Making sense of Hofmeister

In preparation for the forthcoming RSC Faraday Discussion on Hofmeister effects, Pavel Jungwirth, Paul Cremer and their coworkers describe how certain ion-specific interactions with peptides occur (K. B. Rembert et al., JACS ja301297g – paper here). They have looked both experimentally (NMR, thermodynamic measurements) and with MD at the interaction of anions such as iodide, sulphate, chloride and thiocyanate with a model 600-residue elastin-like peptide. Large, soft and weakly hydrated anions such as I- and SCN- bind to an amide nitrogen and the adjacent alpha-carbon, where there is a slight positive charge, promoting peptide solubility and disrupting peptide-water hydrogen bonds. However, at higher concentrations, saturation of these binding sites can eventually lead to salting out because the anions then increase the surface tension of purely hydrophobic portions of the peptide-water interface. In contrast, chloride, further up the Hofmeister series, binds only weakly at these sites, while sulphate, further up still, is repelled from them. The work shows once again how the Hofmeister behaviour is determined by direct interactions between the ions and peptide.

Paul Cremer and colleagues have also considered how carboxylate side-chain groups in a peptide affect Hofmeister cationic effects (J. Kherb et al., J. Phys. Chem. B 116, 7389; 2012 – paper here). These ions will interact directly with the carboxylates to form ion pairs. The influence of the monovalent cations is generally consistent with the ‘law of matching water affinity’, which posits that anions and cations of similar hydration energies will preferentially form ion pairs. One exception here is ammonium, which can form hydrogen bonds with carboxylate. Lithium is also anomalous, perhaps because its tight binding of water leads to the formation of an ion pair mediated by shared waters rather than via direct contacts.

More on Hofmeister effects comes from Juan Luis Ortega-Vinuesa at the University of Granada and colleagues, who look at ion-specific effects in colloidal interactions, specifically on the aggregation of hydrophobic particles (polystyrene nanobeads) (T. López-León et al., ChemPhysChem 13, 2382; 2012 – paper here). They find that anions early in the series (sulphate, chloride) behave in accord with the predictions of DLVO theory, but large, polarizable anions (nitrate, thiocyanate) do not. The (fractal) morphology of the aggregates also differs for different ions. The findings are intriguing, but the suggested interpretation in terms of ion effects on ‘water structure’ is indirect (as the authors acknowledge) and, I think, open to debate.

Enzymes can become inactivated by covalent binding in the active site, a process known as mechanism-based inactivation, which can have important physiological consequences. Hajime Hirao and colleagues at Nanyang Technological University in Singapore have investigated how this occurs for cytochrome P450, which can for example be inactivated by acetylenes (J. Phys. Chem. B jp302592d – paper here). Their density-functional theory calculations suggest that the reaction is mediated by a catalytic water molecule in the binding site, which donates a proton to the acetylene group.

Do the various types of denaturation of proteins share any common features? Cristiano Dias of the Free University of Berlin suggests a unifying picture of cold and pressure denaturation that involves solvation of nonpolar residues by a monolayer or so of water (Phys. Rev. Lett. 109, 048104; 2012 – paper here). Because these solvated states have a lower volume and H-bond free energy than others, they are favoured at high pressure and at low temperature. Using this picture in a very simple bead-spring model of folded proteins, with water described by the two-dimensional Mercedes-Benz model, Dias can reproduce the ellipsoidal NPT phase diagram of proteins.

Pablo Debenedetti at Princeton and his coworkers have been exploring much the same issues using another simple model: a lattice model of homo- and heteropolymers in two and three dimensions (e.g. Patel et al., J. Chem. Phys. 128, 175102 (2008) and Matysiak et al., J. Phys. Chem. B 10/1021/jp3039175 (2012)). They now show that a hydrophobic polymer in the 3D system (with tetrahedral water), which adopts a collapsed conformation within a certain temperature range, will undergo both thermal denaturation and cold denaturation above and below this regime (S. R.-V. Castrillón et al., J. Phys. Chem. B jp3039237 – paper here). The mechanisms of denaturation are distinct at high and low temperatures: the former is an extended random coil, while the latter is still relatively compact but strongly water-penetrated.

The role of small-molecule denaturants is also becoming steadily more clear. Ruhong Zhou at IBM Yorktown Heights, working with colleagues in Hangzhou, has looked at whether the direct-interaction mechanism of destabilization seen previously for lysozyme (Hua et al., PNAS 105, 16928; 2008) applies also to other proteins and peptides (Z. Yang et al., J. Phys. Chem. B jp304114h – paper here). They find that indeed urea seems to act in the same way here, via direct dispersion interactions with the peptide, rather than due to ‘structure-breaking’ effects on water.

The stability of biological macromolecules under conditions which mimic those of the intracellular environment, rich in cosolutes and subject to crowding, has become a focus of increasing attention. Shu-ichi Nakano and colleagues at Konan University in Kobe have investigated the hydration and conformational stability of DNA hairpin oligonucleotides under such conditions, specifically in the presence of PEG and other alcohols (Biophys. J. 102, 2808; 2012 – paper here). They find that aliphatic alcohol cosolutes decrease the stability of base-pair interactions, and may also destabilize the hairpins when the alcohols have fewer hydroxyl groups. In general it appears that binding of water molecules to DNA is promoted by structural order in the oligonucleotide.

Proton transport in many proton channels is believed to happen via hopping along a chain of hydrogen-bonded water molecules. This process in gramicidin A is studied by MD simulations, in particular to model the 2D infrared spectra of amide C=O stretches, by Jasper Knoester and colleagues at the University of Groningen (C. Liang et al., J. Phys. Chem. B 116, 6336; 2012 – paper here). They find that smearing of the isotope-labelled spectra seems to result from cooperativity of proton hopping and water rotation in the channel.

Water in protein channels is also the subject of a preprint from Diego Prada-Gracia and Francesco Rao at the Freibrug Institute for Advanced Studies (arxiv:1207.6953; paper here). They consider how a water molecule in the pore of a potassium channel (KcsA) affects the selectivity. In the presence of water, the equilibrium position for a sodium ion in the channel is shifted relative to the position both it and potassium occupy in a water-free channel. Such a water molecule does not actually enter the pore in the case of potassium – waters instead stay only outside, but close to, the pore entrance and exit, interacting more weakly with the ion inside. Thus the channel is able to bind both ions but on a quite different structural basis, with implications for its selectivity.

Three papers on hydration of small molecules: hydrophobic, the other hydrophilic. John Tatini Titantah and Mikko Karttunen at the Universities of Western Ontario and Waterloo respectively have performed first-principles MD simulations of the hydration of tetramethylurea with a view to testing the old ‘iceberg’ picture of Frank and Evans (JACS 134, 9362; 2012 – paper here). They note that while femtosecond IR measurements by Rezus and Bakker indicate a slowing of water dynamics in the hydration shell of hydrophobes (Phys. Rev. Lett. 99, 148301; 2007), Qvist and Halle recently found in contrast that the dynamics revealed by NMR are faster than the bulk (JACS 130, 10345; 2008). The authors here do find that some (but by no means all) of the hydration waters of TMU are slowed, and that some have a long residence time in the hydration shell. That would seem to support Rezus and Bakker, but unless I missed it, I can’t find an explanation for the discrepancy with the NMR results.

In the second paper, Laura Lupi at the Università degli Studi di Perugia and colleagues have used MD simulations to disentangle the dynamics of water molecules in the hydration sphere of trehalose from those in the bulk, clarifying the two dynamical populations revealed by light-scattering experiments (J. Phys. Chem. B 116, 7499; 2012 – paper here). They find that waters in the first hydration shell are retarded by a factor of about 5, while beyond this there is little significant perturbation of the dynamics relative to the bulk.

In contrast, an oxygen-17 NMR spin relaxation study of trehalose hydration by Bertil Halle and colleagues at Lund shows more modest perturbation of hydration water, with the rotation rates slowed only by a factor of around 1.6 in the first hydration shell (L. R. Winther et al., J. Phys. Chem. B jp304982c – paper here). They agree, however, that any effect on waters beyond this first shell is negligible. Moreover, these authors report significant clustering of the solute, which may or may not be linked to trehalose’s protein-stabilizing influence.

I believe I’ve mentioned previously work by Martin Weik at Grenoble and colleagues on the hydration of intrinsically disordered proteins, and how, by comparing folded globular and membrane proteins, the authors conclude that there is a gradient in the coupling between the dynamics of a protein and its hydration water that depends on the degree of intrinsic structure in the protein. In any event, it is now published (F.-X. Gallat et al., Biophys. J. 103, 129; 2012 – paper here). Using elastic incoherent neutron scattering, the authors find that the coupling seems to decline from a high value for these disordered proteins to more moderate coupling for globular proteins and then weak coupling for membrane proteins. As they put it, “A coherent picture thus emerges in which hydration water, rather than being a mere epiphenomenon, is an integral part of the biologically active protein.”

Teresa Head-Gordon at Berkeley and colleagues have devised a method for optimizing water-solute van der Waals interactions so as to accurately reproduce experimental small-peptide solvation energies in force-field calculations (P. S. Nerenberg et al., J. Phys. Chem. B 116, 4524; 2012 – paper here). Meanwhile, Wilfred van Gunsteren and colleagues at ETH present a coarse-grained water model that reproduces hydration behaviour of proteins at low computational cost (S. Riniker et al., J. Phys. Chem. B jp304188z – paper here). This entails the inclusion of a 0.8 nm later of atomistic water to capture hydrogen-bonding interactions with the protein while allowing only for coarse-graining beyond this – speeding up simulation times by about an order of magnitude.

The merits of the many water models currently in use, such as SPC, TIP4P, TIP5P etc, are much debated. Francesco Rao and coworkers at the University of Freiburg show that all seven widely used models produce the same hydrogen-bonding topology of water, albeit somewhat shifted to differing temperatures (R. Shevchuk et al., J. Phys. Chem. B 116, 7538; 2012 – paper here). The full microscopic geometry and radial distribution functions are not, however, identical in all cases.