Further efforts toward a molecular dynamics force field for simulations of peptides in 40% trifluoroethanol-water.

A set of force field parameters for trifluoroethanol (TFE) was shown in earlier work from this lab to give a good accounting of system density, translational diffusion coefficients, and the solvent fluorine-solute hydrogen NMR cross-relaxation parameter (∑HF) for acetate dissolved in 40% TFE-water. It has since been found that this parameter set performs poorly when used to predict ∑HF for interactions of TFE with the hydrogens of an octapeptide, [val(5)]angiotensin. In the present work, adjustment of nonbonded force field parameters for interactions of fluorine with hydrogen was explored with the goal of improving these predictions. Six sets of TFE parameters were examined. When used in conjunction with the TIP5PE water model, all gave values for system density, translational diffusion coefficients, and ∑HF for acetate-TFE interactions that were similar to experimental results. Increasing the Lennard-Jones σHF for fluorine-hydrogen interactions by 10% led to calculated solvent-solute cross-relaxation parameters that are in better agreement with experiment for many of the peptide hydrogens. Changing the Lennard-Jones εHF parameters for the same interactions had little effect on calculated ∑HF values. There was no discernible influence of the TFE model used on the radius of gyration of the peptide or on the backbone conformational angles of the peptide, implying that the conformational properties of the peptide are not strongly influenced by changes in the force field description of TFE in 40% TFE-water. A recent parameter set for TFE proposed by Vymetal and Vondrasek (2014), which reproduces well various physical properties of neat TFE and TFE-water mixtures, was shown to predict cross-relaxation terms for this system poorly.

Interactions of nonprotic organic solvents with [val5]angiotensin in water.

Intermolecular solvent-solute nuclear Overhauser effects have been used to explore interactions of the organic component of acetonitrile-water, acetone-water, and dimethyl sulfoxide-water mixtures with the peptide hormone [val(5)]angiotensin. As reported by the NOEs, many cross relaxation terms for interactions of these organic cosolvents are adequately accounted for using a hard spheres interaction model in which encounters of peptide and cosolvent molecules take place by mutual diffusion. However, there are indications of localized solvent-peptide interactions that are not well described by this model. In dimethyl sulfoxide-water at 0 °C, organic solvent near the C-terminal Phe8 residue and the Val3 residue produce strongly enhanced cross-relaxation terms. NOEs for all peptide N-H protons and the protons of the Tyr4 aromatic ring were significantly more positive than expected in 33% acetone-water (v/v) at 0 °C, while those for most side-chain protons were close to predictions of the hard sphere model. All peptide-organic solvent NOEs in 35% acetonitrile water (v/v) at 0 °C are consistent with the hard spheres interaction model.

Conformations of Betanova in aqueous trifluoroethanol.

Conformations of the designed peptide Betanova in 42% trifluoroethanol/water (v/v) were explored. Circular dichroism (CD) observations provided no evidence for the presence of significant amounts of beta-structures in water, in TFE/water, or in ethanol/water. Nuclear magnetic resonance (NMR) diffusion experiments showed no significant difference in the hydrodynamic radius of the peptide in water and in 42% TFE/water. However, calculations indicated that the hydrodynamic radii of the triple-stranded beta-sheet, originally proposed for Betanova by Kortemme et al. (Science 1998, 281, 253-256), and a variety of partially folded forms of Betanova would be similar and likely could not be convincingly distinguished by diffusion experiments. Temperature coefficients (Deltadelta/DeltaT) of the peptide N--H chemical shifts are similar in water and 42% TFE/water, implying that most of these protons are highly solvent exposed in both solvents and likely do not participate in intramolecular hydrogen bonding interactions. Possible exceptions to this conclusion are the Lys9 and Lys15 residues, where a more positive coefficient may indicate that these residues are involved to some extent in local turn structures. Peptide proton-solvent fluorine intermolecular nuclear Overhauser effect (NOE)s at 25 degrees C were consistent with the presence of a mixture of conformations, which could include the triple-stranded beta-sheet structure as a minor component. At 0 degrees C, peptide-TFE NOEs indicated that TFE interacts strongly enough with many protons of Betanova that alcohol-peptide interactions persist for times of the order of nanoseconds, appreciably longer than the encounter time characteristic of mutual diffusion of TFE and the solute.

Interactions of 2,2,2-trifluoroethanol with melittin.

Melittin dissolved in 42% trifluoroethanol-water at pH 2 has been shown to be alpha-helical between residues 6 and 12 and between residues 13 and 25, with the two helical regions separated by a bend at the Leu13 residue. The inter-helix angle was found to be 154 +/- 3 degrees at 0 degrees C and 135 +/- 3 degrees at 25 degrees C. The dominant conformation of the peptide is thus similar to those observed by previous workers for the peptide in a variety of media. At 25 degrees C, intermolecular nuclear Overhauser effects arising from nuclear spin dipole-dipole interactions between melittin hydrogens and fluorines of the solvent are essentially those expected for a system that is homogeneous as regards concentration and translational diffusion of the peptide and fluoroalcohol components. However, at 0 degrees C, peptide-trifluoroethanol cross-relaxation terms are negative, a result consistent with the conclusion that fluoroalcohol molecules associate with the peptide for times (approximately 1 ns) that are long compared to the time of a typical peptide-fluoroalcohol diffusive encounter (approximately 0.2 ns). Such interactions may be responsible for the reduction of the translational diffusion coefficient of trifluoroethanol produced by dissolved peptides.

Solvent interactions with [Val(5)]angiotensin II in ethanol-water.

Intermolecular NOE experiments have been used to explore the interactions of water and ethanol molecules in 35% ethanol/65% water (v/v) with the octapeptide hormone [val (5)]angiotensin II at temperatures from 0 to 25 degrees C. Magnetic dipole-dipole cross relaxation terms sigma(HH)(NOE) and sigma(HH)(ROE) for interaction of both solvent components suggest that ethanol molecules interact with the peptide backbone atoms strongly enough to associate for times comparable to the rotational correlation time of the peptide; comparison of observed ROE and NOE cross relaxation terms indicate that lifetimes of these interactions are of the order 0.4 ns at 5 degrees C. Formation of such peptide-ethanol complexes can also account for larger-than-expected values of the cross relaxation terms at higher temperatures. Alternative explanations of the observations reported are shown to be unlikely, primarily because they require unreasonable and highly localized concentrations of the ethanol near the peptide. Side chains of the peptide appear to experience no unusual interactions with ethanol. Cross relaxation terms for water-peptide backbone interactions indicate long-lived interactions of water with the backbone atoms although the nonpolar side chains of the peptide (Val3, Val5, Pro7, and possibly Phe8) do not interact in any specific way with water molecules. Cross relaxation terms for protons of the polar (Tyr4 and His6) side chains may reflect strong interactions with water, but analysis of these is confounded by solvent proton exchange and possible spin diffusion effects.

Solvent interactions with the Trp-cage peptide in 35% ethanol-water.

Intermolecular NOE experiments have been used to explore interactions of water and ethanol molecules in 35% ethanol/65% water (v/v) with the peptide Trp-cage at temperatures from 5 to 25 degrees C. Magnetic dipole-dipole cross-relaxation terms sigma(HH) (NOE) and sigma(HH) (ROE) for interaction of solvent components with spins of the peptide suggest that ethanol molecules associate with backbone atoms for times of the order of nanoseconds at 5 degrees C. Formation of peptide-ethanol complexes can also account for the larger-than-expected values of cross-relaxation terms at higher temperatures. Hydrocarbon side chains of the peptide do not appear to experience such interactions with ethanol. Cross relaxation resulting from water-peptide interactions are consistent with long-lived water interactions with the backbone atoms. Water cross relaxation with nonpolar side chains of the peptide (Leu2, Ile4, Leu7, and proline residues) are only those expected for bulk solvent. However, long-lived association of both water and ethanol with the polar side chains of Tyr3 and Trp6 is indicated by the data.

Interactions of trifluoroethanol with [val5]angiotensin II.

Intermolecular 1H{19F} NOE experiments have been used to explore the interactions of trifluoroethanol (TFE) with the octapeptide hormone [val5]angiotensin II at temperatures from 5 to 25 degrees C. Circular dichroism spectra indicate that 40% trifluoroethanol has an influence on the conformations of the peptide, probably leading to beta-structures. Diffusion experiments show that the mean hydrodynamic radius of the peptide in 40% trifluoroethanol-water is about 8 A, consistent with significant folding of the peptide in this medium. Distance constraints derived from intramolecular NOESY data along with observed vicinal coupling constants (3JCalphaHNH) were used to develop conformations consistent with available data. Assuming that intermolecular 1H{19F} NOEs are the result of diffusive encounters of TFE and peptide molecules, it is shown that no single conformation is consistent with the experimental values of the sigmaHF cross-relaxation parameters. It is argued that the disagreements between observed and expected values of sigmaHF are the result of formation of long-lived (approximately 0.5 ns) fluoroalcohol-peptide complexes, a conclusion consonant with similar studies of other peptide-fluoroalcohol systems. Complex formation appears to be especially prevalent near the charged amino acid side chains of the hormone.

Interactions of trifluroethanol with the Trp-cage peptide.

It has been suggested that aggregation of fluorinated alcohols in water solutions is involved with the abilities of these alcohols to provoke conformational changes in peptides and proteins. The extent of fluoroalcohol aggregation depends on the degree of fluorination: hexafluoroisopropanol (HFIP) is more extensively aggregated than is TFE. We previously described a study of the interactions of HFIP with the peptide Trp-cage and provided evidence for the formation of long-lived complexes between this fluoroalcohol and the peptide. In the present work, we have examined the interactions of the less-fluorinated TFE with Trp-cage, in order to probe the role of fluoroalcohol aggregation in the phenomena observed. Intermolecular (1)H{(19)F} nuclear Overhauser effects arising from interactions of TFE with the hydrogens of the peptide in a solution containing 42% TFE were determined at sample temperatures from 5 to 45 degrees C. It is shown that the folded state of the peptide under these conditions is essentially the same as that observed in water and in 30% HFIP-water. The observed peptide-solvent NOEs indicate formation of complexes of Trp-cage with TFE that persist for times of the order of 1 ns. The interactions leading to complexes with TFE are somewhat weaker than those involved in complex formation with HFIP. There are no indications that the aggregation of fluoroalcohol is a necessary concomitant of the interactions of TFE or HFIP with Trp-cage. Rather, the stronger and more long-lived interactions of HFIP with Trp-cage appear to be primarily the result of the greater hydrogen-bonding ability and hydrophobicity of this fluoroalcohol.

Interactions of hexafluoro-2-propanol with the Trp-cage peptide.

Fluoro alcohols present in aqueous solutions can alter the dominant conformations of peptides and proteins. The origins of these effects likely are related to the details of solute-fluoro alcohol interactions. Preferential interaction of the fluoro alcohol component of a fluoro alcohol-water mixture with peptide solutes has been demonstrated by several experimental approaches. In the present work, we have used 1H{19F} intermolecular NOE experiments to examine interactions of hexafluoro-2-propanol in a 30% fluoro alcohol-50 mM phosphate buffer solvent mixture with the "Trp-cage" peptide (NLY IQW LKD GGP SSG RPP PS). The results show that the peptide is selectively solvated by hexafluoro-2-propanol to the extent that the fluoro alcohol concentration near the peptide may be 3 to 4 times higher than the nominal concentration of fluoro alcohol in the bulk sample. The observed NOEs indicate that peptide-fluoro alcohol interactions persist for times of the order of 1 ns at 5 degrees C. As the sample temperature is increased, the lifetimes of fluoro alcohol interactions with several exposed side chains decrease to the extent that the peptide hydrogen-solvent fluorine interactions appear to become diffusive in nature, with interaction lifetimes of approximately 0.03 ns. It is known that protein molecules can provide specific sites for binding small organic solvent molecules. Our work suggests that small peptides also have this ability and that the dynamics for such interactions can be site-specific.

Interactions of trimethylamine N-oxide and water with cyclo-alanylglycine.

The osmolyte trimethylamine N-oxide (TMAO) is one of a family of compounds found in living systems that can stabilize biomolecular tertiary structures. As a step in exploring the interactions between this material and polyamino acids, we have determined intermolecular 1H{1H} nuclear Overhauser effects (NOEs) between the protons of cyclo-alanylglycine and protons of solvent components in TMAO-water solutions. Comparison of the results to effects predicted on the basis of the molecular shape of the dipeptide and experimental translational diffusion coefficients suggests that both water and TMAO molecules have properties in the vicinity of the dipeptide that are different from those in the bulk solution. Changes of local concentrations of water and TMAO and changes in the diffusive behavior of these components near the dipeptide are rejected as possible explanations of the discrepancies between observed and calculated Overhauser effects. Rather, it is concluded that TMAO molecules, and the water molecules associated with them, participate to some extent in the formation of long-lived solute-solvent complexes. The aliphatic alcohol tert-butyl alcohol is structurally similar to TMAO. Overhauser effect studies of its interaction with cyclo-alanylglycine in tert-butyl alcohol-water suggest similar kinds of interactions are present in this system but that they are significantly weaker, presumably because of the lower polarity of this alcohol compared to TMAO.

Selective solvent interactions in a fluorous reaction system.

Mixtures of chloroform and perfluoro(methylcyclohexane) can be used as solvents for "fluorous" biphase reactions since they exist as two separate phases at low temperature but become a single phase at higher temperatures. Intermolecular nuclear Overhauser effects have been used to investigate the interactions of solvent components with the protons and fluorines of 3-heptafluorobutyrylcamphor in both phases of this biphasic system at 25 degrees C as well as the single phase at 54 degrees C. The results indicate that at 25 degrees C in the perfluorocarbon-rich phase, both solvent components interact with the solute selectively. There are no indications of unusual solute interactions of either solvent component in the chloroform-rich phase and only weak suggestions of selective interactions in the high-temperature phase. Various mechanisms for the enhancement of solute spin-solvent spin cross relaxation rates in the perfluoro(methylcyclohexane)-rich phase are considered. It is suggested that the solvation layer around the solute has a composition and possibly hydrodynamic properties different from those of the bulk solution in this phase. There are indications of appreciable regioselectivity of chloroform interactions with the hydrocarbon part of the solute in all phases.