Loss of in Vitro Metal Ion Binding Specificity in Mutant Copper-Zinc Superoxide Dismutases Associated with Familial Amyotrophic Lateral Sclerosis*

The presence of the copper ion at the active site of human wild type copper-zinc superoxide dismutase (CuZnSOD) is essential to its ability to catalyze the disproportionation of superoxide into dioxygen and hydrogen peroxide. Wild type CuZnSOD and several of the mutants associated with familial amyotrophic lateral sclerosis (FALS) (Ala4 → Val, Gly93 → Ala, and Leu38 → Val) were expressed inSaccharomyces cerevisiae. Purified metal-free (apoproteins) and various remetallated derivatives were analyzed by metal titrations monitored by UV-visible spectroscopy, histidine modification studies using diethylpyrocarbonate, and enzymatic activity measurements using pulse radiolysis. From these studies it was concluded that the FALS mutant CuZnSOD apoproteins, in direct contrast to the human wild type apoprotein, have lost their ability to partition and bind copper and zinc ions in their proper locations in vitro. Similar studies of the wild type and FALS mutant CuZnSOD holoenzymes in the “as isolated” metallation state showed abnormally low copper-to-zinc ratios, although all of the copper acquired was located at the native copper binding sites. Thus, the copper ions are properly directed to their native binding sites in vivo, presumably as a result of the action of the yeast copper chaperone Lys7p (yeast CCS). The loss of metal ion binding specificity of FALS mutant CuZnSODsin vitro may be related to their role in ALS.

More than 70 different point mutations of CuZnSOD have been implicated in the fatal motor neuron disease, familial amyotrophic lateral sclerosis (FALS) (5). It has been demonstrated that the disease is initiated not by a loss of function, i.e. decrease in catalytic SOD activity, but by the gain of a new and toxic property. Suggestions for the nature of this property include the increased ability of mutant CuZnSODs to catalyze oxidation reactions by hydrogen peroxide (6 -8), to promote the formation of nitrotyrosine (9 -12), and to induce protein aggregation (13)(14)(15), each of which might be acting alone or in combination. Many theories concerning the gain of function of FALS mutant CuZnSOD are based on the assumption that the new toxic property is related to the enzyme-bound metal ions, particularly the copper ions. The detailed mechanism for the catalytic disproportionation of O 2 Ϫ is well understood for human, bovine, and yeast CuZn-SODs (16 -22). The mechanism involves sequential reduction (Reaction 2) and reoxidation (Reaction 3) of the copper(II) center, where both reactions are relatively pH-independent over the pH range of 5.0 -9.5 and proceed with diffusion-controlled rate constants of 2 ϫ 10 9 M Ϫ1 s Ϫ1 (16). REACTION 3 In the oxidized form (Cu II ZnSOD), the Cu 2ϩ ion is bound in a distorted five-coordinate geometry by ligands from His 46 , His 48 , His 120 , His 63 , and a water molecule. The Zn 2ϩ is bound in a distorted tetrahedral geometry by ligands from His 71 , His 80 , Asp 83 , and His 63 (23-25) (see Scheme 1).
In vitro remetallation of apo bovine and human wild type CuZnSODs has been extensively studied. It has been established that these proteins bind 2 copper ions and 2 zinc ions/ apoprotein dimer and that these metal ions can be readily directed to their native metal binding sites when the metal ions are added slowly to the apoprotein (26,27). It is also known that the SOD activity of wild type CuZnSOD depends critically on the presence of the copper ion at its native metal binding sites and that the SOD activity of remetallated derivatives is proportional to the amount of copper bound in the native copper site (2,28,29). The zinc ions influence SOD activity indirectly by altering the local structure and binding affinities of the metal binding regions as well as stabilizing the global structure of the dimeric enzyme (30,31). Thus, the placement and amount of both the copper ions and the zinc ions is critical to the SOD activity.
Six different states of metallation have been examined for wild type bovine and human CuZnSOD proteins (see Table I).
1) The apoprotein itself has no SOD activity because it contains no copper (16,26). 2) The zinc-free derivative, CuESOD, in which copper ions are bound to the copper site of each subunit and the zinc sites are empty, has full SOD activity at low pH on a per copper basis, but the SOD activity is pH-dependent (31)(32)(33). 3) Addition of 2 more copper ions, for a total of 4 Cu 2ϩ per dimer (CuCu) yields full SOD activity per copper in the native copper site (34). 4) Properly remetallated CuZnSOD, prepared either by the addition of 2 zinc ions followed by 2 copper ions at pH 5.5 (29), or by the addition of 2 copper ions at pH 3.8 followed by 2 zinc ions at pH 7.8 (26,35), yields full SOD activity on a per copper basis. 5) CuCoSOD with Co 2ϩ bound at the native zinc site is also fully SOD active on a per copper basis (26,36). 6) The AgCuSOD derivative has virtually no SOD activity despite the presence of copper bound to the protein because copper ions in the zinc sites confer no SOD activity on the enzyme (35,37).
The UV-visible electronic absorption spectra of the remetallated derivatives of human wild type apo-CuZnSOD that contain Cu 2ϩ or Co 2ϩ provide valuable information concerning the nature of the metal binding sites to which they are bound. This information, combined with SOD activity profiles obtained by pulse radiolysis, has been used frequently to deduce the locations of Cu 2ϩ or Co 2ϩ ions bound to the wild type CuZnSOD. In the current study, we have applied this same approach to the FALS mutant human CuZnSODs G93A, A4V, and L38V and have found significant differences in the metal ion binding properties of the FALS mutant apoproteins relative to those of the wild type protein.

EXPERIMENTAL PROCEDURES
Materials-All solutions were prepared using distilled water that had been passed through a Millipore ultrapurification system. EDTA and sodium formate were purchased from Sigma. Monobasic potassium phosphate buffer (10 mM unless otherwise noted) (Ultrex; J.T. Baker Inc.), adjusted to the proper pH (pH ϳ5-12), was used in all activity measurements. The solution pH for the activity measurements was adjusted by adding H 2 SO 4 (double distilled from Vycor, GFC Chemical Co.) and NaOH (Puratronic, Alfa/Ventron Chemicals). Cupric sulfate (Cu(II)SO 4 ⅐5H 2 O, Fisher), zinc sulfate (Zn(II)SO 4 ⅐7H 2 O, Mallinckrodt), and cobaltous sulfate (Co(II)SO 4 ⅐7H 2 O, Mallinckrodt) were used for the remetallation of SOD. The enzymes were titrated in pH 5.5, 100 mM sodium acetate buffer. The as-isolated enzymes were in pH 7. 0, potassium phosphate buffer. Diethylpyrocarbonate (DEPC) was purchased from Sigma and prepared in 100 mM sodium phosphate buffer (pH 6.0).
Protein Expression, Purification, Characterization, and Preparation of Derivatives-Methods were identical to those published previously (38). In the case of L38V, extra care was required to titrate slowly to avoid overtitration with metal ions, specifically copper, because the remetallated derivatives of L38V were especially prone to precipitation.
Remetallation and Metal Content of CuZnSOD Derivatives-The CuZnSOD apoproteins were remetallated prior to SOD activity measurements as described previously, with zinc ion added prior to copper ion to form the CuZnSOD derivative at pH 5.5 (38). The CuCoSOD derivative was made in a similar fashion, with cobalt ion added prior to the addition of copper ion. For further descriptions on remetallating CuZnSOD, see the reference by Valentine and Pantoliano (29). The metal content was measured using a Pye-Unicam atomic absorption instrument. The various metallated states of the WT and mutants, including as-isolated, CuZn-, CuE-, CuCu-, and CuCoSOD were measured for the total content of copper, zinc, and cobalt after washing several times in Centricon-10 filters (Amicon) with 3-5 times the total sample volume, with either sodium phosphate buffer (pH 7.0) for the as-isolated enzymes or sodium acetate buffer (pH 5.5). This washing removed any free or weakly bound metal ions. The metal content was measured in triplicate with an experimental error Ͻ5%.
Pulse Radiolysis and SOD Activity-Samples to be analyzed by pulse radiolysis were washed as described above. Methods for determination of SOD activities were identical to those published previously (38). A first order rate for the catalytic disproportionation of O 2 Ϫ was obtained by monitoring the change in absorbance at 260 nm with respect to time. The reported rate constants for as-isolated CuZnSOD and remetallated CuZn-, CuE-, and CuCoSOD were calculated by dividing the observed rates by one-half of the total concentration of copper bound to the dimeric enzyme. The reported rate constants for CuCuSOD were calculated by dividing the observed rates by one-fourth of the concentration of bound copper. This calculation takes into account the fact that the 2 copper ions that are bound per protein dimer in the CuZn derivative are in the copper sites and both are therefore expected to be catalytically active in the SOD reaction, while only 2 of the 4 copper ions per protein dimer in the CuCu derivative are expected to be active since copper in the zinc sites of CuZnSOD derivatives does not have SOD activity. The activity from free copper was eliminated using EDTA in the measurements and also by washing the enzyme samples prior to measurement.
Metal Ion Titrations of Apoenzymes-Titrations of the apoproteins with solutions of cupric and cobaltous sulfate solutions were monitored by UV-visible spectroscopy using a Cary 3 spectrophotometer (Varian, Sunnyvale, CA). A typical volume of 200 l of a 0.1-0.3 mM dimeric enzyme solution was titrated with the appropriate metal sulfate solutions. CuZnSOD derivatives were made by slowly titrating in 1.9 -2.0 eq of zinc sulfate per dimer, stirring overnight, followed by titrating in 1.9 -2.0 eq of cupric sulfate. The CuCoSOD derivatives were made similarly with the substitution of cobalt sulfate for the zinc sulfate. The total volume of metal solution was no more than 20 l (0.20 -0.60 mM) or 10% of the total volume of apoprotein titrated. All titrations were in 100 mM sodium acetate, pH 5.5 buffer solution. End points were reached when the spectra of successive metal additions overlapped.
DEPC Studies-Stock solutions (0.5 M) of DEPC (Sigma, stored at 4°C, in a brown bottle, under argon) were prepared in 100 mM sodium phosphate buffer (pH 6.0) and stored on ice just prior to use. The reaction mixture included 98 l of SOD (12.5 M, 0.2 mM histidine based on a total of 16 histidines/dimer) and 2 l DEPC (for a final concentration of 10 mM in phosphate buffer). The following SOD derivatives were studied: apo, apo ϩ 2Zn 2ϩ , apo ϩ 2Zn 2ϩ ϩ 2Cu 2ϩ and as-isolated enzyme. Metallated protein samples were incubated with metal ions for at least 12 h prior to the addition of DEPC. The absorbance at 240 nm was monitored for 30 min; the reaction usually reached equilibrium by 20 min. The final absorbance of the DEPC-histidine adduct (⑀ 240 ϭ 3200 M Ϫ1 cm Ϫ1 ) was used to calculate the number of modified histidine residues. Results are reported with respect to the number of protected histidines for each of the WT and mutant derivatives. Measurements were made in triplicate, except for L38V, which was measured in SCHEME 1. The active site and monomeric subunit of human wild type CuZnSOD. The active site Cu(II) (lower left of one monomer) and Zn(II) (upper right) ions are shown, in stereo view, with their respective ligands in the crystal structure of a single monomeric subunit.
duplicate. The experimental error is ϳ5-10%, as indicated in the bar graphs in Fig. 5. DEPC has been used previously to measure the number of accessible histidine residues in CuZnSOD (30).

Metal Binding: A Quantitative Analysis of Metal Ion Content of As-isolated and Remetallated Human CuZnSODs-Wild
type, A4V, G93A, and L38V human CuZnSODs were expressed in Saccharomyces cerevisiae, purified, and demetallated to form apoprotein as described under "Experimental Procedures." The apoproteins were remetallated as follows: (a) with 2 eq each of Cu 2ϩ and Zn 2ϩ (i.e. 1 eq of each metal per subunit) to form CuZnSOD, (b) with 2 eq of Cu 2ϩ to form CuESOD, (c) with 4 eq of Cu 2ϩ to form CuCuSOD, (d) with 2 eq each of Cu 2ϩ and Co 2ϩ to form CuCoSOD. Previous spectroscopic metal titrations show that an end point is reached when 4 eq of metal ion are titrated into an apo-WT CuZnSOD solution (2,29), indicating relatively tight binding of the 4 available metal ion binding sites per dimer. The remetallated protein samples were washed with buffer to eliminate any free or loosely bound metal ions and analyzed by atomic absorption. Table II gives the copper, zinc, and cobalt contents of each of the as-isolated and remetallated (CuZn-, CuE-, CuCu-and CuCoSOD) derivatives for hWT, hA4V, hG93A, and hL38V.
The atomic absorption data indicate that the remetallated CuZn-and CuCoSOD derivatives each retained roughly 2 eq of copper and 2 eq of either zinc or cobalt after the washing procedure. The CuESOD derivatives retained roughly 2 eq of copper while the CuCuSOD derivatives were found to contain variable amounts of copper after the washing process. The wild type and A4V proteins retained most of the 4 eq that had been added, whereas G93A and L38V lost significant amounts of copper. Interestingly, the as-isolated wild type and mutant human CuZnSODs were all found to contain significantly lower amounts of copper than zinc (discussed below).
SOD Activity and Spectral Characterizations of Enzyme Metal Binding-The rate constants as a function of pH for the SOD-catalyzed disproportionation of superoxide were determined as described previously (38). Solutions of the enzyme (0.4 -1.2 M) were pulse-irradiated, and the disappearance of superoxide was followed at 250 -270 nm on a fast time scale (microseconds). The catalytic rate constants of the as-isolated human and mutant CuZnSODs and the remetallated hWT and mutant CuZn-, CuE-, CuCu-and CuCoSOD are plotted as a function of pH in Figs. 1a, 2a, 3a, and 4a. (Note that the rate constant calculations are based on the concentration of copper tightly bound to the enzyme rather than the concentration of the enzymes themselves, since only copper-containing enzymes are active; see "Experimental Procedures.") Figs. 1-4 (b-g) show the UV-visible spectra obtained from the titration studies. Conditions for each spectrum are reported in the figure legend.
Human Wild Type CuZnSOD Activities and Metal-binding Titrations-The hWT CuZnSOD is the control for comparison of the mutants. The noteworthy features are that the rate constants for as-isolated WT CuZnSOD and for remetallated WT CuZn-, CuCu-, and CuCoSOD each show the characteristic, pH-independent (pH 5-10) fast rate constants (ϳ1-2 ϫ 10 9 M Ϫ1 s Ϫ1 ) (see Fig. 1a). The high SOD activity indicates that all or nearly all of the copper in the CuZn-and CuCoSOD is as active as the copper in the copper site of fully active CuZnSODs. The same is true for half of the copper in the CuCuSOD derivative. The SOD activity of WT CuESOD was also high, but in contrast it was found to be pH-dependent, as reported previously for bovine WT CuESOD (32).
The spectral characteristics of the hWT are also used as the control for comparison with the FALS mutant enzymes. The UV-visible metal titration data for recombinant human wild type CuZnSOD expressed in and isolated from yeast is given in Fig. 1 (b-g). The UV-visible absorption spectra of the as-iso- lated derivative reveals a Cu(II) d-d transition peak at 680 nm, which is characteristic of Cu(II) in the distorted five-coordinate geometry found in wild type CuZnSODs. The characteristic ligand-to-metal charge transfer band assigned to the bridging imidazolate side chain of His 63 (26,39,40) is observed at 450 nm. In Fig. 1c the result of titrating with 2 eq of Zn(II) followed by 2 eq of Cu(II) is presented. Again the spectrum reveals Cu(II) in its characteristic native copper site geometry. The CuCuSOD titration in Fig. 1e clearly shows the first 2 eq of Cu 2ϩ binding to the copper site yielding an absorption peak at 680 nm and the final 2 Cu 2ϩ eq binding to the zinc site yielding a peak at 810 nm. The CuCoSOD titration is presented in Fig.  1 (f and g), where Fig. 1f represents the apoprotein with 2 eq of Co 2ϩ followed by Fig. 1g, where 2 eq of Cu 2ϩ have been added. The characteristic Co(II) peak due to Co(II) bound to the native zinc site is seen in the 500 -600 nm region, and the 680 nm peak corresponding to Cu(II) bound to the native copper site is observed. Fig. 1g represents the completely metallated Cu-CoSOD derivative of human WT.
SOD Activities and Metal Ion Titrations of Human Mutant G93A CuZnSOD-The rate constant as a function of pH is shown for G93A in Fig. 2a. The as-isolated enzyme retains a high amount of catalytic activity (1.2 ϫ 10 9 M Ϫ1 s Ϫ1 ), as does the CuCuSOD derivative. The high activity of these enzymes confirms that, when copper is in the native copper site, this mutant enzyme is as catalytically active as the WT enzyme. By contrast, the SOD activities of the remetallated CuZn-and CuCoSOD derivatives are considerably lower (4 -5 ϫ 10 8 M Ϫ1 s Ϫ1 ).
The UV-visible spectrum is shown for the as-isolated G93A mutant enzyme in Fig. 2b. The Cu(II) d-d band at 680 nm combined with the shoulder at 450 nm (characteristic of the imidazolate bridge; see above) suggest that the Cu(II) is in the native copper site and that the geometry is similar to that of the WT enzyme. By contrast, the remetallated CuZnSOD derivative of G93A shows a peak at 600 nm (Fig. 2c), shifted from the expected peak at 680 nm, indicating that the ligand environment of the Cu(II) is different than that in the as-isolated derivative and different from that of WT. The spectra for the remetallated CuESOD and CuCuSOD derivatives (Fig. 2, d  and e) are similar to one another in appearance (peak maxima at 610 nm), differing predominantly in the magnitude of the absorbance, suggesting that the environment of the copper ions is similar in both cases. This result is in sharp contrast to the spectra of the remetallated CuESOD and CuCuSOD WT derivatives, which show evidence for preferential binding of copper ions first to the copper sites and to the zinc sites only after the copper sites are occupied (compare Fig. 2, d and e, with Fig. 1,  d and e). In the case of the ECoSOD and CuCoSOD derivatives of G93A (Fig. 2, f and g), we know from the atomic absorption results that 2 eq of Co 2ϩ are bound to the protein dimer, but that the intensities of the Co(II) d-d bands are significantly lower than those of the comparable WT derivatives (compare with Fig. 1, f and g). This indicates that the Co(II) is bound in a different location or geometry in G93A (41,42). Fig. 3a shows the SOD activity profile versus pH for derivatives of A4V CuZnSOD. Once again, the as-isolated and remetallated CuCuSOD derivatives of A4V are as active as the human WT, while the remetallated CuZn-and CuCoSOD derivatives show low SOD activity (2-4 ϫ 10 8 M Ϫ1 s Ϫ1 ). Similar to G93A, the as-isolated A4V shows the 680 nm peak characteristic of Cu(II) in a site like that of the human WT native copper site whereas the remetallated CuZnSOD derivative does not, the Cu(II) d-d band having shifted to about 650 nm. The titration of apo-A4V with copper yielding CuE-and CuCuSOD (Fig. 3, d and e) shows simultaneous peak development at 680 and 810 nm. This simultaneous increase in absorbance suggests that copper is binding to the copper and zinc sites with similar affinities throughout the titration, in direct contrast to WT, which binds copper selectively first to the copper site and then to the zinc site (Fig. 1, d and e). In the case of A4V, unlike G93A, the final CuCuSOD spectrum is very similar to that of WT CuCuSOD (compare Figs. 1e, 2e, and 3e), suggesting that the geometries of the metal binding sites of these fully metallated 4-copper derivatives are also similar. As observed with G93A, the Co(II) d-d bands of ECo-and Cu-CoSOD derivatives of A4V (Fig. 3, f and g) have lower intensities relative to WT (Fig. 1, f and g), indicating that in the mutant enzymes a greater amount of Co(II) is bound in a different ligand geometry than in the WT protein. Fig. 4a shows the SOD activity profile of L38V derivatives versus pH. As was observed with the other enzymes studied, the activities of this as-isolated protein and the remetallated CuCuSOD derivative are relatively high (1.5-2.0 ϫ 10 9 M Ϫ1 s Ϫ1 ), while the remetallated CuCo-and CuZn-SOD show lower activity. In this mutant the activity is dramatically lower (1-2.5 ϫ 10 8 M Ϫ1 s Ϫ1 ). The corresponding UVvisible titrations are shown in Fig. 4 (b-g). The as-isolated derivative (Fig. 4b) again has a WT-like copper site d-d band at 680 nm. The remetallated CuZnSOD derivative (Fig. 4c), by contrast, has a Cu(II) band at 640 nm with a shoulder at 500 nm, suggesting that copper is bound in two different environments. The copper titration in Fig. 4 (d and e) shows a Cu(II) peak at ϳ570 nm. This peak increased in intensity throughout the titration to give first CuESOD and then CuCuSOD, similar to G93A. The ECo-and CuCoSOD derivatives (Fig. 4, f and g) did not show d-d bands characteristic of tetrahedral or fivecoordinate Co(II) geometry (41,42).

SOD Activity and Metal Ion Titrations of Human Mutant L38V CuZnSOD-
Histidyl Residue Modification Using Diethylpyrocarbonate-Human CuZnSOD contains 8 histidyl residues per subunit, i.e. 16/dimer. From the crystal structure of the WT CuZnSOD enzyme, it is clear that His 46 , His 48 , and His 120 are exclusively copper site ligands, His 71 and His 80 are exclusively zinc site ligands, His 63 bridges the copper and zinc sites, His 110 is on the surface, and His 43 is buried in the ␤-barrel and thus inaccessible to solvent. Bovine CuZnSOD also has 16 histidines/protein dimer, and previous studies have established that the histidyl modification reagent DEPC reacts with all 16 histidines of the apoprotein, whereas it reacts with only 2 histidines/dimer in the case of either CuZnSOD or EZnSOD (30). We have repeated these experiments for the human WT and FALS mutant enzymes, and the resulting data are summarized in Fig. 5, presented as the number of histidines protected from reaction with DEPC. For the apoproteins of hWT, G93A, A4V, and L38V, all 16 histidines were modified with DEPC, i.e. no residues were protected. For the 4 as-isolated enzymes, only 2 histidines/dimer reacted with DEPC, i.e. 14 were protected. For the remetallated WT derivatives CuZnSOD or EZnSOD, the results were identical to those that had been obtained earlier for the bovine enzyme, i.e. 14 histidines were protected from reaction with DEPC (30). Strikingly, for both the EZn-and the CuZnSOD derivatives of the remetallated FALS mutants, only 3-8 histidines were protected from DEPC modification. DISCUSSION Proper insertion of copper ions into CuZnSOD in vivo has recently been shown to require the presence of a "copper chaperone," termed CCS in the case of the copper chaperone for human CuZnSOD and termed Lys7 for S. cerevisiae (43)(44)(45)(46). This requirement was not immediately apparent when CuZn-SOD was first characterized because WT bovine and human CuZnSOD can be demetallated and remetallated correctly in vitro in the absence of the copper chaperone (29). However, the in vitro remetallation procedures that were developed to direct binding of the copper and zinc ions to their correct binding sites in the bovine and human apoproteins require conditions of low pH that are far from physiological conditions. At higher pH, Cu 2ϩ ions tend to bind pairwise to the copper and zinc sites in a single subunit, rather than binding exclusively to the copper sites (31). More recently, it has been recognized that yeast apo-CuZnSOD cannot be fully reconstituted in vitro using these same low pH procedures and that one "phantom" subunit remains devoid of bound metal ions (47). The studies described here demonstrate that the apoproteins of the FALS mutant CuZnSODs G93A, A4V, and L38V likewise cannot be properly remetallated in vitro using the same conditions that have been successful in the case of both bovine and human WT CuZnSOD.
Complete Filling of the Four Metal Binding Sites with Copper Gives High SOD Activity-The UV-visible spectra of the Cu-CuSOD derivatives of hWT, G93A, A4V, and L38V contain contributions from Cu 2ϩ in the copper sites and Cu 2ϩ in the zinc sites of each protein. High SOD activities of comparable values are observed for the remetallated CuCuSOD derivatives of WT, G93A, A4V, and L38V (see Figs. 1e, 2e, 3e, and 4e). If one assumes that this high SOD activity is dependent on a WT-like configuration of copper in the native copper site of these four proteins, then we can conclude that the Cu 2ϩ must be bound to the copper sites in a WT-like fashion. Therefore, the contributions of the copper site-bound copper to the resulting UVvisible spectra should be the same. The UV-visible spectra of the CuCuSOD derivatives of hWT and A4V are alike (compare Figs. 1e and 3e), suggesting that the copper ions in both the copper and zinc sites of A4V adopt the same geometries as the WT CuCuSOD. However, the UV-visible spectra of the CuCu-SOD derivatives of G93A and L38V are markedly different from those of hWT CuCuSOD, (compare Figs. 1e, 2e, and 4e). If one makes the assumption, based on SOD activities, that the Cu 2ϩ in the copper site is bound in the same manner for each protein, then one would expect the spectral contributions due to this copper site Cu 2ϩ to be equivalent. This leads to the conclusions that: 1) the differences in the UV-visible spectra come mainly from a change in the geometry of the Cu 2ϩ ions bound to the native zinc sites of G93A and L38V, and 2) the change of geometry at the zinc site in these cases has little or no effect on the SOD activity of the SOD-active copper bound at the copper site.
Half-filling of the Four Metal Binding Sites with Copper (CuESOD) Yields Dramatically Different Behavior for the WT and the Mutants-In an earlier study, the SOD activity of the CuESOD derivative of bovine WT CuZnSOD was found to be strongly pH-dependent between pH 5 and 8 (32), and it was concluded that high, pH-independent SOD activities between pH 5 and 9.5 for CuZnSOD derivatives were possible only when copper was bound in the copper site and another metal ion was bound in the zinc site. In the present study, we have confirmed that the same behavior holds for the CuESOD derivative of human WT CuZnSOD (Fig. 1a).
The CuESOD derivatives of G93A, A4V, and L38V CuZn-SOD, by contrast, have pH-independent SOD activities between pH 5 and 9.5, leading us to conclude that all of the SOD activity measured for these derivatives is derived from CuCu-SOD subunits, i.e. subunits containing copper bound in both the native copper and zinc sites. However, if all of the bound copper ions were distributed pairwise in subunits of the mutant CuESOD derivatives, the resulting SOD activities (on a molar basis of total active enzyme) would be expected to be half those of the mutant CuCuSOD derivatives, whereas the activities reported here are in fact substantially less than half (see Figs. 2a, 3a, and 4a). We conclude, therefore, that a substantial fraction of the copper ions occupy only the zinc sites in the mutant population of CuESOD and that they, like copper ions in zinc sites of WT derivatives such as AgCuSOD, do not confer any SOD activity (see Table I). In other words, the CuESOD derivatives of G93A, A4V, and L38V CuZnSOD contain, on average, one copper per subunit of enzyme, but the metal ions are distributed such that some subunits contain copper in both the copper and zinc sites (active enzyme) and some subunits contain copper only in their zinc sites (inactive enzyme). The UV-visible spectra of the CuESOD derivatives of G93A, A4V, and L38V CuZnSOD are very different from those of the same derivative of WT and thus are consistent with this interpretation, but the spectra cannot provide strong corroborative evidence because the characteristic UV-visible spectrum for Cu(II) bound at the zinc sites of the FALS mutants is unknown.
The CuZnSOD and CuCoSOD Derivatives Also Show Dramatically Different Behavior for the WT and the Mutants-The remetallated hWT CuZn-and CuCoSOD show similar, high, pH-independent SOD activities (Fig. 1a). By contrast, the remetallated CuZn-and CuCoSOD derivatives of G93A, A4V, and L38V CuZnSOD have SOD activities that are markedly reduced from those of the CuCuSOD derivatives of the same mutant proteins, although they maintain the characteristic pH independence found in hWT SOD. Since the mutants still bind approximately 2 eq of each metal (Table II), the reduced activity must result from improper metal placement. Using the decrease in SOD activity relative to that of WT hCuZnSOD as a measure of metal misplacement, we conclude that only 20 -25% of the copper is in the copper-binding site for the Cu-CoSOD derivative of G93A. The effect is more dramatic for the A4V CuCo derivative, where the decrease in activity suggests that only 10 -20% of the copper is in the copper-binding site, and is most marked in the L38V CuCoSOD where only 5-10% of the copper is in the copper-binding site. The UV-visible spectra of these derivatives reinforce this conclusion, especially in the mutant CuCoSODs where the characteristic absorbance at 580 nm, corresponding to the cobalt in the zinc site, is reduced in all three mutants to varying degrees.
The Copper Bound to the As-isolated Mutant and WT CuZn-SODs Has High SOD Activity, but the As-isolated Proteins Contain Less Copper than Expected-On a per copper basis, the SOD activities of the as-isolated mutant and WT CuZnSODs are maximal (Figs. 1a, 2a, 3a, and 4a). Moreover, the UVvisible spectra of the as-isolated proteins are all similar, with copper(II) d-d bands at ϳ680 nm and shoulders at ϳ450 nm. These bands have been previously assigned to an imidazolateto-Cu(II) charge transfer transition in bovine CuZnSOD (26,39,40), suggesting that the geometry about the copper ions in these proteins is the same. However, examination of the metal ion content of the as-isolated proteins revealed that the copper content was markedly low. As this is a human CuZnSOD expressed in a yeast cell with a yeast copper chaperone system (Lys7/yCCS), the low copper content may be a result of the mismatch between these two systems.
Histidine Modification Using DEPC Suggests Conformational Differences between the hWT and FALS Mutant Enzymes-Earlier studies of the reaction of DEPC with bovine CuZnSOD and its derivatives established, first, that all 16 histidines in the apoprotein react with DEPC; second, that 2 histidines/dimer in holo-CuZnSOD react with DEPC; and finally, that 2 histidines react with DEPC when 2 eq of zinc per protein dimer were bound to the zinc sites, despite the fact that the copper sites were empty and thus unprotected by metal binding (30). We repeated these experiments with the reconstituted and as-isolated human WT and G93A, A4V, and L38V CuZnSODs to ascertain whether the conformational stability of these proteins would result in similar patterns of histidine accessibility.
In the case of the as-isolated human WT and G93A, A4V, and L38V CuZnSODs, 2 histidines reacted with DEPC, consistent with the fact that they all bound stoichiometric amounts of total metal. However, in the case of the remetallated (2-zinc only and 2-zinc plus 2-copper derivatives) hWT and FALS mutant CuZnSODs, only the hWT showed the expected protection of 14 histidines from reaction with DEPC. Strikingly, the remetallated FALS mutant CuZnSODs were not well protected from reaction with DEPC. This lack of protection strongly suggests a conformational difference among the mutants upon the in vitro addition of metals.
Since the major difference between the as-isolated and remetallated FALS mutant CuZnSODs is the presence of copper in the zinc sites of the latter, we tentatively conclude that the zinc sites of each protein dimer must contain zinc, rather than copper, to achieve the maximum protection of histidines from reaction with DEPC. The WT bovine and human apoproteins bind zinc selectively to the zinc sites at pH 5.5 and thus achieve maximum protection from reaction with DEPC upon  addition of 2 eq of zinc alone. The FALS mutant proteins have lost this selectivity for zinc binding exclusively to the zinc site when Cu 2ϩ is also present. The DEPC results further support the global effects that are caused by the differences in metal binding of the FALS mutants compared with human WT CuZnSOD.
Implications for ALS?-The FALS mutant human apo-CuZnSODs reported here, i.e. G93A, A4V, and L38V, have been found to differ from wild type human apo-CuZnSOD in that the mutant proteins, unlike wild type, lack the ability in vitro to bind Cu 2ϩ or Zn 2ϩ selectively to the proper copper and zinc sites of the protein. However, it is important to remember that these dramatic alteration in in vitro metal ion binding selectivity of G93A, A4V, and L38V apoproteins relative to wild type may not be reflected in aberrations in metal ion binding in vivo, since metal ion binding to apo-CuZnSOD in vivo is mediated by a copper chaperone. Nevertheless, the similar behavior of the three FALS mutant apoproteins suggests that each of the individual mutations has exerted a similar effect on the global properties of the enzyme, possibly explaining how the more than 70 spatially scattered mutations on CuZnSOD can all lead to the same unknown property of the mutant enzymes that causes FALS.
We have also found that the alterations caused by these three FALS mutations do not diminish the inherent catalytic ability of the copper ion in the copper site to catalyze the SOD reaction, but they may well lead to an increase in the flexibility of the protein in the vicinity of the active site channel. Such increased flexibility could result in an enhancement in the reactivity of copper ions in the copper site with oxidants other than superoxide, e.g. peroxides or peroxynitrite, and it could also provide enhanced access of other oxidizable substrates to the active site as well. Our future studies of the mutant enzymes will address the nature in the changes in the structure, stability, and reactivities of the FALS mutant CuZnSODs in a continuing effort to look for solutions to the puzzle of what causes CuZnSOD-associated FALS.