal 5S RNA (Pieler and Erdmann, 1982; Trifonov and Bolshoi, 1983) and were previously proposed on theoretical grounds (Studnicka et al., 1978). Tertiary interactions of this kind were called knotted or pseudoknotted structures depending on whether or not they could give rise to real knots in the RNA chain, especially when the resulting stem regions are in the range of one turn of an RNA double helix (Cantor, 1980).