Molecular Dynamics Simulations (PKU)

From Bioinformatikpedia
Revision as of 11:50, 29 August 2012 by Boidolj (talk | contribs) (additional first info)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Oh, good. For a moment there I thought we were in trouble.

Short task description

This week, we will start the molecular dynamics simulation of our protein with the wildtype and two mutations. The simulation will be analyzed at a later time, see the complete task description for details. Our journal briefly shows the commands used to create the simulations.

Selected Models

Besides the wildtype, we chose the hyperphenylalaninemia causing ALA322GLY mutation, since this mutation only reduces the enzyme activity and should have a visible but not severe effect on the protein, and the PKU causing ARG408TRP mutation, since this mutation should have a large impact on energy and flexibility of the protein, that has already been visible in the previous minimization simulations.


The wildtype 1J8U had problems equilibrating and AGroS aborted several times before the production runs, so we prepared the simulation manually. The final run finished in 11:32, the simulation seems successfull. The mutations ALA322GLY and ARG408TRP finished without problems in 04:29 and 04:34 respectively, the simulations seem also successfull.

The Molecular Dynamics Simulation

We simulated the movement of the three selected structure with AGRoS, an automated Gromacs simulation pipeline.

One AGRoS run consists first of basic checks of the input structure for breaks and gaps. Then, water in close distance to to the protein is saved as structural water, that might interact with the protein and be neccessary for protein stability, the other crystal water molecules are remoced. scwrl is used to complete sidechains and the complete structure is minimized for the first time in vacuum. Then, a surrounding box is defined and filled with water and ions to simulate solvent. First, the whole protein is fixed in place and the solvent minimized, then the backbone is fixed in place and the sidechains are minimized in solvent, finally the system is minimized with a different integration algorithm, again with fixed backbone.

After these preparations, two short MD runs try to relax the system from the crystalized to a more native state. The final simulation spans 10 nanoseconds in 2 million steps of 5 femtoseconds.

Intermediate Structures

During the simulation, a lot of intermediate PDB files are created. Most of them only have minimal differences (explained in the list below) but we show the differences after solvation and the three minimizations in the next subsections.

e.g. for the ARG408TRP mutant (model file mut408.pdb)

  • mut408.pdb: input structure with clashing water removed.
  • mut408_br.pdb: just the protein, no DNA, crystal water, etc..
  • mut408_br_0.pdb: 1 file for every chain, if several are present
  • mut408.pdb2: structure with hydrogens removed
  • mut408_repair.pdb: renumbered residues
  • mut408_repair_0.pdb
  • mut408_dna.pdb: extracted DNA, empty in our case
  • mut408_water.pdb: contains structural water <15A from the protein, empty in our case
  • mut408_sc.pdb: sidechains completed/adjusted with scwrl
  • mut408_nh.pdb: removed hydrogens, added structural water (and DNA)
  • mut408_solv_tmp.pdb: added ions
  • mut408_solv.pdb: duplicate water removed (none in our case)
  • mut408_solv.pdb2: just protein and added DNA, input to create individual files for each chain (see next step)
  • mut408_solv_0.pdb: used to create restriction files for every chain (only 1 here)
  • mut408_solv_min.pdb: structure with minimized solvent (protein fixed)
  • mut408_solv_min.pdb2: just protein of previous structure, input to create individual files for each chain (see next step)
  • mut408_solv_min_0.pdb: used to create restriction files for every chain (only 1 here)
  • mut408_solv_min2.pdb: minimization with only backbone restrained
  • mut408_solv_min3.pdb: final minimization minimization with only backbone restrained, but different integration algorithm (conjugate gradient insttead of steepest descent)

Wildtype

<figure id="fig:1J8U_solvation">

The starting structure for the molecular dynamics simulation of the wildtype in red and the solvated structure and the simulated water in a box in orange.

</figure>

<figure id="fig:1J8U_solvation_min1">

The starting structure for the molecular dynamics simulation of the wildtype in red and the structure in orange after the first minimization with restrained protein.

</figure>


<figure id="fig:1J8U_solvation_min2">

The starting structure for the molecular dynamics simulation of the wildtype in red and the structure in orange after the second minimization with unrestrained sidechains.

</figure>


<figure id="fig:1J8U_solvation_min3">

The structure for the molecular dynamics simulation of the wildtype in orange after the second minimization and in white after the third minimization with unrestrained protein.

</figure>

<xr id="fig:1J8U_solvation"/> shows the protein in the solvent box, taken from the *solv.pdb file. Not visible are the 47 Na+/Cl- ions added to the solvent.

In <xr id="fig:1J8U_solvation_min1"/> showing the first minimization in solvation there is one discernable difference: The turn between the beta sheets around 410 and 422 is apparently trying to form a helix that is not present in the crystal structure and some of the loops shifted slightly. This is very surprising as the protein structure should have been fixed during this minimization.

<xr id="fig:1J8U_solvation_min2"/> shows the second minimization with loosened sidechains. This minimization converged in 4519 steps. There are only minor differences to the next step in minimization with unrestrained backbone, visible in <xr id="fig:1J8U_solvation_min3"/> but the helix stump in the turn mentioned before disappeared, which lets us suspect that the helix is just a mistake or artifact of the pymol visualization. We checked the dssp asignment and the structure prediction did not change there. This minimization terminated after 2933 steps, before the maximum of 5000 steps was reached.

ALA322GLY

<figure id="fig:ALA322GLY_solvation">

The starting structure for the molecular dynamics simulation of mutant ALA322GLY in red and the solvated structure and the simulated water in a box in blue.

</figure>

<figure id="fig:ALA322GLY_solvation_min1">

The starting structure for the molecular dynamics simulation of mutant ALA322GLY in red and the structure in blue after the first minimization with restrained protein.

</figure>

<figure id="fig:ALA322GLY_solvation_min1_detail">

The starting structure for the molecular dynamics simulation of mutant ALA322GLY in red and the structure in blue after the first minimization with restrained protein, close-Up of the sidechains.

</figure>


<figure id="fig:ALA322GLY_solvation_min2">

The starting structure for the molecular dynamics simulation of mutant ALA322GLY in red and the structure in blue after the second minimization with unrestrained sidechains.

</figure>


<figure id="fig:ALA322GLY_solvation_min3">

The structure for the molecular dynamics simulation of mutant ALA322GLY in blue after the second minimization and in green after the third minimization with unrestrained protein.

</figure>

<xr id="fig:ALA322GLY_solvation"/> shows the protein in the solvent box, taken from the *solv.pdb file. In <xr id="fig:ALA322GLY_solvation_min1"/> showing the first minimization in solvation the first differences are noticable: The beta sheet around 410 to 422 is bent and some of the loops shifted slightly. This is surprising as the protein structure should have been fixed and we suspected this to be due to the display style of pymol, but a more detailed view in <xr id="fig:ALA322GLY_solvation_min1_detail"/> shows that the protein structure has indeed moved.

<xr id="fig:ALA322GLY_solvation_min2"/> shows the second minimization with loosened sidechains and now the previously more stable helices begin to move a bit. There are only minor differences to the next step in minimization with unrestrained backbone, visible in <xr id="fig:ALA322GLY_solvation_min3"/>. This minimization also terminated after 4094 steps, before the maximum of 5000 steps was reached.


ARG408TRP

<figure id="fig:ARG408TRP_solvation">

The starting structure for the molecular dynamics simulation of mutant ARG408TRP in red and the solvated structure and the simulated water in a box in blue.

</figure>

<figure id="fig:ARG408TRP_solvation_min1">

The starting structure for the molecular dynamics simulation of mutant ARG408TRP in red and the structure in blue after the first minimization with restrained protein.

</figure>

<figure id="fig:ARG408TRP_solvation_min1_detail">

The starting structure for the molecular dynamics simulation of mutant ARG408TRP in red and the structure in blue after the first minimization with restrained protein, close-Up of the sidechains.

</figure>


<figure id="fig:ARG408TRP_solvation_min2">

The starting structure for the molecular dynamics simulation of mutant ARG408TRP in red and the structure in blue after the second minimization with unrestrained sidechains.

</figure>


<figure id="fig:ARG408TRP_solvation_min3">

The structure for the molecular dynamics simulation of mutant ARG408TRP in blue after the second minimization and in green after the third minimization with unrestrained protein.

</figure>

<xr id="fig:ARG408TRP_solvation"/> shows again the protein in the solvent box, taken from the *solv.pdb file. In <xr id="fig:ARG408TRP_solvation_min1"/> showing the first minimization in solvation we see similar differences as with ALA322GLY: The beta sheet around 410 to 422 is bent and some of the loops shifted slightly. The introduced mutation lies just before this, in the loop between the helix and the sheet. <xr id="fig:ARG408TRP_solvation_min1_detail"/> shows the details of the warped sheet.

<xr id="fig:ARG408TRP_solvation_min2"/> shows the second minimization with loosened sidechains and now we see e.g. the short helix on the left (around residues 370-376) shifting slightly as well as the loops and the sheet on the left. This second minimization terminated just before the maximum of 5000 steps after 4999 steps. There are only minor differences to the next step in minimization with unrestrained backbone, visible in <xr id="fig:ARG408TRP_solvation_min3"/>. This minimization also terminated after only 2341 steps, before the maximum of 5000 steps was reached.

Altogether there is little influence of the mutation visible, despite the large difference of the introduced amino acid, when we compare only the images of the minimization steps with the images of the wildtype and the other mutant.


additional first info

<figure id="fig:surfacewithna1J8U" >

Close-up of the site, of the 1J8U protein after simulation. Here in blue, a Sodium ion is bound and hold within a pocket by three residues which are shown in sticks. The whole protein is colored according to the element color codec of PyMol.

</figure>As we had a look at the pdb files with PyMol, we stumbled across some pictures which show the protein with ions in water, but those ions where integrated into the protein. As we did not want to analyze to much in this weeks task, we just present some pictures here (<xr id="fig:surfacewithna1J8U" />), which show an example. Not that this picture is a close-up and can be found in any of our simulations, which shows, that this is the expected behavior of the protein wildtype and not any coeffect of a mutant.
As well as for a static binding of the ions we decided to look at the trajectories of the protein throughout the simulation. As the animation is to big for this wiki, we encourage the reader to have a look at the movement of the protein alone and the [anderer link hier protein with ions]. Note that these animated pictures are all simulated in water, but they were removed for a better overview in the pictures.