Susan Q. Newton

The case of glycine continued: some contradictory SCF results

Abstract The energies of five conformations of glycine were determined by ab initio HF/SCF gradient geometry optimization using 13 different basis sets. The conformations considered are the stretched form ( I ), a heavy-atom framework planar ( II ) and non-planar ( III ) cyclic form, and a planar ( IV ) and non-planar ( V ) bifurcated structure. Form I is the global energy minimum in all calculations, but contradictory results are obtained for the stabilities of the other forms. Of the cyclic forms, II is a local energy minimum at the 4-31G, 6-31G, 6-31+G, and (10s,5p/4s) levels, but a saddle point in calculations with both smaller and larger basis sets (including 6-31G ∗∗ , 6-31 + G ∗∗ , and 6-31 + + G ∗∗ ). Similarly contradictory results are obtained for IV . There is considerable scatter in calculated energy differences and in the optimized values of the non-planar N-C-CO torsions. Form I and the planar cyclic form II were previously found in the microwave spectrum of glycine. The significance of the new calculations for interpretations of the microwave data is discussed. It is concluded that assumption of slight non-planarity of the cyclic form, with a minimum at III and a low lying transition state at II is not necessarily in conflict with the available microwave data. In addition it is possible to conclude that a third conformer, IV or V , hitherto not observed, may exist in glycine vapors in amounts which are similar to those of the observed conformer II .

Ab initio conformational analysis of alanine

Abstract The ab initio geometries of 13 conformations of alanine were optimized without any constraints at the HF/6-31G ∗∗ , HF/6-311G ∗∗ , MP2/6-31G ∗∗ , and MP2/6-311G ∗∗ levels of theory. By comparing structures devoid of electron correlation (i.e. HF-optimized) with the corresponding structures in which correlation was switched on (i.e. MP2-optimized), the structural effects of electron correlation can be detected. It is found that the magnitude of the effects (up to 0.03 A in bond distances and about 3 ° in angles) follows the sequence C  O > C  N . C  C , and H  O  C > H  N  C > H  C  C . For the assignment of the spectroscopic data in a previous microwave study of alanine, it was an important finding that, at the HF-computational level, only the 6-31G ∗∗ geometry of conformation 1 agreed with one of two experimental sets of rotational constants and dipole moment components. In contrast, we find that, at the MP2-computational level, a second less stable conformer, form 5 , is also in good agreement with the same set of rotational constants and dipole components. Thus, observation of conformer 1 rather than form 5 can be established only by reference to additional evidence; e.g. the calculated energies, the gas electron diffraction data of alanine, and the most stable form found for glycine. Finally, in a comparison of alanine with alanine dipeptide, the NCCO torsional angles of the characteristic low energy regions of peptides, C eq 7 , C 5 , C ax 7 , α R , and α L , are found close to the minimum energy regions of the free acid studied here.

Molecular dynamics simulations of the adsorption of proteins on clay mineral surfaces

Abstract Some initial results of molecular dynamics simulations of the adsorption of proteins on clay mineral surfaces are being reported. Specifically, the interactions of pyrophyllite surfaces with crambin, rubredoxin, and several oligopeptides were investigated. It is found that clay mineral surfaces can have a denaturing effect on adsorbed proteins for two reasons: (1) they are dehydrating agents, because they perturb the random environment of water molecules that globular proteins need to maintain their native structure; and (2) clay surfaces can establish non-bonded interactions with proteins which compete effectively with the interactions inside a peptide chain. The changes in secondary and tertiary protein structure induced by adsorption to pristine surfaces, or surfaces coated with water, lead to backbone torsions away from the most populated regions of φ,ψ-space, to regions which are not frequently populated in unperturbed proteins. β- and α R -conformations, specifically, are not stable on pyrophyllite but undergo transitions, with some preference, to an area close to C 7 eq . Because of the size of the adsorbed systems, unit cells with adsorbed peptides may be distorted, bulging at the site of adsorption, and displaying a continuously varying interlayer space between the empty parts and those that are occupied by an adsorbate. As a result, warped, or S-shaped basal planes are found.

Molecular dynamics simulations of the adsorption of methylene blue at clay mineral surfaces.

Molecular dynamics simulations were performed of the adsorption of methylene blue (MB) on model beidellite, montmorillonite, and muscovite mica surfaces, using a previously determined empirical force field developed for dioctahedral clays. The simulations show that the adsorption of MB on mineral surfaces can result in a variety of configurations, including single and double layers of MB parallel to the basal surface, and irregular clusters. The d(001) values of ~12.3 and ~15.7 Å are assigned to dry phases with parallel single and double layers of MB, respectively, in agreement with X-ray studies. At intermediate MB loadings, stacks inclined to basal surfaces are formed. The stacks of MB ions inclined by 65–70° relative to the (001) plane of muscovite are not found on dry surfaces, in contrast to previous studies. Configurations similar to those proposed by others form spontaneously in the presence of H2O, but the ions in the model systems are not quite as ordered and not ordered in exactly the same way as the ones previously described, and they display a mobility that is not compatible with strict atomic order. The formation of a triple layer of H2O interspersed with ions may occur in the interlayer. Overall, the results of the simulations confirm that the MB-ion method must be used with great caution in surfacearea determinations, because of the multiplicity of possible configurations. At the same time, the ability for adsorption to occur as either single or multiple MB layers is useful to determine cation-exchange capacity over a wide range of surface-charge densities.

Electron correlation effects in aliphatic non-bonded interactions: comparison of n-alkane MP2 and HF geometries

The geometries of several n-alkanes were determined by HF/6-311G∗∗ and MP2/6-311G∗∗ gradient optimization. The results make it possible to study the effects of electron correlation on non-bonded aliphatic interactions by comparing structures devoid of dispersion forces (HF/6-31 IG∗∗) with those in which the dispersion forces are switched on (MP2/6-311 G∗∗). Conformations with trans bonds T (C-C-C-C torsions 180°), and gauche bonds G (C-C-C-C torsions 60°) were investigated, including T and G n-butane; TT, TG, and GG n-pentane; and TTT, GTT, TGG, GTG, and GGG n-hexane. When geometries are optimized at the MP2/6-311G∗∗ level, rotamer energies do not increase with the number of individual G bonds as expected, because GGG O.3 kcal mol−1) is larger in the MP2-optimized geometries than previously allowed (0.16 kcal mol−1). In general, in all rotamers with G torsions, small contractions in torsional angles (< 5°) cause non-bonded distances in the attractive region of the van der Waals potential to be shorter in MP2 structures than HF geometries, in contrast to 1,4 interactions, which are practically invariant. Specifically, average 1,5 non-bonded distances in GG pentane, and TGG and GGG hexane shrink by 0.22 to 0.27 A; by 0.06 to 0.08 A in TG pentane, and GTT and GTG hexane; and they are essentially unchanged in TT sequences. The results emphasize the importance of accurate geometries in conformational analyses: when the bond lengths and angles of a molecular model are wrong, calculated energies are also wrong, because non-bonded interactions are incorrectly evaluated.

Overlapping anomeric effects in a sucrose analogue

Previous calculations with the molecular mechanics program MM3 gave unusually high energies (as much as 5.5 kcal/mol) for sucrosyl geometries found in single-crystal diffraction studies of oligosaccharides. Comparable MM3 energies for observed interresidue linkage conformations of disaccharides such as maltose and cellobiose are all within 2.8 kcal/mol. These results suggest that some energies calculated by MM3 for the linkage between anomeric centers of a pyranose ring and a furanose ring are too high. In the present paper, ab initio calculations at the 4-21G level and MM3 were used to study the conformational energies and geometry of a sucrose analogue, tetrahydro-2-[(tetrahydro-2-furanyl)oxy]-2H-pyran. The range of energies of the observed structures was substantially reduced (to 2.4 kcal/mol) with the 4-21G calculations for the analogue despite an increase for the analogue (to 7.5 kcal/mol) based on new MM3 calculations. Besides the improved energy values, the 4-21G calculations also reproduced the observed variations in the endocyclic C-O bond lengths better than did MM3.

Ab Initio Study of the Nonequivalence of Adsorption of D- and L-Peptides on Clay Mineral Surfaces

Ab initio geometry optimizations were performed of the structures of the L and D enantiomers of the model dipeptide, N-formyl alanine amide (L-Ala and D-Ala, respectively) adsorbed on the mineral surface in the interlayer space of the 2:1 clay, nontronite. Density functional procedures were used as implemented in the ab initio program CASTEP, in fully periodic calculations in which the properties of a model unit cell are determined by including the effects of its neighboring cells in an infinite crystal. Geometry optimization included the atomic positions of the mineral and the adsorbates within the unit cell, in addition to unit cell lengths and angles, achieving full optimization of the entire crystal system. In agreement with previous studies of the adsorption of organic compounds on clay mineral surfaces, in the most stable arrangement of L-Ala and D-Ala found here, the two molecules reside flat on the mineral basal plane, forming a parallel monolayer. The resulting L-Ala/mineral cocrystal is more stable, by 6 kcal/mol, than D-Ala. A characteristic structural difference exists for the enantiomers in that, in the optimized structure of L-Ala, the C(α)–C(β) bond is directed away from one of the mineral basal planes toward the dioctahedral cavity of the other, while the side group of D-Ala is parallel to the mineral basal plane and pointing to the nearest neighbor in an adjacent unit cell. As expected, the reverse results are obtained for the adsorption of L-Ala and D-Ala on a nontronite surface that is the enantiomer of the one used above. The geometry optimizations illustrate the structural compatibility of clay mineral lattices with peptide structures. That is, balance of adsorption energies and peptide interaction energies, and mineral lattice structure and periodicity allow for adsorption structures, which involve the entire backbone of a single peptide molecule and can, at the same time, immobilize the adsorbates in such a way that stabilizing intermolecular interactions occur across unit cell boundaries. Compatibility of the repeat distances on the clay surface with repeat distances of peptides should be an important property when clay minerals serve as templates for protein synthesis.

Ab initio investigations pertaining to aluminum in tetrahedral versus octahedral sites of clay minerals

Abstract The structures and atomic charges of several aluminum and silicon oxide systems—[Al(OH)4]−, Al(OH3)(H2O)3, Al(OH)2(H2O)2(OH)2Al(OH)2(H2O)2, [Al(OH)3OSi(OH)3]−, and Al(OH)2(H2O)3OSi(OH)3—representing compounds with tetrahedral and octahedral coordination, were investigated by ab initio RHF and MP2 geometry refinements and by density functional calculations. In addition, structures and CHELPG charges were determined for the silicic acid dimer, [Si(OH)4]2, at equilibrium and at various displacements from the equilibrium intermolecular separation, in order to determine the effects of molecular association on partial atomic charges. The calculations were performed because the investigated compounds are fragments representative of phyllosilicate soil minerals such as smectite clays, and the results can be used to augment the database available for developing force field parameters for molecular dynamics simulations of adsorption phenomena at the clay mineral/aqueous solution interface. Furthermore, the results make it possible to discuss the effects of electron correlation on structures of this kind.

Geometry optimization, energetics and solvation studies on four- and five-membered cyclic and disulfide-bridged peptides, using the programs QUANTA 3.3 and CHARMm 22

Abstract Force field parameters for peptides and proteins in the charm m 22 force field have been refined using the results of MP2/6-311G∗∗ calculations on N-formyl alanine amide (Frey et al., J. Am. Chem. Soc., 114 (1992) 5369) and other data (Remek et al., J. Mol. Struct. (Theochem), 235 (1991) 1). Significant changes in the peptide backbone torsional energy terms in charm m 22 were made in order to be consistent with the MP2 results relative to previously fitted HF results. To test the new force field parameters against experimental data, molecular mechanics calculations were carried out on small four- and five-membered cyclic and disulfide-bridged peptides. It was found that in general these molecules are in their calculated minimum energy conformation or a low energy conformation in the X-ray structure. The average deviation of the dihedral angles calculated for the seven energy minimized vacuum structures from the X-ray values is 10°. The dihedral angle deviations of the minimized and dynamically averaged conformations are similar for studies which included explicit TIP3P solvent molecules. Upon solvation and minimization, internal vacuum energies increase by 4–7 kcal mol−1. In addition to calculated structural results, we report some of the force field parameters used for the calculations, energy decompositions, and harmonic vibrational free energies.

ø/ψ-Torsional dependence of peptide backbone bond-lengths and bond-angles: comparison of crystallographic and calculated parameters

Abstract The crystallographic NC(α), C(α)C′ and NC(α)C′ backbone parameters of 43 oligopeptides and the NC(α)C′ angles of 37 proteins were compared with peptide conformational geometry surfaces derived from ab initio calculations of N-acetyl N′-methyl alanine amide. The calculated values were obtained by spline-function representations of ab initio dipeptide conformational geometry maps which allow one to predict backbone bond lengths and angles in peptides and proteins as functions of the o[NC(α)]/ψ[C(α)C′]-torsions. When the parameters are ordered by regions in o/ψ-space defined by a 30° grid and region-average values calculated, the rms deviations between the crystallographic and calculated parameters in the most populated regions of the oligopeptides are 0.009A, 0.013A and 1.2°, for NC(α), C(α)C′ and NC(α)C′ respectively; and 1.2° for the NC(α)C′ angles in proteins. The flexibility in o/ψ-space is significant for the NC(α)C′ angles (observed variations of > 8°), but of lesser importance for the bond lengths (conformational variations of ∼0.02 A ). Thus, torsion dependent ideal geometry functions are recommended for the former, but not necessarily the latter, for use in various areas of protein study, such as protein crystallography and empirical molecular modeling procedures.