Structure based mutation analysis of GBA

Introduction

A detailed list of the ten mutations analyzed in this section, can be found in Task 6.

The different tools and the corresponding steps, applied in this analysis are described in this workflow for reasons of clarity.

Structure Selection

To carry out a structure-based analysis of the mutations chosen in Task 7 a crystal structure had to be chosen. According to Uniprot 19 different crystal structures of glucocerebrosidase exist. The table below shows the six different structures with a resolution of or better than 2 Angstrom. 2NT0 is chosen as template for the analysis carried out in this section, as no residues are missing, the R-value is quite low, and it has the best resolution among the structures without missing residues. Only incomplete structures have been resolved near the physiological pH (7.4), therefore a structure resolved at a more acid pH had to be chosen. The structure can either be downloaded from the PDB website or by using the script fetchpdb, which validates the ID and downloads the corresponding structure.

PDB ID	Resolution [Å]	R-factor	Coverage	pH	# Missing Residues (A/B)
1OGS	2.00	0.195		4.6	0
2NT0	1.79	0.181	.	4.5	0
2V3D	1.96	0.157		6.5	9/8
2V3E	2.00	0.163		7.5	7/7
2V3F	1.95	0.154		6.5	8/14
3GXI	1.84	0.193		NULL	0

Mutation Mapping

The ten positions at which the mutations analyzed in this task take place, are hilighted in the structure of 2NT0 shown in Figure 1. As already mentioned in Task 5 and 6, one can clearly see that two mutations are next to the active site residues Glu235 and Glu340, namley the positions 120 and 311. The wildtype residues at these positions (Arg120 and His311) are known to form hydrogen bonds with the active sites and should therefore be quite important for function and structure. <ref>Kim et al., Crystal Structure of the Salmonella enterica Serovar Typhimurium Virulence Factor SrfJ, a Glycoside Hydrolase Family Enzyme. Journal of Bacteriology, 2009, p. 6550-6554, Vol. 191, No. 21 </ref> The other eight mutation positions are located all over the protein.

Figure 1: 2NT0 with hilighted mutation positions (red) and active site residues (blue).

Figure 2: Close-up of active site of 2NT0 with hilighted mutation positions (red) and active site residues (blue).

SCWRL

SCRWL4 was applied ten times, once for each mutation. The resulting conformations of the mutants are visualized in Figure 3. For this representation the hydrogens of the SCWRL mutant structures have been removed to simplify the comparison with the other two structures. Figure 3 additionally shows the wildtype amino acids and the mutants created with the mutagenesis method of pymol. The conformations, created with SCWRL4 and pymol vary greatly. Only in mutation 9 they seem to be quite similar. Figure 4 shows a superposition of the wild type protein and the mutated proteins in cartoon representation. This shows that SCWRL did not only change the mutant residues, but also changed some beta sheets at the bottom of the structure (shown in green). This fact may be due to the different polar interactions of the mutants.

Figure 3: Wildtype amino acids (red) and mutations created with SCWRL (green) and pymol mutagenesis (orange) hilighted on the structure of 2NT0.

Figure 4: Cartoon representation of 2NT0, chain A (gray) superimposed with the resulting structures of SCWRL (green).

The table below shows the residues forming polar interactions with the mutant/wild type amino acid. The polar interactions can be seen in Figure 3 as well, but one may not clearly distinguish, whether an interaction is only formed by e.g. the wild type or by wild type and mutant. To determine the interaction partners, the tool Pymol was used [actions -> find -> polar contacts -> to other atoms in object].

Mutation	Wild Type	Pymol	SCWRL
1	S8	S8 F9	S8
2	K413 Y418	Y22	K413 Y418
3	T471		G62
4	G83 M85 S177 D282 N234 E340	G83 M85	G83 M85 S177 S339
5	S181 L185	S181 L185	S181 L185
6	L268 H273 H274	L268 H273 H274	L268 H274
7	S366 Y373 V375	S366 Y373 V375	S366 Y373 V375
8	D282 D283 R285 S339	D283 R285	D282 R285 Y313 E340
9	L420	L420	L420
10	T482	T482 V468	T482

Energy Comparison

SCWRL

Mutation	1	2	3	4	5	6	7	8	9	10
Minimal Energy	351.329	348.659	350.017	355.416	473.454	364.148	352.615	362.604	354.98	375.976

FoldX

The total energies calculated with FoldX are shown in the table below. The differences between the wild type and the different mutant structures have been calculated and are listed in the table as well.

Mutation	Total Energy	Difference
WT	-372.60	0
1	-225.82	-146.78
2	-228.18	-144.42
3	-226.97	-145.63
4	-226.38	-146.22
5	-196.84	-175.76
6	-224.01	-148.59
7	-228.48	-144.12
8	-217.29	-155.31
9	-221.71	-150.89
10	-218.65	-153.95

Gromacs

Wildtype

Correlation between nsteps and runtime of mdrun

The following table shows the measured time for mdrun for different nsteps and force fields. It also shows the steps it really took to calculate the lowest energy. With AMBER03 and CHARMM27 it took less than 500 steps, so the values for 500, 1000 and 5000 nsteps should be similar. The same you can see in figure 5.

nsteps	AMBER03	CHARMM27	OPLS/AA
10	real: 0m4.615s user: 0m2.910s sys: 0m0.230s Steps: 10	real: 0m10.760s user: 0m2.760s sys: 0m0.150s Steps: 10	real: 0m3.801s user: 0m2.280s sys: 0m0.320s Steps: 10
50	error occured	real: 0m34.086s user: 0m10.830s sys: 0m0.260s Steps: 50	real: 0m13.042s user: 0m9.200s sys: 0m0.620s Steps: 50
100	real: 0m29.400s user: 0m22.060s sys: 0m0.640s Steps: 100	real: 0m23.616s user: 0m19.410s sys: 0m0.590s Steps: 100	real: 0m25.184s user: 0m18.330s sys: 0m0.740s Steps: 100
500	real: 1m9.519s user: 0m53.700s sys: 0m1.010s Steps: 242	real: 3m15.607s user: 1m1.790s sys: 0m1.070s Steps: 307	real: 3m51.264s user: 1m7.870s sys: 0m1.590s Steps: 500
1000	real: 2m9.160s user: 0m52.990s sys: 0m1.190s Steps: 242	real: 3m22.812s user: 1m2.120s sys: 0m1.170s Steps: 307	real: 5m9.851s user: 1m38.390s sys: 0m1.860s Steps: 732
5000	real: 2m13.786s user: 0m53.020s sys: 0m1.030s Steps: 242	real: 3m15.591s user: 1m2.730s sys: 0m1.560s Steps: 307	real: 5m5.631s user: 1m37.240s sys: 0m1.830s Steps: 732

The following plot shows the correlation between nsteps and the runtime of mdrun. We therefore used the real time.

Figure 5: Correlation between nsteps and the runtime of mdrun for different force fields