Wednesday, January 31, 2007

Proteins through a glass darkly

There is a long-standing idea that proteins undergo something like a glass transition at around 200-220 K: above this temperature the peptide chains can diffuse, while below it they are trapped and exhibit harmonic vibrations. It has also seemed to be the case that this behaviour is intimately connected to that of the hydration water, which apparently undergoes a similar transition. (See E. W. Knapp, S. F. Fischer & F. Parak, J. Am. Chem. Soc. 86, 5042-5047 (1982); B. F. Rasmussen, A. M. Stock, D. Ringe & G. A. Petsko, Nature 357, 423-424 (1992); R. F. Tilton, J. C. Dewan & G. A. Petsko, Biochemistry 31, 2469-2481 (1992); I. V. Borovykh, P. Gast & S. A. Dzuba, J. Phys. Chem. B 109, 7535-7539 (2005); V. Reat, R. Dunn, M. Ferrand, J. L. Finney, R. M. Daniel & J. C. Smith, PNAS 97, 9961-9966 (2000); A. R. Bizzarri, A. Paciaroni & S. Cannistraro, Phys. Rev. E 62, 3991-3999 (2000); C. F. Wong, C. Zheng & J. A. McCammon, Chem. Phys. Lett. 154, 151-154 (1989); C. Arcangeli, A. R. Bizzarri & S. Cannistraro, Chem. Phys. Lett. 291, 7-14 (1998); A. L. Tournier, J. Xu & J. C. Smith, Biophys. J. 85,1871-1875 (2003)). Bizzarri and Cannistraro (J. Phys. Chem. B 106, 6617-6633; 2002) have speculated that the dynamics of the protein and solvent are so strongly coupled that they ‘should be conceived as a single entity with an unique rough energy landscape.’ In other words, the protein motions are not simply ‘slaved’ to those of the solvent, but ‘the very topological structure of the protein energy landscape could be deeply altered by the spatial organization, as well as by the dynamical behaviour of the hydration water.’

But is this truly glassy behaviour that we're seeing in the proteins and the hydration sphere? Jürgen Köhler at the University of Bayreuth and his colleagues don't think so. They've looked at spectral diffusion in individual molecules of the light-harvesting LH2 protein complex in purple bacteria (see paper here) at 1.4 K, and say that the dynamics don’t fit the standard two-level-system model used to understand spectral diffusion in glasses.

Gene Stanley at Boston University and his coworkers don't think so either. In a paper published in Phys. Rev. Lett. towards the end of last year they argued that the dynamical transition in proteins is in fact driven by a change in the diffusivity of the hydration water, which is itself caused not by a glass-like transition but by a crossover in dynamics related to the critical point of a liquid-liquid phase transition in water, predicted to occur at around this temperature (200 K) but at high pressure. Beyond the critical point (that is, in the one-phase region of the liquid), a 'ghost' of the first-order transition remains in the form of a 'Widom line' where the response functions of the liquid are maximal. That would certainly provide a reasonable explanation for why DNA seems to show the same kind of dynamical change at much the same temperature (around 247 K) – in both cases, the change in macromolecular behaviour is driven by a change in the solvent. Thus, says Gene, the protein glass transition is not a transition, not a glass, and not protein.

Tuesday, January 23, 2007

Mostly hydrophobic

With a daunting glut of papers this week, I can do little more than list them. Daniel Blankschtein and colleagues at MIT have a series of three papers the aim to quantify the hydrophobic effect responsible for aggregation of amphiphiles in solution. At issue here is the question of how changes in hydration of the different parts of amphiphile molecules on aggregation into structures such as micelles (but also, of course, bilayer membranes) provide a driving force for the self-assembly process. Blankschtein and colleagues explore this through computer simulation. The three articles look respectively at aggregation of oils, non-ionic surfactants, and ionic/zwitterionic surfactants. The general aim is to be able to model and predict the changes in hydration, and the consequent hydrophobic driving force for aggregation, sufficiently to be able to predict bulk parameters such as the critical micelle concentration. The work is, as far as I can see, at this stage concerned with developing and validating the methodology; presumably it could at some stage be used to look at the more complex structures formed by lipids, and perhaps at protein aggregation and self-assembly too.

That's in a sense also the topic of a paper in Biochemistry by Ronald McElhaney of the University of Alberta and colleagues, which asks how hydrophobic an alpha-helical peptide needs to be in order to get inserted stably into a phospholipid bilayer. Specifically, they have looked at Leu-Ala sequences, where it seems that Leu/Ala ratios of more than 7/17 are needed in the helices for a stable transmembrane association. OK, so that's just a number, though presumably the kind of thing one needs to know in some areas of protein design. It also helps, perhaps unsurprisingly, to put the hydrophobic residues together on one side of the helix.

These questions of hydrophobic hydration are often examined by looking at small model hydrophobes. Benzene is a classic example, though it's not exactly a simple hydrophobe – it's been clear for some time that water molecules can hydrogen-bond to the pi-ring system. Markus Allesch at the Graz University of Technology and colleagues have studied the hydration of benzene, and also of hexafluorobenzene, in some detail using first-principles calculations. They find that both molecules act as hydrophobes equatorially, but that the pi-water interactions are quite subtle in the axial regions. A water molecule typically points towards the ring hydrogen-first for benzene, but lone-pair first for hexafluorobenzene. So one clearly shouldn't generalize even about something as seemingly simple as phenyl-group hydration.

I remember Jacob Israelachvili and Hakan Wennerström ruffling some feathers with their review article in Nature on the hydration of hydrophilic surfaces. It was controversial then, and it still is now. One of the issues is why phospholipids bilayers repel one another at short ranges. Alexander Pertsin of the University of Heidelberg and his colleagues have explored the matter in a paper in Langmuir using Monte Carlo simulation. They point out that there are two leading theories: the repulsion is either entropically driven (confinement by proximity of the bilayers suppresses fluctuations and so decreases entropy), or it arises from a perturbation of water structure in the intervening film. (Israelachvili and Wennerström were dismissive of 'water structuring' theories). There are, for example, suggestions that orientation of the interfacial water plays a role. I'm still chewing on this paper, but it clearly supports the view that the repulsion arises from hydration changes rather than entropic effects or protruding lipid headgroups. I've got a feeling this isn't going to be the last word on the matter.

Finally – well, not quite, but for now – Barry Ninham and his collaborators in Italy have been probing the puzzles of the Hofmeister effect: ion-specific effects of electrolytes on proteins. The classic effect identified by Hofmeister himself was on protein precipitation, but there's plenty else to this phenomenon. In one study, Barry and co. find that different sodium salts have markedly different effects on the enzymatic activity of a lipase: NaSCN can inactivate it completely, while sodium sulphate activates it and sodium chloride has little effect at all. What's going on? The researchers eschew generalized and, I think, questionable ideas about whether particular ions 'make' or 'break' water structure (that is, whether they are kosmotropes or chaotropes), and focus instead on how the specific ions might interact with the protein and its hydration shells. The devil is in the details.

A second paper looks at Hofmeister effects in phospholipids aggregation. Again, the issue is exactly how the ions interact with the organic molecules and their hydration shells: in this case, how they affect hydration and packing of lipid headgroups in micellar aggregates. That, it seems, may depend in this case on the anionic polarizabilities. It all gives me some faith that structure-making and structure-breaking is a concept on the way out – though the alternative will be messy, and it's not yet clear that it will yield handy generalizations.

Wednesday, January 17, 2007

What urea does to water

Urea is a model small hydrophilic solute, but also a denaturant for proteins. Both of these things make it of much interest to know how urea is solvated. It’s commonly said that urea acts as a ‘structure breaker’ of water, promoting the formation of a dense, disordered liquid structure. That suggestion is reiterated, but then challenged, in a paper in JPC B by Yoshihito Hayashi and colleagues from the labs of Sony in Japan. They’ve used dielectric spectroscopy to study the solvation of urea, and conclude that urea ‘fits’ rather easily into the tetrahedrally H-bonded structure of water and thus doesn’t greatly disrupt the structure. It seems that dielectric spectroscopy is a somewhat blunt tool for probing such questions, but as Hayashi et al. point out, their conclusions are consistent with those of Alan Soper and colleagues (Biophys. Chem. 105, 649; 2003), who studied the problem using neutron scattering. Strangely, perhaps, Hayashi et al. examine the question using the framework of the two-state model of Frank and Franks (J. Chem. Phys. 48, 4746; 1968), which posits an ‘ice-like ordered’ phase of water coexisting with a ‘dense disordered’ phase. I don’t know that anyone really thinks such a model is realistic any longer for water at room temperature… Neither am I convinced that the whole concept of structure-making and -breaking is very helpful for understanding hydration of solutes.

Anyway, here’s what I say about Soper’s study, and the more general issue of urea hydration, in my review article:

A neutron-scattering study shows that the urea molecule can ‘substitute’ quite readily for water in the hydrogen-bonded network: the radial distribution function of urea around water in a 1:4 solution looks remarkably like that of water around water. Although urea has nearly three times the molecular volume of water, the structure of liquid water is sufficiently ‘open’ that a urea molecule appears to displace just two waters, offering up to eight hydrogen bonds in place of the displaced pair. Despite this apparently ‘easy’ substitution, however, incorporating urea into the network appears to disrupt it, creating a local compression of the second hydration shell around water molecules in a manner similar to the effect of high pressure on the liquid. Yet the orientational dynamics of the water molecules seem largely unaffected even at urea concentrations high enough for all the water to be part of hydration shells. Only one water molecule per urea, on average, has a significantly slower reorientational time constant (about six times that of bulk water) – which can be rationalized according to a hydration structure in which one water is complexed to the urea molecule via two hydrogen bonds.

Postscript: Dave Thirumalai at the University of Maryland has drawn my attention to some previous papers that he and others have published on the solvation of urea and its influence in denaturation. In JACS 120, 427 (1998), he finds from MD simulations that urea, far from destabilizing the hydrophobic interaction, actually stabilizes the attraction between two solvated methane molecules, which would lead to the expectation of it acting as a renaturant. But the situation is changed for hydrophilic/charged species, to which urea becomes absorbed. In proteins, this would lead to a repulsion between hydrophilic groups, causing swelling, which would expose and destabilize the ‘buried’ hydrophobic residues. This electrostatic origin of urea denaturation was supported by further calculations in 2003 (JACS 125, 1950; 2003), where Dave and Ray Mountain simulated a hydrocarbon chain in water with slight charges at each end. Urea solvates these charged chain ends, which destablizes the collapsed chain.
For the hydration of urea itself, Dave and Ray found in 2004 (JPC B 108, 6826; 2004) that hydrogen-bonding is not the whole story – there is an important excluded-volume contribution too: the larger size of urea limits its ability to form as many hydrogen bonds as is theoretically possible. The simulations also showed that the water H-bond network around urea is perturbed significantly – it is too simplistic to say merely that urea slots nicely into the network.

More on the hydrophobic gap

Alongside Steve Granick’s paper below on the hydrophobic gap, I should have mentioned a study by Harald Reichert of the MPI in Stuttgart and colleagues, published in PNAS in December, which comes to much the same conclusions: X-ray reflectivity shows a definite depletion layer for water against an alkylsilane monolayer. They’re not able to pin down the width of the gap beyond limiting it to 1-6 Å, but the integrated density deficit (density x width) is 1.1 Å g/cm3 – and like Granick et al., they rule out the possibly influence of dissolved gases forming nanobubbles.

Even more gratifyingly, this is now accompanied by a study in Langmuir
by Marco Maccarini at McGill and colleagues, which uses neutron reflectivity to probe density depletions at solid-liquid interfaces for both water and non-polar liquids, the latter in contact with hydrophilic surfaces. This is precisely the sort of comparison that has been previously neglected: rather than focusing on water as something unusual, we need to establish to what extent it behaves just as other liquids do. And indeed Maccarini et al. find that a solvophobic gap is a quite general phenomenon: they say “The results show that the density deficit of a fluid in the boundary layer is not unique to aqueous solid-liquid interfaces but is more general and correlated with the affinity of the liquid to the solid surface.”

This all seems to add up to a rather consistent and satisfying story. But I remain troubled by one thing. Work in this area generally seems to posit the problem as one motivated by previous contradictions and discrepancies in the experimental data but about which we can say nothing that is not empirical. Being led by experiment seems like a good principle; but there is a well established literature on the theory of inhomogeneous fluids and their behaviour at surfaces, with which these discussions rarely connect. This theoretical work provides a general framework for thinking about the problem which does not immediately plunge into considerations of hydrophobicity, dangling hydrogen bonds and so on but starts, as it should, from consideration of wetting and contact angles, and the structure of simple liquids. (Maccarini et al. do touch on this.) The prediction for hard spheres against hard walls – that is, purely repulsive (contact) forces – is clear: the density profile is oscillatory, with a sharp peak at 1r (r = molecular radius) and decaying oscillations with a period of around 2r after that. In other words, there is molecular layering due solely to packing effects, even before one starts to take attractive forces into account. The question is how this picture is modified for more realistic fluids. A Lennard-Jones potential gives something similar – the short-ranged repulsion dominates the structure. Density functional theories show similar density profiles, with the layering getting flattened out as the reduced temperature T/Tc starts to approach 1. Getting appreciable depletion – partial drying – near the surface requires rather large contact angles, as I recall (this was 20 years ago).

Now, the directionality of the hydrogen bond in water may well change this picture in significant ways, but it seems logical to me to start with the expectations for a simple liquid and to go gradually from there to the complexities of water, so that we can see what is generic and what is not. So I’m keen to see these illuminating new results cast in the context of the theory of inhomogeneous fluids, so that we can start to develop a unified view and not treat water as a unique case, nor indeed enshrine the concept of a ‘depletion layer’ without relating it to our understanding of wetting and drying phenomena. The study by Maccarini et al. is a step in that direction.

Wednesday, January 10, 2007

Hydrophobes and halophiles

The clincher for starting this blog was a glut of deeply interesting papers over the past couple of weeks. In Phys. Rev. Lett. (see here) Steve Granick and colleagues have what they call conclusive evidence for a depletion layer where water meets a hydrophobic surface. This has been a long-standing point of debate, with prior claims ranging from complete drying at the surface, or depletion layers several nm think, to no depletion at all. The issue has also been complicated by the possible presence of nanobubbles of dissolved gases. Granick and co. now report evidence from X-ray reflectivity for a depletion of more than 60% of the bulk density over a layer thickness of 2-4 angstroms. That’s a distance of the order of the diameter of a water molecule, so at least there is no new length scale mysteriously entering the picture. But will this be the last word?

In Biophys. J. (see here), Florin Despa and Stephen Berry have taken on another contentious issue – the origin of the long-range hydrophobic attraction. They say the interaction is electrostatic, caused by induced dipoles on the surfaces of hydrophobic solutes. I’ve only seen the abstract of this paper, but hope to take a good look soon.

Joe Zaccai at the ILL and colleagues have a deeply interesting, not to say perplexing, paper in PNAS in which they report very slow translational diffusion coefficients for water inside the cells of the halophilic archaea Haloarcula marismortui from the Dead Sea. The idea that cell water has different dynamics from bulk water goes back a long way, at least to NMR work by Ray Damadian in the 1970s. But it’s never been shown definitively. Zaccai and colleagues use inelastic neutron scattering to measure relaxation times of the water in situ in the archaeal cells, and say that 75% of it diffuses no less than 250 times slower than bulk water. That’s too slow to be explained away as the dynamics of macromolecular hydration shells. Nothing of the sort is seen in E. coli. So what’s going on? I can’t figure out quite what their hypothesis is – it makes sense to link it to the high salt concentrations (specifically potassium), but beyond that the authors just talk about ‘structured water around K+ ions’ in the presence of proteins, similar to that seen in potassium channels. Hmm… clearly potassium doesn’t ‘structure’ water so significantly in simple salt solutions, so what’s the idea? I’m hoping Joe can enlighten me.

What this blog is about

Water in the living cell is a bit like quantum mechanics: anyone who says they understand it probably doesn’t. At least, for everyone who would make such a claim, there will be someone who disagrees with them. The simplistic descriptions of liquid water and macromolecular hydration given in biochemical textbooks (these things might not be considered at all in cell biology textbooks) are often caricatures that bear little relation to the current state of knowledge (or rather, of ignorance).

And yet the question of what cell water is like matters hugely. It determines how macromolecules acquire their biochemically active conformations, how they interact with one another, how they move about, how substances are partitioned in the cell, how signals are transmitted, and much more. To neglect water in a description of biomolecular function is rather like neglecting it in a description of swimming.

Here are just a few of the outstanding questions:
- what is the structure of bulk liquid water (yes, truly, this is still controversial)?
- is water in the cytoplasm bulk-like or not?
- what is the effect of nanoscale confinement of water?
- how does water behave close to hydrophobic and hydrophilic surfaces
- what is the structure of water around proteins, and around nucleic acids?
- …and what is the dynamics?
- how does the hydrophobic force work?
- is there a long-ranged hydrophobic force, and if so, why?
- what accounts for the Hofmeister series, the power of ions to precipitate proteins?
- is water necessarily unqiue as a solvent for life?

I have written a long review article that attempts to bring together what we know about these things – it is currently under review for Chemical Reviews, but a preprint can be downloaded from my web site (or can be very soon). But I’m trying to hit a moving target. Almost every week there is some relevant new disclosure about water and its roles in molecular biology. This blog will serve as a repository for such contributions. My hope is that it will be used by anyone who wishes, to keep a record of new results, what they might mean, and why they are important. And I would like this to be a forum for debate – because there is plenty to debate. Anyone who wishes to be among the posting group for this site should contact me for the login details. I’ll attempt to keep some kind of order, but not to be responsible for the site’s content.