nobahar said:
As a kind of addendum. How far along generally are folding algorithms, are there any good mathematical theories to predict protein folding. I would be very suprised if you couldn't predict folding.
There are two general approaches to predicting protein structure. The first approach is inspired by the central principle underlying biology: evolution. For these approaches, you look for proteins that are similar in sequence to your target protein with the assumption that proteins with similar sequences were likely to have evolved from a common ancestor and likely have the same or similar structure. By comparing the search hits to databases of known protein structures, one can model the structure of the target by threading it through the structures of the closely matching hits creating what is called a homology model.
The other general approach is based more in physics/chemistry. Here, you start with a "force field" which defines the energies of the interactions between the different chemical groups in a protein. You then use Monte-Carlo methods to search for the lowest energy configuration of the amino acid chain, which you assume to be the folded state of the protein. This is challenging, however, because the energy landscape over which you need to search is very rough and contains many local minima which can complicate the Monte-Carlo simulations. This approach, however, has been used successfully to predict the structures of small proteins from first principles, an impressive feat in itself.
Obviously these two approaches can be combined to be even more successful: the homology modeling can create a good initial guess of the protein structure which can then be refined via Monte-Carlo methods.
A problem with these approaches, however, is that they do not give you much information about the actual mechanism of protein folding (this is especially true for the biology-inspired approach, some have used Monte-Carlo simulations to try to gain some insight into protein folding). To computationally simulate protein folding, you need to perform molecular dynamics calculations. Here, you take the protein you want to study and calculate the forces between all of the atoms in the system using your force field, write out Newton's equations for each atom in your system, let the atoms move slightly, recalculate all the forces, and repeat. The step sizes between recalculating the forces need to be very small (~ one femtosecond) or else the atoms in the simulation basically explode (bonds start breaking and atoms fly all over the place). Obviously, this approach is very computationally intensive for systems as large as proteins. Until very recently, computers were limited to simulating at most a few nanoseconds-to-microseconds of time whereas proteins fold on the millisecond or longer timescale. Recently, a research group specifically designed a supercomputer to perform long molecular dynamics simulations and were able to achieve 1 ms long simulations, long enough to observe the folding of a small protein (Shaw et al. 2010 Atomic-Level Characterization of the Structural Dynamics of Proteins. Science, 330: 341-346. http://dx.doi.org/10.1126/science.1187409, free news writeup: http://pubs.acs.org/cen/news/88/i42/8842notw1.html ).
But there are some proteins that have to be folded in certain environments, or 'assissted' in folding (such as the use of chaperones). I have never understood why it is necessary. Surely when you transfer the protein to another environment (presumably one in which it would not have folded correctly, hence the need for chaperones) wouldn't it's shape alter? Or is it such features as disulfide bridges that help preserve the shape? Since hydrophobic/philic interactions, and ionic interactions, are subject to change based on the environment much more than covalent bonds are. But since sometimes 'assistance' is required, that means that either predicting shape is going to be based on probabilities or the protein is fairly rigid and retains it shape under some degree of environmantal pressure to change.
One thing to keep in mind that the cell is a very crowded environment. Furthermore, the free energy of folding of most proteins is very small, on the order of the energy of only a handful of hydrogen bonds. At physiological temperature, protein structures "breath", thermal energy can cause them to transiently open up slightly. This "breathing" can expose the hydrophobic core of the proteins and if they collide with another "breathing" protein, the hydrophobic cores can stick together and you can get some misfolded proteins. Chaperon proteins are required to identify and refold these proteins or target them for degradation. Chaperons are also required to keep other proteins from disturbing the folding of newly synthesized proteins as they come off the ribosome and before they have a chance to fold.
Last edited by a moderator: