A Role for Pet100p in the Assembly of Yeast Cytochrome c Oxidase INTERACTION WITH A SUBASSEMBLY THAT ACCUMULATES IN A pet100 MUTANT*

The biogenesis of multimeric protein complexes of the inner mitochondrial membrane in yeast requires a number of nuclear-coded ancillary proteins. One of these, Pet100p, is required for cytochrome c oxidase. Previous studies have shown that Pet100p is not required for the synthesis, processing, or targeting of cytochrome c oxidase subunits to the mitochondrion nor for heme A biosynthesis. Here, we report that Pet100p does not affect the localization of cytochrome c oxidase subunit polypeptides to the inner mitochondrial membrane but instead functions after they have arrived at the inner membrane. We have also localized Pet100p to the inner mitochondrial membrane in wild type cells, where it is present in a subassembly (Complex A) with cytochrome c oxidase subunits VII, VIIa, and VIII. Pet100p does not interact with the same subunits after they have been assembled into the holoenzyme. In addition, we have identified two subassemblies that are present in pet100 null mutant cells: one subassembly (Complex A (cid:1) ) is composed of subunits VII, VIIa, and VIII but not Pet100p, and another subassembly (Complex B) is composed of subunits Va and VI. Because pet100 null mutant cells lack assembled cytochrome c oxidase but accumulate Complexes A (cid:1) and B it appears likely that these subassemblies of cytochrome c oxidase subunits are intermediates along an assembly pathway for holocytochrome dried, and resuspended protein Sucrose Gradient Analysis of Lysed Mitochondria— To identify con- ditions that give the best solubilization of yeast mitochondrial mem-branes and recovery of active cytochrome c oxidase, we compared a variety of detergents (Triton X-100, lauryl maltoside, cholate, deoxy-cholate, octyl glucoside, and Zwittergent-14) and detergent/protein ra- tios. These studies revealed that Triton X-100 at a ratio of 4 mg of Triton X-100 to 1 mg of mitochondrial protein was best for membrane solubilization and that optimal recoveries of enzyme activity and peak sharpness were achieved on 5–50% sucrose gradients containing 7 m M Triton X-100. These conditions were incorporated into the following protocol. An aliquot (1 mg of protein) of freshly isolated mitochondria was washed in SEM buffer (250 m M sucrose, 1 m M Na 2 EDTA, 10 m M MOPS, pH 7.2) and centrifuged for 4 min at 4 °C at 16,000 (cid:3) g in an Eppendorf centrifuge. The resulting pellet was re-suspended in 300 (cid:3) l of 2 (cid:3) lysis buffer (20 m M KPO 4 , 15% glycerol, 2 m M EDTA, pH 7.4). After a 5-min incubation at room temperature, Triton X-100 (Surfact-Amps X-100, Pierce Chemicals) was added to the suspension to a final concentration of 3 mg/mg mitochondrial protein. The incubation was continued for another 10 min at room temperature. Lysed mitochondria were applied to a linear 5–50% sucrose gradient containing 10 m M KPO 4 (pH 7.0) and 0.1% Triton X-100 made with a Hoefer Scientific Instruments SG series gradient maker. The gradient was centrifuged overnight (12–16

The biogenesis of multimeric protein complexes of the inner mitochondrial membrane in yeast requires a number of nuclear-coded ancillary proteins. One of these, Pet100p, is required for cytochrome c oxidase. Previous studies have shown that Pet100p is not required for the synthesis, processing, or targeting of cytochrome c oxidase subunits to the mitochondrion nor for heme A biosynthesis. Here, we report that Pet100p does not affect the localization of cytochrome c oxidase subunit polypeptides to the inner mitochondrial membrane but instead functions after they have arrived at the inner membrane. We have also localized Pet100p to the inner mitochondrial membrane in wild type cells, where it is present in a subassembly (Complex A) with cytochrome c oxidase subunits VII, VIIa, and VIII. Pet100p does not interact with the same subunits after they have been assembled into the holoenzyme. In addition, we have identified two subassemblies that are present in pet100 null mutant cells: one subassembly (Complex A) is composed of subunits VII, VIIa, and VIII but not Pet100p, and another subassembly (Complex B) is composed of subunits Va and VI. Because pet100 null mutant cells lack assembled cytochrome c oxidase but accumulate Complexes A and B it appears likely that these subassemblies of cytochrome c oxidase subunits are intermediates along an assembly pathway for holocytochrome c oxidase and that Pet100p functions in this pathway to facilitate the interaction(s) between Complex A and other cytochrome c oxidase subassemblies and subunits.
Cytochrome c oxidase is a multimeric protein complex of the mitochondrial inner membrane in eukaryotes. It is assembled from polypeptides encoded by both nuclear and mitochondrial genomes (1) and plays an important role in the regulation of cellular energy production (2,3). In Saccharomyces cerevisiae, active cytochrome c oxidase is composed of nine different polypeptide subunits as well as four redox-active metal centers (heme a, heme a 3 , Cu A , and Cu B ) (1,4). All nine subunit polypeptides are required for a fully active holocytochrome c oxidase (1,4,5). Subunits I, II, and III are products of mitochondrial genes (COX1, COX2, and COX3, respectively) (1,6) and make up the catalytic core of the enzyme (7,8). All three of these polypeptides are translated on mitochondrial ribosomes and are inserted co-translationally into the mitochondrial inner membrane (9), a process termed mitochondrial export (10,11). Subunits IV, V, Va or Vb, VI, VII, and VIIa are encoded by nuclear COX genes (COX4, COX5a or 5b, COX6, COX7, COX8, and COX9, respectively). Some of them function to modulate catalysis (12,13), whereas others function to affect assembly or stability of the holoenzyme (1). These polypeptides are translated by cytosolic ribosomes, imported into the mitochondrion and assembled, along with the mitochondrially encoded subunits and the prosthetic groups, into a functional holoenzyme.
The two different pathways followed by compartmentally separated gene products, import of nuclear-encoded proteins (14) and export of mitochondrially encoded proteins (10,11), must converge at some point into an assembly pathway during the biogenesis of holocytochrome c oxidase. Although the mitochondrial import pathways have been elucidated (for review, see Ref. 15) and the mitochondrial export pathway is beginning to yield to analysis (16), the assembly pathway for cytochrome c oxidase is still poorly understood. However, it is now clear that assembly of cytochrome c oxidase requires proteins encoded by a number of different nuclear genes (1,(17)(18)(19)(20)(21)(22). Some of the proteins act globally and affect assembly of many mitochondrial proteins. These include heat shock proteins, which function as part of the import or sorting pathways that localize nuclear-coded polypeptides to the mitochondrial inner membrane (23), and ATP-dependent proteases (i.e. members of the YTA10 -12 complex) that function in the turnover of unassembled mitochondrially encoded polypeptides (20,24). Some other nuclear-coded proteins appear to be more specific and affect only cytochrome c oxidase assembly.
Among the genes with specific effects on cytochrome c oxidase assembly are three genes (COX10, COX15, and YAH1) required for the synthesis of heme A (25)(26)(27)(28), four genes (SCO1, COX11, COX17, and COX23) required for the recruitment and uptake of copper into mitochondria and its assembly into holocytochrome c oxidase (29 -37), and one gene (OXA1/PET1402) that is part of a translocation channel involved in mitochondrial export (16). In addition to the above genes, there are at least five other genes (PET100, PET117, PET191, COX14, and COX16) (38 -42) that encode proteins that have been proposed to function as "assembly facilitators" (1). These assembly facilitators do not appear to affect other cytochromes of the respiratory chain.
To examine how assembly facilitators function we have been studying Pet100p. Previously, we cloned and characterized the PET100 gene and found that Pet100p is not required for synthesis, processing, or localization of cytochrome c oxidase subunits to mitochondria (38), does not affect the biogenesis of other mitochondrial cytochromes, and is not required for heme A synthesis. We have also found that the COOH-terminal 62 amino acids in Pet100p are important for Pet100p function (41). In addition, we have found that Pet100p is not required for normal mitochondrial morphology (43) and, hence, does not affect cytochrome c oxidase assembly indirectly through a generalized effect on mitochondrial structure.
In this study, we sought to determine when and how Pet100p functions in the assembly of holocytochrome c oxidase. First, we determined the intracellular location of Pet100p. Next, we examined its relationship to holocytochrome c oxidase, its polypeptide subunits, and cytochrome c oxidase subcomplexes. Finally, we looked at the effects of a pet100 null mutation on the localization of cytochrome c oxidase subunits and the intracellular levels of subassemblies in which they reside. Isolation of Mitochondria and Cytosol-For preparation of mitochondrial and cytosolic fractions, cells were grown to mid-exponential phase, harvested, and spheroplasted as described (12). All subsequent steps were performed at 4°C. Spheroplasts were harvested by centrifugation (5 min at 3,000 ϫ g), washed gently in post-spheroplast buffer (1.5 M sorbitol, 1 mM Na 2 EDTA, 0.1% BSA, 1 pH 7.0), and sedimented at 3,000 ϫ g for 5 min. Washed spheroplasts were resuspended in lysis buffer (0.6 M mannitol, 2 mM Na 2 EDTA, 0.1% BSA, pH 7.4), lysed in a Sorvall Omnimixer (Newton, CT) at low speed for 3 s and at full speed for 45 s, and then centrifuged for 5 min at 1,900 ϫ g to pellet unbroken cells, nuclei, and debris. The resulting supernatant was decanted and centrifuged for 10 min at 12,100 ϫ g to pellet mitochondria. The mitochondrial pellet was washed by resuspension in mitochondrial lysis buffer minus BSA (pH 7.0), homogenized with a glass/Teflon homogenizer, and centrifuged at 1,651 ϫ g for 5 min. The resulting supernatant was decanted and centrifuged at 23,500 ϫ g for 10 min to pellet the mitochondria. The post-mitochondrial supernatant, collected after the 23,500 ϫ g centrifugation, was used as the cytosolic fraction.

Strains and
Subfractionation of Mitochondria-Mitochondrial subfractions were prepared by a modification of the digitonin fractionation procedure (46). To prepare mitoplasts, mitochondria were suspended at a protein concentration of 10 mg/ml of mitoplast suspension buffer (0.6 M mannitol, 10 mM NaPO 4 , pH 7.0) and a 1-ml portion of the suspension was treated for 1 min on ice with 0.35 mg of digitonin/mg of mitochondrial protein. All subsequent steps were performed at 4°C. The suspension was diluted immediately with 5 volumes of mitoplast suspension buffer and centrifuged for 10 min at 12,000 ϫ g max . The pellet is the mitoplast fraction. The supernatant was carefully separated from the pellet and centrifuged for 60 min at 144,000 ϫ g max . The resulting pellet and the supernatant were used as the outer mitochondrial membrane and the intermembrane space fractions, respectively. The inner membrane and matrix fractions were prepared from the mitoplasts, collected above, as follows. Mitoplasts were resuspended at 10 mg of protein/ml in 100 mM NaPO 4 (pH 7.0) and sonicated for 20 s at 50 watts with a Branson Sonifier (model W 185) equipped with a microtip. Sonicated mitoplasts were centrifuged for 20 min at 95,000 ϫ g max . The pellet is the inner membrane fraction and the supernatant is the matrix fraction. The pellet was washed by resuspension in 100 mM NaPO 4 (pH 7.0) and recovered by centrifugation at 95,000 ϫ g max for 20 min.
Carbonate Extraction of Mitochondria-An aliquot (1 mg of protein) of isolated mitochondria was resuspended in 0.1 M Na 2 CO 3 (pH 12) to a final concentration of 0.15 g of protein/ml and incubated on ice for 30 min (29). Non-extractable integral membrane proteins were pelleted by centrifugation at 100,000 ϫ g max for 1 h. Extractable peripheral and soluble proteins were recovered from the supernatant by precipitation with ice-cold trichloroacetic acid. The supernatant was adjusted to a final concentration of 10% trichloroacetic acid, incubated on ice for 15 min, and centrifuged for 10 min at 15,000 ϫ g max . Both non-extractable and extractable protein pellets were washed once in ice-cold 100% acetone, air dried, and resuspended in SDS-PAGE protein dissociation buffer (20 mM sodium phosphate, pH 6.8, 4% SDS, 8% glycerol, 40 mM dithiothreitol).
Sucrose Gradient Analysis of Lysed Mitochondria-To identify conditions that give the best solubilization of yeast mitochondrial membranes and recovery of active cytochrome c oxidase, we compared a variety of detergents (Triton X-100, lauryl maltoside, cholate, deoxycholate, octyl glucoside, and Zwittergent-14) and detergent/protein ratios. These studies revealed that Triton X-100 at a ratio of 4 mg of Triton X-100 to 1 mg of mitochondrial protein was best for membrane solubilization and that optimal recoveries of enzyme activity and peak sharpness were achieved on 5-50% sucrose gradients containing 7 mM Triton X-100. These conditions were incorporated into the following protocol. An aliquot (1 mg of protein) of freshly isolated mitochondria was washed in SEM buffer (250 mM sucrose, 1 mM Na 2 EDTA, 10 mM MOPS, pH 7.2) and centrifuged for 4 min at 4°C at 16,000 ϫ g in an Eppendorf centrifuge. The resulting pellet was re-suspended in 300 l of 2ϫ lysis buffer (20 mM KPO 4 , 15% glycerol, 2 mM EDTA, pH 7.4). After a 5-min incubation at room temperature, Triton X-100 (Surfact-Amps X-100, Pierce Chemicals) was added to the suspension to a final concentration of 3 mg/mg mitochondrial protein. The incubation was continued for another 10 min at room temperature. Lysed mitochondria were applied to a linear 5-50% sucrose gradient containing 10 mM KPO 4 (pH 7.0) and 0.1% Triton X-100 made with a Hoefer Scientific Instruments SG series gradient maker. The gradient was centrifuged overnight (12-16 h) at 284,000 ϫ g max in a Beckman SW 41 rotor at 4°C. The gradient was fractionated into 400-l aliquots, collected from the top of the gradient, using an Isco Density Fractionator (model 185). Sucrose densities were determined by measuring the refractive indices of gradient fractions with a refractometer. Aliquots (50 l) of each fraction were precipitated with 200 l of 100% acetone containing 25 g of BSA at Ϫ20°C for 30 min. The precipitate was collected by centrifugation in a Microfuge (10 min at 4°C, 16,000 ϫ g). Pellets were washed once with 80% acetone for 10 min at 4°C and collected by centrifugation as above. The supernatant was removed by aspiration and the remaining acetone was allowed to evaporate. Pellets were resuspended in 50 l of 2ϫ SDS-PAGE protein dissociation buffer (20 mM NaPO 4 , pH 6.8, 4% SDS, 40 mM dithiothreitol, 8% glycerol) and stored frozen until ready for analysis by SDS-PAGE.
Purification of Cytochrome c Oxidase and Subunit Polypeptides-Yeast holocytochrome c oxidase was prepared by Method 1 described in Poyton et al. (4). Holoenzyme preparations of various purities were obtained by taking fractions from different points in the purification procedure. To prepare holocytochrome c oxidase for subunit isolation we used Method 2 (4). Cytochrome c oxidase subunit polypeptides were prepared by reversed phase high performance liquid chromatography, as described (47). The purity of each subunit was assessed by SDS-PAGE and microsequencing (47). Only subunit preparations that contained a single amino-terminal microsequence were used.
Preparation of Antibodies-Polyclonal antibodies were prepared to Pet100p, holocytochrome c oxidase, and subunits I, II, IV, Va, VI, VII, VIIa, and VIII of cytochrome c oxidase. For Pet100p, an antibody was obtained against a peptide synthesized to correspond to residues 93 to 111 at the carboxyl terminus of the PET100 gene product. The peptide was made by BioSynthesis Inc. and coupled to maleimide-activated keyhole limpet hemacyanin (Pierce) according to the manufacturer's directions. The peptide-keyhole limpet hemacyanin conjugate was mixed with Freund's Complete adjuvant (Sigma) and 1 mg of conjugate was injected subcutaneously into New Zealand White rabbits. Booster injections (0.5 mg) were given with Freund's Incomplete adjuvant at 2 and 8 weeks. Antisera were collected at 2-week intervals with the terminal collection at 12 days following the final boost. Antibodies to the holocytochrome c oxidase and subunits were prepared as described (48). The antiserum to subunit V was made to a synthetic peptide that duplicates the 20-amino acid sequence common to both Va and Vb at their carboxyl terminus. The IgG fraction was prepared from antiserum by ammonium sulfate fractionation. An aliquot (2 ml) of antisera was diluted with an equal volume of 10 mM NaPO 4 (pH 7.5), 15 mM NaCl. IgG was precipitated twice with ammonium sulfate at 40% (w/v) with stirring for 10 min at 25°C, and collected by centrifugation at 3,000 ϫ g for 5 min. The final precipitate was resuspended in 1 ml of 10 mM NaPO 4 (pH 7.5), 15 mM NaCl, dialyzed against phosphate-buffered saline, and microcentrifuged at 14,000 ϫ g for 1 min. The supernatant was collected and used as the IgG fraction.
Immunoprecipitation-The IgG fraction of each antisera was bound to Protein A-Sepharose CL-4B beads (Amersham Biosciences), prepared in 100 mM NaPO 4 (pH 7.5), 100 mM NaCl, as follows. An aliquot (100 l) of the IgG fraction was absorbed to 100 l of prepared beads in 1 ml of immunobuffer (10 mM NaPO 4 (pH 7.4), 15 mM NaCl, 1% Triton X-100) at 4°C for 5 h with rocking. Unbound antiserum was removed by washing beads twice with immunobuffer. For immunoprecipitation, IgG bound beads were incubated with either lysed mitochondria or sucrose gradient fractions from lysed mitochondria, as indicated in the appropriate figure legends. In each case, the total volume was adjusted to 1.3 ml with immunobuffer and the samples were incubated at 4°C for 1 h with rocking. To remove unbound protein, beads were washed four times with immunobuffer and then once with water. For Western analysis, protein was released from antibody beads in an equal volume of dissociation buffer (20 mM NaPO 4 (pH 6.8), 4% SDS, 8% glycerol, 40 mM dithiothreitol) and boiled 2 min prior to loading on SDS-polyacrylamide gels.
Miscellaneous Methods-Western analysis and SDS-PAGE were performed on 16% polyacrylamide gels containing 10% glycerol and 3.6 M urea, as described (38). Protein determination was performed by the Lowry assay (49) using BSA as a standard. Cytochrome c oxidase activity assays were performed as described (4).

PET100 Is Not Required for Localization of Cytochrome c
Oxidase Subunits to the Inner Mitochondrial Membrane-Previously, we have found that PET100 is required for the biogenesis of holocytochrome c oxidase but is not essential for the synthesis of cytochrome c oxidase subunit polypeptides, their processing, or their localization to the mitochondrion (38). These findings suggested that Pet100p functions late in the biogenesis of cytochrome c oxidase, either to sort the subunit polypeptides to the inner mitochondrial membrane or to assemble them into the holoenzyme once they have arrived at the inner membrane. To address which of these possibilities is correct we compared the distribution of cytochrome c oxidase subunits in cell and mitochondrial subfractions from JM43 and JM43GD100, a pet100 null mutant. From the results shown in Fig. 1A it is clear that cytochrome c oxidase subunit polypeptides could be detected in all cell fractions from JM43 that contained the inner mitochondrial membrane (i.e. mitochondria, mitoplasts, and the isolated inner membrane fraction). They were not detected in the cytosol or intermembrane space fractions. Low levels of subunits IV and VI were also detected in the matrix fraction. This is not surprising because these polypeptides are hydrophilic peripheral proteins, are easily dislodged from the membrane and holoenzyme (4,50), and their mammalian homologues (subunits Vb and Va, respectively) (13) are bound on the matrix side of the inner membrane (7,8). Low levels of yeast cytochrome c oxidase subunits I, II, III, VII, VIIa, and VIII were also detected in the outer membrane fraction. These polypeptides are hydrophobic integral proteins of the inner membrane that are only dislodged in the presence of detergent (47). Consequently, their presence in the outer membrane fraction most likely results from a low level of cross-contamination of the outer membrane fraction by fragments of the inner membrane.
Cytochrome c oxidase subunits are detectable in the mitoplast and inner membrane fractions, as well as in mitochondria from JM43GD100 (Fig. 1B). Although the levels of subunits I, II, and III are slightly reduced in JM43GD100 mitochondria (38) they are, nevertheless, detected only in those fractions that contain the inner membrane. One interesting difference between JM43 and JM43GD100 is that subunit IV is present in the matrix fraction but not the inner membrane from JM43GD100 (Fig. 1). This finding suggests that subunit IV is less tightly bound to the inner mitochondrial membrane of JM43GD100 than that of JM43, probably because of the absence of assembled cytochrome c oxidase. These results with JM43GD100 indicate that, with the possible exception of subunit IV, Pet100p is not required for the localization of subunits to the inner membrane.
Pet100p Is an Integral Protein of the Inner Mitochondrial Membrane-The above findings suggested that Pet100p functions inside of the mitochondrion. This, together with the fact that the DNA sequence of PET100 predicts the presence of a mitochondrial targeting sequence (38), led us to examine whether Pet100p does, in fact, reside in the mitochondrion. Subcellular and submitochondrial fractions were subjected to SDS-PAGE and analyzed by immunoblotting with an anti-Pet100p antibody. This polyclonal antibody, anti-Pet100C, was prepared against a 19-amino acid peptide synthesized to correspond to that predicted from the carboxyl terminus of the coding region of the PET100 gene ( Fig. 2A). This antibody detects an antigen in a whole cell extract from JM43 but not from JM43GD100 (Fig. 2B). This antigen has an apparent molecular weight of 13,500, which is consistent with that predicted for Pet100p from the sequence of its gene (38). Because the antigen recognized by anti-Pet100C is present in JM43 but absent from JM43GD100 and has a size similar to that predicted for Pet100p, we conclude that anti-Pet100C recognizes and is specific for Pet100p.
From Fig. 2C it is clear that Pet100p is present in mitochondria but not in the cytosol. It is also clear that Pet100p is located predominantly in the inner mitochondrial membrane; it is not present in the matrix or inter-membrane space. The faint antigenic band seen in the outer membrane fraction is most likely because of partial cross-contamination of the outer membrane fraction with a small amount of inner membrane during the mitochondrial subfractionation procedure.
To determine whether Pet100p is an integral or peripheral membrane protein of the inner mitochondrial membrane we incubated JM43 mitochondria with sodium carbonate (pH 12) (51). These conditions convert mitochondria into open sheets and release both soluble proteins of the matrix and intermembrane space as well as peripheral proteins; they do not release integral proteins. Pet100p is not extracted with sodium carbonate (Fig. 2D), indicating that it behaves like an integral membrane protein. In contrast, a portion of cytochrome c oxidase subunit IV is extracted with sodium carbonate. The extracted subunit IV probably represents the population of this polypeptide that is not assembled into the holo-enzyme or its subcomplexes but which is present in the mitochondrial matrix (Fig. 1). The remainder of subunit IV is not extracted, presumably because it is assembled in the holo-enzyme. The finding that Pet100p is an integral membrane protein is consistent with its predicted sequence, which contains a hydrophobic region that is characteristic of a trans-membrane domain (38).
Pet100p Co-sediments in a Sucrose Gradient with Holocytochrome c Oxidase and a Subcomplex Containing Subunits VII, VIIa, and VIII-To ask if Pet100p interacts directly with holocytochrome c oxidase or one or more of its subunit polypeptides we first analyzed the distribution of holocytochrome c oxidase and Pet100p in 5-50% sucrose gradients containing 7 mM Tri-ton X-100. A representative gradient is shown in Fig. 3A. Cytochrome c oxidase activity from mitochondrial lysates sediments between 18 and 22% sucrose. As expected, these fractions contain all of the subunit polypeptides of cytochrome c oxidase (Fig. 3B). Confirmation that holocytochrome c oxidase sediments in this region of the gradient comes from the observation that a cox6 null mutant strain, which lacks assembled cytochrome c oxidase, also lacks subunits in this region of sucrose gradients from lysed mitochondria (see below). From Fig. 3B it is clear that some subunits of cytochrome c oxidase are present in gradient fractions that do not include the holoenzyme. Subunits IV, VI, and to a lesser extent III appear in fractions with lighter densities (i.e. 8 to 12% sucrose) than the holoenzyme, whereas subunits VII, VIIa, and VIII appear in gradient fractions with heavier densities (i.e. 39 -48% sucrose). The presence of subunits in these fractions is not an artifact resulting from the disassembly of holocytochrome c oxidase during centrifugation because when the holoenzyme fraction (18 -22% sucrose) is collected and re-sedimented on a sucrose gradient, subunit polypeptides are only found in those gradient fractions that contain the active enzyme (data not shown).
Pet100p is distributed in the sucrose gradient in bimodal fashion (Fig. 3C). One peak co-sediments with holocytochrome c oxidase, between 18 and 22% sucrose, whereas the other co-sediments with subunits VII, VIIa, and VIII, between 39 and 48% sucrose. The observation that some subunits of cytochrome c oxidase do not migrate with the holenzyme in the sucrose gradient shown in Fig. 3B and the finding that Pet100p co-sediments with both the holoenzyme and subunits VII, VIIa, and VIII (Fig. 3C), raised two questions. First, does Pet100p associate with holocytochrome c oxidase and a subcomplex composed of subunits VII, VIIa, and VIII? Second, are the subunits that do not migrate with the holoenzyme assembled with each other into subcomplexes?
Pet 100p Co-Immunoprecipitates with a Subassembly Containing Subunits VII, VIIa, and VIII-To determine whether the Pet100p that sediments between 39 and 48% is in a complex with subunits VII, VIIa, or VIII, we pooled the fractions from this region of the gradient and subjected them to immunoprecipitation using anti-Pet100C sera. The immunoprecipitates were then immunoblotted with antisera to subunits VIIa, VII, or VIII. We also examined immunoprecipitates formed when the pooled sucrose gradient fractions are incubated with antisera to either subunit VIIa or VIII. Each antiserum is specific to the subunit to which it was made (Fig. 4A). From Fig.  4B it is clear that subunits VII, VIIa, and VIII are precipitated by anti-Pet100C (lane 1), that subunits VII, VIIa, VIII, and Pet100p are precipitated by anti-VIIa (lane 2), and that subunits VII, VIIa, VIII, and Pet100p are precipitated by anti-VIII. Preimmune serum does not precipitate subunits VII, VIIA, or VII from these pooled fractions (Fig. 4C, lane 2). The findings that Pet100p, subunit VII, subunit VIIa, and subunit VIII co-sediment on a sucrose gradient and co-immunoprecipitate indicate that Pet100p forms a subassembly with subunits VII, VIIa, and VIII. This subassembly is designated here as Complex A.
Pet100p Does Not Form a Stable Complex with Holocytochrome c Oxidase-To ask if the Pet100p that sediments between 18 and 22% sucrose is complexed with holocytochrome c oxidase we subjected pooled fractions from this region of the gradient to immunoprecipitation with anti-Pet100C. We also used an antiserum to cytochrome c oxidase subunit VIII as a positive control and preimmune serum as a negative control. Immunoprecipitates were subjected to SDS-PAGE and immu- FIG. 3. Sucrose gradient analysis of detergent-solubilized JM43 mitochondria. Panel A, sedimentation of cytochrome c oxidase activity from lysed JM43 mitochondria on a sucrose gradient. Mitochondria (1 mg of protein) were lysed with Triton X-100 and sedimented on a linear 5-50% sucrose gradient as described under "Experimental Procedures." Samples of every gradient fraction were taken to determine enzyme activity (f), sucrose concentration (E), and protein concentration (ϫ). Enzyme activity is expressed as a first-order rate constant, K. Panels B and C, Western blots of sucrose gradient fractions from panel A. Aliquots of each sucrose gradient fraction were concentrated by acetone precipitation and subjected to SDS-PAGE. Gels were then transferred to nitrocellulose, and blotted with an antiserum that recognizes all cytochrome c oxidase subunits (panel B) or anti-Pet100C (panel C), and detected by horseradish peroxidase-linked secondary antibodies. The asterisk denotes a contaminant antigen that runs below subunit VI; it is not a subunit of cytochrome c oxidase. The sucrose concentrations are indicated above the fraction numbers and the location of Complex A in the gradient is indicated above the gel. The lane at the left (M) was loaded with 20 g of yeast mitochondrial protein.
noblotted with a mixture of antibodies to all subunits of cytochrome c oxidase (Fig. 5A, lanes 2-4). The heavy band seen in lanes 2-4 is the IgG heavy chain, which is recognized by the anti-rabbit horseradish-linked donkey secondary antibody. Because it has the same apparent molecular weight as cytochrome c oxidase subunit I, these studies do not allow for an analysis of subunit I. From Fig. 5A (lane 2) it is clear that anti-subunit VIII immunoprecipitates subunits II, III, Va, VI, VII ϩ VIIa, and VIII. In contrast, anti-Pet100C does not specifically immunoprecipitate any cytochrome c oxidase subunits from this region of the gradient (Fig. 5A, lane 3). Although trace amounts of some cross-reactive bands are seen in the anti-Pet100C immunoprecipitate they probably result from nonspecific precipitation because the levels of these bands are essentially identical to levels seen with preimmune serum (Fig.  5A, lane 4). These findings make it unlikely that the Pet100p, which sediments between 18 and 22% sucrose, is associated with the cytochrome c oxidase that sediments in the same region of the gradient. It is more likely that the co-sedimentation of holocytochrome c oxidase and Pet100p in the gradient is coincidental.
To further address whether Pet100p physically interacts with holocytochrome c oxidase we examined cytochrome c oxidase preparations in various stages of purification for the presence of Pet100p. From Considered together, the above findings indicate that Pet100p does not form a stable complex with holocytochrome c oxidase, and suggest that its co-sedimentation with holocytochrome c oxidase is fortuitous. They also indicate that although Pet100p binds to a subassembly composed of subunits VII, VIIa, and VIII it does not bind to the same subunits once they have been assembled with the holoenzyme.
A pet100 Null Mutant Lacks Assembled Cytochrome c Oxidase and Accumulates Cytochrome c Oxidase Subassemblies-To ask how Pet100p affects assembly of cytochrome c oxidase we analyzed mitochondrial lysates from JM43GD100. This strain lacks cytochrome c oxidase activity (38); consequently there is no peak of cytochrome c oxidase activity in the sucrose gradient. Moreover, there is no concentration of subunit polypeptides in the region of the gradient (i.e. between 18 and 22% sucrose) where the holoenzyme sediments (Fig. 6A). This finding is important because it rules out the existence of an assembled but inactive cytochrome c oxidase. It also clearly establishes that Pet100p is required for the assembly of holocytochrome c oxidase and not for "activation" after it is assembled. The distribution of subunit polypeptides in a sucrose gradient from JM43GD100 differs from that seen with JM43 ( Fig. 6) in at least three ways. First, some subunits (e.g. III) are broadly distributed throughout the gradient (Fig. 6A). Second, some subunits are present in more than one region of the gradient. For example, subunit Va sediments between sucrose concentrations of 6 and 12% and 39 and 48%, and subunit VIIa sediments between 5 and 8%, 17 and 21%, and 39 and 49% sucrose. Third, from comparing sucrose gradients from JM43 (Fig. 2B) and JM43GD100 (Fig. 6A) it is clear that the levels of subunits VII, VIIa, and VIII in the region of the gradient where Complex A sediments are increased in JM43GD100 compared with the levels in JM43.
Immunoprecipitation experiments with anti-subunit VII (Fig. 6B, lane 2), anti-subunit VIIa (Fig. 6B, lane 3), or antisubunit VIII (Fig. 6B, lane 4) reveal that subunits VII, VIIa, and VIII that sediment between 39 and 49% co-immunoprecipitate, indicating that they are complexed with one another. In  Fig. 3B (lane 2), or pooled fractions 19 -24 of the experiment shown in Fig. 6A were incubated with preimmune serum, and centrifuged to collect a precipitate. The precipitate was subjected to SDS-PAGE and immunoblotted with antisera that recognized cytochrome c oxidase subunits IV, Va, VI, VII, VIIa, and VIII. contrast, preimmune serum does not precipitate any subunits from this region of the gradient (Fig. 4C, lane 3). Because this subassembly of subunits VII, VIIA, and VIII lacks Pet100p and is distributed over a range of sucrose concentrations in a sucrose gradient that is slightly broader than that for Complex A we designate it as Complex AЈ. By using gel quantitation software and different exposures of gels like those shown in Figs. 2B and 6A we have determined that the levels of subunits VII ϩ VIIa, and VIII that sediment between 39 and 49% sucrose in JM43GD100 are 5-fold higher than they are in JM43 (Table I). The other subunits that sediment between 39 and 49% sucrose do not co-precipitate with subunits VII, VIIa, or VIII (data not shown).
The presence of subunits IV, Va, and VI in gradient fractions between 6 and 12% sucrose was of interest because a cytochrome c oxidase subassembly containing subunits COX4 and COX5a, the human homologues of yeast cytochrome c oxidase subunits Va and VI, respectively (13), has been reported to accumulate in human fibroblasts carrying mutations that diminish cytochrome c oxidase assembly (52). To determine whether these subunits are in a complex we subjected pooled gradient fractions to immunoprecipitation with antisera specific for subunits IV or VI, followed by immunoblotting with antisera to holocytochrome c oxidase. The subunit specificities of the anti-IV and anti-VI sera are shown in Fig. 7A. Anti-VI serum precipitates nearly equal amounts of subunits VI and Va (Fig. 7B, lane 3), whereas anti-IV immunoprecipitates subunit IV and only a trace of subunit Va (Fig. 7B, lane 2). None of the other subunits (i.e. I, II, or III) that sediment between 8 and 12% sucrose are co-precipitated by either antiserum. These findings suggest that subunits Va and VI, but not IV, form a subassembly, designated here as complex B, in a pet100 null mutant.
To further explore the possibility that subunits V and VI form a complex with one another and that subunit IV is not part of this complex, we analyzed mitochondrial extracts from strains JM43GD4, a cox4 null mutant, and JM43GD6, a cox6 null mutant (Fig. 8A). Both strains lack cytochromes aa 3 and cytochrome c oxidase activity (44,53). Mitochondria from JM43GD4 lack subunit IV but retain the other nuclear-coded subunits of cytochrome c oxidase (Fig. 8A, lane 3). In contrast, mitochondria from JM43GD6 lack both subunit VI, the product of the deleted COX6 gene, as well as subunit Va (Fig. 8A, lane  2). These findings are consistent with the conclusion that subunit VI, but not subunit IV, is required for the stability of subunit Va (50). They also indicate that subunit IV levels are   6. Complex A accumulates in a pet100 null mutant. Panel A, Western blot of lysed JM43GD100 mitochondria separated on a sucrose gradient. Mitochondria (1 mg of protein) were lysed and sedimented on a sucrose gradient as described in the legend to Fig. 3B. Aliquots of each fraction were concentrated by acetone precipitation and subjected to SDS-PAGE on a 16% polyacrylamide gel containing 10% glycerol and 3.6 M urea. Gels were then transferred to nitrocellulose, blotted with an antiserum that recognizes all cytochrome c oxidase subunits, and detected by horseradish peroxidase-linked secondary antibodies. The asterisk denotes a contaminant antigen that is not a subunit of cytochrome c oxidase. The not affected by the absence of subunits Va and VI. Sucrose gradient analysis of mitochondrial extracts from JM43GD6 reveals that there is no concentration of subunit polypeptides in the region of the gradient where the holoenzyme sediments (i.e. between 18 and 22% sucrose) and that subunit IV sediments between 8 and 12% sucrose (Fig. 8B), exactly as it does in JM43, indicating that its position in the gradient is not affected by the presence of either subunits Va or VI. DISCUSSION In our continuing effort to understand nuclear-mitochondrial cross-talk we have been examining the role of the nuclearcoded protein Pet100p in the biogenesis of yeast cytochrome c oxidase. The results reported here provide important new insight concerning both the assembly of yeast cytochrome c oxidase and the function of Pet100p. First, they clearly demonstrate that Pet100p acts on holocytochrome c oxidase assembly per se. Second, they identify subassemblies, composed of some of the nuclear-coded subunits of holocytochrome c oxidase. And third, they provide evidence that Pet100p interacts with a subassembly and suggest a role for Pet100p in a holoenzyme assembly pathway. To our knowledge, this is the first time a role has been assigned to a cytochrome c oxidase assembly facilitator that does not function in the synthesis of heme A or the recruitment and/or insertion of copper into the holoenzyme.
Cytochrome c Oxidase Subassemblies-We have been able to detect three different subassemblies that contain subunits of cytochrome c oxidase. One of these subassemblies (Complex A) contains Pet100p, whereas the other two complexes (Complex AЈ and B) do not. These subcomplexes are likely biosynthetic intermediates in an assembly pathway because Complex A is absent in a pet100 null mutant, whereas Complexes AЈ and B accumulate at elevated levels, and because Pet100p is a component of Complex A.
The identification of a subassembly, Complex B, containing yeast subunits Va and VI is interesting in view of the crystal structure of bovine cytochrome c oxidase (7,8), and the observation that yeast subunit VI is required for stability of subunit Va (Ref. 50 and this study). The crystal structure shows that the bovine subunits (bovine IV and Va) with homology to yeast subunits Va and VI (13) are adjacent to one another. The bovine homolog to yeast subunit VI is on the matrix side of the inner membrane and binds to the hydrophilic amino-terminal domain of the bovine homolog of yeast subunit Va, which is anchored in the membrane by a trans-membrane ␣-helix. The finding that subunit VI protects its neighbor, subunit Va, from proteolytic degradation, together with our observation that they form a subassembly that exists in a pet100 null mutant suggests that these two subunits form a close and stable association. It is also interesting that a similar subassembly has been reported recently to accumulate in human fibroblasts carrying certain mutants that affect cytochrome c oxidase assembly (52). A global analysis of the yeast proteome using TAP-tagged proteins has identified a protein complex containing subunits IV, Va, VI, and VIIa (54). It is difficult to evaluate this finding, however, because the same study did not reveal holocytochrome c oxidase as a complex of interacting proteins. Moreover, we have not found a complex composed of subunits IV, Va, VI, and VIIa in our studies. Although it has also been proposed that subunit IV interacts with subunit I to form an early intermediate in an assembly pathway of mammalian cytochrome c oxidase (55), we find no evidence that these subunits from yeast are co-immunoprecipitable in any form other than when assembled in the holoenzyme. We also find no evidence that a complex of subunits I and IV co-sediment in sucrose gradients of wild type, pet100 null, or cox6 null mutant cells.
Our finding, from sucrose gradient analysis, that unassembled subunits and subassemblies exist in steady stategrown yeast cells is novel but not surprising. Indeed, several previous studies have hinted at the presence of pools of unassembled nuclear-coded cytochrome c oxidase subunits in both S. cerevisiae and Neurospora crassa mitochondria. For example, a portion of subunits I, II, and III that are synthesized in isolated yeast mitochondria assemble with nuclear-coded subunits into a holoenzyme (56,57). Those subunits that do not assemble are degraded rapidly (58). The degradation of unassembled subunits I, II, and III in isolated mitochondria requires ATP (58,59) and is probably brought about by the YTA10-12 complex (24). Because the isolated mitochondria used in these experiments were not importing the nuclearcoded subunits of cytochrome c oxidase, these findings imply that pools of these subunits are available within mitochondria for assembly into holocytochrome c oxidase. In vivo pulse labeling studies with N. crassa have also suggested pools of unassembled subunits (60 -62). Until now, the concept of unassembled pools of nuclear-coded cytochrome c oxidase subunits has presented an apparent paradox. Why are unas- . SDS-solubilized mitochondria (10 g) from each strain were electrophoresed on SDS-PAGE gels and blotted with an antiserum against holo-cytochrome c oxidase, and incubated with horseradish peroxidase-linked secondary antibodies. Panel B, sucrose gradient analysis of lysed mitochondria from JM43GD4 and JM43GD6. Mitochondria (5 g of protein) from JM43GD6 were lysed with Triton X-100 and sedimented on a linear 5-50% sucrose gradient as described in the legend to Fig. 1. Samples of every gradient fraction were taken to determine sucrose concentration, then concentrated by acetone precipitation, and subjected to SDS-PAGE on a 16% polyacrylamide gel containing 10% glycerol and 3.6 M urea. Gels were blotted to nitrocellulose, and probed with an antiserum that recognizes all cytochrome c oxidase subunits. Sucrose concentrations are indicated above the fraction numbers. The lane at the left (M) was loaded with 10 g of JM43 mitochondrial protein.
sembled nuclear-coded subunits stable when unassembled mitochondrially encoded subunits are not? The finding that some nuclear-coded subunits are in subassemblies may provide the answer. Perhaps these subunits are protected from degradation when assembled into these complexes. Conversely, perhaps the mitochondrially encoded subunits either do not form complexes or form unstable complexes that are subject to degradation unless they are assembled with the nuclear-coded subunits. This is supported by the finding that null mutants that carry gene disruptions in the structural genes for subunits IV and VI have reduced levels of subunits I, II, and III (11) and that the rates of subunit I degradation are enhanced in a null mutant that carries a disruption in the gene for subunit VII (63).
Pet100p as a Molecular Chaperone-Pet100p fits the original definition of a molecular chaperone as a protein that mediates the correct assembly of another protein but is not itself a component of the final functional structure (64). However, one characteristic of Pet100p that distinguishes it from other known mitochondrial chaperones (65) is its degree of specificity. It is required for the assembly of cytochrome c oxidase but not the biogenesis of other mitochondrial cytochromes (38). As yet, we do not know whether Pet100p facilitates the assembly of other (non-cytochrome) mitochondrial proteins. Also unclear is whether Pet100p acts alone or in a complex with other proteins. In this context, it is important to note that Pet100p has a bimodal distribution on a sucrose gradient and that most of it is not in Complex A. Preliminary analysis of the major population of Pet100p by fast protein liquid chromatography has revealed two subpopulations: one that has an apparent molecular weight of 425,000 and another with an apparent molecular weight of 255,000. 2 Insofar as the predicted molecular weight of monomeric Pet100p is 13,298 it is clear that Pet100p that sediments between 18 and 22% sucrose is present in high molecular weight complexes. Neither of these complexes elutes from the fast protein liquid chromatography column with cytochrome c oxidase, which migrates with an apparent molecular weight of ϳ380,000. It is not yet known if these complexes are homo-oligomers composed solely of Pet100p or if they are hetero-oligomers containing Pet100p and other polypeptides components.
Role of Pet100p in Cytochrome c Oxidase Assembly-Previously, we have shown that Pet100p is required for cytochrome c oxidase activity and were able to rule out a role in cytochrome c oxidase subunit synthesis, processing, or targeting to the mitochondrion (38). We were also able to show that a pet100 null mutant lacked a heme aa 3 spectral signature but could rule out a role for Pet100p in heme A biosynthesis. In the studies described here we have found that Pet100p is also not required for the localization of subunits to the inner mitochondrial membrane. Thus, all of the early steps in the expression pathway for cytochrome c oxidase subunits (i.e. transcription, translation, targeting, and sorting to the correct mitochondrial subcompartment) are maintained in the absence of Pet100p. An important finding of the present study is that subunits of cytochrome c oxidase are localized to the inner mitochondrial membrane but are not assembled into holocytochrome c oxidase in a pet100 null mutant and that Pet100p is an inner mitochondrial membrane protein. This clearly indicates that Pet100p is required for the assembly of cytochrome c oxidase subunits into holocytochrome c oxidase in the inner mitochondrial membrane. Because Complexes AЈ and B are present in a pet100 null mutant it is clear that Pet100p is not required for the formation of these subassemblies. And because these sub-assemblies are elevated in a pet100 null mutant, it is likely that Pet100p is required at a downstream step in their assembly into the holoenzyme. The finding that Pet100p associates with a subassembly (Complex AЈ) of subunits VII, VIIa, and VIII but does not associate with the same subunits when they are in the holoenzyme provides a useful clue for this downstream function. Indeed, this finding, together with those mentioned above, provide support for a role of Pet100p in bringing Complex AЈ together with other cytochrome c oxidase subunits and/or subassemblies during the assembly process.
Clearly, further work is required to elucidate how Pet100p functions. Indeed, our findings raise several questions. What are the subunit stoichiometries in Complexes A, AЈ, and B? Do all, or only some, of the cytochrome c oxidase subunits in complex A have direct physical contact with Pet100p? Does Pet100p associate in a complex with other proteins that facilitate cytochrome c oxidase assembly? What subunits and assemblies form with Complexes AЈ and B downstream in the assembly pathway? When, during assembly is Pet100p removed from Complex A? Current studies are underway to address these questions.