Friday, April 27, 2012

Now with pictures (what took me so long?)

Despite that previous post, I have found a bit of time while on travel, and so here's a catch-up.

Now here’s something very interesting. The ‘dewetting’ theory of protein assembly, according to which hydrophobic surfaces are drawn together at small separations by a cooperative, abrupt drying transition, has been much debated. Simulations have shown that dewetting can happen for flat hydrophobic plates, and also for hydrophobic polymers, but it seems to be rather rare for real proteins, in part because of their chemical heterogeneity: only a few hydrophilic groups on the surfaces are sufficient to suppress the transition, leading instead to gradual molecule-by-molecule expulsion of water. Now Song-Ho Chong and Sihyun Ham of Sookmyung Women’s University in South Korea have looked carefully at one particular protein assembly process: the dimerization of amyloid-beta, a 42-residue peptide linked to Alzheimer’s disease (PNAS doi:10.1073/pnas.1120646109 – paper here). The researchers study the aggregation with MD, and then apply the integral-equation theory of liquids to the simulated conformations to deduce the thermodynamics of solvation. They see no dewetting, but instead two distinct regimes of assembly. At relatively large separations it is the hydrophilic groups that seem to drive the attraction of the monomers, via an enthalpic interaction. Then as the protein surfaces come into contact, there is a switch to an entropically driven process that is water-mediated and involves dehydration of both hydrophobic and hydrophilic groups. I have not before seen anything much like this picture of protein assembly, although it does somewhat bring to mind the water-mediated contacts over long distances (1 nm or so) adduced in protein folding by Papoian et al. (PNAS 101, 3352; 2004).

The dewetting or capillary-evaporation picture of hydrophobic assembly has so far paid scant attention to the rate of water explusion in the hydrophobically confined region. Sumit Sharma and Pablo Debenedetti fill this gap, with some striking conclusions (PNAS 109, 4365; 2012 – paper here). They find that the (predominantly enthalpic) free-energy barrier to evaporation depends sensitively on the separation between the surfaces (assuming here a slit-like geometry), such that a change in separation from 9 to 14 Å results in a rate of evaporation that differs by ten orders of magnitude.

Another view of protein aggregation is offered by Alfonso De Simone at Imperial College in London and colleagues, who have looked at the formation of amyloid structures with non-pathological, functional significance: the hydrophobins that form robust protein coats on fungal spores, making them resistant to wetting (A. De Simone et al., PNAS doi:10.1073/pnas.1118048109 – paper here). What is striking about these proteins is that they are ‘designed’ to remain soluble unless they come into contact with a hydrophobic-hydrophilic interface, such as the air-water interface. The simulations in this study suggest that a section of the peptide chains that is highly flexible and disordered in bulk solution acts to suppress aggregation in that environment. At the air-water interface the proteins have drastically reduced access to his ensemble of conformations.

In contrast to this view, Martin Scholtz of Texas A&M and colleagues say that keeping proteins soluble seems to involve negative surface charge (R. M. Kramer et al., Biophys. J. 102, 1907; 2012 – paper here). They reach this conclusion by studying the effects on solubility of a range of proteins of adding two types of precipitant (that do not denature the proteins they precipitate): ammonium sulphate and polyethylene glycol 8000. They think that the observed correlation between solubility and surface negative charge is probably due to the strong water-binding propensity of glutamate and aspartate groups.

It is in fact the strong hydration promoted by these very groups that has been proposed as the reason for the stability of proteins in halophilic organisms, which have to resist the unfolding and aggregation that high salt concentrations might ordinarily induce. Bertil Halle and colleagues at Lund have probed this idea by looking at the water dynamics hydrating a halophilic protein Kx6E using oxygen-17 NMR (J. Qvist et al., J. Phys. Chem. B 116, 3436; 2012 – paper here). They find that these dynamics are not significantly different from those of the non-halophilic counterpart of this protein, challenging the notion of very tightly bound water molecules in the hydration sphere of halophilic proteins and, for that matter, the suggestion that cell water in extreme halophiles has slower dynamics than that in other cells (M. Tehei et al., PNAS 104, 766; 2007).

In a theoretical study of the hydration of hydrophobic surfaces using density-functional theory, Y. Djikaev and E Ruckenstein at SUNY at Buffalo support this notion that the hydration free energy here is primarily enthalpic (J. Phys. Chem. B 116, 2820; 2012 – paper here). They say that hydration is generally unfavourable at room temperature, and that the hydration free energy increases with increasing temperature, partly because entropic effects become increasingly negligible.

What is the hydrophobic interaction anyway? According to Snyder et al. (PNAS 108, 17889; 2011), we should start thinking in terms of a multiplicity of such effects; the work above seems to offer some endorsement for that view. Classically, the signature of a hydrophobic effect in protein folding has been considered to be a large difference between the heat capacities of the native and unfolded forms. But Robert Baldwin at Stanford, reviewing the history of how hydrophobic effects have been identified and rationalized (PNAS 10.1073/pnas.1203720109; 2012 – paper here), suggests that new definitions of the hydrophobic free energy are needed.

Some months ago, Umeda et al. reported a very beautiful crystal structure of photosystem II at 1.9 Å resolution, including a detailed picture of the hydration sphere (Nature 473, 55; 2011). They argued that several features of the hydration water distribution suggest that it forms a hydrogen-bonded network with specific roles in the catalytic mechanism of water splitting, such as relaying protons or other water molecules to active regions. Brandon Polander and Bridgette Barry at Georgia Tech have now looked at this idea in some detail (PNAS 109, 6112; 2012 – paper here). They focus on the oxygen-evolving complex, containing a Mn-Ca complex in which two water molecules ligate Mn and two Ca. By perturbing this H-bonded network with ammonia (which may substitute for water) and trehalose (which reverses the effect of ammonia), the authors deduce that the network is indeed critical for the photo-oxidation of water leading to oxygen evolution.

More on urea-induced protein denaturation. Yuguang Mu of Nanyang Technological University in Singapore and colleagues use MD simulations to study the salting in/out effects of urea and NaCl on amino acids and on a small amide mimicking a peptide backbone (W. Li et al., J. Phys. Chem. B 116, 1446; 2012 – paper here). They find, in line with experiments, that NaCl induces salting out, and urea salting in. By calculating the thermodynamic driving forces of these effects, they deduce that urea’s effect is caused not by the formation of hydrogen bonds with the peptide side-chain (or here, their mimics) but by an attraction mediated by van der Waals forces. This supports the idea that urea denaturation is mediated by direct interactions with the protein, but challenges the view that these interactions are hydrogen-bonding ones.

Susmita Roy and Biman Bagchi of the Indian Institute of Science in Bangalore present a picture of protein hydration in terms of water molecules in the hydration shell that are separated from the bulk by a free energy barrier against their escape (J. Phys. Chem. B 116, 2958; 2012 – paper here). They present MD simulations of hydration of the chicken villin head piece, which show that the residence time of hydration waters is strongly sensitive to secondary structure. In particular, a small subset of waters in one particular location, constituting about 5-10% of the total hydration shell, have residence times of around 100 ps. They seem to form a cluster that plays a central role in stabilizing the protein conformation by clamping together two of the protein’s three helices. Here then is another example of ‘quasi-bound’ waters becoming what one might consider as an intrinsic (albeit ‘loose’ and exchangeable) part of a protein’s secondary structure.


The effect of molecular crowding on hydration, for too long neglected, seems now to be getting the attention it deserves. The latest study comes from Michael Feig, currently at RIKEN in Kobe, and colleagues (R. Harada et al., JACS 134, 4842; 2012 – paper here). They have looked at how crowding affects water structure in concentrated solutions of a segment of streptococcal protein G by itself and combined with chicken villin head piece. They find that differences in local water density close to the dilute and crowded (>30% protein by volume – a physiological relevant situation) proteins are more pronounced beyond the first hydration shell; in particular, relatively ordered regions in the dilute case are often disrupted by crowding. What’s more, water diffusion is significantly reduced in the crowded case, and the dielectric constant is reduced. That latter change is especially significant, since it should reduce hydrophobic attraction and, by reducing electrostatic shielding, enhance the strength of hydrogen bonds and salt bridges. This in turn might be expected to stabilize secondary structure while destabilizing tertiary structure. These issues must surely be taken seriously in cell biology: how have proteins adapted, as one must assume they have, to withstand these likely perturbations in the strength of the forces that hold them together? What would this imply for protein-ligand binding constants and dissociation rates?

Proteins undergo slow conformational fluctuations, but they also exhibit higher-frequency collective vibrations akin to those seen in disordered matter. What role, if any, does the solvent play in these fast vibrations? Alessandro Paciarone at the Università degli Studi of Perugia and colleagues have investigated this issue by using terahertz spectroscopy to probe the rapid collective vibrations of essentially dry maltose binding protein (A. Paciarone et al., J. Phys. Chem. B 116, 3861; 2012 – paper here). In the dry state these motions tend to be faster, suggesting that as, one might expect, a reduction in hydration brings about an increase in rigidity.

The distribution of many ions in aqueous solution is perturbed at the air-water interface, where the ions may be either selectively adsorbed or depleted. It is not yet clear what drives this segregation, but a paper by Richard Saykally and colleagues at Berkeley sets out to elucidate that (D. E. Otten et al., PNAS 109, 701; 2012 – paper here). They use resonant UV SHG spectroscopy to look at the distribution of thiocyanate ions (classified in the old-style Hofmeister schemes as a typical ‘chaotrope’) at the interface as a function of temperature, in order to measure the thermodynamic variables. They find that the adsorption enthalpy and entropy changes are both negative. The former reflects a balance between hydration and surface energies of the solvent (for example, how much an ion ruptures solvent-solvent hydrogen bonds), whereas the entropy change seems to result from the way adsorbed ions alter capillary-wave fluctuations at the surface. An ion’s Hofmeister effect should therefore reflect a balance of these two, essentially independent factors.

The homogeneity or otherwise of aqueous solutions, particularly those of alcohols and polyols, is a subject of ongoing debate. There seems good reason to suspect that even rather simple alcohols show a significant degree of aggregation. This seems to be true of 2-butoxyethanol, as revealed in simulations by Rini Gupta and G. N. Patey of UBC in Canada (J. Phys. Chem. B 115, 15323; 2011 – paper here). They find that aggregates of the alcohol begin to appear at mole fractions (X) of more than about 0.005, and by X~0.04-0.02 these are micelle-like and rather large, so that large simulation systems (around 32,000 molecules) are needed to accommodate them. The size of the aggregates at these mole fractions – around 4 nm – is big enough to be seen in light-scattering and SANS experiments, as has already been reported.

There’s increasing interest on the structure, dynamics and thermodynamics of biological macromolecules at the air-water interface. Ozge Engin and Mehmet Sayar at Koç University in Turkey have looked at this issue for amphiphilic peptides designed to sit at such interfaces and to self-assemble into ordered systems and nanostructures (J. Phys. Chem. B 116, 2198; 2012 – paper here). Specifically, they look at small peptides designed to fold into beta-hairpins in solution, which could act as models for studying, e.g. amyloid formation. Perhaps unsurprisingly, the segregation of hydrophobic and hydrophilic residues at the water surface promotes hairpin formation above what is observed in bulk solution, and the molecules form well-ordered monolayers with anti-parallel stacking.

There’s a potential synergy here with a paper by David Vaux at Oxford and colleagues on amyloid formation at the air-water interface (L. Jean et al., Biophys. J. 102, 1154; paper here). They find that the interface promotes amyloidogenesis, largely due to a simple concentration effect: the (representative) amyloid peptides they study are surface-active, and so there is enrichment at the air-water boundary (i.e. this does not speak to the issue of whether, say, hydrophobic interactions are altered at the interface). The implications are partly cautionary: such surface effects could be complicating in vitro studies of amyloid formation and drug screening. But as the authors point out, complex interfaces are also present in vivo.

Forgive me if I’ve mentioned this before, but I think somehow I haven’t: there is an excellent summary of David Chandler’s ideas on dewetting and water fluctuations in a preprint with Patrick Varilly prepared for the International School of Physics "Enrico Fermi", Course CLXXVI - "Complex materials in physics and biology" in Varenna, Italy (a very lovely place), which happened back in July 2010. It’s at arxiv/1101.2235 (here), but doesn’t seem to have been published in a proceedings yet.

Friday, April 13, 2012

Small and often

My new resolution is to try to post more often and in smaller chunks, not least to avoid an impression of this site being moribund. With that in mind, I acknowledge that there is a fair bit of old stuff to catch up on which is not yet covered here. I also hope, when I have a moment, to find a way of posting that will make this blog accessible in China, which it is not at present because it seems the whole hosting network is blocked due to Google-related wrangles.

Water in protein hydration shells has retarded dynamics, for example in terms of reorientation. But how much and why? Reports vary, from slowing by a factor of a few to several orders of magnitude. Damien Laage and colleagues at ENS in Paris have tried to clarify the situation with simulations of lysozyme (F. Sterpone et al., JACS 134, 4116; 2012 – paper here). They find that most (80%) of the hydration water is slowed by a factor of just 2-3, and that the dynamics seem to be dominated by the same kind of activated jumps between H-bond acceptors as in the bulk. This slowdown seems to be due to an excluded-volume effect from the proximity of the protein surface, which reduces the number of transition-state configurations for reorienting jumps. The remaining water may be slowed to a greater degree, apparently due to water bound within clefts and pockets on the protein surface, where generally it is bound to H-bond acceptors.

Ben Corry and Michael Thomas at the University of Western Australia say that water plays a role in the selectivity of voltage-gated sodium channels (JACS 134, 1840; 2012 – paper here). Simulations based on a recent crystal structure show that unlike sodium, potassium ions can’t fit between a plane of glutamate residues with water molecules bridging the ion and the carboxylate groups – so these latter ions are excluded even though in principle they could pass through the pore with their complete hydration shell. This suggests that there are more subtle structural factors at work than (as has been suggested previously) simply the free-energy penalty of ion dehydration.

The hydrophobic effect and its role in the assembly of hydrophobic particles has generally been considered from the perspective of an equilibrium process, with no account of hydrodynamic factors. Bruce Berne and his coworkers at Columbia now seek to rectify this (J. A. Morrone et al., J. Phys. Chem. B 116, 378; 2012 – paper here). They simulate the interactions of two fullerenes in water, taking into account how molecular-scale hydrodynamics affects solvent density fluctuations and drying transitions. Perhaps unsurprisingly, a continuum picture breaks down at the smallest length scales : for example, the friction coefficient deviates from the continuum prediction at small particle separations, and can become non-monotonic due to layering. In general, these hydrodynamic effects can significantly reduce the diffusion-controlled rate constant for hydrophobic assembly. As the authors say, how such effects would become manifest in the crowded environment of the cell is another matter.

When is a protein unfolded? It’s not such an easy question as one might suppose: Sunilkumar Puthenpurackal Narayanan and colleagues in Japan say (Biophys. J. 102, L08 – paper here) that Catherine Royer and colleagues have recently found that staphylococcal nuclease can show a proton NMR signal at high pressures indicative of complete folding while Trp fluorescence data suggest significant unfolding. Narayanan et al. see something comparable for a subdomain of the transcription factor c-Myb R2 at high pressure and low temperature. This, they say, seems to be explicable on the basis that the protein remains folded but extensively hydrated, owing to water filling of a large internal cavity. This adds to the ongoing debate about the mechanism of pressure-induced denaturation, which some say is caused by the intrusion of water. More on this in a later post.

I wish I had a better grasp of a paper by Vladimir Sirotkin and Aigul Khadiullina of the Kazan Federal University in Russia on excess partial enthalpies of water and proteins (J. Chem. Phys. B 115, 15110; 2011 – paper here). But the key point seems to be that the progression from more or less dehydrated to fully hydrated states of several different proteins follow the same trajectory, at least in thermodynamic terms. The excess partial enthalpy is initially dominated by water, but then up to a weight fraction of 0.06 both protein and water contribute significantly. At this point the charged groups on the protein are hydrated and the proteins become flexible. And by a weight fraction of 0.5, hydration is complete and further changes are due only to contributions from the proteins. This seems loosely to be in accord with notions of a roughly universal ‘critical coverage’ for a protein hydration shell.