Along the pathway from DNA to functional protein, much is already understood. DNA is transcribed into RNA which is then translated into the primary structure of a protein. Similarly, for many proteins the mechanisms by which they function is well documented. However, the processes by which an unfolded string of amino acids folds into a complex secondary and tertiary structure is still shrouded in mystery.
Unfortunately, the variable, dynamic, and short-lived nature of even the most important and stable states along the folding pathway make experimental analysis difficult. Even in simulations, the folding pathway cannot be easily examined because of the incredible computational demand of checking all possible steps that might be taken by the folding protein. These problems are addressed using the microscopically reversibility of protein folding. By studying unfolding pathways in molecular dynamics simulations—a much less computationally intensive task—a great amount of insight can be gained about the folding pathway.
From unfolding simulations, it is possible to characterize all important points along the folding/unfolding pathway from the native and near-native states to the transition state to important shared intermediates and the denatured state. By characterizing all of these states, major events in unfolding become clear—loss of hydrophobic core packing or hydrogen bonding, change in dihedral angles, formation of non-native contacts, etc. When these events are understood in the unfolding direction, it can then be shown how they affect the folding direction. For instance, loss of hydrophobic core packing in the unfolding transition state would indicate formation of the hydrophobic core in the folding transition state or highly conserved non-native contacts in a shared unfolding intermediate may indicate that this non-native contact is important to the formation of other parts of the native structure in the folding direction. These sorts of insights can lead to a well characterized and understood folding/unfolding pathway for any protein or protein family.
Misfolding in prion protein (PrPprot) simulations. Left: starting structure. Right: mid pH simulation 3 at 49.9 ns. Hydrophobic residues 134, 137, 139, 141, 205, 209, and 213 are shown with translucent surface and stick representations of the side chains. E4 region is colored in orange. (image adapted from DOI)
Thermal unfolding pathways of SOD1. (A) Early unfolding events of the dominant unfolding pathway, which is followed by simulations 498–1,498-3 and 498-5. Structures and associated times are from 498-1. (B) Alternate early unfolding events. Structures and associated times are from 498-4, which shows the largest difference from the dominant pathway. (image source)
Towards this end, we have simulated over 800 distinct proteins under both native and unfolding conditions for a total of over 6000 simulations. For over 1300 of these simulations representing more than 180 distinct proteins, the transition state has been identified and characterized, and for five of these the transition state was verified against available experimental data. When general properties such as solvent exposed surface area and number of native contacts were calculated for all of these transition state ensembles, it was found that the standard deviations were small indicating that there are common rules influencing the structure and properties of all transition states and folding pathways. The overarching goal then becomes to elucidate these rules through the further study of molecular dynamics simulations.