Alpha subunit of mitochondrial F1-ATPase from the fission yeast. Deduced sequence of the wild type and identification of a mutation that alters apparent negative cooperativity.

The nuclear gene atp1 encoding the mitochondrial ATP synthase alpha subunit of the fission yeast Schizosaccharomyces pombe was sequenced. It contains a 1,608-base pair-long open reading frame interrupted by two introns of 175 and 269 base pairs, located near the 5'-end of the gene. The initiation site of transcription AAAC was located 60 nucleotides upstream of the translation initiation codon. The deduced polypeptide sequence contains a 27-amino acid residue presequence, presumably involved in mitochondrial targeting, preceding a mature protein of 509 amino acid residues. The atp1 alleles from mutant A2313 (Bouty, M., and Goffeau, A. (1982) Eur. J. Biochem. 125, 471-477) and its related phenotypic revertant R351 (Falson, P., Di Pietro, A., Darbouret, D., Jault, J. M., Gautheron, D. C., Boutry, M., and Goffeau, A. (1987) Biochem. Biophys. Res. Commun. 148, 1182-1188) were also cloned and sequenced. A single nonsense mutation CAA-TAA (Gln173-stop) in mutant A2313 became a missense mutation TAA-TTA (stop-Leucine) in revertant R351. Glutamine 173 is located in the first putative element of the nucleotide binding site. Its substitution by a leucine residue appears responsible for the lower enzyme affinity toward ADP and for the loss of cooperativity of F1-ATPase activity.

To whom correspondence should be addressed. Tel.: 32-10-473621; Fax: 32-10-473872. 1988). Among the sequences known for the a subunit, only one has been reported for a lower eukaryote, the budding yeast Saccharomyces cerevisiae (Takeda et al., 1986). The structural gene encoding the a subunit of the fission yeast Schizosaccharomyces pombe has been isolated (Boutry et al., 1984) and located on chromosome I (Vassarotti et al., 1984).
Beside the / 3 subunit, which plays a major role in catalysis and regulation of the ATP synthase (for a review, see Vignais and Lunardi, 1985;Wang, 1988;Futai et al., 1989), the a subunit seems to have an important role although more recently documented. Indeed, chemical modifications of the a subunit have given evidence for a structural (Nalin et al., 1985) or functional (Falson et al., 1986) asymmetry. Immunological approaches have concluded a regulatory role for this subunit (Moradi-Ameli et al., 1989). Nucleotides or analogs have been found to bind to one site on the a subunit (Dunn et al., 1980;Tagaya et al., 1988;Rao et al., 1988;Lee et al., 1990). Several bacterial mutants, characterized by a single residue substitution of the a subunit and showing either a low activity or a defective assembly, have been isolated (Maggio et al., 1988;Yohda et al., 1988;Soga et al., 1989;Pagan and Senior, 1990).
In eukaryotes, the genetic approach was successfully carried out with S. pombe since among several Fl-ATPase defective mutants a strain (A2313), with no immunodetectable a subunit, was isolated (Boutry and Goffeau, 1982). Moreover, a phenotypic revertant recovered the a subunit in the complex but conserved an altered ATPase activity (Falson et al., 1987). Biochemical studies of the purified or membrane-bound Fl moiety of the revertant enzyme revealed (i) a lack of apparent negative cooperativity correlated to a decreased affinity for ADP; (ii) no bicarbonate activation of ATPase activity and a lower azide inhibition; (iii) a lower amount of endogenous nucleotides than the wild-type enzyme; and (iv) a modified intrinsic fluorescence of F1-tryptophan residues (Falson et al., 1987;reviewed by Di Pietro et al., 1989).
In the present work, we have determined the nucleotide sequence of the wild-type, mutant, and revertant alleles of the atpl gene of the fission yeast. The CAA codon encoding glutamine 173 in the wild strain is replaced by a nonsense TAA codon in the mutant, resulting in the synthesis of a twothirds truncated protein. The phenotypic reversion is correlated to the replacement of TAA with TTA, establishing a leucine residue. The results show the important role of glutamine 173 in the nucleotide interaction with the a subunit and the role of this site in the enzyme cooperativity.
Enzymes and Reugents"T4 DNA ligase, DNA polymerase I (Klenow fragment), T4 polynucleotide kinase, nuclease S1, kilobase sequencing system, and restriction enzymes were from Bethesda Research Laboratories. Thermostable T q DNA polymerase was from Promega (The Netherlands). [3ZP]dCTP, 35S-dATP, T7 DNA polymerase, and a random priming kit were purchased from Amersham Corp. A Sequenase sequencing kit and plasmidpTZ18U were provided by the U. S. Biochemical Corp. Plasmid pBluescript SK' was purchased from Stratagene. Synthetic oligonucleotides were from Eurogentec (Belgium). Novozyme SP 234 was generously given by Novo (Denmark). Gene-Clean I1 was purchased from BIO 101, Inc.
Preparation of Nucleic Acids-Yeast genomic DNA was prepared according to a plublished procedure (Davis et al., 1980) scaled up' as follows. A 200-ml instead of a 5-ml culture was used. Cells were washed with cold distilled water, centrifuged, and suspended at 35 "C for 10 min in 20 ml of 1 M sorbitol, 25 mM EDTA, 50 mM dithiothreitol, pH 8.0 (NaOH). After centrifugation, the pellet was washed in 1 M sorbitol, spun again, and suspended in 10 ml of 1 M sorbitol, 0.1 M sodium citrate, pH 5.8 (HCI), 10 mM EDTA. Novozyme (Beach and Nurse, 1981) was added at a final concentration of 3 mg/ml, and the suspension was incubated at 35 "C for 1 h. After a 5-min centrifugation at 3,000 rpm (rotor JA20, Beckman), the spheroplast pellet was suspended in 5 ml of 50 mM Tris-C1, pH 7.4, 20 mM EDTA, and 0.5 ml of 10% sodium dodecyl sulfate was added. The suspension was incubated at 65 "C for 30 min. After the addition of 2 ml of 5 M potassium acetate, the suspension was incubated at 0 "C for 1 h. After centrifugation at 18,000 rpm (JA20) for 30 min, the clear supernatant was incubated at room temperature for 40 min with 2 pg of freshly boiled RNase. After the addition of an equal volume of isopropyl alcohol, the solution was centrifuged at 7,000 rpm (JA20) for 10 min. The pellet was dried under vacuum and then solubilized in 4 ml of 10 mM Tris-C1, pH 7.5, 1 mM EDTA. DNA was purified by cesium chloride centrifugation and then submitted to a phenol-chloroform extraction and ethanol precipitation as described by Sambrook et al. (1989). Total RNA from the wild-type S. pombe strain was a generous gift from E. Capieaux (this laboratory). Bacterial plasmids were prepared as described by Sambrook et al. (1989).
Isolation of atpl from Different Strains-Genomic libraries of BamHI digests of the wild-type and revertant R351 nuclear DNA were constructed in the plasmid pTZ18U. Cloning experiments were carried out as described by Sambrook et al. (1989). The probe was a 432-base pair ClaI fragment containing the NHn-terminal coding sequence of atpl (see Fig. l), labeled either by nick translation or random priming (Sambrook et al., 1989). atpl of the A2313 mutant was isolated directly from amplified genomic DNA as described below. In this case, the cesium chloride gradient step was omitted. Ends of the amplified fragment were filled in with the Klenow fragment, phosphorylated with T4 polynucleotide kinase (Sambrook et al., 1989), and inserted into the plasmid pBluescript SK' as described under "Results and Discussion." Amplification by the Polymerase Chain Reaction-Direct amplification of atpl was carried out as described by Mullis and Faloona (1987) and adapted to the thermostable Taq DNA polymerase provided by Promega. The buffer contained 10 mM Tris-C1, pH 9.0, 50 mM KCI, 1.5 mM MgCIz, 0.1% Triton x-100, 20 pg/ml gelatin, dNTPs (0.2 mM each), primer 785 (1 pM; TCGGCCTAATA-ACTGTAATTTGTC), and primer 1276 (1 pM; CCCACTCCTCCT-CTCAGTTTAACCG). About 50-200 ng of genomic DNA was diluted into 100 pl of the amplification buffer, denatured for 2 min at 93 "C, and chilled rapidly on ice. After the addition of 2.5 units of Taq DNA polymerase and a few drops of mineral oil, the tubes were submitted to 30 cycles of 2 min at 93 "C (denaturation), 2 min at 55 "C (hybridization), and 3 min at 72 "C (extension) in a homemade apparatus.
During the last cycle, the extension step was prolonged for 10 min. The amplified DNA was separated from the reaction buffer by electrophoresis on a 0.8% agarose gel and recovered with the Gene-Clean I1 kit.
DNA Sequence Determination-Overlapping clones of the wildtype atpl gene were obtained using the progressive deletion strategy described by Barnes et al. (1983). Synthetic oligonucleotides were used to sequence either the minus strand 785, 534 (GGGACTTG-CTCTTC), 535 (GCTGAGACATCAC), 536 (CGGTAGCAGCAAC), ' M. Ghislain, personal communication. 537 (CACGACGGCGTTC), 527 (CCAGACTCCATCA), 1,097 (GATTCCAACACCG) or part of the plus strand 1,276 and 528 (CCTGTTGATCGTG). The atpl genes cloned from R351 and amplified from A2313 were sequenced using synthetic oligonucleotides. The dideoxynucleotide chain termination method (Sanger et al., 1977) was used to sequence either single-stranded DNA (Sequenase kit, kilobase sequencing kit) or double-stranded DNA (T7 DNA polymerase). S1 Nuclease Mapping-S1 nuclease protection experiments were performed essentially as described by Davis et al. (1986). Synthetic oligonucleotide 527 was hybridized to the appropriate single-stranded DNA template derived from the pTZ18U vector containing the 5 'upstream region of a w l and extended with the Klenow polymerase in the presence of labeled dCTP. The resulting double-stranded DNA was cleaved at the unique EcoRV site (position -233; see Fig. 2), generating a fragment of 927 base pairs. The single-stranded labeled DNA was purified on a denaturating acrylamide-urea gel. Labeled DNA probe (20,000 cpm) was hybridized with 50 pg of either RNA prepared for the wild type or tRNA as a control. After incubation at 42 "C for 10 h in 10 mM Pipes,' pH 6.4 (HCI), 100 mM NaCI, and 80% (v/v) formamide, the hybridization mixture was diluted 8.5-fold in 30 mM sodium acetate, pH 4.6, 250 mM NaCI, 1 mM ZnS04, and 20 pg/ml salmon sperm DNA, containing 150 units of nuclease SI and digested at 30 "C for 30 min. Protected fragments were separated on a 4.5% acrylamide sequencing gel and revealed by autoradiography. A size scale was made by primer extension and dideoxy sequencing with the same primer (527).
Determination of NHz-terminal Protein Sequence-About 225 pg ofpurified F,-ATPase  was dissolved in the sample buffer (Laemmli, 1970), heated at 100 "C for 5 min, and loaded on a 10% sodium dodecyl sulfate-polyacrylamide gel. After electrophoresis, the upper band corresponding to the a subunit was transferred onto a glass microfiber membrane soaked in poly(4-vinyl-N-methylpyridium)iodide as described by Vandekerckhove et al. (1985). Transfer was carried out in 50 mM Tris, 50 mM boric acid, pH 8.3 (HCI), for 7 h at 4 V/cm. The membrane was then washed in 10 mM sodium borate, pH 8.0 (HCI), 25 mM NaCl and dried. The fixed polypeptide was stained with a solution of fluorescamine (1 mg/600 ml of acetone) and localized under UV light. The membrane area containing the protein was cut and placed in the reaction chamber of an automated sequencing apparatus (Applied Biosystems model 477A) coupled with a phenylthiohydantoin amino acid analyzer (Applied Biosystems model 120A). The analytical program was described by Hewick et al. (1981), and the gradient applied on the reverse phase column was described by Hunkapiller and Hood (1983).
Computer Aid and Analysis-DNA and protein analysis was made with the software PCGENE 6.00 of A.
Bairoch provided from IntelliGenetics, IncJGenofit, s.a. (Switzerland). Alignment of homologous proteins was made with the program MULTALIN developed by Corpet (1988). Protein secondary structure analysis was carried out with the software ANTHEPROT (Delbage et al., 1988).

RESULTS AND DISCUSSION
Isolation of the atpl Genomic Sequence-The previously isolated genomic clone pMal (Boutry et al., 1984) was SUSpected not to contain the whole atpl transcription promoter since transformants of the mutant A2313 by this clone only recovered a moderate growth on a respiratory substrate. Therefore this initial clone was used to screen a S. pombe BamHI genomic library and to isolate a clone containing the whole gene included in a 5.2-kb insert (Fig. 1). The atpl sequence was determined on both strands either from subclones or with synthetic oligonucleotides. More than 97% of the gene was thus sequenced on both strands, and when not, at least two sequences were obtained in the same orientation.
Structural Organization of atpl- Fig. 1 shows the general structure of atpl as determined from nucleotide sequence data displayed in Fig

TATCATTATCATGAAATAAATTTAACTACTTTTCCGACAAATTACAGT
" " " vening sequences. Their position was first assumed by identifying consensus sequences proposed to be involved in intron splicing in S. pombe (Hindley and Phear, 1984;Russell, 1989) which are GTANG at the 5'-end, CTPuAN at the internal site near the 3'-end, and NAG at the 3'-end. Indeed, we found these three sequences, namely GTATG, CTAAT, and TAG (underlined and at positions 64, 225, and 236, respectively, for the first intron and at positions 299, 550, and 565 for the second intron in Fig. 2). The coding frame and the position of the first intron could be unambiguously determined since the amino acid sequence from the NH2-terminal residue to the region surrounding the intron junction was determined (see below). Upstream of the first intron, the initiation codon could be identified since the single ATG was found preceded in frame by a stop codon at -33. Direct experimental localization of both introns was obtained by nuclease S1 protection assays, as shown in Fig. 3.
The 5'-end of the radioactive probe was located 47 nucleotides downstream of the 3'-end of the second intron. Consequently, three bands were expected with a predictable size of 47 and 60 nucleotides for two of them. These correspond to the protected regions of the second and third exons, respectively (Fig. 3, lower panel). As shown in Fig. 3, three main spots were revealed S1, S3, and S4. The major band of the lower spot S1 corresponded to a size of 49 nucleotides whereas the spot S3 was composed of several bands centered on 64 nucleotides. Both sizes (49 and 64) are in good agreement with the expected ones (47 and 60), and the 2-or +base difference could be explained by a difficulty of the nuclease to digest the probe a t exon junctions. The weak spot S2 could correspond to a breathing of the RNA/DNA at the AT-rich region of the 5'-end of the second exon.
The largest signal (S4) allowed us to locate the initiation site of transcription about 120 nucleotides upstream of the 3'-end of the first exon or 60 nucleotides upstream of the translation initiation codon. The 5"boundary corresponds to the consensus AAAC sequence reported previously (Russel, 1983). It was thus concluded that atpl contains a 1,608nucleotide-long open reading frame interrupted by two introns of 175 and 269 base pairs located near the 5'-end of the gene. Both introns with sizes of 175 and 269 nucleotides are the longest S. pombe introns ever reported in the literature (for review, see Russell, 1989). Recently, Gatermann et al. (1989) have concluded that in S. pombe, introns having a size between 160 and 300 nucleotides were spliced with a very low efficiency. Our present data show that such splicing occurs effectively in vivo, a t least for atpl.
A broad search for putative promoter sequences upstream of the transcription start site (Breathnach and Chambon, 1982) allowed us to identify a putative TATA box a t -182 (Fig. 2, dotted-underlined sequence); TATATAXXG was located 124 nucleotides upstream of the transcription start site. This observation is not consistent with the proposal that in S. pombe (Russell, 1989) as well as in higher eukaryotes (Breathnach and Chambon, 1982), TATA box elements are located closer to the transcription initiation site. On the contrary, such a situation has been reported for S. cerevisiae (Russell, 1983). Sixty-seven nucleotides downstream of the 3'-end of the coding sequence (Fig. 2, position 2122) is located the consensus hexanucleotide AATAAA involved in the 3'end processing of eukaryotic mRNA (Proudfoot and Brownlee, 1976;Monte11 et al., 1983).
The codon bias index (Bennetzen and Hall, 1982) was calculated according to the codon usage table proposed recently for S. pombe (Russell, 1989). Surprisingly, an intermediate value of 0.50 was obtained, predicting that atpl would not be highly expressed. As a comparison, we calculated the codon bias index value of S. cerevisiae atpl (nucleotide sequence available from Takeda et al., 1986), and a value of 0.57 was found. The lower codon bias index value obtained for atpl of S. pombe is in good agreement with the observation of Russell and Hall (1983) that the codon usage bias for genes of equivalent function is less severe in S. pombe than in S.

cerevisiae. Deduced Precursor and Mature Protein
Sequences of a Subunit-The predicted protein encoded by atpl is composed of 536 amino acid residues (Fig. 2)) accounting for a calculated molecular weight of 58,587, which is consistent with the size of the in vitro synthesized product of hybrid selected RNA (Boutry et al., 1984). Assignment of the atpl product to the ATP synthase a subunit was achieved by primary structure comparison with known a subunits; the identities found (not shown) were between 54% (E. coli) and 74% (S. cerevisiae). As atpl is a nuclear gene encoding a mitochondrial protein, we could expect the presence of a mitochondrial targeting peptide (Von Heijne et al., 1989) in the NH2-terminal region of the precursor. Thus, as described under "Materials and Methods," we sequenced the 20 1st-amino acid residues of the NH2-terminal region of the mature a subunit. We identified Ala2* of the precursor (Fig. 2) as the 1st residue of the mature subunit. The latter consists of 509 amino acid residues with a calculated molecular weight of 55,577. This molecular weight is consistent with gel electrophoresis estimation . However, the deduced isoelectric point of 8.8 is significantly different from the observed one, 7.3 (not shown), indicating that the experimental determination is biased probably by a strong association with the y subunit? (Williams et al., 1984;Williams and Pedersen, 1986). The targeting peptide (underlined in Fig. 2) displays the general features reported for mitochondrial targeting sequences: no acidic residue and enrichment in basic, hydrophobic, and polar amino acids (Von Heijne et al., 1989). Secondary structure analysis (Garnier et al., 1978) and estimation of the amphiphilicity (Eisenberg et al., 1982) of the 50 1st residues of the precursor a subunit revealed two amphiphilic structured segments: an a-helix Met'-Leu'" and a short @-sheet Ilez'-Leu23. The @-sheet is followed by a coil L y~'~-G l y~~ and then followed by an a-helix T~r~~-A r g~; .
These data suggest that the first predicted ahelix probably interacts with mitochondrial membranes in the early steps of targeting. Such a hypothesis is supported ' P. Falson, unpublished results. by the recent results of Lemire et al. (1989) showing that an amphiphilic NH2-terminal a-helix is required for targeting and of Endo et al. (1989) showing that the interaction of the a-helix with micelles increases the level of @-helix in the NH2terminal moiety of the presequence. A further hypothesis is that the cleavage of presequence is probably favored by the presence of a coil between the amphiphilic @-sheet and the large a-helix.
Comparison of the primary structures of the mature a subunits from different species reveals the presence in S.
pombe of 2 tryptophan residues (217 and 284 in Fig. 2) which are conserved in S. cerevisiae but not in other species in which they are replaced by tyrosine or phenylalanine residues. These tryptophans are responsible for the intrinsic fluorescence observed recently with wild-type F1-ATPase from S. pombe and which could be possibly modulated by single amino acid substitutions of the a or @ subunits (Di . In addition, the S. pombe a subunit contains a great number of cysteine residues. This observation is consistent with the particular F1-ATPase sensitivity to thiol modifiers reported previously (Falson et al., 1986). Isolation and Sequencing of the Mutant and Revertant atpl Genes-The mutant strain A2313 was characterized by the lack of immunodetectable a subunit and by no ATPase activity in a mitochondrial fraction (Boutry and Goffeau, 1982). The coding region of the mutant atpl allele was obtained by direct amplification using the polymerase chain reaction from genomic DNA. Polymerase chain reaction experiments were carried out as described under "Materials and Methods.'' Two synthetic oligonucleotides (primers 785 and 1276, Fig. 1) of 24 mer each were used to amplify a DNA fragment of 2.1 kb. Fig. 4 shows that a DNA band of the expected size was obtained without any contaminating fragment. The amplified fragment was inserted in a pBluescript SK' vector as follows. A HindIII cut (indicated in Fig. 1) generated two shorter fragments of 0.8 and 1.3 kb which were ligated to the vector. After transformation, we isolated several positive clones containing both fragments. No random mutation introduced during the amplification step was observed.
The revertant atpl of the R351 strain was cloned from a BamHI library, as described for the wild-type gene. The using synthetic oligonucleotides as described under "Materials and Methods." Total sequencing revealed no more than a single or a double mutation in atpl of the mutant and revertant strains, respectively. In the mutant gene, cytosine 1042 was converted into thymidine (Fig. 5), creating a stop codon TAA instead of CAA which encodes glutamine 173 of the mature a subunit (Gln200 of the precursor protein in Fig. 1). Thus, it appears that the mutation leads to the synthesis of only one-third of the protein. Since no a subunit was immunodetected (Boutry and Goffeau, 1982), this suggests either that the truncated subunit is not integrated into the complex or that the shortened polypeptide does not contain any epitope recognized by the polyclonal antibodies used for the immunodetection. The same single mutation was observed in the revertant atpl followed by a second one modifying adenosine 1043 into a thymidine and leading to the replacement of the stop codon TAA by TTA, which encodes leucine. Such a double mutation restores the capacity of the revertant strain to synthesize an entire subunit that was actually observed by electrophoretic analysis (Falson et al., 1987). Recently, several mutants of the E. coli a subunit have been obtained by in vitro random mutagenesis (Pagan and Senior, 1990). Three of them were due to nonsense mutations: Gln272* 382-4m-stop (281*31(o, in S. pombe), as in the A2313 strain described here but located further in the sequence.
Structural and Biochemical Properties of the Modified Enzyme-The whole process of mutation-reversion led to the replacement of the wild-type glutamine 173 into leucine. Each residue takes up approximat$y the same volume, 161.1 (glutamine) and 167.9 (leucine) A3 (Chothia, 1975) whereas each of them has a different hydrophobicity constant, 0.36 (glutamine) and 1.34 (leucine) (Sasagawa et al., 1982), indicating that the disappearance of glutamine suppresses potential hydrogen bonds and that the leucine residue locally induces a new hydrophobic environment. Secondary structure analysis (Garnier et dl., 1978;Deliiage et al., 1988) revealed that the coil + turn region containing the mutation is not dramatically modified but probably shortened. This glutamine residue is conserved in all a subunits described up to date. It is located in the segment 170-GDRQTGKT-177, corresponding to the common "glycine-rich loop" GXXXXGK, found in all nucleotide-binding proteins such as adenylate kinase, EF-Tu, Ras-P21, and the F1 / 3 subunit (Walker et al., 1982;Fry et al., 1986). It is thus the first report of a nondirected mutation on the a subunit affecting this loop, which is assumed to interact I WILD TYPE STRAIN

3' A G T T T 5'
3'AATTT 5'  with the phosphate chain of the nucleotide. Near the glutamine residue, lysine 175 (176 in S. pombe) was modified by site-directed mutagenesis and found to play a critical role in either nucleotide binding in E. coli (Rao et al., 1988) or in subunit assembly in Thermophilic bacterium PS3 (Yohda et al., 1988). More recently, an E. coli mutant in which alanine 177 (178 in S. pombe) was replaced by a valine residue was obtained, and preliminary results indicated a partial impairment of catalytic turnover (Pagan and Senior, 1990). The modified properties of the S. pombe partial revertant strain and of its purified mitochondrial F,-ATPase were described previously (Falson et al., 1987; and are summarized in Table I. (i) The growth of the strain in either fermentative or oxidative conditions was still partially impaired. (ii) The mutation made the enzyme more resistant to azide inhibition and markedly decreased its affinity for ADP. (iii) A lower amount of endogenous nucleotides, almost 1 mol/mol of enzyme, was titrated after incubation in the presence of magnesium. (iv) The most dramatic effect was the loss of apparent negative cooperativity whereas the maximal rate of ATP hydrolysis was not markedly decreased.

3' A A
Such an alteration was deduced from double-reciprocal plots of ATP hydrolysis versus ATP concentration in conditions under which ATP concentration (0.03-3 mM) was much higher than that of the enzyme. A Hill number of 1.0 was found for the mutant instead of 0.7 for the wild type. Consistent with this was the observation that bicarbonate had no activating effect.
Which Role for the a Subunit?-As these results show strongly a modified interaction between the enzyme and ADP, it was proposed that the lower enzyme affinity for ADP means that during ATP hydrolysis the step of ADP release is no longer limiting (Di . The mutation presented here is located in a region proposed by Senior and coworkers (for the E. coli enzyme) to be a nucleotide-binding domain (Maggio et al., 1987(Maggio et al., ,1988, and the modified enzyme affinity observed here is in agreement with this hypothesis. However, the ATPase activity being largely restored, it is clear that glutamine 173 of the (Y subunit has no direct catalytic role. This more likely suggests that binding of adenine nucleotide to the a subunit may control ATP hydrolysis, resulting in an apparent negative cooperativity of ATP hydrolysis. Similar effects, i.e. loss of negative cooperativity and increased dissociation constant for ADP, were obtained in the T. bacterium PS3 by another 01 subunit mutation: substitution of the aspartic acid residue 261 (Aspz7' in S. pombe) by an asparagine residue (Yohda et al., 1988). Both types of results suggest a regulatory role for the a subunit. Residues Gln'73 and Aspz7' in S. pombe are located in two segments of the a subunit, homologous to adenylate kinase segments (Fry et al., 1986), as indicated below adenylate kinase 0 15 -GGPGSGKGT -023 . . . 1 13 -LLYVD -1 19 a subunit 170-GDRQTGKTA-177 . . 266-LWYp-270 X-ray crystallography and NMR spectroscopy (Fry et al., 1986) have shown that both segments of adenylate kinase are spatially located in the same region, the first one interacting with the phosphate chain of ATP and the second one interacting with magnesium. As the effects of mutation "-* Leu and Aspz6' (270 in S. pombe) -+ Asn are quite similar, it is tempting to propose that these residues and their respective segments are spatially close together, as in adenylate kinase.
In conclusion, the results presented here show clearly that isolation of mutant and selection of related revertants are a powerful method to study yeast ATP synthase. Indeed, such a procedure was also used for the E. coli enzyme (Senior et aL, 1984) although identification and localization of reversion have not been reported yet. Further isolation of other revertants from the A2313 and other mutants is in progress to obtain new data on the ATP binding site of the a subunit.