Wednesday, March 26, 2008

Chemistry vs geometry

Pablo Debenedetti and colleagues have carried out precisely the kind of study that is needed to tease apart the various factors that might be at play in hydrophobic association of proteins (N. Giovambattista et al., PNAS 105, 2274-2279; 2008 – paper here). One can anticipate that the potential for effects such as abrupt drying transitions as the two surfaces approach is affected both by surface chemistry – by the distribution of hydrophilic and hydrophobic groups – and by geometry. Certainly, both have been implicated as playing a role in how real proteins behave, as for example in the Berne group’s study of protein associations for BphC and melittin (Liu et al., Nature 437, 159-162; 2005; Zhou et al., Science 305, 1605-1609; 2004). Melittin monomers enclose a tubelike space, for example, whereas BphC is slablike. Moreover, melittin, like many proteins, has a rough surface with concavities. To decouple the effects, Pablo and colleagues have simulated the association of a mutated melittin dimer in which the distribution of hydrophobic and hydrophilic groups is retained but the surface is artificially flattened. The results suggest that the flattened melittin behaves as an intermediate case between ideal, flat hydrophobic and hydrophilic surfaces, and that the drying seen in the case of ‘real’ melittin happens only at very small separations (about one intervening water layer) for the flattened case, being localized to a central region where an apolar residue resides. It can be suppressed by replacing that residue. In other words, drying seen for ideal hydrophobic plates is probably stronger than it is for real proteins, where it is likely to be highly sensitive to small variations in surface chemistry.

Michael Geisler and colleagues at the Technical University of Munich have looked at Hofmeister effects in the adhesion of spider silk proteins to a solid surface, using single-molecule AFM force spectroscopy (Langmuir 24, 1350-1355; 2008 – paper here). They find that the desorption forces follow the Hofmeister series, but can’t yet develop a clear interpretation of what is going on. The hydrophobicity of the silk protein also plays a part: ions that stabilize adhesion do so less when the protein is less hydrophobic, ‘indicating that hydrophobic and Hofmeister effects are closely related’ – but how?

Dusan Bratko and Alenka Luzar have attempted to unravel the much vexed question of how dissolved gases affect the hydrophobic interaction (Langmuir 24, 1247-1253; 2008 – paper here). They have used simulations to look at how various gases influence water structure close to a hydrophobic surface, and solvation forces between two such surfaces. They say that although there does seem to be accumulation of dissolved gas at the interface, it doesn’t have a big effect either on putative water depletion or on solvation forces – something that several experiments seem to bear out. One of the nice aspects of this work is that it enables the authors to make a link between capillary evaporation of pure water induced by hydrophobic confinement and evaporation nucleated by an excess of dissolved gas at the interface – two things that are sometimes not so clearly distinguished. But the simulations can’t follow the possible formation of nanobubbles and the effect this might have on the hydrophobic interaction.

Finally, I have good reason to think that my recent Essay in Nature on water (here) might be seen by some as an endorsement of the ‘new view’ of water structure championed by Anders Nilsson and Lars Pettersson. It’s not, as I think is clear if you read carefully. I merely point out that, first, it is remarkable that such fundamental disagreements about water are still occurring (I know, of course, that Anders and Lars’ idea has been strongly criticized), and secondly, that the implications are rather more far-reaching than might be naively supposed. I must apologize, incidentally, for giving the impression that the experimental work on which their new model is based was done by Lars at Stockholm, rather than by Anders at Stanford.

Tuesday, March 11, 2008

Antifreeze: what the sugar does

First, an historical note: I recently discovered that this very nice paper by Charles Tanford on the history of the hydrophobic effect is available online. Much of this stuff appears in his books The Hydrophobic Effect (Wiley, 1980) and Nature’s Robots (OUP, 2001), but it’s a very nice summary of it.

Joe Zaccai has sent me a preprint of a paper just accepted by EMBO Reports that uses neutron scattering to look at water dynamics in vivo in E. coli. It shows that these dynamics are ‘normal’ and bulk-like, contrary to suggestions that water is ‘tamed’ in the cytoplasm. Bertil Halle and his coworkers have a paper in press with PNAS that reports precisely the same conclusion based on NMR data. So together, these papers ought to bury one more water myth.

There’s an interesting study here (JACS 130, 2928-2929; 2008) by Robert Ben and colleagues at Ottawa of the effect of sugar hydration on the antifreeze behaviour of glycoproteins. By substituting various sugars on antifreeze glycoprotein analogues, they find that the sugar conformation and thus hydration is important for inhibition of ice recrystallization. Here’s the punchline: “our data indicate that the compatibility of a hexose with the three-dimensional hydrogen-bonded network of water is inversely proportional to recrystallization-inhibition activity” – a finding they associate with the consequent free-energy change of transferring a water molecule to the ice lattice.

Also in JACS (130, 3120-3126; paper here), Greg Voth and his coworkers Feng Wang and Sergei Izvekov report ab initio MD simulations showing that hydronium ions form unusual cation pairs in concentrated aqueous HCl, stabilized by delocalization of the excess charge of the hydrated proton. This is consistent with Greg’s earlier work showing that hydronium seems to display amphiphilic behaviour – one can regard this as a kind of amphiphilic clustering.

Water does interesting stuff around benzene, which is hydrophobic around the edges but can form hydrogen bonds via the pi orbitals over the ring faces. So how does this translate to C60? Dahlia Weiss, Tanya Raschke and Michael Levitt have addressed that question using MD simulations in a paper here (J. Phys. Chem. B 112, 2981-2990; 2008). They say that the waters in the first hydration shell become more oriented, and have an increased number of hydrogen-bonding contacts, but that hydrogen bonding is disrupted between the first and second hydration shells. In general, the hydration shell is dense and ‘well-structured’ – I’d guess consistent, at a glance, with the kinds of orientational ordering described by Jan Engberts and W. Blokzijl in their 1993 article on hydrophobicity (Angew. Chem. Int. Ed. 32, 1545-1579), as opposed to the old notion of a hydrophobic ‘iceberg’. In this regard, the authors say that “C60 behaves as a large hydrophobic solute.”

Tuesday, March 4, 2008

Solvent not included

I talked a little bit in my review article about the difficulty of understanding and/or predicting the energetics of water expulsion from the active site of a protein when it binds its ligand, and the potential value of being able to do so for drug design. Richard Friesner, Bruce Berne and their colleagues have now reported a computational model which they say allows them to make this calculation in an efficient manner (JACS 130, 2817-2831; 2008 – paper here). They test it out on ligand binding in factor Xa, a potential anti-thrombosis drug target. They imply that this approach, considering a molecularly resolved rather than a continuum solvent, is needed for accurate prediction of the significant contributions that such displacements can make to the binding energies.

There’s more on this issue by Anthony Davis and colleagues at Bristol (E. Klein et al., Angew. Chem. Int. Ed. 10.1002/anie.200704733; paper here), who look at the role of displaced water in binding of carbohydrates by synthetic receptors (which they argue to be good analogues of carbohydrate-binding proteins). They say that hydrophobic interactions – which I think means here the expulsion of water from hydrophobic-hydrophobic contacts – play a significant role in binding.

Also somewhat related is a paper by Ken Raymond and colleagues at Berkeley, who have probed the influence of solvation on supramolecular encapsulation processes (Leung et al., JACS 130, 2798-2805; 2008 – paper here). They look at the subtle compensation effects between enthalpic and entropic contributions to encapsulation free energy: in water, desolvation releases water molecules to form more hydrogen bonds in the bulk, which is enthalpically favourable but entropically not. They conclude that the primary driving force of encapsulation, in water and other polar protic solvents, is the rearrangement of the hydrogen-bonding network in the solvent.

A recent paper on segregation of hydronium ions at air-water (and by extension, hydrophobic) surfaces, claiming that these have elevated pH (Buch et al., PNAS 104, 7342; 2007) stirred up some controversy. Some others claim that in fact such water surfaces are enriched with hydroxide, not hydronium. Konstantin Kudin and Roberto Car have now looked at both cases, using ab initio molecular dynamics simulations (JACS doi:10.1021/ja077205t; paper here). They say that both hydroxide and hydronium act as amphiphiles at these interfaces, with one end even more hydrophilic than water and the other essentially hydrophobic. The effect is larger for hydroxide, which implies that these ions accumulate more readily at the surface, giving it a negative charge. That’s indeed what seems to be observed in practice, as James Beattie pointed out to me when I wrote about the Buch et al. paper. But the results also seem consistent with Greg Voth’s predictions that hydronium acts as an amphiphile (e.g M. K. Petersen et al., J. Phys. Chem. B 108, 14804; 2004).

It’s very heartening to see in such a prominent place (Science 319,1197-1198; 2008) Douglas Tobias and John Hemminger’s head-on challenge to the notion of generalized structure-making and structure-breaking of water as an explanation for Hofmeister (specific-ion) effects. Tobias and Hemminger’s piece is a perspective on two recent papers mentioned earlier on this blog (Smith et al., JACS 129, 13847; 2007 and Mancinelli et al., J.Phys. Chem. B 109, 13570; 2007). I won’t outline those papers again, but simply point out that they both, from different perspectives, highlighted shortcomings of the traditional picture. T&H point out that recent work on specific ion absorption or depletion at surfaces by Jungwirth, Saykally, Pegram and Record, Berne and others are beginning to point to a rather more complicated picture of electrolyte effects that has nothingto do with modifications of the bulk structure of water.

Julio Fernandez and colleagues (first author Lorna Dougan at Columbia) argue here (PNAS 105, 3185-3190; 2008) that the mechanical functions of proteins, which involve conformational changes, are highly sensitive to the solvent because of solvent bridges between parts of the polypeptide chain. This is consistent with earlier work by Jose Onuchic and collaborators on protein folding (e.g. PNAS 99, 685; 2002). Dougan et al. use single-molecule force spectroscopy on a repeating-sequence domain of titin, a component of muscle tissue, to study how stretching it out changes when the solvent is switched to deuterium oxide or glycerol. The results are consistent with simulations in which the solvent molecules bridge adjacent beta-strands in the unfolding transition state. For water, several bridges of one molecule each seem to be involved; for glycerol, with a longer hydrogen-bonding ‘reach’, this transition state corresponds to a wider strand separation. Here the unfolding is an intrinsic part of the protein’s biological role, but presumably the same considerations would be expected to apply to denaturation of globular proteins too.

Fengshou Zhang at the Beijing Normal University has sent me a preprint of his paper now published in Phys. Rev. Lett. 100, 088104 (2008), in which he and his colleagues report MD simulations of conformational changes in DNA brought about by changes in solvent. Specifically, they consider ‘modified water’ with a tetrahedral structure but with variable dipole moment, ranging from ‘over-polarized’ (relative towater) to under-polarized. In the former case the double-helical B form is maintained but becomes stiffer (smaller fluctuations); as polarity decreases, the A form becomes increasingly favoured. The authors relate this to changes in phosphate screening, which is effected mainly by solvent molecules in more polar solvents and by counterions in less polar ones. I’m interested that Ruth Lynden-Bell is thanked for discussions; Ruth has pioneered this notion of a kind of counterfactual exploration of water’s role in structural biology, as a way of investigating the notion of ‘fine-tuning’ of water in biology (or should it be, of biology in water?).

There’s more good stuff to come, but that’s enough for now. It is very nice to be getting sent these things...