Wednesday, January 24, 2018

Hydration as a design element in biomaterials

Given what is known about water’s active role in the structure, dynamics and function of biomolecules, can it be used as a design element in synthetic molecular and nanoscale structures derived from them, which might have more amenable levels of complexity? Sam Stupp at Northwestern and colleagues ask this question in relation to the amphiphilic peptide-based nano-architectures they have been developing. They have used Overhauser dynamic nuclear polarization relaxometry to characterize the hydration of such a peptide nanofibre [J. H. Ortony et al., JACS 139, 8915; 2017 – paper here]. They find a wide range of water dyamics, from rapid translational diffusion within the hydrophobic interior to near-immobilization of water on the hydrophilic surface. Water further from the core and close to the inner peptide domain is slowed relative to the bulk, and more so for fibres in the gelled than in the solution state. The same applies to water close to the (charged) nanofibre surface, which may be slowed to a degree comparable to that of water confined within protein cavities. MD simulations echo these findings. Sam and colleagues suggest that the surface water might play an active role in gelation and interactions with biomolecules, and in consequence that “water dynamics not only play an important role in the function of nanostructured biomaterials, but… may also be tunable in soft matter at molecular length scales to optimize performance for a variety of biomedical applications.”

In a similar spirit, Geraldine Richmond at Oregon and coworkers use vibrational sum frequency spectroscopy to look at water orientational ordering close to the surface of nanoemulsion aggregates: micelles and reverse micelles [J. K. Hensel et al., PNAS 114, 13351; 2017 – paper here]. Water seems to be electrostatically ordered by the surfactant head groups and their counterions at the interface.


Proposed orientation of water at surfactant (AOT)-counterion surfaces, both for the curved surfaces of micelles and the planar surfaces of lamellae.

How exactly does one quantify water order or disorder? Various measures have been suggested; Fabio Sterpone of the Université Paris Diderot and colleagues suggest that it might be sought in the connectivity of the hydrogen-bond network (O. Rahaman et al., JPCB 121, 6792; 2017 – paper here). They have looked at this measure in simulations of the protein (here lysozyme) dynamical transition at around 240 K. There is a fairly abrupt change in the connectivity at the transition, in that the degree of percolation is constant below about 220-240 K but decreases with temperature above it. A closer analysis shows that at 240 K the hydration water network begins to sample a larger number of configuration states, corresponding to an abrupt increase in configurational entropy. Fabio and colleagues interpret these findings in terms of the hydration water acting as an ‘entropic reservoir’ that, above the dynamical transition, enables the protein to undergo configurational changes.


Probability distribution of hydration-water percolation propensity for lysozyme as a function of temperature.

Not strictly water in biology, but Daniel Munoz-Santiburcio and Dominik Marx in Bochum describe ab initio simulations of proton transport in alkaline water films confined between the layers of the mineral mackinawite [Nat. Commun. 7, 12625; 2016 – paper here]. They find that orientation and coordination of hydroxide can alter the hydrogen-bond network in ways that make transport of protons and mobility of negative-charge defects sensitive to the width of the confining space: different structures in the layered water films support qualitatively different transport mechanisms. Daniel and Dominik conclude that “it should be possible to rationally design different nanostructures that will allow for different charge transport rates of confined alkaline aqueous solutions.” Meanwhile, in another paper they show that water within mackinawite layers has a boosted tendency to self-dissociate (Phys. Rev. Lett. 119, 056002; 2017 – paper here).

And in a third paper they consider the effect of water confinement (between mineral layers) on a whole set of chemical reactions, particularly those involved in peptide formation – which may serve as a model of prebiotic chemistry at nanoconfined mineral surfaces (Chem. Sci. 8, 3444; 2017 – paper here). Under these conditions both the energetics and the mechanisms of the reactions can be significantly altered relative to the bulk.

X-ray free electron lasers are already transforming structural biology with their ability to provide diffraction data on very small samples in some cases down to the single-molecule scale, at high time resolution for studying dynamical processes. Jessica Thomaston and coworkers have used the Japanese Spring-8 XFEL source to obtain a new structure of the influenza M2 proton channel under ambient conditions (J. L. Thomaston et al., PNAS 114, 13357; 2017 – paper here). They find that an ordered network of hydrogen-bonded water molecules spans the pore at pH5.5, providing a proton-conduction pathway and stabilizing the protonated His37 residue present in the intermediate open state of the channel. These ordered waters decrease in number as pH increases and the open state becomes less stable.


The room-temperature XFEL structures of the M2 channel at different pH. Red spheres show waters with full occupancy, light and dark blue with half-occupancy.

The idea has been around for some time that water might play an important role in protein (mis)folding into amyloid fibrils associated with neurodegenerative diseases. Mei Hong at MIT and colleagues have now used solid-state NMR to characterize the reservoirs of water within and around fibrils of the Alzheimer’s β-amyloid peptide Aβ-40 (T. Wang et al., JACS 139, 6242; 2017 – paper here). They find five distinct pools, ranging from surrounding bulk-like matrix water to mobile interfibrillar water channels and relatively immobile peptide-bound reservoirs. Exchange of water happens between the two dynamic pools on very long (second) timescales. The balance between the various pools is shifted in some mutant fibrils.


The five water pools in wild-type Aβ-40 amyloid fibrils.

Much attention has been given to the roles of hydration water in the binding of ligands by proteins, but less to the idea that solvation is also important for unbinding. Paolo Carloni at Jülich and colleagues have simulated the unbinding kinetics of an anti-inflammatory agent (a urea derivative) that binds to p38 MAP kinase (R. Casanovas et al., JACS 139, 4780; 2017 – paper here). They find that a rotation of the urea group during unbinding creates a more –solvent-exposed state in which the hydrophobic interactions between the t-butyl substituent of the ligand and a hydrophobic part of the cavity are weakened. Water molecules that enter into the cavity play an essential role in this step, forming hydrogen bonds to the urea group and mediating those with residues in the cavity.

An important role for water-mediated contacts is identified in the DNA binding of transcription factors of the ETS family by Gregory Poon and colleagues at Georgia State University (S. Xhani et al., JPCB 121, 2748; 2017 – paper here). One in particular, denoted PU.1, shows a strong osmotic sensitivity in its binding, which the researchers attribute to the water-mediated binding between a tyrosine residue and DNA. Mutation of this residue removes this sensitivity. In the homologue Ets-1, which lacks this sensitivity, it seems that the same Tyr residue, while apparently engaging in the same water-mediated contact in the bound state, actually plays a different role in binding, by influencing the local dynamics of the free protein. I guess this underscores once more how hard it is to identify general rules of thumb for how hydration structures affect biomolecular function and molecular recognition.



The same and not the same: these two water-mediated contacts between a Tyr residue and DNA in the homologous transcription factors PU.1 (top) and Ets-1 (bottom) actually play different roles in the molecular-recognition process.

Cytochtome c oxidase (CcO) pumps protons across membranes using the energy of dioxygen reduction. The mechanism seems to involve a change in hydration – a switch between ‘wet’ and ‘dry’ configurations – in an internal cavity that is connected to the proton’s exit channel. Qiang Cui and colleagues at the University of Wisconsin have studied that process using MD simulations (C. Y. Son et al., PNAS 114, E8830; 2017 – paper here). They show how the protonation state of one residue acts as a switch between the dry and wet states, and that in the wet state the pKa of a Glu residue is lowered to facilitate proton transfer. Here, then, is a functional role for the dry-wet switching of the cavity.

What about the proton channel itself in CcO? It’s not in fact yet clear what route it takes – three possible channels (labelled D, K and H) have been proposed. Vivek Sharma at the University of Helsinki and colleagues use simulations to show that the H channel alone doesn’t seem able to do the job, unless a buried histidine residue is protonated (which seems unlikely); otherwise there’s a gap that the transient water networks in the channel can’t span (V. Sharma et al., PNAS 114, E10339; 2017 – paper here).

Here’s a rather old paper that I only noticed recently. Molecular recognition in proteins is still often discussed in biochemistry textbooks in terms of the induced-fit model. But there’s more to it. An alternative (if not necessarily mutually exclusive) picture has been developing recently in which binding occurs preferentially for particular, perhaps weakly populated conformations in an ensemble of protein states, stabilizing that conformation. Oliver Lange at the MPI Göttingen and colleagues used NMR to characterize the conformational states of ubiquitin in solution, and their results offer support for this notion of conformational selection (O. F. Lange et al., Science 320, 1471; 2008 – paper here). This work strengthens the idea that protein dynamics – which of course are highly influenced by hydration and by solvent fluctuations – have a prominent role in their functionality, an issue I discussed briefly in a recent column in Chemistry World (here https://www.chemistryworld.com/opinion/snapshots-of-lifes-dancers/3008393.article). The Lange et al. work was nicely summarized in a Perspective article by David Boehr and Peter Wright at Scripps (Science 320, 1429; 2008 – paper here).

In an intriguing simulation study, Sridip Parui and Biman Jana at the Indian Association for the Cultivation of Science in Kolkata have looked at the hydrophobic interactions between two hydrocarbon molecules, and also two rod-like model hydrophobes, at low temperatures (240 K) (JPCB 121, 7016; 2017 – paper here). They find that there is a second solvent-separated minimum for both systems, roughly 1 nm apart. This corresponds to weaker hydrophobic interactions between the solutes, due to stronger water-water interactions. The authors suggest that such a state could play a role in cold denaturation of proteins.

The effect of osmolytes on the denaturation of a model protein (stem bromelain) is considered in detail by Pannuru Venkatesu of the University of Delhi and coworkers (A. Rani et al., JPBC 121, 6456; 2017 – paper here). They use a whole battery of experimental techniques (fluorescence, UV-Vis and circular dichroism spectroscopy and dynamic light scattering), along with MD simulations, to study the stabilizing effect of a series of osmolytes (proline, betaine, arginine, sarcosine) against thermal denaturation. In all cases the effects seem to be caused by direct interactions. Betaine, sarcosine and arginine all interact with the protein via hydrogen-bonding. Proline, meanwhile, which confers the greatest stabilizing effect, binds in the protein’s active site.

Michael Feig at Michigan State and others have published a nice review article on biomeolecular crowding (M. Geig et al., JPCB 121, 8009; 2017 – paper here). This considers the effects of crowding on molecular diffusion, conformation and dynamics, but also on the solvent, which – as the authors say – can have altered structure, dynamics and dielectric response in typical crowded geometries.


A typical crowded environment in the cell.

Coralie Pasquier at the University of Lund and coworkers have studied the effect of electrolytes with multivalent (trivalent yttrium) ions on protein-protein interactions, using MD and MC simulations (C. Pasquier et al., JPCB 121, 3000; 2017 – paper here). The Y ions bind to the surface of the proteins (human serum albumin), but the consequences are complex. In a coarse-grained model, increasing Y3+ concentration could increase protein repulsion at low ionic strength but increase it at high ionic strength. The first situation is due to double-layer effects, the second is a Coulombic repulsion due to high charging of the protein surface. These interactions are water-mediated, and screened out by addition of NaCl, resulting in protein attraction. At intermediate concentrations of YCl3 there is also a net attraction between the proteins, due to ion-ion correlations.

Takeshi Yamada of CROSS in Naka, Japan, and coworkers have used quasi-elastic neutron scattering to look at the dynamics of water sandwiched between phospholipid bilayers (T. Yamada et al., JPCB 121, 8322; 2017 – paper here). They see three distinct populations of water molecules: free and almost bulk-like, loosely bound and tightly bound to the phospholipid head groups.

Concentrated ionic solutions aren’t so commonly encountered in biology, but they are important in some technologies, such as rechargeable aqueous batteries. Water is known to have slowed rotation in such solutions, and Wei Zhuang of the Fujian Institute of Research on the Structure of Matter and coworkers suggest why (Q. Zhang et al., PNAS 114, 1123; 2017 – paper here). Their simulations suggest that the key contribution comes from a coupling of the slow, collective component of rotation with ion clusters, rather than from faster single-molecule motions. They say that there are similarities with water rotations near large biomolecules.

Do we finally have an accurate, predictive, tractable ab initio model of water? Mohan Chen at Temple University and collaborators think so (M. Chen et al., PNAS 114, 10846; 2017 – paper here). They call their model the strongly constrained and appropriately normed (SCAN) density functional, and say that it captures many of the structural, dynamic and electronic properties of liquid water as well as the density difference with ice Ih.

Talking of such things, Tomotaka Oroguchi and Masayoshi Nakasako of Keio University say that to accurately simulate the directionality of hydrogen bonds from donor atoms in hydrophilic amino acid residues, one needs to include in the force fields off-atom charge sites that mimic the lone-pair electrons (Sci. Rep. 7, 15859; 2017 – paper here).