Examination of Solvent Interactions with Trp-Cage in 1,1,1,3,3,3-Hexafluoro-2-propanol-water at 298 K through MD Simulations and Intermolecular Nuclear Overhauser Effects.

MD simulations of the peptide Trp-cage dissolved in 28% hexafluoroisopropanol (HFIP)-water have been carried out at 298 K with the goal of exploring peptide hydrogen-solvent fluorine nuclear spin cross-relaxation. The work was motivated by the observation that most experimental fluoroalcohol-peptide cross-relaxation terms at 298 K are small, both positive and negative, and not always well predicted from simulations. The cross-relaxation terms for hydrogens of the caged tryptophan residue of Trp-cage are substantially negative, a result consistent with simulations. It was concluded that hexafluoroisopropanol interactions near this part of the peptide are particularly long-lived. While both HFIP and water are present in all regions of the simulation box, the composition of the solvent mixture is not homogeneous throughout the system. HFIP generally accumulates near the peptide surface, while water molecules are preferentially found in regions that are more than 1.5 nm from the surface of the peptide. However, some water remains in higher-than-expected amounts in the solvent layer surrounding 6Trp, 9Asp, Ser13, and Ser14 residues in the helical region of Trp-cage. As observed in simulations of this system at 278 K, HFIP molecules aggregate into clusters that continually form and re-form. Translational diffusion of both HFIP and water appears to be slowed near the surface of the peptide with reduction in diffusion near the 6Trp residue 2- to 3-fold larger than calculated for solvent interactions with other regions of Trp-cage.

Examination of Interactions of Hexafluoro-2-propanol with Trp-Cage in Hexafluoro-2-propanol-Water by MD Simulations and Intermolecular Nuclear Overhauser Effects.

All-atom molecular dynamic simulations of the peptide Trp-cage in 30% hexafluoro-2-propanol- water (V/V) at 278 K have been carried out with the goal of exploring peptide hydrogen-solvent fluorine nuclear spin cross relaxation. Force field parameters for HFIP reported by Fioroni et al. along with the fluorine parameters of the TFE5 model reported by this lab were used. Water was represented by the TIP5P-Ew model. Peptide modeling used the AMBER99SB-ILDN force field. Translational diffusion coefficients of solution components at 278 K were predicted to within 35% of experimental values using these parameter sets. The simulations indicate that the solvent mixture is not homogeneous, with HFIP molecules clustered into aggregates as large as 53 fluoroalcohol molecules. The solvent environment of surface atoms of Trp-cage fluctuates between being HFIP-rich and more water-rich about every 10 ns. In accord with previous studies by other groups, the average concentration of HFIP near the surface of the peptide is significantly enhanced over the concentration of HFIP in the bulk solvent. In the simulations, ∼7% of the initial contacts between HFIP molecules and Trp-cage develop into peptide-fluoroalcohol interactions that persist for times as long as 8 ns. Most of the available experimental nuclear spin cross-relaxation rates (ΣHF) for hydrogens of the Trp-cage in 30% HFIP-water are reproduced from the MD trajectories to within uncertainties of the experimental data and the simulations. However, a few calculated ΣHF values for hydrogens of the Trp-cage do not agree with experiment. These tend to be situations where long-lived peptide-HFIP interactions are predicted. The disagreements between observed and calculated ΣHF in these instances signal defects in the modeling parameters and procedures that are presently unrecognized.

Examination of Trifluoroethanol Interactions with Trp-Cage in Trifluoroethanol-Water at 298 K through Molecular Dynamics Simulations and Intermolecular Nuclear Overhauser Effects.

Molecular dynamics simulations of the protein model Trp-cage in 42% trifluoroethanol (TFE)-water at 298 K have been carried out with the goal of exploring peptide hydrogen-solvent fluorine nuclear spin cross-relaxation. The TFE5 model of TFE developed in a previous work was used with the TIP5P-Ew model of water. System densities and component translational diffusion coefficients predicted by the simulations were within 20% of the experimental values. Consideration of the calculated relative amounts of TFE and water surrounding the hydrogens of Trp-cage indicated that the composition of the solvent mixture beyond ∼1.5 nm from the van der Waals surface of the peptide is close to the composition of the bulk solvent, but as observed by others, TFE accumulates preferentially near the peptide surface. In the simulations, both TFE and water molecules make contacts with the peptide surface; water molecules predominate in contacts with the peptide backbone atoms and TFE molecules generally preferentially interact with side chains. Translational diffusion of solvent molecules appears to be slowed near the surface of the peptide. Depending on the location in the structure, TFE molecules form complexes with the peptide that may persist for up to ∼7 ns. Many of the peptide spin-solvent fluorine cross-relaxation parameters (ΣHF) for which experimental values are available are reasonably well-predicted from the simulations. However, the calculated ΣHF values were too small for some hydrogens of the 6Trp indole ring and the amino acid hydrogens near this residue in the native structure, whereas ΣHF values for hydrogens on the side chains of 1Asn, 4Ile, and 7Leu are too large. In 42% TFE-water, persistent conformations of Trp-cage are found, which differ from the conformation found in water by the orientation of the 3Tyr ring.

Examination of Trifluoroethanol Interactions with Trp-Cage through MD Simulations and Intermolecular Nuclear Overhauser Effects.

Molecular dynamics simulations of the protein model Trp-cage in 42% 2,2,2-trifluoroethanol (TFE)-water at 318 K have been carried out with the goal of exploring solvent fluorine-peptide hydrogen nuclear spin cross-relaxation. TFE5 and TFE6 models of TFE developed in previous work from this laboratory were used with the TIP5PE model of water. System densities and component translational diffusion coefficients were well predicted by the simulations, as were many of the cross-relaxation parameters (ΣHF) for which experimental values are available. However, the calculated ΣHF values were too small for some hydrogens of the Trp6 indole ring and for amino acid hydrogens near this residue in the native structure. Simulations carried out with unfolded versions of Trp-cage suggest that underestimates of ΣHF are the result of insufficient sampling of conformational motions that expose these hydrogens to interactions with solvent molecules during simulations of the native peptide. Consideration of the relative amounts of TFE and water surrounding the Trp-cage structure indicates that the composition of the solvent mixture at distances beyond ∼1.5 nm from the surface of the peptide is close to the composition of the bulk solvent, but, as observed by others, TFE tends to accumulate preferentially near the surface of the peptide. Both TFE and water molecules make contacts with the peptide surface; water molecules predominate in contacts with the peptide backbone atoms, and TFE molecules generally preferentially interact with side chains. Translational diffusion of solvent molecules appears to slow down near the surface of the peptide.

Toward a molecular dynamics force field for simulations of 40% trifluoroethanol-water.

Various computational models of trifluoroethanol (TFE) and water have been explored with the goal of finding a system for molecular dynamics (MD) simulations that reliably predict properties of 40% TFE-water (v/v) and can be used in studies of peptide-solvent nuclear cross-relaxation. Models derived by modification of TFE parameters developed by Fioroni et al. (J. Phys. Chem. B 2000, 104, 12347), in combination with either TIP4P-Ew or TIP5P-E water, were most successful. Simulations of 40% TFE-TIP4P-Ew water evidenced separation of the system into large TFE-rich and water-rich domains. With TIP5P-E water, simulations showed aggregation of each solvent component into small clusters. Nuclear spin dipolar interactions between solvent fluorines and the methyl hydrogens of acetate ion dissolved in 40% TFE-water were calculated. The cross-relaxation parameter σHF reckoned for the TFE-TIP5P-E system agreed with experiment while the value calculated using the TFE-TIP4P-Ew system was too low. While the TFE-TIP5P-E model of 40% TFE-water leads to good predictions of the system density, translational diffusion coefficients, and a solvent-solute cross-relaxation parameter, this model performs poorly in predicting the enthalpy of mixing. Preliminary studies of 20% TFE-water and 50% TFE-water suggest that the model will perform with the same characteristics for mixtures that have compositions near 40% TFE-water.

Investigation of ethanol-peptide and water-peptide interactions through intermolecular nuclear overhauser effects and molecular dynamics simulations.

Molecular dynamics simulations have been used to explore solvent-solute intermolecular nuclear Overhauser effects (NOEs) on NMR (nuclear magnetic resonance) signals of [val5]angiotensin dissolved in 35% ethanol-water (v/v). Consideration of chemical shift, coupling constant and intramolecular NOE data suggest that conformations of the peptide are adequately sampled by simulations of up to 0.6 µs duration. Calculated cross relaxation terms at 0 and 25 °C are compared to experimental values and to terms predicted using a particulate model of the solvent. Many calculated solvent NOEs are in agreement with experimental results; disagreements are particularly striking for hydrogens of the Phe8 residue of the peptide. Calculations show that individual molecules of either solvent component can spend many ns in association with the peptide but dipolar interactions within such a complex account for only a few percent of an observed cross relaxation rate. Most parts of the peptide interact selectively with ethanol. Diffusion of both solvent components is slowed when they are close to the peptide. Solvent-solute cross relaxation terms for acetic acid in the same solvent obtained from simulations agree with experiment. Preferential interactions of solvent molecules with acetic acid are largely absent, as are effects of this solute on solvent diffusion rates.

Investigation of methanol-peptide nuclear Overhauser effects through molecular dynamics simulations.

Intermolecular nuclear Overhauser effects (NOEs) produced by interactions of methanol with [val(5)]angiotensin in 25% methanol-water at 0 °C were examined through molecular dynamics (MD) simulations and compared to experimental results. Calculated average (3)J(NHCαH) spin coupling constants, conformation-sensitive chemical shift changes, and intramolecular (1)H-(1)H NOEs indicated that peptide conformations present over the course of simulation trajectories of 100-300 ns are likely similar to those present in the experimental system. Calculated cross-relaxation terms for the methanol-peptide interactions showed the same trends as corresponding experimental data but were about a factor of 3 too large. The lack of agreement between observed and calculated cross-relaxation terms probably has origins in characteristics of the simulations that lead to overestimation of translational diffusion coefficients of the system components. Simulations confirmed the heterogeneity of the methanol-water solvent at the molecular level, with clusters of methanol and water molecules changing their size and composition on a subpicosecond time scale. Most peptide hydrogens are preferentially solvated by interactions with methanol molecules. Simulations suggest that diffusion of water and methanol molecules near the peptide is slowed as these species approach the peptide backbone.

Cross relaxation in liquid methanol.

The cross relaxation rate for intermolecular dipole-dipole interactions between methyl protons in liquid methanol at 0 °C was measured and compared to the rate predicted from MD simulations of the experimental system. The experimental and calculated values agree well, even though the translational diffusion coefficient and bulk viscosity of the sample are not well-predicted by the simulations.

Interaction of alcohols with [val5]angiotensin in alcohol-water mixtures.

Intermolecular solvent-solute NOE experiments have been used to probe interactions of various alcohols with the peptide hormone [val(5)]angiotensin II at 0 degrees C. It is found that these NOEs are detectable but dependent on the kind of alcohol present and the conformation of the peptide. Solvent-solute NOEs in 100% methanol and 89% methanol-water are basically those predicted by a hard sphere model for intermolecular spin dipole interactions. NOEs at the peptide backbone (N-H, C alpha-H) protons in 25% methanol-water and 36% ethylene glycol-water mixtures indicate that alcohol interactions near these groups are also adequately described by this model. However, in 35% ethanol-water, interactions of alcohol methyl protons with the peptide result in unexpectedly negative NOEs, probably signaling that peptide-alcohol interactions in this solvent take place on a significantly slower time scale than that defined by mutual diffusion of these species. Some side chain-alcohol interactions result in NOEs up to 8 times larger than expected. Possible reasons for these enhanced effects are discussed.

TMAO promotes fibrillization and microtubule assembly activity in the C-terminal repeat region of tau.

Alzheimer's disease most closely correlates with the appearance of the neurofibrillary tangles (NFTs), intracellular fibrous aggregates of the microtubule-associated protein, tau. Under native conditions, tau is an unstructured protein, and its physical characterization has revealed no clues about the three-dimensional structural determinants essential for aggregation or microtubule binding. We have found that the natural osmolyte trimethylamine N-oxide (TMAO) induces secondary structure in a C-terminal fragment of tau (tau(187)) and greatly promotes both self-aggregation and microtubule (MT) assembly activity. These processes could be distinguished, however, by a single-amino acid substitution (Tyr(310) --> Ala), which severely inhibited aggregation but had no effect on MT assembly activity. The inability of this mutant to aggregate could be completely reversed by TMAO. We propose a model in which TMAO induces partial order in tau(187), resulting in conformers that may correspond to on-pathway intermediates of either aggregation or tau-dependent MT assembly or both. These studies set the stage for future high-resolution structural characterization of these intermediates and the basis by which Tyr(310) may direct pathologic versus normal tau function.

Structure and solvation of melittin in hexafluoroacetone/water.

Intermolecular (1)H[(19)F] and (1)H[(1)H] nuclear Overhauser effects have been used to explore interaction of solvent components with melittin dissolved in 50% hexafluoroacetone trihydrate (HFA)/water. Standard nuclear Overhauser effect experiments and an analysis of C(alpha)H proton chemical shifts confirm that the conformation of the peptide in this solvent is alpha-helical from residues Ala4 to Thr11 and from Leu13 to Arg24. The two helical regions are not collinear; the interhelix angle (144 +/- 20 degrees ) found in this work is near that observed in the solid state and previous NMR studies. Intermolecular NOEs arising from interactions between spins of the solvent and the solute indicate that both fluoroalcohol and water molecules are strongly enough bound to the peptide that solvent-solute complexes persist for > or =2 ns. Preferential interactions of HFA with many hydrophobic side chains of the peptide are apparent while water molecules appear to be localized near hydrophilic side chains. These results indicate that interactions of both HFA and water are qualitatively different from those present when the peptide is dissolved in 35% hexafluoro-2-propanol/water, a chemically similar helix-supporting solvent system.

Structure and solvation of melittin in 1,1,1,3,3,3-hexafluoro-2-propanol/water.

Fluorinated alcohols can induce peptides and proteins to take up helical conformations. Nuclear Overhauser effect (NOE) spectroscopy experiments and analysis of C(alpha)H proton chemical shifts show that the conformation of melittin in 35% hexafluoro-2-propanol/water is alpha-helical from residues Ile-2 to Val-8 and from Leu-13 to Gln-25. As has been found in other solvent systems, the two helical regions are not colinear; the interhelix angle (73 +/- 15 degrees ) in 35% 1,1,1,3,3,3-hexafluoro-2-propanol/water is smaller than the angle found in other fluoroalcohol-water mixtures or in the crystal. Intermolecular (1)H(19)F and (1)H(1)H nuclear Overhauser effects were used to explore interaction of solvent components with melittin dissolved in this solvent mixture. The NOEs observed indicate that fluoroalcohol and water molecules are both tightly bound to the peptide in the vicinity of the interhelix bend. For the remainder of the molecule, solute-solvent NOEs are consistent with preferential solvation of the peptide by the fluoroalcohol component of the solvent mixture.

Solute-solvent interactions probed by intermolecular NOEs.

Nuclear Overhauser effects arising from the interactions of spins of solvent molecules with spins of a solute should reveal the "exposure" of solute spins to collisions with solvent. Such intermolecular NOEs could, therefore, provide information regarding conformation or structure of the solute. Determinations of solute-solvent NOEs of 1,3-di-tert-butylbenzene in solvents composed of perfluoro-tert-butyl alcohol, tetramethylsilane, and carbon tetrachloride have been carried out. A crude, but apparently reliable, method for prediction of intermolecular solvent-solute NOEs based on hard (noninteracting) spheres was developed. Comparison of experimental to predicted NOEs indicates that tetramethylsilane interacts with the solute according to the model. By contrast, intermolecular NOE data indicate attractive interactions between the solute and perfluoro-tert-butyl alcohol. All NOE results and the corresponding predictions confirm that proton H2 of the solute is protected by the flanking tert-butyl groups from interactions with solvent molecules.

Intermolecular Overhauser effects in fluoroalcohol solutions of cyclo-alanylglycine.

Interactions between the diketopiperazine cyclo-alanylglycine and four fluorinated alcohols in water-fluoroalcohol mixtures were examined by (1)H[(19)F] intermolecular nuclear Overhauser effects (NOE) experiments. The alcohols studied were trifluoroethanol, hexafluoroacetone trihydrate, 1,1,1,3,3,3-hexafluoroisopropanol and perfluoro-t-butanol. The experimental methods used permit detection of solvent-solute NOEs of 0.1% or less. Solute and solvent diffusion coefficients were determined and apparent molecular radii of the fluoroalcohols estimated. Using these data, it was shown that observed (1)H[(19)F] intermolecular NOEs are consistent with expectations based on theory. A method for extending conventional theory to take into account the shape of a solute and the exposure of its hydrogens to solvent is described. This approach gives reasonable agreement with experimental results, particularly if it is assumed that solute-solvent interactions take place in such a way that the fluorines of a fluoroalcohol are preferentially oriented toward the solute during solute-solvent encounters. The results support the suggestion that intermolecular (1)H[(19)F] NOEs may become a useful tool for studies of peptide and protein conformations in fluoroalcohol-water solvent mixtures.

Intermolecular (1)H[(19)F] NOEs in studies of fluoroalcohol-induced conformations of peptides and proteins.

Mixtures of fluorinated alcohols and water can selectively stabilize certain secondary structures of peptides and proteins. Such mixtures may also be of use in solubilizing hydrophobic or membrane-bound proteins. We show that intermolecular dipolar interactions between the fluorine nuclei of such solvents and the protons of a dissolved protein lead to readily detected (1)H[(19)F] nuclear Overhauser effects. These NOEs can potentially provide information about solvent exposure of particular groups as well as indicate the formation of long-lived fluoroalcohol-solute complexes. Results obtained with HEW lysozyme in solutions containing trifluoroethanol illustrate these possibilities.

Spin relaxation and chemical exchange in NMR simulations.

Theory for describing the density matrix of a spin system experiencing chemical exchange and relaxation during the steps of an NMR experiment is presented in a form suitable for computation. Features in the theory that arise from exchange are discussed in detail, and comparisons to the exchange-free situation are made. A general computer program to carry out simulations of NMR experiments is described, and several examples of its performance are presented.

Pulsed field gradients in simulations of one- and two-dimensional NMR spectra.

A method for the inclusion of the effects of z-axis pulsed field gradients in computer simulations of an arbitrary pulsed NMR experiment with spin (1/2) nuclei is described. Recognizing that the phase acquired by a coherence following the application of a z-axis pulsed field gradient bears a fixed relation to its order and the spatial position of the spins in the sample tube, the sample is regarded as a collection of volume elements, each phase-encoded by a characteristic, spatially dependent precession frequency. The evolution of the sample's density matrix is thus obtained by computing the evolution of the density matrix for each volume element. Following the last gradient pulse, these density matrices are combined to form a composite density matrix which evolves through the rest of the experiment to yield the observable signal. This approach is implemented in a program which includes capabilities for rigorous inclusion of spin relaxation by dipole-dipole, chemical shift anisotropy, and random field mechanisms, plus the effects of arbitrary RF fields. Mathematical procedures for accelerating these calculations are described. The approach is illustrated by simulations of representative one- and two-dimensional NMR experiments.

Tritium NMR studies of the human carbonic anhydrase I-benzenesulfonamide complex.

Tritium NMR spectroscopy has been used to examine the complex formed by [4-3H]benzenesulfon-amide and human carbonic anhydrase I. The results show that in solution the inhibitor forms a 1:1 complex with the enzyme. A 100-spin computational model of the system, constructed with reference to crystallographic results, was used to interpret tritium relaxation behavior and 3H{1H} NOEs. The analysis shows that the rate of dissociation of the enzyme-sulfonamide complex is 0.35 s-1 and that the aromatic ring of the inhibitor undergoes rapid rotation while complexed.

Effects of fluorine substitution on the structure and dynamics of complexes of dihydrofolate reductase (Escherichia coli).

Fluorine NMR experiments with a protein containing fluorinated amino acid analogs can often be used to probe structure and dynamics of the protein as well as conformational changes produced by binding of small molecules. The relevance of NMR experiments with fluorine-containing materials to characteristics of the corresponding native (nonfluorinated) proteins depends upon the extent to which these characteristics are altered by the presence of fluorine. The present work uses molecular dynamics simulations to explore the effects of replacement of tryptophan by 6-fluorotryptophan in folate and methotrexate complexes of the enzyme dihydrofolate reductase (DHFR) (Escherichia coli). Simulations of the folate-native enzyme complex produce local correlation times and order parameters that are generally in good agreement with experimental values. Simulations of the corresponding fluorotryptophan-containing system indicate that the structure and dynamics of this complex are scarcely changed by the presence of fluorinated amino acids. Calculations with the pharmacologically important methotrexate-enzyme complex predict dynamical behavior of the protein similar to that of the folate complex for both the fluorinated and native enzyme. It thus appears that, on the time scale sampled by these computer simulations, substitution of 6-fluorotryptophan for tryptophan has little effect on either the structures or dynamics of DHFR in these complexes.

Cross-correlation effects on NMR lineshapes and peptide conformation.

Information about molecular structure and dynamics can potentially be obtained by studying dipole-dipole and chemical-shift anisotropy (CSA) auto-correlation and dipole-CSA cross-correlation effects in high-resolution NMR spectra. Equations for the lineshapes of the HN multiplet in the fragment- 15NH-CH- as a function of NH-CH dihedral angle are derived by including these effects within the framework of the Redfield treatment of relaxation. To test the utility of the theoretical results, 1H[15N] HSQC proton lineshape data for a variant of the enzyme staphylococcal nuclease in which all valine residues are labeled with 15N have been analyzed to obtain the conformational angle (phi) between the N-H and adjacent C-H bonds. The results are generally in good agreement with values of phi obtained from crystal structure data. Considerations in the further development of the analysis of the lineshape of the HN multiplet for experimental determinations of phi are discussed.