Search

link to homepage

Institute for Advanced Simulation (IAS)

Navigation and service


GS-α3W: A 67 residue 3-helix bundle

GS-α3W is a designed 67 residue 3 helix bundle protein. It's structure was determined by NMR and has the PDB id 1LQ7. It is a relatively simple 3 helix bundle, with the two turns between the 3 helices composed of 4 Glycines each. Several simulations of this protein have been performed with ProFASi since our force field article, and this protein is now one of our test systems. In the test phase of ProFASi version 1.5, this system was simulated using 2048 cores of a Intel Nehalem based supercomputer for 70 minutes. 394 replicas had found states with RMSDs 5 Å or less, at least once.

3 helix bundle 3 helix bundle
Left: One snapshot from the free energy minimum seen in the simulations (colour) superimposed on the PDB structure (gray). Right: More detailed view of the free energy minimum structure. Hydrophobic side chains are shown with thick lines, hydrophilic side chains with thin lines. Positively charged side chains are blue, negatively charged ones are brown. Glycine is green.

In the figure above, we see that the native state packs in the hydrophobic residues in the space between the helices, and leaves the charged residues to the outside. Not only is the native state stable, it is found relatively easily in Monte Carlo simulations, starting from random initial conformations. The following animation shows one instance of the formation of the native state in a parallel tempering simulation. The least of the benefits of such an animation is that it shows the Monte Carlo search is not about small perturbations of the native state.

 

One instance of a folding event for 1LQ7 using parallel tempering MC simulations with ProFASi. In the animation, hydrophobic side chains are shown as thick white lines, secondary structure elements are colour coded with red for helices and green for β-sheets.

Although this protein is larger than CFr, it has a much simpler structure, and folds relatively easily. Every hour of simulation time (2.4 GHz, AMD Opteron) with a 32 replica parallel tempering run finds more than 1 independent events similar to the one depicted in the above animation. Such animations are only to help visualise the Markov chains created in the simulations. The above is not a "video recording" of a moving protein. Such a thing does not exist. Please read the disclaimer on these visualisations in the small peptides page.

The following figure shows how different properties of the system vary in the simulation for the interval shown in the above animation.

3 helix bundle 3 helix bundle
The plots show the span of MC time covered in the above animation. The vertical dashed line is at the point where RMSD goes below 3 Å for the first time in this interval. Left: RMSD, helix content(H) and backbone hydrogen bonds (EHB) vs MC time. Right RMSD, hydrophobicity energy (EHp) and total energy (Etot) for the same span of MC time.

Notice that the helix content rises and falls several times without any trace of it appearing on the RMSD curve. This is because large RMSD values such as 20 Å can result from completely unstructured chains as well as "unbundled" folded helices. Notice also that the hydrogen bond energy is related almost trivially with the helix content. The most probable hydrogen bonds in this system are those associated with the α-helices. Notice that the hydrophobicity energy only reaches its optimum values when the RMSD reaches the native-like values, whereas hydrogen bond (in the left figure) saturates a little before that. Taken together, these plots point towards a folding process in which the helices form independently, before being assembled into a 3-helix bundle. The helix formation is driven mainly by hydrogen bonds where as arrangement of the folded helices is driven by hydrophobicity.

References

The results discussed in this page have been published in the following articles.
  • An Effective All Atom Potential for Proteins, Anders Irbäck, Simon Mitternacht and Sandipan Mohanty, (2009) PMC Biophysics 2:2
  • Protein folding and Structure Prediction at the Simulation Laboratory Biology, Jan H. Meinke, Sandipan Mohanty, Olav Zimmermann, (2010) NIC syposium (submitted)

 


last change 20.11.2009 | SL Biology | Print


Servicemeu

Homepage