For extended objects, such as amyloid fibrils, diffusion NMR experiments become difficult to interpret because in addition to translational diffusion they are also sensitive to rotational diffusion. We have constructed a mathematical theory describing the outcome of PFG NMR experiments on rod-like fibrils. The effect of rotational diffusion is indeed significant, accounting for up to 30% of the diffusion-induced signal loss. We have tested the validity of our theory on Sup35NM fibrils, where the presence of disordered M domain makes it possible to observe spectral signals from the fibrils by solution NMR. The experimental results are in good agreement with the theoretical predictions, paving the way for diffusion-based investigation of complex amyloidogenic systems.
Oftentimes, diffusion NMR measurements are conducted on samples with low protein concentration and/or low stability, which complicates accurate determination of diffusion parameters. The situation is compounded by baseline distortions, caused by residual solvent signals or by instrumental reasons. In response to these challenges, we have developed a new data processing scheme to treat the data from 1D diffusion NMR experiments. The algorithm, which uses the ideas from the theory of optimal filtration, has been implemented in a form of web server named DDfit (Diffusion Data fit, https://ddfit.bio-nmr.spbu.ru). In our tests, the new processing scheme proved to be both more accurate and more precise compared to the standard schemes available in the programs TopSpin and MestreNova.
Diffusion NMR potentially offers an insight into conformational landscape of intrinsically disordered proteins. In particular, diffusion data can be used to validate MD models of IDPs. Along these lines, we have shown that MD simulations of the disordered N‑terminal tail of the histone H4 using TIP4P-D and OPC water models are consistent with the experimental diffusion data. In contrast, MD simulations in TIP4P-Ew water produce overly compact conformational species (as also corroborated by the 15N relaxation data). In our work, the diffusion coefficients have been extracted directly from the series of MD trajectories recorded in the simulation cells of increasing size; alternatively, application of the program HYDROPRO to the MD snapshots can lead to misleading results.