Quantifying protein dynamics in the ps–ns time regime by NMR relaxation

Abstract

Both 15N chemical shift anisotropy (CSA) and sufficiently rapid exchange linebroadening transitions exhibit relaxation contributions that are proportional to the square of the magnetic field. Deconvoluting these contributions is further complicated by residue-dependent variations in protein amide 15N CSA values which have proven difficult to accurately measure. Exploiting recently reported improvements for the implementation of T1 and T1ρ experiments, field strength-dependent studies have been carried out on the B3 domain of protein G (GB3) as well as on the immunophilin FKBP12 and a H87V variant of that protein in which the major conformational exchange linebroadening transition is suppressed. By applying a zero frequency spectral density rescaling analysis to the relaxation data collected at magnetic fields from 500 MHz 1H to 900 MHz 1H, differential residue-specific 15N CSA values have been obtained for GB3 which correlate with those derived from solid state and liquid crystalline NMR measurements to a level similar to the correlation among those previously reported studies. Application of this analysis protocol to FKBP12 demonstrated an efficient quantitation of both weak exchange linebroadening contributions and differential residue-specific 15N CSA values. Experimental access to such differential residue-specific 15N CSA values should significantly facilitate more accurate comparisons with molecular dynamics simulations of protein motion that occurs within the timeframe of global molecular tumbling.

13C NMR Relaxation Analysis of Protein GB3 for the Assessment of Side Chain Dynamics Predictions by Current AMBER and CHARMM Force Fields

“…The [2- 13 C]glycerol labeled sample of GB3 yielded 29 1 H α − 13 C α bond vectors for which the seven force fields tested all predict S 2 values greater than 0.85. Utilizing the previously determined axial ratio of 1.35 for its diffusion tensor 37,66 and the fact that the experimentally determined principal axis of the diffusion tensor differs by only ∼5°from the calculated inertial tensor, 67 the effective 1 H α − 13 C α bond length and average global rotational diffusion time were jointly optimized to yield a best fit to the experimental R 1 and R 2 data. This optimization is facilitated by the fact that variations in the bond length proportionally alter the predicted values for both R 1 and R 2 , while changes in the tumbling time have opposite effects on the R 1 and R 2 values for these relatively immobilized 1 H α − 13 C α bond vectors.…”

“…An E. coli-optimized codon sequence for the Δ1-5, T6M, T7Q variant of the B3 domain of streptococcal protein G (Genscript) was cloned into the expression vector pET11a. 37 After transformation of this plasmid into a T7 promoter-containing variant of the E. coli strain DL323 38 lysogenized with the phage T7 promoter expression system, two samples of isotopically enriched GB3 were prepared by growth on a carbon source of either [70% U-2 H,2-13 C]glycerol + NaH 13 CO 3 or [70%U-2 H,1,3-13 C 2 ]glycerol in minimal media containing 70% 2 H 2 O. 38 For the growth on [70%U-2 H,1,3-13 C 2 ]glycerol, compressed air containing 5% CO 2 was bubbled through the medium during growth to suppress the reincorporation of biosynthetically generated 13 CO 2 .…”

“…39 The GB3 samples were purified and concentrated to 2 mM protein in 25 mM NaH 2 PO 4 buffer (pH 6.3) containing 6% 2 H 2 O and 0.01% sodium azide as previously described. 37 2.4. NMR Spectroscopy.…”

“…The 13 C enriched glycerol compounds (Cambridge Isotopes) were catalytically deuterated by refluxing with Raney nickel . The GB3 samples were purified and concentrated to 2 mM protein in 25 mM NaH 2 PO 4 buffer (pH 6.3) containing 6% 2 H 2 O and 0.01% sodium azide as previously described …”

Prediction of Bond Vector Autocorrelation Functions from Larmor Frequency-Selective Order Parameter Analysis of NMR Relaxation Data

“…In comparing the dynamics modeled by MD simulations with the experimentally observed NMR relaxation values, the theoretical nuclear dipolar coupling factor must be rescaled to reflect the fact that MD simulations standardly utilize rigid models for hydrogen-containing covalent bonds so that the contributions to relaxation arising from high frequency classical/quantum mechanical librations are not adequately predicted . While this rescaling of the theoretical coupling factor to optimally match the experimental data is typically expressed in terms of an altered effective 1 H– 15 N bond length, in reality any error in the assumed average 15 N chemical shift anisotropy (CSA) value for NMR data at a given field strength is simultaneously corrected for as well . Effective bond length optimizations for the 1 H– 15 N data were carried out under the assumption of a uniform 15 N CSA value of 168 ppm .…”