Thermophilic F1-ATPase Is Activated without Dissociation of an Endogenous Inhibitor, ε Subunit*

Subunit complexes (α3β3γ, α3β3γδ, α3β3γε, and α3β3γδε) of thermophilic F1-ATPase were prepared, and their catalytic properties were compared to know the role of δ and ε subunits in catalysis. The presence of δ subunit in the complexes had slight inhibitory effect on the ATPase activity. The effect of ε subunit was more profound. The (−ε) complexes, α3β3γ and α3β3γδ, initiated ATP hydrolysis without a lag. In contrast, the (+ε) complexes, α3β3γε and α3β3γδε, started hydrolysis of ATP (<700 μm) with a lag phase that was gradually activated during catalytic turnover. As ATP concentration increased, the lag phase of the (+ε) complexes became shorter, and it was not observed above 1 mm ATP. Analysis of binding and hydrolysis of the ATP analog, 2′,3′-O-(2,4,6-trinitrophenyl)-ATP, suggested that the (+ε) complexes bound substrate only slowly. Differing fromEscherichia coli F1-ATPase, the activation of the (+ε) complexes from the lag phase was not due to dissociation of ε subunit since the re-isolated activated complex retained ε subunit. This indicates that there are two alternative forms of the (+ε) complex, inhibited form and activated form, and the inhibited one is converted to the activated one during catalytic turnover.

other subunits of F 1 -ATPase and Fo are involved in the rotational coupling is currently receiving attention.
The ␦ and ⑀ subunits, together with ␥ subunit, are considered to form a stalk portion that connects F 1 -ATPase to membraneembedded proton channel, Fo (13)(14)(15). ␦ subunit has an ␣helical, elongated structure (16 -18), and close proximity to the N-terminal region of ␣ subunit (19 -23) and Fo subunit (24) has been suggested. ␦ subunit may lie outside of the ␣ 3 ␤ 3 hexamer since the central cavity is mostly occupied by ␥ subunit and only little space is left there (4). Thus, ␦ subunit is currently considered as a stator that connects the ␣ 3 ␤ 3 hexamer ring to Fo. According to the recent structural study by NMR, ⑀ subunit of Escherichia coli F 1 -ATPase (EF 1 ) 1 consists of N-terminal ␤-sandwich domain and C-terminal ␣-helical domain (25). ⑀ subunit is known as an endogenous ATPase inhibitor of EF 1 and F 1 -ATPases from chloroplast (CF 1 ) (16,26,27). ⑀ subunit of EF 1 tends to dissociate from EF 1 resulting in gradual activation of ATPase activity after initiation of the ATPase assay (26). ATPase activity of CF 1 is enhanced by either reduction of the disulfide bond of ␥ subunit or removal of ⑀ subunit (28,29). Recently, Capaldi and co-worker (30) have extensively studied the interaction of ⑀ subunit with the ␣, ␤, and ␥ subunits in EF 1 by using cross-linking and chemical modification. Interestingly, ⑀ subunit changes the partner subunit of cross-linking dependent on the nucleotide in the solution. (31).
In this study, to know the function of ␦ and ⑀ subunits in catalysis of F 1 -ATPase, we compared kinetics of the homogeneous preparation of ␣ 3 ␤ 3 ␥, ␣ 3 ␤ 3 ␥␦, ␣ 3 ␤ 3 ␥⑀, and ␣ 3 ␤ 3 ␥␦⑀ complexes of the F 1 -ATPase from thermophilic Bacillus strain PS3 (TF 1 ). The results indicate that the ⑀-containing (ϩ⑀) complex can exist in two forms, inhibited form and activated form, and the former is converted to the latter without dissociation of ⑀ subunit during catalytic turnover.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmid of TF 1 ␦ Subunit-DNA fragment containing TF 1 ␦ subunit gene was prepared by polymerase chain reaction. Primer oligonucleotides were designed to introduce new restriction sites at both ends of the ␦ subunit gene (5Ј-AAGAATTCATAT-GAACCAAGAAGTGATCGCC-3Ј (EcoRI and NdeI) and 5Ј-AAGGATC-CTTAGCCGATCAGCTGCCGC-3Ј (BamHI), introduced restriction sites are shown in parentheses). Polymerase chain reaction was carried out with recombinant Taq DNA polymerase as described by the manufacturer (Takara, Japan) using the plasmid that contained TF 1 ␦ * 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  subunit gene as a template. The amplified fragments were digested with EcoRI and BamHI and cloned into EcoRI-BamHI sites of pUC118 (32) to create pUC118-␦. pUC118-␦ was digested with EcoRI and PstI, and the 0.5-kbp fragment containing TF 1 ␦ gene was cloned into EcoRI-PstI sites of pKK223-3 (33) to create pKD2 in which TF 1 ␦ gene was placed under the tac promoter. pKD2 was used as the expression plasmid for E. coli strain JM109 (34).
Construction of Expression Plasmid of TF 1 ⑀ Subunit-pTF1 (7) carrying entire TFoF 1 genes was digested with EcoRI and its termini were blunted. Then the fragment was further digested with PstI, and the resulting 6.7-kbp fragment containing TFoF 1 genes was cloned into PstI-HindIII sites of pTD-tac (35) whose HindIII-digested terminus had been blunted. Resulting plasmid, pTD-tac-TFoF 1 , was digested with SmaI, and the resulting 3.3-kbp fragment containing TF 1 ⑀ gene was self-ligated to create pTD-tac-⑀. pTD-tac-⑀ was then digested with EcoRI and HpaI. The resulting 0.6-kbp fragment containing TF 1 ⑀ gene was cloned into EcoRI-HincII sites of pTD-T7 to create pTD-T7-⑀. For efficient expression, the upstream region of the TF 1 ⑀ gene was removed (52 base pair) by the loop-out mutagenesis (36). Primer oligonucleotide (5Ј-GATCGTTTTCATAGCTGTTTCCTG-3Ј) containing both complementary sequence to 5Ј-terminus of ⑀ subunit gene and the downstream region of the T7 promoter was used. Resulting plasmid was named pTE2, in which TF 1 ⑀ gene was under the control of T7 promoter. pTE2 was used as the expression plasmid for E. coli strain BL21(DE3) (37), which harbors T7 RNA polymerase gene under the inducible lacUV5 promoter in its genomic DNA.
Purification of TF 1 ␦ and ⑀ Subunits-E. coli strain JM109/pKD2 was grown in 3 liters of Terrific Broth containing 50 g/ml ampicillin in a jar fermenter with strong agitation and aeration at 37°C. When A 600 reached to about 0.6, an inducer, isopropyl-␤-D-thiogalactopyranoside was added to final concentration of 0.7 mM. After 7 h from induction, cells were harvested by centrifugation at 5000 ϫ g for 15 min at 4°C. About 6 g (wet weight) of JM109/pKD2 cells per liter of culture media were obtained. All the following procedures were performed at 4°C. About 5 g of cells were suspended in 50 mM Tris-HCl (pH 7.5) and 1 mM EDTA (buffer A) at 0.2 g cells/ml and then were disrupted by a French pressure system (1400 kgf/cm 2 ). About half of the expressed protein was recovered in the soluble fraction of the cell lysate, and the other half was in the insoluble fraction. Soluble fraction was subjected to successive centrifugations, at 6000 ϫ g for 15 min and at 200,000 ϫ g for 20 min. Then saturated ammonium sulfate solution (pH 7.7, adjusted by ammonia solution) was added to 35% saturation. The fraction containing TF 1 ␦ was precipitated by centrifugation for 10 min at 12,000 ϫ g. The precipitant was dissolved in buffer A, and saturated ammonium sulfate solution (pH 7.7) was added to 10% saturation. Then the fraction was applied to a Butyl-Toyopearl column (3 ϫ 1.4 cm, Tosoh, Japan) equilibrated with buffer A containing 10% saturated concentration of ammonium sulfate. The column was washed with the same buffer (30 ml), and a 10 -0% saturated ammonium sulfate linear gradient in buffer A (total 60 ml) was applied. A trace amount of TF 1 ␦ was eluted at this step. With a gradient between buffer A and distilled water (total 60 ml), TF 1 ␦ was eluted as a peak fraction at the end of the gradient. The purified protein solution was frozen by liquid nitrogen and stored at Ϫ80°C until use. About 3 mg of purified ␦ subunit was obtained from 1 g of wet cells. ⑀ subunit was prepared from E. coli strain BL21(DE3)/ pTE2 as described previously (38).
Measurement of Catalytic Activities-Steady-state ATPase activities at 1 M to 5.3 mM ATP were measured spectrophotometrically by using an ATP regenerating system (39) at 25°C as described previously (40). ATP-Mg solution was prepared by mixing ATP-Tris salt (Sigma) with equal molar of MgCl 2 . The reaction was initiated by the addition of the subunit complexes (typically 10 l of 0.1 mg/ml) to the ATPase assay solution (1.0 or 1.2 ml). The absorbance at 340 nm were measured every 1 s for 20 min in a spectrophotometer UV-2200 (Shimadzu, Kyoto, Japan), and the data were stored in an on-line computer. Single-site catalysis was measured using TNP-ATP as a substrate. A reaction mixture (50 l) containing 50 mM Tris-HCl (pH 8), 4 mM MgCl 2 , 200 mM KCl, and 0.3 M TNP-ATP was incubated at 25°C. The reaction was initiated by the addition of equal volumes of 1 M subunit complexes of TF 1 in 50 mM Tris-HCl (pH 8). The reaction was quenched by the addition of 5 l of ice-cold 24% perchloric acid. In ATP-chase experiments, 10 l of 30 mM ATP-Mg was added instead of perchloric acid. After 5 s, the reaction was quenched by the addition of 5 l of ice-cold 24% perchloric acid. The amounts of TNP-ATP and TNP-ADP were measured by HPLC as described previously (41).

Measurement of TNP-ATP Binding Monitored by Fluorescent
Increase-TNP-ATP binding to the TF 1 subunit complexes were measured in a spectrofluorometer FP-777 (JASCO, Tokyo, Japan) under the condition of single-site catalysis (42). The excitation and emission wavelengths were 410 and 548 nm, respectively. The slit widths of excitation and emission were set at 10 and 20 nm, respectively. The assay mixture contained 50 mM Tris-HCl (pH 8), 100 mM KCl, 2 mM MgCl 2 , and 0.15 M TNP-ATP. The assay mixture (1.2 ml) was transferred to a glass cuvette and incubated at 25°C. Analysis of the Activated Complex with Gel-filtration HPLC-␣ 3 ␤ 3 ␥⑀ complex (1 mg/ml) was incubated with 4 mM ATP-Mg in 50 mM Tris-HCl (pH 8) at room temperature (130 l). After the 2-min incubation, 15 l of the enzyme solution was added to 1.2 ml of ATPase assay mixture that contained no ATP (49 M ATP final concentration). The ATPase activity presented in Fig. 6A was taken from the velocity at 30 -50 s after the initiation of the ATPase assay. The rest of the sample (100 l, 3-min incubation with ATP) was subjected to gel-filtration HPLC (G3000SWXL (Tosoh, Japan)) equilibrated with 50 mM Tris-HCl (pH 7.1), 100 mM KCl, and 1 mM ATP-Mg. The column was eluted at a flow rate of 0.5 ml/min. A peak fraction was concentrated with centrifuge concentrator Microcon 100 (cut off M r ϭ 100,000, Amicon), and their subunit composition was analyzed by SDS-and native-PAGE.
Other Materials and Procedures-Restriction endonucleases were obtained from Toyobo and Takara. Other chemicals were the highest grade commercially available. ␣ 3 ␤ 3 ␥ complex of TF 1 was prepared as described previously (7). Purified ␣ 3 ␤ 3 ␥ complex contained less than 0.1 mol of adenine nucleotides/mol of enzyme when analyzed by HPLC (41). TF 1 was prepared as described previously (43). One unit of enzyme activity was defined as that producing 1 mol of ADP per min at the specified ATP concentration. Recombinant DNA procedures were performed as described in a manual (44). E. coli strain JM109 was used for preparation of plasmids, and CJ236 (36) was used for generating uracilcontaining single-stranded plasmids for site-directed mutagenesis. TNP-ATP was prepared according to Hiratsuka and Uchida (45,46). Protein concentration was determined by the method of Bradford (47) using bovine serum albumin as a standard or by the UV absorbance using the factor 0.45 at 280 nm as 1 mg/ml (48). Polyacrylamide gel electrophoresis in the presence of 0.1% SDS was performed as described by Laemmli (49). The protein bands in gels were stained with Coomassie Brilliant Blue R-250. N-terminal amino acid sequencing was performed as described previously (50) and confirmed that both ␦ and ⑀ subunits had the same N-terminal amino acid sequences, starting with the N-terminal methionines, as those of subunits contained in the authentic TF 1 (51).

Homogeneous Preparations of
The subunit complexes with desired subunit combinations were obtained by a simple procedure: mixing the ␣ 3 ␤ 3 ␥ with ␦ and/or ⑀ subunits, short incubation at room temperature, and repeated ultrafiltration to remove excess subunits. Analysis of thus obtained subunit complexes by SDS-PAGE showed that the preparations contained the expected subunits (Fig.  1A). Staining intensities of ␦ and/or ⑀ bands relative to those of the bands of ␣ and ␤ subunits in each lane were apparently the same as those of the authentic TF 1 , suggesting the same subunit stoichiometry of the complexes as that of authentic TF 1 . Each preparation of the subunit complex was electrophoresed in native-PAGE as a single band with different electrophoretic mobility (Fig. 1B). Interestingly, the mobility was not simply parallel to the molecular masses of the complexes; in the order This enabled us to distinguish each complex and ensured the homogeneity of the prepared subunit complexes. The mobility of the subunit complex reconstituted from the ␣ 3 ␤ 3 ␥ complex and ␦ and ⑀ subunits was the same as that of the authentic TF 1 in native-PAGE (Fig. 1B, lanes 5 and 6), confirming again that the reconstituted subunit complex had normal subunit stoichiometry (␣ 3 ␤ 3 ␥ 1 ␦ 1 ⑀ 1 ). When 1 mM ATP-Mg was included in all of the solutions used for electrophoresis, the sample buffer, electrode buffer, and gels of native-PAGE, the same electrophoretic patterns as those in the absence of ATP were observed (Fig. 1C). If total or partial dissociation of ␦ and/or ⑀ subunits from the complexes occurred during ATP hydrolysis, the electrophoretic mobility of the band should have shifted or the band should have split into two bands. Therefore, the fact that each complex was electrophoresed in the presence of ATP-Mg as a single band with its characteristic mobility indicates that the complex is stable during ATP hydrolysis.
ATP Hydrolysis by the Complexes-Time courses of ATP hydrolysis by subunit complexes were examined at various ATP concentrations (1 M-5.3 mM). As shown in Fig. 2, the presence of ␦ subunit in the subunit complexes has slight inhibitory effect on ATPase activity. This was most distinct at 5.3 mM ATP (Fig. 2C) where about 20% of inhibitions were observed. The effect of ⑀ subunit was more profound than that of ␦ subunit, and the complexes can be classified into two groups according to the similarity of profiles of time courses, that is, the ⑀-less (Ϫ⑀) complexes (␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 ␥␦) and the ⑀-containing (ϩ⑀) complexes (␣ 3 ␤ 3 ␥⑀ and ␣ 3 ␤ 3 ␥␦⑀). When the (Ϫ⑀) complexes were mixed with the assay solution containing a low concentration of ATP, 50 M for instance as shown in Fig.  2A, ATP hydrolysis proceeded with three phases as previously reported for mitochondrial F 1 -ATPase (MF 1 ) and for ␣ 3 ␤ 3 ␥ complex of TF 1 ; initial fast phase (Ͻ10 s), partially inhibited intermediate phase (10 -300 s), and final, reactivated steadystate phase (Ͼ300 s) (52,53). The (ϩ⑀) complexes showed different time courses. There was a long lag phase, that is, an initial inhibited rate of hydrolysis was slowly activated to the final rate. Values of the ATPase activities at the final phase of the (ϩ⑀) complexes were still smaller than that of the final phase activity of the (Ϫ⑀) complexes. Hydrolysis of 500 M ATP by the (ϩ⑀) complexes occurred with a shorter lag period, and the final value of the ATPase activity reached almost the same magnitude as that of the (Ϫ⑀) complexes (Fig. 2B). At ATP concentrations above 1 mM, as shown in Fig. 2C where ATP was 5.3 mM, a lag period apparently disappeared and profiles of time courses by the four complexes became similar to each other except that the initial high activity phase (Ͻ20 s) was more pronounced for the (Ϫ⑀) complexes. Reflecting the ATP concentration dependence of the extent of activation, s-v plot for the (ϩ⑀) complexes exhibited apparently sigmoidal shape (data not shown). Reasonably, kinetic behavior of the authentic TF 1 in these experiments were all the same as those of the ␣ 3 ␤ 3 ␥␦⑀ complex although the specific activity of the former was somehow slightly lower (approximately 70%) than that of the latter.
Hydrolysis and Binding of Substoichiometric TNP-ATP-As reported previously (8), ␣ 3 ␤ 3 ␥ complex hydrolyzed a substoichiometric amount of TNP-ATP, and this hydrolysis was promoted by chase-added ATP (Fig. 3A). In a very similar manner, ␣ 3 ␤ 3 ␥␦ complex also hydrolyzed TNP-ATP, and the hydrolysis was promoted by chase-added ATP (Fig. 3B). The typical single-site catalysis and chase-promotion indicate that most TNP-ATP bound rapidly to a single high affinity catalytic site of the (Ϫ⑀) complexes upon mixing, and ATP occupation of (or ATP hydrolysis at) the second catalytic site accelerates otherwise slow hydrolysis of bound TNP-ATP. In contrast, the (ϩ⑀) complexes and authentic TF 1 hydrolyzed substoichiometric TNP-ATP only slowly, and the chase-added ATP did not promote it (Fig. 3, C-E). These different behaviors of the (ϩ⑀) complexes can be explained by assuming slow TNP-ATP binding or impairment of catalytic site cooperativity. To know which was the case, we directly measured TNP-ATP binding to the complexes with fluorescence change of TNP-ATP accompanied by its binding to the enzyme (42) (Fig. 4). The concentrations of TNP-ATP and the complexes were the same as those in Fig. 3. Upon mixing the (Ϫ⑀) complexes with TNP-ATP solution (Fig. 4, traces ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 ␥␦), fluorescence jumped within a dead period of measurement and a slow, small increase of fluorescence followed. This suggested that most TNP-ATP bound rapidly to the (Ϫ⑀) complexes although a small residual fraction of TNP-ATP bound slowly. As shown previously (8), produced TNP-ADP did not dissociate from the complexes since intensity of fluorescence was not decreased even after it reached saturation. In contrast, fluorescence jumped only slightly upon mixing with the ␣ 3 ␤ 3 ␥⑀ complex and a slow, large increase followed (Fig. 4, trace ␣ 3 ␤ 3 ␥⑀). In the case of the ␣ 3 ␤ 3 ␥␦⑀ complex, a slow, large increase followed after a small jump and a small bump of fluorescence (Fig. 4, trace ␣ 3 ␤ 3 ␥␦⑀). The cause of a small jump and bump observed is not known at present, but it is obvious that most TNP-ATP bound to the (ϩ⑀) complexes slowly. For all measurements, further addition of excess ATP to the solutions with saturated fluorescence slowly reverted the intensity of fluorescence nearly down to the level of that in the absence of the complexes (data not shown). Slow fluorescence change by the binding of TNP-ATP to the (ϩ⑀) complexes observed in Fig.  4 is almost parallel to the slow hydrolysis of TNP-ATP in Fig.  3, indicating that the substrate binding is the rate-limiting step in the hydrolysis of substoichiometric TNP-ATP by the (ϩ⑀) complexes.
Effect of Free ⑀ Subunit and Dilution of the Complexes on ATPase Activity-It was reported that the ␣ 3 ␤ 3 ␥⑀ complex of EF 1 was also gradually activated after initiation of ATP hydrolysis (16). This activation was attributed to the dissociation of inhibitory ⑀ subunit because the inclusion of free ⑀ subunit in the ATPase assay solution suppressed the activation (26). The activation of the ␣ 3 ␤ 3 ␥⑀ complex of TF 1 , however, was not suppressed by the presence of 4000-fold molar excess free ⑀ subunit in the ATPase assay solution (Fig. 5A). Similarly, inclusion of free ⑀ subunit in the ATPase assay solution did not induce inhibition of the ATPase activity of the ␣ 3 ␤ 3 ␥ complex of TF 1 and even a slight activation by the free ⑀ subunit was observed (Fig. 5A). Other evidence for the dissociation of ⑀ subunit from EF 1 was the observation that simple dilution of EF 1 in the ATPase assay solution resulted in apparent enhancement of specific activity of the enzyme (26,40). However, the dilution of the ␣ 3 ␤ 3 ␥⑀ complex of TF 1 did not change the profiles of time courses of ATP hydrolysis (Fig. 5B), and the specific activity of the ␣ 3 ␤ 3 ␥⑀ complex at the final activated phase at 500 M of ATP was hardly influenced by dilution of the enzyme; 7.2 u/mg (0.44 nM of ␣ 3 ␤ 3 ␥⑀ complex), 7.9 units/mg (1.1 nM), 8.6 units/mg (2.2 nM), and 7.9 units/mg (4.4 nM). Thus, the rate and the extent of activation of the ␣ 3 ␤ 3 ␥⑀ complex of TF 1 were not dependent on protein concentration.
⑀ Subunit Did Not Dissociate during ATP Hydrolysis-Results described in the previous paragraph suggested a possibility that the ⑀ subunit of TF 1 did not dissociate from the (ϩ⑀) complexes even during ATP hydrolysis. Indeed, the results of native-PAGE showed that all four kinds of the complexes were stable during ATP hydrolysis. To obtain further evidence for the retention of ⑀ subunit in the complex during activation, we re-isolated the fully activated (ϩ⑀) complex and analyzed the subunit composition. The ␣ 3 ␤ 3 ␥⑀ complex was incubated with 4 mM ATP-Mg, and it was fully activated to a magnitude similar to that of the ␣ 3 ␤ 3 ␥ complex (Fig. 6A, third column). The control sample remained inhibited (Fig. 6A, second column). Then, the fully activated complex was subjected to gel-filtration HPLC equilibrated and eluted with the buffer containing 1 mM ATP-Mg, and the complex was re-isolated (Fig. 6B). SDS-PAGE analysis of the re-isolated complex clearly showed that a stoichiometric amount of ⑀ subunit was contained in the complex (Fig. 6B, inset a). Homogeneity of the complex was further demonstrated by native-PAGE in which the complex was elec- trophoresed as a single band (Fig. 6B, inset b). Altogether, it is evident that ⑀ subunit remains bound to the (ϩ⑀) complex even after completion of the activation. Gradual activation of the (ϩ⑀) complex of TF 1 during catalytic turnover is not due to the dissociation of inhibitory ⑀ subunit from the complex. Instead, it appears that the (ϩ⑀) complex can exist in two forms, an inhibited form and an activated form, and conversion of the former to the latter is stimulated during catalytic turnover.

␦ Subunit of TF 1 Has Only Little Effect on ATPase Activity-
Although it was reported that EF 1 lacking ␦ subunit exhibited different kinetics from the EF 1 containing ␦ subunit (54), the effect of ␦ subunit on ATPase activity of TF 1 is not significant. Some small inhibitory effect was observed (Fig. 2C). The functional and physiological meaning of this marginal effect of ␦ subunit on ATPase activity is not understood at present.
⑀ Subunit of TF 1 Is an Inhibitory Subunit-Twenty years ago, Yoshida et al. (48) reported that ⑀ subunit did not have significant effect on the ATPase activity of TF 1 ; ␣ 3 ␤ 3 ␥␦⑀ complex had ATPase activity similar to that of ␣ 3 ␤ 3 ␥␦ complex. At almost the same time, inhibitory effect of ⑀ subunit of EF 1 was reported by Smith and Sternweis (16). This discrepancy is now settled in a conclusion that ⑀ subunit of TF 1 is also an inhibitory subunit.
However, the manner of inhibition by TF 1 ⑀ subunit is different from that by EF 1 ⑀ subunit. Inhibitory effect of TF 1 ⑀ subunit is clearly observed only at an initial phase of ATP hydrolysis at low ATP concentration as a long lag period, and when ATP concentration is high, for example at 5 mM which Yoshida et al. (48) used in the previous paper, inhibition by ⑀ subunit apparently disappears (Fig. 2C). On the contrary, inhibition by EF 1 ⑀ subunit was reported as noncompetitive inhibition; an increase of ATP concentration does not rescue the enzyme from the inhibition (55). The kinetic step assigned to be affected by the ⑀ subunit is also different between TF 1 and EF 1 . Because the (ϩ⑀) complexes of TF 1 binds TNP-ATP more slowly than the (Ϫ⑀) complexes (Fig. 4), it appears that the substrate binding step is the step affected by the ⑀ subunit of TF 1 . However, it was proposed for EF 1 that the rate of product release was slowed down by the ⑀ subunit (56). It is interesting to note that, despite the difference described above, the holoforms (consisting of five kinds of subunits) of both TF 1 and EF 1 fail to show typical "chase promotion" of single-site catalysis (Fig. 3) (54). Rather, the (Ϫ⑀) complex of TF 1 and the (-␦) complex of EF 1 exhibit the chase promotion similar to MF 1 .
⑀ Subunit of TF 1 Does not Dissociate from the Complexes-The most remarkable difference between TF 1 ⑀ and EF 1 ⑀ subunit is that the former does not dissociate from the (ϩ⑀) complex during catalytic turnover at room temperature while the latter does. Apparently similar time courses of the (ϩ⑀) complexes of TF 1 and EF 1 under appropriate conditions, a slow inhibited phase followed by gradual activation, arise from different causes. For EF 1 , dissociation of the ⑀ subunit is responsible for the activation, but for the (ϩ⑀) complexes of TF 1 , conversion from the inhibited form to the activated form without changing subunit compositions should be the reason for the gradual activation. We suppose that EF 1 can also be activated without dissociation of ⑀ subunit, but the activated complex is more unstable than the corresponding form of TF 1 so that ⑀ subunit is lost from the complex in a short period. Lotscher et al. (57) reported that ␣ 3 ␤ 3 ␥⑀ complex of EF 1 was activated 5-6-fold without dissociating ⑀ subunit when a detergent, lauryldimethylamine oxide, was present in the assay solution, and it was reverted to low activity form by diluting out the detergent. Although it was reported later that lauryldimethylamine oxide activated ␣ 3 ␤ 3 ␥ complex of EF 1 (58) and of TF 1 (53), it is A, activation of ␣ 3 ␤ 3 ␥⑀ complex. ␣ 3 ␤ 3 ␥⑀ complex was incubated with 4 mM ATP-Mg or without ATP. An aliquot was taken out, and hydrolysis of 49 M ATP was measured. The ATPase activity of ␣ 3 ␤ 3 ␥ complex is taken as 100%. B, gel-filtration analysis of the activated ␣ 3 ␤ 3 ␥⑀ complex. The rest of the sample was subjected to a G3000SWXL gelfiltration column that was equilibrated and eluted with a buffer containing 1 mM ATP-Mg. A peak fraction at elution volume of 6.6 ml was recovered and was analyzed by 14% SDS-PAGE (inset a) and by 6% native-PAGE (inset b worth considering the fact that this detergent-induced activation accompanied the movement of ⑀ subunit in the complex as reflected by the greatly reduced yield of chemical cross-linking between ␤-⑀ subunits (57). For CF 1 , activation without dissociating ⑀ subunit has been achieved by reduction of the disulfide bond of the ␥ subunit (28). With the reduction, the ⑀ subunit shifted its location or conformation because the affinity of CF 1 for the ⑀ subunit decreased (59).
Inhibited Form versus Activated Form of the (ϩ⑀) Complexes-Considering the very low initial activity of the (ϩ⑀) complexes at low ATP concentrations, we suppose that the inhibited form of the (ϩ⑀) complexes have very low ATPase activity. This means that the ATP hydrolysis observed for the (ϩ⑀) complexes in our experiments was mostly catalyzed by the activated form of the (ϩ⑀) complexes. Since the magnitude of ATPase activity of fully activated (ϩ⑀) complexes at ATP concentrations above 150 M are almost the same as those of the (Ϫ⑀) complexes, the ATPase activity of the activated form of the (ϩ⑀) complexes might be the same as or very close to those of the (Ϫ⑀) complexes. The conversion of the inhibited form to the activated one is dependent on ATP concentration. When ATP concentration is low (Ͻ150 M), the conversion is slow and incomplete ( Fig. 2A). At intermediate concentrations of ATP, the conversion becomes faster and reaches 100% yield (Fig.  2B). At high ATP concentrations (Ͼ1 mM), the conversion is so fast to reach completion upon exposure to ATP-Mg that time courses of ATP hydrolysis by the (ϩ⑀) complexes are almost indistinguishable from those by the (Ϫ⑀) complexes (Fig. 2C). Rates of the activation process increased linearly without saturation as ATP concentration increased under the conditions examined ([ATP]ϭ 9 -700 M). The final phase activity of ␣ 3 ␤ 3 ␥⑀ complex expressed as percent of that of ␣ 3 ␤ 3 ␥ complex depended on ATP concentration with a half-maximum at 40 M ATP.
We have noticed that some of the previously published results of kinetics of TF 1 are apparently contradictory to each other and to the results reported here. TF 1 hydrolyzed 1-700 M ATP with a lag (Fig. 2, A and B). A lag phase was also reported when TF 1 hydrolyzed 2 M ATP (60). However, it was reported that hydrolysis of 50 M ATP by TF 1 started with a burst but not a lag (61). Hydrolysis of TNP-ATP was reported to be chase-promoted by ATP (41), but no chase-promotion was observed in the experiment in this report (Fig. 3). Several preparations of TF 1 available for experiments in this laboratory showed the same kinetics as reported here. Although the real reason for these discrepancies is not known and being pursued, we suspect that the preparations of TF 1 used in the previous works were predominantly the activated form of TF 1 , whereas the recent preparations used in this work were the inhibited one.
Is the Shift of ⑀ Subunit a Part of Regulatory System or a Step of Rotational Catalytic Cycle?-This study has revealed that the (ϩ⑀) complexes of TF 1 , including intact TF 1 itself, can exist in two forms without changing subunit composition. Details are not known yet but binding and/or hydrolysis of nucleotide induce the transition of location and/or conformation of ⑀ subunit in TF 1 , which results in the conversion from the inhibited form to the activated form. As the authors of the previous papers on the inhibitory effect of the EF 1 ⑀ subunit suggested (16,26), this transition is possibly reminiscent of a regulatory system of ATP synthase in vivo. TF 1 exists predominantly as the activated form at physiological concentrations of ATP (Ͼ1 mM), and, therefore, protection of cellular ATP during the course of assembly of ATP synthase by ⑀ subunit (55) may not be the case. However, if, for example, the interconversion of the inhibited and activated forms occurs in ATP synthase and is influenced by the activity of respiratory chain as demonstrated for mitochondrial ATP synthase containing inhibitor proteins (62), it can work as a part of the regulatory system in responding to the growth condition of the cell.
Capaldi and co-workers (31,(63)(64)(65) reported nucleotide-dependent transition of the ⑀ subunit in EF 1 . Most remarkably, when EF 1 was incubated with ATPϩMg or ADPϩP i ϩMg, ⑀ subunit was predominantly cross-linked to the ␤ subunit, whereas it was cross-linked to the ␣ subunit with AMP-PNPϩMg (31). They considered this change as a transient shift of ⑀ subunit, which is accompanied by the rotation of the ␥-⑀ subunits in the center of EF 1 . However, another interpretation not excluded is that this change reflects the conversion of the two alternative forms of EF 1 , high-and low-activity forms. The relationship between the role of ⑀ subunit as an endogenous regulatory subunit and the presumed role of ⑀ subunit as a part of the rotor apparatus in rotational catalytic cycle is still unclear and should be clarified by experiment.