MOLECULAR-DYNAMICS OF THE HA-RAS PROTEIN - NUCLEOTIDE ATOM-CENTERED CHARGES WITHIN THE AMBER FORCE-FIELD

Citation
Ga. Worth et al., MOLECULAR-DYNAMICS OF THE HA-RAS PROTEIN - NUCLEOTIDE ATOM-CENTERED CHARGES WITHIN THE AMBER FORCE-FIELD, JOURNAL OF MOLECULAR MODELING, 1(3), 1995, pp. 123-142
Citations number
50
Categorie Soggetti
Biophysics,Biology,Chemistry
Journal title
JOURNAL OF MOLECULAR MODELING
ISSN journal
16102940 → ACNP
Volume
1
Issue
3
Year of publication
1995
Pages
123 - 142
Database
ISI
SICI code
1610-2940(1995)1:3<123:MOTHP->2.0.ZU;2-R
Abstract
Molecular dynamics simulations have become an essential tool for the s tudy of biological systems. The Ha-vas protein, is a system suitable f or such studies. Despite much recent progress, it is stilt not known e xactly how the protein functions in the cell growth cycle. In this wor k atom-centred point charges for the guanosine nucleotide ligands are calculated and tested. To be compatible with the other AMBER force fie ld parameters these are fitted to a molecular electrostatic potential derived from an ab initio wavefunction. The smallest basis set able to produce a stable wavefunction for the negatively charged GDP and GTP molecule ions was 3-21G with diffuse functions added on the phosphate groups. To maintain force field integrity these charges were scaled t o be equivalent to STO-3G derived values. This procedure is seen to pr oduce a good magnesium-phosphate interaction potential when compared t o 6-31++G ab initio calculations. With the nucleotides fixed in the b inding site conformation, it was found essential to include the electr ostatics of the binding site in the calculation of the charges. It was also found to be inappropriate to divide the nucleotide into constitu ent parts for the calculations. From the calculated charges and experi mental data, the nucleotide protonation states in the protein are dedu ced. It is unlikely that GDP is protonated, GTP probably binds one pro ton. The charges were tested in MD simulations of a protein modelled o n the crystal structure of Tong et al. [1], during which the dynamics of the nucleotide and binding site residues were in good agreement wit h the crystal structure data. The model is seen to be sensitive, not o nly to the inclusion of explicit solvent, but to the number of waters ligating the magnesium ion and the conformation of the loop between re sidues 60 and 66; both pieces of information are lacking in the crysta l structure data.