Human SOD1 before Harboring the Catalytic Metal

SOD1 has to undergo several post-translational modifications before reaching its mature form. The protein requires insertion of zinc and copper atoms, followed by the formation of a conserved S-S bond between Cys-57 and Cys-146 (human numbering), which makes the protein fully active. In this report an NMR structural investigation of the reduced SH-SH form of thermostable E,Zn-as-SOD1 (E is empty; as is C6A, C111S) is reported, characterizing the protein just before the last step leading to the mature form. The structure is compared with that of the oxidized S-S form as well as with that of the yeast SOD1 complexed with its copper chaperone, CCS. Local conformational rearrangements upon disulfide bridge reduction are localized in the region near Cys-57 that is completely exposed to the solvent in the present structure, at variance with the oxidized forms. There is a local disorder around Cys-57 that may serve for protein-protein recognition and may possibly be involved in intermolecular S-S bonds in familial amyotrophic lateral sclerosis-related SOD1 mutants. The structure allows us to further discuss the copper loading mechanism in SOD1.

Cu,Zn-SOD 2 is a very efficient enzyme that catalyzes the dismutation of superoxide to oxygen and hydrogen peroxide (1). It is a dimer in all eukaryotes, whereas in prokaryotes it exists as either a dimer or a monomer. The mature, active form of SOD1 contains, in each subunit, a copper ion essential for catalysis and a zinc ion that has primarily a structural role.
In the active site, a narrow channel is present that is large enough to admit only superoxide, water, and small anions and ligands such as imidazole and peroxynitrite. In the lining of the channel there is a positively charged side chain from an arginine residue (Arg-143, human numbering). The cationic residue generates an electrostatic gradient proposed to be responsible for steering the superoxide anion toward the active site. Site-directed mutagenesis of residue 143 to a neutral or ani-onic residue produced proteins with dramatically reduced or abolished activity (2).
A peculiar feature of the mature form of SOD1 is the presence of a kinetically stable disulfide bond. Once formed, the disulfide bond is maintained in the reducing cytoplasmic environment where the majority of the SOD1 inside the cell is located. The presence of intramolecular disulfide bonds is common in secreted proteins, where their primary purpose is for protein stabilization. However, disulfide bonds are rare in intracellular proteins because of the highly reducing environment and low concentration of dioxygen in the cytosol (3,4). It has been previously shown that intramolecular disulfide bonds in intracellular proteins (like SOD1) can play more than just a structural role and have functional significance (3). The disulfide bond between Cys-57 and Cys-146 is fully conserved in all SOD1 structures. This bond links loop IV, which contains Cys-57, with strand ␤8, containing Cys-146. The linkage of the secondary structure elements contributes to the stabilization of the SOD1 fold.
The active form of the protein is obtained when one zinc and one copper ion per subunit are bound. Although nothing is known on how the protein acquires the zinc ion, a description of the insertion of copper is available. The process is aided by a copper metallochaperone called CCS (copper chaperone for superoxide dismutase) (5,6). Yeast mutants lacking CCS express a form of SOD1 protein that is essentially apo for copper but contains zinc (5,7). However, a CCS-independent activation of mammalian SOD1 is possible (8). A mechanism of the copper transfer from CCS to SOD1 in yeast has been proposed based on the x-ray structure of the heterodimeric CCS-SOD1 complex (9). In this picture, SOD1 first loads zinc ions from an unknown source, and subsequently copper transfer takes place through the formation of an intermolecular S-S bond, involving Cys-57 of SOD1 and a Cys of CCS. The disulfidereduced, copper-depleted form of SOD1 therefore represents the state of the protein before its complex formation with copper(I) CCS (10).
Additionally, this protein state may play a relevant role in pathological conditions. The presence of exposed cysteine residues may modulate its ability to interact with other macromolecules and to give rise to large molecular aggregates (11). Aggregates containing SOD1 have been found in some fALS patients and in some fALS SOD1 transgenic mice and rats (12,13). The fALS pathology has been related to mutations in the protein Cu,Zn-SOD1, some of which have a tendency to form fibrils (14). An increasing number of pathological states involve protein aggregates, which represent the final state of biomolecules with altered folding properties and/or structural conformations (15). To avoid interference effects arising from Cys-6 and Cys-111 (i.e. the two other Cys present in SOD1 that are in the reduced state), we studied the thermostable mutant of human SOD1 (C6A, C111S), which is structurally and functionally equivalent to the wild type protein (16 -18) and is commonly used in the study of fALS mutants to prevent intermolecular aggregation (4, 19 -21).
Currently, there are a huge number of structures of SOD1 (Ͼ50, ϳ10 of which of human SOD1) deposited in the Protein Data Bank in various states (22)(23)(24)(25), including several mutants, but all with the disulfide bond intact. Therefore, the structural characterization of E,Zn-SOD1 with the disulfide bond reduced that is reported here (hereafter E,Zn-hasSOD1 SH-SH ) is particularly meaningful as it is the first structure of SOD1 with the disulfide bond broken. It represents the step before the protein acquires copper but also a state that is emerging to be relevant in the fALS-related SOD1 mutants (11,26). Therefore, it provides useful new data on the features of SOD1.

MATERIALS AND METHODS
Sample Preparation-Human asSOD1 (hasSOD1) was expressed in the Escherichia coli TOPP1 (Stratagene) or BL21(DE3) strain. The 15 N and 15 N, 13 C, 2 H-labeled protein samples were obtained as previously reported (25). The protein was isolated and purified according to previously published protocols (16). The triple-labeled dimeric hasSOD1 contained ϳ70% 2 H. Metal ions were removed according to published protocols (27), and zinc reconstitution was obtained as described in Ref. 17. The reduction of disulfide bridge was accomplished by addition of dithiothreitol. Fully reduced and empty copper protein hasSOD1 (E,Zn-hasSOD1 SH-SH ) was prepared under a nitrogen atmosphere in an anaerobic chamber. The thiol-disulfide status of purified hasSOD1 was determined by chemical modification with the thiol-specific reagent 4-acetamide-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid (Molecular Probes, Inc.) (10).
The NMR samples of the E,Zn-hasSOD1 SH-SH were in 20 mM sodium phosphate buffer, pH 5, 90% H 2 O/10% D 2 O. The final protein concentration ranges between 1 and 1.5 mM in the presence of 20 mM dithiothreitol. ϳ0.6 ml of sample was loaded into high quality NMR tubes that were capped with latex serum caps in the Vacuum Atmospheres chamber.
NMR Measurements and Structure Calculation-The NMR spectra were acquired on Avance 900, 800, 600, and 500 Bruker spectrometers. All of the triple resonance probes used were equipped with pulsed field gradients along the z-axis. The 800-and 500-MHz spectrometers were equipped with a triple resonance cryoprobe.
The NMR experiments, used for the backbone and the aliphatic side chain resonances assignment and for obtaining structural restraints, recorded on 2 H/ 13 C/ 15 N/, 13 C/ 15 N-enriched and unlabeled E,Zn-hasSOD1 SH-SH samples, are summarized in supplemental Table S1. The 1 H, 13 C, and 15 N resonance assignments of E,Zn-hasSOD1 SH-SH are reported in supplemental Table S2. In total, 88% of the 1 H, 98% of the 15 N, and 80% of the 13 C have been assigned. The combined variation of the 15 N and 1 H amide chemical shifts of E,Zn-hasSOD1 SH-SH , Cu,Zn-hasSOD1 SS , and E,Zn-E133QM2-asSOD1 SS are shown in supplemental Fig. S1. 1 H-1 H nuclear Overhauser enhancement spectroscopy (NOESY) experiment (80-ms mixing time) (28) was carried out to identify connectivities involving histidines of the binding sites at 298 K. All the spectra were collected at 298 K, processed using the standard Bruker software (XWINNMR), and analyzed through the CARA program (29).
An automated CANDID approach combined with DYANA torsion angle dynamics algorithm (30) was used to assign the ambiguous NOE cross-peaks and to have a preliminary protein structure. Structure calculations were then performed through iterative cycles of DYANA (31) followed by restrained energy minimization with AMBER 5.0 (32) applied to each member of the final DYANA family. The assessment of the structures was evaluated using the program PROCHECK-NMR (33). The root mean square deviation values for the backbone and heavy atoms of one subunit are shown in supplemental Fig. S2. 15 N R 1 , R 2 , and steady-state heteronuclear NOEs, which can provide information on internal mobility as well as on the overall protein tumbling rate, were measured with pulse sequences as described by Farrow et al. (34). R 2 were measured using a refocusing time of 450 ms. In all experiments the water signal was suppressed with the "water flip back" scheme (35). Average R 1 , R 2 , and 1 H- 15 N NOE values of 0.81 Ϯ 0.10, 23.7 Ϯ 2.9, and 0.82 Ϯ 0.10 s Ϫ1 are found, respectively, at 600 MHz. A correlation time of 17.3 Ϯ 2.3 ns was estimated from the R 2 /R 1 ratio (the experimental relaxation rates are shown in supplemental Fig. S3).
The exchange with the solvent of the backbone amide proton was investigated through a series of 1 H-15 N heteronuclear single quantum correlation (HSQC) experiments performed over 4 days on the E,Zn-hasSOD1 SH-SH protein previously frozen dry and then dissolved in D 2 O. Exchange with the bulk solvent in D 2 O solution was measured following the intensity of amide proton moieties in 15 N-HSQC spectra. The amide protons can be grouped in three groups: fast exchanging (the peak disappears in less than 20 min), intermediate exchanging (the peak is still present after 20 min but disappears within 10 h), and slow exchanging (the peak persists after 10 h). The behavior indicates not only solvent exposure but also the presence of hydrogen bonding, which may slow down the exchange of the exposed amide protons. This is the case for almost all residues located in regions having a defined secondary structure, as they are involved in extensive H-bond networks that stabilize the ␤-barrel structure typical of this protein.
Copper(I) Insertion-The copper-depleted protein was added with the copper(I) complex (Cu(CH 3 CN) 4 ) ϩ in a 1:1 ratio in an anaerobic chamber. The protein was then washed extensively in the presence of phosphate buffer 20 mM at pH 5. The metal uptake of the protein was established through one-dimensional NMR spectrum in the region of imidazole NHs (10 -16 ppm).

RESULTS AND DISCUSSION
The structure of the copper-depleted thermostable form (C6A, C111S) of human SOD1 with the disulfide bond reduced (E,Zn-hasSOD1 SH-SH ) has been determined in solution by NMR with high resolution (root mean square deviation 0.66 Ϯ 0.14 and 1.15 Ϯ 0.13 for the backbone and heavy atom, respectively, considering the segments involved in the ␤-barrel). The structure quality and the statistics on constraint violations for the final family is shown in Table 1. The protein is in a dimeric state and completely maintains the global fold of mature SOD1 (Fig. 1). Surprisingly, the breakage of the disulfide bond does not significantly destabilize the protein structure. This can suggest a functional rather than structural role for this highly conserved moiety. In addition, the backbone conformation of the region at the dimer interface (residues 53-55, 149 -151) is very similar to that found in x-ray (22,36) and solution structures (25) of human and yeast Cu,Zn-SOD1 SS . Local conformational rearrangements are found essentially only in the immediate surrounding of the two cysteines (Cys-57 and -146) involved in the intrasubunit disulfide bond. This protein region, and in particular residue 57, becomes extremely more solvent accessible with respect to the oxidized protein (Fig. 2). The solvent exposure of cysteine 57 is a peculiarity that is never observed in SOD1 structures, not even in the apo form of the protein (Fig. 2 and supplemental Fig. S4). The presence of the disulfide bond, in fact, prevents solvent exposure by forcing the Cys-57 to reach the Cys-146 that is located within the protein core. Solvent exposure is also confirmed by H 2 O/D 2 O solvent exchange measurements. The region spatially close to the cysteine residues responsible for the disulfide bridge (residues 57 and 146) appears more exposed to the solvent. This was not observed in the structures of Cu,Zn-hasSOD1 SS and E,Zn-E133QM2-asSOD1 SS .
Interestingly, the metal binding ligands in E,Zn-hasSOD1 SH-SH have a conformation very close to that found in Cu,Zn-hasSOD1 SS and in the monomeric copper-depleted form (24,25), indicating that the protein is ready to harbor the catalytic metal. In the active site channel of Cu,Zn-SOD1 SS , located between the electrostatic loop VII (121-142) and loop IV (49 -84), some H-bonds form a highly conserved network that is important for generating the optimal electrostatic field necessary to attract and drive the substrate toward the copper site. In the present structure most of the H-bonds are maintained with the exceptions of the copper ligand His-48 hydrogen bonded to Cys-57 via the carbonyl oxygen of Gly-61 and the catalytically important side chain of Arg-143. This is also assessed by the chemical shift perturbation of the amide proton of Gly-61 (supplemental Fig. S1). The total solvent exposure of Cys-57 located in loop IV is also reflected in an increased solvent exposure of residues 48 and 146, which are buried in Cu,Zn-hSOD1 SS . Accordingly, the relative amide protons display fast hydrogen/deuterium exchange in E,Zn-hasSOD1 SH-SH (data not shown).
The conformation of Arg-143 is extremely sensitive to the position of Cys-57 (37)(38)(39). Arg-143 is located at the end of the electrostatic loop (loop VII) whose conformation greatly affects the enzymatic activity. In the dimeric Cu,Zn-hasSOD1 SS solution structure, Cys-57 is quite rigid along with loop VII. As a consequence, the side chain of Arg-143 is highly ordered and is in the optimal conformation to interact with the  The number of meaningful constraints for each class is reported in parentheses. c Medium range distance constraints are those between residues (i,iϩ2), (i,iϩ3), (i,iϩ4), and (i,iϩ5). d As it results from the Ramachandran plot analysis.  superoxide anion bound to copper. In the monomeric Cu,Zn-E133QM2-asSOD1 SS species, loop IV containing Cys-57 experiences conformational equilibria and Arg-143 moves away from copper, determining a sizable drop in SOD activity (40). In the present structure, loop IV is highly disordered due to the S-S breaking resulting in extensive disorder of the Arg-143 side chain. Consequently, the overall electrostatic field is highly perturbed. The comparison of our solution structure with that of monomeric E,ZnE133QM2-asSOD1 SS (24) reveals that the observed disorder of residue Arg-143 is not a consequence of the absence of copper but of the relaxation of loop IV due to the S-S breaking. The breakage of the disulfide bond results in a perturbed distribution of the electrostatic charges in the active site channel essential to guide the superoxide ion in the catalytic site. Two mechanisms have been proposed for copper uptake and consequent activation of human SOD1; the first requires the presence of CCS (9,41), and the second involves the reduced form of glutathione (8). The first mechanism, described in yeast, is based on the x-ray structure of the heterodimeric yeast CCS-SOD1 complex. Although a direct comparison with the yeast E,Zn-SOD1 SH-SH is not possible because of the lack of its structure in the Protein Data Bank, the present solution structure shares some specific features with that of the yeast mutant H48F SOD1 in the heterodimeric complex with CCS (9) (Fig. 3). All SOD1 residues involved in the interaction with domain II of CCS (residues 51, 114, 152) experience similar conformation in E,Zn-hasSOD1 SH-SH regardless of the differences in the primary sequence. It has been proposed that in yeast the presence of Pro-142 and Pro-144 in place of Ser-142 and Leu-144 may account for the loss of ability to acquire copper independently of CCS (8). Because yeast E,Zn-SOD1 SH-SH has been found to be monomeric (10), we can speculate that these differences in the primary sequence destabilize the structure of this region, leading to monomerization (Leu-144 is inserted in the hydrophobic part of the protein called the plug of the barrel). The correct conformation for acquiring copper could be restored only by the formation of the heterodimeric complex with CCS facilitated by interaction with domain III. However, the complexity of the issue is demonstrated by the increased propensity to monomerization displayed by a subgroup of fALS mutants (for example G85R, D125H, L38V) that have the mutation far away from the dimer interface (11,19,26).
In the crystal structure of the heterodimeric complex, an intermolecular disulfide bond between Cys-57 of SOD1 and Cys-229 of CCS was found. If the present solution structure is superimposed on the yeast H48F SOD1 in the heterodimeric complex, a similar distance of the sulfur atom of Cys-57 from that of Cys-229 of CCS is found (0.20 nm in the heterodimer complex and 0.28 nm in the superimposed solution structure of E,Zn-hasSOD1 SH-SH ). This feature demonstrates that Cys-57 has a favorable conformation to form the same disulfide bond. Also, in the heterodimer complex the side chain of Arg-143 is far away from the copper site and is hydrogen bonded to the backbone of Ala-230 of yeast CCS; the mobility of Arg-143 could have the role of making its side chain available for this kind of interaction.
On the basis of the present structure we can speculate that the copper ion could enter in its site with the assistance of Cys-57, which, being close to the copper site, could constitute a ligand in a hypothetical intermediate (Fig. 4). Indeed, the high affinity of copper for thiol groups and the complete exposure of the side chain of Cys-57 could be significant . Proposed mechanism for acquisition of copper by E,Zn-hSOD1 SH-SH (displayed as an ellipsoid). Copper and zinc are represented as spheres (dark and light gray, respectively) and their sites as empty circles. Thiol groups are shown as small spheres. A hypothetical mechanism for acquisition of copper ion from CCS or glutathione (GSH) complex (shown as a thick wavy line) could take place through the formation of a metalbridged intermediate. for the incorporation of copper ion into the protein, regardless of whether this occurs through the interaction with CCS (9,41) or with other copper-containing species (e.g. copper-glutathione complex) (8).
Independently of the mechanism by which the copper enters the site in vivo, we have observed that E,Zn-hasSOD1 SH-SH is able to acquire copper(I) in vitro. The acquisition of copper can, in fact, be easily monitored by one-dimensional NMR spectrum of imidazole NH protons. Supplemental Fig. S5 shows the signals originated by copper binding; the figure also shows that the spectra of Cu(I),Zn-hasSOD1 SH-SH and Cu(I),Zn-hSOD1 SS are very similar (and different from E,Zn-hasSOD1 SH-SH ), thus demonstrating that the metal enters correctly in its site. Subsequent oxidation of the disulfide bond would allow the formation of the hydrogen bond network constraining the side chain of Arg-143, making the protein fully functioning.
Exposed cysteines can also easily form interprotein disulfide bonds, thus giving rise to aggregates. The status of the disulfide bond in fALS SOD1 mutants has not been extensively characterized, although in all crystal structures of fALS mutants such bonds are intact (42)(43)(44). However, the disulfide bonds in fALS mutants are more susceptible to reduction than wild type SOD1 (26), and it has been suggested that in some fALS mutants, fibrils are formed through intersubunit S-S bonds (11). This implies that under cellular disulfide-reducing conditions at physiological pH and temperature, the fALS SOD1 mutants are more prone to aggregation. The present structure provides an important piece of information for understanding the copper uptake mechanism in SOD1 and for the debate about which is the relevant form of SOD1 for the onset of the fALS pathology.