This week David Shaw set another milestone in the protein-folding field. Two years ago he was the first to show how twelve “fast folding” proteins fold. Now he is the first to show how a “slow” protein folds. As test case he used the well-studied ubiquitin protein. The main goal was to prove the folding rules they found for “fast folders” is also applicable to “slow folding” proteins.
This paper is a follow-up of the Science paper from 2011 (Lindorff-Larsen, Piana, Dror, & Shaw, 2011) , where “fast folding” (<50μs) proteins with an average size of ~50AA were folded in a MD simulation. In their recent paper (Piana, Lindorff-Larsen, & Shaw, 2013) ubiquitin, with a folding time of ~3 ms and a size of 76AA is folded and unfolded in a staggering 8 ms of simulation time. To achieve this they used the custom-build MD machine Anton running an altered CHARMM force field. In order to bring the simulation close to the melting temperature a simulation temperature of 390°K (=117°C) was chosen, in other words; ubiquitin was boiling. The authors have a good reason to do this, since in vitro experiments are usually carried out in highly acidic conditions due to the high stability of ubiquitin. In order for this in silico simulation to run at neutral pH they just had to increase the temperature.
The folding landscape of ubiquitin highlighting the different structures in the ‘energy basins’. As can rapidly seen it folds from an unfolded state via a distinct intermediate into the folded state. It is interesting to note that major secondary structures and an ubiquitin like topology is already visible in the transition state ensemble (TSE). The colored lines on top denote the structural properties of the folding trajectory (blue, number of helical residues; orange, number of sheet residues; green, Cα RMSD; purple, contact order; red, fraction of native contacts). Copyright by PNAS from(Piana et al., 2013).
In this post I would like to highlight two results, first of all the free-energy surface we all know from the classical textbooks. In their paper it is nicely described as can be seen in Figure 1. Next to an ensemble of unfolded states they were able to highlight the transition state ensemble (TSE), which is, using experimental techniques, impossible to visualize. From the energy landscape and kinetic analysis (check out the paper, Figure 1C for that one!) it is also clear ubiquitin behaves as a two-folder, meaning it is most of the time either folded or unfolded. Although this paper showed some meta-states are occupied. In particular there is a misfolded meta-state (MF3) which has near-native conformation (Cα RMSD 2 Å) which it is unstable. This example just shows that Cα RMSD doesn’t always tells you the truth.
The second subject I would like to highlight is the experimental validation of this simulation. Although the authors did not perform experimental studies themselves, they used previously gathered data of the influence of mutations on ubiquitin. These mutations are then analyzed by comparing the Φ value’s. A Φ value has nothing to do with regular usage of the Φ (phi) in protein science, where it usually describes a part of the backbone dihedral angle. In this case it is used to describe the influence of an amino acid on the transition state. The method was developed more than 20 years ago by Alan Fersht (Fersht, Matouschek, & Serrano, 1992; Matouschek, Kellis, Serrano, & Fersht, 1989) and is still used today. It is based on the principle that the structure of the protein is closely related to the energy landscape. It’s a two step experiment, first making one-by-one mutations and secondly constructing a free-energy profile of the mutant. A value of Φ = 0 implies the mutation has no effect on the structure of the transition state where as a value of Φ = 1 implies the structure is as folded in the transition state as in the folded state.
Ubiquitin (PDB 1 UBQ) showing the α-helix in red and the first 2 β-strands in blue. From the Φ analysis it becomes apparent this is the folding nucleus.
From this paper it becomes clear the folding nucleus is localized the N-terminal region of the protein and involving the α-helix and first two β-strands of ubiquitin. This conclusion is based on the Φ value analysis, because this showed high Φ-values of the first three β-strands and a part of the α-helix (figure 2). This perfectly aligns with previous studies which elucidated the Φ value in an experimental manner and also pinpointed the initial folding nucleus at the N-terminal part of the protein.
Taken together they were able to extend the conclusion from their previous work (Lindorff-Larsen et al., 2011) that the native state topology and secondary structures tend to form before the majority of the long-range contacts (which are formed after the transition state). Now lets wait another year and see which new proteins gets boiled by Anton.
References
Piana, S., Lindorff-Larsen, K., & Shaw, D. E. (2013). Atomic-level description of ubiquitin folding. Proceedings of the National Academy of Sciences, 110(15). (Featured article)
Lindorff-Larsen, K., Piana, S., Dror, R. O., & Shaw, D. E. (2011). How fast-folding proteins fold. Science (New York, N.Y.), 334(6055), 517–20.
Matouschek, A., Kellis, J. T., Serrano, L., & Fersht, A. R. (1989). Mapping the transition state and pathway of protein folding by protein engineering. Nature, 340(6229), 122–126.
Fersht, A. R., Matouschek, A., & Serrano, L. (1992). The folding of an enzyme: I. Theory of protein engineering analysis of stability and pathway of protein folding. Journal of Molecular Biology, 224(3), 771–782.