Purification, Characterization, and Localization of Yeast Cox17p, a Mitochondrial Copper Shuttle*

Cox17p was previously shown to be essential for the expression of cytochrome oxidase in Saccharomyces cerevisiae. In the present study COX17 has been placed under the control of the GAL10 promoter in an autonomously replicating plasmid. A yeast transformant harboring the high copy construct was used to purify Cox17p to homogeneity. Purified Cox17p contains 0.2–0.3 mol of copper per mol of protein. The molar copper content is increased to 1.8 after incubation of Cox17p in the presence of a 6-fold molar excess of cuprous chloride under reduced conditions. An antibody against Cox17p was obtained by immunization of rabbits with a carboxyl-terminal peptide coupled to bovine serum albumin. The antiserum detects Cox17p in both the mitochondrial and soluble protein fractions of wild type yeast and of the transformant overexpressing Cox17p. Exposure of intact mitochondria to hypotonic conditions causes most of Cox17p to be released as a soluble protein indicating that the mitochondrial fraction of Cox17p is localized in the intermembrane space. These results are consistent with the previously proposed function of Cox17p, namely in providing cytoplasmic copper for mitochondrial utilization.

Cox17p is a low molecular weight protein of yeast (1) and mammalian cells (2). In Saccharomyces cerevisiae, mutations in COX17 cause a respiratory defect stemming from a specific deficiency in cytochrome oxidase, the activity of which depends on the presence of copper (1). The ability of high concentrations of copper in the growth medium to rescue the mutant phenotype indicated that the COX17 gene product is involved in mitochondrial copper metabolism, most likely by targeting cytoplasmic copper to the organelle (1). A role of Cox17p in copper homeostasis is also supported by its small size and high cysteine content, both features being common to other copper proteins such as metallothioneins (3) and Atx1p (4). The latter has recently been shown to mediate the transfer copper to Ccc2p (5), which functions in the maturation of the high affinity iron transporter Fet3p (6).
To better understand the role of Cox17p in cytochrome oxidase assembly we undertook to purify the protein with the aim of confirming its metal-binding properties and to eventually develop an in vitro system for studying the mechanism of copper addition to apocytochrome oxidase. Here we report a procedure capable of yielding a homogeneous preparation of Cox17p from a strain of yeast transformed with a multicopy plasmid containing COX17 under GAL10 regulation. The availability of pure Cox17p together with an antibody against a 20-amino acid long carboxyl-terminal peptide has enabled us to quantitate the cellular concentration of the protein, to study its subcellular distribution, and to demonstrate its ability to bind copper.

Construction of pG74/ST31
An 0.8-kilobase pair HindIII fragment containing COX17 was ligated to the HindIII site of YEp52 (7) yielding pG74/ST30, placing the gene behind the GAL10 promoter. To further increase the level of expression of COX17, this plasmid was modified by introducing a 4.8-kilobase pair BamHI fragment with GAL4 under the control of the ADH1 promoter. The configuration of the two genes in the plasmid, pG74/ST31, is shown in Fig. 1.

Preparation of an Antibody against Cox17p
The 20-mer NFIEKYKECMKGYGFEVPSAN, synthesized by Research Genetics (Huntsville, AL), was coupled to bovine serum albumin (BSA) 1 by a modification of the method of Hancock and Evan (8). The peptide (20 mg) was dissolved in 30 mM dithiothreitol and incubated at room temperature for 30 min. After removal of insoluble material by centrifugation, the reduced peptide was separated from dithiothreitol by passage through a Sephadex G-25 desalting column. Fractions absorbing at 280 nm were pooled and concentrated to 2 ml with a final yield of 12.5 mg of protein. BSA (30 mg) was dissolved in 1.5 ml of 10 mM KPO 4 , pH 7.5, dialyzed against the same buffer for 1.5 h, and activated by slow addition of 0.23 ml of the N-hydroxysuccinimide ester of 3-maleimidobenzoic acid dissolved in dimethylformamide at a concentration of 3 mg/ml. The mixture was stirred for 30 min at room temperature and passed through a Sephadex G-25 desalting column. Fractions with the activated BSA were pooled, concentrated in a SpeedVac, and adjusted to pH 7.4 with 1 M NaOH. Coupling was effected at a protein ratio of 4 mg of peptide per 5 mg of BSA by overnight incubation at room temperature. Analysis of the product by PAGE indicated that most of the peptide had been conjugated. This solution was used as antigen after the addition of 0.1 volume of sodium dodecyl sulfate. Rabbits were initially inoculated with 3 mg of protein in complete Freund's adjuvant. After 4 weeks they were boosted at 2-week intervals, the first two times with 1 mg of protein followed by 0.1 mg of protein. Maximal titer was attained 16 -20 weeks after the initial injection.

Purification of Cox17p
W303⌬COX17/ST31 is grown overnight in 100 ml of YPD medium. The overnight culture (15 ml) is transferred to 800 ml of YPD contain-* This work was supported by Research Grant GM50187 from the National Institutes of Health, United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ing 75 M CuSO 4 and grown for 17 h at 30°C with shaking. Cells are harvested under semi-sterile conditions by centrifugation at 800 ϫ g av , inoculated to 800 ml of YPGal supplemented with copper (2% galactose, 1% yeast extract, 2% peptone, 75 M CuSO 4 ), and shaken at 30°C for 6 -8 h. At this stage the culture is stored in the cold overnight for further processing. The following procedure is based on cells obtained from 20 flasks, equivalent to 16 liters of medium.
Step 1-Cells are harvested at 800 ϫ g av and washed with 1.5 liters of 1.2 M sorbitol. The yield should be 200 -300 g of cells, wet weight.
Step 2-The washed cells are suspended in Zymolyase digestion buffer (30 ml per 10 g of cells). The Zymolyase buffer consists of 1.2 M sorbitol, 30 mM KPO 4 , pH 7.5, 0.15 M ␤-mercaptoethanol, 0.5 mg/ml Zymolyase 20,000 (ICN Biochemicals). The cell suspension is incubated at 37°C for 2 h or until at least 80 -90% of the cells have been converted to spheroplasts.
Step 3-The spheroplast suspension is centrifuged at 2,600 ϫ g av for 10 min in a Sorvall centrifuge and washed twice with 1 liter of 1.2 M sorbitol. The spheroplasts are lysed in 600 ml of 0.4 M sorbitol, 20 mM Tris-HCl, pH 7.5, and 1 mM phenylmethylsulfonyl fluoride and homogenized in a Waring blendor for 40 s. The suspension is centrifuged at 640 ϫ g av for 10 min. The supernatant is separated from the loose pellet and centrifuged a second time in the same way to remove any remaining cells debris.
Step 4 -The supernatant from step 3 is centrifuged at 14,700 ϫ g av for 30 min to sediment mitochondria. The post-mitochondrial supernatant obtained from this centrifugation contains most of the Cox17p expressed from the plasmid. The post-mitochondrial supernatant can be stored at Ϫ80°C before proceeding to the next step.
Step 5-Solid ammonium sulfate is added to the post-mitochondrial supernatant to bring it to 55% saturation. The precipitate is removed by centrifugation at 14,000 ϫ g av for 20 min, and the supernatant is adjusted to 80% saturation in ammonium sulfate. The 55-80% ammonium sulfate precipitate is collected by centrifugation at 14,000 ϫ g av for 20 min, dissolved in the minimum volume of 20 mM Tris-HCl, pH 7.5 (30 -40 mg of protein/ml), and desalted on a 480-ml column of Sephadex G-50 (fine) equilibrated with 20 mM Tris-HCl, pH 7.5, 50 mM NaCl. Cox17p elutes as a trail after the main peak (Fig. 2). Fractions containing Cox17p clear of the main peak are pooled.
Step 6 -The pool from step 5 is applied to an 8-ml Pharmacia Mono Q (FPLC) column equilibrated with 20 mM Tris-HCl, pH 7.5. Following sample loading, the column is washed with 20 ml of 20 mM Tris-HCl, 50 mM NaCl and 80 ml of a 0.050 M to 0.4 M linear NaCl gradient in 20 mM Tris-HCl, pH 7.5. Fractions of 8 ml are collected and analyzed for Cox17p. Generally the protein is spread over 5 fractions between 0.1 and 0.2 M NaCl that are pooled and concentrated on a 1-ml hydroxyap-atite column (Bio-Rad HTP gel) equilibrated with 20 mM Tris-HCl, pH 7.5. After sample loading, the buffer is changed to 150 mM KPO 4 , pH 7.5. Fractions of 1 ml are collected. Cox17p elutes in approximately 4 ml as a single protein peak.
Step 7-The concentrated fraction from step 6 is loaded at a flow rate of 0.25 ml/min on a 22-ml column (1 cm in diameter) of Red Sepharose ("Procion," Pharmacia type RL-6B). The column is equilibrated and eluted with 20 mM Tris-HCl, pH 7.5. Cox17p does not adsorb on this column, and its elution is best monitored by silver staining. Contaminating proteins are recovered as a second protein peak.
Step 8 -The pooled fractions from the Sepharose Red column are concentrated on the Mono Q column (equilibrated in 20 mM Tris pH 7.5). The protein is eluted with 0.35 M NaCl and diluted with 4 volumes of 0.05% solution of trifluoroacetic acid. The sample is loaded on a 1 ϫ 25 cm C4 reverse-phase HPLC column (Primesphere 5 , Phenomenex, CA) equilibrated with 0.05% trifluoroacetic acid. The column is developed with a linear 0 -50% acetonitrile gradient containing 0.05% trifluoroacetic acid at a flow rate of 2.0 ml/min. Cox17p elutes as a single peak at 30 -35% acetonitrile. All the contaminating proteins elute between 35-50% acetonitrile.

Copper Assay
Copper was assayed by a modification of the method of Felsenfeld (9). Samples of protein adjusted to 2.1 ml with water were treated with 0.23 ml of 50% trichloroacetic acid and 50 l of 30% hydrogen peroxide and heated at 90°C for 3 min. Precipitated protein was removed by centrifugation at 400 ϫ g av for 5 min. The following reagents were added sequentially with mixing following transfer of 2 ml of the clarified solution to a fresh tube: 1) 0.5 ml of 0.25% 2,2Ј-biquinoline in glacial acetic acid, 2) 0.1 ml of 10% hydroxylamine, 3) 0.5 ml of saturated sodium acetate. The copper biquinoline complex was extracted into 0.6 ml of isoamyl alcohol by vigorous mixing for 15 s. The upper phase was separated by centrifugation at 400 ϫ g av , transferred to a spectrophotometer cuvette, and clarified by addition of 30 l of ethanol. Absorbance was read at 535 nm against a reagent blank treated as above. The concentration of copper was estimated from a standard curve obtained from a known solution of cupric sulfate.

Preparation of Mitochondria
For the localization and distribution of Cox17p, yeast mitochondria with intact outer membranes were prepared by the method of Glick and Pon (10). The supernatant fraction after sedimentation of mitochondria was centrifuged at 100,000 ϫ g av for 10 min. The 100,000 ϫ g supernatant is referred to as the post-mitochondrial supernatant.

Miscellaneous Procedures
Standard procedures were used for the preparation and ligation of DNA fragments and for transformation and recovery of plasmid DNA from Escherichia coli (11). Proteins were analyzed by PAGE with the buffer system of Laemmli (12). The concentration of acrylamide used was generally 15% with 15% glycerol added to the separation gel.  Western blots were treated with antibodies against the Cox17p carboxyl-terminal peptide followed by a second reaction with 125 I-protein A (13). Protein concentrations were determined by the method of Lowry et al. (14).

RESULTS AND DISCUSSION
Subcellular Localization of Cox17p in Wild Type and W303⌬COX17/ST31-In earlier experiments, Cox17p was detected as a biotinylated protein with a 7-kDa carboxyl-terminal extension containing a bacterial biotinylation signal (15). The introduction of the biotin tag was necessitated by the lack of an antibody against Cox17p. Most of the biotinylated Cox17p was found in the post-mitochondrial supernatant fraction, corresponding to the soluble cytosolic proteins of yeast. This was true irrespective of whether the COX17/BIO fusion gene was expressed from a multicopy plasmid or from a single chromosomally integrated copy of the gene (1).
The availability of an antibody against Cox17p made it possible to re-examine the subcellular localization of the native protein. These studies were done with the parental wild type strain W303-1B and with W303⌬COX17/ST31, a cox17 null mutant transformed with the wild type gene on an episomal plasmid (this strain will be referred to as "ST31 transformant"). The results of the Western analyses indicated that a substantial fraction of the protein is associated with mitochondria in both the wild type and the ST31 transformant (Fig. 3). The relative distributions of the protein in mitochondria and in the soluble post-mitochondrial supernatant fractions are different in the two strains. The concentrations of Cox17p in mitochondria and post-mitochondrial supernatants of wild type yeast are 0.65 g and 0.04 g of Cox17p per mg of protein, respectively. When normalized to the total protein recovered in the two fractions, these values indicate that approximately 60% of Cox17p is associated with mitochondria and 40% with the post-mitochondrial supernatant fraction. It is unlikely that the Cox17p detected in the post-mitochondrial supernatant is due to leakage. Western blot analysis of cytochrome b 2 failed to detect this intermembrane protein in the post-mitochondrial supernatant fraction even when as much as 200 g of protein of this fraction was applied to a single lane of a polyacrylamide gel (data not shown).
The concentrations of Cox17p in the ST31 transformant are 8 times higher in mitochondria and 100 times higher in the post-mitochondrial supernatant fraction. The transformant, therefore, has 9 times more Cox17p in the post-mitochondrial supernatant fraction than in mitochondria. The presence of 90% of Cox17p in the post-mitochondrial supernatant fraction of the ST31 transformant is also inconsistent with leakage of the protein from mitochondria during their preparation, but rather it suggests a limit to the amount of protein that can be taken up by the organelle.
The difference in the subcellular distribution of native and biotinylated Cox17p is surprising based on other studies in which the presence of the amino-terminal extension with the biotinylation sequence did not affect correct targeting/localization of the proteins (16 -19). All of the previously biotin-tagged proteins were inner membrane or matrix constituents that are translocated by the energy-dependent protein transport machinery of mitochondria (20). Cox17p differs from these proteins in lacking a mitochondrial import sequence and in having an intermembrane localization (see below). The absence of both a normal matrix targeting presequence and of an intermembrane sorting presequence (21,22) in Cox17p suggests that, like cytochrome c (23), it may be internalized by a different mechanism. The presence of a 7-kDa sequence could be inhibitory to this process.
Purification of Cox17p-The purification of Cox17p was attempted from several sources. Attempts to obtain the protein from a strain of E. coli transformed with COX17 on a high expression vector were unsuccessful. The best source of the protein was the post-mitochondrial supernatant of the ST31 transformant. Although the ST31 transformant was routinely grown in medium supplemented with 75 M CuSO 4 , the addition of extra metal did not affect the copper content of the purified protein. Cox17p was also purified from the mitochondrial fraction of the transformant and from the post-mitochondrial supernatant of commercial bakers' yeast. The yield of protein on a cell weight basis from the commercial source is about 100 times less, thereby requiring manipulation of much larger volumes.
In the absence of a functional assay for Cox17 it was necessary to use an immunological method of detection during the purification. As indicated in Table I, size separation of the 55-80% ammonium sulfate precipitate on Sephadex G-50 achieves a substantial initial purification of Cox17p. To minimize contamination by the major peak corresponding to proteins excluded by the sieving column, some loss of Cox17p is incurred (Fig. 2). Chromatography of the pool from Sephadex  a The concentration of Cox17p was determined by quantitative Western blot analysis. Purified Cox17p was used to obtain a standard curve relating amounts of protein to the density of the signals obtained with the antiserum against the carboxyl-terminal peptide. The density of the signals was quantitated with a Visage 110 Bioimager (Millipore Corp.). To increase the accuracy of the measurement, only a 3-fold range in the concentrations of the Cox17p was used for the standard curve, and the amount of the different fractions applied to the gel was adjusted so that the signals would fall within this range. b ND, not determined.
G-50 on Mono Q followed by a negative Red Sepharose step removes most of the contaminating proteins. The low levels of extraneous proteins still present at this stage are separated on a C4 reverse phase column (Fig. 4). The purification procedure results in a 240-fold increase in the signal relative to protein as judged by Western blot analysis, with a final yield of 8 -10% (Table I).
Properties of Purified Cox17p-The purity of Cox17p obtained either from the post-mitochondrial supernatant or mitochondria was confirmed by protein sequencing. The aminoterminal 5 residues matched the sequence encoded by COX17 starting from the second residue (threonine). The molar yield of amino acids was consistent with the amount of protein sequenced. There was no evidence of secondary contaminating sequences. The absence of the initiator methionine in the protein indicates that this residue is processed. Cox17p migrates with an apparent molecular mass of 14 kDa in SDS-polyacrylamide gels (see Fig. 4). The discrepancy in size is unlikely to be due to modifications or dimerization of the protein because the mass measured by mass spectrometry is in accord with the molecular weight calculated from the translated gene sequence.
The copper content of purified Cox17p was determined colorimetrically with 2,2-biquinoline as a copper-specific chelator and by atomic emission. Both methods yielded comparable values of 0.3 mol of copper per mol of protein. The latter method also failed to detect any cobalt, nickel, or zinc in the protein.
Based on its cysteine content Cox17p is expected to bind 2-3 mol eq of copper. The substoichiometric recovery of copper in the final protein suggests that the metal is lost during the purification procedure.
The copper binding capacity of Cox17p was assessed by reconstitution of dithiothreitol-reduced protein with cuprous chloride under limiting oxygen to prevent reoxidation. Under these conditions 1.8 mol of copper was bound per mol of protein.
No binding occurred when cupric sulfate was used under aerobic conditions. The Sephadex G-25 chromatography step used in the reconstitution achieves a complete separation of free from protein-bound copper (Fig. 5). Chromatography of the reconstituted protein on the sieving column leads to a release of most of the bound copper, indicating a low binding constant for the metal. This would suggest that the real stoichiometry of copper binding is probably greater than that measured by this method. Cox17p purified from the mitochondrial fraction of the ST31 transformant also had a copper content of approximately 0.3 mol/mol of protein.
Localization of Cox17p in the Intermembrane Space-As indicated above, 60% of Cox17p in wild type yeast is detected in mitochondria. It was of interest to determine whether the protein is located in the matrix or intermembrane space. Mitochondria prepared under conditions favoring maximal integrity of the outer membrane (10) were diluted either with buffered sorbitol or with buffer alone to induce rupturing of the outer membrane. The membrane fractions (mitochondria or mitoplasts) were separated from the soluble proteins released by the two treatments and the distributions of Cox17p, an intermembrane marker (cytochrome b 2 ), and a matrix protein (␣ketoglutarate dehydrogenase) were assessed by Western blot analysis. Almost all of Cox17p, cytochrome b 2 , and ␣-ketoglutarate dehydrogenase remained associated with mitochondria diluted in the presence of an osmotically stabilizing concentration of sorbitol (Fig. 6). Approximately 60% of Cox17p, however, was released into the soluble protein fraction when mitochondria were diluted in the absence of sorbitol. In contrast virtually all of cytochrome b 2 was recovered in the supernatant indicating quantitative lysis of the outer membrane. As expected all of ␣-ketoglutarate, the matrix marker, remained associated with the mitoplast fraction. The solubilization of most of Cox17p by the hypotonic treatment indicates that the Binding of copper to Cox17p. All steps were preformed anaerobically, with deaerated solutions saturated with and kept under nitrogen gas. To reduce Cox17p, 0.7 mg of protein was incubated for 10 min at room temperature in the presence of 150 mM dithiothreitol. The protein was separated from dithiothreitol on a 10-ml Sephadex G-25 (superfine) column equilibrated with 20 mM Tris, pH 7.5, and was immediately incubated for 20 min at room temperature with 6 molar equivalents of cuprous chloride. Free copper was removed on the same Sephadex G-25 desalting column. Fractions of 1 ml were collected and assayed for copper. Elution of Cox17p was monitored by absorbance at 280 nm and by Western blot analysis. All the protein eluted with the first faster migrating peak. Copper was assayed colorimetrically and is reported as absorbance at 535 nm corresponding to the absorption peak of the 2,2-biquinoline copper complex (solid bars). The absorption at 280 nm coeluting with the free copper appears to be due to some low molecular weight material that is stripped from the column by copper. It is also seen when the column is charged with either cuprous or cupric salts alone. protein is located in the intermembrane space of mitochondria (10). The complete release of cytochrome b 2 by the osmotic shock of mitochondria indicates that the Cox17p recovered in the membrane fraction is not due to partial lysis of the outer membrane. The fraction of Cox17p present in the mitoplasts is not located in the matrix compartment because of its susceptibility to proteinase K. An alternative explanation to account for its incomplete release is that a fraction of Cox17p may be physically complexed to another component of mitochondria. It is unlikely that Cox17p is bound to Sco1p since its relative distribution is not altered in yeast harboring SCO1 on a multicopy plasmid (data not shown).
In addition to high concentrations of copper in the medium, the cytochrome oxidase deficiency of cox17 mutants can be partially suppressed by SCO1 when present on a high copy plasmid (24). SCO1 has been shown by Schulze and Rodel (25) to be essential for expression of cytochrome oxidase. These observations were taken to indicate that Sco1p functions in the pathway of copper addition to apocytochrome oxidase (24). A role of Sco1p in mitochondrial copper metabolism was also supported by the presence in the protein of a domain with sequence similarity to subunit 2 of cytochrome oxidase (24). This domain includes two conserved cysteine residues that are known to be copper ligands of subunit 2 (26). Sco1p is an insoluble protein of mitochondria that was localized in the inner membrane (27). The extra membrane domain of Sco1p probably faces the intermembrane space. This is supported by the sensitivity of Sco1p to proteinase K in mitoplasts but not in intact mitochondria (Fig. 7). This suggests that the proposed copper binding site of Sco1p and the copper site of subunit 2 of cytochrome oxidase (26) face the same mitochondrial compartment as Cox17p. Such a localization would be in accord with our previous model in which Cox17p targets copper specifically to mitochondria by transferring the metal to Sco1p. At present it is still not clear whether Sco1p is a carrier that transports copper to the matrix or whether it acts as a transferase that adds copper directly to cytochrome oxidase. If copper addition occurs after membrane insertion of subunit 2, a role of Sco1p as a copper transferase would the more attractive of the two models (Fig. 8). Since cytochrome oxidase is the only known copper-containing protein of yeast mitochondria, it is not clear whether the presence of copper in the matrix is required. FIG. 8. Models of the roles of Cox17p and Sco1p in cytochrome oxidase assembly. The model assumes that Cox17p shuttles between the cytoplasm and the intermembrane space of mitochondria where it delivers copper to Sco1p. In the upper half of the cartoon, Sco1p functions as a carrier that translocates copper to the matrix. In the lower half Sco1p transfers copper to the copper-bearing subunits 1 and 2 of cytochrome oxidase. The steps between the Ctr1p-dependent uptake of copper (28) and the binding of cytoplasmic copper by Cox17p are not known.
FIG. 6. Localization of Cox17p in the mitochondrial intermembrane space. Mitochondria prepared from the wild type strain W303-1B by the method of Glick and Pon (10) were suspended in 20 mM Hepes, pH 7.5, 0.6 M sorbitol. The mitochondria were lysed by dilution with 7 volumes of 10 mM Hepes, pH 7.5. Control samples of mitochondria were treated identically except that the dilution buffer contained 0.6 M sorbitol. The diluted samples were centrifuged at 100,000 ϫ g av for 10 min, and the pellets consisting of mitochondria or mitoplasts were dissolved in Laemmli buffer (12). Samples of the pellets (P) and supernatants (S), adjusted for volume, were separated on a 15% or 10% polyacrylamide gels. Panels A, the amount of mitochondrial/mitoplast protein loaded per lane was 70 g. Cox17p was detected immunochemically as described in the legend to Fig. 2. Panels B, the amount of mitochondrial/mitoplast protein loaded per lane was 8.25 g. The Western blot was treated with antibody against cytochrome b 2 . Panels C, the amount of mitochondrial/mitoplast protein loaded per lane was 8.25 g. The Western blot was treated with antibody against the dehydrogenase component of yeast ␣-ketoglutarate dehydrogenase complex. Panels D, mitochondria were lysed in Hepes buffer and separated into soluble intermembrane proteins and mitoplasts. Samples of both fractions were digested with proteinase K at a final concentration of 0.1 mg/ml for 45 min at 4°C. The amount of mitoplast proteins loaded per lane was 16 g. The experimental conditions were identical to those described in the legend to Fig. 6, except that the dilutions with 0.6 M sorbitol/Hepes or Hepes alone were done in the presence and absence of 100 g/ml proteinase K. Mitochondrial/mitoplast (10 g) proteins were separated on a 12% polyacrylamide gel and, after transfer to nitrocellulose, probed with antibodies to Sco1p and to subunit 5 of cytochrome oxidase.