Mechanism of Cu,Zn-Superoxide Dismutase Activation by the Human Metallochaperone hCCS *

The mechanism for copper loading of the antioxidant enzyme copper, zinc superoxide dismutase (SOD1) by its partner metallochaperone protein is not well understood. Here we show the human copper chaperone for Cu,Zn-SOD1 (hCCS) activates either human or yeast enzymes in vitro by direct protein to protein transfer of the copper cofactor. Interestingly, when denatured with organic solvents, the apo-form of human SOD1 cannot be reactivated by added copper ion alone, suggesting an additional function of hCCS such as facilitation of an active folded state of the enzyme. While hCCS can bind several copper ions, metal binding studies in the presence of excess copper scavengers that mimic the intracellular chelation capacity indicate a limiting stoichiometry of one copper and one zinc per hCCS monomer. This protein is active and unlike the yeast protein, is a homodimer regardless of copper occupancy. Matrix-assisted laser desorption ionization-mass spectrometry and metal binding studies suggest that Cu(I) is bound by residues from the first and third domains and no bound copper is detected for the second domain of hCCS in either the full-length or truncated forms of the protein. Copper-induced conformational changes in the essential C-terminal peptide of hCCS are consistent with a “pivot, insert, and release” mechanism that is similar to one proposed for the well characterized metal handling enzyme, mercuric ion reductase.

Eukaryotic and prokaryotic cells accumulate essential first row transition metals for various cellular functions. Copper for instance, is maintained at a total intracellular concentration in the 10 Ϫ4 -10 Ϫ5 M range. Despite this abundance, it has recently been shown that copper ions are typically unavailable in the cytoplasm for direct substitution into metalloenzymes. In fact, the steady state concentration of the free or labile form of copper ion in the cytoplasm of yeast is far less than one ion per cell (1). This finding suggests that the intracellular milieu has a significant overcapacity for metal chelation, and raises the issue of how apo-proteins acquire the correct metal cofactor.
A class of metal ion trafficking proteins is emerging that act in distribution of essential transition metal cofactors to specific targets in the cytoplasm, to the cell surface or integral membranes, or to various intracellular compartments. Several newly discovered metal-receptor proteins called metallochaperones are now known to serve as key agents in cellular trafficking of these cofactors (2)(3)(4). These cytosolic proteins are characterized by a metal-exchange partnership with one or more specific intracellular target proteins and by a metalspecific binding activity.
A prototypical member of the metallochaperone family, Atx1 of Saccharomyces cerevisiae, is now well characterized at the genetic, biochemical, and structural levels (5)(6)(7)(8). This cytosolic 8-kDa protein functions to deliver copper ions to a P-type ATPase cation transport protein (Ccc2) localized in membranes of trans-Golgi vesicles. A human homologue of Atx1, HAH1 (or Atox1), has also been identified (9). Crystallographic and NMR solution structures of Atx1 (10) 1 and HAH1 (12), as well as the cytosolic domains of their ATPase target proteins including the human Wilson and Menkes Disease proteins (13,14), reveal that they adopt a common "ferridoxin-like" ␤␣␤␤␣␤ fold. In each structurally characterized case, the metal ion is bound by two cysteine thiolate moeities in a highly conserved MXCXXC sequence motif. A mechanism has been proposed that allows for rapid metal exchange between the otherwise tightly bound Cu(I)-protein complexes involving formation and decay of twoand three-coordinate complexes of the Cu(I) ion with the cysteines of both proteins (5,7,12). Both acquisition and delivery of the copper ion between these proteins is expected to occur at the single copper-binding site, although this may not necessarily be the case for the more complex metallochaperones.
Physiological incorporation of copper into the eukaryotic antioxidant enzyme Cu,Zn-superoxide dismutase (SOD1) 2 requires the CCS metallochaperone (copper chaperone for SOD1) (1,15,16). Both yeast and human CCS proteins have been identified, as well as potential homologues in plants and insects (17,18). These 26 -30-kDa proteins possess an Atx1-like sequence at the N terminus, complete with the MXCXXC metal-binding motif, which is fused to a sequence homologous to its SOD1 target. The C-terminal region of CCS has high sequence homology with CCS proteins from other species, and also contains two highly conserved cysteine residues. While the latter region is essential for CCS activity in vivo, a requirement for the Atx1-like region is only apparent under copper-limiting conditions (17). The three regions correspond to domain I (Atx1-like), domain II (SOD1-like), and domain III (unique CCS sequence). Recent crystallographic structures for yeast and human CCS show that domain II from yeast and human fold in a "Greek key" ␤-barrel conformation that is quite similar to its target, SOD1 (19 -21). The x-ray structure of yeast CCS solved by Rosenzweig and co-workers (19) also reveals a fold for domain I that is similar to Axt1. To date, no structural data is available for either domain III or a copper-bound form of any CCS protein.
Based on a combination of genetic and biochemical experiments with yeast CCS, mechanistic functions for each of the three domains in yeast SOD1 activation have been hypothesized (17). An isolated peptide corresponding to domain III of yCCS is capable of binding Cu(I) ions and is hypothesized to function independently of domain I, which is thought to act in copper acquisition from the cellular source. Domain III is proposed to carry out insertion of the copper ion into apo,(Zn)-ySOD1 (17). Domain II, with highly conserved SOD1-like dimer interface residues (17,19,20), is assumed to provide a specific interaction with SOD1 to target the delivery of copper ions. Recent studies of the metal-binding environments in purified human and tomato CCS support the involvement of both domains I and III in metal ion binding (18,22). Although these studies contend that both domains bind metal ions simultaneously, the spectroscopic data on isolated CCS may describe only "transport stage" of the metallochaperone as it diffuses through the cytoplasm.
To elucidate the mechanism of SOD1 activation by CCS, we have investigated the biochemical basis of copper transfer for human CCS protein. Direct activation of both human and yeast SOD1 by human CCS is demonstrated by in vitro assays with denatured apo-SOD1. Intriguingly, the denatured human SOD1 enzyme always requires the chaperone for optimal SOD activity. Under the same experimental conditions, the yeast SOD1 can be activated by simple copper salts (1). Furthermore, conformational changes in the human CCS protein observed upon binding of copper support the proposal that domains I and III carry out independent functions, with domain III directing the exchange of copper ion to the active site of SOD1. The differences between the biochemistry of the human and yeast SOD1/CCS partnership thus reveal several new insights into how one family of metallochaperones execute the direct transfer of copper ions.

EXPERIMENTAL PROCEDURES
Human CCS (hCCS) and hCCS Domain Truncations-The gene for the full-length, wild-type human CCS protein was cloned into the expression vector pET24d (Novagen) and transformed into Escherichia coli strain BL21(DE3). Several liters of this transformed strain were grown to A 600 ϭ 0.6 in LB media with 40 g/ml kanamycin and induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside. Purification of hCCS was accomplished through freeze-thaw extraction with 0.2 mM 4-(2-aminoethyl)benzenesulfonyl fluoride protease inhibitor in the extraction buffer, followed by precipitation with 40% ammonium sulfate, salt removal with several buffer exchanges, anion chromatography through Uno-Q12 (Bio-Rad), and gel filtration chromatography through Superdex 75 26/60 column (Amersham Pharmacia Biotech). The purified full-length hCCS protein was confirmed by a mass of 28,906.4 Ϯ 4.7 Da determined by ESI mass spectrometry, which corresponds to the mass predicted for hCCS missing the N-terminal methionine residue (28,909.4 Da). In this report, residue numbering is maintained with methionine as residue 1, giving a sequence of 2-274 for the isolated protein we refer to as full-length human CCS. A correction factor for the Bradford protein assay, [ Human CCS polypeptides corresponding to the Atx1-homolous segment, residues 9-79 (referred to as the DI polypeptide), the SOD1-homologous region from Gly 77 -Lys 241 (DII polypeptide), or to a truncation missing the C-terminal-most domain Ala 2 -Lys 241 (DI, II polypeptide) were isolated as follows. DNA fragments corresponding to the appropriate regions of the human CCS gene were amplified by polymerase chain reaction and inserted into complementary sites NcoI and EcoRI in pET 24d overexpression vector (Novagen). The plasmid products were transformed into E. coli strain BL21(DE3). Cultures were grown to A 600 ϭ 0.6 in LB media with 40 g/ml kanamycin and induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside. Purification of hCCS DI, II, or DII was accomplished through freezethaw extraction followed by precipitation with 40% ammonium sulfate, salt removal with several buffer exchanges, anion chromatography through Uno-Q12 (Bio-Rad), and gel filtration chromatography through Superdex 75 26/60 column (Amersham Pharmacia Biotech). Purification of the domain I polypeptide was accomplished through freeze-thaw extraction followed by salt removal with several buffer exchanges and gel filtration chromatography through Superdex 75 26/60 column (Amersham Pharmacia Biotech). The masses of the purified hCCS polypeptides were confirmed by ESI mass spectrometry (hCCS-DI ESI/MS ϭ 7,555.6 Ϯ 0. 26 Analytical Gel Filtration Chromatography-Human CCS proteins were injected on Superose 12 gel filtration column (Amersham Pharmacia Biotech) calibrated with 100 M samples of protein standards. Nitrogen-purged and Chelexed 50 mM Tris, pH 8.0, buffer with 100 mM NaCl was used as the elution buffer. Standards included blue dextran (5000 kDa), bovine serum albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (30 kDa), and cytidine (243 Da).

Metal Incorporation of hCCS and hCCS Domain Constructs in Vivo-
The expression strains for full-length hCCS or hCCS domain polypeptides were grown as described above, with the exception that after 30 min of protein induction, the culture was spiked with CuSO 4 to a final media concentration of 1 mM. After 4 h of induction, the cells were separated by centrifugation and cell pellets were resuspended in Chelex-treated buffer (50 mM Tris, 200 mM NaCl, pH 8.0) to remove residual media and extracellular CuSO 4 . Purification of copper-bound proteins was accomplished through a combination of freeze-thaw extraction, 0.5% streptomycin sulfate treatment, 40% (NH 4 ) 2 SO 4 precipitation, and preparative gel filtration chromatography. All buffers used during these preparations contained fresh 1 mM DTT. Initial freeze thaw extraction buffers also contained 0.2 mM 4-(2-aminoethyl)benzenesulfonyl fluoride to inhibit protease activity. Ion exchange chromatography was removed from this purification procedure to avoid stripping of metal ions from the hCCS protein constructs.
Metal Incorporation of hCCS in Vitro-All stages of the in vitro metal complexation of purified hCCS were carried out under anaerobic atmosphere in a VAC Atmospheres glove box. Buffers were thoroughly treated with Chelex to remove all trace metal impurities prior to use. A solution of 80 M hCCS protein in 50 mM Tris, 200 mM NaCl, 10 mM DTT, 10 mM reduced glutathione (GSH), 10 mM histidine, pH 7.8, was mixed slowly with an excess of ZnSO 4 and Cu(I)(CH 3 CN) 4 PF 6 (final concentration of each was 240 M, corresponding to a 3-fold molar excess to hCCS protein monomer) followed by overnight incubation at 15°C. The excess metal ions, either free in solution or bound to nonspecific sites on the protein, were first removed by five exchanges with the same Tris/NaCl/DTT/GSH/histidine buffer in an Amicon ultrafiltration stir cell, followed by five exchanges with a buffer containing only 50 mM Tris, pH 8.0, to remove all metal-binding competitors from the protein solution.
Human and Yeast Cu,Zn-Superoxide Dismutase-Human SOD1, holo-wild type form, was kindly donated by the laboratory of I. Bertini, University of Florence. Yeast SOD1 was overexpressed in E. coli BL12(DE3) and purified as described previously (1). The apo-form of hSOD1 for CCS activation assays was prepared by first removing the bulk (about 80 -85%) of the metal by dialysis against 50 mM ETDA in 100 mM NaOAc, pH 3.8, followed by dialysis versus 100 mM MgCl 2 in 100 mM NaOAc, pH 3.8, and then 100 mM NaOAc, pH 5.5. Since the apo-hSOD was found to retain 10 -15% of its native activity, a further treatment was employed to reduce the residual activity of the SOD1 and therefore enhance the sensitivity for detection of SOD activation by CCS or copper controls. The "apo"-hSOD1 obtained after dialysis demetallation (3 mg/ml, 100 M) was incubated with a mixture of 10 mM EDTA, 1 mM BCS, and 20 mM sodium ascorbate in 15% acetonitrile, 10% methanol, and 0.03% trifluoroacetic acid for 3 h at 4°C. The apo, denatured hSOD1 protein was purified of the chelating reagents and residual metal by reverse phase high performance liquid chromatography separation through a Vydac C4 column with H 2 O/CH 3 CN/trifluoroacetic acid gradient elution. The protein-containing eluent was stripped of solvent under vacuum and resuspended in Chelex-treated potassium phosphate buffer (50 mM), pH 7.8, immediately prior to use in SOD1 activation assays.
The yeast SOD1 protein purified from E. coli contained only 10 -20% of its holo-metal content. This protein sample was therefore directly applied to the second stage demetallation/denaturation with organic solvents and high performance liquid chromatography clean-up for use in the activation assays.
SOD1 Activation Assays-Design of an in vitro assay for CCS activity poses several challenges. Based on EXAFS data and metal-complexation methods, the copper bound to hCCS is concluded to be in the reduced, ϩ1 oxidation state as applied in the assay. Given the possibility of copper-catalyzed air oxidation of the metal-binding cysteine thiol ligands, the metal-loading and transfer events are evaluated under strictly anaerobic atmosphere. Copper ion transfer to apo-SOD1 in this assay is conveniently assessed by the gain of SOD activity in classic enzyme assays, yet these require aerobic buffers to produce superoxide substrate. Another challenge inherent in determination of CCS function is the well documented ability of apo-SOD1 to acquire uncomplexed, or "free" copper ions from solution without the aid of a chaperone protein.
Because of this self-activating ability of SOD1 with copper ions from solution, additional considerations were made to distinguish between SOD1 activation by direct copper insertion mechanisms (facilitated activation), and SOD1 activation by simple co-equilibration mechanisms (incidental activation).
To address these challenges, an SOD1 activation assay for CCS function has been designed with partitioned anaerobic and aerobic stages in combination with the application of appropriate metal chelates in each stage (schematic illustrated in Fig. 5A). The copper source as either Cu(I)-glutathione complex or Cu-bound CCS was combined with ZnSO 4 and reduced GSH in one tube, the apo-SOD1 and BCS chelate competitor in another. Both solutions were prepared anaerobically in deoxygenated and Chelex-treated, 50 mM potassium phosphate buffer, pH 7.8, and the total copper concentration was normalized for all reaction samples. The two solutions were then combined and incubated in a heating block at 34°C for 2 h. An aliquot of each reaction mixture was extracted from the anaerobic chamber and diluted into an aerobic 50 mM potassium phosphate buffer, pH 7.8, for assay of SOD activity by standard in-gel assay with NBT staining (23) or cytochrome c assay (24). The aerobic buffers (both loading and running buffers for in-gel assay) included 1 mM EDTA to scavenge any Cu(II) formed from oxidation of any Cu(I) source.
Proteolytic Analyses of hCCS-The multidomain structure of hCCS was established by partial proteolysis with trypsin. A 400-l solution with 1 g/l apoZn-hCCS was proteolyzed by 1 g of trypsin in 50 mM Tris, pH 8.0, for 30 min at 15°C in an anaerobic chamber. Proteolysis was halted after 30 min with 2 l 6 M phenylmethylsulfonyl fluoride. Prior to extracting the sample from the anaerobic chamber for analysis by MALDI-TOF MS, 2 l of 1 M DTT solution was added to prevent disulfide cross-linking of cysteine thiols on the peptide fragments. A 3-l aliquot of the whole protein digest was added to a standard sinapinic acid/H 2 O/CH 3 CN/trifluoroacetic acid matrix solution for spotting on a MALDI-TOF sample plate. MALDI-TOF spectra of the digest mixture were collected with delayed extraction (300 ns) in linear mode on a Voyager DE Pro spectrometer from PE Biosystems. An internal standard of myoglobin (horse skeletal muscle, Sigma) was used in a parallel MALDI-TOF sample spot for accurate calibration of the digest fragment masses. Fragment masses were matched to predicted tryptic sequences of hCCS with the MS Digest program on the ProteinProspector web site authored by the UCSF Mass Spectrometry Facility.
To examine copper-induced conformational changes and localization of the copper-binding site in hCCS, limited trypsin proteolysis of E,Zn-hCCS, and Cu,Zn-hCCS was carried out with similar methods as above. For these studies, the trypsin digest was monitored as a function of time to elucidate relative proteolytic rates. Aliquots of the proteolysis mixture were extracted and terminated with phenylmethylsulfonyl fluoride at digest times ranging from 5 min to 3 h. For examination of metal content in the proteolytic fragments, the digest aliquots were mixed in a matrix solution without trifluoroacetic acid.
Thiol Quantification/Mapping-5,5Ј-Dithiobis(nitrobenzoic acid) assay for quantifying the reduced cysteine thiols of E,Zn-hCCS was carried out according to the published procedures of Riddles et al. (25). Localization of the thiol and disulfide cysteines was accomplished by first alkylating a 20 M hCCS protein sample (no DDT, anaerobically) with 10 mM iodoacetamide in 6 M guanidine-HCl, 50 mM potassium phosphate, adjusted to pH 7.2. After removal of excess iodoacetamide and guanidine-HCl with several buffer exchanges in an Amicon ultrafiltration stir cell, the alkylated protein was subject to total proteolysis with 10 g/l trypsin at 34°C for 20 h. The alkylated cysteine residues, indicating those not involved in disulfide linkages on the original protein, were identified with MALDI-TOF MS by a mass increase of 58 Da for the predicted mass of a specific tryptic peptide. Disulfide-linked peptides were first identified by a covalent combination of two predicted peptide masses and then confirmed by the disappearance of the combined-peptide mass signal in a digest sample treated with DTT.

RESULTS
CCS Notation-The metal-complexed forms of human CCS are abbreviated in this report according to a similar system as that used for Cu,Zn-SOD (26). Unlike the yeast form of the CCS protein, human CCS binds both copper and zinc. We have designated the copper-deficient protein as E,Zn-hCCS rather than apo-hCCS, and the copper-occupied as Cu,Zn-hCCS rather than holo-hCCS. The Cu,Zn-hCCS designation does not represent the protein with copper bound to the potential SOD1 site within domain II of hCCS. As shown below, the copper ions in Cu,Zn-hCCS in fact bind to sites other than in domain II.
Human CCS Consists of Three Protein Domains in Solution-The human CCS protein consists of three proteolytically separable domains in solution that closely correlate to the independent domain structure predicted for yCCS (17,19). MALDI-TOF mass spectrometry of a limited trypsin digest mixture of hCCS isolated with no bound copper (E,Zn-hCCS form) reveals three protease-resistant fragments corresponding closely to the predicted Atx1-like domain I (dI, residues 2-76), SOD1-like domain II (dII, residues 77-241), and the fusion of domains I and II (dI,II, residues 2-241) (Fig. 1). Domains determined by proteolytic susceptibility are denoted by lowercase (i.e. dI,dII, etc.) while expressed and purified proteins that correspond to given domains are denoted by the uppercase (i.e. DI). For example, the expressed protein corresponding to the Atx1-like region of hCCS was constructed for amino acids 9 -79 while the observed trypsin fragment of hCCS corresponding to the domain I consists of amino acids 2-76. Both position 76 and 79 are within the expected 8 -12 amino acid junction between domains I and II of hCCS based on the crystallographically characterized yeast protein (19). The trypsin-accessible domain junctions in hCCS are also equivalent to the trypsin cleavage sites observed for yeast CCS protein (17).
Unlike domains I and II, the third domain (dIII) is rapidly cleaved from the human CCS protein with either trypsin or chymotrypsin proteases, indicating that this C-terminal peptide segment has little conformational stability in the absence of copper. There are only two trypsin sites near the dI-dII junction (Fig. 1A): site A is adjacent to Arg 71 and site B follows Lys 76 . Under a variety of proteolytic conditions, hCCS is preferentially cleaved at Lys 76 between dI and dII, although the consistent detection of a small portion of domain II fragment with residues 72-241 indicates appreciable cleavage at Arg 71 . Trypsin proteolysis at Lys 76 in hCCS is homologous to the cleavage junction between domains I and II that we previously reported for yCCS. In tryptic digests of E,Zn-hCCS the stable dII and dI,II fragments both terminate with the Lys 241 proteolytic site (Fig. 1A, site C).
The hCCS protein (E,Zn-hCCS) was also analyzed for cysteine disulfide linkages with a combination of thiol quantification by 5,5Ј-dithiobis(nitrobenzoic acid) assay and sequence localization with iodoacetamide alkylation and tryptic peptide mapping. Prior to examination of thiol/disulfide content, samples of freshly purified hCCS were transferred into a anaerobic glove box and DTT (typically at 5 mM concentration) reductant that was used in all buffers during purification was removed by several buffer exchanges in an ultrafiltration stir cell. 5,5Ј-Dithiobis(nitrobenzoic acid) colorimetric assay of the guanidine-denatured hCCS revealed 5-6 thiols per protein from the total 9 cysteines residues in hCCS. Alkylation of all free thiols in hCCS with iodoacetamide followed by MALDI-TOF MS of a total tryptic digest allows the localization of cysteine thiol and disulfide sequence sites. The expected disulfide in the SOD-like domain II (between Cys 141 and Cys 227 ) is identified by a covalent combination of peptides 113-163 and 225-232 for assignment of a fragment mass signal at 6,301 Da (with alkyl groups for the two nondisulfide cysteines in the 113-163 peptide). No other disulfide-linked peptides were detected between separate tryptic peptides, however, a significant population of peptides 2-23 and 113-163, each containing three cysteine residues, were identified with only one alkylated cysteine, implicating possible intra-fragment disulfides. A dominant mass signal of 1,769 Da for peptide 242-255 with two alkyated cysteines indicates that the cysteines of the CXC motif of hCCS-dIII remain entirely in the reduced state.
In addition to the full-length hCCS protein, the biochemical properties of polypeptides corresponding to isolated domains were examined. Human CCS domain I, domain II, and domain I,II proteins were each expressed and purified from E. coli to a yield of ϳ20 mg/liter LB media. In contrast to the yeast CCS protein, both the copper-depleted and copper-bound forms of full-length hCCS exhibit an apparent molecular mass of 61.1 kDa corresponding to the dimeric species in analytical gel filtration chromatography experiments (Table I). The isolated hCCS-DII and hCCS-DI,II proteins elute with apparent molecular masses of 33.2 and 46.7 kDa, respectively, also near the expected values for dimeric species. The isolated hCCS-DI protein, however, migrates as a monomeric polypeptide at an apparent molecular mass of 9.7 kDa.
Metal Binding Properties of hCCS-Human CCS protein is isolated with one equivalent of zinc (0.95 eq) per protein monomer and no significant amount of copper (0.04 eq) when E. coli expression is carried out under standard growth conditions (no metal supplementation of LB media). Isolated constructs of hCCS-dI,II (DI,II) and hCCS-dII (DII) also contain 1 eq of zinc per protein monomer (0.98 eq for DI,II and 0.92 eq for DII), whereas hCCS-dI does not. As additional evidence for the localization of the zinc ion in expressed full-length hCCS, the MALDI mass spectrum of a limited tryptic digest of E,Zn-hCCS in matrix solution without trifluoroacetic acid shows a significant ϩ65 Da adduct on both ϩ1 and ϩ2 mass signals for the dII fragment, whereas no such adduct is seen for the dI fragment (Fig. 2).
Copper loaded forms of hCCS were obtained in two ways, one of which involved expression of the human gene in E. coli grown in copper supplemented medium. Supplementation of the growth media with 1 mM final concentration of CuSO 4 during hCCS expression yields protein bound with 1-1.5 eq copper per isolated protein monomer. Similar in vivo copper binding techniques applied during expression of hCCS-dI and hCCS-dI,II resulted in comparable copper contents in the isolated proteins. In contrast, isolated hCCS-dII expressed with CuSO 4 did not retain a significant amount of copper.
The metal binding capacity of hCCS was also examined in vitro with the isolated protein. Since copper ions can readily occupy many types of sites, including those that are physiological zinc-binding sites, a competitive introduction of an equal amount of each metal was introduced to hCCS and incubated with excess small-molecule metal ligands that also act as potent competitors. Anaerobic incubation of E,Zn-hCCS with 3 eq each of Cu(I)(CH 3 CN) 4 PF 6 and ZnSO 4 was carried out under stringent competition with excess DTT (10 mM), histidine (10 mM), glutathione (10 mM), and NaCl (200 mM), followed by several buffer exchanges to remove the unbound metal ions and nonprotein reagents. After thorough buffer exchanges with and without competitor reagents, the hCCS protein retained slightly more than 1 eq of copper (1.2 Cu per protein monomer) and nearly 1 eq of zinc ion (0.90 Zn/protein monomer). Gel filtration chromatography of copper loaded forms of Cu,Zn-hCCS prepared by either in vitro or in vivo methods indicated that most of the protein remained a dimer, although a small fraction repeatedly migrated with a higher apparent mass (molecular mass ϭ 118 kDa) consistent with a tetrameric hCCS species (molecular mass ϭ 115.6 kDa) (Fig. 3). The higher apparent mass fractions were found to contain approximately twice as much copper as monomer protein, whereas the dimeric fractions contained closer to 1 eq of copper per protein monomer. Zinc concentrations in these fractions followed a 1:1 Zn: monomer protein stoichiometry throughout all gel filtration fractions. The higher and lower mass fractions were collected separately for the SOD1 activation assays described below.
Samples of in vitro Cu-loaded hCCS (1.2 equivalents per monomer) were separated into domains by limited trypsin proteolysis and analyzed by MALDI-TOF MS in the absence of added acid. Under these conditions, metal-protein interactions were observed. As in the MALDI-TOF mass spectrum of trypsin digested E,Zn-hCCS, the spectrum for the trypsin digest of Cu,Zn-hCCS reveals a zinc adduct for the domain II fragment and no additional adduct for copper (Fig. 2B). Control samples of holo bovine-SOD1 indicate that both the copper and zinc complexes with the protein can be observed by MALDI-TOF   (Fig. 2C). Therefore, none of the entire 1.2 eq of bound copper in the in vitro reconstituted hCCS protein is bound in the canonical copper site of SOD1. It is thus presumed that all the copper is complexed in sites that incorporate residues of domain I and/or III.

Copper-induced Conformational Changes in hCCS-
A clear difference in hCCS protein conformation was observed with comparison of time-resolved proteolytic maps of E,Zn-hCCS and Cu,Zn-hCCS. As we have reported for the yeast CCS protein (17), domain III is more slowly cleaved from hCCS by either trypsin or chymotrypsin when copper is bound to hCCS (data not shown). Two specific copper-induced conformational changes in hCCS are observed by comparison of tryptic maps from E,Zn-hCCS and Cu,Zn-hCCS (Fig. 4). First of all, cleavage after Arg 71 is significantly inhibited upon copper loading, implicating that this segment of hCCS is likely to be important in copper manipulation by domain I. The second difference in the tryptic profiles is the appearance of a stabilized 77-255 fragment in the time-resolved digest of Cu,Zn-hCCS. This fragment, corresponding to dII and much of dIII, is never a dominant species in the time-resolved digest of E,Zn-hCCS (trypsin cleavage at site D, Fig. 1A). The stabilized portion of hCCS-dIII with copper binding includes the CXC potential metal binding motif and the segment of dIII with highest sequence homology to CCS proteins from various other organisms (17).
Human CCS Activates Human and Yeast SOD1 in Vitro-An SOD1 activation assay was designed to measure the activity of the purified hCCS metallochaperone in vitro. The assay procedure is summarized by the scheme in Fig. 5A and described in detail under "Experimental Procedures." The ability of the various copper donors to activate apo-hSOD1 in this reaction system was assessed by the increase in SOD1 activity as measured by the standard native PAGE, in-gel NBT qualitative assay and by the more quantitative cytochrome c kinetic assay. Both in vivo and in vitro reconstituted Cu,Zn-hCCS can activate hSOD1 in this in vitro assay in the presence of excess chelate competitor, BCS (Fig. 5). As expected from previous studies with yCCS, the reduced glutathione complex of Cu(I) (Cu-GSH) is incapable of hSOD1 activation with BCS competitor present. Furthermore, the visible absorbance of the Cu-GSH/apo-hSOD1 reaction mixture with BCS at 483 nm (⑀ 483 ϭ 12, 250 M Ϫ1 cm Ϫ1 for Cu(BCS) 2 (27)) yields a concentration for Cu(BCS) 2 complex of 9.7 M, indicating that essentially all the Cu(I) ion supply from the original Cu(I)-GSH complex was acquired by BCS rather than the apo-SOD1 target. Cu,Zn-hCCS collected from gel filtration fractions at either the apparent dimeric mass or tetramer mass (Fig. 3) activated apo-hSOD to comparable levels. Cu,Zn-hCCS prepared by in vitro copper loading of the isolated protein also activated apo-hSOD to a similar extent (not shown).
A significant difference between human and yeast SOD1 enzymes in this activation assay is the low activity that resulted from apo-hSOD1 exposed to Cu-GSH complex without BCS competitor present (Fig. 5, B and C, sample 3). Yeast SOD1 that had been inactivated through denaturation, demetallation, and subsequent high performance liquid chromatography purification demonstrated the ability to reactivate with an unchallenged Cu-GSH supply to a level comparable to that from Cu-yCCS activation (1). However, human SOD1 that was inactivated with similar denaturing procedures for these experiments did not show this ability to efficiently reactivate with a simple Cu(I) complex.  Table I). The protein eluted as two species, tetrameric and dimeric. The fractions containing each species were confirmed as hCCS by SDS-polyacrylamide gel electrophoresis and the protein content was measured by Bradford assay using a standard correction factor. Copper concentrations were measured by graphite furnace atomic absorption (GFAA) spectroscopy and zinc concentrations were measured by ICP AES. Human CCS protein concentrations are displayed as molarity of protein monomer. To further test the correspondence of these biochemical results to features of the CCS activity observed in the cell, hCCS was tested for the ability to activate the yeast SOD1 (ySOD1) enzyme in vitro. As was previously shown by yeast complementation studies in vivo (15), Cu,Zn-hCCS is capable of activating this nonphysiological partner protein in vitro (Fig. 6). Efficient ySOD1 activation by Cu,Zn-hCCS in the presence of excess BCS demonstrates that this activation must also occur via direct transfer of copper ion.
Human CCS Itself Does Not Have SOD Activity-The remarkable similarity of the second domain of hCCS to the SOD1 target (15,20) raises the possibility that this metallochaperone itself can act as an SOD catalyst. Recent studies from Culotta and co-workers (28) have shown that lysates from yeast expressing wild-type hCCS do not show SOD activity attributable to the hCCS protein, although a D201H mutation of hCCS yields an SOD-active protein (28). SOD activity assays of both purified Cu,Zn-hCCS and purified hCCS-DII incubated with CuSO 4 and ZnSO 4 in vitro indicate that neither are SOD active (Fig. 7), corroborating the in vivo observations. It is thus unlikely that the SOD1-like domain of hCCS functions physiologically as an SOD-like enzyme.

Human CCS Activates SOD1 by Direct Insertion of the Copper Ion
Cofactor-The in vitro activation results demonstrate direct transfer of copper from the hCCS metallochaperone to hSOD1 and furthermore, reveal that transfer occurs through an intermediate in which the copper is bound by both proteins. When low M r copper donors such as the copper-glutathione complex are mixed with apo-SOD1 in the presence of stringent copper chelating agent BCS, any released or loosely bound copper ion is quickly scavenged in this reaction system. Based on this control, it is clear that any copper released to solution by Cu,Zn-hCCS would also be effectively trapped as the Cu(BCS) 2 complex. We conclude that activation of hSOD1 by Cu,Zn-hCCS in this in vitro reaction system must occur without release of copper by either protein during the transfer process. A direct transfer process has also been shown in yeast. This allows for protection of the cofactor from competing intracellular chelators that otherwise maintain free copper concentrations below the level of one atom per cell (1).
The lack of significant activation of hSOD1 by the Cu-GSH in the absence of BCS competition stands in contrast to the yeast SOD1 system (1). As with most SOD1 enzymes, hSOD1 demetallated by nondenaturing methods such as dialysis against excess EDTA is readily reactivated by simple copper and zinc salts. However, hSOD1 cannot be reactivated in this manner when it has been denatured with organic solvents and reducing agents. This unexpected result implies that the complete function of the human CCS metallochaperone may be more complex than copper delivery alone. The human SOD1 enzyme is notably less stable to standard chloroform-ethanol purification procedures than the yeast and bovine forms of the protein (29,30). Given that organic solvents were employed in these experiments for optimal inactivation of the apo-enzyme prior to application in the in vitro assay, it is likely that apo-hSOD1 prepared in this manner cannot refold properly upon addition of copper salts alone. The fact that Cu,Zn-hCCS is able to activate this pre-denatured form of apo-hSOD1 may indicate that hCCS serves an additional mechanistic role. This raises the possibility that hCCS also functions to stabilize an optimal folding state of hSOD1 or catalyze the formation of the conserved disulfide within the hSOD1 enzyme.
Chemical mechanisms for direct copper transfer from a metallochaperone have been reported for Atx1 and its physiological partners (5,7,10,12). We have previously proposed that CCS metallochaperones transfer copper ions to apo-SOD1 through a similar series of ligand exchange steps wherein the metal is never required to be released into solution as a free ion (1,17,19). Below we discuss two aspects of how the direct transfer of copper ion might occur: 1) the location and ligand environment of the copper bound to CCS, and 2) the orientation of this copper site relative to the active site in SOD1 when docked to CCS.
Interaction of Copper Ion with Human CCS-The copper binding and proteolytic protection data for hCCS are consistent with the two-hand mechanism of copper manipulation previously proposed for yeast CCS (17). Metal binding experiments indicate that in the presence of stringent physiological competitors, copper can bind to the cysteine motifs of both domains I and III. Furthermore, structural studies indicate that the substitution of an aspartate for a single SOD1-homologous copper ligand (histidine) may greatly decrease the affinity of the dII site in hCCS for copper ions (20).
Data for the copper bound state of human CCS described here corroborates our earlier results with yeast CCS (17), and are also consistent with the conclusions from Co(II)-substitution experiments of tomato CCS (18) and Cu-EXAFS studies human CCS (22); however, there remains a question of whether hCCS carries only one copper ion at a time, or two (perhaps more) copper ions as seen in some samples prepared for EXAFS studies (22). Although observations from all of these studies are consistent with the involvement of both domain I and domain III in copper ion manipulation by hCCS, none resolve the possibility that CCS proteins may adopt several different copper-binding modes to carry out the individual stages of its function including: (a) copper-acquisition from source, (b) protection of copper during transport, and (c) insertion of copper into hSOD1 target (Scheme 1).
Key insights into the mechanism of hCCS interaction with copper ions are also revealed by the differential proteolytic protection experiments. First of all, protection of Arg 71 upon copper binding to hCCS is consistent with a movement of this residue to a site near the metal ion. A similar copper-induced change is observed for the homologous residue (Lys 65 ) in the NMR structural studies of the Cu(I) complex of Atx1 (11) and the x-ray structure of Hg-Atx1(10). This highly conserved basic residue is found in both families of copper chaperones but is substituted with a hydrophobic aromatic residue in homologous domains of the copper-acceptor proteins (13). Furthermore, mutation of this residue can disrupt the function of Atx1 in vivo (8). These observations have led us to conclude that the basic residue at this position can protect the copper before the protein docks with partner and play a switching role in the release steps (12). Without a metal ion bound at this site in dI, the loop containing Arg 71 must either be significantly more flexible in conformation or oriented in a more solvent-exposed position. The occlusion of Arg 71 backbone induced by metal ion binding to this site suggests similar movement to that observed in the apo and Cu-Atx1 solution structures (11).
Copper Binding Alters Domain III Structure-Although very little is known about the mechanistic function of the third domain of CCS, it is essential to its activity in vivo. Copper binding to hCCS induces stabilization of the dII-dIII junction along with a portion of domain III extending through the CXC motif (Fig. 4). Interestingly, this stabilization does not require the presence of dI, suggesting that dIII is capable of binding copper independently of dI. This observation may also reflect an association of the CXC-containing portion of dIII with dII upon copper binding. Given the high sequence homology and very similar overall structure between hCCS-dII and the hSOD1 target (20), we speculate that such an interaction of dII and dIII upon copper binding emulates the interaction of copper-loaded dIII with the SOD1 target enzyme. In this model, dIII would be the sole portion of the CCS protein responsible for insertion of the copper ion into SOD1. This copper insertion mechanism is further supported by in vivo studies in yeast that showed a yCCS-dII,III truncated protein is still capable of ySOD1 activation in vivo, whereas the yCCS-dI,II is not (17). Models of a docked yeast CCS⅐SOD1 complex reveal that a domain III extension from the C terminus of domain II to the CXC motif is capable of spanning the distance to the CXXC copper site in domain I as well as to the SOD1 copper active site (19).
Precedent for Metal Insertion by Cysteine Residues of a Cterminal Domain-Enzymatic and structural studies of Hg(II) transfer into the flavin-containing active site of mercuric ion reductase (MerA) (31-33) provide a framework for understanding metal transfer between CCS and SOD1. While this protein shows little overall homology to CCS, analysis of its domain structure reveal several striking similarities (Fig. 8A). Like CCS, biochemical and structural studies of MerA distinguish three separate regions in the protein structure. The N-terminal most domain is homologous to MerP (35% identity), which has the same fold as Atx1 and includes the MXCXXC metal binding motif. The next domain or "core" is highly homologous to glutathione reductase and has a pair of mercury-binding cysteine residues in the active site. Finally, the homodimeric protein has a 15-amino acid C-terminal extension from the core of one monomer that is embedded into the mercury-binding active site of the partner monomer (Fig. 8, B and C) (34). As with CCS, the N-terminal domain is not essential for activity and the Cterminal extension contains a pair of highly conserved cysteines that are essential for activity. In the case of MerA, these residues are involved in steady state turnover of the MerA enzyme (35). This C-terminal domain of MerA exhibits significant similarities to the CCS-dIII sequence, including a conserved spacing from the end of the large second domain to the proposed metal binding motifs: Cys-X-Cys for copper in CCS or Cys-Cys for mercury in MerA (Fig. 8B).
The function of the C-terminal protein domain of MerA in metal ion transfer has been demonstrated most recently in a study by Engst and Miller (31) that concluded the C-terminal cysteines facilitate the movement of Hg(II) into the buried active site of an adjacent monomer (Fig. 8D). If the cysteine residues in the C-terminal domain of MerA are mutated, cells lose resistance to mercury and only Hg(II) complexes with small ligands such as chloride can serve as substrates. Even low molecular weight exogenous thiols are too bulky and prevent entry of the Hg(II) into the active site of truncated MerA. In contrast, the full-length protein can readily obtain Hg(II) from most donors, presumably by initial binding to the CC motif in this C-terminal extension which subsequently hands off the metal to the active site thiols. The CXC motif of domain III of CCS is likely to play a similar role in delivering copper to the buried active site of SOD1. These and other features of the biochemistry, metal binding, structure, mutagenesis, and function of the Hg-MerA homodimer provide precedents for the mechanism of CCS proposed below.
Proposed Mechanism of SOD1 Activation by hCCS-We have established that hCCS activates either human or yeast apo-SOD1 by direct insertion of the copper ion cofactor, and propose further that the transfer process occurs through the cysteine pair of domain III via ligand exchange with copper-binding residues of the SOD1 active site. Since the CXXC region of domain I can also be involved in copper binding to hCCS and is proposed to act in the acquisition of copper ions from an unknown source (17), domain III could interact alternatively with the dI site and the SOD1 target site in a presumed switch-like translocation of the copper ion (Fig. 9A). A key issue yet to be resolved is the orientation of interaction between the CCS and SOD1 proteins that allows this direct copper transfer to occur. One significant difference in the solution properties of the human and yeast CCS proteins is that hCCS is a dimer regardless of copper content or protein concentration. The apo-form of yeast CCS protein is monomeric, but forms a mixture of monomers and dimers upon binding copper ion (17). Despite the difference in oligomerization state of the metallochaperone itself, hCCS is able to act as yCCS in activation of ySOD1 in the in vitro assay. Taken together, these observations leave open two mechanistic possibilities after the homodimeric Cu,Zn-hCCS encounters the homodimeric apo-hSOD1: 1) Cu,Zn-hCCS simply activates apo-hSOD1 enzyme through a dimer-to-dimer transfer of copper ion, or 2) the Cu,Zn-hCCS and apo-SOD1 dimers "swap" monomers to yield a pair of heterodimers or a heterotetramer prior to copper transfer (17,18,20,21) (Fig.  9B). In the first scenario, disruption of the stable dimeric form of apo-SOD1 is unnecessary, and localization of a dI-dIII cocomplex of copper can be envisioned at a site adjacent to the copper-binding site of SOD1 in a model tetrameric complex (21). The second mechanistic possibility is supported by the conservation of residues responsible for SOD1 dimerization in all CCS proteins, and modeling studies demonstrating that dIII can access both dI and the active site of the adjacent SOD1 in a docked heterodimer that uses this interface (20).
The thermodynamic stability of the interaction between SOD1 monomers has been argued to support the first scenario, yet it is not necessarily grounds for eliminating the second. Although the interface between monomers of SOD1 is known to be very strong in a thermodynamic sense, exchange of monomers between dimeric proteins can nonetheless be kinetically facile, especially if the two have similar dimerization interfaces. It has been shown that tight homodimers of similar Greek key ␤-barrel proteins can readily exchange monomers under physiological conditions. For example, the ␤2and ␤3crystallin proteins of the mammalian eye lens have been shown to display this monomer-swapping property, even though their homodimeric interactions have dissociation constants measured in the micromolar range (36). Furthermore, the dimeric interaction in SOD1 is known to be significantly weakened with removal of the metal ions as well as with reduction of the conserved disulfide (11). The necessity for the involvement of metal binding residues of domain I in direct ligand exchange with residues of the SOD1 active site is also unfounded. This proposed exchange is incompatible with the ability of the yCCS lacking domain I to activate ySOD1 in vivo (17). The requirement for dI of yCCS is only observed when copper concentrations are limiting in the growth media. In addition, a close inspection of the putative CCS sequence from Drosophila melanogaster (gene product CG17753) reveals that this protein lacks the CXXC motif in domain I while maintaining high homology to hCCS throughout the remaining portions of the protein.
Given the data in hand, we favor the second mechanism wherein monomers are swapped prior to copper transfer and the resulting heterodimer interface is established that involves the most highly conserved residues between SOD1 and CCS (19,20). The domain III peptide in this docked complex will have access to, but not does not require the presence of, the metal-binding site of domain I. A pivoting action at the domain II,III junction would allow movement of this key metal binding peptide to the active site of the adjacent SOD1 enzyme active site. Repositioning of the copper in the CXC motif of CCS domain III proximal to a histidine residue in the SOD active site could then account for the direct ligand exchange transfer of copper from metallochaperone to target. The MerA system provides precedence for the extension of a metal-binding domain from one protein into another and leads to several testable postulates. The first conserved aromatic residue in the third domain of each protein (corresponding to Phe 237 in hCCS or Phe 619 in RC607 MerA) is positioned to serve as an anchor FIG. 9. Proposed mechanism of CCS activation of SOD1. A, proposed mechanism of CCS copper ion insertion into apo-(E,Zn)SOD1. B, possible monomerswap or rearrangement mechanisms for the docked transfer of copper from human CCS to human or yeast SOD1 utilizing the conserved dimer interface between CCS-dII and SOD1.
around which the remainder of the flexible C-terminal peptide might pivot and extend into the active site of the adjacent monomer (Fig. 9A). In the case of CCS, the dIII peptide would bind copper, pivot toward the adjacent partner, insert into the active site in a "pivot, insert, and release" mechanism. It remains to be seen how the copper might be released from a thiol-rich site in domain III to the nitrogen-rich site in SOD1.
Conclusions-Results from these biochemical investigations on the human CCS protein suggest that this metallochaperone can adopt several conformations to fulfill its role in delivery of copper to SOD1. The "copper acquisition state," wherein CCS obtains copper from a yet to be identified donor, most likely involves domain I since copper-rich growth conditions render this domain unnecessary in vivo (17). The "transiting state," in which CCS protects the copper from metal-binding scavengers in the cellular environment, most likely corresponds to the copper loaded forms of the holo protein examined in these in vitro studies. Here, in the absence of the donor or target enzyme, the copper is most likely bound by the cysteine motifs of both dI and dIII. Last, a "copper insertion state" is induced upon capture of the target protein which releases the copper ion to dIII and subsequently to the SOD1 active site. The observed conformational changes upon copper binding to hCCS and the mechanistic precedents in the analogous MerA metalloprotein support a proposal that dIII is independently responsible for the ligand-exchange transfer of copper to the SOD1 active site.