Structural investigation on the electrostatic loop of native and mutated SOD1 and their interaction with therapeutic compounds

Abstract

A β-sheet cytosolic enzymatic protein, [Cu–Zn superoxide dismutase 1, (Cu–Zn SOD1)] not only neutralizes superoxide radicals but also catalyses their disproportionation into hydrogen peroxide and molecular oxygen. The structure and activity of the SOD1 protein depend on two important loops, viz. Zn-binding and an electrostatic loop [E-loop]. In this work, the electrostatic loop of the SOD1 protein in the native and mutated states with single point mutation in the loop, viz. D124V, D125H and S134N, is subjected to molecular dynamics simulation for a period of 50 000 ps. The mutated E-loops are docked with four different chemical compounds, viz. 5-fluorouridine, isoproteranol, 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline and uridine-5-monophosphate. The structures with larger binding affinity are again subjected to molecular dynamics simulation for a period of 50 000 ps. The structure and electrostatic properties of the mutated E-loops before and after interacting with the chemical compounds are compared with those of the E-loop in the native state. It is found that the chemical compounds reduce the effect of the E-loop in the mutated SOD1 proteins by reorienting the mutated E-loop in such a way that it exhibits structural and electrostatic properties close to those of the E-loop in the native state. Among the different chemical compounds of interest, uridine-5-monophosphate is found to exhibit better interaction, which helps to reduce the rapid aggregation of the SOD1 protein due to Familial Amyotrophic Lateral Sclerosis [FALS] mutations. Molecular dynamics simulation for the mutated E-loops generates an ensemble of configurations, and the conformational sampling in these configurations is examined using the free energy profile.

---

Introduction

One of the most invariably fatal and progressive motor neuron diseases [MNDs] is Amyotrophic Lateral Sclerosis [ALS], which occurs due to degeneration of the cortical motor neurons in the brain stem and spinal cord resulting in paralysis and respiratory depression, and causes death within 2–5 years.^1,2 The only available compound for the treatment of ALS is riluzole, capable of providing a 9% chance of survival for 1 year after the initiation of treatment.³ The first genetic link to the familial form of ALS was reported in 1993 as mutations in the gene encoding the Cu–Zn superoxide dismutase.¹ Dismutation of a charged substrate, the superoxide anion, is found to be catalysed by the metalloenzyme Cu–Zn SOD1.⁴ It is a homodimeric antioxidant enzyme consisting of 153 amino acids where each monomer has one Cu²⁺ and one Zn²⁺ ion.⁵ The electrostatic loop produced by the charged residues is important for a deeper understanding of the catalytic mechanism.⁶ The structure and activity of the enzyme is based on two important functional loops.⁷ One such loop is the loop VII or the electrostatic loop involving the charged and polar residues 121–144.⁷ These residues are found to enhance the enzymatic function by providing electrostatic guidance to the superoxide (O₂⁻) substrate towards the active site centred around the Cu²⁺ ion. It has been found that the catalytic properties and the enzymatic behaviour of the SOD1 protein are not only influenced by the electrostatic effect of the charged residues but also by the structure of the electrostatic loop and the location of the charged residues.⁶ A positive electric field generated by the metals and the charged residues of the electrostatic loop surrounded by the active-site channel leads to diffusion of the negatively charged substrate towards the catalytic metal.^8–12 It has been found that modifications in the electrostatic loop of a mutated Cu–Zn SOD1 protein lead to a change in SOD1 activity.¹⁴ The structural and dynamic changes affect the electrostatic loop when the E-loop is subjected to site-directed mutagenesis.¹³ The electrostatic loop perturbation represents ∼50% of the SOD1 perturbation.¹³ Disorder in the electrostatic loop is associated with increased intermolecular interactions, responsible for fibril formation.^13,15 So far, the biological relevance of the E-loop is not fully elucidated.

In this work, we studied the electrostatic loop in its native state along with three single point mutations in the E-loop. The selected three mutations are D124V, D125H and S134N, which are selected because these three residues are found to be directly linked to the metal-binding region. The D124V mutation is selected because the residue D124 is found to be essential for the integrity of the metal-site structure.¹⁶ It is referred to as second-sphere ligand of Cu²⁺ and Zn²⁺ ions, because the D124 residue forms hydrogen bonds with the copper ligand H46 and zinc ligand H71.¹⁶ The structure of E-loop and the hydrogen bond network is dramatically changed⁶ even if the total charge is maintained due to replacement of negatively charged aspartic acid residue with a neutrally charged valine. For the D125H mutation, aspartic acid is found to be a negatively charged residue, which is replaced by the positively charged histidine residue. Here, the zinc coordination geometry is displaced to a certain distance from the plane formed by the nitrogen atoms of three histidine ligands.¹⁷ For the S134N mutation, Elam et al. found significant structural disorder due to non-native contacts between the electrostatic loop of one domain (monomer A) and an exposed cleft made by the β-strands V and VI of a neighbouring domain (monomer B).¹⁵ The S134N mutation leads to major structural and dynamical changes in the E-loop, which result in non-native interactions leading to aggregation.^13,14 The asparagine at the 134^th position is found to be involved in the hydrogen bonding network which links the E-loop with the Zn-binding loop, and its replacement with serine leads to decreased metallation.¹⁸ The S134N mutated SOD1 crystallizes into filament-like arrays with increased intermolecular interactions and deprotection of the exposed β-strands.^9,13 A patient with a S134N mutation will have an average life expectancy of less than three years after the onset of the disease.

The E-loop in native WT-SOD1 [PDB code: 1HL5] is shown in Fig. 1, where the loop is highlighted in magenta indicating the position of the E-loop in the WT-SOD1 protein and the residues of the loop are marked in the enlarged image in Fig. 1. The orientations of the E-loop in mutated SOD1 with the single point mutations, viz. D124V, D125H, S134N, are given in Fig. S1† and the mutated E-loops are highlighted in magenta, while the residues of the loop are shown in the enlarged mutated E-loops. The residues of the E-loops in the native and mutated SOD1 proteins, with single point mutation on the E-loop, are subjected to molecular dynamics analysis for a time period of 50 000 ps. We initiated protein–ligand interactions by interacting the mutated E-loops with the following chemical compounds: isoproteranol [5FW], 5-fluorouridine [5UD], uridine-5-monophosphate [U5P] and 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline [MDTQ]. 5-Fluorouridine is a fluorinated analogue of uridine monophosphate, but lacks the phosphate group at the 5′ position. It has been expected to bind between zinc and the electrostatic loop.^3,19 Isoproteranol is structurally related to the neurotransmitters dopamine and epinephrine.¹⁹ Hence, fluorouridine and isoproteranol are considered as SOD1 stabilizers.¹⁹ Uridine-5′-monophosphate is found to be directly bound to SOD1 by its pyrimidine base, and there is no direct contact between the protein and ribosyl group.³ MDTQ is patented as a modulator of ALS.²⁰ It is composed of a quinazoline ring system with a homopiperazine group connected at C4 and trifluoromethyl at C2.²⁰ The structures of the various chemical compounds are shown in Fig. S2.† The above-mentioned chemical compounds were made to interact with the mutated E-loops using docking techniques. Among the various conformers obtained through the docking studies, the conformers with higher binding affinity and stability are again subjected to molecular dynamics simulation for a period of 50 000 ps. The docking structures for the three different mutations with uridine-5′-monophosphate are shown in Fig. 2, since this compound shows higher binding affinity with the mutated structures. The docking structures of the remaining chemical compounds are shown in Fig. S3.† The hydrogen bonds formed between the ligand and the mutated E-loops in the docking structures are listed in Table 1. The structural disorder of the E-loop has been studied using root-mean-square deviation, the flexibility vs. rigidity, end-to-end distance, electrostatic potential, hydrogen bond analysis, and free energy analysis. Analyses were performed based on structural parameters, charge distribution, the conformational variation and loss of enzymatic activity of the mutated E-loops, which in turn contribute to the toxicity and misfolding of ALS-associated SOD1.

Fig. 1 Native WT-SOD1 protein with the E-loop coloured in magenta. The enlarged E-loop is shown on the left side as a magenta coloured ribbon labelled with the residue names and numbers.

Fig. 2 The best docking structures of the mutated E-loops interacting with uridine-5′-monophosphate [U5P]: (a) D124V + U5P, (b) D125H + U5P, and (c) S134N + U5P.

Table 1 Atoms involved in hydrogen bonding between the mutated E-loops and ligands with their corresponding bond lengths

Complexes	Atoms involved in H-bond formation	H-bond lengths
D124V + 5UD [3H2P + 5-fluorouridine]	GLU 121 H2⋯5UD O2	1.856 Å
	SER 142 O⋯5UD HA	2.104 Å
D124V + 5FW [3H2P + isoproteranol]	ALA 140 O⋯5FW HAB	2.341 Å
	GLU 121 H2⋯5FW OAA	2.295 Å
	LYS 122 O⋯5FW HAA	1.810 Å
	LYS 122 O⋯5FW HAD	1.968 Å
D124V + MDTQ [3H2P + 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline]	GLU 121 H2⋯MDTQ N3	2.216 Å
D124V + U5P [3H2P + uridine-5-monophosphate]	GLU 121 H2⋯U5P O1P	2.321 Å
	GLU 121 H2⋯U5P O3P	2.239 Å
	GLY 141 O⋯U5P HB	2.456 Å
	SER 142 O⋯U5P HB	2.020 Å
D125H + 5UD [1P1V + 5-fluorouridine]	ARG 143 HH21⋯5UD O4	2.694 Å
	ARG 143 HE⋯5UD O4	1.835 Å
	ALA 140 H⋯5UD O2	1.844 Å
	GLY 141 H⋯5UD OA	2.237 Å
	ASN 139 OD1⋯5UD HA	1.880 Å
	THR 137 HG1⋯5UD O4′	1.840 Å
D125H + 5FW [1P1V + isoproteranol]	LYS 128 O⋯5FW HAA	2.070 Å
	LYS 128 O⋯5FW HAD	2.168 Å
	GLY 138 H⋯5FW OAB	2.120 Å
D125H + MDTQ [1P1V + 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline]	GLU 121 H2⋯MDTQ N3	2.337 Å
D125H + U5P [1P1V + uridine-5-monophosphate]	LYS 136 H⋯U5P O1P	2.457 Å
	THR 137 H⋯U5P O1P	2.325 Å
	THR 137 HG1⋯U5P O4′	2.089 Å
	ASN 139 OD1⋯U5P HA	2.079 Å
	ARG 143 HH21⋯U5P O4	2.026 Å
S134N + 5UD [1OZU + 5-fluorouridine]	ASN 131 O⋯5UD HA	1.950 Å
	ASN 139 HD22⋯5UD OG	2.109 Å
S134N + 5FW [1OZU + isoproteranol]	THR 137 HG1⋯5FW OAA	2.170 Å
	ASN 139 OD1⋯5FW HAD	2.393 Å
S134N + MDTQ [1OZU + 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline]	ARG 143 HH21⋯MDTQ N1	2.160 Å
	ARG 143 HE⋯MDTQ N1	2.470 Å
S134N + U5P [1OZU + uridine-5-monophosphate]	ALA 140 H⋯U5P O2	2.093 Å
	THR 137 HG1⋯U5P O4′	1.805 Å
	THR 137 H⋯U5P O1P	2.585 Å
	LYS 136 H⋯U5P O1P	1.849 Å
	ARG 143 HE⋯U5P O4	1.866 Å
	ARG 143 HH21⋯U5P O4	2.513 Å

---

Materials and methods

The crystal structure of the E-loop of human SOD1 in its native state used in our simulation is taken from the protein databank²¹ [PDB code: 1HL5].²² Similarly, the crystallographic X-ray structures of the E-loop with single point mutation are found to possess the following PDB codes 3H2P, 1P1V, and 1OZU^15,17,23 for the D124V, D125H and S134N mutations, respectively.

Docking method

To dock the ligands with the E-loops of the mutated SOD1 proteins, the PDBQT files for the E-loop with single point mutation and the ligands were prepared using the graphical user interface program AutoDock Tools 1.5.6. The Broyden–Fletcher–Goldfarb–Shanno [BFGS] method in AutoDock Vina has been employed for local optimization.²⁴ Reliable results were obtained when the exhaustiveness of the global search is set equal to 100. The maximum and minimum energy difference between the best binding modes differs for the different ligands. We have chosen the centre of mass of the binding site residues as the grid centre to cover all the possible space to bind the ligand. The centre of the grid is placed at the centre of mass of the four different ligands, with a grid spacing of 1 Å. The box dimensions and centre of mass coordinates are found to be different for the three mutated E-loops, viz. D124V, D125H and S134N of SOD1 docked with different ligands. The box dimensions and centre of mass coordinates for the D124V mutated E-loop docked with 5-fluorouridine, isoproteranol, 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline and uridine-5-monophosphate are size_X = 15.00 Å, size_Y = 18.75 Å and size_Z = 15.00 Å, and center_X = 3.79, center_Y = 34.81 and center_Z = 36.77, respectively. Similarly, the box dimensions and centre of mass for the D125H mutated E-loop docked with 5-fluorouridine, isoproteranol, 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline and uridine-5-monophosphate are size_X = 22.50 Å, size_Y = 18.75 Å and size_Z = 20.62 Å, and center_X = 10.62, center_Y = 31.27 and center_Z = 33.71, respectively. In the same way the box dimensions and centre of mass for the S134N mutated E-loop docked with 5-fluorouridine, isoproteranol, 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline and uridine-5-monophosphate are size_X = 22.50 Å, size_Y = 22.88 Å and size_Z = 20.25 Å, and center_X = 4.10, center_Y = 40.17 and center_Z = 41.19, respectively. These dimensions are found to be large enough in order to cover the E-loop along with the ligand. The grid spacing is sufficient enough to sample all possible conformations inside the binding pocket. With random starting positions of the mutated E-loops and ligands, 10 most energetically favourable binding positions are generated. The atoms in the E-loop with single point mutations are set to flexible degrees of freedom. Strong binding modes for the different ligands with the different single point mutated E-loops were employed for subsequent molecular dynamics simulation.

Molecular dynamics simulation

Molecular Dynamics [MD] simulations for the SOD1 protein and protein–ligand complexes were performed for 50 000 ps. The force-field used is GROMOS96 43a1.²⁵ This force-field demonstrates the usefulness in studying the folding and aggregation of peptides.^24,26–29 All the systems are solvated using the Single Point Charge Extended [SPC/E] water model because the SPC/E model is known to yield better pure solvent properties.^30–32 The residues of the E-loop in its native state, and those of the E-loops with single point mutations and the single point mutated E-loops interacting with chemical compounds, are placed in the centre of the cubic box with a diameter of 1 nm, under periodic boundary conditions in the X, Y and Z directions. The different solvated systems are then subjected to energy minimization, where all the atoms of the different systems are allowed to relax in the last step of energy minimization. The energy minimization is then followed by NVT and NPT equilibration for a time period of 100 ps. A leapfrog integrator is used to integrate the equations of motion for NVT and NPT equilibration with a time step of 2 fs. The bonds between the atoms in the protein and the chemical compounds are constrained to their equilibrium values using the LINCS algorithm.³³ The bond lengths and bond angles of water molecules are constrained using the SETTLE algorithm.³⁴ In order to calculate Lennard-Jones interactions, a cut-off of 1.4 nm is used. The particle-mesh Ewald method [PME] is used to treat long-range electrostatic interactions.³⁵ A cut-off of 0.9 nm is used for evaluating real space interactions, while reciprocal space interactions are evaluated on a 0.16 nm grid with cubic interpolation. All the simulations are performed with parrinello–Rahman coupling³⁶ of 1 atm in an isotropic ensemble. Hence, the proteins and ligand molecules are kept unconstrained throughout the simulation. For the different systems, the isothermal compressibility is set to 4.5 × 10⁻⁵ bar, with a pressure coupling time of 2 ps throughout the simulation. The temperature is controlled at 300 K by a V-rescale thermostat.³⁷ The coupling time to couple the system with the temperature bath is 0.1 ps. In the production process, the MD integrator is used with an integration time step of 2 fs, while a time interval of 2 ps is used to save the trajectories for further analysis. MD simulations were carried out using the GROMACS 4.5.5 simulation package.³⁸ The GROMACS suite of programs is used for the analysis of the trajectories.

Results and discussion

Root-mean-square deviation [RMSD]

Fig. 3 represents the RMSD for the backbone atoms of the E-loop in the native and mutated states. Similarly, Fig. 4 represents the RMSD for the backbone atoms of the mutated E-loops interacting with the chemical compounds, viz. U5P and MDTQ, and that of the native state E-loop. The RMSD values for the backbone atoms of the mutated E-loops interacting with 5UD and 5FW, along with those of the native state E-loop, are shown in Fig. S4.† The RMSD values calculated for the backbone atoms of the E-loop in the native state can be considered as important benchmark for the measurement of convergence of the E-loops with three single point mutations. The deviation of the RMSD values for the D124V, D125H and S134N mutations from the native state of the E-loop describes the loss of stability with the single point mutations. The stability of the E-loops with the single point mutations D124V, D125H and S134N is in the following order D124V-mutated E-loop > S134N-mutated E-loop > D125H-mutated E-loop. In the case of the E-loop with D124V mutation, the RMSD values are found to be equivalent to the RMSD values of the native state E-loop in the time period of 0 to 10 000 ps. After 10 000 ps, the RMSD values decrease till 33 000 ps. The RMSD values of the D124V-mutated E-loop are in close agreement with the RMSD values of the native state E-loop during 33 000 to 37 000 ps, and again get reduced till the end of simulation. In the case of the D125H mutation, the structural deviation of the backbone atoms rises abruptly after 17 000 ps, reaching a maximum value of around 0.95 nm until the end of the simulation. Similarly, the S134N-mutated E-loop shows an increase in RMSD values of the backbone atoms from the native state after 4000 ps until the end of the simulation, which attains a maximum deviation of 0.91 nm. The structural disorder of the mutated E-loops leads to rapid conformational shifting accompanied by the loss of functional activity of the E-loops. When the stability and functional activity of the E-loop is lost, the electrostatic guidance for the superoxide substrate towards the catalytic copper site will be affected thereby reducing the enzymatic activity of the SOD1 protein due to the mutation in the E-loop. The Cu–Zn SOD1 protein depends on the electrostatic make up around the active-site channel to maximize its superoxide turnover rate.³⁹ The decrease in conformational stability of the E-loop with single point mutations contributes to the toxicity associated with FALS.⁴⁰ Structural distortions of the E-loops with D125H and S134N single point mutations partially affect the Zn-binding loop, leading to rotation of the catalytically important residue Arg 143 away from the active site. The D124V single point mutated E-loop shows a small structural variation from the native state E-loop, which accounts for the loss of native coordination of the Zn²⁺ ion and disorderliness has been found in the Zn-binding loop.⁴¹ The hydrogen bonds formed by D124 are lost due to the mutation of aspartic acid to valine leading to incorrect orientation of the D124V-mutated E-loop.⁴²

Fig. 3 RMSD for the backbone atoms of the E-loop in the native and mutated states.

Fig. 4 RMSD for the different mutated E-loops interacting with uridine-5-monophosphate [U5P] and 4-(4-methyl-1,4-diazepan-1-yl)-2-(trifluromethyl) quinazoline [MDTQ] along with that of the native state E-loop: (a) the D124V-mutated E-loop interacting with MDTQ and U5P, (b) the D125H-mutated E-loop interacting with MDTQ and U5P, and (c) the S134N-mutated E-loop interacting with MDTQ and U5P.

In the D125H single point mutated E-loop, the exchange of the negatively charged aspartic acid to histidine indicates a decrease in the enzymatic rate of the mutated E-loop than that of the native state E-loop. In the native state, the E-loop residues 132–137 are found to be in an ordered fashion and possess a consistent conformation.⁴³ The conformational shift of the E-loop due to S134N single point mutation leads to a structural rearrangement of the residues 132–137 within the mutated E-loop on a picosecond time scale. The S134N single point mutation did not affect the hydrophilic side-chain interaction of the loop because serine, a hydrophilic residue, is replaced by another hydrophilic residue, asparagine. The significant and adverse structural distortion of the S134N-mutated E-loop is responsible for the diffusion pathway of an anionic substrate towards the Cu²⁺ binding site. When the D124V-, D125H- and S134N-mutated E-loops interacted with the chemical compounds, the structural distortion was found to be less when compared with that of the free, mutated E-loops. The chemical compounds 5FW, 5UD, U5P, & MDTQ are effective in bringing the structure of the mutated E-loops towards the standard orientation of the native state E-loop by enhancing the crystal packing interactions of the residues in the mutated E-loop structures. The interaction of chemical compounds with the single point mutated E-loop increases the anionic affinity of the electrostatic loop channel, and recognizes the superoxide anion substrate towards the catalytic copper centre. The intermolecular interaction can also be reduced to a certain extent by interacting with the chemical compounds, thereby preventing protein oligomerization. When the E-loop is disordered, it allows the SOD1 dimers to combine, producing amyloid-like linear or helical filaments.³ The chemical compounds 5FW, 5UD, U5P and MDTQ are found to be effective in blocking aggregation and altering the disease course caused due to mutations in the E-loop of SOD1.

Flexibility vs. rigidity

In Fig. 5 the level of fluctuation and dynamic behaviour of the E-loops in the native and mutated states are represented by spectra with the corresponding eigenvalues. The spectrum with higher fluctuation corresponds to the flexibility of the E-loop, while the spectrum with reduced fluctuation corresponds to the rigidity of the E-loop in the native and mutated states. The spectral values ranging from 0.004 to 0.16 nm and 0.01 to 0.26 nm correspond to the rigidity and flexibility of the E-loop in the native state. In the case of the D124V-, D125H- and S134N-mutated E-loops, the spectral values from 0.01 to 0.25 nm, 0.02 to 0.21 nm, and 0.03 to 0.31 nm correspond to the rigidity, while the spectral values from 0.07 to 0.32 nm, 0.04 to 0.6 nm and 0.04 to 0.8 nm correspond to the flexibility of the above mutated E-loops, respectively. The rigidity of the mutated E-loops when compared with that of the native state E-loop is in the following order S134N < D124V < D125H. Due to a loss of compactness, the mutated E-loops possess the ability to attain expandable and flexible conformations when compared with the conformation of the native state E-loop.⁴⁴ The expanded conformations of the mutated E-loops are attributed towards more conformational breathing motion of the E-loop⁴⁵ as a result of which a large scale structural rearrangement arises. The flexibility of the mutated E-loops is in the following order D124V < D125H < S134N when compared with that of the native state E-loop. When the single point mutated E-loops become more flexible, this may affect the catalytic function of the protein, because the flexible structure is prone to less interaction with neighbouring residues. In order to function properly, the protein molecules require both flexibility and rigidity. The significant changes in the flexibility of the mutated E-loops arise because interactions of the solvent molecules enhance the steric constraints. The increase in flexibility of the D124V-, D125H- and S134N-mutated E-loops alter the E-loop geometry. The balance between flexibility and rigidity of the D124V, D125H and S134N mutations leads to variations in the entropy–enthalpy compensation mechanism.⁴⁶ The perturbation in flexibility and rigidity of the mutated E-loop is largely accompanied by changes in the hydrogen bond network. Hence, the altered rigidity and flexibility can also play a role in the pathogenesis of FALS.

Fig. 5 Flexibility and rigidity of all atoms in the native and mutated E-loops for a period of 50 ns: (a) flexibility vs. rigidity for the native E-loop, (b) flexibility vs. rigidity for D124V, (c) flexibility vs. rigidity for D125H, and (d) flexibility vs. rigidity for S134N.

End-to end distance

Partial unfolding of the E-loops with single point mutation is measured by end-to-end distance analysis, as shown in Fig. 6. This involves the distance between the Cα atoms of residues 121 and 144. The increase in distance between the Cα atoms of residues 121 and 144 in the mutated E-loop leads to the detachment of the E-loop from the β-sheet structure. The conformational variation in the native state and the mutated E-loops D124V, D125H, and S134N are shown in the graphs in Fig. 6. The unfolding of the E-loop in the D124V-mutated structure occurs after 10 000 ps till 33 000 ps at a faster rate. The misfolding rate decreases from 34 000 ps to 37 000 ps. After 38 000 ps, the D124V-mutated E-loop is found to misfold rapidly till the end of the simulation. In the case of the D125H mutation, the rapid misfolding of the E-loop occurs from 10 000 ps till the end of the simulation. In the S134N mutation, the hydrophilic residue serine is replaced by another hydrophilic residue, asparagine, which accounts for the reduced misfolding rate of the mutated SOD1 protein. In the unfolded mutated E-loops, the end-to-end distance is found to be deviated from that of the native state E-loop. The trajectories of the partially unfolded mutated E-loops for 50 000 ps illustrate the variation in the thermodynamics of the E-loop. The average distance between glutamine 121 and leucine 144 in the mutated E-loops failed to illustrate the binding affinity of the loop with the β-strands VII & VIII. The detachment of the N- and C-termini of the mutated E-loops from the β-strands leads to enhanced exposure of hydrophobic residues compared to the native state E-loop; as a consequence, the tendency for aggregation of the SOD1 protein is increased. The variation in the Cα–Cα distance affects the stiffness of the E-loop with D124V, D125H, and S134N mutations, because the Cα–Cα distance depends on the number of peptide units that constitute the E-loop.

Fig. 6 Time dependencies of the end-to-end distances of the three mutated E-loops for a trajectory of 50 ns: (a) end-to-end distance for the D124V-mutated E-loop compared with that of the native E-loop, (b) end-to-end distance for the D125H-mutated E-loop compared with that of the native E-loop, and (c) end-to-end distance for the S134N-mutated E-loop compared with that of the native E-loop.

Electrostatic potential

Fig. 7 represents the spatial distribution of the electrostatic potential across the box in the native and mutated E-loops and the mutated E-loops interacting with uridine-5-monophosphate [U5P]. The electrostatic potentials for the mutated E-loops interacting with other chemical compounds are given in Fig. S5.† The single point mutations D124V, D125H, and S134N are found to result in different electrostatic potentials of the loop, causing the overall charge of the loop to become more negative than that of the native state E-loop. The electrostatic potential is calculated by dividing the box into a number of slices and summing up the charges per slide, which is then followed by double integration of the charge distribution. When the mutated E-loop interacts with uridine-5-monophospahte, the electrostatic potential is positive and nearer to the value of the electrostatic potential of native state. Hence, U5P is effective in favouring the electrostatic interactions between the charged residues of the mutated E-loop, and also reorients the distorted conformation of the mutated E-loops towards the native state conformation, when compared with other chemical compounds. For both the D124V- and D125H-mutated E-loops interacting with 5-fluorouridine and isoproteranol, the electrostatic potential is found to be slightly greater than that of the mutated E-loops. The D124V- and D125H-mutated E-loops interacting with MDTQ are found to have a reduced electrostatic potential than that of the native state. The S134N-mutated E-loop, interacting with 5UD, has a higher electrostatic potential than that of the S134N-mutated E-loop. The chemical compounds 5FW and MDTQ interacting with the S134N-mutated E-loop exhibit a reduced electrostatic potential than that of the S134N-mutated E-loop. The loss of electrostatic potential in the mutated E-loops not only destabilizes the stable conformation of the loop but also affects the pairing of the residues in E-loop with the β-strands.⁴⁷

Fig. 7 Spatial distribution of the electrostatic potential across the box. (a) Electrostatic potentials for the native and mutated E-loops, and (b) electrostatic potentials for the native and mutated E-loops before and after interacting with uridine-5-monophosphate [U5P].

U5P enhances the catalytic activity of the D124V-, D125H- and S134N-mutated proteins by increasing the electrostatic interactions between the charged residues of the loop. The increase in electrostatic potential possesses the tendency to screen the hydrophobic regions of the loop, and hence aggregation of the SOD1 protein associated with the FALS mutations D124V, D125H and S134N in the E-loop, is reduced by the interaction of U5P with the mutated E-loop.⁴⁸ U5P enhances the side-chain interactions of charged residues, thereby allowing the mutated E-loops to be able to bind to metal ions. As a result of this, the SOD1 protein leads to proper folding and stability. The weak electrostatic potential due to reduced electrostatic interactions in the D124V-, D125H- and S134N-mutated E-loops affect the ionic atmosphere of the SOD1 protein, leading to a separation of the E-loop from the protein structure. The mutated E-loops interacting with U5P result in an amplified electrostatic potential which contributes to greater binding energies of the charged peptide units in the E-loop.⁴⁹ The significant change in electrostatic potential of the mutated E-loop after interacting with U5P allows intermolecular interactions of the loop through redistribution of electrons in the local electric field.⁵⁰ The negative electrostatic potential in the D124V-, D125H- and S134N-mutated E-loop can result in attraction of protons in the loop by the concentrated electron density of the loop.⁵⁰ When the charge distribution in the electrostatic loop is changed, the structure of the loop and the hydrogen bond network is dramatically modified even if the total charge is maintained.

Interaction energies

The interaction energies for the D124V-, D125H- and S134N-mutated E-loops and uridine-5-monophosphate are given in Fig. 8, while the interaction of the mutated E-loops with other chemical compounds is shown in Fig. S6.† The non-bonded interaction energy is composed of the van der Waals interaction energy and electrostatic interaction energy.⁵¹ The van der Waals interaction energy is given by the Lennard-Jones potential and the electrostatic interaction energy is given by the coulombic potential.⁵¹ The interaction energy is calculated using the module g_energy in GROMACS:⁵¹

U_Coul(r_ij) = q_iq_j/4πε₀r_ij, 1/4πε₀ = 138.935485 (3)

where, σ_ij = the average diameter of the two atoms, and ε_ij = a constant characteristic of the two atom types.

Fig. 8 Interaction Energies between the mutated E-loops and uridine-5-monophosphate [U5P]: (a) the D124V-mutated E-loop interacting with U5P, (b) the D125H-mutated E-loop interacting with U5P, and (c) the S134N-mutated E-loop interacting with U5P.

The interaction energy of D124V mutated E-loop interacting with 5UD is −270 kJ mol⁻¹ at 4000 ps. At time interval between 8800 ps to 10 800 ps there is no interaction between protein and ligand. In case of D124V mutated E-loop interacting with 5FW, the interaction energy is −187 kJ mol⁻¹ at 2200 ps. Absence of interaction has also been noticed between the mutated E-loop and ligand at the time interval between 37 000 ps and 38 000 ps. The interaction energy is −240 kJ mol⁻¹ around 44 000 ps and −335 kJ mol⁻¹ around 42 000 ps for MDTQ and U5P respectively. The interaction energy between various chemical compounds for the above mutated E-loop are in the following order U5P + D124V > 5UD + D124V > MDTQ + D124V > 5FW + D124V. The electrostatic loop with D125H single point mutation interacting with 5UD exhibits the average interaction energy of −260 kJ mol⁻¹ from 23 000 ps till at the end of the simulation. Here the van der Waals interaction plays significant role than electrostatic interaction. The interaction between 5FW and D125H mutated E-loop, the interaction energy is −228 kJ mol⁻¹ near 8000 ps. The nature of interaction is found to be electrostatic. The interaction energy for MDTQ and the specified E-loop is found to be reduced up to 30,000 ps reaching to a minimum of −295 kJ mol⁻¹. Here, the van der Waals interaction is found to be predominant.

The electrostatic interaction between U5P and the D125H mutated E-loop results in interaction energy of −385 kJ mol⁻¹ around 27 500 ps. The interaction energy for the chemical compounds with D125H mutated E-loop is found to be in the following order U5P + D125H > MDTQ + D125H > 5UD + D125H > 5FW + D125H. When S134N mutated E-loop is interacted with 5UD, the interaction energy is −380 kJ mol⁻¹ around 11 500 ps. The nature of interaction for 5FW, interacting with S134N mutated E-loop is found to be electrostatic and the interaction energy is −215 kJ mol⁻¹ during the initial period of simulation. The MDTQ interacting with S134N mutated E-loop exhibit electrostatic interaction up to 12 500 ps beyond this time period, the van der Waals interaction is found to be visible till at the end of the simulation period. The interaction energy is −280 kJ mol⁻¹ around 33 000 ps for the above protein–ligand complex. During the interaction of U5P with S134N mutated E-loop, the van der Waals interaction is remarkable and the interaction energy is −445 kJ mol⁻¹ during the initial period of simulation around 7500 ps. The interaction energy of S134N mutated E-loop with different chemical compounds are in the following order U5P + S134N > 5UD + S134N > MDTQ + S134N > 5FW + S134N. Hence, the chemical compounds U5P and 5FW are found to have stronger and weaker interaction respectively than all other chemical compounds in all three single point mutated E-loops viz., D124V, D125H and S134N. The strong interaction exhibited by the mutated E-loops with different ligands restricts the conformational disorder of the mutated E-loops. The difference in interaction affinity exhibited by the mutated E-loops with four different ligands arises when the inward facing conformers of the mutated E-loops are transformed to outward facing conformers.⁵¹ The chemical compound U5P prevent the structural destabilization of E-loop and deprotection of β-strands VII & VIII. Hence, the misfolding of SOD1 protein associated with FALS mutations can be reduced.

Hydrogen bond analysis

The number of hydrogen bonds in the native and mutated E-loops helps to understand the stability of the E-loops. The number of hydrogen bonds between the residues of the E-loop and the surrounding water molecules in the native and mutated states is shown in Fig. 9. The number of hydrogen bonds formed in the E-loop of the native state is 25, whereas in the case of the D124V-, D125H- and S134N-mutated E-loops the numbers are 9, 12, and 15, respectively. Thus, the number of hydrogen bonds within the mutated E-loops are in the following order S134N > D125H > D124V in correlation with that of the native state. Similarly, the number of hydrogen bonds between the E-loop and the surrounding water molecules in the native state is found to be 105, while for the D124V-, D125H- and S134N-mutated E-loops, the numbers are around 47, 79, and 95, respectively, throughout a simulation period of 50 000 ps. The reduction in intramolecular hydrogen bonds within the mutated E-loops destabilizes the conformations of the loop, thereby facilitating the unfolding and misfolding processes of the mutated E-loops due to the loss of dehydration based on hydrophobic interactions.⁵² The reduced number of hydrogen bonds alters the electrostatic state of the loop responsible for the charge distribution, which in turn affects the binding interfaces.⁵³ The E-loop with single point mutation will exhibit a rapid change in flexibility of the loop in comparison with that of the native state E-loop. For the E-loop with single point mutation, the side-chains of the polar residues do not form structural coupling with the surrounding water molecules, because the energies of the water molecules and the interacting polar residues are found to be inequivalent,⁵⁴ hence resulting in a reduced number of intermolecular hydrogen bonds when compared with that of the native state. Since, the hydrogen bonds are dominated by electrostatic interactions, the reduced number of hydrogen bonds in the mutated E-loops corresponds to the reduced electrostatic interactions of the D124V-, D125H- and S134N-mutated E-loops.⁵⁵ The reduction in number of hydrogen bonds produce structural deformation in the mutated E-loops and also in the β-strands VII & VIII at their hinges. Furthermore, the structural rearrangement in the E-loop with S134N mutation affects the major open turn of the E-loop, which is being stabilized in the native state E-loop by forming hydrogen bonds between Ser 134 and O₁₂₇, N₁₂₈, and N₁₂₉.⁴³ Hence, the loss of hydrogen bond interactions in the mutated E-loops affects the α-helix dipole interactions prevalent in the native state E-loop.⁴³

Fig. 9 (a) The number of hydrogen bonds in the native and mutated E-loops, and (b) the number of hydrogen bonds formed between the atoms of the E-loop with the surrounding water molecules in the native and the three different mutated E-loops.

The mutation which decreases the net charge of the protein or increases the hydrophobicity of the protein will lead to aggregation of the protein.^56,57 The positively charged E-loop attracts negatively charged superoxide anions and guides them towards the Cu²⁺ metal-binding centre where the superoxide anions are disproportionated into hydrogen peroxide and molecular oxygen. Due to the metal-binding mutations D124V, D125H and S134N, the charge of the E-loop is found to be decreased, so it could not attract and guide superoxide anions towards the catalytic metal centre. As a result of this, the enzymatic activity of the protein is lost. The loss in enzymatic activity is found to be proportional to the degree of Cu²⁺ ion deficiency.⁵⁸ The binding of the Cu²⁺ ion is one of the post-translational processes. Hence, the retardation of the post-translational process will increase the chance for the SOD1 protein to be misfolded and aggregated.⁵⁹ Due to charge–dipole interactions, hydrogen bonds are formed, where the ionizable charge groups interact with the partial charges or dipoles of surrounding polar groups.⁶⁰ So, the decrease in charge of the mutated E-loops will reduce the interactions of the mutated E-loops with the surrounding water molecules. Hence, the D124V-, D125H- and S134N-mutated E-loops possess a lower number of H-bonds with the surrounding water molecules than that of the native state E-loop.

In both the D124V- and D125H-mutated E-loops the significant number of hydrogen bonds formed between Asp 124 Oδ and Asp 125 N is lost.⁴² The disorder in the mutated E-loops due to the loss of native hydrogen bond interactions not only affects the interactions at the metal-binding site but also affects the crystal packing interactions.⁴² Due to the loss of hydrogen bond interactions within the S134N-mutated E-loop due to the replacement of Ser 134 with Asn, the segment of the loop involving the residues 129–137 will exhibit extensive mobility on the nanosecond timescale.¹⁴ The side-chain of Ser 134 in the native state E-loop is found to be important because it controls the conformation of the loop through hydrogen bond interactions with the side-chain of Asp 125.⁶¹ The loss of this combined interaction due to the mutations of Asp 125 to His and Ser 134 to Asp does not allow the E-loop to bind to Zn; hence, it distorts the structural organization of the Zn-binding loop.^62,63 Even though the S134N-mutated E-loop possesses a higher number of hydrogen bonds than those of the D124V- and D125H-mutated E-loops, the residues 125–131 are found to be involved in inter-dimeric contacts between the pathogenic SOD1 dimers. These interactions induce the formation of cross β-fibrillar structures.⁶⁴ A loss of exchange rate between the protons of the mutated E-loops and the solvent molecules arises because the exchange rates are not governed by the inductive effect of the nearest neighbouring residues in the loop.⁶⁵ The loss of the hydrogen bond network within the mutated E-loop not only results in conformational instability of the protein but also results in charge–charge repulsion and rearrangement of the hydrophobic core on the surface of the protein. The number of hydrogen bond interactions for the D125H- and S134N-mutated E-loops are found to be higher than that of the D124V-mutated E-loop, hence resulting in a lower potential for aggregation of the protein.⁶⁶ This variation in the aggregation rate for D124V, D125H and S134N arises because of the difference in enzymatic activity, protein thermostability, mutation position and degree of change in protein charge.⁶⁶

Free energy analysis

Fig. 10 depicts the free energy landscapes for the native and mutated E-loops along with the data of RMSD and R_g obtained from the trajectories. The free energy landscapes for the E-loops with the single point mutations D124V, D125H and S134N interacting with 5UD, 5FW, U5P and MDTQ are given in Fig. S7.† In all the figures, the colour scale represents energy values ranging from 0 to 3.5 kcal mol⁻¹. The free energy values are found to increase from deep blue to dark brown for all the structures. The deep blue region corresponding to lower energy values accounts for the global energy minimum for both the native and mutated E-loops.⁶⁷ In the native state E-loop, the stable conformer with average values of RMSD and R_g of 0.5 nm and 0.8 nm corresponds to the global energy minimum. The free energy profile for the D124V-mutated E-loop exhibits the global free energy minima for a few conformers with a RMSD value of 0.5 nm and a R_g value of 0.58 nm. Most of the non-native expanded and flexible conformers possess a Gibbs free energy value between 1.8 and 3.5 kcal mol⁻¹, with RMSD and R_g values reaching to a maximum of 0.6 nm and 0.85 nm, respectively. As a result of this, the highly irreversible change in the conformational structure of the E-loop with D124V single point mutation enhances the tendency of the FALS-associated SOD1 protein to form toxic aggregates. The reduction in the number of conformers exhibiting a global free energy minimum in the D124V-mutated E-loop allows the loop to disperse away to a larger extent, because the global free energy minimum in the native state E-loop behaves like an attractive funnel that prevents the structures from dispersing away to a larger extent.⁶⁷ In the case of the D124V-mutated E-loop interacting with 5UD, 5FW, U5P and MDTQ, the global free energy minimum is attained by a few conformers exhibiting RMSD values between 0.4 and 0.6 nm and R_g values between 0.5 and 0.6 nm. Most of the conformers of 5UD interacting with the D124V-mutated E-loop possess a Gibbs free energy value between 0.9 and 2.6 kcal mol⁻¹. Similarly, the conformers of 5FW, U5P and MDTQ interacting with the D124V-mutated E-loop exhibit Gibbs free energy values from 0.9 to 1.8 kcal mol⁻¹. The free energy profile of the D125H-mutated E-loop indicates that a few conformers exhibit global free energy minima with a RMSD value between 0.4 and 0.6 nm and a R_g value around 0.76 nm. Most of the conformers of the D125H-mutated E-loop are found to possess a Gibbs free energy value of around 1.8 kcal mol⁻¹ with the RMSD and R_g values reaching a maximum of 1.0 nm. The conformers of 5UD, 5FW, U5P and MDTQ interacting with the D125H-mutated E-loop exhibit global free energy minima with a RMSD value between 0.6 and 0.7 nm and a R_g value between 0.7 and 0.8 nm. Even though the position of the conformers with maximum free energy in the free energy landscape for 5UD, 5FW and U5P interacting with the D125H-mutated E-loop are similar, the structural characteristics and ensemble distribution of the conformers are different because of a greater RMSD value exhibited by the conformers, contributing to an energy value of 3.5 kcal mol⁻¹ in the free energy profile. In the case of the S134N-mutated E-loop, the global free energy minimum is exhibited by conformers with average RMSD and R_g values of around 0.78 nm. The variation in the orientation of the side-chains in the residues of the S134N-mutated E-loop arises because the conformers possess similar RMSD and R_g values with a larger variation from the RMSD and R_g values of the native state E-loop. This in turn allows the non-native state with highly flexible conformers to populate more at an energy value of 3.5 kcal mol⁻¹ in the free energy conformational space. Hence, the increased proportion of unstable conformers in the S134N-mutated E-loop induces the unfolding rate of the loop.

Fig. 10 Conformational free energy profiles with RMSD and R_g values from the trajectories of 50 ns in 2D space: (a) the E-loop in the native state, (b) D124V-mutated E-loop, (c) D125H-mutated E-loop, and (d) S134N-mutated E-loop.

The chemical compounds 5FW, U5P and MDTQ interacting with D124V and S134N, and MDTQ interacting with the D125H-mutated E-loops induce the conformers to attain global free energy minima. The distribution of conformers with an energy value of 3.5 kcal mol⁻¹ is reduced in the free energy landscape for the D124V- and S134N-mutated E-loops after interacting with the 5FW, U5P and MDTQ molecules, and similarly for the D125H-mutated E-loop interacting with MDTQ. The chemical compounds 5FW, U5P and MDTQ interacting with the D124V-, D125H- and S134N-mutated E-loops result in the formation of conformers with intermediate energy states which in turn reduces the unfolding and aggregation rate of the E-loop with single point mutation. The formation of intermediate energy states occurs for the above protein–ligand complexes because the mutated E-loop with a reduced net charge due to a reduced number of charged residues once interacting with the above chemical compounds, allows the loop to flip between the intermediate and unstable energy states in the free energy landscape. Hence, the conformers with these intermediate energy states, appearing to be very close to the global free energy minimum, correspond to metastable conformers in the free energy landscape. As a result of this, the aggregation propensity of the mutated E-loops is found to be reduced to a greater extent after interacting with 5FW, U5P and MDTQ. Most of the conformers in the D124V- and S134N-mutated E-loops with an average Gibbs free energy value around 3.5 kcal mol⁻¹ accounts for the loss of thermal stability of the loop. The dispersion of the D125H-mutated E-loop is reduced than those of the D124V- and S134N-mutated E-loops because a few conformers of the E-loop with D125H mutation partially form an attractive funnel, preventing the structure from dispersing away to a larger extent.⁶⁷ The vast difference in the free energy profiles of the mutated E-loops from that of the native state E-loop describes the unfolding behaviour exhibited by the mutated E-loops. The unfolded state of the mutated E-loops is confirmed by the regions of the free energy profile with higher RMSD values, which in turn correspond to a higher potential energy of this region. In the native state, the region with a lower potential energy in the free energy profile corresponds to lower RMSD values indicating the folded state of the E-loop. The free energy profile with poorly defined minima for the E-loops with single point mutation leads to clustering of unfolded conformers. At very high RMSD values, the E-loop with single point mutation is found to be dissimilar to the native state E-loop, and in such regions of conformational free energy space the gradient of the RMSD value is found to be irregular, hence exhibiting a sudden jump in the free energy values.⁶⁸ The significant free energy minima for the mutated E-loops interacting with 5FW, U5P and MDTQ correspond to the prevention of unfolded and expanded conformations of the loop. The interactions of 5FW, U5P and MDTQ with the mutated E-loops are found to possess favourable hydrophobic interactions with the β-strands VII and VIII.⁶⁹ Thus, the RMSD and R_g based free energy profiles indicate that the mutated E-loops interacting with the chemical compounds reduce the free energy barrier.

Conclusions

The RMSD values exhibited by the mutated E-loops correspond to the structural disorder of the loop. The chemical compounds 5FW, 5UD, U5P and MDTQ interacting with the mutated E-loops exhibit RMSD values close to the RMSD value of the E-loop in its native state. Thus, the chemical compounds are useful in reorienting the structurally disordered mutated E-loops, which in turn reduces the unfolding propensity and aggregation rate of mutated SOD1. Flexibility vs. rigidity analysis for native and mutated E-loops shows that the mutated E-loops have a higher flexibility than that of the native state E-loop which corresponds to the loss of activity of the loop. Hence, the catalytic function of the entire SOD1 protein is lost. Similarly, the variation in rigidity of the mutated E-loops compared with the rigidity of the native state E-loop accounts for the loss of thermodynamic stability of the protein. The larger end-to-end distances of the mutated E-loops result in weaker hydrogen bonding between the E-loop and β-strands, thereby resulting in the detachment of the E-loop from the β-strands. The chemical compounds U5P and 5FW are found to exhibit higher and lower interaction energies, respectively, with the three different mutated E-loops than those of the other chemical compounds. The reduced electrostatic potentials for the mutated E-loops account for the alteration in the intramolecular hydrogen bonding network found within the loop. The steric clash induced by the single point mutations between the residues of the loop is found to account for the reduced electrostatic potentials. The electrostatic potentials of the mutated E-loops interacting with U5P are found to approach the electrostatic potential of the native state E-loop. The lower number of hydrogen bonds in the mutated E-loop affects the conformational stability of the loop. Hence, destabilization of the mutated E-loops has been observed as a good cause for FALS. The reduced number of hydrogen bonds between the mutated E-loops and the surrounding water molecules indicates that the surface of the mutated E-loops is hydrophobic in nature. The conformational free energy profiles were drawn based on the R_g and RMSD values, which characterize the unfolding behaviour of the mutated E-loops. From the molecular dynamics simulation, we found that the E-loop in the native and mutated states exhibits perpetual fluctuations. The varied conformations of the mutated E-loops before and after interacting with various chemical compounds enables the estimation of the molecular properties of the loop based on the free energy landscape.

References

1. G. Marucci, L. Morandi, I. Bartolomei, F. Salvi, A. Pession, A. Righi, G. Lauria and M. P. Foschini, Neuromuscular Disord., 2007, 17, 673–676.
2. M. S. Rotunno and D. A. Bosco, Front. Cell. Neurosci., 2013, 7, 1–16.
3. S. Antonyuk, R. W. Strange and S. S. Hasnain, J. Med. Chem., 2010, 53, 1402–1406.
4. M. R. Ciriolo, A. Battistoni, M. Falconi, G. Filomeni and G. Rotilio, Eur. J. Biochem., 2001, 268, 737–742.
5. A. Das and S. S. Plotkin, J. Mol. Biol., 2013, 425, 850–874.
6. L. Banci, P. Carloni and P. L. Orioli, Proteins: Struct., Funct., Genet., 1994, 18, 216–230.
7. E. D. Getzoff, D. E. Cabelli, C. L. Fisher, H. E. Parge, M. S. Viezzoli, L. Banci and R. A. Hallewwell, Nature, 1992, 358, 347–351.
8. F. Polticelli, A. Battistoni, P. O’Neill, G. Rotilio and A. Desideri, Protein Sci., 1998, 7, 2354–2358.
9. K. S. Molnar, N. M. Karabacak, J. L. Johnson, Q. Wang, A. Tiwari, L. J. Hayward, S. J. Coales, Y. Hamuro and J. N. Agar, J. Biol. Chem., 2009, 284, 30965–30973.
10. A. Desideri, M. Falconi, F. Polticelli, M. Bolognesi, K. Djinovic and G. Rotilio, J. Mol. Biol., 1992, 223, 337–342.
11. E. D. Getzoff, J. A. Trainer, P. K. Weiner, P. A. Kollman, J. S. Richardson and D. C. Richardson, Nature, 1983, 306, 347–351.
12. I. Klapper, R. Hagstrom, R. Fine, K. Sharp and B. Honig, Proteins, 1986, 1, 47–59.
13. W. H. Koppenol, Biophys. J., 1980, 29, 493–507.
14. L. Banci, I. Bertini, N. D’Amelio, E. Gaggelli, E. Libralesso, I. Matecko, P. Turano and J. S. Valentine, J. Biol. Chem., 2005, 280, 35815–35821.
15. J. S. Elam, A. B. Taylor, R. Strange, S. Antonyuk, P. A. Doucette, J. A. Rodriquez, S. S. Hasnain, L. J. Hayward, J. S. Valentine, T. O. Yeates and P. J. Hart, Nat. Struct. Biol., 2003, 10, 461–467.
16. R. M. Adasme, C. M. Suomivuori, A. Fierro, J. Pesonen and D. Sundholm, J. Biol. Inorg. Chem., 2013, 18, 931–938.
17. J. S. Elam, K. Malek, J. A. Rodriguez, P. A. Doucette, A. B. Taylor, L. J. Hayward, D. E. Cabelli, J. S. Valentine and P. J. Hart, J. Biol. Chem., 2003, 278, 21032–21039.
18. C. Kayatekin, J. A. Zitzewitz and C. R. Matthews, J. Mol. Biol., 2010, 398, 320–331.
19. G. S. A. Wright, S. V. Antonyuk, N. M. Kershaw, R. W. Strange and S. S. Hasnain, Nat. Commun., 2013, 4, 1758–1767.
20. G. Wright, Fragment based drug discovery for SOD1-ALS and solution X-ray scattering studies on the copper chaperone for SOD1 (hCCS) and its functional complexes with SOD1, PhD Thesis, University of Liverpool, 2011, pp. 1–174.
21. H. M. Berman, K. Henrick and H. Nakamura, Nat. Struct. Biol., 2003, 10, 980.
22. R. W. Strange, S. Antonyuk, M. A. Hough, P. A. Doucette, J. A. Rodriguez, P. J. Hart, L. J. Hayward, J. S. Valentine and S. S. Hasnain, J. Mol. Biol., 2003, 328, 877–891.
23. S. V. Seetharaman, D. D. Winkler, A. B. Taylor, X. Cao, L. J. Whitson, P. A. Doucette, J. S. Valentine, V. Schirf, B. Demeler, M. C. Carroll, V. C. Culotta and P. J. Hart, Biochemistry, 2010, 49, 5714–5725.
24. O. Trott and A. J. Olson, J. Comput. Chem., 2010, 31, 455–461.
25. W. F. van Gunsteren, S. R. Billeter, A. A. Eising, P. H. Hünenberger, P. Krüger, A. E. Mark, W. R. P. Scott and I. G. Tironi, Biomolecular Simulation: The GROMOS96 Manual and User Guide, Vdf Hochschulverlag AG an der ETH Zürich, Zürich, Switzerland, 1996, pp. 1–1042.
26. P. H. Nguyen, G. Stock, E. Mittag, C. K. Hu and M. S. Li, Proteins: Struct., Funct., Genet., 2005, 61, 795–808.
27. P. H. Nguyen, M. S. Li, G. Stock, J. E. Straub and D. Thirumalai, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 111–116.
28. H. B. Nam, M. Kouza, H. Zung and M. S. Li, J. Chem. Phys., 2010, 132, 165104–165114.
29. P. H. Nguyen, M. S. Li and P. Derreumaux, Phys. Chem. Chem. Phys., 2011, 13, 9778–9788.
30. H. J. C. Berendsen, J. R. Grigera and T. P. Straatsma, J. Phys. Chem., 1987, 91, 6269–6271.
31. O. M. Becker, A. D. MacKerell Jr, B. Roux and M. Watanabe, Computational BioChemistry and BioPhysics, CRC Press, New York. 2001, pp. 1–512.
32. M. Yadav, Computational Chemistry, Discovery Publishing House Pvt. Ltd, New Delhi. 2009, pp. 1–265.
33. B. Hess, H. Bekker, H. J. C. Berendsen and J. G. E. M. Fraaije, J. Comput. Chem., 1997, 18, 1463–1472.
34. S. Miyamoto and P. A. Kollman, J. Comput. Chem., 1992, 13, 952–962.
35. T. Darden, D. York and L. Pedersen, J. Chem. Phys., 1993, 98, 10089–10092.
36. M. Parrinello and A. Rahman, Phys. Rev. Lett., 1980, 45, 1196–1199.
37. G. Bussi, D. Donadio and M. Parrinello, J. Chem. Phys., 2007, 126, 014101–014108.
38. D. Van der Spoel, E. Lindahl, B. Hess, A. R. Van Buuren, E. Apol, P. J. Meulenhoff, D. P. Tieleman, A. L. T. M. Sijbers, K. A. Feenstra, R. Van Drunen and H. J. C. Berendsen, Gromacs User Manual version 4.5.4, https://www.gromacs.org, 2010.
39. C. L. Fisher, D. E. Cabelli, R. A. Hallewell, P. Beroza, T. P. Lo, E. D. Getzoff and J. A. Tainer, Proteins, 1997, 29, 103–112.
40. J. A. Rodriquez, J. S. Valentine, D. K. Eqqers, J. A. Roe, A. Tiwari, R. H. Brown Jr and L. J. Hayward, J. Biol. Chem., 2002, 277, 15932–15937.
41. R. M. Adasme, F. Mendizábal, M. Gonzalez, S. M. Rojas, C. O. Azar and D. Sundholm, Inorg. Chem., 2012, 51, 5561–5568.
42. S. Antonyuk, J. S. Elam, M. A. Hough, R. W. Strange, P. A. Doucette, J. A. Rodriguez, L. J. Hayward, J. S. Valentine, P. J. Hart and S. S. Hasnain, Protein Sci., 2005, 14, 1201–1213.
43. H. E. Parge, R. A. Hallewell and J. A. Tainer, Proc. Natl. Acad. Sci. U. S. A., 1992, 89, 6109–6113.
44. B. Ranjith and C. George Priya Doss, Appl. Biochem. Biotechnol., 2013, 169, 1659–1671.
45. P. L. Wintrode, D. Zhang, N. Vaidehi, F. H. Arnold and W. A. Goddard III, J. Mol. Biol., 2003, 327, 745–757.
46. T. Li, M. B. Tracka, S. Uddin, J. Casa-Finet, D. J. Jacobs and D. R. Livesay, PLoS One, 2014, 9, e92870–e92883.
47. A. López de Victoria, P. Tamamis, C. A. Kieslich and D. Morikis, PLoS One, 2012, 7, e49925–e49939.
48. V. Sivozhelezov, E. Pechkova and C. Nicolini, J. Theor. Biol., 2006, 241, 73–80.
49. S. Gavryushov and P. Zielenkiewicz, J. Phys. Chem. B, 1997, 101, 10903–10909.
50. K. Bozek, E. E. Nakayama, K. Kono and T. Shioda, Front. Microbiol., 2012, 3, 1–6.
51. J. Zhang, T. Sun, L. Liang, T. Wu and Q. Wang, Soft Matter, 2014, 10, 438–445.
52. A. Fernández and H. A. Scheraga, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 113–118.
53. C. G. Priya Doss and N. NagaSundaram, PLoS One, 2012, 7, e31677–e31687.
54. G. A. Petsko and D. Ringe, Protein Structure and Function, New Science Press, London. 2004, pp. 1–189.
55. S. Kim and T. A. Cross, Biophys. J., 2002, 83, 2084–2095.
56. A. Tiwari, Z. Xu and L. J. Hayward, J. Biol. Chem., 2005, 280, 29771–29779.
57. E. Sandelin, A. Nordlund, P. M. Andersen, S. S. L. Marklund and M. Oliveberg, J. Biol. Chem., 2007, 282, 21230–21236.
58. Y. Furukawa and T. V. O’Halloran, Antioxid. Redox Signaling, 2006, 8, 847–867.
59. Y. Furukawa, Neurol. Res. Int., 2012, 2012, 1–6.
60. C. N. Pace, G. R. Grimsley and J. M. Scholtz, J. Biol. Chem., 2009, 284, 13285–13289.
61. A. Radunović and P. N. Leigh, J. Neurol., Neurosurg. Psychiatry, 1996, 61, 565–572.
62. C. Kayatekin, The Coupling Between Folding, Zinc Binding, and Disulfide Status of Human Cu, Zn Superoxide Dismutase: A Dissertation, Graduate School of Biomedical Sciences, University of Massachusetts Medical School, 2010, pp. 1–266.
63. J. Danielsson, M. Kurnik, L. Lang and M. Oliveberg, J. Biol. Chem., 2011, 286, 33070–33083.
64. B. F. Shaw, A. Durazo, A. M. Nersissian, J. P. Whitelegge, K. F. Faull and J. S. Valentine, J. Biol. Chem., 2006, 281, 18167–18176.
65. J. A. Rodriguez, B. F. Shaw, A. Durazo, S. H. Sohn, P. A. Doucette, A. M. Nersissian, K. F. Faull, D. K. Eggers, A. Tiwari, L. J. Hayward and J. S. Valentine, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 10516–10521.
66. M. Prudencio, P. J. Hart, D. R. Borchelt and P. M. Andersen, Hum. Mol. Genet., 2009, 18, 3217–3226.
67. N. K. Banavali and B. Roux, J. Am. Chem. Soc., 2005, 127, 6866–6876.
68. J. Hénin, G. Fiorin, C. Chipot and M. L. Klein, J. Chem. Theory Comput., 2010, 6, 35–47.
69. C. S. Schwaiger, S. I. Börjesson, B. Hess, B. Wallner, F. Elinder and E. Lindahl, PLoS One, 2012, 7, e45880–e45890.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra00286a

---