Prediction of Collision-Induced Dissociation Spectra of Deprotonated Oligonucleotides.

Abstract

Developing DNA/RNA-based therapeutics (siRNA, antisense oligonucleotides, mRNA, etc.) requires comprehensive tools for structural characterization. MS/MS remains a main tool for structural characterization of short synthetic DNA/RNA or oligonucleotides produced from the digestion of longer mRNA. To facilitate reliable oligonucleotide identification by MS/MS, a mechanistic model similar to the previously described peptide fragmentation model is developed. The model simulates the kinetics of the oligonucleotide fragmentation process, accounting for competing pathways including charge-remote and charge-directed base losses and backbone cleavages. Oligonucleotides implemented in the model include regular DNA and RNA oligonucleotides, as well as synthetic RNA with 2'-fluoro and 2'-O-methyl modifications. In addition to the five common bases (A, C, G, T, and U), methoxyuracil, N1-methyl pseudouracil, pseudouracil, dihydrouracil, hypoxanthine, and abasic groups are also considered. Phosphate linkages include normal phosphodiester, phosphorothioate, inverted phosphodiester, and inverted phosphorothioate. The model is trained on over 19 000 MS/MS spectra of known deprotonated oligonucleotide precursors and evaluated by k-fold cross validation. The optimized model is capable of predicting the MS/MS spectra of deprotonated oligonucleotides across various collision energies.