The use of 31P NMR and principal components analysis to determine sameness or differences between diastereomeric composition of phosphorothioate oligonucleotides

• Phosphorothioate (PS) groups are introduced in synthetic oligonucleotides therapeutics; this leads to formation of mixtures of diastereomers, which are indistinguishable to most analytical methods. For Mongersen, a 21-mer, there are 2 20 ≈ 1 million diastereomers. They can display different biological activity. • 31 P-NMR is specific for phosphorous centres and provides distinct signals for (R)/(S) configured PS linkages in a oligonucleotide. The 31 P-NMR is the envelope of the 20 × 2 20 phosphorous signals. The first two principal components of PCA explain about 90% of the variance of the spectra. • The repeatability and reproducibility of 31 P-NMR are assessed by analysing specimens of 2 different production batches. The confidence interval of the 31 P-NMR spectra and of the subsequent PC1 and PC2 are determined. • Two independent teams of operators analysed the same production batches on two different NMR instruments at two different fields, resulting in (PC1,PC2) scores which are within the confidence intervals. • The analytical method is extremely simple, it only requires access to a high field NMR (at least 11.7T/500 MHz) instrument for a few hours, a few mg of substance are required, practically no handling (only dissolving in deuterated water). • The analytical method is robust and can be operated in different laboratories providing results that remain within the tolerance. Synthetic oligonucleotides often contain chemically modified backbones to prevent degradation and improve bioactivity. Phosphorothioate (PS) linkage in place of phosphodiester is the most used one. Substituting an oxygen for a sulfur atom introduces a stereogenic element on each PS, which leads to an exponentially large number of diastereomers, if the synthetic process is not stereoselective. Variable conditions during synthesis may lead to unpredictable variation in the manifold of diastereomers, with heavy fallouts in bioactivity. We developed an analytical protocol based on 31 P-NMR followed by principal components analysis (PCA) to fingerprint each manifold. We assess the precision of this method by investigating repeatability and reproducibility. By working on two instruments at different magnetic fields, we also assess the problem of robustness and ultimately the possibility of exchanging data between laboratories. We describe and critically assess a method for scoring the stereoisomer composition of phosphorothioate oligodeoxynucleotides based on 31 P NMR spectroscopy followed by principal components analysis. The method is operated on real drug substance samples , ready for clinical tests and reproduced by two independent teams of operators on two different NMR instruments.