View source for Task 9: Structure-based mutation analysis

<css>
table.colBasic2  {
margin-left: auto;
margin-right: auto;
border: 1px solid black;
border-collapse:collapse;
width: 40%;
}
 
.colBasic2 th,td {
padding: 3px;
border: 1px solid black;
}
 
.colBasic2 td {
text-align:left;
}
 
/* for orange try #ff7f00 and #ffaa56
for blue try #005fbf and #aad4ff
 
maria's style
blue: #adceff
grey: #efefef
*/
.colBasic2 tr th   { background-color:#efefef; color: black;}
.colBasic2 tr:first-child th { background-color:#adceff; color:black;}
</css>

[[Lab_Journal_Hemochromatosis_Task9]]
== Structure Selection ==

<figtable id="structures">
{| class="colBasic2"
! PDB ID || Res [A] || R-value (obs) || pH || missing residues || coverage
|-
| 1A6Z || 2.60 || 2.33 || 6.5 || 1-3 || 83.4%
|-
| 1DE4 || 2.80 || 2.31 || 8.0 || 1-3 || 83.4%
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>''' Table 1:''' List of available structures for the human hemochromatosis protein in the PDB.
|}
</figtable>

From the two available structures listed in <xr id="structures"/>, we chose 1A6Z, because it has a slightly higher resolution and a nearly identical R-value compared to 1DE4. On the downside, 1A6Z was resolved at a pH value of 6.5, which is more distant to the physiological pH than the resolution pH of 1DE4. 1A6Z was chosen nevertheless, because it was used in all previous tasks, in order to keep consistency.

== Mutations ==

<figtable id="mutations">
{| class="colBasic2" style="width:30%"
! Mutation || Disease causing ?
|-
| Val53Met || Yes
|-
| His63Asp || Yes
|- 
|  Met97Ile || No
|-
|  Thr217Ile || No 	
|-
| Cys282Tyr || Yes
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>''' Table 2:''' List of selected mutations. The mutation positions correspond to the Uniprot entry [http://www.uniprot.org/uniprot/Q30201 Q30201].
|}
</figtable>
We chose 5 mutations from the ones described in task 8, some of which are disease causing and some not. They are listed in <xr id="mutations"/>.

<figure id="mut_overview" >
{| style="width: 40%; border-width:0px" align="center"
| [[File:Mut_1_lab.png|center|thumb|500px]] || [[File:Mut_2_lab.png|center|thumb|500px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 1:''' The PDB structure 1A6Z of HFE_HUMAN with the MHC I domain colored in limegreen and the Ig C1-set domain colored in blue. The disease causing mutations are shown in red and the non-disease causing in orange. The cysteine that is the disulfide bridge binding partner of C282 is colored in white.
|}
</figure>

<xr id="mut_overview"/> shows the 3D structure of the HFE protein and the 5 mutations. The disease causing ones are marked in red and the non-disease causing ones in orange.

== Structure Mutation using SCWRL ==

SCWRL was used to mutate the HFE protein. The mutated structures are shown and analysed in the following.

=== Val53Met ===
<figure id="53_mut">
{| class="colBasic2"
| [[File:53_wt.png|center|thumb|300px]] || [[File:53_mut.png|center|thumb|300px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 2:''' Visualisation of the HFE protein (green) with the Val53Me mutations, The wild type valine is shown in grey and the mutant in red. Hydrogen bonds are colored in yellow.
|}
</figure>

<xr id="53_mut"/> shows that the mutation to methionine does not change the polar contacts of the residue. But methionine extends further into the binding pocket than valine, which might disturb the structure of the binding pocket and inhibit the binding process.

=== His63Asp ===

<figure id="63_mut">
{| class="colBasic2"
| [[File:63_wt.png|center|thumb|300px]] || [[File:63_mut.png|center|thumb|300px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 3:''' Wild type (grey) and mutant (red) residues for position 63.
|}
</figure>

Regarding the polar contacts of residue 63, there is no change, because both variants do not exhibit polar interactions (<xr id="63_mut"/>). Also, the mutation lies in a loop region and does not disturb an ordered secondary structure.
But nevertheless, the mutation is disease causing, which might be due to the fact that the loop where it is located is still part of the binding interface to ferritin and that the change from an aromatic, mainly uncharged residue to a negatively charged residue disturbs this interface.

=== Met97Ile ===

<figure id="97_mut">
{| class="colBasic2"
| [[File:97_wt.png|center|thumb|300px]] || [[File:97_mut.png|center|thumb|300px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 4:''' Wild type (grey) and mutant (red) residues for position 97 with hydrogen bonds indicated in yellow.
|}
</figure>

Both variants show the same polar contacts, which are only the intra-backbone hydrogen bonds that stabilize the alpha helix (<xr id="97_mut"/>). The isoleucine is slightly smaller than methionine, but the residue stays uncharged and nonpolar and thus, although it is located directly at the binding interface to ferritin, the mutatiion is neutral. 


=== Thr217Ile ===

<figure id="217_mut">
{| class="colBasic2"
| [[File:217_wt.png|center|thumb|300px]] || [[File:217_mut.png|center|thumb|300px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 5:''' Wild type(grey) and mutant(red) residues for position 217 with hydrogen bonds colored in yellow.
|}
</figure>

While the change from a polar to a non-polar side chain reduces the number of hydrogen bonds at this loop and thus the stability of the loop (<xr id="217_mut"/>), it is not enough to affect the function of the protein, because the loop is not near a binding interface or an essential structural part.

=== Cys282Tyr ===

<figure id="282_mut">
{| class="colBasic2"
| [[File:282_wt.png|center|thumb|300px]] || [[File:282_mut.png|center|thumb|300px]] || [[File:282_clash.png|center|thumb|300px]] || [[File:282_mut2.png|center|thumb|300px]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 6:''' Wild type(grey) and mutant(red) residues for position 282. The disulfide bridge binding partner of C282 is shown in white. Hydrogen bonds are shown in yellow.
|}
</figure>

Although for this mutation, the polar contacts to the neighbouring beta sheet remain unchanged (<xr id="282_mut"/>), it is evident why this mutation is the major cause for hemochromatosis. The replacement of cysteine with tyrosine causes a break of the disulfide bond that connects the two beta sheets of the Ig c1-set domain. Consequently, the structure of this domain is destabilized, which probably inhibits the formation of the HFE,ferritin, beta-micorglobulin complex. 
It is also noteworthy, that SCWRL does not change the conformation of the cysteine binding partner, resulting in a clash of tyrosine and cysteine.

== Comparison of SCWRL and FoldX ==

<figure id="fx_scwrl" style="border-width: 0px">
{| class="colBasic2"
| [[File:53_both.png|center|thumb|300px|V53M: Practically, there is no difference between the rotamers generated by the two programs.]] || [[File:63_both2.png|center|thumb|300px| H63D: In this case, there is also nearly no difference between the placements of the programs. ]] || [[File:97_both.png|center|thumb|300px| M97I: As in the two cases before, the results do not differ significantly.]]
|-
| [[File:217_scwrl2.png|center|thumb|300px| T217I SCWRL: For this mutation, the rotamer of the mutant differs, but this has nearly no effect, since the side chain sticks out of the protein and has no interaction partner inside of the protein. The major difference between the programs in this case is, that FoldX  minimises the protein after a suitable rotamer has been found and thus changes the conformation of the surrounding residues. In this particular case, the minimisation results in a weakening or loss of the polar interactions with neighbouring residues.]] || [[File:217_fx2.png|center|thumb|300px| T217I FoldX: See neighbouring image description.]]
|-
| [[File:282_scwrl.png|center|thumb|300px| C282Y SCWRL: In this case, the two rotamers from SCWRL and FolX do again not differ significantly. SCWRL was not able to resolve the clash between the partner cysteine and the tyrosine.]] || [[File:282_foldx.png|center|thumb|300px| The minimisation step from FoldX resolves the clash between the partner cysteine and the tyrosine.]] || [[File:282_both.png|center|thumb|300px| Superposition of the tyrosine in the SCWRL (green) and FoldX (cyan) mutated structures. Both rotameres are very similar.]]
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Figure 7:''' Pymol visualisation of the mutations performed by SCWRL (green) and FoldX (cyan). The structures were superimposed for better comparison.
|}
</figure>

FoldX was also used to mutate the HFE proteins. The mutated structures from SCWRL and FoldX were superimposed and are compared in <xr id="fx_scwrl"/>

== Energy Comparisons ==

It might be possible to infer whether a mutation has a functional effect or not, if the calculated energies of a protein differ significantly before and after the 3D mutation. But the energies from different force fields should not be compared, because every force field has different parameters and usually a slightly different formula. Therefore, in the following only the energies from one force field are compared.  
<figtable id="energies">
{| class="colBasic2"
! ||  colspan="3" | SCWRL || colspan="5" | FoldX
|-
! Mutant || Energy ||Mutant/WT || Prediction || Energy Mutant || Energy WT || Mutant/WT || Prediction || Actual
|-
| WT	|| 247.94 || 1.0 || - || || || || ||
|-
| V53M || 264.93 || 1.07 || maybe || 164.92 || 164.24 || 1.00 || - || +
|-
| H63D || 245.20 || 0.99 || - || 167.83 || 166.43 || 1.01 || - || +
|-
| M97I || 247.68 || 1.00 || - || 166.08 || 164.84 || 1.01 || - || -
|-
| T217I || 260.18 || 1.05 || maybe || 169.74 || 167.99 || 1.01 || - || -
|-
| C282Y || 389.54 || 1.57 || + || 182.49 || 167.24 || 1.09 || + || +
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>''' Table 3:''' Comparison of the wildtype and mutant energies of the structures created with SCWRL and FoldX. For SCWRL, the energy of the WT protein was computed and compared to the energies given after performing each single mutation. For FoldX, every mutation had a different WT energy that was compared to the energy of the mutant.
|}
</figtable>
To assess the effect of a mutation on a proteins energy relative to its wildtype energy, the ratio between the wild type protein energy and the mutant protein energy was computed. Thereafter, we empirically predicted whether a mutation has a functional effect, based on the energy ratio.
As <xr id="energies" /> shows, the energies between mutant and wildtype structures differ only slightly most of the time. Only in the case of the C282Y mutation, a large relative energy difference can be observed. Thus, it can be concluded that the energy difference yields correct hints at the structural effect of a mutation only in extreme cases.

== Minimisation ==

<figtable id="minimise">
{| class="colBasic2"
! Method || Mutation || Iter. 1 || Iter. 2 || Iter. 3 || Iter. 4 || Iter. 5
|-
! - || WT 
|| -3724.2 || -5003.5 || -5118.4 || -5198.3 ||style="background-color: #66FF66"| -5301.4
|-
! rowspan="5" | SCWRL 
||  V53M || -5022.7 ||style="background-color: #66FF66"| -5295.7 || -5154.7 || -5272.5 || -5260.4
|-
| H63D || -4940.5 ||style="background-color: #66FF66"| -5212.8 || -5084.4 || -5190.7 || -5189.6
|-
| M97I || -5025.3 || style="background-color: #66FF66"| -5291.5 || -5146.5 || -5247.7 || -5246.2
|-
| T217I || -5037.7 || style="background-color: #66FF66"| -5307.9 || -5171.54 || -5277.0 || -5269.3
|-
| C282Y || -2596.7 || -5107.7 || -5037.12 || -5159.1 || style="background-color: #66FF66"| -5191.7
|-
!rowspan="5" | FoldX  
|| V53M || -5323.9 ||style="background-color: #66FF66"| -5544.4 || -5450.0 || -5377.1 || -5436.9
|-
|  H63D || -5284.4 ||style="background-color: #66FF66"| -5493.8 || -5437.6 || -5364.6 || -5454.7
|-
| M97I  || -5264.6 ||style="background-color: #66FF66"| -5482.1 || -5405.7 || -5343.1 || -5255.8
|-
| T217I || -5275.8 ||style="background-color: #66FF66"| -5492.3 || -5416.5 || -5343.9 || -5431.5
|-
|  C282Y || -3376.9 || -5217.0 || -5194.0 || -5231.1 ||style="background-color: #66FF66"| -5290.4
|-
|+ style="caption-side: bottom; text-align: left" |<font size=2>'''Table 4:''' The energies after each minimisation iteration are shown for all mutations and the wildtype. For each minimisation series, the step that yields the lowest energy is marked in green.
|}
</figtable>

The results of the minimisation runs (see <xr id="minimise"/>) nearly all have their lowest energy after iteration 2. The only exception to this are the C282Y mutant structures and the wildtype structure. 
The reason for this probably is, that the local minimum is already reached after iteration 2 and the computation of the gradient at a local minimum points to a higher energy state. The method "overshoots" its target. In the case of two consecutive energy increases, the following reasoning might apply: the minimisation first finds one local minimum that it then leaves to reach a second local minimum. But from the second local minimum, it does not find its way back to the first one, but finds a third one with even higher energy instead.
The wildtype and the C282Y mutant structures probably reach a local minimum only after the 5th iteration. 

When comparing the energies of the mutant structures between the two programs after the first iteration, it stands out, that the energies of the FoldX mutants are always significantly lower than those of the SCWRL mutants. This is due to the additional minimisation step carried out by FoldX. Interestingly, the overall best energy of every FoldX mutant structure is lower than that of the corresponding SCWRL structure.

<figure id="282_min2">
[[File:Min2_282_compare.png|center|thumb|400px| '''Figure 8:''' The conformation of the C282Y mutants is shown after 2 iterations of minimisation. The SCWRL mutant strucure is shown in green and the FoldX structure in cyan.]]
</figure>

<xr id="282_min2"/> shows that the rotamers for the tyrosine and cysteine residues generated by FoldX and SCWRL are nearly identical after minimisation.

In conclusion, it can be said that FoldX produces better initial results than SCWRL, but after minimisation, there is nearly no difference between the two methods.