Highlights
Primary processes in protein folding
For more than a decade, short peptides that fold to form helices or hairpins have been studied as miniature versions of protein folding. Also these systems are small enough for us to run atomistic MD in explicit water, facilitating comparison with experiment. Using master equation / Markov state models (MSMs) I have determined the rate of formation of a helix nucleus in the W1H5 peptide, the fastest protein folding related process (ref 10). We also have solved a problem in the estimate of the rates from MD simulations for a hairpin forming peptide (see ref 13).
The origin of "internal friction"
The contribution of internal friction to protein dynamics was first proposed over 20 years ago. Usually thought to slow down the overall rate of the folding due to frustrated interactions, a molecular description of internal friction has been lacking. Using atomistic MD and the MSM methodology I have recently found that in fact the experimental signature usually interpreted as internal friction can be explained as a result of the crossing of sharp torsional barriers (i.e. from the alpha to beta region in the Ramachandran map, see refs 16, 19, 25). This is sufficient to explain experimental results for helix forming peptides and mini-proteins, although other sources internal friction may exist in larger systems.
Gas diffusion in enzymes
The entry of ligands into enzyme active sites is oftentimes essential for their transport or for catalysis to take place. Using molecular dynamics (MD) simulations and Markov state models (MSMs) we have calculated the rate of entry of gases into myoglobin (ref. 20) and hydrogenases (16, 27). More importantly our approach informs on possible ways of modulating the entry of ligands. In the case of FeFe hydrogenases this is particularly important, as in this may prevent the oxidation of the active site. This research avenue can result in new biotechnological approaches for producing clean energy using hydrogenases.
Intrinsically disordered proteins
In the last few years we have realized of the importance of intrinsically disordered proteins, which fold to well defined 3D structures only upon binding to their targets. I use coarse grained structure based (Go) models to study the mechanism of folding/binding for this type of systems. However given the especial properties of IDPs we need to adapt the models so that they reproduce the disorder and the dimensions in the unbound state. Recently we have shown how non-native interactions and electrostatic forces may be key in understanding the ultrafast binding of some of these IDPs to their targets (ref 11). In a collaborative project with Prof Jane Clarke's group we have shown how the results from coarse-grained simulation models can be integrated with experimental observables to provide a global view of the coupled folding and binding process of IDPs (ref 17).
Statistical mechanics models to study experiments
Simple models can be extremely helpful in our understanding of experimental data. My work in one such models allowed fitting the rates of folding and unfolding in a carefully curated database of two-state proteins. We were able to extract the variations in stabilization energy and barrier effects to folding (ref 8). This analysis showed that regardless of having stabilities ranging 5-50 kJ/mol and rates spanning 6 orders of magnitude proteins seemed to be surprisingly homogeneus. We later implemented an algorithm,PREFUR, to predict protein folding and unfolding rates from the information of size and structural class (ref 9).
Coarse grained models for protein folding
During my PhD in Antonio Rey's Group, in the Department of Chemical Physics I of UCM, I designed genetic algorithms to study knowledge-based potentials for protein folding simulations. These potentials are widely utilized in coarse-grained simulations of protein folding and binding and structure prediction. We first carefully designed the optimization algorithm and later studied models for different types of interaction (mainly hydrogen bonds and hydrophobic contacts) that we took from the literature. Our results were useful to understand how detailed modelling can be critical to reach the native conformation instead of one of its topo-isomers (refs 1-6).