Computational selection of nucleic acid biosensors via a slip structure model

Abstract

Aptamers have been shown to undergo ligand-dependent conformational changes, and can be joined to ribozymes to create allosteric ribozymes (aptazymes). An anti-flavin (FMN) aptamer joined to the hammerhead ribozyme yielded an aptazyme that underwent small, FMN-dependent displacements in the helix that joined the aptamer and ribozyme. This ‘slip structure’ model in which alternative sets of base-pairs are formed in the absence and presence of ligand proved amenable to energetic and computational modeling. Initial successes in modeling the activities of known aptazymes led to the in silico selection of new ligand-dependent aptazymes from virtual pools that contained millions of members. Those aptazymes that were predicted to best fit the slip structure model were synthesized and assayed, and the best-designed aptazyme was activated 60-fold by FMN. The slip structure model proved to be generalizable, and could be applied with equal facility to computationally generate aptazymes that proved to be experimentally activated by other ligands (theophylline) or that contained other catalytic cores (hairpin ribozyme). Moreover, the slip structure model could be applied to the prediction of a ligand-dependent aptamer beacon biosensor in which the addition of the protein vascular endothelial growth factor (VegF) led to a 10-fold increase in fluorescent signal.