No abstract

We report here an approach for predicting charge distributions in molecules for use in molecular dynamics simulations. The input data are experimental atomic ionization potentials, electron affinities, and atomic radii. An atomic chemical potential is constructed by using these quantities plus shielded electrostatic interactions between all charges. Requiring equal chemical potentials leads to equilibrium charges that depend upon geometry. This charge equilibration (QEq) approach leads to charges in excellent agreement with experimental dipole moments and with the atomic charges obtained from the electrostatic potentials of accurate ab initio calculations. QEq can be used to predict charges for any polymer, ceramic, semiconductor, or biological system, allowing extension of molecular dynamics studies to broad classes of new systems. The charges depend upon environment and change during molecular dynamics calculations. We indicate how this approach can also be used to predict infrared intensities, dielectric constants, and other charge-related properties.

A representative set of high resolution x-ray crystal structures of nonhomologous proteins have been examined to determine the preferred positions and orientations of noncovalent interactions between the aromatic side chains of the amino acids phenylalanine, tyrosine, histidine, and tryptophan. To study the primary interactions between aromatic amino acids, care has been taken to examine only isolated pairs (dimers) of amino acids because trimers and higher order clusters of aromatic amino acids behave differently than their dimer counterparts. We find that pairs (dimers) of aromatic side chain amino acids preferentially align their respective aromatic rings in an off-centered parallel orientation. Further, we find that this parallel-displaced structure is 0.5-0.75 kcal/mol more stable than a T-shaped structure for phenylalanine interactions and 1 kcal/mol more stable than a T-shaped structure for the full set of aromatic side chain amino acids. This experimentally determined structure and energy difference is consistent with ab initio and molecular mechanics calculations of benzene dimer, however, the results are not in agreement with previously published analyses of aromatic amino acids in proteins. The preferred orientation is referred to as parallel displaced -stacking.Attractive nonbonded interactions between aromatic rings are seen in many areas of chemistry, and hence are of interest to all realms of chemistry. Porphyrin aggregation (1), the conformation of diarylnaphthalenes (2) and phenylacetylene macrocycles (3), and the strength of Kevlar (4) can be attributed, at least in part, to aromatic-aromatic interactions. Aromatic-aromatic interactions have been implicated in catalytic hydroformylation (5), the catalytic formation of elastomeric polypropylene (6), and the asymmetric cis dihydroxylation of olefins (7). The vast majority of medicinal agents contain aromatic substituents and their differential recognition by proteins is likely dominated by aromatic-aromatic interactions (8). In biologically related areas of chemistry, aromatic-aromatic interactions are crucially involved in protein-deoxynucleic acid complexes where interactions between aromatic residues and base pairs are seen in x-ray crystal structures (9, 10).Because aromatic-aromatic interactions are so prevalent across chemistry, a large body of experimental and theoretical work has focused on determining the gas phase structure of the prototype, benzene dimer (11-14, 30). As summarized recently by Sun and Bernstein (15), the experimentally observed structure depends heavily upon the observation technique. Off-centered parallel displaced, 1p, and T-shaped, 1t, structures are the most commonly cited orientations (Structure 1).Large scale ab initio electronic structure theory suggests that the off-centered parallel displaced and T-shaped structures are nearly isoenergetic (11)(12)(13)(14)30). As reported by Sun and Bernstein (15), empirical force field studies favor either off-centered parallel displaced or T-shaped structures depending upon t...

We present calculations for the nonbonded interactions in the dimeric complexes:  methane dimer, ammonia dimer, water dimer, H2O·(NH3), CH4·(NH3), and (FHF)- as a function of theory level (HF, DFT(B3LYP), MP2, LMP2, MP3, MP4, CCSD(T), and others) and basis set (6-31G**, cc-pVXZ, X = D, T, Q, 5). Dimer minimum energy structures are determined at the MP2 theory level for the cc-pVTZ basis set employing analytical second derivatives. For HF and DFT levels of theory, methane dimer and one structure of CH4·(NH3) are not bound. The basis set superposition error (BSSE) begins to converge (becomes systematically small) for basis sets larger than cc-pVTZ. For hydrogen-bonded systems, most levels of theory seem to give reasonable estimates of the experimentally known binding energies, but here, too, the BSSE overwhelms the reliability of the binding energies for the smaller basis sets. The CH4·(NH3) dimer has two minimum energy conformations with similar binding energies, but very different BSSE values especially for small basis sets (cc-pVXZ, X ≤ T). On the basis of these calculations, we present a discussion of ab initio calculations of nonbonded interactions for molecules, such as phenethylamine, that have different conformations. Suggestions for possible next steps in the calculation of nonbonded interactions are presented.

No abstract