Modulation of Nucleotide Specificity of Thermophilic FoF1-ATP Synthase by ϵ-Subunit

The C-terminal two α-helices of the ϵ-subunit of thermophilic Bacillus FoF1-ATP synthase (TFoF1) adopt two conformations: an extended long arm (“up-state”) and a retracted hairpin (“down-state”). As ATP becomes poor, ϵ changes the conformation from the down-state to the up-state and suppresses further ATP hydrolysis. Using TFoF1 expressed in Escherichia coli, we compared TFoF1 with up- and down-state ϵ in the NTP (ATP, GTP, UTP, and CTP) synthesis reactions. TFoF1 with the up-state ϵ was achieved by inclusion of hexokinase in the assay and TFoF1 with the down-state ϵ was represented by ϵΔc-TFoF1, in which ϵ lacks C-terminal helices and hence cannot adopt the up-state under any conditions. The results indicate that TFoF1 with the down-state ϵ synthesizes GTP at the same rate of ATP, whereas TFoF1 with the up-state ϵ synthesizes GTP at a half-rate. Though rates are slow, TFoF1 with the down-state ϵ even catalyzes UTP and CTP synthesis. Authentic TFoF1 from Bacillus cells also synthesizes ATP and GTP at the same rate in the presence of adenosine 5′-(β,γ-imino)triphosphate (AMP-PNP), an ATP analogue that has been known to stabilize the down-state. NTP hydrolysis and NTP-driven proton pumping activity of ϵΔc-TFoF1 suggests similar modulation of nucleotide specificity in NTP hydrolysis. Thus, depending on its conformation, ϵ-subunit modulates substrate specificity of TFoF1.

F o F 1 -ATP synthase (F o F 1 ) 2 is ubiquitously found in membranes of bacteria, chloroplast, and mitochondria and synthesizes ATP by the energy of proton flow driven by the proton motive force. F o F 1 also is able to catalyze the reverse reaction, ATP hydrolysis-driven proton pumping, which actually occurs in some cases and conditions. F o F 1 is a motor enzyme composed of two rotary motors, membrane integral F o , which converts the proton motive force into rotation, and water-soluble F 1 , which converts the rotation into synthesis of ATP (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). F 1 has a subunit composition of ␣ 3 ␤ 3 ␥␦⑀ and acts as ATPase when isolated. It has been known for the typical bacterial enzymes from thermophilic Bacillus PS3 (TF o F 1 ) and Escherichia coli, that the smallest subunits of F 1 , ⑀ acts as an endogenous inhibitor of ATPase activity under some conditions (11)(12)(13)(14). It is a 15-kDa protein composed of an N-terminal ␤-sandwich (ϳ80 residues) and C-terminal ␣-helical (ϳ50 residues) domains (15,16). Two ␣-helices in the C-terminal domain undergo large conformational transition between the "up-state," in which the helices are extended as a long arm, and the C terminus reaches near the catalytic sites inside the ␣ 3 ␤ 3 -ring, and the "downstate," in which the helices are retracted as a hairpin beside a globular domain of the ␥-subunit (17)(18)(19). When ⑀ is in the up-state, ATP hydrolysis is suppressed and transition to the down-state accompanies activation (20 -22). The transition is induced by ATP (but not ADP) binding to F 1 , and then the down-state is further stabilized by binding of ATP directly to ⑀ (21,23,24). Consequently, ⑀ favors the up-state in the absence of ATP even when high concentration of ADP is present, and it undergoes the transition to the down-state as ATP concentration raises. Apparent binding affinities of ATP (K d ) to cause the transition is 140 M for TF o F 1 at 59°C, a physiological temperature of Bacillus PS3 (25). These values obviously are lower than the normal cellular ATP concentration of bacteria (ϳ3 mM in E. coli) (26). Therefore, the function of C-terminal helices is assumed as "emergency brake" to prevent ATP hydrolysis, which works only when cellular ATP concentration decreased drastically under severely starving conditions (25,27).
These contentions have been deduced mostly from analysis of ATP hydrolysis, and here, we studied the effect of conformational state of ⑀ on ATP synthesis. As we have shown previously (20,21), ⑀ in TF 1 is predominantly in the up-state when ATP in the solution is eliminated by hexokinase. As a control mimicking TF o F 1 with the down-state ⑀, we used a mutant ⑀⌬c-TF o F 1 , in which ⑀ lacks C-terminal helices (27)(28)(29)(30). Comparison of nucleotide specificity of wild-type TF o F 1 (WT-TF o F 1 ) and ⑀⌬c-TF o F 1 in NTP (ATP, GTP, UTP, or CTP) synthesis reactions in the presence of hexokinase suggests that, when ⑀ adopts the up-state conformation, TF o F 1 gains higher nucleotide specificity to ATP. Results of NTP-driven proton pumping by ⑀⌬c-TF o F 1 and NTP synthesis in the presence of adenosine 5Ј-(␤,␥imino)triphosphate (AMP-PNP) by TF o F 1 are interpreted consistently by the above contention.

EXPERIMENTAL PROCEDURES
Preparation of Inverted Membrane Vesicles-In this study, WT-and ⑀⌬c-TF o F 1 were expressed from plasmids pTR19-ASDS (31) and pTR19-ASDS-⑀⌬c (29 DK8 (bglR, thi-1, rel-1, HFrPO1, ⌬(uncB-uncC), ilv:Tn10) (31). Cultivation of the recombinant strains and preparation of inverted membrane vesicles (IMVs) were performed as described (31). ATP hydrolysis activities of IMV containing WT-and ⑀⌬c-TF o F 1 were inhibited efficiently (Ͼ80%) by dicyclohexylcarbodiimide. IMV was also prepared by the same procedures from thermophilic Bacillus PS3 cells that were cultured as described (33). NTP Synthesis Assay with HPLC-NTP synthesis activity of IMV was measured with HPLC using an anion-exchange column, Super-Q 5PW (TOSOH Corp.). The column was equilibrated with 300 mM potassium phosphate buffer (pH 7.5) containing 10 mM CDTA at a flow rate of 1 ml/min. The NTP synthesis reaction was performed in 50 mM potassium phosphate buffer (pH 7.5) containing 5 mM MgCl 2 , 15 M adenosine-penta-5Ј-phosphate, 0.3 mM NDP, and IMV (72 g proteins/ml). After incubation of the assay mixture for 2 min at 45°C, 5 mM sodium succinate was added to initiate synthesis reaction. After the reaction (500 s), 10 mM CDTA was added to terminate the reaction. It was confirmed that 10 mM CTDA completely stopped synthesis and hydrolysis reactions (data not shown). The reaction mixtures were applied to a centrifugal concentrator (100 kDa, AmiconUltra) to remove IMV. Then, the pass-through fraction was applied to the column. Elution was monitored with absorbance at 260 nm. Separately, several concentrations of NTP and NDP were applied to the column to obtain the calibration curves by which peak areas were converted into nucleotide amount.
NTP Synthesis Assay with NDP-regenerating System-NTP synthesis activity of IMV was monitored in real time with an NDP-regenerating system at 45°C in 25 mM potassium phosphate buffer (pH 7.5) containing 5 mM MgCl 2 , 15 M P 1 ,P 5di(adenosine-5Ј)-pentaphosphate, 200 units/ml thermophilic hexokinase (TOYOBO) that can use NTP as a substrate, 25 units/ml thermophilic glucose-6-phosphate dehydrogenase (TOYOBO, Japan), 10 mM D-glucose, 0.5 mM NADP, IMV (72 g proteins/ml), and 0.3 mM NDP. After preincubation of the assay mixture for 2 min at 45°C, 5 mM sodium succinate was added to the mixture to generate electrochemical potentials of protons by the E. coli respiratory chain. At the end of the measurements, 1 g/ml FCCP was supplemented to the reaction mixture to dissipate the potential and to know the slope of the base-line. NTP synthesis activity was calculated from the change in A 340 during the period of 100 to 300 s after the succinate addition, using a molar absorbance coefficient of NADPH, 6220 A/M Ϫ1 cm Ϫ1 . The activity that synthesized one mol of ATP per min was defined as one unit. Activities of synthesis of other NTPs, GTP, UTP, or CTP, were determined by introducing GDP, UDP, or CDP to the reaction mixture instead of ADP. The reaction mixture contained excess amount of hexokinase that converted NTPs to NDPs during the synthesis reactions. Judged from HPLC analysis, contamination of ATP in the ADP and AMP-PNP used in the experiments was Ͻ0.1%.
ATP Synthesis Assay with Luciferase-ATP synthesis in the presence of GTP, UTP, and CTP was measured by firefly luciferase that uses only ATP to emit luminescence (32). To avoid a large accumulation of ATP that could cause the conformational transition of ⑀-subunit, the lowest possible amount of IMV was used. Because of the heat-unstable nature of luciferase, reaction was performed at 37°C. Reaction mixture contained 20 mM potassium phosphate (pH 7.5), IMV (3 g proteins/ml), 5 mM MgCl 2 , 0.3 mM ADP, 15 M P 1 ,P 5 -di(adenosine-5Ј)pentaphosphate, and 1/10 volumes of the CLSII solution containing luciferin/luciferase (ATP bioluminescence assay kit, Roche Applied Science). The synthesis reaction was initiated by addition of 5 mM sodium succinate, and synthesis of ATP was monitored with the luminescence at 37°C. At the end of each measurement, 1.7 M ATP was added to the assay mixture several times to calibrate the luminescence intensity into ATP concentration. To view the effects of other NTPs (GTP, UTP, and CTP) and AMP-PNP on the activity, some or all of these components were supplemented to the reaction mixture at the final concentrations of 0.3 mM prior to the reaction. No significant luminescence was observed before addition of sodium succinate due to high specificity of luciferase toward ATP.
FRET Measurement-The conformational state of ⑀-subunit in TF o F 1 was analyzed by FRET measurement 45°C as described previously (25). Briefly, IMV containing TF o F 1 labeled by fluorescent dyes (a donor Cy3 at the N terminus of the ␥-subunit and an acceptor Cy5 at the C terminus of the ⑀-subunit) was prepared and suspended in 50 mM HEPES/KOH buffer (pH 7.5), 100 mM KCl, 5 mM MgCl 2 , 0.5 mg/ml pyruvate kinase, and 2.5 mM phosphoenolpyruvate. Various concentrations of ATP was added, and the degree of the transition (upstate to down-state) was assessed after 40 min of incubation at 45°C from the donor fluorescence at 566 nm.
Other Analytical Methods-NTP-driven proton pumping was monitored by fluorescence quenching of 9-amino-6chloro-2-methoxyacridine (excitation at 410 nm, emission at 480 nm) at 43°C as described previously (31). NTPase activities were measured at 45°C in 50 mM HEPES/KOH buffer (pH 7.5) containing IMV (10 g proteins/ml), 100 mM KCl, 5 mM MgCl 2 , 0.2 g/ml FCCP, indicated concentration of NTP, and the NTP-regenerating system composed of 0.5 mg/ml pyruvate kinase (from rabbit muscle, Roche Applied Science), 0.2 mg/ml lactate dehydrogenase, 2.5 mM phosphoenolpyruvate, 0.2 mM NADH, and 2.5 mM KCN (31,34). Average hydrolysis rates in a time period from 3 to 6 min after initiation of the reactions were obtained. The activity that hydrolyzed one mol of NTP per min was defined as one unit. Protein concentrations were determined by the BCA Protein Assay kit from Pierce, with bovine serum albumin as a standard. All data shown in the present study were measured at least in triplicate. Error bars in figures represent the standard deviations. Molar absorbance coefficients of 15,400 (259 nm), 13,700 (253 nm), 10,000 (262 nm), and 9000 A/M Ϫ1 cm Ϫ1 (271 nm) were used for quantification of AT(D)P, GT(D)P, UT(D)P, and CT(D)P throughout this study.  (25), and ⑀⌬c-TF o F 1 , which lacks the C-terminal two helices (Ile 88 -Lys 133 ) of ⑀, is thought to act as the TF o F 1 with ⑀ in the downstate. They were expressed in E. coli cells, and the IMV prepared from these cells was used for NTP synthesis measurements. SDS-PAGE analysis confirmed that the content of ⑀⌬c-TF o F 1 in IMV was almost the same as that of WT-TF o F 1 as reported previously (30). Nucleotides were identified and quantified with anion-exchange HPLC. Retention time of ADP was different from those of other NDPs, and we confirmed the absence of detectable contamination of ADP in the GDP, UDP, and CDP used in the experiments (Fig. 1A). The reaction mixtures contained 0.3 mM of one of NDPs and IMV containing WT-TF o F 1 or ⑀⌬c-TF o F 1 . The synthesis reactions were started by addition of succinate to the mixture to generate the proton motive force through oxidation by the E. coli respiratory chain. After incubation for 500 s at 45°C, the reactions were terminated by addition of 10 mM CDTA, and nucleotides were analyzed with HPLC (Fig. 1B). Control samples supplemented with an uncoupler, FCCP, or with 10 mM CDTA from the beginning gave the background elution profiles with no net NTP synthesis. IMV prepared from F o F 1 -deficient E. coli (DK8) did not show net NTP synthesis (supplemental data). The amounts of synthesized NTP were obtained from net increase of the peak areas, and the synthesis activities were calculated (Fig. 1C). Both WT-TF o F 1 and ⑀⌬c-TF o F 1 had apparently the same activity of ATP synthesis. However, ⑀⌬c-TF o F 1 had a 2-fold higher activity of GTP synthesis than WT-TF o F 1 . A small amount of UTP synthesis was observed for ⑀⌬c-TF o F 1 but not for WT-TF o F 1 . Net synthesis of CTP was not detected both for WT-TF o F 1 and ⑀⌬c-TF o F 1 .

NTP Synthesis by WT-TF o F 1 and ⑀⌬c-TF
NTP Synthesis in Presence of NDP-regenerating System-In the above HPLC experiments, the back reaction (NTP hydrolysis) and up-to-down transition of ⑀ in WT-TF o F 1 became possible as the synthesized NTP accumulated with time. To avoid this possibility, we next included hexokinase in the assay to convert NTP into NDP. As shown previously (20,21), ⑀ in TF o F 1 predominantly adopts the up-state conformation in the presence of hexokinase. The hexokinase used in the experiments accepts any NTPs as a substrate, and the produced glucose 6-phosphate was monitored (35). The reaction was started by addition of succinate to the reaction mixture containing 1 mM NDP and IMV ( Fig. 2A). NTP synthesis proceeded at a constant rate after a short initial lag (Ͻ30 s). The activity values and nucleotide specificity of NTP synthesis of WT-TF o F 1 measured with NDP-regenerating system at 0.3 mM NDP were almost the same as those measured with HPLC; ATP synthesis (ϭ100%) Ͼ GTP synthesis (40 -50%) Ͼ Ͼ UTP and CTP synthesis (ϳ3%) (Fig. 2B, left panel). For ⑀⌬c-TF o F 1 , the activity values of ATP, GTP, and UTP synthesis were Ͼ2-fold higher than those measured with HPLC. Furthermore, small but significant CTP synthesis was detected in the presence of NDP-regenerating system. It appeared that, different from WT-TF o F 1 , ⑀⌬c-TF o F 1 catalyzed hydrolysis of the synthesized NTP during the synthesis reaction and prevention of the hydrolysis by NDPregenerating system made the synthesis activity larger. Nucleotide specificity in NTP synthesis by ⑀⌬c-TF o F 1 observed with NDP-regenerating system showed the trend similar to that observed with HPLC; ATP synthesis (100%) Ϸ GTP synthesis Ͼ The reaction mixture contained 0.3 mM NDP, 50 mM potassium phosphate, and IMV (72 g proteins/ ml). The reaction was initiated by addition of 5 mM sodium succinate. After incubation for 500 s at 45°C, the reaction was terminated by addition of 10 mM CDTA, and the nucleotides in the reaction mixtures were analyzed with HPLC (red trace). In the control samples, 10 mM CDTA (black traces) or 1 g/ml FCCP (green traces) were further included in the mixtures from the start of the reaction. The amount of synthesized NTP was determined from the net increase of the peak areas (red and black). C, comparison of NTP synthesis activities of WT-TF o F 1 and ⑀⌬c-TF o F 1 . Detailed experimental conditions are as described under "Experimental Procedures." abs, absorbance.
UTP synthesis (20%) Ͼ CTP synthesis (17%). The concentrations of substrate NDP in the above experiments were 0.3 mM and the difference between WT-TF o F 1 and ⑀⌬c-TF o F 1 potentially could be attributed to their different dependence on NDP concentration. However, it was not the case because essentially the same results were obtained at 1 mM NDP, indicating that the activities were already saturated as V max at 0.3 mM NDP (Fig. 2B, right panel). ATP and GTP Synthesis by WT-TF o F 1 and ⑀⌬c-TF o F 1 in Presence of AMP-PNP-It has been known that a nonhydrolyzable ATP analogue, AMP-PNP, can stabilize the down-state ⑀ in F 1 (18,20,21). Taking advantage of the fact that AMP-PNP does not significantly inhibit ATP synthesis (36), we examined the effect of AMP-PNP on NTP synthesis by WT-TF o F 1 and ⑀⌬c-TF o F 1 . ATP and GTP synthesis activities of ⑀⌬c-TF o F 1 were, respectively, only slightly and not at all inhibited by AMP-PNP. In a sharp contrast, ATP and GTP synthesis activities of WT-TF o F 1 were stimulated, respectively, nearly and Ͼ2-fold by AMP-PNP. ATP synthesis activity of WT-TF o F 1 was stimulated as AMP-PNP concentration increased, and it reached the same level of that of ⑀⌬c-TF o F 1 at 1 mM AMP-PNP (Fig. 3A). Similarly, GTP synthesis activity of WT-TF o F 1 was stimulated by AMP-PNP, and the stimulated activity was almost the same as that of ⑀⌬c-TF o F 1 at 2 mM AMP-PNP (Fig. 3B). AMP-PNP alone at 2 mM without ADP and GDP did not show any NTP synthesis activities, eliminating a possibility that contaminating ADP (or GDP) in AMP-PNP was responsible for the synthesis. Nucleotide specificity of the synthesis reaction by WT-TF o F 1 was also changed by AMP-PNP. As seen in the activity of GTPsynthesis relative to ATP-synthesis (Fig. 3C), AMP-PNP stimulated GTP synthesis of WT-TF o F 1 more markedly than ATP synthesis, apparently making nucleotide specificity of WT-TF o F 1 similar to that of ⑀⌬c-TF o F 1 . Apparent dissociation constant of AMP-PNP (K d(AMP-PNP) ) in the stimulation of WT-TF o F 1 was estimated to be 0.1ϳ0.3 mM. These results clearly show that WT-TF o F 1 behaves as if it were ⑀⌬c-TF o F 1 in the presence of sufficient concentration of AMP-PNP where ⑀ in WT-TF o F 1 adopts the down-state.
We also measured ATP synthesis reaction in the presence of 0.3 mM of GTP, UTP, and CTP. The synthesized ATP was assayed by luciferase that is highly specific to ATP. As seen (Fig.  3D), the presence of these nucleotides did not interfere with ATP synthesis by WT-TF o F 1 and ⑀⌬c-TF o F 1 . Also, the effect of AMP-PNP on ATP synthesis was not influenced by these nucleotides. Therefore, GTP, UTP, and CTP have actually no effect on ATP synthesis; they do not compete with adenine nucleotides (ADP, ATP, and AMP-PNP) for catalytic or other modulatory nucleotide binding sites of TF o F 1 under the conditions tested.

ATP and GTP Synthesis by Authentic TF o F 1 of Bacillus Cells-Similar experiments were performed for IMV prepared from
Bacillus PS3 cells. Measurement at 45°C showed severalfold stimulation of ATP and GTP synthesis activities of authentic Bacillus PS3 TF o F 1 by AMP-PNP (Fig. 4A) with an apparent K d(AMP-PNP) of 0.1ϳ0.3 mM. The thermostable auxiliary enzymes hexokinase and glucose-6-phosphate dehydrogenase enabled us to raise the assay temperature up to 58°C, a physiological temperature of Bacillus PS3. At 58°C, AMP-PNP stimulated ATP-and GTP-synthesis activities severalfold, but higher AMP-PNP concentration was needed to saturate the stimulatory effect (Fig. 4B). At either 45 or 58°C, GTP synthesis activity increased more sharply than ATP synthesis as the AMP-PNP concentration rose, and both activities became almost equal at 2 mM (45°C) or 3 mM (58°C) AMP-PNP (Fig.  4C). Thus, AMP-PNP stimulates NTP synthesis activity of Bacillus PS3 TF o F 1 , either from recombinant E. coli cells or Bacillus PS3 cells, with concomitant loss of strict nucleotide specificity.
NTP Hydrolysis and Proton Pumping by WT-TF o F 1 and ⑀⌬c-TF o F 1 -We measured NTP hydrolysis activities (NTPase) of IMV containing WT-TF o F 1 and ⑀⌬c-TF o F 1 . Activities were measured at 45°C as NADH oxidation with auxiliary enzymes, pyruvate kinase and lactate dehydrogenase. Pyruvate kinase used in the experiments has a broad substrate specificity toward NTPs (34), and it keeps the initial NTP concentrations during reaction time by regenerating NTP from NDP. When ATPase activity at the same ATP concentration was compared between WT-TF o F 1 and ⑀⌬c-TF o F 1 ; both had similar activities at an ATP concentration Ͼ100 M (Fig. 5A). Below 100 M, however, ATPase activities of WT-TF o F 1 became much smaller than those of ⑀⌬c-TF o F 1 (Fig. 5A, inset). From the plot of activity of WT-TF o F 1 relative to ⑀⌬c-TF o F 1 as a function of ATP concentration, an ATP concentration that gave half-maximal activation of ATPase activity of WT-TF o F 1 was estimated to be ϳ52  M (Fig. 5B). We also measured the up-to-down transition of ⑀ in WT-TF o F 1 at 45°C by change of FRET between dyes attached on the C terminus of ⑀ and the N terminus of the ␥-subunit (Fig. 5B) (25). At low ATP concentration, donor fluorescence was weak, and ⑀ mostly adopted the up-state conformation. As ATP concentration increased, donor fluorescence increased reflecting increase in population of the down-state ⑀. An ATP concentration that gave half-maximum donor intensity was estimated to be ϳ58 M. This value was smaller than the previous one (140 M at 59°C) because temperature was lower (25). Good coincidence between the two plots in Fig. 5B suggests that the up-to-down transition of ⑀ in TF o F 1 occurs at ϳ55 M ATP accompanying activation of ATPase activity.
When GTP was a substrate, on the contrary, WT-TF o F 1 had severalfold smaller activity than ATPase, whereas ⑀⌬c-TF o F 1 hydrolyzed GTP at almost the same rate as ATP (Fig. 5A). Less strict nucleotide specificity of ⑀⌬c-TF o F 1 was also observed in hydrolysis of UTP and CTP (Fig. 5C). WT-TF o F 1 was almost unable to hydrolyze 1 mM UTP and CTP, but ⑀⌬c-TF o F 1 hydrolyzed 1 mM UTP at a significant rate (ϳ15% of ATPase) (Fig. 5D).
The nucleotide specificity of NTP-driven proton pumping activity of ⑀⌬c-TF o F 1 was also compared with that of WT-TF o F 1 (Fig. 5E). Upon addition of 1 mM nucleotide, a small, immediate shift of fluorescence level of 9-amino-6-chloro-2methoxyacridine occurred due to direct interaction of 9-amino-6-chloro-2-methoxyacridine with nucleotide. The extent of the shift differed among nucleotide, with GTP as the largest. The shift was followed by relatively slow quenching of fluorescence that reflected acidification of the inside lumen of IMV. Proton pumping by WT-TF o F 1 was driven efficiently by ATP, weakly by GTP, and not at all by UTP and CTP. In the case of ⑀⌬c-TF o F 1 , GTP was able to drive proton pumping as efficiently as ATP, and UTP showed a barely detectable pumping activity. From the results of ⑀⌬c-TF o F 1 , we can conclude that TF o F 1 with down-state ⑀ can utilize GTP as efficiently as ATP in hydrolysis reactions. As to WT-TF o F 1 , ATP hydrolysis and ATP-driven proton pumping should be catalyzed by WT-TF o F 1 with the down-state ⑀ because substrate ATP at 1 mM was present. On the contrary, conformational state of ⑀ in 1 mM GTP should be a mixture of the up-state and the down-state (21), hampering the clear interpretation.

DISCUSSION
The ⑀-subunit has been long known as an intrinsic inhibitor of ATPase activity of F 1 and F o F 1 . Recent studies revealed that ⑀ can adopt two conformations, and the upstate conformation is responsible for the inhibition. Furthermore, it was found that ATP can bind to and stabilizes the down-state ⑀ in some bacteria (13). Here, we added another novel aspect of the functions of ⑀, that is, modulation of nucleotide specificity. Given that AMP-PNP would stabilize the down-state ⑀ in TF o F 1 under the experimental conditions, we can indicate that the native TF o F 1 of Bacillus PS3 might also modulate nucleotide specificity by ⑀ at a physiological temperature (Fig. 4). As well, ⑀ is likely to modulates nucleotide specificity in ATP hydrolysis (Fig. 5).
From a structural aspect, higher nucleotide specificity of TF o F 1 with the up-state ⑀ is not surprising because the residues of the C-terminal region of ⑀ can reach the location close to the catalytic site when C-terminal helices of the up-state ⑀ are extended like a long arm into the inside of ␣ 3 ␤ 3 ring. We think it is likely because the introduced cysteines at ␤Lys 334 near the catalytic site and ⑀Glu 131 at the third position from the C ter- minus can form a disulfide cross-link when ⑀ is up-state in the absence of nucleotide. 3 In the actively growing E. coli cell, cellular ATP concentration is in the range of ϳ3 mM (26) and concentrations of other three NTPs are estimated to be ϳ1.5 mM or 0.7-1.2 mM (GTP), and ϳ0.7 mM (UTP and CTP) (37,38). Therefore, hydrolysis of the three NTPs by F o F 1 with the down-state ⑀ and, synthesis as well, can occur in the cell. Although its physiological significance is yet unclear, it is tempting to speculate that ⑀ acts as an ATP-dependent switch that changes the mode of the enzyme operation. When the cell energy becomes poor and ATP concentration drops, TF o F 1 would limit ATP and GTP consumption, and the synthesis activity is restricted to ATP.
In summary, our results suggest that the ⑀-subunit of F o F 1 -ATP synthase from a thermophilic Bacillus regulates not only magnitude of activity but also modulates specificity of substrate nucleotide both in synthesis and hydrolysis reactions, depending on its extended and retracted conformations.