Fast NMR-Based Assessment of Cancer-Associated Protein Glycosylations from Serum Samples

Nuclear magnetic resonance (NMR) spectra of blood serum and plasma show signals arising from metabolites, lipoproteins, and <i>N</i>-acetyl methyl groups of <i>N</i>-glycans covalently linked to acute-phase proteins. These glycan signals often called glycoprotein A (GlycA) and glycoprotein B (GlycB) arise from <i>N</i>-acetyl methyl groups and have been proposed as biomarkers, initially for cardiovascular diseases, but also for other inflammatory conditions. For the detection of glycan resonances, <i>J</i>-edited, diffusion, and relaxation filtered NMR spectroscopy (JEDI) has been proposed to suppress the lipoprotein signals. JEDI is however limited to measure those acetyl signals, whereas all other glycan resonance cannot be observed. For improved glycoprotein profiling, the signals arising from the pyranose ring protons are essential. Here, we show how selective frequency excitation combined with scalar coupling filtering can be used to dramatically increase the number of <i>N</i>-glycan signals observable in NMR spectra of serum and plasma samples, facilitating glycosylation profiling in less than 30 min. This approach grants selective detection of sialylation, galactosylation, <i>N</i>-acetylglucosaminylation, and fucosylation of dominant <i>N</i>-glycans and, to some extent, <i>N</i>-glycan branching complexity. Notably, sialylated and nonsialylated Lewis^x and Lewis^a antigens can also be observed. Lewis^a antigen is well established as a cancer biomarker, known as CA19-9. NMR glycosylation profiles from nine isolated serum glycoproteins show excellent agreement with well-established UHPLC-MS analysis. The proposed NMR method facilitates the detection of glycoprotein biomarkers without the need for enzymatic treatment of serum or plasma and provides a robust read-out as exemplified by samples from 33 patients with hepatocellular carcinoma.