Thursday, November 22, 2012

Softening up proteins

It’s commonly thought that hydration of hydrophobic solutes happens with the positioning of water-water hydrogen bonds tangential to the solute surface. That notion seems to be confirmed in a neutron-scattering study by Jeremy Smith at Oak Ridge and colleagues (I. Daidone et al., Biophys. J. 103, 1518-1524; 2012 – paper here). They show that this information can be extracted from the low-Q spectra of dipeptides in water, in conjunction with MD simulations. Moreover, the degree of connectivity of the hydration water clusters is greater than that in the bulk, and increases with increasing solute hydrophobicity.

In a second paper Jeremy, with Alexei Sokolov and coworkers, use neutron scattering to look at the dynamics of (fully deuterated) GFP and its hydration shell (J. D. Nickels et al., Biophys. J. 103, 1566-1575; 2012 – paper here). The interaction between solvent and solute dynamics is complex. At lower temperatures (below about 200 K) the hydration water suppresses protein motions, albeit itself exhibiting a dynamical change around 180-190 K that seems to correspond to the glass transition of the protein. This suppression of protein motion seems to be stronger for GFP than for many other proteins, owing to its barrel-like shape. But at higher temperatures the hydration water has the opposite effect, enhancing protein dynamics. Here, however, at least on the ps-ns timescale probed, the hydration waters behave diffusively while the protein motions are confined to within 3 ångstroms or so.

In something of a companion piece, Jeremy and his coworkers again use neutron scattering and MD to probe protein dynamics, this time looking at the atomic motions in lysozyme at different levels of hydration on timescales from ps to ns (L. Hong et al., Phys. Rev. Lett. 108, 238102; 2012 – paper here). They ask whether the effect of hydration on protein motions is local to the interface or propagates into the protein core. Previously, hydration has tended to be considered in terms of its effect on conformational dynamics: how it changes the ability of protein atoms to jump between potential wells. But the researchers say that, on these timescales, the principal effect of hydration seems to be to enlarge the volume of the well within which protein atoms can diffuse locally. Nevertheless, because of strong coupling between such wells – I guess this means cooperativity of atomic motions – the effect can propagate into the protein interior and soften the whole molecule. Consistent with the Biophys. J. paper above, they observe the same kind of behaviour in GFP.

A very interesting new take on hydration is offered by Franci Merzel of the National Institute of Chemistry in Ljubljana and colleagues (Jeremy Smith is among these too – busy man) (A. Godec et al., Phys. Rev. Lett. 107, 267801; 2012 – paper here). Their MC simulations suggest that as a solute becomes more hydrophilic/polar, the hydration water tends to fractionate into a more highly ordered and less highly ordered component. This allows the former to rearrange its hydrogen-bonded network so that the bond angles are optimal, while the latter is released from the network. So while the hydration water of hydrophobic solutes is less orientationally flexible, it is NOT more tetrahedrally ordered, as is commonly thought; rather, hydration water of hydrophilic solutes is in fact more tetrahedral, on average, but there is less of it. In any event, say the authors, the picture is considerably more complex than any notions of ‘icebergs’ around hydrophobes.

Chang Sun at Nanyang Technological University in Singapore and his coworkers have been developing interesting ideas about how some of water’s properties can be rationalized by the differences between the covalent and hydrogen-bonded portions of the O-H-O links in the water network, in which Coulomb repulsion between the unevenly bound electron pairs can produce stiffening of the shorter covalent bond and softening and lengthening of the hydrogen bond. This idea is developed in a preprint (C. Sun et al., in which Sun and colleagues say that this effect can account for the stiffening of phonon modes in molecular clusters of water molecules, surface skins and ultrathin water films. The latter have been found to exhibit ice-like (or glue-like) properties and to be somewhat hydrophobic, as Haiping Fang and coworkers have reported (C. Wang et al., Phys. Rev. Lett. 103, 137801 (2009)).

Mihail Barboiu at the Institute Europeen des Membranes in Montpellier has reviewed some of the ways to make artificial water channels for bilayer membranes (Angew. Chem. Int. Ed. 51, 11674; 2012 – paper here), which are of course of great potential interest for water purification. This piece brought to my attention Mihail’s earlier work on channels made by the self-assembly of quartets of ureido imidazole molecules, which will stack into channels held together by hydrogen bonding to internal water wires and allow water and protons to permeate lipid bilayers (Y. Le Duc et al., Angew. Chem. Int. Ed. 50, 11366; 2011 – paper here).

The prevailing view of the hydration of small hydrophobes now seems to be that they are accommodated without a significant perturbation of water’s hydrogen-bonded network. This idea is supported by ab initio MD calculations of Fabio Sterpone at the Université Paris Diderot and his colleagues, who look at the hydration of methane (M. Montagna et al., J. Phys. Chem. B 116, 11695; 2012 – paper here). They say that the water molecules form a clathrate-like structure around the solute, although by this they do not imply anything static: the hydration shell is rather dynamic, and fails to produce any features in the IR spectrum significantly different from that of the bulk, although there are small signatures in the OH stretching and librational bands.

Steven Boxer at Stanford and colleagues have used time-resolved vibrational spectroscopy to examine the dynamics of water molecules in the active site of a ketosteroid isomerase (S. K. Jha et al., J. Phys. Chem. B 116, 11414; 2012 – paper here). They find that waters in the active site are much more rigid during the catalytic cycle than those in bulk, and suppose that this is by design rather than an epiphenomenon: the rigidified waters may help to preserve the particular electrostatic environment required for catalysis. In other words, catalysis here doesn’t provoke water ordering but is preceded and enabled by it.

Here’s an old(ish) paper of which I was recently made aware, in which water-assisted binding of a protein and ligand is dissected in detail by Alfonso García-Sosa at the University of Tartu in Estonia and Ricardo Mancera at the Curtin University of Technology in Australia (Mol. Informatics 29, 589; 2010 – paper here). Using MD simulations, they calculate the free-energy cost of removing the tightly bound water molecule from the SH3 domain of Ableson tyrosine kinase with several bound ligands (this molecule mediates protein-protein interactions), by adding functional groups to the peptide substrate that displace the water. They find that this process is unfavourable in all cases except for three substituents: hydroxyl, ethyl and formamide. The key point is in the methodology: the thermodynamics of such substitutions can be examined, and the role of the water assessed, with implications for drug design.

And here’s another paper that I missed last year. Lignin is one of the recalcitrant components in the conversion of biomass to biofuels, and pretreatments are needed in order to break down the lignin barrier that slows cellulose hydrolysis. Lignin is a hydrophobic polymer which is dense, aggregated and glassy in water at room temperature but softens and becomes more extended at higher temperatures. Jeremy Smith and colleagues have investigated the hydration changes that accompany this transition using MD simulations (L. Petridis et al., JACS 133, 20277; 2011 – paper here). They find that at high temperature (above about 420 K) the polymers form fractal crumpled globules, while the low-temperature collapse (around 300 K) is driven by density fluctuations of water removed from the hydration shell: the fluctuations in the bulk are slightly greater than those in the hydration shell, producing an entropic driving force for the collapse. Ultimately these insights might be of use for improving the efficiency of the pretreatment process.

How many water molecules does it take to make ice? Around 275 ± 25, according to Thomas Zeuch of the University of Göttingen and colleagues. Their IR spectra of size-selected water clusters show the first signs of crystallization, while the clear ice-like spectral feature of the 3200 cm-1 OH stretch is evident by clusters of around 475 molecules (C. C. Pradzynski et al., Science 337, 1529; 2012 – paper here). The researchers think it should be possible to follow the evolution of the cluster at least to sizes of around 1000 molecules, where one might begin to see bulk-like behaviour.

Considerably more to come. I have a long plane trip coming up, so you might get it soon.