Miniprotein folds in two ways at the same time

Proteins are the molecular workhorses of all living organisms. Amongst others, they play a crucial role in the cell structure, in the transportation of nutrients and in the production of body characteristic matter. Each somatic cell has a nucleus containing the complete genetic material, DNA (deoxyribonucleic acid). On this DNA you find the genes, containing the molecular code that shapes the proteins. The proteins themselves have a compound three-dimensional structure.

The building blocks of proteins are amino acids. Indeed, we know the sequence of the building blocks of most proteins, but we do not know how they have been folded to a three-dimensional structure in their active shape. The big question is whether we are able to predict which folding is stablest. Therefore, we need computer simulation, but the problem is that there are many ways at which a straight chain of protein building blocks is able to fold to a three-dimensional protein. A medium-sized protein, consisting of hundreds of amino acids, is theoretically able to fold up to about 10⁹⁰ possible ways. For small proteins computer simulations try to discover which of the possibilities do occur in practice, and which do not. For the time being, medium-sized and large proteins are still out of computational reach of the current simulation techniques. Once we know how the protein structure looks like and how it is being established, then we are able to understand the function of the protein better.

Artificial miniprotein
Researcher Peter Bolhuis - appointed on a FOM-‘stepping stone’ position and recently appointed half-time professor in the University of Amsterdam - and doctoral student Jarek Juraszek have been studying the folding of an artificial miniprotein, consisting of twenty amino acids. This miniprotein has been devised especially for folding up quickly, while it still contains all characteristics of a protein. As a specific natural protein has a folding time between milliseconds and one second, the investigated protein folds up in four microseconds. This will considerably shortens the time of the computer simulations.

All atoms of the protein (about three hundred) are being taken to the simulation. Other than in previous simulations, the watery solution, in which the protein is being found in a genuine cell, is also simulated. A total of about three thousand water molecules enclose the simulated protein. Without using an extra tric, simulation of spontaneously folding of the small protein would be taken up some years.

Common practice to the living cell is the dynamic balance of the protein in a folded and an unfolded state: both states occur side by side. The protein has to overcome an energy barrier in order to move from an unfolded state into a folded one - and also vice versa.
The researchers have been accelerating the simulations by a factor of one thousand by zooming in critically on this energy barrier. Thus for the first time, Bolhuis and Juraszek have been carrying out a simulation that computed all folding routes of a protein in a solution without simplifying.

Dominant folding route
Although theoretically many possibilities of folding exist, practically a protein often turns out to follow only one overall route. It is as if travelling from Southern Germany to Italy through the Alps that in theory offers you many possibilities but in practice you choose a route that runs via the Brenner Pass. However, the simulations of the researchers at the University of Amsterdam prove that the investigated miniprotein will choose the one folding route (L) in eighty percent of the cases and in twenty percent of the cases it will choose the other route (I). Previous simulations carried out by a less accurate technique showed an opposite result: the I-route proved to be the dominant one.

To their surprise, the researchers discovered that each of the two routes (I and L) appeared to be a role model for one of the two general classes of protein folding. In the first way local units such as helixes, build up inch by inch, after which the units collide by diffusion and eventually, they build up a folded protein structure. In the second way a kind of condensation nucleus exists, around which the remains of the protein will condense, so to speak. It had not been observed before in a simulation that the two completely different mechanisms are able to exist side by side in the folding of one kind of protein. However, it is not yet known to what extent this phenomenon will also occur at larger, natural proteins.

A third and important result is the fact that simulating a watery solution is fundamental for having the correct protein folding originate. Previous computer simulations took the effect of the water only in an average way, without having individual water molecules correlate with the folding protein.

A better understanding of protein function
The research by Bolhuis and Juraszek has shown that theoretically, it is now possible to predict the exact folding mechanisms of proteins in a watery solution by way of detailed computer simulation. In the future this will lead to a better understanding of the function of proteins.

The article is entitled:
Sampling the multiple folding mechanisms of Trp-cage in explicit solvent, Proceedings of the National Academy of Sciences of the United States of America (pnas), Jarek Juraszek and Peter Bolhuis.

For more information, please contact Jarek Juraszek (020) 525 64 92 or dr. Peter Bolhuis (020) 525 64 47.