(109d) Combinatorial Synthesis and Cheminformatics Modeling of Antibiotics-Based Polymers for Transgene Expression

Gene delivery, using viral or non-viral vectors is an attractive therapeutic modality for correcting diseases of genetic origin or consequence. Non-viral vectors, especially polymers, are limited by low transgene expression levels and cytotoxicity. Consequently, new synthesis and predictive approaches that facilitate rapid discovery of effective polymeric vehicles for transgene expression are necessary. Here, we developed a library of fifty six polymers based on aminoglycosides, several of which are approved by the FDA as antibacterial drugs. Parallel screening of transgene expression activity resulted in the identification of several candidates that demonstrated lower cytotoxicities and higher efficacies than 25kDa poly(ethyleneimine), a current standard for polymer-mediated transgene delivery. We first correlated transgene expression efficacies of polymers with physicochemical properties obtained using their ‘building block’ structures. Predictive Quantitative Structure-Activity Relationship (QSAR) models were generated using Support Vector Machine (SVM) regression. Different methods, includin K-fold cross-validation, model evaluation metrics and Y-Scrambling, were applied during the model construction process to ensure predictive and stable QSAR models. According to our structure based QSAR modeling, polymers with smaller sized aminoglycosides, hydrophilic oxygen atoms and shorter hydrocarbon chain between ether oxygen atoms in the cross-linkers are favorable for efficient transgene delivery. For further analyzing the influencing factors of the transfection efficacies of our polycation library, we also constructed several Quantitative Structure-Properties Relationship (QSPR) models to correlate polymer structures to the their experimentally determined polymer/polyplex properties, including size, zeta-potential, molecular weight, hydrophobicity, amine content and DNA binding efficacy. Taken together, this combined combinatorial synthesis, parallel screening and cheminformatics modeling approach can led to materials with high efficacies of transgene expression, fundamental understanding into polymer physicochemical properties that facilitate this activity, and predictive approach for the rational a priori design of efficient polymers.