Computational Design of DNAzymes

DNAzymes are characterised by their enzymatic activity, cleaving RNA molecules at purine-pyrimidine junctions, and are comprised of two substrate-binding arms and a catalytic core. Applications for DNAzymes, based on the different type of chemical reactions, include therapeutic DNAzymes, ligating DNAzymes and those that are used as biosensors. The catalytic DNAs, used for RNA cleavage, are being developed as a therapeutic approach for gene inactivation due to their efficiency, specificity and multiple turnover. DNAzymes are currently identified through large costly and time consuming screens (in vitro selection), which has hampered the development of this technology. This project aims to develop a computational design tool that could prove useful in DNAzyme design. The current model developed by Pine, Brooke and Marco (unpublished data) determines, using the interaction free energy, hairpin(internal) and dimer parameters, for a given transcript, the most effective DNAzymes. The previous DNAzyme program was refactored and a series of new parameters were introduced, specifically internal structure energy, hairpin energy, base pair probability and entropy. The parameters were analysed in relation to the efficiency of the molecule. While energy of the DNA enzyme was a strong indicator for efficiency, the overall approach could be improved. Initially linear models were developed, but these incorrectly assumed that DNAzyme efficiency is influenced by one parameter. Models for prediction of efficiency included linear regression, logistic regression and beta-binomial out of which the beta- binomial proved best. This was used to identify potential DNAzymes for testing in the laboratory, Lead DNAzymes, targeting key drivers of prostate cancer and cervical cancer, the androgen receptor and HPV16 respectively, were subsequently tested using in vitro and cellular models of cancer. The results demonstrated that the computational approach could successfully identify efficient DNAzymes and therefore has potential to accelerate the design of these molecules.