Solution-state methyl NMR spectroscopy of large non-deuterated proteins enabled by deep neural networks

• Published in: Nature communications 2024-06, Vol.15 (1), p.5073-12, Article 5073
• Main Authors: Karunanithy, Gogulan, Shukla, Vaibhav Kumar, Hansen, D. Flemming
• Format: Article
• Language: English
• Subjects: 101/6; 119/118; 140/131; 631/114/1305; 631/535/878/1263; 631/57/2266; Artificial neural networks; Biomolecules; Carbon Isotopes - chemistry; Deep Learning; Deuteration; E coli; Escherichia coli - metabolism; Escherichia coli Proteins - chemistry; Escherichia coli Proteins - metabolism; Humanities and Social Sciences; Machine learning; Magnetic resonance spectroscopy; Magnetic Resonance Spectroscopy - methods; Malate synthase; Malate Synthase - chemistry; Malate Synthase - metabolism; multidisciplinary; Neural networks; Neural Networks, Computer; NMR; NMR spectroscopy; Nuclear magnetic resonance; Nuclear Magnetic Resonance, Biomolecular - methods; Proteins; Proteins - chemistry; Proteins - metabolism; Science; Science (multidisciplinary); Spectra; Spectroscopy; Spectrum analysis
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-024-49378-8

Methyl-TROSY nuclear magnetic resonance (NMR) spectroscopy is a powerful technique for characterising large biomolecules in solution. However, preparing samples for these experiments is demanding and entails deuteration, limiting its use. Here we demonstrate that NMR spectra recorded on protonated, uniformly 13 C labelled samples can be processed using deep neural networks to yield spectra that are of similar quality to typical deuterated methyl-TROSY spectra, potentially providing information for proteins that cannot be produced in bacterial systems. We validate the methodology experimentally on three proteins with molecular weights in the range 42–360 kDa. We further demonstrate the applicability of our methodology to 3D NOESY spectra of Escherichia coli Malate Synthase G (81 kDa), where observed NOE cross-peaks are in good agreement with the available structure. The method represents an advance in the field of using deep learning to analyse complex magnetic resonance data and could have an impact on the study of large biomolecules in years to come. Here, the authors show deep neural networks can be trained to transform crowded 1 H, 13 C-NMR spectra of large proteins into readily interpretable spectra, negating the requirement for uniform deuteration to analyse complex NMR spectra.