We present the development of a force field for simulation of nucleic acids and proteins. Our approach began by obtaining equilibrium bond lengths and angles from microwave, neutron diffraction, and prior molecular mechanical calculations, torsional constants from microwave, NMR, and molecular mechanical studies, nonbonded parameters from crystal packing calculations, and atomic charges from the fit of a partial charge model to electrostatic potentials calculated by ab initio quantum mechanical theory. The parameters were then refined with molecular mechanical studies on the structures and energies of model compounds. For nucleic acids, we focused on methyl ethyl ether, tetrahydrofuran, deoxyadenosine, dimethyl phosphate, 9-methylguanine-1 -methylcytosine hydrogen-bonded complex, 9-methyladenine-1 -methylthymine hydrogen-bonded complex, and 1,3-dimethyluracil base-stacked dimer. Bond, angle, torsional, nonbonded, and hydrogen-bond parameters were varied to optimize the agreement between calculated and experimental values for sugar pucker energies and structures, vibrational frequencies of dimethyl phosphate and tetrahydrofuran, and energies for base pairing and base stacking. For proteins, we focused on , ' maps of glycyl and alanyl dipeptides, hydrogen-bonding interactions involving the various protein polar groups, and energy refinement calculations on insulin. Unlike the models for hydrogen bonding involving nitrogen and oxygen electron donors, an adequate description of sulfur hydrogen bonding required explicit inclusion of lone pairs.

We describe a computer program we have been developing to build models of molecules and calculate their interactions using empirical energy approacaes. The program is sufficiently flexible and general to allow modeling of small molecules, as well as polymers. As an illustration, we present applications of the program to study the conformation of actir iycin D. In particular, we study the rotational isomerism about the D-Val-, L-Pro, and L-Pro-Sar dmide bonds as well as comparing the energy and structure of the Sobell model and the x-ray structure of actinomycin D.

Electrostatic forces have been implicated in a variety of biologically important molecular interactions including drug orientation by DNA, protein folding and assembly, substrate binding and catalysis and macromolecular complementarity with inhibitors, drugs and hormones. To examine enzyme-substrate interactions in copper, zinc superoxide dismutase (SOD), we developed a method for the visualization and analysis of an enzyme's three-dimensional electrostatic vector field that allows the contributions of specific residues to be identified. We report here that the arrangement of electrostatic charges in SOD promotes productive enzyme-substrate interaction through substrate guidance and charge complementarity: sequence-conserved residues create an extensive electrostatic field that directs the negatively charged superoxide (O-2) substrate to the highly positive catalytic binding site at the bottom of the active-site channel. Dissection of the electrostatic potential gradient indicated the relative contributions of individual charged residues: Lys 134 and Glu 131 seem to have important roles in directing the long-range approach of O-2, while Arg 141 has local orienting effects. The reported methods of analysis may have general application for the elucidation of intermolecular recognition processes.

Color-coded computer graphics representations of the electrostatic potentials of trypsin, trypsin-inhibitor, prealbumin and its thyroxine complex, fragments of double-helical DNA, and a netropsin-DNA complex illustrate the electrostatic and topographic complementarity in macromolecule-ligand interactions. This approach is powerful in revealing intermolecular specificity and shows promise of having predictive value in drug design.The electrostatic potential is a powerful tool that has provided insights into intermolecular association and molecular properties of small molecules (1), actions of drug molecules and their analogs (2), the biological function of hemoglobin (3), and enzyme catalysis (4). We thought that, given the electrostatic potentials ofboth a ligand and its receptor, much greater progress might be made in the "rational design" of the optimal ligand (5-7). Subsequent studies of the electrostatic potentials of macromolecules (8-10) demonstrate the feasibility ofthis approach, but there has been no satisfactory way to represent the electrostatic potential in a way useful for understanding electrostatic complementarity. This problem is effectively solved by displaying the electrostatic potential molecular surface with a realtime interactive color graphics facility (11) using the molecular surface definition of Richards (12) What should one use for the ionization state ofexposed acidic and basic residues (9)? An assumption that all residues are in their predominant state of ionization in aqueous solution at pH 7 exaggerates the importance of the charged residues, whereas an assumption of neutrality ignores the effects of the charged groups. Here we adopt the following compromise: for protein calculations the charges were retained, and for nucleic acids we either explicitly included counterions or neutralized the phosphate charges.The molecular surfaces for each structure were calculated from x-ray and energy-refined coordinates (19-21) by using the program MS (22). The electrostatic potential was calculated 1.4 A along the surface normal vector from a given molecular surface point and represented at the surface point itself. We calculated the contribution of each atom within 20 A of the electrostatic potential point F and used the method of neutral spheres (23, 24), which corrects for dipoles split by the finite cutoff radius by adding an additional interaction (at that radius) to ensure that the net charge inside the sphere is zero. The surfaces were displayed at the University of California (San Francisco) Computer Graphics Laboratory with the display program HANDLE (written by M. Connolly). RESULTSTrypsin-Trypsin Inhibitor. First we consider the prototype protein-protein interaction of trypsin and bovine pancreatic trypsin inhibitor (BPTI) (25). There is a large negative pocket (due to in trypsin that is the binding pocket for the positively charged Lys-15 on the lower right of BPTI (Fig.

No abstract