Characterization of ATP11 and detection of the encoded protein in mitochondria of Saccharomyces cerevisiae.

In Saccharomyces cerevisiae, expression of functional F1-ATPase requires two proteins encoded by the ATP11 and ATP12 genes. Mutations in either gene block some crucial late step in assembly of F1, causing the alpha and beta subunits to accumulate in mitochondria as inactive aggregates (Ackerman, S. H., and Tzagoloff, A. (1991) Proc. Natl. Acad. Sci. U.S.A. 87, 4986-4990). In the present study we have cloned and determined the sequence of ATP11. The encoded product is protein of 37 kDa with no obvious homology to any known protein. In vitro import assays of ATP11 precursor and immunochemical evidence indicate that the protein is located in mitochondria. A fusion was made between ATP11 and a short sequence coding for 78 amino acids with the biotination signal of bacterial transcarboxylase. The protein expressed from this construct complements atp11 mutants, indicating that the addition of the extra 78 amino acids at the carboxyl terminus of the ATP11 protein does not compromise its function. The hybrid protein is detected in mitochondria with antibodies and with peroxidase-conjugated avidin. Biotinated ATP11 protein can be partially purified by affinity chromatography on monomeric or tetrameric avidin coupled to Sepharose. A fraction eluted from the avidin column and enriched for the biotinated ATP11 protein also contains the alpha and beta subunits of F1-ATPase.

In ~arccharomyces cerevisiae, expression of functional F1-ATPase requires two proteins encoded by the ATP11 and ATP12 genes. Mutations in either gene block some crucial late step in assembly of FI, causing the a and B subunits to accumulate in mitochondria as inactive aggregates ( In vitro import assays of ATPll precursor and immunochemical evidence indicate that the protein is located in mitochondria. A fusion was made between ATPll and a short sequence coding for 78 amino acids with the biotination signal of bacterial transcarboxylase. The protein expressed from this construct complements a t p l l mutants, indicating that the addition of the extra 78 amino acids at the carboxyl terminus of the ATPll protein does not compromise its function. The hybrid protein is detected in mitochondria with antibodies and with peroxidase-conjugated avidin. Biotinated ATPl 1 protein can be partially purified by affinity chromatography on monomeric or tetrameric avidin coupled to Sepharose. A fraction eluted from the avidin column and enriched for the biotinated ATPll protein also contains the a and B subunits of F1-ATPase. F1 is an important catalytic component of mitochondrial, chloroplast, and bacterial energy-transducing ATPases (1,Z). F1-ATPases are hetero-oligomers composed of five different subunit polypeptides. The two major subunits, referred to as a and @, are each present in three copies per oligomer (3). The other subunits are present in single copy. At present very little is known about the requirements and pathway for assembling this rather complex structure. Is there an obligatory order in which the subunits interact with one another? Is the final quaternary structure achieved through the intervention of other nonstructural proteins, or is it a spontaneous process guided solely by protein-protein recognition determinants on the surfaces of the folded subunits? These and other related questions remain unanswered.
*This research was supported by Research Grant HL22174 and National Research Service Award GM12435 (to S. H. A.) from the National Institutes of Health. 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 U.S.C. Section-1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper hos been submitted to the GenBankTM/EMBL Data Bank with accession numbeds) M87006.
We have recently reported two genetically distinct groups of Saccharomyces cerevkiae mutants whose inability to grow on nonfermentable carbon sources was ascribed to a block in F1 assembly (4). These mutants define two nuclear genes, ATPll and ATP12, with important roles in the expression of a functional mitochondrial ATPase. Even though mitochondria of atpll and atp12 mutants are grossly deficient in ATPase activity, they have nearly normal concentrations of mature size CY and @ subunits. Instead of being part of the F1-Fo complex, however, the F1 subunits are present as aggregated proteins readily separable from the membrane fraction. Based on their phenotype, atpll and atp12 mutants were proposed to be arrested in ATPase assembly at some step after import and cleavage of the F, subunit precursors.
To facilitate further studies of the ATPll and ATP12 proteins and to learn more about their functions, we have characterized their respective genes. In an earlier paper we reported the sequence of ATP12 and some properties of the encoded product (5). In the present studies we demonstrate that ATP11, like ATP12, is synthesized as a larger precursor and is cleaved during its import into mitochondria. The cloned ATPll gene has enabled us to construct two different gene fusions. One fusion was used to prepare antibodies against the ATP11 protein. The second fusion has permitted the protein to be tagged in vivo with biotin (6). The biotinated ATPll protein is functional based on its ability to restore respiratory competence to atpll mutants. The possible association of the ATPll protein with the CY and / 3 subunits of F1 has been assessed by partial purification of the biotinated derivative on avidin affinity columns.
Transformation of Yeast-The atpll mutant aC15/U3 (cu,ura3-1,atpll-2) was grown in 50 ml of YPGal to a density of lo7 cells/ml. Approximately 5 X lo8 cells were transformed with a yeast genomic library (5 pg of DNA) by the procedure of Beggs (10  Preparation of Yeast Mitochondria-Yeast was grown aerobically in YPGal media at 30 "C to early stationary phase. The procedure of Faye et al. (12) was used to prepare mitochondria except that Zymolyase 20,000 (Miles Corp.) instead of Glusulase was used for the conversion of cells to spheroplasts. Phenylmethylsulfonyl fluoride was added at a final concentration of 10 pg/ml during the homogenization step to minimize proteolysis.
Construction of Plasmid for Expression of ATPll mRNA-The 1.6kb BamHI-KpnI fragment with the entire ATPl 1 reading frame was transferred to pGEM3 (13) with an orientation such that transcription is initiated from the SP6 promoter of the vector. This vector was used as the template for in vitro transcription of the cloned ATPll gene as described previously (5). The transcript was translated with a nuclease-treated rabbit reticulocyte lysate, and import of the product into isolated yeast mitochondria was measured as described previously (14).
Preparation of Antibodies to the ATPll Gene Product-A 1.1-kb Sau3A fragment containing the carboxyl two-thirds of the ATPll reading frame, and 3'-flanking region was ligated, in frame, to the amino-terminal half of the E. coli trpE gene in the expression vector pATH2 (15). Two independent transformants, verified to harbor the plasmid with the hybrid gene, were used to express the fusion protein (15). After cell lysis most of the overexpressed protein was found in the supernatant fraction. It was recovered in the precipitate obtained at 0-30% saturation in ammonium sulfate and was further fractionated on DE52 cellulose (Whatman) by stepwise elution with 0.1, 0.2, 0.4, and 0.8 M NaCl in the presence of 20 mM Tris-HC1, pH 7.5. The protein eluted with 0.4 M NaCl was applied to a hydroxylapatite column equilibrated with 5 mM NaPO,, pH 7.5. This column was developed with 10,20, and 50 mM NaPO,, pH 7.5. The protein eluted a t 50 mM NaP04 was concentrated by precipitation at 50% saturation in ammonium sulfate. The precipitate was dissolved in a buffer containing 10 mM Tris-HC1, pH 7.5, 0.9% NaCI, and 0.1% sodium dodecyl sulfate and dialyzed against the same buffer. Rabbits were immunized as described previously (16).
Construction of ATPl 1 -Transcarboxylase Fusion Gene-To construct an ATP2 1 gene with a carboxyl-terminal hiotination sequence, the native gene was modified by eliminating the termination codon and replacing it with an XbaI site. A 24-nucleotide-long oligomer, 3'-TACCTTTTAAGATCTTCTTAATGA-5', was used together with the M13 pentadecamer universal primer to synthesize the modified ATP11 gene. The synthetic primer with the XbaI site shown above is complementary to nucleotides +946 to +969 of the sequence reported in Fig. 3, except for three base changes needed to form the new XbaI site. A plasmid containing the entire ATP11 reading frame plus 5'and 3'-flanking sequences was used as a template for the polymerase chain reaction. The product obtained after 25 cycles was digested with a combination of BamHI and XbaI. The digested fragment was purified by electrophoresis on 1% agarose and ligated to the BamHI and XbaI sites in the multiple cloning region of the yeast/E. coli shuttle vector YEp352 (17). After linearization with XbaI and PstI, the plasmid was ligated to a 270-base pair XbaI-BamHI fragment coding for the biotination sequence of Propionibacterium shermanii transcarboxylase. This fragment, obtained from the plasmid pCY66 and coding for the carboxyl-terminal 78 amino acids of the transcarboxylase, has been shown to be a substrate for in uivo biotination in yeast (6). The ligation of the bacterial sequence resulted in an inframe fusion of the entire ATP11 gene to the biotination signal sequence (at the XbaI site). The ligation of the downstream BamHI to the PstI of the vector caused both sites to be destroyed.
Miscellaneous Procedures-Standard procedures were employed for restriction endonuclease analysis of DNA, purification and ligation of DNA fragments, transformations of and recovery of plasmid DNA from E. coli, and nick translation of DNA (18). The conditions of Myers et al. (19) were used for Southern hybridizations. DNA was sequenced by method of Maxam and Gilbert (20). For Western blot analysis, proteins were separated on 12% polyacrylamide gels run in the electrophoresis system of Laemmli (21) with the separation buffer adjusted to pH 8. The running buffer contained 0.05 M Tris, 0.38 M glycine, and 0.1% SDS. After transfer to nitrocellulose the Western blots were reacted with antibodies against various mitochondrial proteins. In most analyses the y-globulin fraction was purified by chromatography of antisera on DE52 cellulose. The blot was then reacted with 1251-protein A and was washed by the protocol of Schmidt et al. (22). Protein concentrations were determined by the method of Lowry et al. (23).

RESULTS
Properties of atpl I Mutants-The phenotype of atpll and atpl2 strains is identical and has already been reported (4). Briefly, the mutants are deficient in F,-ATPase activity both in the mitochondrial and postmitochondrial supernatant fractions. The absence of enzyme activity correlates with a defect in assembly of the a and / 3 subunits of F1 into the normal oligomer. Mature size subunits are detected in mitochondria, but instead of being associated with the hydrophobic Fo unit of the inner membrane they are present as large aggregates that can be quantitatively separated from the membranes by centrifugation of sonically disrupted mitochondria on isopycnic sucrose gradients (4).
Mitochondrial ATPase mutants can be distinguished by their genetic properties depending on whether the mutations affect the Fo or F, component. As a rule mutations in genes coding for Fo subunits cause a marked instability in mitochondrial DNA as a result of which such mutants convert to secondary p-and po derivatives at a high frequency (24-26). Mutations in F1 also express the absence of ATPase but have no significant effect on the stability of mitochondrial DNA (27,28). It is of interest that both atpll and atpI2 strains have genetic properties consistent with lesions in F,. Like most other pet mutants (nuclear mutants of yeast defective in mitochondrial respiration) they produce 1-5% p -derivatives.
Cloning and Sequencing of ATPl1-The ATPl 1 gene was cloned by complementation of aC15/U3 with a yeast genomic library. This mutant was transformed with a recombinant plasmid library consisting of 5-15 kb of partial Sau3A fragments of yeast genomic DNA ligated to the BamHI site of YEp24 (11). A single respiratory-competent and uracil-independent clone (aC15/U3/T1) was obtained by transformation of approximately 5 X 10' cells with 5 pg of the plasmid library. The acquired uracil prototrophy and respiratory competence of aC15/U3/Tl cosegregated indicating both phenotypes to be dependent on the presence of an autonomously replicating plasmid. Plasmid pG13/T1, isolated from clone aC15/U3/Tl, was ascertained by restriction mapping to have a nuclear DNA insert of 10.5 kb (Fig. 1).
The gene was localized by transferring different regions of the pG13/T1 insert to the E. colilyeast shuttle vector YEp352 (17), and the resultant plasmids were tested for their ability to complement an atpll mutant. The results of the transformations with these constructs indicated the gene to be internal to a 3-kb BamHI fragment defined by pG13/ST3 (Fig. 1).
The complementing region was further narrowed down to a 1.6-kb BamHI-KpnI fragment (pG13/ST4) by removal of 1.4 kb of DNA from pG13/ST3. This plasmid restored oligomycin-sensitive ATPase in mitochondria and conferred growth on the nonfermentable substrate glycerol. The 1.6-kb BamHI-KpnI fragment cloned in pG13/ST4 was sequenced by the method of Maxam and Gilbert (20) by 5' end labeling of the restriction sites shown in Fig. 2. All the sites were crossed from neighboring sites, and both strands were sequenced. The sequence of this region disclosed only a single full-length reading frame that was identified as ATPll based on the subcloning results and the properties of a mutant with a partial deletion of the gene (see in situ disruption of ATPl 1 below). The gene starts at an ATG at nucleotide 1 and ends with an ochre stop at nucleotide 955 of the sequence reported in Fig. 3. The primary translation product consists of 318 amino acid residues with a molecular weight of 36,874. The amino-terminal 40 residues are rich in basic and hydroxylated amino acids, suggestive of a mitochondrial targeting sequence (29). This is consistent with evidence presented later showing that the protein is processed during import into mitochondria. A search of the deduced protein sequence encoded by ATPll against GenBank and Swiss-Prot failed to detect any entries with significant primary structure similarity.  The sequence of the BamHI-KpnI fragment has enabled us to map ATP1 I relative to two other genes. Analysis of the sequence revealed a second open reading frame initiated from an ATG 146 nucleotide downstream of ATP11. Only the region of the gene coding for the amino-terminal 92 amino acids is present in the pG13/ST4 insert. This partial sequence showed significant homology to the bifunctional E. coli chorismate mutase/dehydratase encoded by pheA (30) (Fig. 4). Deletion of the sequence coding for the amino-terminal part of this protein confers a requirement of phenylalanine but not tyrosine or tryptophan for growth. The primary sequence homology and the phenylalanine auxotrophy suggest that this gene corresponds to yeast prephenate dehydratase encoded by PHA2 (31, 32). As discussed later this assignment is substantiated by genetic complementation tests. The second partial sequence included in the RamHI-KpnI fragment is the first six codons of the DAD2 gene coding for a protein required for induction of the allantoin degradation pathway (33). DAD2 is transcribed from the opposite strand and begins 166 nucleotides upstream of the ATPll ATG. Even though the sequence of the entire 1.6-kb BamHI-KpnI fragment has been determined in connection with the characterization of DAL82 (33), neither ATPll nor the PHA2 gene was reported because of several missing nucleotides in the reported sequence.
In Situ Disruption of ATPIl---To confirm that the restoration of respiratory function in atpll mutants by pG13/ST4 is caused by complementation rather than extragenic suppression, the one-step gene substitution procedure (34) was used to create a mutant with a deletion in the gene. The chromosomal copy of ATPll in the respiratory competent haploid strain W303-1A was replaced with a mutant allele deleted for all but the first 17 codons of the gene. The construction of the deletion allele is illustrated in Fig. 5. A 3.2-kb EcoRI fragment containing the entire reading frame of ATPl 1 and 5'-and 3"flanking sequences was isolated from pG13/T1 and transferred to YEp352E (this plasmid lacks the multiple cloning sequence of YEp352 and has in its place a single EcoRI site). Digesbion of the resultant plasmid with PuuII removed the ATPll reading frame starting from codon 18 of the gene and an additional 6.50 base pairs of 3"flanking sequence. The deletion included the amino-terminal coding region of the downstream open reading frame. The gapped plasmid was ligated to a blunt-ended 1.8-kb fragment of DNA containing the yeast HIS3 gene (Fig. 5). The deleted allele, Aatpll::HXS3, was recovered as a linear 3.1-kb EcoRI fragment and was used to transform W303-1A. Selection of transformants on minimal glucose medium supplemented with all the auxotrophic requirements of W303-1A except histidine yielded several respiratory deficient His+ clones. The His+ phenotype did not segregate, indicating stable integra-tion of HIS3 into chromosomal DNA. The respiratory deficiency of one such clone (W303AATPll) was checked by crosses to be complemented by a pa but not by an atpll mutant.
The failure of W303AATPll to be complemented by the atpll tester was consistent with the substitution of the AatplI::HZS3 allele for the wild-type gene. This was supported by the results of the genomic Southern analysis shown in largest fragments correspond to the two cross-hydridizing bands seen in W303-1A. The other two fragments of 1.6 and 0.7 kb have sizes consistent with the presence of the disrupted allele based on the known locations of BgZII sites in the HIS3bearing insert (Fig. 5). T o confirm that the respiratory defect of W303AATPll is a consequence of faulty F1 assembly we also measured the ATPase activity of isolated mitochondria and analyzed the physical properties of a and p subunits of F1 by isopycnic centrifugation as described previously (4). In all respects W303AATPll is identical to the original atpll mutant used to characterize the ATPase lesion (data not shown). The genotype and biochemical phenotype of W303AATPll are, therefore, consistent with the correct identification of the gene.
The deletion in W303AATPll encompassed both the A T P l l reading frame and a significant part of the downstream gene whose product was found to be homologous to E. coli chorismate mutase/prephenate dehydratase (30). W303AATPll grows very slowly on media supplemented with the usual auxotrophic requirements of W303-1A, indicating that the partial removal of the downstream gene elicits a new growth requirement. Growth of the transformant on minimal glucose medium was restored by supplementing the normal requirements of W303-1A with phenylalanine. The phenylalanine requirement was not complemented by XS1007-2A, a haploid strain with a mutation in PHA2, a gene identified previously to code for yeast prephenate dehydratase (31).
I n Vitro Import and Processing of the A T P l l Precursor Protein-The mitochondrial location of the ATPll protein was assessed by in vitro import assays and by immunochem-ical assays. T o study the transport of A T P l l protein into mitochondria, the gene was transferred to the pGEM3 vector in an orientation allowing transcription from the SP6 promoter. This plasmid, as well as a previously described construct with the Neurospora crassa F1 @ subunit gene (35), served as a template for synthesis of the A T P l 1 and p subunit mRNAs with SP6 polymerase. The ATP11 and p subunit precursors were synthesized by programming a reticulocytefree translation system with the two mRNAs in the presence of [35S]methionine. The N. crassa p subunit was used as an internal control, having been previously shown to be an efficient substrate for import into yeast mitochondria (35). The transfer and concomitant processing of both radioactive precursors into a protease-protected compartment of mitochondria are confirmed by the results in Fig. 6 antibody prepared against a protein expressed from a hybrid trpE-ATP11 gene. This protein consisted of the carboxyl twothirds of the ATPll sequence starting from Asp112 fused to the amino-terminal half of component I of anthranilate synthetase (see "Materials and Methods"). As shown in Fig. 7, the antibody detects a 32-kDa protein in wild-type mitochondria. The identity of the 32-kDa protein as the ATPl 1 product is supported by its absence in W303AATPll containing the Aatpll::HIS3 allele and its presence in substantially higher concentrations in an atpll mutant transformed with ATPl 1 on a multicopy plasmid.
As an alternative approach, a sequence coding for a short polypeptide with the biotination site of bacterial transcarboxylase, was fused in frame to the last codon of ATP11. The hybrid gene in the multicopy shuttle vector YEp352 was able to complement the respiratory deficiency of C15/U1, indicating that the presence of the biotination signal at the carboxyl end of the ATP11 protein does not compromise its function. Mitochondria prepared from wild-type yeast and from a transformant expressing the biotinated ATPll protein were fractionated to yield a soluble and a membrane fraction. Mitochondria and the soluble and insoluble protein fractions were separated by SDS-polyacrylamide gel electrophoresis and were transferred to nitrocellulose. Similar fractions were obtained from a strain with the disrupted A T P l l allele and from a transformant harboring normal ATPl 1 on a multicopy plasmid. The Western blot was first developed with peroxidase-conjugated avidin to visualize proteins with bound biotin and was then reacted with the antibody against ATPll protein. The results of this experiment show that the antibody detects a protein some 8 kDa larger than wild-type ATPll protein in mitochondria obtained from the transformant harboring the fusion gene (Fig. 8). This protein (BT-ATP11) is also detected with the peroxidase-conjugated probe, indicating the presence of covalently bound biotin. The apparent molecular weight of BT-ATP11 based on its properties on SDS-gel electrophoresis is 40,000 and is consistent with the expected size increase of ATP11 protein because of the extra 78 amino acids contributed by the biotination signal. Approximately 50% of the overproduced biotinated ATPll protein is recovered in the soluble protein fraction after disruption of mitochondria by sonic irradiation (Fig. 8). A similar distribution was observed in a transformant harboring the wildtype A T P l l gene in a multicopy plasmid. In wild-type yeast, however, with a single chromosomal copy of ATPl1, almost all of the ATPll protein is released into the soluble fraction by sonic disruption of mitochondria. These results suggest that in overexpression strains approximately half of the ATPll protein detected in mitochondria is probably present in an aggregated form because of improper folding after transport.
A biotinated protein of the same size and antigenic properties as BT-ATP11 was detected in the crude postmitochondrial supernatant fraction of transformants harboring the A T P l l fusion gene (data not shown). Because the size of BT-ATPll in the postmitochondrial supernatant is identical to that found in mitochondria it probably corresponds to protein that leaked out of mitochondria during their isolation.

Purification of the Biotinated A T P l l Protein by Chromatography of an Avidin Affinity Column-
The observation that biotinated ATP11 protein can be recovered in the soluble protein fraction of mitochondria made it possible to to use an avidin affinity column for its partial purification. Both high affinity columns with tetrameric avidin coupled to Sepharose (37) and lower affinity columns containing monomeric avidin (38) have been used to purify biotin-containing proteins from relatively crude protein mixtures. The monomeric column has the advantage that biotin-containing proteins can be displaced under nondenaturing conditions with biotin (38). The release of proteins fixed onto tetrameric avidin requires extreme conditions of pH or the use of denaturing detergents.
The fractionation of the biotinated protein on both types of avidin columns is shown in Fig. 9. Biotinated ATPll protein was assayed by Western analysis of the fractions using either antibody to ATP11 or peroxidase-conjugated avidin. Equivalent amounts of a mitochondrial extract from the transformant C15/Ul/ST16 (lane 1 ) were passed over a monomeric and a tetrameric avidin column. Even though less than 50% of the protein applied to the monomeric column was retained (lane 2), a substantial portion eluted in the biotin wash (lane 3 ) . The smaller fraction released in the SDS wash (lune 5 ) was probably complexed to some residual tetrameric avidin present in the column. A much higher proportion of biotinated A T P l l protein was adsorbed on the tetrameric column (lane 6). None of the complexed protein, however, was eluted with biotin (lune 7) although its recovery was quantitative in the hot SDS wash (lane 8). To obtain a better idea of the purification achieved on the monomeric column, the fraction eluted with biotin (lune 3) was separated by SDSgel electrophoresis and the proteins visualized by silver staining. The gel revealed the most prominent band detected by the stain to be protein whose migration was identical to BT-A T P l l detected with peroxidase-conjugated avidin (Fig. 10). This protein was absent in the comparable fraction obtained by fractionation of a mitochondrial extract from a wild-type yeast strain although the other bands of higher molecular weight were present. BT-ATP11 was also a prominent band in the hot SDS wash of the tetrameric column although the pattern of proteins was considerably more complex (data not shown).
Does the A T P l l Protein Interact with Fl Subunits andlor the ATP12 Protein?-The fact that atpll and atpl2 mutants exhibit identical phenotypes (4) suggests that the encoded products may have a related function and therefore might be part of a single complex. The existence of an ATP11-ATP12 complex in mitochondria was tested in two ways. In previous studies, the ATP12 protein was determined to have a native molecular weight approximately twice that of the monomer as estimated by SDS-gel electrophoresis (5). The discrepancy in size could indicate that the native ATP12 protein is a homodimer of a hetero-oligomer complexed with some other protein. Because A T P l l seemed a reasonable candidate, molecular weight determinations were done on the ATP12 protein in W303AATPl1, a strain unable to express ATP11. Within experimental error, no difference could be detected in the size of the native ATP12 protein in this mutant and in a wild-type strain. Avidin affinity chromatography also failed to provide evidence for the existence of an ATP11-ATP12 protein complex. Fractions obtained from the tetrameric avidin column were separated by SDS-gel electrophoresis and probed with antibody against the ATP12 protein and against the CY and p subunits of F1. As shown in Fig. 11, no ATP12 protein could be detected in the hot SDS wash containing almost all of the biotinated A T P l l protein applied to the column. Interestingly, a significant amount of both a and subunits was detected in this fraction. An identical fractionation of a mitochondrial extract from wild-type yeast is also shown in Fig. 11. Although the SDS eluate from this column also had signals corresponding the the F1 subunits they were significantly weaker. The results of the fractionations indicate that ATP11 protein is unlikely to be complexed to ATP12 but could have some affinity either for F1 subunits or for the oligomeric enzyme itself.

DISCUSSION
The ATPll and ATP12 genes of S. cereuisiae have been shown previously to be required for the assembly of the F1 subunits into a functional oligomer (4). The present studies were undertaken with several objectives in mind. The first was to characterize the A TPl1 gene and learn more about the properties of the protein. Second, it was important to establish whether ATPll protein is located in mitochondria. Finally, we explored a recently described method for in vivo tagging proteins with biotin for purposes of their detection and purification (6).
The ATPl 1 gene was cloned by transformation of an atpl I mutant from complementation group G13 with a plasmid library of yeast nuclear DNA. The gene codes for a product of 39 kDa with a hydrophilic amino acid composition suggestive of a water-soluble protein. The amino-terminal 30-40 residues have been confirmed by in vitro import assays to constitute a mitochondrial import signal. The primary translation product expressed from an RNA transcribed from ATPl 1 is translocated by an energy-dependent process into a protease-resistant compartment of yeast mitochondria. The mature protein is some 3-4 kDa smaller than the precursor. The transport-coupled processing of the ATPll precursor constitutes strong evidence for its mitochondrial localization. This was confirmed by immunological assays of fractionated yeast. An antibody against a hybrid protein consisting of ATPll fused to the amino-terminal half of the component I of anthranilate synthetase was found to react with a mitochondrial protein of 32 kDa present in wild-type yeast but not in a mutant with a deleted copy of A T P l l . The protein detected by the antibody was identical to the processed ATPll protein seen after its import into mitochondria.
The primary structure of the ATP1 I product, deduced from the sequence of the gene, does not provide any obvious clues concerning the role of this protein in F1 assembly. The ATPll protein is not homologous to any entries in either the Gen-Bank or EMBL databases; nor does it have a domain that is recognized by programs designed to search for protein sequence motifs indicative of enzyme functions or substrate binding properties. The elucidation of how the ATPll protein promotes F1 assembly will therefore require that it be available in a form sufficiently pure for functional studies. Our initial trials at purification have been hampered by the low abundance of the protein in mitochondria. As an alternative approach we constructed a fusion gene of ATPl 1 and a sequence coding for the carboxyl-terminal biotination domain of P. shermanii transcarboxylase (6). The hybrid protein expressed from this gene has a carboxyl-terminal extension of 78 residues with a lysine situated 34 amino acids from the new carboxyl terminus that serves as the biotin acceptor (39). The ability of this gene to complement the respiratory defect of an atpll mutant indicates that the presence of the bacterial sequence does not abolish the normal function of the ATPll protein. Furthermore, analysis of mitochondria from the transformant harboring a multicopy plasmid with the fusion gene disclosed the presence of a protein with the expected size which reacts with the anti-ATP11 antibody and with peroxidase-conjugated avidin. The detection of the hybrid protein with the avidin probe confirmed the in vivo biotination of the hybrid protein.
In addition to providing a convenient and sensitive means for assaying the ATP11 protein, the presence of covalently attached biotin makes it possible to use an avidin affinity column to remove most matrix proteins. Chromatography of a crude mitochondrial extract on a monoavidin column yields a fraction in which ATPll protein is the major stained band detected with silver stain. Because the protein is displaced from the avidin column with biotin it should retain its native conformation. A more quantitative recovery of the ATP11 protein is obtained after affinity chromatography on Sepharose coupled to native tetrameric avidin. However, elution of the protein from this column requires harsher conditions such as SDS because of the higher affinity of tetrameric avidin for biotin. The native avidin column can be used to test for possible association of other proteins with the biotinated component. The results of such analyses have failed to provide any evidence of an ATP11-ATP12 protein complex. They do, however, suggest that the ATPll protein may interact with the a and/or @ subunits of F1-ATPase.