Revealing the global map of protein folding space by large-scale simulations

Abstract

The full characterization of protein folding is a remarkable long-standing challenge both for experiment and simulation. Working towards a complete understanding of this process, one needs to cover the full diversity of existing folds and identify the general principles driving the process. Here, we want to understand and quantify the diversity in folding routes for a large and representative set of protein topologies covering the full range from all alpha helical topologies towards beta barrels guided by the key question: Does the majority of the observed routes contribute to the folding process or only a particular route? We identified a set of two-state folders among non-homologous proteins with a sequence length of 40-120 residues. For each of these proteins, we ran native-structure based simulations both with homogeneous and heterogeneous contact potentials. For each protein, we simulated dozens of folding transitions in continuous uninterrupted simulations and constructed a large database of kinetic parameters. We investigate folding routes by tracking the formation of tertiary structure interfaces and discuss whether a single specific route exists for a topology or if all routes are equiprobable. These results permit us to characterize the complete folding space for small proteins in terms of folding barrier ΔG(‡), number of routes, and the route specificity RT.

The Sensitivity of Computational Protein Folding to Contact Map Perturbations: The Case of Ubiquitin Folding and Function

“…The effect of contact strength variation on folding routes has been examined both theoretically and computationally in detail. 24,25,72,73 One result that has emerged from these studies is that, given the choice of 20 amino acids, naturally occurring contact strength variation is unlikely to affect the folding routes of topologically complex proteins such as β and α/β proteins. 24,25 Nevertheless, functional constraints can determine the choice of not just isolated but groups of interacting amino acids, and in such cases, energetic heterogeneity can affect the folding routes of even topologically complex proteins.…”

“…24,25,72,73 One result that has emerged from these studies is that, given the choice of 20 amino acids, naturally occurring contact strength variation is unlikely to affect the folding routes of topologically complex proteins such as β and α/β proteins. 24,25 Nevertheless, functional constraints can determine the choice of not just isolated but groups of interacting amino acids, and in such cases, energetic heterogeneity can affect the folding routes of even topologically complex proteins. 23,74 sensitivity of the folding route of Ub to contact map perturbation is an example of such an effect of functional constraints on folding.…”

“…Different flavors of C α -SBMs have been constructed with differing but structure derived dihedral terms, , contact definitions, − attractive and repulsive potentials, ,,− methods for assigning contact strengths, ,− etc. For most proteins, as long as the structure is encoded in the SBM and little frustration in the form of non-native interactions exists, protein folding routes calculated using differing models agree with each other, with atomistic simulations and with experiments. , This is because the differences in local energetic stabilization caused by the differences in potential energy function are not sufficient to overcome the topology (pattern of contacts) inherent to the protein structure.…”

CoCoNet—boosting RNA contact prediction by convolutional neural networks

“…Such a better metric is not trivial, as it would have to (A) weight each predicted contact in its assumed benefit for 3D modeling ideally in (B) the context of the other predicted contacts. (A) could be realized, for example, by weighting contacts with measures such as contact order from the protein folding field ( 60 ). (B), however, is likely more difficult to achieve.…”

Simulation of FRET dyes allows quantitative comparison against experimental data

“…29 They are based on energy landscape theory and the principle of minimal frustration. [30][31][32][33] SBMs are used successfully to study a wide range of phenomena 34 ranging from, e.g., protein folding, 35,36 misfolding, 37,38 structure prediction, 39 and conformational dynamics 40,41 to large biomolecules such as the ribosome 42 or RNA. 43 SBM simulations show good agreement with experimental results, e.g., they are used to reproduce transition state ensembles and "en-route" intermediates, 44 and folding rates comparable to experimental measurements.…”