Thursday, December 18, 2008

A nice Christmas package

There are some important and provocative papers in this batch…

Teresa Head-Gordon and her coworkers have extended their recent work on quasi-elastic neutron scattering in peptide hydration shells (e.g. Russo et al., J. Phys. Chem. B 108, 19885 and 109, 12966 (2005); Russo et al., Biophys. J. 86, 1852 (2004)) by using MD simulations to explore the way in which the hydration dynamics are affected by the heterogeneous, amphiphilic nature of most protein surfaces (M. E. Johnson et al., J. Phys. Chem. B doi:10.1021/jp806183v – paper here). The notion they proposed earlier is that the dynamics are most perturbed at the interfaces of hydrophobic and hydrophilic patches, due to the frustration created by different styles of hydration in the adjacent regimes. This now seems to be borne out by the simulations, where the water dynamics seen experimentally are reproduced for an amphiphilic peptide but not a hydrophilic one. The strongest dynamical perturbations are found for the first hydration shell of hydrophobic residues.

Jeetain Mittal and Gerhard Hummer have used simulations to try to clarify exactly what goes on at the interface of a hydrophobic surface and water (PNAS doi:10.1073/pnas.0809029105 – paper here). They are in particular examining the vexed question of whether there is a depletion layer in water density close to the surface, as proposed first by Frank Stillinger and invoked in the Lum-Chandler-Weeks model of dewetting-induced hydrophobic collapse (K. Lum, D. Chandler & J. D. Weeks, J. Phys. Chem. B 103, 4570; 1999). Experiments have now shown some evidence for a depletion layer perhaps 1-2 Å thick. But is there a sharp transition between a liquid-like and vapour-like phase, or a gradual thinning? In the former case, capillary waves are expected to blur the interface, so it’s hard to tell the difference. Mittal and Hummer find, for a purely repulsive spherical solute particle, that the interface is indeed rather sharp, but broadened by capillary waves in line with what theory predicts for a free air-water interface. The ‘dry’ layer looks to be instantaneously about 2 Å or so thick. The result is a flickering interface with patches that are intermittently dry and wet (in proportions that depend on the solute size), and transitions between them that are slow on a molecular timescale. This is all very illuminating, but I’m hard to satisfy – what happens when van der Waals forces between solvent and surface are included, I wonder?

Roland Netz at the TU Munich and his colleagues have also explored the depletion-layer problem from a very different angle. They have used MD simulations to examine how the slip length for water flow past a hydrophobic surface depends on the contact angle (D. M. Huang et al., Phys. Rev. Lett. 101, 226101; 2008 – paper here). Experimental studies in this area have given confusing and conflicting results, with slip lengths orders of magnitude different for surfaces that seem very similar. But the simulations show a rather systematic (though nonlinear) dependence of slip length on static contact angle. Moreover, they see depletion layers of molecular dimensions, whose average width varies with the ¼ power of the slip length. Thus, anything that influences the width of the depletion layer (dissolved gases) should have a marked effect on the slip length.

I referred recently to a study that challenged the notion of a dynamical transition for protein hydration water at 220 K and its interpretation as a fragile-to-strong crossover (S. Khodadadi et al., J. Phys. Chem. B 112, 14273-14280; 2008). Now here comes another one, from Michael Vogel at the Technical University of Damstadt (Phys. Rev. Lett. 101, 225701; 2008 – paper here). He has used deuterium NMR to study reorientational dynamics of hydration water for elastin and collagen, and sees no sign of a transition at 225 K. There is one at 200 K, but Vogel says that it corresponds to the onset, at lower temperatures, of thermally activated jumps in tetrahedral coordination, perhaps related to defect motion in the hydrogen-bonded network.

Fabio Sterpone and colleagues in Rome argue that the thermostability of proteins is primarily determined by protein-water interactions, with the intra-chain interactions between packed portions of the polypeptide being of only secondary importance (F. Sterpone et al., J. Phys. Chem. B doi:10.1021/jp805199c – paper here). They looked, using simulations, at the thermal stability and flexibility of three homologous proteins – one mesophilic, one thermophilic, and one hyperthermophilic. As thermal stability increases, so the proteins seem to be encased in an increasingly persistent hydration shell linked by hydrogen bonds. The idea, crudely speaking, seems to be that this shell supplies an increasingly robust protective coat against the penetration of water into the folded protein.

At the recent Hangzhou workshop I heard about the work of Shengfu Chen of Zhejiang University and colleagues on anti-fouling films that incorporate heterogeneously charged peptides. The idea is that the ability of these films to resist non-specific protein adsorption is linked to the nature of hydration of the surface chemical groups: the ‘more’ hydration there is, the stronger the disrupting influence of an incoming adsorbate and thus the more its attachment is inhibited. Shengfu and his workers in Washington and Taiwan develop this idea in a paper here (J. Phys. Chem. B doi:10.1021/jp8065713). They introduce a method for deducing the number of water molecules hydrating a given solute, and find that the greater the ‘hydration capacity’ of a solute, the greater its ability to resist protein adsorption in anti-fouling films.

Haiping Fang, my co-chair at that meeting, has an intriguing paper on the effect on water flux through a nanotube on the nature of the ‘outside structure’, in this case meaning whether the nanotube threads through a single, double or multiple sheets of graphene (X. Gong et al., Phys. Rev. Lett. 101, 257801; 2008 - paper here). In simulations, they find that the flux of water can be quite different in the various cases. For example, with two graphene sheets separated by a vacuum, the flux and flow both increases as the separation increases. And if water surrounds the nanotube in the space(s) between sheets, the flux is lowered. They deduce that interactions between water molecules inside the nanotube and the species outside the tube are responsible for the differences, emphasizing how sensitive, in this confined geometry with more or less single-file molecular traffic (where molecular motions are strongly correlated), the water transport is to the internal configurations of water molecules.

It seems clear that nanobubbles can form on hydrophobic surfaces, and very likely that these play a key role in the long-ranged hydrophobic interaction that is sometimes observed between such surfaces. The question has remained of how such bubbles, with a very high radius of curvature, can be stable when that curvature creates a large Laplace pressure which should lead to rapid diffusive efflux of gas out of the bubble. Michael Brenner and Detlef Lohse have considered this question (Phys. Rev. Lett. 101, 214505; 2008 – paper here). They say that the outflux can be balanced by an enhanced influx of gas at the contact line of bubble and surface, owing to the attraction of dissolved gas to the hydrophobic surface. They acknowledge that this is a non-equilibrium situation which suggests that in the long term the bubbles should disappear. But there haven’t yet been any long-term studies of closed systems to see whether that is the case.

Apparently sobering news from Michael Levitt and colleagues: MD simulations for protein structure refinement perform worse in explicit solvent than implicit solvent (G. Chopra et al., PNAS doi:10.1073/pnas.0810818105 – paper here). This seems to be because the potential in explicit solvent is more rugged, and so there is more chance of getting stuck in local minima unless the simulation is very long. So there are some situations in which it is still best not to consider the hydration shell molecule by molecule.

Angel Garcia at RPI and coworkers have calculated the stability diagram of the well-studied Trp-cage miniprotein (D. Paschek et al., PNAS 105, 17754; 2008 – paper here). They derive some insights into the role of hydration in pressure-induced denaturation, which they link in part to tighter packing of water around nonpolar atoms as pressure increases.

The debate over the ‘pH’ of the air-water interface continues. First-principles empirical-valence-bond calculations by Greg Voth and colleagues seem to indicate that the preference of hydrated protons for the surface (as claimed in their earlier work) is energetically (rather than entropically) promoted, due to the amphiphilic nature of the hydrated proton (S. Iuchi et al., J. Phys. Chem. B doi:10.1021/jp805304j – paper here). They say that much the same applies for a water-hydrophobe interface too.

Tuesday, December 9, 2008

A lot about interfaces

Janamejaya Chowdhary and Branka Ladanyi at Colorado State have used MD simulations to look at the dynamics of H-bonds at a water-hydrocarbon interface (J. Phys. Chem. B ASAP doi:10.1021/jp; paper here). They find that the reorientation of the O-H bond is anisotropic, and quantify the effects of cooperativity in the dynamics.

Robert Woods and colleagues at the University of Georgia study how bound water mediates the binding of concanavalin A to its target carbohydrate ligand (R. Kadirvelraj et al., JACS ASAP; paper here). Or rather, they look at a modified ligand of the natural trisaccharide, with a hydroxylethyl side chain that may or may not displace a conserved water in binding of the natural ligand. The crystal structure reported here shows that this water is retained, though its position is distorted. This helps to explain the previous thermodynamic data on ligand specificity for Con A, showing that there is no entropic component for the synthetic ligand arising from water displacement.

Roger Tam and colleagues in Ottawa have looked at the inhibition of ice recrystallization by mono- and disaccharides (JACS ASAP; paper here). Specifically, they look for correlates of ice-growth inhibition in the degree of hydration of the sugars, and find that, rather than using the total number of tightly bound water molecules, a better predictor of inhibiting ability is a hydration index in which the hydration number is divided by the molar volume. The researchers conclude that the inhibition arises from a disruption of water ‘pre-ordering’ at the ice-water interface.

Joe Zaccai and colleagues have measured water dynamics in human red blood cells using quasielastic incoherent neutron scattering (A. M. Stadler et al., JACS ASAP; paper here). In line with their previous work on E. coli, they find that most (90%) of the cell water has similar translational diffusion to the bulk, while about 10% is slower, this presumably being the water hydrating haemoglobin.

Sherwin Singer and colleagues at Ohio State have looked at the hydration dynamics of myoglobin using MD simulations (T. Li et al., J. Phys. Chem. B 10.1021/jp803042u; paper here). Specifically, they look at the time-dependent fluorescence Stokes shift after photoexcitation of the Trp-7 residue, a measure of the relaxation dynamics of the chromophore’s environment. The question is whether the water dynamics are due to constraint of the water by interactions with the protein, or whether they are controlled by the dynamics of the protein itself. This distinction should be revealed by arresting the protein in the simulations. Singer and colleagues find that doing so significantly changes the Stokes shift, suggesting that the intrinsic protein flexibility is important. They caution, however, that this does not necessarily imply that the water dynamics exhibit no intrinsic slow component of relaxation; rather, the protein and water dynamics are so intimately coupled that either slow water dynamics or slow protein dynamics (or both) could alter the Stokes shift.

Shekhar Garde and colleagues at RPI have conducted simulations of hydrophobically induced polymer collapse near to the interface with air or a hydrophobic wall (S. N. Jamadagni et al., J. Phys. Chem. B 10.1021/jp806528m – paper here here). They find that the driving force for collapse is smaller at the water-alkane interface, and all but vanishes at the air-vapour interface, where the polymer remains unfolded. They think that both the weaker hydration of the polymer and the enhanced density fluctuations of water at the interface produce faster conformational switches in the folded chain. The results throws up lots of interesting questions, most obviously of course what this implies for the conformational flexibility of two peptide chains approaching one another via the hydrophobic interaction.

Hangjun Lu and colleagues at Zhejiang Normal Univerity have looked at how an external charge of +1e near a carbon nanotube will affect the filling and emptying by water (H. Lu et al., J. Phys. Chem. B 10.1021/jp802263v – paper here here). It seems that the charge stabilizes the water-filled state when it is at the midpoint of the nanotube, but much less so if it is moved towards the ends. The implication is that this is a method that might be exploited by protein channels to control water transport via the positioning of ionized residues.

The freezing-point depression of water that hydrates phospholipid membranes has been studied using NMR by Dong-Kuk Lee at Seoul National University of Technology and coworkers (D.-K. Lee et al., Langmuir 24, 13598 (2008) – paper here). They find that water molecules still show liquid-like signatures below -20 C in bilayers, and that the freezing behaviour is depressed still further by cholesterol, a known cryoprotectant.

I have a kind of follow-up to my Chem. Rev. article in a forthcoming issue of ChemPhysChem, which has now appeared online (here). This will form part of a special issue on the subject of water at interfaces, stemming from a meeting of the DFG Forschergruppe 436 in Dortmund last summer.