Thursday, December 6, 2012

Salts: how far do they go?

In my previous post I commented on a recent paper by Jeremy Smith and colleagues (L. Hong et al., Phys. Rev. Lett. 108, 238102; 2012) in which it was shown that the coupling between the dynamics of a protein and its hydration shell can propagate from the surface to the core to soften up the entire molecule. Martin Weik at Grenoble, Frans Mulder at Aarhus and their coworkers have now investigated this idea experimentally via neutron scattering from the small protein calbindin D9k P43G deuterated in the one case on the “outside” and in the other case on the “inside” (K. Wood et al., Angew. Chem. Int. Ed. 10.1002/anie.201205898 – paper here). They find that both the exterior and the core of the protein are sensitive to hydration: the dynamics of both regions undergo a hydration-dependent dynamical transition in the same temperature range of around 250 K. In other words, this transition does appear to be a “global” one.

Ruhong Zhou at IBM Yorktown Heights is continuing to investigate how denaturants affect protein conformation and folding. In a paper with Eugene Shaknovich and coworkers, he reports the surprising finding that in a mixture of the denaturants urea and guanidinium chloride, hen egg-white lysozyme and protein L will both collapse, albeit with an increase in non-native hydrophobic contacts (Z. Xia et al., JACS 134, 18266; 2012 – paper here). The collapse (it is evidently not ‘folding’ in the proper sense) is induced by the specific interactions of the denaturants with the protein surface: GdmCl is absorbed there due to electrostatic interactions, while urea also accumulates near the first hydration shell and introduces crowding. This is a reminder of how specific, rather than generic, interactions of proteins with denaturants and osmolytes will dominate the behaviour, sometimes in non-intuitive ways.

Another counter-intuitive result is reported by Haiping Fang of the Shanghai Institute of Applied Physics and coworkers, relating to the evaporation of water molecules from solid surfaces. They find that the evaporation rate doesn’t change monotonically as one progresses from hydrophilic to hydrophobic surfaces, as one might expect, but has a maximum (S. Wang et al., J. Phys. Chem. B 116, 13863; 2012 – paper here). This follows from the fact that, while interactions between water and the surface dominate in situations where the surface is well wetted, for highly hydrophobic surfaces the water gathers into isolated ‘surface droplets’, in which case water evaporation is controlled by water-water interactions. The maximum results from a crossover between these two competing mechanisms. The findings might be important not only for, e.g. water retention in soils but also for the possibility of drying transitions in hydrophobic aggregation.

It’s become clear that even in relatively dilute solution some soluble small organic molecules may not be homogeneously dispersed. That notion is backed up by neutron-scattering experiments of Lorna Dougan at the University of Leeds and colleagues, in which they investigate glutamine solutions (N. H. Rhys et al., J. Phys. Chem. B 116, 13308; 2012 – paper here). Polyglutamine stretches of proteins are quite common and apparently important – they feature, for example, in some proteins that aggregate to induce neurodegeneration. Such aggregates seem to have collapsed polyglutamine regions, even though glutamate might be expected to form favourable hydrogen-bonding interactions with water. This motivated the present study, to investigate the interactions of glutamine monomers in water. It appears that glutamine forms some dimers via hydrogen bonds in both the ‘backbone’ and ‘side chain’ of the molecules, even at concentrations of 30 mg per mL, revealing a strong propensity to self-associate. Later work will address glutamine oligomers and polymers.

Lorna, in collaboration again with Alan Soper, has also used neutron scattering to investigate the clustering/microsegregation of the cryoprotectant glycerol in water (J. J. Towey et al., J. Phys. Chem. B 116, 13898; 2012 – paper here). They find that glycerol’s cryobiological effects don’t seem to stem from any effect it has on the hydrogen-bonding ability of water. Rather, glycerol molecules simply replace water molecules to allow the waters to retain their hydrogen-bonding capacity over a wide range of glycerol concentration. However, the presence of glycerol does segregate water into clusters at higher concentrations. So the researchers propose that the cryoprotectant activity stems not from any disruption of ‘water structure’ per se, but from glycerol’s very ability to substitute for water, while suppressing ice formation because of water segregation.

It is sobering to see that the dissociation of an ion pair such as NaCl –that’s to say, how and why salt dissolves – is still not fully understood. Andrew Ballard at the University of Maryland and Christoph Dellago of the University of Vienna present simulations of this process, using TIP4P water (J. Phys. Chem. B 116, 13490; 2012 – paper here). They say that the ion dissociation is favoured energetically but opposed entropically because of the water molecules entering the hydration shells of the ions. As in Dellago’s earlier work with Philip Geissler and David Chandler (J. Phys. Chem. B 103, 3706 (1999)), they find that the inter-ion separation is not a good reaction coordinate, as structures with different relaxation behaviour can occur for the same separation. The solvent fluctuations continue to play a role in the process over relatively long ranges, even into the third hydration shell.

It would be nice to know how that finding of a relatively long-ranged perturbation of water sits with the apparently surprising results of Sheeba Jem Irudayam of the UNC at Chapel Hill and Richard Henchman at Manchester. Their MD simulations of the hydration of alkali metal and halide ions show that this perturbation extends over remarkably long distances (J. Chem. Phys. 137, 034508; 2012 – paper here). They find non-bulk-like structural characteristics over virtually the whole simulation box (about 2.2 nm on a side for most ions, 3.3 nm for lithium and fluoride, corresponding to around 375 and 1200 TIP4P water molecules respectively). Specifically, there is a very slight but detectable excess of H-bond acceptors around halide ions, and of donors around alkali metal ions, which compensates for the shorter-ranged perturbations to water structure in the first and second hydration spheres. They also find long-ranged deviations for noble gas solutes: slightly enhanced tetrahedrality and an oscillating excess and deficiency of donors and acceptors. Similar perturbations are found for the air-water interface. The authors say that such small effects are likely to be invisible to standard spectroscopic techniques, although there have been some hints of them in neutron-scattering studies by Alan Soper and coworkers. To what extent they might have important thermodynamic consequences remain to be seen, but these results are highly intriguing.

In concentrated salt solution, ion-dependent effects on the O-D stretch are seen for both the anion and cation in pump-probe IR spectroscopic experiments by Michaal Fayer and colleagues at Stanford (C. H. Giammanco et al., J. Phys. Chem. B 116, 13781; 2012 – paper here). They find that in concentrated solutions it no longer suffices to divide water’s relaxational modes into ion-associated and water-associated fractions, because there is no longer any bulk-like component of the solvent. In this situation, water reorientational motions are highly cooperative.

Ionic hydration in confined spaces is important for a number of reasons not connected to water in biology, not least for understanding the interfacial behaviour in supercapacitors and batteries with nanoporous carbon electrodes. Tomonori Ohba of Chiba University in Japan and colleagues have explored this issue using synchrotron XRD of electrolytes inside carbon nanotubes (JACS 134, 17850; 2012 – paper here). They conclude that hydration is significantly different inside nanotubes with an average internal pore diameter of 2 nm, relative to the bulk. They can’t evaluate hydration numbers or detailed hydration structures, but say that the hydration structuring is stronger under confinement and that the hydrogen-bonded network of the solvent is correspondingly stretched and weakened. I must confess that I struggle to find an intuitive picture of what is happening here – but this perturbation is at least consistent with experiments showing a pore-size dependence of capacitance in double-layer capacitors (e.g. Chmiola et al., Science 313, 1760; 2006).

Transport of water through nanopores such as carbon nanotubes and aquaporin has important implications both for biology (e.g. functioning of ion channels) and technology (water purification). Kuiwen Zhao and Huiying Wu of Shanghai Jiao Tong University have used MD simulations to study water and ion transport through arrays of carbon nanotubes, driven by osmotic pressure (J. Phys. Chem. B 116, 13459; 2012 – paper here). In particular, they have looked at the effect of the packing density of the pores, and find that at high packing densities there can be steric interference of ions (and their hydration spheres) entering the pore mouths. There may therefore be an optimal packing density for efficient transport through the pore array, rather than simply trying to pack pores as densely as possible.

Acid or base? Yes, it’s the air-water interface again, and this time Agustín Colussi at Caltech and colleagues report experiments which seem to indicate that the interface is Brønsted basic (H. Mishra et al., PNAS 109, 18679; 2012 – paper here). They say that the controversies and discrepancies that have previously plagues this question stem at least partly from a failure to recognize acidity as a relative concept referring not so much to proton or hydroxide concentrations as to the extent of proton sharing between conjugate acid/base pairs. The researchers use electrospray ionization mass spectrometry of interfacial layers to measure the degree of dissociation of carboxylic acids both in the dissolved aqueous phase and when it collides in the gas phase with a water jet – probing, respectively, the ‘inner’ and ‘outer’ side of the surface. The detection of carboxylate ions indicates the presence of hydroxide at the surface, for all pH>2. Moreover, this surface excess of hydroxide can account for the observed negative charge at the air-water surface.

Mischa Bonn and coworkers at Mainz say that the bond orientational behaviour of water at the air-water interface is sensitive to nuclear quantum effects (Y. Nagata et al., Phys. Rev. Lett. 109, 226101; 2012 – paper here). They report quantum MD simulations showing that, while H2O and D2O have indistinguishable structures at the interface, HDO is quite distinct, with OD bonds oriented into the liquid and OH bonds oriented towards the gas phase. This would be because OD groups are able to form relatively stronger hydrogen bonds owing to a quantum isotope effect. They say this finding is in good quantitative agreement with SFG spectroscopic studies, e.g. by Geri Richmond.

The use of fluorescence microscopy to study freeze-dried biological samples can reveal details of water and ionic content of cells at the sub-cellular, nanoscale level, according to a paper by Jean Michel at the Université de Reims Champagne Ardenne in France and coworkers (F. Nolin et al., J. Struct. Biol. 180, 352; 2012 – paper here). Elemental (e.g. ion) distributions can be deduced from EDXS analysis, while specific protein densities can be studied by GFP-labelling. It looks like a nice method for deducing larger-scale patterns of hydration and ion distribution, for example that occasioned by chromatin compaction.

Water can undergo capillary evaporation from between two hydrophobic surfaces at small separations, but exactly how this happens hasn’t been fully elucidated. Sumit Sharma and Pablo Debenedetti at Princeton have investigated the process using MC simulations, and find that the outcome depends on the size of the surfaces (J. Phys. Chem. B 116, 13282; 2012 – paper here). If they are sufficiently large (3 nm square), evaporation involves the formation of a tubular cavity spanning the gap of the slit-like pore – an activated event as described by classical nucleation theory. But for 1 nm square surfaces there is too little space to accommodate such a vapour cavity, and the gap simply empties entirely.

It’s well documented that water in such confined spaces may also show reduced diffusional mobility. Hiroki Matsubara at Tohoku University in Japan and colleagues have attempted to figure out why this happens for liquids in general using MD simulations (Phys. Rev. Lett. 109, 197801; 2012 – paper here). They find that OMCTS (an alkylsiloxane) between two mica surfaces has a diffusion coefficient that depends on the surface separation, at least in the range 23-64 Å (3-7 molecular layers). This is because diffusion is an activated process, and the activation energy increases both because of the entropic constraint on some molecular configurations under confinement and the lowering of the average molecular potential energy in the gap. To what extent these effects operate and are modified for water remains to be seen.

How water mediates the association of a protein with its ligand is one of the most interesting issues concerning water’s role in molecular biology. Francesco Paesani of the University of California at San Diego and his colleagues present MD simulations which further support the notion that the water is involved as an active participant (R. Baron et al., J. Phys. Chem. B 116, 13774; 2012 – paper here). They are interested in extracting signatures of time-dependent changes in water structure and dynamics accessible to ultrafast vibrational spectroscopy. They model the binding between an apolar cavity and a spherical hydrophobic ligand, and find that the expulsion of disordered water from the cavity and suppression of slow water density fluctuations on binding result in an unfavourable entropic contribution to the binding free energy. Meanwhile, the reorientational dynamics of the water hydrating the ligand speed up as it approaches the cavity, and this water becomes less tetrahedral. This should be accompanied by the appearance of a shoulder on the O-D stretch mode for mixtures of HOD/H2O, as used in some recent spectroscopic studies, owing to the increasing concentration of dangling O-D bonds.

Francesco and his coworkers also present a new ab initio model for water, called HBB2-pol (V. Babin et al., J. Phys. Chem. Lett. 3, 3765; 2012 - paper here). They say it provides agreement with experiment from water dimers to the liquid state, capturing both the observed structural and dynamic properties while a avoiding the computational complexity that has previously rendered models such as WHBB intractable beyond small clusters.

Hydrophobicity may be the driving force behind the formation of some knots in protein structures. This idea is supported by work by Jeremy Sanders at Cambridge and colleagues, who show that a synthetic molecule composed of hydrophobic polyaromatics linked by hydrophilic amino acids will trimerize into a trefoil knot as the most effective way to ‘bury’ the hydrophobic surfaces (N. Ponnuswamy et al., Science 338, 783; 2012 – paper here).