Yeast contain a non-proteinaceous pool of copper in the mitochondrial matrix.

The yeast mitochondrion is shown to contain a pool of copper that is distinct from that associated with the two known mitochondrial cuproenzymes, superoxide dismutase (Sod1) and cytochrome c oxidase (CcO) and the copper-binding CcO assembly proteins Cox11, Cox17, and Sco1. Only a small fraction of mitochondrial copper is associated with these cuproproteins. The bulk of the remainder is localized within the matrix as a soluble, anionic, low molecular weight complex. The identity of the matrix copper ligand is unknown, but the bulk of the matrix copper fraction is not protein-bound. The mitochondrial copper pool is dynamic, responding to changes in the cytosolic copper level. The addition of copper salts to the growth medium leads to an increase in mitochondrial copper, yet the expansion of this matrix pool does not induce any respiration defects. The matrix copper pool is accessible to a heterologous cuproenzyme. Co-localization of human Sod1 and the metallochaperone CCS within the mitochondrial matrix results in suppression of growth defects of sod2Delta cells. However, in the absence of CCS within the matrix, the activation of human Sod1 can be achieved by the addition of copper salts to the growth medium.

Copper is an essential cell nutrient acting as a cofactor in nearly 20 enzymes (1). However, excess accumulation of copper ions results in toxicity. Evidence for the effectiveness of copper ions as a toxin comes from its historic use as a fungicide, molluscide, and algicide. Homeostatic mechanisms exist in cells to regulate the cellular concentration of copper ions, thus maintaining copper balance and minimizing deleterious effects. Cells appear to maintain a quota for essential metal ions; this quota is primarily the quantity necessary to metallate the various copper proteins (2,3). Copper ions are required for at least three key enzymes in Saccharomyces cerevisiae. The cuproenzymes include the cytosolic superoxide dismutase Sod1, 1 the plasma membrane ferroxidase Fet3, and the mitochondrial inner membrane enzyme cytochrome c oxidase (CcO). The copper quota of the yeast S. cerevisiae is about 5 ϫ 10 5 atoms per cell (3,4).
It is unclear what fraction of the 5 ϫ 10 5 copper atoms per cell is from copper in Sod1, Fet3, and CcO. Expression of these copper-binding proteins varies with growth conditions, suggesting that the copper may be distributed differently depending on growth conditions. Clearly, a significant fraction of the cellular copper is associated with Sod1; however, not all Sod1 molecules are metallated (4,5). Fet3 requires four copper ions for activity, but levels of this protein are dependent on iron status of the medium (6). CcO levels vary depending on whether the cells are grown by fermentation or respiration. In addition, a varying quantity of cellular copper exists bound to two metallothioneins, Cup1 and Crs5 (7,8). Expression of CUP1 and CRS5 is regulated by copper levels through the copper-responsive transcription factor Ace1 (9,10). An increase in the free Cu(I) ion pool activates Ace1 through formation of a polycopper cluster (11). Activated Ace1 induces transcription of CUP1, CRS5, and SOD1. Each protein is capable of buffering the copper ion concentration in the cytoplasm. Copper buffering regulated by Ace1 must be highly efficient because yeast cells are predicted to lack a pool of free copper ions in the cytoplasm (4). Cells concentrate copper by several orders of magnitude from the culture medium to achieve the copper quota (2). This gradient is generated by multiple metal ion permeases on the plasma membrane. Copper ion uptake is mediated by high affinity Ctr1 and Ctr3 and low affinity Smf1 and Fet4 permeases (12). Within the cell other transporters are necessary for transmembrane movement of copper ions. Translocation of copper ions into the lumen of trans-Golgi vesicles is achieved by P-type ATPase transporters (Ccc2 in yeast, ATP7A and ATP7B in animal cells) (13). Translocation of copper ions across the vacuolar membrane is mediated by Ctr2 (14).
Copper ions are shuttled to sites of utilization by proteinmediated transfer (15,16). Metallochaperones facilitate the transport of copper ions to specific sites through a chelating environment and prevent the inappropriate interaction of copper ions with other molecules. Copper insertion into Sod1 in yeast requires the function of the Lys7 (CCS) metallochaperone (4,(17)(18)(19). In the absence of CCS copper ions are not inserted into yeast Sod1 unless cells are incubated in medium containing high levels of copper salts, and this requirement is attenuated by a decrease in levels of the metallothioneins by a deletion of ACE1 (4). This is an indication of the cytosolic copper-buffering capacity of Cup1. Copper insertion into Fet3 occurs within post-Golgi vesicles. The copper is translocated across the vesicular membrane by the Ccc2 P-type ATPase (13). Cu(I) ions are shuttled to the Ccc2 translocase by the Atx1 metallochaperone (20,21). Deletion of either Ccc2 or Atx1 results in apo-Fet3 and an inability to grow on iron-deficient medium (20). Cu(I) ions translocated by the Ccc2 pump can also be used to metallate heterologous molecules within trans-Golgi vesicles (22).
The mechanism of copper ion routing to the mitochondrion for assembly of cytochrome c oxidase is unknown. Because copper ion delivery to Sod1 and Ccc2 is protein-mediated, the prediction is that copper shuttling to the mitochondrion will also be protein-mediated. Two proteins implicated in copper ion translocation to the mitochondrion are Cox17 and Cox19 (23)(24)(25). Both proteins are conserved in eukaryotic cells and exhibit a dual localization in both the cytosol and the mitochondrial intermembrane space (IMS) (24,25). The IMS is an aqueous space between the mitochondrial inner membrane (IM) and outer membrane (OM) and is interrupted by junction points in which the IM and OM are in contact (26). Cox17 has been implicated in copper ion delivery to the mitochondria based on its dual localization and the fact that the respiratory defect of cox17-1 cells is suppressed by high levels of copper in the growth medium (23). A copper transport function for Cox17 is plausible because it binds Cu(I) through three critical cysteinyl residues in a Cys-Cys-Xaa-Cys sequence motif (27). Cox19 shows weak sequence similarity to Cox17 in the conservation of three C-terminal Cys residues (25); however, these Cys residues in Cox17 are not part of the Cu(I) binding motif (27). Yeast cells lacking Cox17 or Cox19 are respiratory-deficient and devoid of cytochrome c oxidase activity. Suppression of these phenotypes by the addition of exogenous copper is observed only in cox17 cells.
Within the mitochondrion two additional inner membrane proteins, Sco1 and Cox11, are implicated in copper ion insertion into cytochrome c oxidase (28 -33). Sco1 and Cox11 are important in reactions of direct copper transfer to the Cu A site and Cu B site, respectively. The two copper centers exist within mitochondrially encoded subunits of cytochrome c oxidase, so copper ions must be shuttled to the mitochondrion for insertion into nascent chains within the inner membrane. Cox17 may deliver Cu(I) to both Sco1 and Cox11 for subsequent donation to the Cu A and Cu B sites. Alternatively, separate shuttle proteins may exist for Sco1 and Cox11.
If Cox17 functions in the delivery of Cu(I) ions to the mitochondrion, it may deliver Cu(I) ions to a mitochondrial OM transporter, in analogy to the Atx1 routing of Cu(I) ions to the Ccc2 trans-Golgi translocase. Alternatively, Cox17 may ferry Cu(I) ions across the semiporous mitochondrial OM. Many proteins are imported across the OM through the translocase of the outer membrane import complex, but these proteins are imported as unfolded proteins (34,35). Import of Cu(I) across the OM by Cox17 would necessitate import of the folded conformer to retain its copper cargo.
The only other known mitochondrial copper protein is Sod1, which exhibits a dual localization within the cytosol and the IMS (36). Approximately 1-5% of total Sod1 is localized within the IMS (36). The fraction of IMS Sod1 is partially dependent on the CCS metallochaperone (37). Directed import of CCS to the IMS compartment using a heterologous mitochondrial import sequence enhances accumulation of Sod1 within the IMS, although Sod1 can still localize to the IMS in the absence of CCS (36,37). Copper ion metallation of Sod1 likely occurs within the IMS because mitochondrial Sod1 import is dependent on the apo conformer (37). The source of copper ions for the CCS-mediated metallation of Sod1 in the IMS is unknown, but Cox17 does not appear to influence formation of active Sod1 within the IMS (36). Likewise, CCS does not affect CcO function (17).
We show that neither Cox17 nor Cox19 is required for the import or accumulation of copper in the mitochondrion, suggesting that neither acts as metallo shuttles for the bulk of mitochondrial copper. We show for the first time that a significant pool of copper exists within the mitochondrion that is not associated with either CcO or Sod1. This pool fractionates with matrix markers and does not appear to be proteinaceous. This pool of copper can be utilized to metallate a heterologous molecule, human Sod1, targeted to the mitochondrial matrix.
Expression Vector-Human SOD1 obtained from Dr. P. J. Hart was inserted into YCpRS316 and fused to the 5Ј 27 codons of the yeast SOD2 ORF that encode the mitochondrial target sequence. The fusion gene, designated SOD2/hSOD1, was under the control of the SOD2 promoter with the hSOD1 terminator. A variant of this fusion gene was constructed in which only the promoter of SOD2 was fused to hSOD1. This fusion, designated hSod1(c), lacked the mitochondrial target sequence of Sod2 and gave cytosolic expression. These constructs were confirmed by dideoxynucleotide sequencing. CCS-encoding plasmids pLS114 and pLS117 were generously provided by Dr. Val Culotta. Both contain CCS (LYS7) fused to segments of CYB2 in pRS313. Plasmid pLS117 encodes a fusion of the CYB2 IMS-targeting sequence to LYS7 (37), whereas pLS114 encodes a chimera of 80 codons of CYB2 fused to LYS7 that targets CCS to the matrix. 2 Plasmids were transformed into yeast by the lithium acetate protocol.
Preparation of Mitochondria-Mitochondria were isolated as described previously (38). Briefly, lyticase was used to create spheroplasts that were gently ruptured in a glass Dounce homogenizer. After the spheroplasts were lysed crude mitochondria were isolated by differential centrifugation in the presence of 0.5 mM phenylmethylsulfonyl fluoride. Cell debris was pelleted at 1,500 ϫ g, mitochondria and microsomes were pelleted at 12,000 ϫ g; the supernatant represented the post-mitochondrial fraction. The crude mitochondria were then loaded onto a discontinuous Nycodenz gradient (16% on 22%) and centrifuged at 150,000 ϫ g for 1 h. The intact mitochondria recovered from the gradient interface were washed in isotonic buffer and pelleted at 12,000 ϫ g before analysis. Mitochondria were quantified using standard Bradford protein reagents. For hypotonic lysis, the intact mitochondria were diluted 20-fold into 10 mM Hepes, pH 7.4, and centrifuged at 25,000 ϫ g to isolate the soluble IMS. The pellet was retained as mitoplasts and later fractionated into soluble and membrane fractions by sonication or carbonate treatment in 100 mM NaCO 3 , pH 11.
Western Blots-10 -25 g of total protein from the mitochondrial or post-mitochondrial fraction were separated on a 15% SDS-PAGE gel system and electrophoretically transferred onto a nitrocellulose membrane. Membranes were blocked in 1ϫ PBS (50 mM Na 3 PO 4 , 100 mM NaCl, pH 7.0), 0.01% Tween 20, and 10% milk solution before protein detection. Antisera to porin (Por1), phosphoglycerol kinase (Pgk1), the vacuolar H-ATPase subunit Vma2, and dolichol phosphate mannose synthase (Dpm1) were obtained from Molecular Probes. Antiserum to cytochrome b 2 (Cyb2) was kindly provided by Dr. R. Stuart. Antisera to ySod1 and ySod2 were provided by Dr. V. C. Culotta and to hSod1 was from Santa Cruz Biotechnology, Inc. Blots were reprobed after stripping in 5 washes of 25 mM glycine, pH 2.0, 100 mM NaCl, and 0.5% Tween20. Loading controls were performed using porin (mitochondria) and phosphoglycerol kinase (cytosol).
ICP-OES Analysis-Three concentrations (1, 3, and 5 mg/ml) of intact mitochondria were digested in sealed, acid-washed tubes at 95°C in 150 l of metal-free 40% nitric acid (Optima). The samples were then diluted to 1 ml into double-distilled water for analysis. Serial dilutions of commercially available mixed metal standards were used to construct a standard curve. Blanks of nitric acid or buffer samples were also digested in the acid-washed tubes for comparison, and spiked controls were analyzed to ensure reproducibility.
Cytochrome c Oxidase Assays-CcO oxidase activity in isolated mitochondria was quantified by monitoring the reduction of 32 M bovine cytochrome c at 550 nm by 5-10 g of isolated mitochondria (40 mM KH 2 PO 4 , pH 6.7, 0.5% Tween 20).
Mitochondrial Fractionation-Purified mitochondria were lysed by three 30-s pulses of sonication at 50% output of a microtip. The soluble fraction was isolated from the insoluble fraction by centrifugation at 15,000 ϫ g. The soluble fraction was diluted into buffer A (20 mM Tris-HCl, pH 7.2) and filtered with a 0.45 M syringe filter. The sample was the loaded onto a HR10/10 Mono Q column equilibrated in 10 column volumes of buffer A. The unbound protein was eluted with 5 column volumes of buffer A. A 25-column-volume gradient of buffer B (20 mM Tris-HCl, pH 7.2, 1 M NaCl) was then initiated. Fractions were collected and analyzed by ICP-OES after dilution in 10% nitric acid. The remainder of the metal-containing fractions was then loaded onto a size exclusion column (HR10/30 Superdex200 or Superdex 75) equilibrated in 20 mM Tris HCl, pH 7.2, 150 mM NaCl. Fractions were collected throughout the run and analyzed by ICP-OES after dilution into 10% nitric acid. The remainder of the sample was used for SDS-PAGE analysis or further desalted for mass spectrometry.

RESULTS
Copper Delivery to the Mitochondria-Cox17 and Cox19 have been implicated as mitochondrial copper ion shuttles for assembly of cytochrome c oxidase based on their dual localization in both the cytoplasm and the IMS of mitochondria (24). In addition, both proteins are capable of binding copper ions. We demonstrated previously that Cox17 binds Cu(I) within a polycopper-thiolate cluster (39). Recombinant Cox19 is also a copper-binding protein. 3 If Cox17 and/or Cox19 are dominant copper ion shuttles for assembly of CcO, the prediction is that cells lacking Cox17 or Cox19 would have depressed mitochondrial copper levels. The copper levels were measured in Nycodenzpurified mitochondria from a variety of cells lacking one of the CcO assembly factors. The amount of copper within the mitochondrion was independent of the presence of Cox17 or Cox19 (Fig. 1A). Although Cox17 and Cox19 cannot complement each other, a potential redundant role was investigated by testing a double cox17,cox19 null strain. As shown in Fig. 1A, the mitochondrial copper was not affected by this double deletion. In fact, the mitochondrial copper concentration was not markedly diminished in yeast lacking Cox2, which is the mitochondrially encoded subunit of CcO that forms the Cu A site, or in rho Ϫ cells, which are devoid of mitochondrially encoded proteins. The level of mitochondrial copper was independent of the carbon source used for culturing yeast (data not shown). The lack of an appreciable diminution in mitochondrial copper in cells lacking a functional CcO complex suggests that mitochondria contain a pool of copper not associated with CcO or CcO assembly proteins.
Non-respiratory Pool of Mitochondrial Copper-To measure the fraction of mitochondrial copper associated with the CcO complex, we quantified the heme A pool within mitochondria. CcO is the only protein to use heme A, and all heme A is associated with the active, oligomeric CcO complex. Yeast mutants exhibiting defects in assembly of CcO (e.g. cox11, cox17, cox19, and sco1) are devoid of heme A within the mitochondrion (23,30,40). Using an established extinction coefficient for heme AϩA 3 , the amount of CcO can be quantified from the difference absorption spectrum of oxidized versus reduced cytochromes (41). Because the active CcO binds three copper ions per complex, the fraction of mitochondrial copper associated with CcO can be determined. This analysis revealed that ϳ10% of the mitochondrial copper was associated with CcO (Fig. 1B). This calculation is based on the assumption that all CcO molecules are copper-loaded. This assumption is reasonable because it is known that mutants that fail to insert copper into CcO result in an unstable complex that does not contain heme A (23,30). The ratio of copper to total protein or heme A was unaffected by growth in a non-fermentable carbon source. This is consistent with the deletion mutant data in Fig. 1A showing that removal of a CcO subunit or several assembly factors has no significant effect on the copper levels. The known assembly factors involved in copper ion metallation of CcO are not likely to contribute to the mitochondrial copper pool because they are low abundance proteins. A recent mitochondrion proteomic study showed Sco1 to be a low abundant protein comparable in levels of the CcO subunit Cox6 (42).
The copper pool is associated with the mitochondrion and not a copurifying organelle. Gradient-purified mitochondrial may contain limited vacuolar and ER contamination (38). Western analyses were conducted on crude and gradient-purified mitochondria to address contamination. Vacuolar and ER markers were readily apparent in the crude mitochondrial fraction, but the marker proteins were significantly reduced in the purified mitochondrial fraction ( Fig. 2A). Despite a marked diminution in vacuolar and ER markers in gradient-purified mitochondria, the copper content (nmol of copper/mg of protein) did not appreciably change. The variance observed in 10 independent gradient purifications was less than 30%, with multiple isolates having a greater copper content in the purified relative to the crude fraction. To confirm that vacuolar contamination is not the source of the mitochondrial copper, mitochondria were purified from vps33 mutant cells that have abnormally small vacuoles (43). The mitochondrial copper content was unchanged as was the ratio of copper to heme A, confirming that 3 K. Rigby, unpublished observation.
FIG. 1. Panel A, quantitation of copper levels in Nycodenz-purified mitochondria (mito) from various respiratory mutants and their isogenic wild-type (WT). Mitochondria were acid-digested, and total metals were quantified by ICP-OES. The copper content is expressed per mg of mitochondrial protein. Panel B, fraction of mitochondrial copper associated with CcO. CcO was quantified in purified mitochondria based on heme A content. Heme A was quantified from the oxidized versus reduced difference absorption spectra in mitochondria solubilized in 0.5% deoxycholate. The spectra represent the dithionite-reduced spectrum minus the ferricyanide-oxidized spectrum. An extinction coefficient of 14 mM Ϫ1 cm Ϫ1 for A 605 nm Ϫ ((A 620 nm Ϫ A 590 nm )/2) was used for heme A quantitation. The samples were then analyzed by ICP to quantify the copper concentrations. A ratio of 3 copper per 2 heme A was used to determine the percentage of copper that is heme A-associated. the copper did not arise from vacuolar contamination (Fig. 2B). To confirm that ER/Golgi vesicle contamination was not a concern, mitochondria isolated from ccc2 null cells were analyzed. Ccc2 is the P-type ATPase copper transporter that is responsible for translocation of Cu(I) into post-Golgi vesicles. The mitochondrial copper content in ccc2 cells was similar to that of wild-type cells (Fig. 2B).
The Mitochondrial Copper Pool Is Not Associated with Known Yeast Copper Proteins-The only two known copper enzymes within mitochondria are CcO and Sod1. A fraction of the cytosolic Cu,Zn Sod1 partitions in the mitochondrial IMS (36). The CCS metallochaperone responsible for copper insertion into Sod1 is involved in the recruitment of Sod1 to the IMS (36). Cells lacking CCS have diminished levels of Sod1 within the IMS. We observed that deletion of LYS7 encoding CCS does not significantly diminish the mitochondrial copper pool (Fig.  3). Two other cytosolic copper-binding proteins include the Cup1 and Crs5 metallothioneins. Both proteins bind multiple Cu(I) ions within polycopper clusters (8). Distribution of these molecules within the mitochondrion may contribute a significant quantity of mitochondrial copper. Expression of CUP1 and CRS5 is regulated by the copper-responsive transcriptional activator Ace1. Deletion of ACE1 results in a marked decrease in basal CUP1 and CRS5 expression in yeast (44). Cup1 and Crs5 do not contribute to the mitochondrial copper pool because ace1⌬ cells have wild-type levels of mitochondrial copper (Fig.  3). Similarly, crs5⌬ cells have at least wild-type levels of mitochondrial copper. The only other known cuproenzymes in yeast are the plasma membrane Fet3 and vacuolar Fet5 ferrooxidases (45,46). Neither of these is expected to contaminate the copper pool in gradient-purified mitochondria. In addition, the mitochondrial copper pool is unperturbed in cells cultured in high iron-containing medium in which Fet3 and Fet5 are lowly expressed. The mitochondrial copper pool is not attenuated by overexpressing cytoplasmic copper ion buffers. Cells expressing CUP1 from the constitutive ADH1 promoter on a high copy plasmid did not diminish the level of mitochondrial copper (data not shown). Likewise, overexpression of human SOD1, which increased markedly the copper binding Sod1 levels in the cytoplasm, did not attenuate the mitochondrial copper pool (data not shown).
Localization of the Copper Pool-The mitochondria has three distinct compartments, the IMS, the intercristal space, and the matrix. Incubation of mitochondria with hypotonic solution results in rupture of the OM and solubilization of proteins localized within the IMS and presumably intercristal space. OM fragments are retained with the mitoplasts. Incubation of yeast mitochondria in hypotonic buffer results in retention of the bulk of mitochondrial copper with the mitoplast fraction (Fig. 4A). Western blot analysis confirmed that endogenous Sod1 was released by the hypotonic treatment consistent with its localization within the IMS (Fig. 4B). As expected, the known matrix protein Sod2 was retained with the mitoplasts. The marked release of Sod1 in hypotonic-treated mitochondria yet retention of the bulk of the copper with mitoplasts is consistent with the data with lys7⌬ cells showing that Sod1 does not contribute significantly to the mitochondrial copper pool.
To address whether copper was associated with the membrane fraction or soluble matrix compartment, mitochondria were subjected to sonication to disrupt both membranes. Analysis of the resulting membrane fraction and soluble fraction revealed that the soluble fraction contained little CcO activity but a majority of the copper and the soluble IMS protein Cyb2 (Fig. 5, A and B). The membrane fraction contained the bulk of the CcO enzymatic activity and the OM protein porin (Por1). These results suggest that the bulk of the mitochondrial copper exists within the matrix compartment. This matrix compartmentalization of copper is unprecedented, with no known cupro-enzymes present in the matrix. Because the IM forms a permeability barrier to ions, the matrix copper pool must require transport across the IM.
Copper Pool Is Dynamic-To determine how the mitochondrial pool of copper responds to changes in copper levels in the growth medium, cells were cultured in synthetic medium containing 0.5 mM CuSO 4 . It was known previously that yeast incubated in copper-supplemented medium take up additional copper well beyond the copper quota. CUP1 expression is dramatically up-regulated in response to the increased cellular copper levels, and a significant fraction of the increased cellular copper is retained in the cytoplasm as CuCup1 complexes (47). We show presently that mitochondria also accumulate a fraction of the increased cellular copper. The mitochondrial copper pool is increased 5-10-fold in cells cultured with 0.5 mM CuSO 4 . The accumulated mitochondrial copper has no negative effect . LYS7 encodes the copper metallochaperone CCS necessary for copper insertion into Sod1. Because Crs5 is induced by the diauxic shift, crs5⌬ cells were independently tested. Copper was quantified as described in Fig. 1. WT, wild type. on respiration as measured by oxygen consumption. Under culture conditions with 0.5 mM CuSO 4 , much of the cytoplasmic copper is associated with the Cup1 and Crs5 metallothioneins (8). The effect of these copper-buffering molecules can be assessed by evaluating ace1⌬ cells with attenuated basal and induced expression of CUP1 and CRS5. Because ace1⌬ cells are sensitive to elevated copper concentrations (10), cells were incubated with 0.01 mM CuSO 4 for only 5 h. Under these conditions mitochondria from wild-type cells show only a modest increase in copper (2.3-fold) (Fig. 6). The bulk of the increased cellular copper is associated with Cup1 within the cytosol. However, an 8-fold increase was evident in mitochondrial copper accumulation in ace1⌬ cells. The accentuated accumulation in ace1⌬ cells suggests that the mitochondrion efficiently imports copper, and this may add to the chelating environment of the cytosol.
The addition of 0.5 mM CuSO 4 to the growth medium of wild-type cells not only enhanced the mitochondrial copper pool but also significantly diminished the mitochondrial zinc and manganese pools (Table I). Likewise, the addition of 0.5 mM ZnSO 4 but not FeCl 2 to the culture medium attenuated the mitochondrial copper pool. The diminution in mitochondrial copper by the addition of ZnSO 4 occurred despite no attenua-tion in total cellular copper. These data suggest that zinc and manganese may compete with copper for mitochondrial uptake.
Analyses of Mitochondrial Copper-The avid uptake of copper and the lack of known copper enzymes in the matrix suggest that additional cuproproteins may exist. After sonication of purified mitochondria the soluble fraction was loaded onto a Mono Q anion exchange column. The bulk of the copper was retained and eluted with application of a salt gradient (Fig. 7,  panel A). The anionic character of the copper pool yields significant purification from mitochondrial proteins, as many matrix proteins are cationic. The copper-containing fraction quantified by ICP-OES were further fractionated by gel filtration chromatography (Fig. 7, panel B). The anionic copper pool chromatographed in fractions corresponding to the elution volume of a globular protein of 13 kDa. This fraction was, however, devoid of absorbance at 280 nm (data not shown). Repeated attempts to identify proteins in the copper-containing fraction by liquid chromatography-MS/MS after trypsin digestion or electrospray ionization-MS analysis for intact polypeptides resulted in no identifiable proteins. In contrast, a different column fraction containing manganese (Fig. 7, dashed line) was similarly analyzed and found to contain MnSod2 by both liquid chromatography-MS/MS for a tryptic digest and electrospray ionization-MS for the intact protein. Additionally, gel filtration fractions containing the copper pool failed to show any visible protein bands using Sypro-Ruby staining of SDS-PAGE gels. The detection limit of Sypro-Ruby staining is 2 ng.
To further probe the possibility of a copper-binding polypeptide, the copper-pool eluant from Mono Q was subjected to proteinase K treatment. After incubation, the sample was reanalyzed by Mono Q chromatography (Fig. 8A). The elution position of the copper fraction in the salt gradient was unchanged by incubation with proteinase K even though polypeptides within the fraction were digested (Fig. 8B). The observed digestion of polypeptides within this fraction yet the unaltered elution volume from Mono Q suggests that copper is ligated by small molecules or peptides rather than a protein of substantial size. Mitochondrial DNA does not contribute to the anionic characteristics of this copper complex as the elution profile was unaffected by treatment with DNase, and furthermore, the complex lacks absorbance at 260 nm (data not shown). In

FIG. 6. Quantitation of copper in purified mitochondria from wild-type (WT) cells (A) and ace1⌬ cells incubated in the presence and absence of 10 M CuSO 4 for 5 h (B).
This dose of copper is not toxic to ace1⌬ cells. addition the profile does not change if the soluble fraction is prepared by detergent extraction or detergents are added after sonication.
Chromatographic analyses were also conducted with purified mitochondria isolated from ace1⌬, cox17⌬ cells and wild-type cells cultured with added additional CuSO 4 . The elution profiles of the copper pool were equivalent in each case (data not shown). Mitochondria from cells cultured with added copper showed a marked increase in the quantity of the anionic copper pool. The similar elution profile of this pool from Mono Q to that of wild-type mitochondria suggests that the anionic pool is easily expanded and ligands are not limited. Fractionation of this pool on gel filtration resulted in fractions containing no observable polypeptides by Sypro-Ruby staining. In fact, one gel filtration elution fraction containing 3 nmol of copper showed no stained band. A polypeptide present at 1-3 nmol would be considerably above the Sypro-Ruby detection limit (data not shown).
hSod1 Complements a sod2⌬ Strain-The import of copper into the mitochondrial matrix and apparent ligation by a small molecule may make this pool available to metallate a heterologous copper-binding protein. Although the yeast Sod1 cannot be populated in the absence of CCS in the cytosol under normal culture conditions, copper population of Sod1 can occur in lys7⌬ cells when the cytosolic copper level is significantly enhanced (4). Unlike yeast Sod1, human Sod1 can be metallated into an active form in the absence of CCS. 2 To address whether the matrix copper pool is accessible to heterologous molecules, human Sod1 was targeted to the mitochondrial matrix in an sod2⌬ strain. Appending the mitochondrial matrix targeting sequence of yeast SOD2 to human SOD1 resulted in the import of hSod1 into the mitochondria (Fig. 9A) and mitoplasts (Fig.  9B). The sod2⌬ strain is sensitive to oxidative stress particularly under conditions of hyperoxia (48,49). The SOD2/hSOD1 chimera (designated Sod1(m) for matrix hSod1 in Fig. 10) was able to suppress the hyperoxia growth impairment of sod2⌬ cells only if cells were cultured on agar plates containing 0.1 mM CuSO 4 (Fig. 10A). The sod2⌬ cells are also impaired in glycerol growth. This phenotype was partially suppressed by SOD2/hSOD1, but glycerol growth was improved with the addition of 0.1 mM CuSO 4 (Fig. 10B). To ensure that the rescue of the sod2⌬ cell phenotype was due to matrix hSod1 and not trace levels of cytosolic hSod1, we expressed hSOD1 under the yeast SOD2 promoter but lacking the Sod2 mitochondrial target sequence. The resulting hSod1 was no longer localized in the matrix (Fig. 9B). No growth of sod2⌬ cells harboring the hSOD1 variant (designated Sod1(c) for cytosolic Sod1 in Fig.  10) was observed under hyperoxia conditions or on glycerol plates regardless of the amount of exogenous copper added to the cultures (Fig. 10, A and B). The lack of suppression of sod2⌬ cells by hSOD1(c) is consistent with the observation that the    The Mono Q elution fraction 10 from Fig. 7 was divided into two fractions that were rechromatographed on Mono Q after dilution of the salt content. One sample was incubated with proteinase K (PK) overnight at 23°C before rechromatography. The peak fraction after chromatography was concentrated, and the retentate was analyzed on a 15% SDS-PAGE gel stained with Sypro Ruby (panel B). The loss of protein bands in the sample incubated with proteinase K confirms the activity of the proteolytic digestion.
E. coli iron superoxide dismutase was ineffective in suppressing the growth defects of yeast sod2⌬ cells when localized to the IMS but was effective when localized within the matrix (50). Although the hSod1(c) was nonfunctional in suppression of hyperoxia growth of sod2⌬ cells, the molecule, as expected, was effective in suppression of hyperoxia growth of sod1⌬ cells (data not shown).
The requirement for added copper salts in the growth medium for efficient suppression of sod2⌬ cell phenotypes was significantly reduced when SOD2/hSOD1 transformants also contained CCS within the matrix. Fusion of CCS (LYS7) to a 5Ј segment of CYB2 (5Ј 80 codons of open reading frame) results in import of CCS into the mitochondrial matrix (51). 2 Cells containing matrix hSOD1(m) and CCS (designated CCS(m) in Fig. 10) were able to propagate in hyperoxia conditions without added copper salts (Fig. 10A). Growth of these cells was marginally improved with the addition of 0.1 mM CuSO 4 to the growth medium. Likewise, cells containing hSOD1(m) and CCS(m) grew in glycerol without added copper to the growth medium (Fig. 10B). The rescue of the sod2⌬ phenotype by hSod1(m) suggests that the copper present in the matrix can be used for metallation of Sod1. The efficiency of matrix metallation of hSod1(m) is improved by the presence of CCS within the matrix. The SOD2/hSOD1 construct does not have a dominant negative effect on respiration when expressed in wild-type cells.

DISCUSSION
Mitochondria are presently shown to contain a pool of copper that is distinct from the only two known mitochondrial cuproenzymes, Sod1 and CcO. Only a small fraction of mitochondrial copper is associated with these two cuproenzymes. The bulk of the remainder is localized within the matrix as a solu-ble, anionic, low molecular weight complex. The mitochondrial copper pool is dynamic, responding to changes in the cytosolic copper level. The addition of copper salts to the growth medium leads to an increase in mitochondrial copper. This increase in mitochondrial copper is more pronounced when Cup1 metallothionein expression is attenuated by deletion of ACE1. The mitochondrial copper pool is not diminished by constitutive expression of CUP1 or SOD1, suggesting that a committed pathway for copper ion uptake into the mitochondrion exists. Storage of copper ions within the mitochondrion may be a component of cellular copper buffering.
The identity of the matrix copper ligand is unknown. Several lines of evidence suggest that the matrix copper fraction is not protein-bound. First, no protein bands were observed in the purified fraction by SDS-electrophoresis and Sypro-Ruby staining. Second, no proteins were observed by mass spectrometry. Third, the elution behavior on anion exchange chromatography did not change after prolonged digestion with proteinase K. The conclusion that the matrix pool is non-proteinaceous rests on the assumption that our isolation protocol for the matrix fraction did not liberate copper ions from a labile protein complex. Exposure to air may liberate copper from an oxygen-labile protein. This situation is unlikely considering copper complexes are reasonable air-stable including the thiolate-rich metallothioneins. It is unlikely that the pool is copper GSH because gsh1⌬ cells cultured in medium containing dithiothreitol, to enable growth, contained the same mitochondrial copper level. Furthermore, the anionic character of this pool from gsh1⌬ cells did not change. A number of organic acids (citrate, oxaloacetate, etc.) exist within the matrix and may be ligands. Considering that the mitochondrial copper pool is dynamic and can expand significantly as a similar anionic complex, the prediction is that ligands are not limiting within the matrix. Genetic and biochemical approaches are being taken to identify the matrix copper ligand(s).
The matrix copper pool is accessible. Co-localization of human Sod1 and CCS within the mitochondrial matrix results in suppression of growth defects of sod2⌬ cells. In the absence of CCS within the matrix, the activation of hSod1 requires the addition of copper salts to the growth medium. Because mitochondrial import of proteins into the matrix occurs as unfolded molecules, the copper metallation of hSod1 must occur within the matrix. The CCS requirement for efficient activation of Sod1 may arise from its role as a copper metallochaperone in directed copper ion insertion in apo-Sod1 (4). Alternatively, CCS may be important in the formation of the essential disulfide bond in Sod1 (37). In the absence of CCS within the matrix, the additional copper necessary for hSod1 activity may contribute to an oxidant activity necessary for disulfide formation.
Suppression of the hyperoxia growth defects of sod2⌬ cell can arise from mutations in genes encoding components of the respiratory chain as reactive oxygen species are diminished (52). The observed suppression of sod2⌬ cell by matrix-localized hSod1 and CCS molecules cannot arise from such respiratory chain mutations because the transformants grow on the nonfermentable carbon glycerol.
The cytoplasm is predicted to contain no free aquo-copper ions (4). Copper ion buffering within the cytoplasm occurs by Sod1 and the two metallothioneins, Cup1 and Crs5. Complexation of copper ions by glutathione may represent a labile pool; however, the Ace1-modulated expression of Cup1 must limit any copper GSH pool. The present data suggest that mitochondrial copper uptake may also contribute to the cytosolic copper buffering.
The transporter responsible for mitochondrial copper ion uptake is unknown. It is clear that neither Cox17 nor Cox19 is responsible for delivering copper to this mitochondrial pool. Consistent with the present results showing that mitochondrial copper levels are unchanged in cells lacking either Cox17 or Cox19, we recently demonstrated that Cox17 is functional when localized exclusively to the IMS (53). The addition of zinc salts to the growth medium decreases mitochondrial copper levels but not cellular copper levels. Thus, copper ions may be imported into the mitochondrial matrix through a transporter(s) that also pumps zinc and manganese ions. If the mitochondrial copper transporter translocates multiple metal ions, the mitochondrion may have a buffering capacity for diverse metal ions. Cells lacking Mtm1 show no Sod2 activity, but mitochondrial Mn(II) levels are normal (54). Thus, mtm1⌬ cells are expected to contain a pool of non Sod2-bound Mn(II).
Mitochondria may contain a storage pool for metals besides copper. The pool of zinc in mammalian mitochondria increases with increasing cytosolic zinc (55,56). Likewise, labile iron levels increase significantly within the mitochondria of yeast mutants defective in iron-sulfur cluster formation and processing (57,58). The increase in the mitochondrial labile iron pool has deleterious consequences. The elevation in labile iron in mitochondria of cells depleted for Yfh1 or Isa1 results in an increase in petite formation (59,60). Whereas increases in the labile iron pool in mitochondria are deleterious, a 5-10-fold increase in the matrix copper pool is not associated with mitochondrial DNA damage. The ligands of the matrix copper must reduce the reactivity of this pool.
The significance of the matrix copper pool is unclear. Copper ions are needed for assembly of Cu A and Cu B sites in CcO as well as metallation of Sod1 within the IMS. A common pathway of copper ion delivery to the IMS may exist to provide copper ions for CcO assembly and metallation of Sod1. Cu(I) delivery to the IMS may occur through a permease within the OM or through a shuttle protein. Alternatively, copper may be delivered to the IMS from the matrix. The matrix copper pool may represent a storage pool of copper available to an IM transporter for delivery to the IMS. Attempts are being made to identify such a transporter.