The Copper Chaperone CCS Directly Interacts with Copper/Zinc Superoxide Dismutase*

Dominantly inherited mutations in the gene encoding copper/zinc superoxide dismutase (SOD1) result in the fatal motor neuron disease familial amyotrophic lateral sclerosis (FALS). These mutations confer a gain-of-function to SOD1 with neuronal degeneration resulting from enhanced free radical generating activity of the copper present in the mutant enzyme. The delivery of copper to SOD1 is mediated through a soluble factor identified as the copper chaperone for SOD1 (CCS). Amino acid sequence alignment of SOD1 and CCS reveals a striking homology with conservation of the amino acids essential for mediating SOD1 homodimerization. Here we demonstrate that CCS and SOD1 directly interact in vitro and in vivo and that this interaction is mediated via the homologous domains in each protein. Importantly, CCS interacts not only with wild-type SOD1 but also with SOD1 containing the common missense mutations resulting in FALS. Our findings therefore reveal a common mechanism whereby different SOD1 FALS mutants may result in neuronal injury and suggest a novel therapeutic approach in patients affected by this fatal disease.

nisms whereby multiple mutations in SOD1, including several that involve the essential copper binding ligands, result in neuronal injury remain unclear, current evidence implicates enhanced free radical generating activity associated with the copper bound by the mutant SOD1 enzymes (10 -12).
The delivery of copper to specific proteins within the cell is mediated by distinct intracellular carrier proteins termed chaperones (13). Consistent with this, recent studies have identified a protein termed the copper chaperone for superoxide dismutase (CCS) as the factor responsible for copper incorporation into SOD1 (14). In this study, comparison of the amino acid sequences of SOD1 and CCS revealed a striking homology between this enzyme and its putative chaperone, including residues involved in SOD1 homodimerization. This observation suggested a mechanism by which copper may be delivered to SOD1 by CCS through direct protein-protein interaction, and thus further analysis was undertaken to examine this possibility in vitro and in vivo. The results validate this concept and suggest a novel paradigm whereby the multiple FALS-associated SOD1 mutations may result in neurodegenerative disease.

EXPERIMENTAL PROCEDURES
Cloning and Protein Expression-The coding region for full-length CCS was amplified by polymerase chain reaction (PCR) and subcloned into the BamHI and EcoRI sites of pGEX 4T-1 (Amersham Pharmacia Biotech). Similarly, the coding region for amino acids 86 -274 (domains B/C) of CCS was amplified by PCR and subcloned into this expression plasmid using the same restriction sites. The coding region for amino acids 1-85 of CCS was amplified by PCR using oligonucleotides with NdeI and EcoRI sites and then subcloned into the pET 28a(ϩ) expression plasmid (Novagen). Full-length and truncated CCS constructs subcloned into glutathione S-transferase (GST) or (His) 6 vectors were expressed in Escherichia coli BL21 or BL21(DE3) cells and purified as described previously (15).
Column Binding Assays-GST and His tag column binding assays were performed as described (16). For these assays, the CCS fusion proteins were immobilized on glutathione-agarose beads (Sigma) or His-binding resin (Novagen) and allowed to interact with cell lysate (50 g of protein) or purified human SOD1 (Sigma) in 100 l of PBS at 4°C for 1.5 h. Following binding, the beads were washed extensively with PBS, and protein complexes were released by the addition of sample buffer with dithiothreitol (DTT) and heating at 100°C for 5 min. Samples were then separated by SDS-PAGE followed by electrophoretic transfer to nitrocellulose. Membranes were blocked with 5% non-fat dry milk and following incubation with specific antisera were developed using enhanced chemiluminescence (Amersham Pharmacia Biotech) as described previously (17). The copper content of the purified CCS and SOD1 was determined utilizing bicinchoninic acid as described (18).
Transfection, Coimmunoprecipitation, and Immunoblot Analysis-COS-1 and HepG2 cells were obtained from the American Type Culture Collection and grown to confluence in medium with fetal calf serum as described (17). COS-1 cells were transiently transfected with pEFBOS vector containing wild-type or FALS-associated mutants as described previously (19). Cells were lysed in PBS supplemented with 1 M pepstatin, 2.5 g/ml leupeptin, and 0.2 mM phenylmethylsulfonyl fluoride. After one cycle of freeze-thaw, lysates were precleared overnight with normal rabbit serum and protein A-agarose beads (Sigma). CCS antiserum was added to 1 mg of total protein, incubated with rocking at 4°C for 1 h, and the immunoprecipitated CCS was extracted from solution by the addition of protein A-agarose beads (16). The coimmunoprecipitated complex was released from the beads by the addition of sample buffer with DTT followed by heating to 100°C for 5 min. Following SDS-PAGE and transfer to nitrocellulose, immunoblot analysis was performed as described (17).

RESULTS AND DISCUSSION
Amino acid sequence alignment of human CCS and SOD1 revealed a region in CCS from residues 86 -234 (domain B), which is 47% identical to SOD1 (Fig. 1). Importantly, the identical residues include all of the known SOD1 copper and zinc ligands except His 120 as well as those amino acids at the SOD1 dimer interface (20) and most of the amino acid residues mutated in FALS (21). The first 85 amino acid residues of CCS (domain A) contain the copper binding consensus sequence MXCXXC found in the copper transporting ATPases (22,23) and the cytosolic copper chaperones ATX1 (24,25) and HAH1 (26,27). Analysis of the carboxyl-terminal region of CCS (domain C), which includes residues 235-274, revealed no homology to SOD1 or other known proteins but did indicate a putative peroxisomal localization sequence (AHL) at the COOH terminus ( Fig. 1) SOD1 is a homodimer and the striking homology between CCS and SOD1 suggested to us that these two proteins may form a heteromeric complex during copper delivery. To investigate this, column binding assays were performed utilizing a GST-CCS fusion protein. GST-CCS was immobilized on glutathione-agarose beads and mixed with either cell lysates from COS-1 cells transiently transfected with the pEFBOS vector containing the SOD1 gene or purified human SOD1 with or without the addition of CuSO 4 or DTT. Following incubation, eluates were subjected to SDS-PAGE and analyzed by immunoblotting with a polyclonal antisera specific for SOD1. As can be seen in Fig. 2A, SOD1 was readily detected following incubation of GST-CCS with cell lysates. Although variability in the amount of SOD1 recovered was observed, no consistent effect on this interaction was detected with the addition of copper or DTT. The CCS-SOD1 interaction is direct and not mediated by additional components in the cell lysate, because incubation of GST-CCS with purified SOD1 gave identical results ( Fig. 2A,  lanes 6 -8). To further evaluate the effect of copper on this interaction, column binding assays were performed utilizing apo-and holo-CCS and apo-and holo-SOD1. The copper status of the purified proteins was verified by bicinchoninic acid assay (18). In each case the copper status of the protein had no effect on the observed binding (data not shown), suggesting that structural elements independent of copper binding are essential for the interaction of CCS and SOD1.
To further examine the specificity of CCS-SOD1 interaction and to determine which domains of CCS function in this process, column binding assays were repeated using GST or Histagged constructs containing the indicated CCS domains or the human copper chaperone HAH1 (Fig. 2B). Cell lysates from COS-1 cells overexpressing human SOD1 were mixed with these proteins complexed to the agarose beads. As anticipated from the homology noted above, a CCS construct containing domains B and C was found to interact with SOD1 in a manner identical to that observed previously with full-length CCS. In contrast, a CCS construct containing only domain A did not interact with SOD1, indicating that the copper-binding MX-CXXC motif is not necessary for CCS-SOD1 interaction. The interaction of CCS and SOD1 observed here is highly specific, as no SOD1 was recovered following incubation of cell lysates with GST or GST-HAH1 (Fig. 2B, lanes 1 and 2).
An A4V mutation in SOD1 is responsible for almost 50% of SOD1 mutations in FALS cases (21). This A4V mutant retains significant superoxide dismutase activity, indicating the presence of copper in the mutant enzyme (19). As CCS is responsible for copper delivery to wild-type SOD1, we examined interaction between CCS and SOD1 A4V utilizing the column binding assay and COS-1 cell lysates expressing this mutant protein. These studies revealed interaction of SOD1 A4V with CCS similar to that observed for wild-type SOD1 (Fig. 2C, lanes   FIG. 1. Amino acid sequence alignment of human CCS and SOD1. Identical amino acids are shaded. The copper binding motif MXCXXC is boxed, and the putative peroxisomal localization sequence (AHL) is underlined. The copper and zinc binding residues in SOD1 are shown in bold, and the residues known to be mutated in FALS are marked with an asterisk.

FIG. 2. In vitro interaction of SOD1 and CCS.
A, the GST-CCS fusion protein was bound to glutathione-agarose beads and interacted with buffer (lane 1), COS-1 cell lysates transfected with pEFBOS expressing wild-type human SOD1 (lanes 2-5), or with purified human SOD1 (lanes 6 -8). In some experiments, 1 mM CuS0 4 and/or 1 mM DTT were added to the incubation mixture as indicated. Following SDS-PAGE in 4 -20% gradient gels, eluates were examined by immunoblotting with SOD1 antisera. B, the indicated GST constructs were bound to agarose beads, incubated with COS-1 cell lysates expressing human wild-type SOD1, and analyzed as indicated above. C, full-length GST-CCS was bound to agarose beads and mixed with buffer (control), cell lysates from COS-1 cells expressing wild-type (lane 2) or mutant human SOD1 (lanes 3-5), or with purified human SOD1 (lane 6) and analyzed as indicated above.

SOD1 and CCS Interaction 23626
2 and 3). Identical results were obtained using lysates from COS-1 cells expressing several other common SOD1 FALS mutants (19) (data not shown). Two FALS-associated SOD1 mutants, H46R (29) and H48Q, are devoid of copper due to loss of an essential histidine copper ligand, and yet both mutations result in significant motor neuron disease (21). Interestingly, these mutants also displayed a similar interaction with CCS (Fig. 2C, lanes 4 and 5) leading to the intriguing possibility that this chaperone may interact with and attempt to deliver copper to SOD1, even when the enzyme cannot readily accept it. Such a finding immediately suggests a common mechanism for motor neuron disease associated with many different FALS mutants dependent upon the process of CCS interaction with an impaired SOD1 target resulting in inappropriate release of free copper.
These in vitro experiments revealed a direct interaction between CCS and SOD1. To examine this process in vivo, we performed coimmunoprecipitation utilizing a rabbit polyclonal antiserum specific for domain A of CCS. Immunoblotting with this antiserum revealed endogenous CCS as a single 34-kDa protein in human liver and HepG2 cell lysates (Fig. 3A, lanes 1  and 2). Immunoprecipitation of lysates from metabolically labeled HepG2 cells with this antisera also detected CCS (Fig.   3A, lane 5). Despite abundant SOD1 detectable in these same cells by immunoblotting and immunoprecipitation (Fig. 3A,  lanes 3 and 4), as anticipated from the lack of homology in the region utilized to generate the CCS antisera (domain A Fig. 1), no SOD1 was detected with the CCS antisera under these stringent conditions (Fig. 3A, lanes 1, 2, and 5).
Having established the specificity of the CCS antisera, we next performed coimmunoprecipitation of HepG2 cell lysates (Fig. 3B). Following immunoprecipitation and washing under gentle conditions, immunoprecipitates were analyzed by SDS-PAGE and the presence of CCS confirmed by immunoblotting (Fig. 3B, lanes 1 and 2). After removal of the CCS signal from this blot, the presence of SOD1 was examined utilizing an SOD1-specific antisera (Fig. 3B, lanes 3 and 4). As can be seen from the data, immunoprecipitation of CCS results in the presence of SOD1 in these same samples. This association of SOD1 with CCS was specific and not the result of washing conditions, as immunoprecipitation with antisera for the copper chaperone HAH1 did not result in any detectable SOD1 (Fig. 3C, lane 1), whereas identical experiments utilizing either SOD1 or CCS antisera resulted in detection of SOD1 (Fig. 3C, lanes 2 and 3). In all cases, no specific bands were detected following immunoprecipitation with preimmune sera, and the band corresponding to SOD1 was specifically blocked when SOD1 antisera was preincubated with purified SOD1 prior to FIG. 4. Immunofluorescent localization of CCS in HepG2 cells. HepG2 cells were processed for indirect immunofluorescence and analyzed at ϫ 60 after incubation with antibodies to CCS alone (a, b) or together with a murine anti-human SOD1 monoclonal antisera (c-f). In some experiments (e, f), cells were permeabilized with digitonin for 15 min prior to processing for immunofluorescence. Arrows indicate regions of cytoplasmic staining (a, b) and residual nuclear signal (e, f).

FIG. 3. Coimmunoprecipitation of CCS and SOD1.
A, immunoblot analysis of 100 g of human liver and HepG2 cell lysate with CCS antibody (lanes 1 and 2). Immunoblot analysis of SOD1 from these same HepG2 cell lysates (lane 3). Immunoprecipitation of SOD1 and CCS from [ 35 S]cysteine-labeled HepG2 cells (lanes 4 and 5). B, HepG2 cell lysates were immunoprecipitated with CCS antisera followed by SDS-PAGE and immunoblot analysis with CCS (lane 1) or SOD1 (lane 3) antisera. Immunoblot analysis of cell lysates with each antibody (lanes 2 and 4) were included as controls. IgG heavy chain was detected by secondary antibody (H). C, immunoprecipitation of HepG2 cells with antisera to HAH1, SOD1, or CCS followed by SDS-PAGE and immunoblotting for SOD1.
immunoblotting (data not shown). As might be anticipated based upon homology, the SOD1 antisera used in these experiments did cross-react with CCS precluding the interpretation of similar coimmunoprecipitation experiments using this antibody.
These studies suggested that copper delivery to SOD1 is mediated via a direct interaction with CCS. Previous studies have revealed that SOD1 is a soluble protein localized in the cytoplasm and nuclei of cells (30,31). We reasoned that if CCS and SOD1 interact in vivo, then these proteins should be detectable in the same intracellular locations. To examine this question, we performed indirect immunofluorescence of HepG2 cells using the CCS antisera and a murine monoclonal SOD antibody that did not react with CCS (data not shown). In these experiments, CCS was found diffusely localized throughout the cytoplasm and nucleus with a somewhat punctate appearance in the cytoplasm (Fig. 4, a and b). Double-immunofluorescence revealed that the distribution of CCS in these cells was identical to that of SOD1 (Fig. 4, c and d). Digitonin permeabilization prior to antibody staining resulted in a marked diminution of the signal for CCS and SOD1 within both the nucleus and the cytoplasm, consistent with the concept that these are freely diffusable within the cell (Fig. 4, e and f). Sequence analysis of CCS revealed a potential peroxisomal localization sequence in the carboxyl terminus (Fig. 1), and although recent studies have suggested that in neurons SOD1 is localized in part to peroxisomes (32), no overlap was observed for either SOD1 or CCS when double-immunofluorescent studies were performed in these same cells utilizing a sheep anti-human catalase antibody that identified peroxisomes (data not shown).
Copper is an essential redox metal, and recent studies have revealed a complex pathway of intracellular copper trafficking mediated by unique chaperones functioning to deliver this metal to specific proteins while protecting cellular constituents from untoward redox-mediated injury (13). Although the mechanism by which mutant SOD1 results in neuronal injury remains unclear, it is apparent that copper is an integral part of this process, as chelation of this metal abrogates the deleterious effects of FALS SOD1 mutants (10,33). Our finding that CCS directly interacts with FALS-associated SOD1 mutants is consistent with the results reported recently that these same mutants can acquire catalytic copper in Saccharomyces cerevisiae via this chaperone (34). As previous studies on these SOD1 mutants have revealed dramatically altered metal binding sites (35), our findings suggest a novel paradigm by which multiple SOD1 FALS mutants may result in neuronal injury. The delivery of copper by CCS to a target protein either unable or less able to incorporate this metal would inevitably lead to copper-mediated toxicity. Such a model is consistent with the failure to observe FALS in transgenic mice lacking SOD1, because under such circumstances no CCS-SOD1 interaction and thus copper transfer will occur (8,9). Taken in this context, our findings place CCS in a central role in pathogenesis of neuronal injury in FALS and suggest a novel therapeutic approach directed toward this CCS-SOD1 interaction to prevent copper delivery to the mutant enzyme.