Strategies for predictive modeling of overloaded oligonucleotide elution profiles in ion-pair chromatography

Due to their potential for gene regulation, oligonucleotides have moved into focus as one of the preferred modalities modulating currently undruggable disease-associated targets. In the course of synthesis and storage of oligonucleotides a significant number of compound-related impurities can be generated. Purification protocols and analytical methods have become crucial for the therapeutic application of any oligonucleotides, be they antisense oligonucleotides (ASOs), small interfering ribonucleic acids (siRNAs) or conjugates. Ion-pair chromatography is currently the standard method for separating and analyzing therapeutic oligonucleotides. Although mathematical modeling can improve the accuracy and efficiency of ion-pair chromatography, its application remains challenging. Simple models may not be suitable to treat advanced single molecules, while complex models are still inefficient for industrial oligonucleotide optimization processes. Therefore, fundamental research to improve the accuracy and simplicity of mathematical models in ion-pair chromatography is still a necessity. In this study, we predict overloaded concentration profiles of oligonucleotides in ion-pair chromatography and compare relatively simple and more advanced predictive models. The experimental system consists of a traditional C18 column using (dibutyl)amine as the ion-pair reagent and acetonitrile as organic modifier. The models were built and tested based on three crude 16-mer oligonucleotides with varying degrees of phosphorothioation, as well as their respective n - 1 and (P = O)1 impurities. In short, the proposed models were suitable to predict the overloaded concentration profiles for different slopes of the organic modifier gradient and column load.