Gaucher Disease: Task 09 - Structure-based mutation analysis

<css>

table.colBasic2 { margin-left: auto; margin-right: auto; border: 1px solid black; border-collapse:collapse; }

.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>

Preparation

1. Choice of a structure to work with

We chose the structure 2V3E, chain B, which has the following properties:

<figtable id="2V3E">

PDB-ID	Resolution (Å)	Chain	Covered residues (UniProt seq.)	Missing residues (ATOM seq.)	Covered residues (ATOM seq.)	R-Value(obs.)	R-Free	pH	Temperature (K)
2V3E	2.0	A/B	40-536 (92.7%)	A: 31, (498-503), B: (-1), (498-503)	A: -1-30, 32-497, B: 0-497	0.163	0.220	7.5	100

Properties of 2V3E, chain B, the chosen reference structure of GBA sequence P04062.

</figtable>

For more information about other candidates and the missing residues, see the lab journal.

2. Visualization of the mutations to work with

We selected the following five mutation from the mutations selected in for this task:

<figtable id="mutations">

Reference	Codon number (UniProt)	Codon number (PDB)	Codon change	Amino acid change	Polarity	Charge (pH)	Disease causing?
rs368786234	77	38	AGC ⇒ AGA	Ser ⇒ Arg (S77R)	polar ⇒ polar	neutral ⇒ positive	FALSE
rs374003673	141	102	AAT ⇒ AGT	Asn ⇒ Ser (N141S)	polar ⇒ polar	neutral ⇒ neutral	FALSE
CM992894	241	202	GGA ⇒ GAA	Gly ⇒ Glu (G241E)	nonpolar ⇒ polar	neutral ⇒ negative	TRUE
CM880036	409	370	AAC ⇒ AGC	Asn ⇒ Ser (N409S)	polar ⇒ polar	neutral ⇒ neutral	TRUE
CM870010	483	444	CTG ⇒ CCG	Leu ⇒ Pro (L483P)	nonpolar ⇒ nonpolar	neutral ⇒ neutral	TRUE

Selected mutations of GBA sequence P04062. Mapping of the UniProt positions onto the PDB ATOM sequence is given.

</figtable>

The following figure visualizes the five residues we are going to mutate on the reference structure, 2V3E, chain B.

<figure id="pymol_mutations" >

• 

  The five residues to be mutated mapped onto the structure 2V3E, chain B. Those residues are represented as green (not disease causing S77R and N141S) and red (disease causing G241E, N409S and L483P) spheres. The ligands bound to the structure at three binding sites during its resolution (BMA, FUC, NAG and NND) are represented as spheres in different shades of orange. Neither of the five residues lies in the proximity of one of the binding sites.

• 

  Same as the first figure, only view from the other side of the protein.

PyMol visualization of five residues of the structure 2V3E, chain B, we are going to mutate as described in Table 2. </figure>

As can be seen on the image, none of the residues to be mutated lies in the proximity of one of the three binding sites. (See also the second subfigures of the figure galleries in the next subsection for close up on the residues and their hydrogen bonds.) However, four of the residues lie within a secondary structure element (beta sheet or helix) and one - Glycine 241 - in a turn near a helix. This implies that exchange of these residues with others with different functional groups, polarity and charge could lead to destruction of some hydrogen bonds within or between the secondary structures (e.g. Asparagine 141). This might lead to structural changes and even to destruction of the secondary structures or important blocks of secondary structure elements. Moreover, an exchange with a side chain of a bigger size might lead to clashed with proximate residues (e.g. with the loop near the Serine 77).

3. Creation of mutated structures

We used SQWRL4 to create the five mutated structures. (See lab journal.) The mutated residues in comparison to the native residues, the hydrogen of the mutants and possible clashes are shown in the following figures.

S77R

<figure id="S77R" >

• 

  Comparison of wild type (wt) Serine 77 (magenta) and mutant Argenine (orange). The mutated residue has a longer side chain.

• 

  Serine 77 (PDB 38) is colored magenta, its chain is in the stick representation and all the backbone residues are in the cartoon representation (the same for the following subfigures). Serine 77 lies in the middle of a beta strand flanked by two other strands in an anti-parallel order. It forms two hydrogen bonds with a residue in a neighboring strand.

• 

  Hydrogen bonds formed by the mutant Argenine 77 remain conserved, as they are formed by the main chain.

• 

  Spheres representation of the mutant Argenine 77 (orange) shows, that its side chain does not clash with the neighboring loop (gray spheres).

PyMol visualization of the mutation S77R (PDB 38) of the structure 2V3E, chain B, with SCWRL. </figure>

According to this images, we would also predict the mutation S77R as not disease causing.

N141S

<figure id="N141S" >

• 

  Comparison of wild type (wt) Asparagine 141 (magenta) and mutant Serine (orange). The mutated side chain is a little shorter and points towards another direction.

• 

  Asparagine 141 (PDB 102) is located in a helix and forms four hydrogen contacts: two with the main-chain carboxyl- and amino-groups with two helix amino acids (one in each direction) and two with the side chain. One of the contacts of the Asparagine side chain also binds to the same residue in the helix as the Asparagine amino-group binds. The second hydrogen bond of the Asparagine side chain is formed with the side chain of a residue in the neighboring helix. Maybe this bond plays a stabilizing role between the two helices.

• 

  Hydrogen bonds formed by the mutant Serine 141. A contact that was formed between the native Asparagine side chain and a side chain pointing from a neighboring helix (showed in sticks) is missing, as well as a second contact with a residue in the same helix. Instead, the mutated Serine side chain forms a contact with another residue in the same helix.

• 

  Spheres representation of the mutant Serine and the missing contact with the neighboring helix.

• 

  Spheres representation of the mutant Serine and a proximate loop (gray spheres) - there is no clash.

PyMol visualization of the mutation N141S (PDB 102) of the structure 2V3E, chain B, created with SCWRL. </figure>

The mutation N141S is benign, therefore the missing stabilizing contact between the two helices is probably not so important.

G241E

<figure id="G241E">

• 

  Comparison of wild type (wt) Glycine 241 (magenta) and mutant Glutamate (orange). The mutated residue has a long side chain.

• 

  Glycine 241 (PDB 202) lies in a turn in the proximity of a helix end and in front of an another helix. Its amino-group forms a hydrogen bond to a residue of the latter helix.

• 

  The mutant Glutamate 241 maintains the same hydrogen bond (formed by the main chain) with a residue in a neighboring helix. The residue lies on the surface and the bigger side chain points to the outside of the protein.

PyMol visualization of the mutation G241E (PDB 202) of the structure 2V3E, chain B, created with SCWRL. </figure>

As no changes in the hydrogen bonds or clashes etc. can be noticed, we would predict the mutation G241E as benign. However, it is a disease causing mutation. Maybe the effect is caused by the polarity and negative charge of the mutant Glutamate, instead of the nonpolar and neutral Glycine.

N409S

<figure id="N409S" >

• 

  Comparison of wild type (wt) Asparagine 409 (magenta) and mutant Serine (orange). The mutated side chain is a little bit smaller, but points towards the same direction.

• 

  Asparagine 409 (PDB 370) is located in a helix. It builds five hydrogen bonds: four with the main chain and one with the side chain. Two of the main chain contacts are formed by the carboxyl-group: one with a helix residue and one with a residue located in a proximate beta-strand beginning, the two binding partners are separated by a short turn. Two other main chain hydrogen bonds are built by the amino-group and two carboxyl-groups of an amino acid in the helix (from the other side). The fifth bond is formed between the side chain amino-group and one of the caroboxy-groups of the last mentioned helix residue.

• 

  Hydrogen bonds formed by the mutant Serine 409. The five hydrogen bonds described in the figure on the left with the Asparagine 409 contacts are conserved.

PyMol visualization of the mutation N409S (PDB 370) of the structure 2V3E, chain B, created with SCWRL. </figure>

Due to the fact that all hydrogen bonds remain conserved in the mutant and that no clashes etc. could be detected, we would predict the mutation N409S as non-effect. Moreover, the both amino acids, Asparagine and Serine, are both polar and neutral. Nevertheless, this mutation is annotated as disease causing.

L483P

<figure id="L483P" >

• 

  Comparison of wild type (wt) Leucine 483 (magenta) and mutant Proline (orange). The mutant Proline is a cycle amino acid, which has very different properties than the hydrophobic Leucine. Is leads to a shift in the main chain conformation "backwards", so that the beginning of the neighboring strand is shifted.

• 

  Leucine 483 (PDB 444) lies at the very beginning of a beta strand. There is another beta strand in the proximity on the right, which is followed by three others, forming a beta sheet. However, the Leucine does not form any contact with the neighboring strand, as its side chain is hydrophobic and points in another direction. The Leucine forms an interesting hydrogen bond with its amino-group to the side chain of a residue located in the neighboring turn. After this loop another beta-strand follows.

• 

  The mutant Proline 483 does not form a hydrogen bond with the residue in the loop, like the Leucine main chain amino-group does (see the corresponding figure). The reason for this is that Proline lacks a free amino-group due to the cycle formation.

PyMol visualization of the mutation L483P (PDB 444) of the structure 2V3E, chain B, created with SCWRL. </figure>

We would predict the mutation L483P as having an effect - and so it is according to annotations.

Energy comparisons

Lab journal

foldX

We applied foldX for the five mutations to predict new structures with energies.

<figtable id="scwrl_foldx">

	SCWRL			FoldX
Mutant	Energy	Mutant-WT	Predicted effect	Energy Mutant	Energy WT	Mutant-WT	Predicted effect	Observed effect
WT	386.356
S77R	386.706	0.350	FALSE	20.90	23.25	-2.35	FALSE	FALSE
N141S	390.966	4.610	TRUE	28.81	28.84	-0.03	FALSE	FALSE
G241E	394.319	7.963	TRUE	32.30	29.31	2.99	TRUE	TRUE
N409S	391.73	5.374	TRUE	29.08	27.77	1.31	TRUE	TRUE
L483P	424.715	38.359	TRUE	38.40	35.55	2.85	TRUE	TRUE

Energies of the WT structure and the five mutant structures calculated by SQWRL and foldX. FoldX minimizes the energy of the WT structure for each mutation. The energy difference between each mutant and the respective WT structure was calculated. If it is (significantly) positive, the mutation is regarded as effect causing. The observed effect of the mutation is given for comparison.

</figtable>

We decided to regard all differences between the mutant and WT energies higher than 1 as significant. SCWRL identifies the three disease causing mutations as such. However, it also predicts one not disease causing mutation (N141S) as effect causing, thought with the smallest positive difference (higher 1) of 4.61. Interestingly, the most prevalent Gaucher disease mutation according to OMIM, L483P (L444P in OMIM), has a highest energy change of 38.359. FoldX makes a correct prediction in all cases, the two benign mutations even have a small negative energy change. However, the disease causing mutations have all only a small energy increase (1.31 - 2.99). In foldX the mutation L483P does not have the highest energy increase, in the contrary to SCWRL.

Next, we superimposed the SCWRL and foldX structures in Pymol and compared them in the following figures.

<figure id="scwrl_vs_foldx" >

• 

  Mutation S77R. There is almost no difference in the conformation. However, foldX predicts two additional hydrogen bonds between the Argenine side chain: one to a residues in a proximate loop and the second is a big Tyrosine side chain from the same beta strand.

• 

  Mutation N141S. There is almost no difference in the conformation and the same hydrogen bonds are predicted.

• 

  Mutation G241E. In foldX the conformation of the Glutamate side chain is a little bit different, but it has no meaning, as it lies on the outside of the protein. Also the binding partner side chain has a little bit different angle, so that two hydrogen bonds, instead of only one in SCWRL, are formed.

• 

  Mutation N409S. SCWRL and foldX agree in the conformation of the mutated and the binding residues, including the hydrogen bonds formed between them, apart from tiny shifts.

• 

  Mutation L483P. Also here SCWRL and foldX predict nearly the same conformation of the Proline (and the residue which forms a hydrogen bond in WT with the Leucine).

Comparison between the mutants created with SCWRL (gray protein, orange mutated residue, yellow polar contacts) and foldX (lime protein, cyan mutated residue, dark cyan polar contacts). </figure>

Minimise

We applied minimise for all mutant structures produced by SCWRL and foldX and the WT structure. We used the output of one minimisation for another run as input 5 times for each structure. The <xr id="minimise"/> summarizes the resulting energies:

<figtable id="minimise">

Method	Mutation	Round 1	Round 2	Round 3	Round 4	Round 5
-	WT	-12360.0	-11661.9	-12079.7	-11636.9	-11630.6
SCWRL	S77R	-3482.3	-12111.1	-11724.2	-11509.3	-11260.2
	N141S	-3593.4	-12199.7	-11787.7	-11568.1	-11328.3
	G241E	-3548.4	-12163.9	-11792.4	-11593.1	-11381.2
	N409S	-3631.0	-12214.2	-11801.9	-11594.4	-11347.0
	L483P	-3469.4	-11900.1	-11610.9	-11665.8	-11330.6
foldX	S77R	-12297.0	-11892.1	-12037.2	-11707.4	-11703.1
	N141S	-12359.7	-11961.2	-11983.4	-11550.9	-11515.9
	G241E	-12356.6	-11944.8	-12103.6	-11782.5	-11764.0
	N409S	-12389.9	-11966.0	-12117.7	-11845.4	-11808.9
	L483P	-12335.4	-11925.2	-12096.9	-11771.4	-11771.3

Energies produced by each of five consecutive runs of the program minimise, applied on the WT structure and the five mutant structures calculated by SQWRL and foldX. The lowest energies for each structure are marked gold.

</figtable>

For the WT, the lowest energy is reached after the first iteration of minimise. This is because the structure is already correct and therefore optimal. Interestingly, the lowest energy of the SQWRL mutants is reached after two minimise iterations, whereas for the foldX mutants the minimal energy is reached already after the first iteration. This may be explained by the fact that foldX already performs a minimisation step. Moreover, foldX minimal energies are always lower, than those calculated by SCWRL.

Another interesting observation is that after reaching the lowest energy in the second minimise iterations in the WT and the foldX mutants, the energy rises constantly in the consequent runs. This probably means that the minimal local energy is reached in the second iteration and then the computation of the gradient at the minimal point can only lead to a higher energy value, maybe to another local minimum with a higher energy. A different pattern happens during the minimisation of the SCWRL mutants: after reaching the local minimal energy already after the first iteration, in the second iteration the energy rises, however it falls a little again in the third iteration, thought not into the same local minimum as in the first iteration. This indicates that a second "suboptimal" local minimum is found. After the forth iteration the energy rises again.

We compared the mutant structures in Pymol again after the minimise runs reaching the best energies: after the second round for SQWRL and the first round for foldX (<xr id="minimise"/>).

<figure id="minimise" >

• 

  Mutation S77R.

• 

  Mutation N141S.

• 

  Mutation G241E.

• 

  Mutation N409S.

• 

  Mutation L483P.

Comparison between the mutants created with SCWRL after two minimise iterations (gray protein, orange mutated residue) and foldX after one minimise iteration (lime protein, cyan mutated residue). There is almost no difference in the conformations of all the mutations. </figure>

To conclude, SQWRL and foldX are both pretty good tools for calculating the energy difference and the structural change of a protein after a point mutation. There is almost no difference in the resulting structures between the two programs, in particular the predicted conformations of the mutated residues and their environments are very similar.