Wednesday, August 20, 2008

Inside chaperonins, and dewetting for amyloids

If I were ever to do something as tendentious as deciding on a ‘paper of the week’, on this occasion it would be this one. Vijay Pande and colleagues at Stanford argue here that water trapped inside barrel-shaped enzymes called chaperonins could be crucial to the way they help proteins to fold (J. L. England et al., JACS doi:10.1021/ja802248m). This encapsulated water has generally been neglected previously. When the chaperonin complex GroEL+ES takes in an unfolded protein, it undergoes a conformational change to expose hydrophilic residues on its inner surface. Using MD simulations, Pande and colleagues show that this can be explained as a way of sequestering water in the cavity, which creates a stronger driving force for folding driven by hydrophobic interactions. Cavity hydrophilicity turns out to be well correlated with refolding rate, and fine differences found for various GroEL mutants can be explained on the basis of different spatial distributions of charged residues. The message is that this enzyme seems to mould the solvent micro-environment to favour folding in a generic way.

Joan-Emma Shea at UCSB and coworkers have teamed up with Bruce Berne, Ruhong Zhou and Lan Hua at Columbia to extend the latter group’s investigations of dewetting transitions in protein aggregation and folding to amyloids (M. G. Krone et al., JACS 130, 11066-11072; 2008 – paper here). They look at the formation of protofilaments from two parallel beta-sheets of segments of the Alzheimer amyloid-beta. Dewetting occurs in some but not all of the simulation trajectories – when it doesn’t, hydrophobic collapse is simultaneous with the expulsion of water from between the hydrophobic faces of the peptides. Dewetting always occurs when the van der Waals forces between the proteins and water are turned off, suggesting that these attractions may in reality be often sufficient to compensate for the loss of hydrogen-bonding in the confined water. Small changes in the simulation temperature can also tip the balance, suggesting both that the results of simulations like this may be highly sensitive to the nature of the molecular force fields used and also, I guess, that the balance between dewetting or not may be rather finely balanced in vitro/vivo as well as in silico.

More on the hydrophobic gap: Mark Schlossman at the University of Illinois at Chicago and colleagues have used X-ray reflectivity to probe the oil-water interface, both for heptane and for the extreme superhydrophobic case of perfluorohexane (K. Kashimoto et al., Phys. Rev. Lett. 101, 076102; 2008 – paper here). In both cases they find that any vapour-like depletion layer can be no thicker than 0.2 Å. It seems the evidence is now fairly overwhelming that a single hydrophobic surface in water is not in any meaningful sense ‘dry’.

Aggrecan, a proteoglycan with a ‘bottle-brush’ structure that is involved in the organization of the extracellular matrix of cartilage, seems to be extremely insensitive to salt, according to scattering experiments (SANS, SAXS, light) by Ferenc Horkay at NIH and colleagues (Phys. Rev. Lett. 101, 068301; 2008 – paper here). They find that the aggregation properties are very insensitive to calcium concentrations. This seems to be a necessary consequence of its biological role: aggrecan assemblies not only protect bone surfaces from wear and lubricate joints but also seem to provide a reservoir of calcium ions for bone mineralization. It’s interesting that nature could find a way of engineering such salt-independent properties into a polyelectrolyte – achieved, apparently, by virtue of the rigidity conferred by the side-chains.

I have tended naively to assume that we knew already all that needed to be known about the differences between heavy and light water. Clearly that isn’t so. Alan Soper and Chris Benmore use X-ray and neutron diffraction and simulation to refine the differences, and say that they have been underestimated. The OH bind length in water is longer than OD by about 3 percent, while the H-bond is about 4 percent shorter – making the H-O---H bond more symmetric than O-D---O (Phys. Rev. Lett. 101, 065502: 2008 – paper here).

Sony Joseph and Narayana Alura at the University of Illinois at Urbana-Champaign say that using electric fields to orient the dipoles of water molecules inside carbon nanotubes introduces a coupling between rotational and translational motions that creates a directional bias for diffusion, which can be used to pump the molecules through the tube (Phys. Rev. Lett. 101, 064502; 2008 – paper here). This is intriguing, although it seems to me that basically much the same result as was reported last year by Haiping Fang and colleagues (X. Gong et al., Nature Naotechnology 2, 709-712; 2008 – paper here).

Urban Johanson at Lund and coworkers have just published a high-resolution crystal structure of human aquaporin 5, with a fine view of the central pore (R. Horsefield et al., PNAS doi:10.1073/pnas.0801466105 – no link available yet). In contrast to other aquaporins, here the passage of gas molecules and ions seems to be prevented by a lipid occluding the central pore.

Bert de Groot at Göttingen has done a lot of work on the transport mechanism of aquaporins, and now he and coworkers have looked at the generic mechanism of ion permeation and gating in narrow peptide channels (G. Portella et al., Biophys. J. 95, 2275-2282; 2008 – paper here). I only have the abstract of this paper, but it appears they find that the free-energy barrier for ion permeation is predominantly entropic, arising from constraints on motion within the channels, rather than from the enthalpic cost of desolvation.

Wednesday, August 6, 2008

Some new perspectives on old debates

Garegin Papoian has extended his previous studies of how water-mediated contacts influence protein folding, with a new paper with Christopher Materese and Christa Goldmon (C. K. Materese et al., PNAS doi:10.1073/pnas.0801850105 – paper here). The basic notion is that a variety of contacts in the folding peptide – hydrophobic, hydrophilic, salt bridges – are ‘tried out’ during the folding process and filtered down to a subset of preferred interactions in a hierarchical branching process. This paper shows that many of these contacts are mediated by bridging water molecules, and that subsets of such interactions are characteristic of certain basins in the folding landscape. It adds to the argument that explicit water is essential for a full picture of the folding process.

Valerie Daggett and colleagues at the University of Washington in Seattle have carried out simulations of aquaporin embedded in a lipid bilayer to study the role of protein fluctuations on water transport (N. Smolin et al., Biophys. J. 95, 1089-1098; 2008 – paper here). Their aim was to look at some of the apparent discrepancies in earlier studies of how water passes through the protein pore in a hydrogen-bonded wire (e.g. Tajkhorshid et al., Science 296, 525-530; 2002; de Greet & Grubmuller, Science 294, 2353-2357; 2001). The key action seems to happen in the narrow constriction at the centre of the channel called the NPA region. Here, the dynamics of two asparagine residues seem to play a crucial role in aligning the water molecules for transport to occur. But beyond the constriction is a ‘valve’ region in which two residues, His76 and Val155, act to control the flow by potentially swinging into the channel to block the passage of the water molecules. The emerging picture, then, is of a remarkably orchestrated collaboration of side-chain and water dynamics to regulate the progress of the water along the ‘wire’.

H. Nagase of Hoshi University in Tokyo and his coworkers have continued their exploration of the molecular mechanisms of anhydrobiosis and how trehalose acts as a bioprotectant in this regard (H. Nagase et al., J. Phys. Chem. B. 112, 9105-9111; 2008 – paper here). They have studied the crystal structure of trehalose anhydrate, and find that it contains a one-dimensional channel threading between the trehalose molecules which may be filled with water in the dihydrate form of solid trehalose. This water uptake facilitates the transformation from the anhydrate to the dihydrate, and effectively makes the crystalline form a potential source and sink of water. If I understand this rightly, I believe the idea is then that this ‘water sponge’ prevents uptake of water by the amorphous (glassy) phase of trehalose thought to be responsible for bioprotection, which would otherwise lower its glass transition temperature.

Fernando Bresme at Imperial College and his coworkers have returned to the controversial question of a ‘hydrophobic gap’ or depletion layer at the interface of water with a hydrophobic surface (Phys. Rev. Lett. 101, 056102; 2008 – paper here). They model this interface as that between water and dodecane or hexane, which they study using computer simulations. Their objective is to decouple the intrinsic width and density profile of the interface with the effect of fluctuations from capillary waves, which will blur the details. They find that at 300 K water at the interface resembles that at the air-water interface – despite the fact that there is no appreciable intervening vapour film because the system is far from the drying transition. And the interface is rather rigid: corrugations remain well below a molecular diameter. But the water structure is significantly perturbed, with layering similar to that seen at a hard surface. Of course, this leaves open the question of what a nanoscopic film of water looks like between two such surfaces (see below), let alone the issue of how (if at all) a more rigid hydrophobic surface changes the situation. But it does seem to support the growing consensus that any ‘hydrophobic gap’ is extremely narrow.

On the same issue, there’s an interesting exchange in Phys. Rev. Lett. between Ben Ocko and colleagues and Steve Granick and coworkers (B. Ocko et al., Phys. Rev. Lett. 101, 039601 and A. Poynor et al., 039602; 2008 – letters here and here). Poynor et al. claimed previously (Phys. Rev. Lett. 97, 266101; 2006) that they see a depletion layer 2-4 Å thick with a reduced water density of at least 40 percent of the bulk. Ocko et al. say that, if the hydrogen-rich methyl groups of the hydrophobic monolayer are taken into account, the density deficit is much reduced, and might in fact be explained instead by local water orientation. Poynor et al. reject the latter interpretation, but agree that, as they stated, their originals density depletion was cited only as an upper bound. They point out that there does now seem to be agreement that a depletion zone exists (and, I guess, that it is very narrow and certainly not gas-like), and argue that the focus now should be on the role of fluctuations in the interfacial density. David Chandler has argued that indeed it’s the fluctuations (as opposed to the average equilibrium state) that matter for any discussion of how dewetting might occur between two such surfaces.

I have been meaning for some time to mention a paper by Alexander Pertsin and Michael Grunze (Langmuir 24, 4750-4755; 2008 – paper here) on simulations of the shear behaviour of water films between hydrophilic surfaces. Perhaps the delay was fortuitous, because into this discussion there now comes an experimental paper by K. B Jinesh and Joost Frenken at Leiden (Phys. Rev. Lett. 101, 036101; 2008 – paper here). Pertsin and Grunze previously simulated water monolayers (Langmuir 24, 135; 2008 – paper here), and found that they could observe essentially solid-like configurations in the confined layer. They now say that solidification can happen for bilayers too when sheared quasi-statically (that is, with an infinitely small shear rate) – but only for a small range of wall-to-wall separation, where the separation between the two monolayers is favourable for the formation of hydrogen bonds between them. The solid-like shear behaviour also depends on the relative alignment and period of the wall lattices. And importantly, the solid-like shear behaviour does not involve film crystallization. For trilayers, there is no solid-like behaviour at all.

So then, a complex picture. Now, already this seems to complicate the picture reported by Zhu and Granick (Phys. Rev. Lett. 87, 096104; 2001), where oscillatory shear of electrolyte films between mica showed no solid-like signature. One might add that Jacob Klein and colleagues have also seen the retention of fluidity in sub-nanometre confined water films under shear (U. Raviv et al., Nature 413, 51-54; 2001). Yet Jinesh and Frenken claim to see a solid-like response in their friction-force measurements of water between a graphite surface and a tungsten tip. Specifically, they see stick-slip behaviour which seems to change, with increasing humidity, from that expected for graphite corrugation to that with a different periodicity (around 0.4 nm), similar to a lattice periodicity of ice. The film thickness here is not known for sure, but is less than about 2 nm. (I notice that they have made this claim before on the basis of different evidence, which has caused a little confusion.)

Now, I’ve seen some criticisms of this latest work – for example, that it seems to attribute thermodynamic transitions from dynamic mechanical measurements, that it ignores the possibility of surface reconstructions of bulk ice or of a tip-sample potential and the role of the lateral spring constant in the cantilever in determining the 0.4 nm periodicity. This is a tricky issue – it would be unfair to level unattributed criticisms at the work, but neither can I pretend I haven’t heard them. I guess I can only say that there seems to be a debate in store, and until that happens we might best regard the results as no more than suggestive. In any event, if Pertsin and Grunze are right, there is likely to be a great deal of further subtlety to the question, not least in terms of the lattice periodicities and the hydrophobicity/hydrophilicity of the surfaces.

Staying with fundamentals, Noam Agmon, Greg Voth and their colleagues present a new look at the details of proton transport in water – classically explained by the Grotthuss hopping mechanism but now known to be a more complex, cooperative process (O. Markovitch et al., J. Phys. Chem. B 112, 9456-9466; 2008 – paper here). Their quantum-chemical calculations offer a fine-grained picture involving a range of time and distance scales: a ‘dance’ that embraces proton motions and water reorientations in both the first and the second hydration shells of the central hydronium ion.

An interesting paper by Haoran Li and colleagues at Zhejiang University in Hangzhou (X. Hu et al., J. Phys. Chem. B doi:10.1021/jp8028903 – paper here) explores not a biologically relevant process per se – the iron-porphyrin-catalysed activation of methane and methanol – but one that has interesting parallels to the functions of some cytochromes and horseradish peroxidise. In both of those latter cases, water molecules have been found to play important roles, particularly by providing bridges for proton transport to or from the heme group. Water has also been found to assist metal-porphyrin-catalysed oxidation in trace amounts, but to suppress the reaction when present in greater amounts. Li et al. find that a water molecule near the iron porphyrin can either assist or inhibit the catalytic processes considered here, depending on where it sits.