Considerations for deconvolution of oligonucleotide mass spectra in quality control

Abstract

Oligonucleotide therapeutics and mass spectrometry are evolving fields in both research and pharmaceutical development. A proper data evaluation needs to be ensured, especially in a regulated environment. Deconvolution is often applied to derive the intact mass from electrospray ionization (ESI) mass spectra. The type of deconvolution algorithm applied as well as the used parameters can affect the outcome of the analysis. This study aims to show differences in the deconvolution algorithm in terms of mass accuracy and how the input mass range affects the mass accuracy and the detected abasic impurities and adducts for polydeoxythymidine (poly-dT) samples with 20, 48, 100 and 200 nucleotides (nts). While Bayesian deconvolution is more suitable for analytes up to ∼15 000 Da in terms of mass accuracy, maximum entropy deconvolution fits best for larger oligonucleotides, with a mass accuracy of 1.6 ppm for a 100 (nts) poly-dT sample. The input mass range is the second important parameter to be tested, as low charge states tend to show more adducts and higher charge states tend to be prone for depurination and depyrimidation, leading to an overrepresentation of the impurity mass in the deconvoluted spectrum. This manuscript also highlights how the findings can be translated to a short DNA-based adjuvant and a sample single guide RNA.