Difference between revisions of "Structure-based mutation analysis BCKDHA"
|Line 249:||Line 249:|
Minimise calculates the energy for a mutation by building a new model for each mutation. Then it calculates the energy for the whole mutated model. The aim by comparing the mutated models with the wildtype is to find out if there is a structural change caused by a mutation. We superposed each mutated protein with the wildtype and focused on the mutated position. The
Minimise calculates the energy for a mutation by building a new model for each mutation. Then it calculates the energy for the whole mutated model. The aim by comparing the mutated models with the wildtype is to find out if there is a structural change caused by a mutation. We superposed each mutated protein with the wildtype and focused on the mutated position. The show the superimposed structures. The pictures showing the wildtype structure display the unmutated residue in bold and vice versa. So we can compare the two pictures to see if there is a change in the structure caused by the mutation of this residue.
Revision as of 18:52, 10 August 2011
- 1 Structure selection
- 2 Mapping of the mutations on the crystal structure
- 3 Comparison energies
- 4 Gromacs
- 4.1 General
- 4.2 Analysis
- 4.3 Mutation analysis
- 5 Links
In order to perform a structure based mutation analysis an appropriate protein structure is needed. Therefore one has to check whether the resolution of the structure is high enough, that means the Å-value should be small. The R-factor (reliability factor or residual factor) is another measure to access the quality of a structure. The R-factor rates how well the crystallographic model predicts the observed data from experimental X-ray diffraction. The smaller the R-factor the better. The coverage is the percentage of UniProt sequence present in the sample structure. This value was obtained from []. The coverage should be quite high. The pH-value indicates the pH-value at which the structure was resolved. Ideally it should be resolved at physiological pH (7.4).
The following table presents the PDB structures for BCKDHA to date. For each structure the resolution in [Å], the R-factor, coverage and pH-value are listed as well.
|PDB id||resolution [Å]||R-factor||% of UniProt sequence present in the sample||%residues observed||ph-value|
The asteriks-marked values indicate that these structures were resolved with the asked experimental quality. As one can see, none of the structures fulfills all conditions.
Furthermode, we could not use any of the PDB structures for BCKDHA because all of them had gaps in the secondary structure which means that some residues were missing. So we took the structure which has the less gaps: 1U5B
- resultion: 1.83
- R-factor: 0.156
- ph-value: 5.8
This structure has to be modified with some programms to close the gaps. Additionally the first residues which are in BCKDHA misses in 1U5B thats why the start position corresponds to position 6 of the BCKDHA -PDB sequence.
As we can see none of the values corresponds to the demands because it was asked for a structure which has a very small R-factor, a pH of 7.4 and a high resolution.
Mapping of the mutations on the crystal structure
Figure 1 shows the structure of BCKDHA with mapped mutations.
Hydrogen bonds are interactions between an hydrogen atom and an electronegative atom. Electronegative atoms which often take part in hydrogen bonds are oxygen, nitrogen and fluorine (not present in amino acid side chains). They serve as a hydrogen bond acceptor, whereas a hydrogen bond donor is a electronegative atom bonded to a hydrogen atom. Hydrogen bonds are essential for the three-dimensional structures of proteins. They play a important role in the formation of helices and beta-sheets and cause proteins to fold into a specific structure.
Showing hydrogen bonds with Pymol: A -> find -> polar contacts -> within selection The respective amino acids were colored by element, s.t. oxygen is red, nitrogen is blue, hydrogen is white and sulfur is yellow.
The following figures show the hydrogen bonds between the wildtype residue and its environment compared to the formation of hydrogen bonds when the corresponding residue is mutated.
Comparing the figures 2 and 3 for the wildtype and the mutated amino acid on position 82, no change in the hydrogen bonding network can be observed. This is due to the similar physiochemical properties of these two amino acids. No atom which could serve as additional hydrogen-bond donor or acceptor was introduced or removed.
The substitution from glutamine to glutamic acid changes the side chain properties completely. A NH2 group is substituted by a negatively charged oxygen. The NH2 which served in the wildtype structure as a hydrogen bond acceptor (see Figure 4) is not present any more, so the hydrogen bonding network changed for this substitution (compare Figure 4 and 5).
Although tyrosine and asparagine both could play a role in the hydrogen bonding network, no hydrogen bond is formed for position 166 (see Figure 6 and 7). Therefore this substitution has no influence on the hydrogen bonding network of the protein.
Introducing a serine on position 249 leads to the formation of several additional hydrogen bonds (see Figure 9). Two of the newly established bonds are due to the new hydroxy group which is very likely to participate in hydrogen bonds. Another additional hydrogen bond is formed using the nitrogen atom as a hydrogen bond acceptor.
Although the amino acids cysteine and tryptophan have very different structures and chemical properties, no change in the hydrogen bonding network occurs (compare Figure 10 and 11).
The mutation from arginine to tryptophan leads to a drastic change in the hydrogen bonding network. Arginine, which contains three nitrogen atoms in its side chain is removed and therefore three hydrogen bond acceptors (see Figure 12) are missing in the mutated protein (Figure 13).
The mutation from isoleucine to threonine doesn't have an influence on the hydrogen bonding network, although the oxygen atom of threonine could serve as an additional hydrogen bond donor (compare Figure 14 and 15).
The phenylalanine side chain in the wildtype protein does not participate in any hydrogen bonds (see Figure 16). The mutation to serine doesn't introduce new hydrogen bonding donors or acceptors, therefore the mutation has no effect on the hydrogen bonding network (see Figure 17).
The hydrogen bond donor property of the amino acid on position 438 is maintained but the bond seems to be between different sidechains now (Compare Figure 18 and 19). This substitution also disturbs the hydrogen bonding network of our protein.
Before we could use SCWRL we first had to get the sequence of our model:
repairPDB bckdha.pdb -seq >> bckdha.seq
When we have the sequence we have to make one file for each mutation. In these files we copied the bckdha.seq and changed the sequence to lower case letters. Then we add the mutation in an upper case letter.
To run SCWRL we used the command:
scwrl -i bckdha.pdb -s mutation1.seq -o mutation1Model.pdb
Total minimal energy of the graph
To use foldX we first build a runscript. It is important to change values of <Temperature> and <pH> to the values of the used protein. These values can be found on the pdb side .
Additionally we had to create one file with all PDB Ids each in a new line (list.txt). We used the command
Foldx -runfile run.txt > Stout.txt to run the programm.
After using foldx we have the total energy for the wiltype protein and for each mutation. The value of the wildtype protein is 401.00 which is already a high value. This means that the protein is quite instabile. To find out which mutation has a high influence on the protein we look at the energies and especially on the difference between the energy of the mutated protein and the wildtype protein. All of the mutated proteins have a much higher energy than the unmutated protein which means that these proteins are less stable. We can see in the table that the proteins can be divided into two groups. The first group has an energy difference of about 31 and the other group has a much higher difference.
It is important to remove the hydrogens and water before using the programm. For this we used the new version of repairPDB of the virtualbox. The programm can be started with the command:
repairPDB bckdha.pdb -nosol out.pdb > Stout.txt
It is also possible to use the old version but then the command is:
repairPDB bckdha.pdb -nosol -noh out.pdb > Stout.txt
It is useful to save the output in a file because it includes the energy.
Minimise calculates the energy for a mutation by building a new model for each mutation. Then it calculates the energy for the whole mutated model. The aim by comparing the mutated models with the wildtype is to find out if there is a structural change caused by a mutation. We superposed each mutated protein with the wildtype and focused on the mutated position. The figures 20-37 show the superimposed structures. The pictures showing the wildtype structure display the unmutated residue in bold and vice versa. So we can compare the two pictures to see if there is a change in the structure caused by the mutation of this residue.
|mutation||wildtype structure||mutated structure|
The first part describes general background information for gromacs as well as how to run those programs. The second part contains the result description and analysis.
The fetch-pdb script first checks, if it was called with an valid PDB-id. If the entered PDB code has 4letters, the script tries to download the pdb-file from the server. The successfully downloaded folder gets unzipped and everything except the actual pdb file is removed.
For repairPDB the following options are available:
|-offset value||offset the residue numbering|
|-chain char||change Chain ID|
|-het||no change of HETATM to ATOM for AA|
|-seq||returns protein sequence from AA in pdb file|
|-seqrs||protein sequence from SEQRES entries|
|-nosol||just Protein, no solvent OR|
|-ssw cutoff||print only waters with B-value below cutoff OR|
|-cleansol||remove overlapping solvent for GROMACS|
We run repairPDB using the following command:
repairPDB bckdha_mod.pdb -noh -nosol > bckdha_clean.pdb
Using this command we removed hydrogens and solvent from our pdb to get just the protein.
SCWRL was executed using the following command:
scwrl -i bckdha_mod.pdb -s extractedPDB.seq -o bckdha_scwrl.pdb
SCWRL returned a pdb including HETATOMS. These solvent atoms needed to be removed before continuing.
(use clean pdb without HEATOMS!)
pdb2gmx -f bckdha_clean.pdb -o bckdha.gro -p bckdha.top -water tip3p -ff amber03
title = PBSA minimization in vacuum cpp = /usr/bin/cpp define = -DFLEXIBLE -DPOSRES implicit_solvent = GBSA integrator = steep emtol = 1.0 nsteps = 500 nstenergy = 1 energygrps = System ns_type = grid coulombtype = cut-off rcoulomb = 1.0 rvdw = 1.0 constraints = none pbc = no
adjust nsteps for the time vs steps analysis
|integrator||a steepest descent algorithm for energy minimization.|
|emtol||tolerance for steep integrator:the minimization is converged when the maximum force is smaller than this value|
|nsteps||maximum number of steps to integrate or minimize, -1 is no maximum|
|nstenergy||frequency to write energies to energy file (last energies are always written)|
|energygrps||groups to write to energy files|
|ns_type||Which atoms to check when construction a new neighbor list|
|coulombtype||which coulomb-type to use, e.g. Cut-off, PME, Ewald|
|rcoulomb||distance for the Coulomb cut-off|
|rvdw||Van-der-Waals cut-of (distance for the LJ or Buckingham cut-off)|
|constraints||constraints for bonds, e.g bonds are presented by a potention, all bonds are converted to constraints,...|
|pbc||Remove the periodicity (make molecule whole again)|
grompp -v -f bckdha.mdp -c bckdha.gro -p bckdha.top -o bckdha.tpr
7. System Minimization
mdrun -v -deffnm bckdha 2> mdrun_out.txt
g_energy -f bckdha.edr -o energy_1.xvg
Wildtype analysis: nsteps vs time
The table below shoes the running time for mdrun depending on different values for nsteps. It also lists the real number of steps carried out to calculate the energy.
|steps||time (real) [s]||time (user) [s]||time (sys) [s]||performed steps|
The following plot shows the correlation between nsteps and the running time for mdrun
Interestingly, the running time is not dependent on the number of nsteps, but just on the number of really performed steps. There is a linear dependency between the calculation time and the number of performed steps. The number of performed steps however is not correlating with the value for nsteps. It is not obvious why the number of performed steps varies so extremely given a certain value for nsteps.
Wildtype analysis: force fields
The different force fields chosen for this task were:
|Force Field||Average||Err. Est.||RMSD||Tot-Drift (kJ/mol)|
|Force Field||Average||Err. Est.||RMSD||Tot-Drift (kJ/mol)|
|Force Field||Average||Err. Est.||RMSD||Tot-Drift (kJ/mol)|
go back to Maple syrup urine disease main page
go back to Task 6 Sequence based mutation analysis
go to Task 8 Molecular Dynamics Simulations