Fragment-Based Approach for the Evaluation of NMR Chemical Shifts for Large Biomolecules Incorporating the Effects of the Solvent Environment.

We present an efficient implementation of the molecules-in-molecules (MIM) fragment-based quantum chemical method for the evaluation of NMR chemical shifts of large biomolecules. Density functional techniques have been employed in conjunction with large basis sets and including the effects of the solvent environment in these calculations. The MIM-NMR method is initially benchmarked on a set of (alanine)10 conformers containing strong intramolecular interactions. The incorporation of a second low level of theory to recover the missing long-range interactions in the primary fragmentation scheme is critical to yield reliable chemical shifts, with a mean absolute deviation (MAD) from direct unfragmented calculations of 0.01 ppm for 1H chemical shifts and 0.07 ppm for 13C chemical shifts. In addition, the performance of MIM-NMR has been assessed on two large peptides: the helical portion of ubiquitin ( 1UBQ ) containing 12 residues where the X-ray structure is known, and E6-binding protein of papilloma virus ( 1RIJ ) containing 23 residues where the structure has been derived from solution-phase NMR analysis. The solvation environment is incorporated in these MIM-NMR calculations, either through an explicit, implicit, or a combination of both solvation models. Using an explicit treatment of the solvent molecules within the first solvation sphere (3 Å) and an implicit solvation model for the rest of the interactions, the 1H and 13C chemical shifts of ubiquitin show excellent agreement with experiment (mean absolute deviation of 0.31 ppm for 1H and 1.72 ppm for 13C), while the larger E6-binding protein yields a mean absolute deviation of 0.34 ppm for 1H chemical shifts. The proposed MIM-NMR method is computationally cost-effective and provides a substantial speedup relative to conventional full calculations, the largest density functional NMR calculation included in this work involving more than 600 atoms and over 10,000 basis functions. The MIM-NMR solvation protocols developed in this work may pave the way for very accurate de novo prediction of NMR chemical shifts of a range of large biomolecules in the future.