Thursday, January 19, 2012

More dynamical transitions?

Francesco Mallamace at MIT/Messina and colleagues have used proton NMR to follow the rearrangements of water during the thermal (heat and cold) denaturation of lysozyme (F. Mallamace et al., J. Phys. Chem. B 115, 14280; 2011 – paper here). Since the chemical shift is considered to probe hydrogen-bonding interactions in the water network, tracking it as the temperature changes take the protein from the folded to the denatured states reveal some aspects of how changes in protein structure are mirrored by those of the hydration water. The researchers conclude that water plays an active role in the process, and that as denaturation proceeds, the average number of H-bonds with which each water molecule is involved changes correspondingly.

How glycerol acts as a cryoprotectant is the subject of a study by J. Towey and L. Dougan at Leeds (J. Phys. Chem. B jp2093862 - paper here). They use neutron scattering to investigate whether, as one hypothesis has it, dissolved glycerol perturbs water structure to modify the hydrogen bonding ability and suppress ice formation. The solution turns out to be well mixed, with many glycerol monomers, but there is no discernible perturbation of the first coordination shell of water. The second shell, however, is perturbed in a manner similar to that caused by elevated pressure. Quite what this means for “the ability of water molecules to form ice” isn’t clear (indeed, I’m not even too sure what that expression means – displacement of the phase boundary?), but Towey and Dougan conclude that any explanation will need to focus not simply on local perturbations (or changes to water’s ability to form hydrogen bonds) but on changes to the extended hydrogen-bonded network.

More on the mechanisms of osmolytes: Dave Thirumalai and colleagues at Maryland have studied the stabilization of compact peptide conformations by trimethylamine N-oxide (TMAO) (S. S. Cho et al., J. Phys. Chem. B 115, 13401; 2011 – paper here). They attribute it to direct interactions between TMAO and the protein surface, which exclude solvent there. This is then a kind of excluded-volume effect analogous to the stabilization of proteins by molecular crowding (in which there is entropic destabilization of the unfolded state) – making TMAO a ‘nanocrowding’ particle.

In a related vein, M. Paulaitis at Ohio State and colleagues describe an integral-equation approach to deduce preferential interactions between cosolvents, which they say can be used to deduce preferential interactions of cosolvents such as osmolytes, naturants and cryoprotectants locally on the surface of proteins (M. H. Priya et al., J. Chem. Phys. B 115, 13633; 2011 – paper here).

Personally, I find it hard to think about hydration in crowded environments, in which marcomolecules might disturb one another’s hydration shells and sometimes temporarily associate with one another in non-specific ways. Sergio Hassan and Peter Steinbach at NIH have tried to provide a context from framing this question (J. Phys. Chem. B 115, 14668; 2011 – paper here). One issue is how incomplete and anisotropic hydration might create electrostatic effects. Another is how solvation forces due to structured hydration shells (layering, for example) manifest themselves on hydrogen-bonding at solute-water interfaces. Using a continuum solvent model, they say that the electrostatic effects of solvent exclusion can have a strong impact on protein-protein binding. But I think it fair to say that at this point the paper is largely presenting the methodology for investigating the problem, rather than reaching general conclusions about how these aspects of crowding affect the molecular biology.

Confinement will, of course, do other things to water itself. Ivan Brovchenko and Alla Oleinikova have predicted that water in slit-like pores just 2.4 nm wide might undergo the liquid-liquid transition predicted in the metastable bulk at low temperature and high pressure (J. Chem. Phys. 126, 214701; 2007). Limei Xu and Valeria Molinero at Utah have now examined that idea in simulations using their mW model of water held within 1.5-nm diameter cylindrical pores (J. Phys. Chem. B 115, 14210; 2011 – paper here). This system is comparable to the pores of MCM-41 nanoporous silica, as used in recent experiments on confined water (e.g. L. Liu et al., Phys. Rev. Lett. 95, 117802 (2005)). They find no evidence for a first-order liquid-liquid transition, but note that smearing of discontinuous transitions is well known in pores (although not in fact inevitable), and therefore that this doesn’t rule out the existence of such a transition in the bulk. For their simulations of the bulk phase, they do see a possible signature of a L-L critical point – a locus of maximum compressibility – but can’t study this in detail because fast crystallization makes it impossible to equilibrate a metastable water phase in this region.

How do proteins remain dynamic while remaining soluble and resistant to aggregation? Fabrizio Chiti at the University of Florence, Chris Dobson at Cambridge, and their colleagues, have sought to answer this by combining NMR relaxation data, H/D exchange experiments and MD simulations (A. De Simone et al., PNAS 108, 21057; 2011 – paper here). They use the fruitfly acylphosphatase as their model system, and find that the wild-type protein has free-energy barriers that limit access to aggregation-prone conformations except under aggrtegation-prone conditions (addition of small amounts of trifluoroethanol). They sum up the situation nicely: “The sensitivity of the energy surfaces of proteins to minor perturbations supports the view that there is a delicate balance between functionality, stability, and solubility, which is encapsulated by the concept of ‘life on the edge’”.

Hydrogen bonds hydrating hydrophobic regions of a protein or peptide seem to have a greater strength than those in bulk water. It’s been suggested that this might be not so much because the H-bonds are genuinely strengthened but because the orientational preferences of H-bonds in such a situation result in a depletion of weaker, strained H-bonds, i.e. a change in the population (Zichi & Rossky, J. Chem. Phys. 83, 797 (1985)). Peter Rossky and colleagues have now explored this idea further using MD simulations of a 16-residue peptide (J. Phys. Chem. B 115, 14859; 2011 – paper here). They find support for the idea, namely, water is depleted of near neighbours around apolar groups and so samples lower-coordination configurations that are undistorted and unstrained. In other words, the phenomenon is primarly a kind of packing effect.

A new angle on Hofmeister effects is offered by Huib Bakker at FOM Institute for Atomic and Molecular Physics in Amsterdam, who have looked at the orientational dynamics of water around various ions (K. J. Tielrooij et al., J. Chem. Phys. B 115, 12638; 2011 – paper here). When the salt consists of a strongly hydrated ion and a weakly hydrated counterion, the water molecules hydrating the former have impeded orientational dynamics, making it strongly anisotropic. In that case, they say that hydration is ‘semi-rigid’ in the first hydration shell: affected along one vector but not along others. If both ions are strongly hydrated, such perturbations of water dynamics extend well beyond the first hydration shell.

And while we’re there: Daryl Eggers and colleagues at San José State University have taken a thermodynamic line of attack on Hofmeister by determining the molar water volumes in various concentrated electrolyte solutions (A. Y. Payumo et al., J. Phys. Chem. B 115, 14784; 2011 – paper here). They find that the solutions are highly nonideal, presumably because of strong competition under these conditions for hydration water. Moreover, the solubility of the small amide diketopiperazine follows the Hofmeister series for all the anions and cations studied, and the authors explain this on the basis that Hofmeister effects are governed by changes in the average free energy of the bulk aqueous phase – that is, if Hofmeister effects are a bulk phenomenon of water.

The tale of the low-temperature dynamical crossover for hydrated proteins and their hydration shells continues to get more complicated. Using dielectric spectroscopy and MC simulations, Gene Stanley, Giancarlo Franzese and their colleagues now report evidence of two such crossovers for the hydration water of lysozyme: one at about 252 K, the other around 181 K (M. G. Mazza et al., PNAS 108, 19873; 2011 – paper here). Marie-Claire Bellissent-Funel and her colleagues have previously seen something similar – two transitions at 220 and 150 K (J.-M. Zanotti et al., PCCP 10, 4865; 2008). Stanley et al. now ascribe the first of these to maximal fluctuations in the making and breaking of hydrogen bonds, and the second to maximal fluctuations in cooperative reordering of the H-bonded network.

Meanwhile, Sol Gruner and colleagues at Cornell report evidence for another protein dynamical transition right down at 110 K, which they say correlates with the transition of the hydration water from a high- to a low-density amorphous state (C. U. Kim et al., PNAS 108, 20897; 2011 – paper here).

Wilfred van Gunsterden at ETH and colleagues show how solvation free energies, as well as the free energies of protein-ligand binding and protein conformational dynamics, can be calculated using a new software package called GROMOS, which van Gunsterden and colleagues have introduced in a paper in press with Comput. Phys. Commun. (S. Riniker et al., J. Phys. Chem. B 115, 13570; 2011 – paper here).

Alan Soper has an intriguing paper in J. Phys. Chem. B (115, 14014; 2011 – paper here) in which he presents a new mixture model of water. This isn’t exactly a two-state model – the two forms are intimately mixed – but it postulates two populations of water molecules, each of which can form hydrogen bonds only with molecules of the other type and not with those of their own type. This is not, as far as I can see, intended as a literal representation of some molecular-scale distinction – both types of water molecule have identical structure – but is imposed as a device for introducing a three-body term into the interactions. The results show good agreement with structural studies using neutron and X-ray scattering, and give rise to a situation where water molecules are H-bonded to some of their neighbours but not others. This, Alan suggests, is perhaps why mixture models of water have been so enduring: not because there really are two distinct populations but because – if I’m understanding this correctly – the three-body terms have the effect of making it appear that way.

Michele Parrinello and colleagues at ETH have investigated the recombination of hydronium and hydroxide ions in water using ab initio MD simulations (A. Hassanali et al., PNAS 108, 20410; 2011 – paper here). They find that the mechanism is rather different from what has traditionally been assumed in terms of a Grotthuss mechanism. The researchers say that the Grotthuss mechanism serves to bring the hydronium and hydroxide to a distance of around 6 Å, when they are bridged by two water molecules as a ‘water wire’. But then there is a collective compression of this water wire that results in a concerted motion of three protons (rather than a series of distinct one-proton hops), converting both ions into water molecules.

Tuesday, January 10, 2012

Welcome to 2012

Lots to catch up on here, and while I’m not going to do that exhaustively now, here is an update to let you know that this blog will still be active in 2012.

I’ve recently written a News & Views article for Nature (478, 467; 2011 – here) about, among other things, the recent paper by George Whitesides’ group (PNAS 108, 17889; 2011 – paper here) on the hydrophobic effect in ligand binding – which, as discussed in the previous post, suggests that there is not any single ‘hydrophobic effect’ operating here, but a delicate and case-specific balance of enthalpic and entropic effects. They noted that in the example they studied, the ‘hydrophobic’ aspect of binding was driven largely by the enthalpic effect of displacing/rearranging water around nonpolar contacts. That notion is supported by a paper by Stephen Martin and colleages at the University of Texas at Austin, who find something rather similar for small peptide binding by the SH2 domain of the growth receptor binding protein Grb2 (J. M. Myslinski et al., JACS 133, 18518; 2011 – paper here). They say that increases in the nonpolar contact area don’t necessarily lead to entropic gains due to release of water, but have a primarily enthalpic influence on binding affinity.

How quickly drugs bind to their target molecules can have important pharmacological implications, but these kinetics are generally poorly understood, and thus to design rationally. Xavier Barril from the University of Barcelona and colleagues show using MD simulations that slow kinetics are often a consequence of the presence of buried polar atoms which form hydrogen bonds that are shielded from water, because of the slowness of dehydration/rehydration (P. Schmidtke et al., JACS ja207494u – paper here). In other words, the pace of events in binding can be set by the degree of water accessibility. They show that this effect can be predicted from structural data, and can thus potentially be accessible to design.

Knowing the hydration structure of proteins is important for small-/wide-angle X-ray scattering (SWAXS) studies of proteins not so much for its own sake but because the contribution of the solvent to the scattering must be subtracted in order to extract information about the protein secondary structure. It’s easy enough to extract the contribution of bulk water, but correcting for the scattering of the hydration shell is complicated. Tobin Sosnick and coworkers at the University of Chicago now show two ways to do this (J. J. Virtanen et al., Biophys. J. 101, 2061-2069; 2011 – paper here). One is to deduce the solvation structure by full MD simulations, which they say gives results that match the SWAXS data closely. But they have also developed a much more computationally less intense solvation model called HyPred, which gives a scattering profile that agrees well both with MD and with experiment.

Bernhardt Trout and his coworkers at MIT present an illustration of just how complex the interactons of proteins and ions can be (D. Shukla et al., JACS ja205215t – paper here). Complex ions such as guanidinium create a particularly intricate picture. Gdm+ can destabilize proteins via the formation of hydrogen bonds and electrostatic interactions, but when paired with an anion that is a hydrogen-bonding acceptor it can form clusters with the ions, which suppresses the effect. Molecules (such as arginine) with multiple Gdm+ groups are sometimes used to suppress protein aggregation, and can do so without compromising protein stability. Trout and colleagues investigate the effect of arginine oligomers (n=1-4) on protein aggregation and conformational stability for two different anions, chloride and sulphate. While monomeric arginine chloride is used as an aggregation suppressor, the n-mers only inhibit aggregation at low concentration – they actually accelerate it at moderate to high concentration. Meanwhile, the sulphates inhibit aggregation at all concentrations. And while the chlorides reduce protein stability, the sulphates enhance it. The researchers explain all this in terms of the balance between ion-ion and ion-protein interactions.

All this mirrors an increasing tendency to consider Hofmeister effects of ions on proteins in terms of direct interactions between the two species, rather than as indirect consequences of changes in hydration. Elena Algaer and Nico van der Vegt at the TU Darmstadt provide some support for this notion with a study of the salting-in and –out of small model amides by various sodium salts (J. Phys. Chem. B jp208583w – paper here). They say, for example, that the salting-in if NiPAM by NaI is mediated by interactions of iodide with the nonpolar groups. Such interactions also explain why, of all the salts studied, the iodide alone fails to induce hydrophobic collapse of polyNiPAM.

How does water permeate cell membranes? Its passage through the water-regulating membrane protein aquaporin is fairly well studied, but little is known about other water transporters. The bacterial sodium-galactose transporter vSGLT and its human homologue the sodium-glucose cotransporter hSGLT1 both have the potential to let water through. Jean-Yves Lapointe at the Université de Montréal and colleagues have used MC and MD simulations to show that indeed these protein pores can be filled with water (a pathway of about 100 molecules) which allows passive water permeation (L. J. Sasseville et al., Biophys. J. 101, 1887-1895; 2011 – paper here). This pathway depends on the proteins’ conformation: there is a constriction at one point which reduces the water bridge to a single-molecule chain which then ‘snaps’ at a ‘hydrophobic plug’, creating a 4.3 Å gap of low water density. But the resulting barrier to water permeation is conformation-dependent, and can be altered by varying the membrane potential. The passage of a sugar molecule also can bring water with it, but the mechanism of this is still open to debate.

Jhih-Wei Chu and colleagues at Berkeley conclude from MD simulations that the insolubility of cellulose in water is an entropic effect due primarily to the reduction of solvent entropy if the glucan chains in a fibril unravel (A. S. Gross et al., J. Phys. Chem. B 115, 13433; 2011 – paper here). There are some lessons here for how to solubilize cellulose in other solvents, such as ionic liquids.

Why and how do ions segregate at the air-water interface? This phenomenon seems well attested, with anions, especially large and polarisable ones, tending to accumulate at the interface. But the reason for this is still debated. Yi Qin Gao if the Beijing National Laboratory for Molecular Sciences and coworkers investigate the question with MD simulations, looking in particular at the differences in how anions and cations are solvated (L. Yang et al., J. Phys. Chem. B jp207652h – paper here). They argue that these differences are due to the charge distribution in the water molecules themselves, and that water can approach anions more closely than cations. They suggest that these differences in hydration account for why anions tend to populate the interface more readily.

And on the same topic, Pavel Jungwirth and colleagues have looked at how the hydration of guanidinium ions affects their orientation at the air-water interface (E. Wernersson et al., J. Phys. Chem. B jp207499s – paper here). Guanidinium is depleted at the interface, but the ions that do stay there are preferentially oriented parallel to the surface: in this configuration, it can sit at the surface without needing to break hydrogen bonds. Another way of looking at this is that the ions can take advantage of the deficit of hydrogen bonds between waters at the surface. The authors suggest that similar reasoning might account for the unexpected orientation of some arginine groups (with an analogous structure) in protein side chains.