Protein force fields transferability

While simulations reach into larger time spans, the inaccuracies of force fields become more apparent on the one hand properties based on free energies, which were never used for parametrization, are computed more accurately and discrepancies show up on the other hand longer simulations, particularly of proteins, show more subtle discrepancies that only appear after nanoseconds. Thus force fields are under constant revision as far as their parameters are concerned, and this process will continue. Unfortunately the form of the potentials is hardly considered and the refinement leads to an increasing number of distinct atom types with a proliferating number of parameters and a severe detoriation of transferability. The increased use of quantum mechanics to derive potentials will not really improve this situation ab initio quantum mechanics is not reliable enough on the level of kT, and on-the-fly use of quantum methods to derive forces, as in the Car-Parrinello method, is not likely to be applicable to very large systems in the foreseeable future. 

Studies of ferredoxin [152] and a photosynthetic reaction center [151] have analyzed further the protein s dielectric response to electron transfer, and the protein s role in reducing the reorganization free energy so as to accelerate electron transfer [152], Different force fields were compared, including a polarizable and a non-polarizable force field [151]. One very recent study considered the effect of point mutations on the redox potential of the protein azurin [56]. Structural relaxation along the simulated reaction pathway was analyzed in detail. Similar to the Cyt c study above, several slow relaxation channels were found, which limited the ability to obtain very precise free energy estimates. Only semiquantitative values were... 

T. Herges and W. Wenzel. Reproducible in-silico folding of a three-helix protein in a transferable all-atom force field. Physical Review Letters (in press), http //www.arXiv.org physics/0310146, 2004. 

Hybrid multiscale models enable us to focus on the relevant part of a system. For example, Leenders et al. studied the proton transfer process in the photoactive yellow protein (Figure 6.3) [9], They used Car-Parrinello molecular dynamics [10], a QM method for dynamics simulations, to describe the chromophore and its hydrogen-bonded network in the protein pocket (middle and right-hand circles). This was combined with a traditional MD force field of 28 600 atoms, simulating the entire protein in water (left-hand circle). 

Quantum dynamics effects for hydride transfer in enzyme catalysis have been analyzed by Alhambra et. al., 2000. This process is simulated using canonically variational transition-states for overbarrier dynamics and optimized multidimensional paths for tunneling. A system is divided into a primary zone (substrate-enzyme-coenzyme), which is embedded in a secondary zone (substrate-enzyme-coenzyme-solvent). The potential energy surface of the first zone is treated by quantum mechanical electronic structure methods, and protein, coenzyme, and solvent atoms by molecular mechanical force fields. The theory allows the calculation of Schaad-Swain exponents for primary (aprim) and secondary (asec) KIE... 

While the structure and force field uniquely determine the vibrational frequencies of the molecule, the structure cannot in general be obtained directly from the spectrum. However, to a useful approximation, the atomic displacements in many of the vibrational modes of a large molecule are concentrated in the motions of atoms in small chemical groups, and these localized modes are to a good approximation transferable between molecules. Therefore, in the early studies of peptides and proteins (Sutherland, 1952), efforts were directed mainly to the identification of such characteristic frequencies and the determination of their relation to the structure of the molecule. This kind of analysis depended on empirical correlations of the spectra of chemically similar molecules. 

---