INTRODUCTION

F₀F₁-ATP synthase catalyzes ATP synthesis/hydrolysis coupled with proton flow across the membrane. It is composed of two distinctive parts, an intrinsic membrane portion, F₀, which consists of three types of subunits and forms a proton channel, and a peripheral portion, F₁, which has ATPase activity and has a subunit composition of ₃₃ (Boyer, 1993; Futai et al., 1989; Senior, 1990). F₁-ATPase has six nucleotide binding sites which are classified into three catalytic and three noncatalytic binding sites (Boyer, 1993; Futai et al., 1989; Penefsky et al., 1991; Senior, 1990). Crystal structure revealed that the  and  subunits, whose overall structures are very similar to each other, are arranged alternatively like the segments of an orange and that catalytic and noncatalytic sites are located at different interfaces of  and  subunits (Abrahams et al., 1994). Catalytic sites reside mostly on  subunits, whereas noncatalytic sites are mostly on  subunits.

A model for energy coupling by F₀F₁-ATP synthase, the binding change mechanism, assumes the rotational participation of three catalytic sites during ATP synthesis (Boyer, 1993; Cross, 1981; Duncan et al., 1995). At a given moment, three catalytic sites are in distinct functional and conformational states, but they contribute equally to catalytic turnover. According to this model, strong positive cooperativity observed for kinetics of ATP hydrolysis by F₁-ATPase is interpreted as a result of stimulation of releasing products from one catalytic site by tight binding of ATP to another catalytic site. This model leads to a prediction that, if one of three catalytic sites is incompetent, the enzyme cannot mediate normal catalytic turnover. However, the argument about the number of catalytic sites necessary for catalytic turnover under steady-state conditions has not been settled. Covalent modification of a single catalytic site by 7-chloro-4-nitrobenzofrazan (Ferguson et al., 1975; Yoshida and Allison, 1990), 5-p-fluorosulfonylbenzoyl inosine (Bullough and Allison, 1986), or 2-azido-adenine nucleotide (Cross et al., 1987; Melese et al., 1988; Van Dongen et al., 1986) is sufficient to inactivate steady-state ATPase activity of F₁-ATPase. Contradictory observations were also reported; the binding of two inhibitory molecules was necessary to cause complete inactivation of F₁-ATPases. Examples of such reagents include azido-naphthoyl-ADP (Lubben et al., 1984), fluoroaluminium- and fluoroberyllium-nucleotide diphosphate complexes (Issartel et al., 1991), 2,3-O-(2,4,6-trinitrophenyl)-ADP (TNP-ADP)¹ (Muneyuki et al., 1994), and N-ethylmaleimide for a cysteine mutant of Escherichia coli F₁-ATPase (Haughton and Capaldi, 1995). In general, labeling of the enzyme, either through covalent or strong noncovalent bonds, could fix the conformation of the enzyme in a state unfavorable for catalysis and cannot be taken as exclusive evidence for the solution of the current problem. A reconstituted hybrid complex containing one or two defective subunit(s) has been another approach to this problem. A hybrid E. coli F₁-ATPase, reconstituted from the 1:2 mixture of mutant and normal  subunits, had as low an activity as that observed for the mutant enzyme (Noumi et al., 1986). The hybrid enzymes containing a defective  subunit also did not show significant ATPase activity (Rao and Senior, 1987). Both experiments, however, were carried out using mixed-population hybrid complexes assuming the same reconstitution efficiency of various possible hybrids. Miwa et al. (1989) improved the method for preparing the hybrid complex of F₁-ATPase using the reconstitution system developed for a thermophilic Bacillus PS3 (TF₁). Their ``solid phase reconstitution'' appeared to be an effective method for obtaining homogeneous populations of hybrid ₃₃ complexes (Miwa et al., 1989). Their results indicated that the complex containing one incompetent  subunit still had significant steady-state ATPase activity. However, the conclusion became ambiguous because it was found later that hybrids used in the above experiments must have contained some amount of ₃₃ complex which could reversibly dissociate into ₁₁ complex in the presence of ATP and hence exchange subunits (Harada et al., 1991).

In pursuit of conclusive results, we have applied a new method for isolating homogeneous preparation of the ₃₃ complex with a definite number of intact catalytic sites. Although the complex with a single intact catalytic site shows the activity of ``single-site catalysis'' and the complex with two intact catalytic sites can mediate ``chase-acceleration'' of the single-site catalysis, all three intact catalytic sites are necessary for normal steady-state ATP hydrolysis.

EXPERIMENTAL PROCEDURES

TNP-ATP and TNP-ADP were prepared according to Hiratsuka and Uchida (1973) and further purified by the method described in Grubmeyer and Penefsky (1981a). Rabbit muscle pyruvate kinase, hog muscle lactate dehydrogenase, and NADH were from Boehringer Mannheim GmbH (Germany). All other products used were of analytical grade.

Recombinant plasmids carrying (wild) or (E190Q) subunit genes were previously constructed (Ohtsubo et al., 1987). Oligonucleotide-directed mutagenesis was carried out as described by Kunkel et. al (1991). The oligonucleotide used to create the Glu·Tag-linked incompetent  subunit, (E190Q + Glu·Tag), was: 5-TAGAGGATCATTCTTCTTCTTCTTCTTCTTCTTCTTCTTCCACTTCGACACC-3. The gene of (E190Q + Glu·Tag) was transferred to an expression vector of pkk (Matsui et al., 1995).

Purification of ₃₃ Complex

Purification of the ₃₃ complex was carried out as described by Matsui et al. (1995) with some minor modifications as follows. The constructed plasmids for the ₃(wild)₃ or ₃(E190Q + Glu·Tag)₃ complexes were transformed into an F₁-ATPase deletion mutant E. coli, JM103(uncB-uncD) (Monticello et al., 1992) and spread on LB plates containing 50 µg/ml ampicillin. A single colony was cultured in 3 ml of x2YT liquid medium (16 g of Tryptone, 10 g of yeast extract, 5 g of NaCl, 50 mg of ampicillin per liter), overnight at 37 °C with intense shaking. The overnight culture was inoculated into 3 liters of a rich medium containing 36 g of Tryptone, 72 g of yeast extract, 6.9 g of KH₂PO₄, 37.6 g of K₂HPO₄, and 12 ml of glycerol and cultured for 18 h at 37 °C with gentle stirring. Before harvesting, 1.2 ml of 1 M isopropyl-1-thio--D-galactoside was added, and cells were incubated successively for 3 h. The collected cells (25 g, wet weight) were stored at 80 °C.

Cells were suspended in 150 ml of 20 mM Tris-HCl, pH 8.0, and 1 mM EDTA (TCE buffer), containing 50 mM NaCl. The suspension was applied to a French press twice, and then incubated for 1 h at 60 °C. The precipitant was removed by centrifugation at 100,000 × g for 1 h at 4 °C. The supernatant fraction was applied to a DEAE-Sephacel (Pharmacia Biotech. Inc.) column (10 × 2.8 cm) that was preequilibrated with TCE buffer. The column was washed with the TCE buffer containing 150 mM NaCl and eluted with 400 ml of a linear NaCl gradient (150-500 mM). All of the eluted fractions were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE), using a 13% gel. Ammonium sulfate was added up to 18% (w/v) saturation to the fractions containing the ₃₃ complex, and the solution was loaded onto a Butyl-Toyopearl M (Tosoh) column (4.5 × 3.0 cm) preequilibrated with TCE buffer containing 18% (w/v) saturated ammonium sulfate. The complex was eluted with 300 ml of an ammonium sulfate linear decreasing gradient (18-0% (w/v) saturation), and the column was further washed successively with 100 ml of TCE buffer. The fractions containing pure ₃₃ complex were combined and stored at 4 °C as an ammonium sulfate precipitate (70% (w/v) saturation).

Fifteen milligrams of stored ₃(wild)₃ or ₃(E190Q + Glu·Tag)₃ complexes in ammonium sulfate suspension were collected respectively by centrifugation and dissolved in 5 ml of 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA. These two solutions were mixed, and the proteins were denatured by addition of 4.8 g of solid urea. The suspension was dialyzed against 3 liters of the 50 mM PIPES-NaOH, pH 7.0, and 0.2 M NaCl (G3000 buffer) for 6 h at room temperature. The medium for dialysis contained 0.2 M NaCl because it has been shown that the ₃₃ complex is destabilized and is not formed in the presence of NaCl (Amano et al., 1994a). The solution was concentrated to 100 µl with Centricon-30 (Amicon) at 25 °C and was subjected to a gel filtration HPLC column TSK-GEL G3000SW (30 × 2.15 cm, Tosoh) which was equilibrated and eluted with the G3000 buffer at a flow rate of 2 ml/min. The peak fraction of the ₃₃ complex was concentrated by Centriprep-30 (Amicon) to 500 µl, and the same volume of 50 mM PIPES-NaOH, pH 7.0, and 1 mM EDTA (QAE buffer) was added. The solution was applied on an anion-exchange HPLC column (COSMOGEL QA glass packed column, 75 × 8 mm, Nacalai tesque, Japan) equilibrated with QAE buffer. The column was eluted at a flow rate of 1 ml/min with a linear gradient of NaCl (0-1.0 M). The four protein peaks were individually collected and applied again to the same column. Each of the first, second, and third peak fractions of the first anion-exchange HPLC was eluted as an isolated, symmetrical peak in the second anion-exchange HPLC. The peak fraction of the second anion-exchange HPLC was reserved and concentrated by Centricon-30 to about 50 µl, and the solution was diluted with 20 mM Tricine-KOH (pH 7.8) more than 20-fold to adjust the absorbance at 280 nm to 0.20. This buffer exchange is essential to suppress subunit rescrambling. As to the fourth peak in the first anion-exchange HPLC, a major peak (Type 0 complex, see Fig. 1) was accompanied by a shoulder peak (Type I complex) in the second anion-exchange HPLC. To avoid contamination by Type I complex, we treated ₃(E190Q + Glu·Tag)₃ complex alone with urea, followed the above procedures, and used the peak fraction of the second anion-exchange HPLC as Type 0 complex. The complexes were stable for 1 day at room temperature without significant subunit rescrambling. In the case of a control experiment to assess the effect of Glu·Tag on the wild-type complex, ₃(wild)₃ and ₃(wild + Glu·Tag)₃ complexes were mixed and treated with the same procedures as described above.

Steady-state ATPase activity at 25 °C was measured in the presence of an ATP-regenerating system (Stiggal et al., 1979). The assay mixture contained 50 mM Tris-HCl, pH 8.0, 100 mM KCl, 2.5 mM phosphoenolpyruvate, 2 mM ATP, 2 mM MgCl₂, 0.2 mM NADH, 20 µg/ml pyruvate kinase, and 20 µg/ml lactate dehydrogenase. Different from assay mixtures used in previous reports which contained sulfate anion, this mixture did not show a long lag phase before reaching the steady-state rate of ATP hydrolysis (Amano et al., 1994a; Odaka et al., 1994). One unit of activity was defined as the activity which produced 1 µmol of ADP/min.

A solution (50 µl) containing 1.0 µM ₃₃ complex in 20 mM Tricine-KOH, pH 7.8, was mixed with an equal volume of the solution containing 0.6 µM TNP-ATP, 40 mM Tricine-KOH, pH 7.8, and 4 mM MgCl₂, and the mixture was incubated at 25 °C. At the indicated time, the reaction was quenched by addition of 5 µl of 24% (v/v) perchloric acid (single-site catalysis). To measure the chase-acceleration, 10 µl of 30 mM ATP or 10 µl of 100 µM ADP containing 2 mM inorganic phosphate were added instead of perchloric acid at the time indicated, and the mixture was incubated for 5 s. Then the reaction was quenched by the addition of perchloric acid. After centrifugation to remove the precipitated protein, the amounts of TNP-ATP and TNP-ADP were measured by reversed-phase HPLC (Hisabori et al., 1992).

Difference spectra induced by binding of TNP-ADP were measured as described previously (Hisabori et al., 1992). The concentration of protein solution was adjusted to 1 µM in 20 mM Tricine-KOH buffer, pH 8.0, containing 1 mM MgCl₂. Ten microliters of 90 µM TNP-ADP were added to 1 ml of the protein solution and, after 5 min, difference absorption spectra were measured at 25 °C with a double-beam spectrophotometer model UV-2200 (Shimadzu, Japan) using a double-sector cuvette.

Protein concentrations were determined by the method of Bradford (1976) using bovine serum albumin as a standard. SDS-PAGE was performed as in Laemmli (1970) using 13% polyacrylamide gels containing 0.1% (w/v) SDS. Protein bands were stained by Coomassie Brilliant Blue G-250.

RESULTS AND DISCUSSION

As a suitable mutant for our purpose, we used a  subunit mutant, (E190Q), in which -Glu¹⁹⁰ was replaced by Gln. TF₁ was inactivated when this residue was labeled by N,N-dicyclohexylcarbodiimide (Yoshida and Allison, 1983; Yoshida et al., 1981, 1982). The ₃(E190Q)₃ complex lost ATPase activity completely and could not hydrolyze even a single ATP molecule (Ohtsubo et al., 1987; Amano et al., 1994a). A water molecule is seen between a carboxyl group of -Glu¹⁹⁰ (numbering is according to TF₁) and -phosphate of AMP-PNP in crystal structure of MF₁ (Abrahams et al., 1994), and almost certainly -Glu¹⁹⁰ plays a critical role in catalysis as a general base to activate a water molecule for an in-line attack to the -phosphate during ATP hydrolysis (Amano et al., 1994b). In addition to this mutation, we introduced a peptide tag of 10 glutamic acids (Glu·Tag) to C terminus of the mutant  subunit. This Glu·Tag enabled us to separate the mixture of the four complexes which contained 0, 1, 2, or 3 mol of mutant  subunit(s)/mol of the complex into each of homogeneous complexes.

Outline of the procedures to isolate pure, homogeneous complexes is as follows (Fig. 1). 1) The same amounts of ₃(wild)₃ and ₃(E190Q + Glu·Tag)₃ were mixed. 2) The proteins in the mixture were denatured by addition of urea. 3) Urea was removed from the solution by dialysis. Each subunit was refolded and assembled into ₃₃ complexes. 4) Unassembled free subunits were removed by gel-filtration HPLC. 5) Each type of the complex was separated with anion-exchange HPLC. 6) Each complex was further purified by the second anion-exchange HPLC. 7) The buffer containing NaCl was replaced by Tricine buffer to suppress rescrambling of subunits.

At step 3), the wild-type and mutant  subunits were randomly incorporated into complexes, and following the four kinds of hybrid complexes should be produced, ₃(wild)₃, ₃(wild)₂(E190Q + Glu·Tag)₁, ₃(wild)₁(E190Q + Glu·Tag)₂, and ₃(E190Q + Glu·Tag)₃. They were designated as Types III, II, I, and 0 complexes, respectively, according to the number of wild-type  subunits, that is, the number of intact catalytic site(s) (Fig. 1). Note that Type III complex is just a wild-type ₃₃ complex. At step 5), four protein peaks, designated as, a, b, c, and d, were developed according to the number of Glu·Tag(s) which attached to incompetent  subunit(s) in the complex (Fig. 2A). At step 6), except the peak d fraction, each of three peak fractions of the first HPLC was eluted in a single peak at the same position as that in the first HPLC. As to the peak d fraction, a second HPLC was not sufficient to remove a small amount of the peak c fraction (data not shown). As described in next paragraph, peak d fraction corresponded to ₃(E190Q + Glu·Tag)₃ (Type 0 complex). To obtain pure Type 0 complex, we treated ₃(E190Q + Glu·Tag)₃ complex alone with the above 2)-7) procedures, and used the peak fraction of the second HPLC as Type 0 complex.

Each of purified peak fractions of Fig. 2A was analyzed with SDS-PAGE (Fig. 2C). Judged from relative staining intensity of  subunit band and  subunit band, the complexes were not contaminated by the ₃₃ complex, or very little if any. Since Glu·Tag-linked  subunit has a larger molecular size (53.2 kDa) than the wild-type  subunit (51.9 kDa), it was electrophoresed in SDS-PAGE slightly more slowly than the wild-type  subunit (Fig. 2C, lanes e and f), and its band appeared between the band of  subunit (54.6 kDa) and that of wild-type  subunit. The complex in the peak a contained wild-type  subunit (Fig. 2C, lane a) and was identified as Type III complex. The complex in the peak d contained (E190Q + Glu·Tag), but not wild-type  subunit (Fig. 2C, lane d), and was verified to be Type 0 complex. The complexes in the peak b and c contained both wild-type  subunit and (E190Q + Glu·Tag) (Fig. 2C, lane b and c) indicating that they were hybrid complexes. Judging from the relative staining intensity of both  subunit bands, it was concluded that the peak b and c contained Type II and I complexes, respectively. This order of elution, that is, the complex with more Glu·Tag was eluted later, was consistent with predicted chromatographic behaviors of the complexes with the anion-exchange HPLC. A relative amount of Type III, II, I, and 0 complexes, calculated from the peak area, was approximately 1:3:3:1 and agreed with the assumption of random subunit scrambling during the step 3) reconstitution procedures.

ATPase activity of each fraction of the first anion-exchange HPLC was measured (Fig. 2A). As expected, the peak a fraction (Type III complex) had ATPase activity. However, the other three peaks did not show activity under the assay conditions. One might argue that the Glu·Tag attached at the C terminus of the  subunit could be the reason for the loss of ATPase activity of peak b, c, and d fractions. To make this point clear, we attached the Glu·Tag to the wild-type  subunit and carried out the same experiment. The ₃(wild + Glu·Tag)₃ complex was fully active in ATP hydrolysis (specific activity, 8.25 units/mg). The ₃(wild)₃ and ₃(wild + Glu·Tag)₃ complexes were mixed and denatured, and the ₃₃ complexes were reconstituted as described above. The anion-exchange HPLC gave almost the same elution profile as that observed for Fig. 2A, and now each of four peaks had ATPase activity with almost the same specific activity (Fig. 2B). Thus, Glu·Tag has only little effect, if any, on ATPase activity of the complex.

The result of Fig. 2A indicated that Type II, I, and 0 complexes did not have ATPase activity. This conclusion was confirmed by the experiments using purified preparations of the complexes. Type III complex had ATPase activity (specific activity, 10.3 units/mg), but none of other complexes had the activity greater than 1% of that of the Type III complex (Fig. 3, Table I). Type II complex showed a trace amount of activity (0.054 unit/mg). It is interesting to note that this activity is close to the value (0.1 unit/mg) obtained for the minimum catalytic unit, most likely ₁₁ complex, of TF₁ (Saika and Yoshida, 1995). Activities of Type I and 0 complexes were below 0.01 unit/mg. The above measurements were performed at 2 mM ATP and 25 °C, an optimum temperature for ATPase activities of TF₁ and ₃₃ complex (Matsui et al., 1995), but essentially the same results were obtained from assays at 10 µM ATP and 25 °C and at 2 mM ATP and 60 °C (not shown). From these results, it was concluded that the ₃₃ complex can catalyze normal steady-state ATP hydrolysis only when all three catalytic sites are functional. This conclusion is apparently in accordance with the ``binding change mechanism'' (Boyer, 1993), which assumes sequential participation of all catalytic sites in steady-state catalysis. It also agrees to a contention that the maximum steady-state activity of E. coli F<sub>1</sub>-ATPase is achieved only when all three catalytic sites are occupied by AT(D)P (Weber et al., 1994). Although the results do not prove ``sequential'' participation of the catalytic sites, they are consistent with such a mechanism.

Table I. Steady-state ATP hydrolysis by the purified complexes Complex Freshly prepared in Tricine-KOH After 24 h In Tricine-KOH In eluted buffera units/mg Type III 10.3 9.65 11.2 Type II 0.05 0.10 0.39 Type I <0.01 <0.01 0.13 Type 0 <0.01 <0.01 <0.01 a  The peak fractions eluted from anion-exchange HPLC, which contained 0.5-0.6 M NaCl, were aged for 24 h at 4 °C without replacing the eluted buffer by Tricine-KOH, and ATPase activities were measured at 2 mM ATP.

The absence of subunit rescrambling among complexes is a prerequisite for the above experiments. We noticed that ATPase activity of the Type II complex increased up to 0.39 unit/mg if the complex was aged for 24 h at 4 °C in the solution containing a high concentration of NaCl (0.6 M) which was used for elution from anion-exchange HPLC (Table I). If all of the subunits in the Type II complex were rescrambled, a probability to reconstitute Type III complex should be 29.6% (i.e. (2/3)³), which should give a ATPase activity of 3.32 units/mg. Therefore, 12% of the Type II complex could be rescrambled in 24 h. Indeed, when an aged preparation of Type II complex was analyzed with anion-exchange HPLC, several new peaks appeared beside the Type II complex peak (not shown). However, when the buffer containing NaCl was exchanged to 20 mM Tricine-KOH, pH 8.0, the rescrambling of the subunit was suppressed (Table I). Based on this knowledge, we replaced the buffer to Tricine-KOH immediately after elution from anion-exchange HPLC, and experiments were carried out within 6 h after the preparation of the complexes.

The unexpected finding that subunit rescrambling occurs in the ₃₃ complex, which had been thought as stable as native TF₁, may explain the results reported by Miwa et al. (1989). They immobilized a cysteine-introduced incompetent  subunit to the surface of sulfhydryl-reactive resins through a disulfide bond and reconstituted a homogeneous hybrid ₃₃ complex by incubation of the resin with , , and wild-type  subunits. Then the complex was released from the resin by treatment with a sulfhydryl reagent. Thus obtained complex containing one incompetent  subunit retained significant steady-state ATPase activity. However, their experiments did not assume subunit rescrambling of the hybrid ₃₃ complex. Moreover, their preparation must have contained some amounts of ₃₃ complex, which is more susceptible to subunit rescrambling than is the ₃₃ complex (Harada et al., 1991). Therefore, the activity they detected for the complex containing one incompetent  subunit might be due to the complex with three intact  subunits generated from subunit rescrambling.

Elementary steps and partial reactions of ATP hydrolysis by F₁-ATPase can be analyzed from experiments of single-site catalysis and chase-acceleration (Cross and Nalin, 1982; Grubmeyer et al., 1982; Cunningham and Cross, 1988; Penefsky, 1988; Penefsky and Cross, 1991). As typically observed for bovine heart mitochondria F₁-ATPase (MF₁), when a substoichiometric molar amount of ATP is added to F₁-ATPase, ATP binds rapidly to a single high affinity catalytic site on one of three  subunits, and then hydrolysis of bound ATP proceeds at a relatively slow rate (single-site catalysis). This slow hydrolysis is greatly accelerated by the addition of an excess amount of ATP (chase-acceleration). Chase-acceleration has been taken as an indication of conformational communication between catalytic sites. Typical chase-acceleration was not observed for TF₁ (Yohda and Yoshida, 1987). However, it was found that TF₁ and the ₃₃ complex showed significant chase-acceleration when TNP-ATP was used as a substoichiometric substrate and ATP as a chase-accelerator (Hisabori et al., 1992; Kaibara et al., 1996). Then we measured single-site and chase-acceleration of the complexes.

As demonstrated previously, Type III complex showed single-site catalysis and chase-acceleration (Fig. 4A) (Kaibara et al., 1996), and Type 0 complex did not have any activity (Fig. 4D) (Amano et al., 1994a). Type II and I complexes could mediate single-site catalysis which had noticeable features. The Type II complex hydrolyzed about two thirds of the added TNP-ATP, and the other one third was not hydrolyzed even after long incubation or after chase-acceleration (Fig. 4B). Similarly, the Type I complex hydrolyzed about one third of the added TNP-ATP but the other two thirds remained unhydrolyzed (Fig. 4C). In general, the amount of remaining TNP-ATP at 300 s was almost proportional to the number of incompetent  subunits in the complex; Type 0 complex, 95%; Type I complex, 60%; Type II complex, 25%; Type III complex, 0%. This result suggests that substoichiometric TNP-ATP binds to one of the catalytic sites of the complex without preference for an intact or incompetent site, and, in cases where it binds to incompetent site(s), TNP-ATP remains bound without hydrolysis.

Interestingly, Type II complexes exhibited chase-acceleration (Fig. 4B, open circles) while Type I complex did not (Fig. 4C, open circles). This indicates that at least two functional catalytic sites are necessary for chase-acceleration. When the second catalytic site is incompetent as in the case of Type I complex, bound ATP at this site is not hydrolyzed, and ADP is not produced. Therefore, either ATP hydrolysis itself or the presence of products at the second site might be responsible for the chase-acceleration. To distinguish these two possibilities, ADP and P_i were chase-added instead of ATP. The result was that none of four kinds of complexes showed chase-acceleration of TNP-ATP hydrolysis (Fig. 4, A-D, open triangles). Probably some step in the process of ATP hydrolysis at the second catalytic site triggers a conformational change which leads to the chase-acceleration at the first catalytic site. Grubmeyer and Penefsky (1981b) reported some weak chase-acceleration of MF₁ using ADP or nonhydrolyzable ATP analog as chase promoters. The reason for the difference between our results and theirs is not clear.

The above argument is based on the assumption that a single high affinity site for the TNP-nucleotide, where single-site catalysis occurs, is retained by the complexes containing the mutant  subunit(s). Our previous results of equilibrium dialysis and difference spectra using TNP-ADP showed that the Type III complex has a single high affinity site for TNP-ADP on one of the  subunits (Kaibara et al., 1996). The incompetent  subunit used in this work has Gln instead of Glu at position 190. Although this residue plays a direct role in catalysis, its contribution to the binding of substrate appears to be little since Glu¹⁹⁰ is not directly involved in substrate binding according to the crystal structure of MF₁ (Abrahams et al., 1994). This was confirmed by analysis of the difference absorption spectra of TNP-ADP induced when it binds to the complexes. Binding of TNP-ADP to the  subunit in the complex induces a characteristic difference spectrum which is easily distinguished from the spectrum induced by its binding to the  subunit in the complex (Hisabori et al., 1992). When TNP-ADP was added at 1:0.9 molar ratio to each of the complexes, shapes and magnitudes of difference spectra of the complexes containing mutant  subunit(s) were almost identical with those of the Type III complex (Fig. 5). It was concluded that all of the complexes have a single high affinity site on one of the  subunits and, therefore, loss of single-site catalysis of the Type 0 complex cannot be attributed to the loss of a single high affinity site.