Does column length still matter? A case study of the effect of column length on resolution of therapeutic oligonucleotides.

Abstract

As the importance of therapeutic oligonucleotides (ONs) continues to grow in the pharmaceutical industry, the importance of high performing analytical methods needed to characterize them also grows. The characteristics of these molecules (e.g., highly charged phosphate backbone, and small but important modifications such as methylation and fluorination) make them difficult to analyze thoroughly using conventional liquid chromatography (LC) conditions. Recently, other research groups have been emphasizing the utility of ultra-short (<< 50 mm) columns for proteins and other large biomolecules, and have remarked that long columns only add unnecessary peak dispersion without providing additional resolution over short columns. These statements naturally call into question the long-established theory for small molecule LC separations that asserts that separation performance is maximized by working at the highest available operating pressure, and then choosing the longest column possible while working at the van Deemter optimum flow rate. This apparent contradiction in turn raises the question - for which types of large biomolecule does the established chromatographic theory no longer apply? In this study we have carried out experiments and calculations aimed at answering this question for ion-pairing reversed-phase separations of therapeutic ONs with masses on the order of 6 kDa. This included measuring isocratic plate heights for these molecules after establishing an empirical relationship between retention, mobile phase composition, and flow rate, because retention of the ONs is extremely sensitive to pressure (20 % increase in k per bar pressure drop), and thus retention varies with flow rate at a constant mobile phase composition. After taking these factors into account, we find that resolution of the oligonucleotides does increase with the square root of column length, as predicted by the well-established theory for small molecules. However, we also find that this relationship is only found when the gradient slope is held constant while varying the column length, and that if this is not done it is actually possible to observe that resolution decreases with increasing column length. Thus, the design of experiments used to evaluate the role of column length in separation performance is critical. In addition to the importance of these findings to development of LC methods for ON separations in general, they will be especially impactful in two-dimensional (2D) separations of ONs where there is more or less freedom to choose parameters from a wide range of possibilities depending on the mode of 2D separation that is used.