computer modeling in molecular biology.pdf

5 Modellinp Nucleic Acids 1275.14 Modelling of Large Nucleic Acid MoleculesFor nucleic acids, starting with the first model of B-DNA [46], modelling has beenused extensively, even before detailed X-ray crystallographic data on fragments wereabundantly available [47]. Specifically, for RNA structures, the modelling of the anticodon-codoninteraction [48] and of a full tRNA [49] constitute both remarquableachievements.The structure proposed by Fuller and Hodgson [48] for the anticodon loop wasbasically correct. The main assumption was that the number of stacked bases in theanticodon loop is a maximum. Although the first two residues of the loop (positions32 and 33) were wrongly exposed toward the solvent, the structure provided most ofthe stereochemical explanations for Crick’s “wobble” hypothesis [50] and for therole of the modified purine at position 38. Levitt’s model [49] was the only“topologically” correct model. Its main features were: AA- and T-stems co-axial; D-and AC-stems co-axial; D-and T-loop interactions (especially G19-C56) ; U8-Al4with Hoogsteen pairing in trans; the R15-Y48 Levitt pair with Watson-Crick pairingbut incorrectly in cis instead of trans; in the AC-loop a la “Fuller-Hodgson”. Themain errors were in the conformation of the T-loop and in the use of five bases inthe syn conformation.Following the tremendous success of the X-ray crystallography method in ourunderstanding of tRNA structure, the modelling of tRNA structures was disregarded.In the early 80s, a revival of interest started following the accumulationof chemical, biochemical, and phylogenetic data on RNA molecules. Indeed, forRNA molecules, a variety of biochemical or chemical probes allowing to explore theaccessibility of most of the atoms implied in maintaining the secondary and the tertiarystructure of nucleic acids have been developed [51]. Also, one could exploitsystematically the enormous amount of information contained in biological sequencesfrom various organisms. Again, the basic assumption is that partlydivergent, but nevertheless functionally and historically related, RNA moleculesfrom various biological origins fold into similar tertiary structures. The comparisonand alignment of such sequences give information about the position and nature ofinvariant residues, which are considered important either for maintaining the 3D foldor for the function, and about those regions presenting insertions or deletions.Besides leading to consensus secondary structures, comparative sequence analysiscan be employed to identify tertiary contacts, as pioneered for tRNA by Levitt [49].Clearly, the method of comparative sequence analysis is restricted to molecules witheither structural or functional roles conserved across the phylogeny (e. g. tRNA, 5srRNA, ...).It should be understood that the use and display of atomic models for large RNAmolecules does not imply that the constructions are valid at atomic resolution in thecrystallographic sense. The fact of being meticulous about distances, angles, con-