Structure refinement using time-averaged J-coupling constant restraints

Abstract

We describe a new penalty function for use in restrained molecular dynamics simulations which allows experimental J-coupling information to be enforced as a time-averaged, rather than instantaneous, quantity. The pseudo-energy term has been formulated in terms of a calculated J value (a measured quantity) rather than the relevant dihedral angle (a derived quantity). This accounts for the distinct non-linearity of the coupling constant with respect to either Cartesian coordinates or dihedral angles. Example simulations of the cyclic decapeptide antamanide show the procedure's ability to enforce experimental restraints while exploring a large region of conformational space and producing a relatively small disturbance of the physical force field.

On using oscillating time-dependent restraints in MD simulation

“…In our work we generally find decreased /-angle rms fluctuations (especially for high force constants) for IR and increased rms fluctuations for TA for both the 3 J-constant and the /-angle. Both effects are expected and have been described before (Torda et al, 1993;Scott et al, 1998). From the rms fluctuations of the 3 J-values no clear trend as to whether BTA and OBTA increase or decrease the fluctuations can be distinguished.…”

“…Previous studies using instantaneous restraining have found no single conformation which could explain the 3 J-values measured (Kessler et al, 1988(Kessler et al, , 1989Bru¨schweiler et al, 1991). It was previously used to test J-value restraining methodology (Torda et al, 1993;Scott et al, 1998). One antamanide molecule immersed in 3002 water molecules in a cubic periodic box with edge length 4.488 nm was simulated at constant volume and temperature, similarly to the butane system.…”

“…Equilibration covered 100 ps and sampling the next 400 ps. This system set-up differs from that in (Torda et al, 1993;Scott et al, 1998): (i) Here solvent molecules are explicitly simulated, whereas in (Scott et al, 1998) their effect was mimicked by stochastic and frictional forces, (ii) the factor ð1 À e t=s Þ was approximated by t=s in (Scott et al, 1998), and (iii) in Scott et al (1998) an older set of force field parameters, set 43A1 (van Gunsteren et al, 1996a;Daura et al, 1998) was used.…”

“…This approach has been successfully applied in structure refinement based on NMR (Torda et al, 1990;Torda et al, 1993) or X-ray (Schiffer et al, 1995) data. This conventional timeaveraging refinement relies on employing a highquality physical force field, because it introduces more conformational freedom for the molecule.…”

Protein structure calculation with data imputation: the use of substitute restraints

“…The structural information contained in NOEs reports on pairwise distances between specific protons and can thus provide unequivocal information about the relative spatial locations of different residues in a protein sequence (Wüthrich 1986). Other experimental information derived from J-couplings (Pardi et al 1984;Kim and Prestegard 1990;Torda et al 1993;Garrett et al 1994), chemical shifts (Cavalli et al 2007;Shen et al 2008), and residual dipolar couplings (Tolman et al 2001;Qu et al 2004;Rathinavelan and Im 2008) has also been used to further improve the quality of NMR structures. Despite these new types of experimental data, distance restraints have remained the single most valuable source of information for the elucidation of high-resolution solution structures by NMR spectroscopy, and therefore, traditional NMR structure determination programs such as CNS (Brunger et al 1998;Brunger 2007) or CYANA (Guntert 2004) require a large number of redundant NOE restraints-typically 15-20 NOE restraints per residue-to obtain a high resolution structure.…”