Inhibition of ATP Hydrolysis by Thermoalkaliphilic F1Fo-ATP Synthase Is Controlled by the C Terminus of the {varepsilon} Subunit

The F₁F_o-ATP synthases of alkaliphilic bacteria exhibit latentATPase activity, and for the thermoalkaliphile Bacillus sp.strain TA2.A1, this activity is intrinsic to the F₁ moiety.To study the mechanism of ATPase inhibition, we developed aheterologous expression system in Escherichia coli to produceTA2F₁ complexes from this thermoalkaliphile. Like the nativeF₁F_o-ATP synthase, the recombinant TA2F₁ was blocked in ATPhydrolysis activity, and this activity was stimulated by thedetergent lauryldimethylamine oxide. To determine if the C-terminaldomain of the ${varepsilon}$ subunit acts as an inhibitor of ATPase activityand if an electrostatic interaction plays a role, a TA2F₁ mutantwith either a truncated ${varepsilon}$ subunit [i.e., TA2F₁( ${varepsilon}$ ^C)] or substitutionof basic residues in the second ${alpha}$ -helix of ${varepsilon}$ with nonpolar alanines[i.e., TA2F₁( ${varepsilon}$ ^6A)] was constructed. Both mutants showed ATP hydrolysisactivity at low and high concentrations of ATP. Treatment ofthe purified F₁F_o-ATP synthase and TA2F₁( ${varepsilon}$ ^WT) complex with proteasesrevealed that the ${varepsilon}$ subunit was resistant to proteolytic digestion.In contrast, the ${varepsilon}$ subunit of TA2F₁( ${varepsilon}$ ^6A) was completely degradedby trypsin, indicating that the C-terminal arm was in a conformationwhere it was no longer protected from proteolytic digestion.In addition, ATPase activity was not further activated by proteasetreatment when compared to the untreated control, supportingthe observation that ${varepsilon}$ was responsible for inhibition of ATPaseactivity. To study the effect of the alanine substitutions inthe ${varepsilon}$ subunit in the entire holoenzyme, we reconstituted recombinantTA2F₁ complexes with F₁-stripped native membranes of strainTA2.A1. The reconstituted TA2F_oF₁( ${varepsilon}$ ^WT) was blocked in ATP hydrolysisand exhibited low levels of ATP-driven proton pumping consistentwith the F₁F_o-ATP synthase in native membranes. ReconstitutedTA2F_oF₁( ${varepsilon}$ ^6A) exhibited ATPase activity that correlated with increasedATP-driven proton pumping, confirming that the ${varepsilon}$ subunit alsoinhibits ATPase activity of TA2F_oF₁.

INTRODUCTION

Top

Introduction

F₁F_o-ATP synthases are membrane-bound enzymes found in mitochondria,chloroplasts, and bacteria that couple the transport of protonsdown an electrochemical gradient to the synthesis of ATP (2,47, 61). Alternatively, the enzyme can work as an ATPase, hydrolyzingATP to generate an electrochemical gradient of protons ( ${Delta}$ µ_H⁺).F₁ contains the catalytic sites of the enzyme and is water soluble,whereas F_o is embedded in the membrane and translocates protonsacross the membrane. Bacterial F₁ consists of five subunitswith a stoichiometry of ${alpha}$ ₃ß₃ ${gamma}$ ${delta}$ ${varepsilon}$ , in which the catalyticnucleotide-binding sites are on the ß subunits andthe noncatalytic nucleotide-binding sites are located on the ${alpha}$ subunits (1, 36, 57). F_o comprises 3 subunits, 1 a subunit,2 b subunits, and a ring of 10 to 15 c subunits.

In eukaryotic organisms, the ATPase activity of F₁F_o-ATP synthasesis subject to regulation. In mitochondria, ATPase activity isregulated by the natural inhibitor protein IF₁, which bindsto the ATP synthase in a pH-dependent manner. Cross-linkingstudies suggest an interaction between IF₁ and the C-terminalregion of the ß subunit (21). No homologue of IF₁has been found in either chloroplasts or bacteria. ATPase activityof the chloroplast F₁F_o-ATP synthase (CF₁F_o) is subject to complexregulation to prevent wasteful ATP hydrolysis under dark conditions.Under such conditions, the enzyme is inhibited by Mg²⁺-ADP (8),the ${varepsilon}$ subunit (38, 39, 42), and the oxidation/reduction stateof the ${gamma}$ disulfide bond (37). Exposure of thylakoid membranesto light results in the generation of a ${Delta}$ µ_H⁺ that stimulatesthe release of inhibitory bound ADP and relieves inhibitionby the ${varepsilon}$ subunit (23). Reduction of the ${gamma}$ disulfide bond furthersimulates ATP synthesis (37).

Several bacterial ATP synthases exhibit low or inhibited ATPaseactivity while being competent in ATP synthesis activity (13-15,17, 35), but the mechanisms responsible for this latency arenot understood. It is noteworthy that these bacteria are aerobicmicroorganisms that are faced with the challenge of blockingATP hydrolysis activity when the cells become limited for oxygenand the ${Delta}$ µ_H⁺ decreases. Studies with the ATP synthase fromParacoccus denitrificans have revealed that the trapping ofMg²⁺-ADP at a high-affinity inhibitory site is responsible forthe low ATPase activity of this enzyme (62).

The strongest candidate for a regulator of ATPase activity inbacterial ATP synthases is the ${varepsilon}$ subunit. The ${varepsilon}$ subunit acts asan inhibitor of ATP hydrolysis activity in F₁ alone, as wellas in the entire F₁F_o enzyme in bacteria (24, 25, 31, 48). The ${varepsilon}$ subunit from F₁F_o-ATP synthases is a two-domain protein whichconsists of an N-terminal part that forms a flattened 10-strandedß-sandwich structure and a C-terminal domain thatforms two ${alpha}$ -helices running antiparallel to one another (53,58, 59). Furthermore, it is apparent that the ${varepsilon}$ inhibitory effectis caused by the C-terminal ${alpha}$ -helical domain of this subunit(25, 29, 60). Two high-resolution structural arrangements havebeen reported for the bovine mitochondrial central stalk inF₁ and an Escherichia coli complex of ${gamma}$ and ${varepsilon}$ (10, 43). Tsunodaet al. (52) have been able to trap conformational changes ofthe ${varepsilon}$ subunit in E. coli using cross-linking studies, thus demonstratingthat at least two distinct conformations of the ${varepsilon}$ subunit existin the enzyme complex of E. coli. In the ${varepsilon}$ conformation withthe C-terminal domain facing toward F₁ (up state), ATP hydrolysisis strongly inhibited but ATP synthesis is not affected. Inthe other conformation of the ${varepsilon}$ subunit with the C-terminal domaintowards the F_o (down state), the enzyme operates with equalefficiency in either direction (i.e., ATP synthesis or hydrolysis).Hara et al. (12) have shown that the electrostatic interactionbetween the positive charges in the C-terminal ${alpha}$ -helical domainof the ${varepsilon}$ subunit and the negative charges in the DELSEED motifin the ß subunit is essential for the inhibitory effectof the ${varepsilon}$ subunit.

Suzuki et al. (51) have investigated the factors that influencethe two transitions. In the up state, the two helices of ${varepsilon}$ arefully extended and insert into F₁. Without an added nucleotide, ${varepsilon}$ is in the up state and is stabilized by the ${Delta}$ µ_H⁺. ATPaddition induces the transition to the down state, and ADP counteractsthe action of ATP. Based on these observations, the authorspropose that increases in the ${Delta}$ µ_H⁺ and ADP concentrationtransform the ${varepsilon}$ subunit into the up-state conformation, thusgearing the enzyme towards ATP synthesis (51). Recent work suggeststhat the ${varepsilon}$ subunits of some bacteria are capable of binding ATP,and Yoshida and coworkers propose that the ${varepsilon}$ subunit acts asa built-in cellular sensor of ATP concentration (26).

A common feature of alkaliphilic F₁F_o-ATP synthases is theirspecific blockage of the ATP synthase in the ATP hydrolysisdirection (7, 15, 17). This feature of the enzyme is more pronouncedin the ATP synthase from Bacillus sp. strain TA2.A1, and evenafter reconstitution into proteoliposomes, ATP-driven protontransport or proton-coupled ATP hydrolysis activity could notbe detected (7). In contrast, low but measurable ATP hydrolysisactivities coupled to proton transport were reported for theATP synthases from Bacillus alcalophilus (18) and Bacillus pseudofirmusOF4 (15). We have previously hypothesized that this propertyof the enzyme may be a necessity to survive at extremes of alkalinepH and temperature under aerobic growth conditions (9). Basedon the observation that alkaliphilic bacteria grow under conditionswhere the ${Delta}$ µ_H⁺ is low due to an inverted pH gradient, thenatural tendency of a freely reversible enzyme will be to operateas an ATPase, thus leading to alkalinization of the cytoplasmand hydrolysis of valuable intracellular ATP. This will be furthermagnified when these aerobic alkaliphilic bacteria become limitedfor oxygen and the ${Delta}$ µ_H⁺ collapses. We have demonstratedthat the preferential blockage of ATP hydrolysis activity andstimulation by lauryldimethylamine oxide (LDAO) were intrinsicto the F₁ moiety (7). Based on these data, it is tempting topropose that the ${varepsilon}$ subunit of the ATP synthase from strain TA2.A1is permanently fixed in a conformation in which the rotationalmovement in the ATP hydrolysis direction is impaired.

In a previous study, the complete nucleotide sequence of theatp operon of strain TA2.A1 was determined (27). Sequence analysisand alignment of the C-terminal domain (residues 89 to 134)of the ${varepsilon}$ subunit with those from other bacilli and E. coli confirmedthe presence of a high proportion of positively charged residueswith a cluster of four basic residues (i.e., Arg, His, Lys,and Arg) at positions 116 to 119 (Fig. 1). This cluster of basicresidues in strain TA2.A1, along with other basic residues,may form a very strong electrostatic interaction with the ³⁸³DELSDED³⁸⁹motif in the ß subunit of strain TA2.A1 and may effectivelyanchor the ${varepsilon}$ subunit in a distinct conformation that contributesto the mechanism of ATP hydrolysis blockage. To address thishypothesis, we have developed a recombinant system to overproduceTA2F₁ complexes in E. coli and study the effect of mutationsin ${varepsilon}$ on latent ATPase activity.

View larger version (24K):
[in this window]
[in a new window]
FIG. 1. Alignment of the C-terminal domain of the ${varepsilon}$ subunit from Bacillus species and E. coli. Basic residues are shown in boldface. Sources of the amino acid sequences were as follows: B. pseudofirmus OF4 (20), Bacillus halodurans (accession number NC_002570), Bacillus megaterium (3), Bacillus subtilis (46), Geobacillus stearothermophilus (accession number D38060), Bacillus sp. strain PS3 (40), and E. coli (54).

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and growth conditions. Escherichia coli DH10B (11) was used for all cloning experiments,and E. coli mutant DK8 (28), lacking the ATP synthase genesof the unc operon, was used to overproduce the TA2F₁ complexesfrom the ATP synthase of Bacillus sp. strain TA2.A1.

Construction of expression plasmids. In a previous study, the genes encoding the F₁ subunits werecloned into the expression vector pTrc99A (Amersham Biosciences)under the control of the strong tryptophan promoter to createplasmid pTrcF₁ (50). Plasmid pTrcF₁ ${varepsilon}$ ^C, lacking the coding sequencefor the C-terminal domain (residues 90 to 135) of ${varepsilon}$ , was constructedby digesting pTA2F₁ (50) with AccI and ClaI and cloning this4.7-kb fragment into pTrc99A digested with AccI. To constructthe ${varepsilon}$ mutant plasmids, pTrc ${varepsilon}$ ^4A and pTrc ${varepsilon}$ ^6A, two 0.23-kb ClaI-HindIIIfragments containing codons for four and six alanine substitutionsin the C-terminal domain of ${varepsilon}$ , respectively, were created bythe overlap extension PCR method (16). These fragments werethen cloned simultaneously with a 4.7-kb EcoRI-ClaI fragment,containing the remainder of the F₁ genes, into pTrc99A digestedwith EcoRI and HindIII. To create the first ClaI-HindIII fragment,two pairs of primers (viz., atpCfwd2 with atpCmutantP1 and atpCmutantP2with pUC8rev) (Table 1) and plasmid pA3 (27) as the templatewere used to generate mutated DNA fragments. These overlappingfragments were then used for overlap extension PCR with theexternal primers atpCfwd2 and pUC8rev, and the product obtainedwas digested with ClaI and HindIII. To create the second ClaI-HindIIIfragment containing codons for a total of six alanine substitutions,the same procedure as above was used except that plasmid pTrc ${varepsilon}$ ^4Aserved as a template and the primer pairs atpCfwd2 with atpCmutantP3and atpCmutantP4 with pTrc99Arev (Table 1) were utilized. ThePCRs were carried out with the Phusion high-fidelity PCR kit(Finnzymes) according to the manufacturer's instructions. Theregion amplified by PCR in the ${varepsilon}$ mutant plasmids was confirmedby DNA sequencing.

View this table:
[in this window]
[in a new window]
TABLE 1. Mutant primers used for overlap extension PCR

Expression and purification of recombinant TA2F₁ complexes. The plasmids were expressed in E. coli DK8 by growth in 2x YTmedium (45) containing 2 g/liter glucose and 100 µg/mlampicillin at 37°C with shaking to an optical density at600 nm of 0.35 to 0.4. An inducer, isopropyl ß-D-thiogalactopyranoside(IPTG) was added (1 mM final concentration), and incubationcontinued for 5 h. Cells were harvested, washed with precooled100 mM MOPS (morpholinepropanesulfonic acid) buffer (pH 8.0),and resuspended in MOPS-Mg buffer (100 mM MOPS, pH 8.0, 2 mMMgCl₂, 10% glycerol). Phenylmethylsulfonyl fluoride (PMSF) wasadded to 0.1 mM, and the cells were disrupted by two passagesthrough a French pressure cell at 20,000 lb/in². PancreaticDNase was added to 0.1 mg/ml, and the mixture was kept on icefor 1 h or until viscosity decreased. Lysate was cleared ofdebris by centrifugation at 8,000 x g for 10 min, and the membraneswere pelleted from the supernatant at 180,000 x g for 1 h at4°C. The supernatant was retained and incubated at 60°Cfor 30 min to denature E. coli proteins, and these were removedby centrifugation at 180,000 x g for 20 min at 4°C. Theheat-treated supernatant was then brought to 5 mM MgCl₂ and0.5 M NaCl before application to a 50-ml column of chelatingSepharose Fast Flow (Amersham Biosciences). Before applyingthe TA2F₁ samples, the chelating Sepharose column was loadedwith 80 ml of 0.3 M CuSO₄, washed with five column volumes ofdistilled water, and equilibrated with five column volumes of20 mM Tris-HCl (pH 8.0) containing 0.5 M NaCl, 5 mM MgCl₂, and10% glycerol. After the TA2F₁ samples (120 ml) were loaded,the column was washed with three column volumes of the samebuffer. The bound TA2F₁ complexes were eluted with one columnvolume of the same buffer containing 100 mM imidazole. The elutedTA2F₁ complexes were dialyzed (molecular weight cutoff [MWCO],6,000) against 20 mM Tris-HCl (pH 8.0) containing 1 mM MgCl₂and 10% glycerol to remove excess salt. The samples (approximately200 mg protein) were applied to a 60-ml POROS HQ 50 column (AppliedBiosystems), which was equilibrated with the dialysis buffer.Bound proteins were eluted by 12 column volumes of a 0 to 600mM linear gradient of NaCl in the same buffer. The ATPase-containingfractions were pooled and concentrated to 30 mg/ml using AmiconUltra centrifugal filter devices (MWCO, 100,000). The ATPasesamples (1 ml) were then applied in aliquots of 0.2 ml to aGPC Superose 6 column (Amersham Biosciences) and fractionatedwith 20 mM Tris-HCl (pH 8.0), 1 mM EDTA, and 100 mM NaCl buffer.The peak between 370 to 380 kDa was pooled and immediately usedfor ATPase assays and reconstitution studies.

Biochemical methods. Protein concentrations were determined by using the bicinchoninicacid protein assay kit from Sigma, with bovine serum albuminas the standard. Purified TA2F₁ complexes were analyzed by 13%polyacrylamide gel electrophoresis (PAGE) in the presence of0.1% sodium dodecyl sulfate (SDS) (30), and protein bands werevisualized by staining with Coomassie brilliant blue.

Proteolysis of TA2F₁ complexes and TA2F₁F_o-ATP synthase. All protease treatments of TA2F₁ complexes and TA2F₁F_o wereperformed in an assay mixture containing 50 mM Tricine [N-tris(hydroxymethyl)methylglycine]-NaOH,pH 8.0, 10% sucrose, 10% glycerol, 2 mM EDTA, and 2.5 mM ATP.TA2F₁ and TA2F₁F_o preparations (1 mg/ml) were incubated at 30°Cwith either trypsin (DPCC [diphenyl carbamyl chloride] treated)or ${alpha}$ -chymotrypsin (TLCK [N ${alpha}$ -p-tosyl-L-lysine chloromethyl ketone]treated) at a ratio of 1:250 to 1:1,000 (wt/wt). At varioustime intervals 5-µl aliquots were withdrawn and immediatelyassayed for ATPase activity in the presence and absence of LDAO.At intervals of 60 to 120 min, proteolysis was stopped by a15-fold dilution of the sample into buffer (50 mM MOPS [pH 7.5]and 2 mM MgCl₂) containing PMSF (100-fold excess by weight overproteases) and stored at –70°C. The subunit compositionwas analyzed by SDS-PAGE.

Reconstitution of TA2F₁ complexes with F₁-stripped membranes. Inverted membrane vesicles of strain TA2.A1 were prepared aspreviously described (7). To strip F₁ from wild-type membranes,approximately 15 mg of washed membranes (500 µl) was resuspendedin 100 ml of 5 mM Tris-sulfate buffer (pH 8.0) containing 0.5mM EDTA and the mixture was incubated for 30 min at room temperaturewith gentle agitation. The homogenate was subsequently centrifugedat 180,000 x g for 45 min. The pellet was resuspended in 100ml of 0.5 mM EDTA (pH 8.0) with stirring for 1 h at room temperature,followed by high-speed centrifugation (180,000 x g for 45 min)to harvest the membranes. The pellet was resuspended in 0.5ml of 50 mM potassium phosphate buffer (pH 7.8) containing 2mM MgCl₂ and 0.1 mM PMSF. The ATPase activity (LDAO stimulated)of the F₁-stripped membranes indicated that the protocol waseffective at removing >95% of the F₁ moiety.

Purified TA2F₁ complexes (1.3 mg) were incubated with freshlyprepared F₁-stripped membranes (1.5 mg) in a total volume of300 µl, and the volume was increased to 5 ml with 50 mMMOPS (pH 7.5) containing 100 mM KCl and 2 mM MgCl₂ (buffer A).The mixture was incubated on ice for 60 min and then centrifugedat 180,000 x g for 45 min. The membranes were washed with bufferA (10 ml) to remove unbound TA2F₁ complexes. This step was repeatedthree or four times or until the supernatant contained no ATPaseactivity.

Determination of ATPase and proton-pumping activity. ATP hydrolysis activity was measured at 45°C by using anATP-regenerating assay. The assay mixture contained 100 mM MOPS(pH 7.5), 3 mM phosphoenolpyruvate, 2 mM MgCl₂, 0.25 mM NADH,0.57 U/ml pyruvate kinase, 3.2 U/ml lactate dehydrogenase, and50 µM or 2 mM ATP, as specified in the figure legends.The reaction was initiated by the addition of the sample into1 ml of assay mixture, and the rate of NADH oxidation was monitoredcontinuously at 340 nm. Nine micrograms of TA2F₁ subunit complexeswas used for measurements. The activity that hydrolyzed 1 µmolof ATP per min is defined as 1 unit.

ATP-dependent proton translocation was determined at 45°Cby the quenching of 9-amino-6-chloro-2-methoxyacridine (ACMA).The 1.5-ml reaction mixture contained 5 mM potassium phosphatebuffer (pH 7.5), 5 mM MgCl₂, 1 µM ACMA, and 150 to 200µg of inverted membranes or reconstituted TA2F_oF₁ complexes.After the fluorescence signal had stabilized (100%), the reactionwas initiated by the addition of neutralized ATP (1.7 mM). Fluorescencewas measured with an excitation wavelength of 410 nm and anemission wavelength of 480 nm (slit width, 10 nm). At the timesindicated in Fig. 7, 1.5 µM of carbonyl cyanide m-chlorophenylhydrazone(CCCP) was added.

View larger version (24K):
[in this window]
[in a new window]
FIG. 7. ATP-driven proton translocation and ATPase activity of reconstituted TA2F₁ complexes with F₁-stripped F_o membranes. ATP-dependent proton translocation was determined at 45°C by the quenching of ACMA as described in Materials and Methods. The reaction was initiated by adding 1.7 mM ATP and terminated with 1.5 µM CCCP at the times indicated by arrows. (A) Native membranes; (B) F₁-stripped membranes; (C) reconstituted F_oF₁( ${varepsilon}$ ^WT) membranes; (D) reconstituted F_oF₁( ${varepsilon}$ ^6A) membranes; (E) reconstituted F_oF₁( ${varepsilon}$ ^C) membranes. (F) ATPase activity of reconstituted F_oF₁( ${varepsilon}$ ^WT), F_oF₁( ${varepsilon}$ ^6A), and F_oF₁( ${varepsilon}$ ^C) membranes assayed at 45°C with the ATP-regenerating assay and the addition of 2 mM ATP.

RESULTS AND DISCUSSION

Top

Results and Discussion

Purification and characterization of recombinant TA2F₁ complexes. To study the role of the ${varepsilon}$ subunit in the ATP synthase of Bacillussp. strain TA2.A1, we developed a heterologous expression systemin E. coli to produce recombinant TA2F₁ complexes with variousmutations in atpC. The genes encoding TA2F₁ were overexpressedin DK8 ( ${Delta}$ atp) and purified using a three-step procedure. Thepurity of the recombinant TA2F₁ complexes was assessed by SDS-PAGEand compared to the subunit composition of purified TA2F₁F_ofrom native membranes (Fig. 2). The five protein bands fromthe recombinant TA2F₁( ${varepsilon}$ ^WT) complex, corresponding to the ${alpha}$ , ß, ${gamma}$ , ${delta}$ , and ${varepsilon}$ subunits, coincided with those from the native TA2F₁F_o-ATPsynthase (Fig. 2, compare lanes 1 and 4). The TA2F₁( ${varepsilon}$ ^C) complex,containing the ${varepsilon}$ subunit from which the C-terminal domain hadbeen removed (Fig. 2, lane 3), and the TA2F₁( ${varepsilon}$ ^6A) complex, wherethe basic residues in ${varepsilon}$ (i.e., Arg¹¹⁶, His¹¹⁷, Lys¹¹⁸, Arg¹¹⁹,Arg¹²³, and Arg¹²⁷) were substituted with nonpolar alanines(Fig. 2, lane 2), exhibited the expected changes at a proteinlevel. The ${varepsilon}$ subunit of TA2F₁( ${varepsilon}$ ^6A) migrated at a slightly lowermolecular weight than TA2F₁( ${varepsilon}$ ^WT).

View larger version (74K):
[in this window]
[in a new window]
FIG. 2. Analysis of recombinant TA2F₁ complexes and TA2F₁F_o-ATP synthase by 13% SDS-polyacrylamide gel electrophoresis. Lane 1, wild type F₁; lane 2, F₁ complex containing mutant ${varepsilon}$ ^6A subunit; lane 3, F₁ complex containing truncated ${varepsilon}$ ^C subunit; lane 4, F₁F_o-ATP synthase purified from native membranes. Fifty micrograms of F₁ complexes (lanes 1 to 3 [from the left]) and 50 µg of F₁F_o-ATP synthase (lane 4) were applied.

To determine whether the recombinant TA2F₁ complexes were functional,ATP hydrolysis assays were carried out. Like the native F₁F_o-ATPsynthase, the TA2F₁( ${varepsilon}$ ^WT) was blocked in ATP hydrolysis activitywith a specific activity of 0.9 units mg protein^–1. ATPaseactivity was activated 30-fold by the addition of LDAO (Fig.3A), and maximal ATP hydrolysis activity was observed at 0.04%LDAO with a specific activity of 28.5 units mg protein^–1.Purified TA2F₁F_o-ATP synthase was also stimulated by LDAO (20-fold,19.4 units mg protein^–1), but 10-fold more LDAO was requiredto reach near-saturation ATPase activity (Fig. 3B).

View larger version (12K):
[in this window]
[in a new window]
FIG. 3. Effect of LDAO on ATPase activity. ATP hydrolysis activity of purified TA2F₁( ${varepsilon}$ ^WT) complex from membrane vesicles (A) and purified TA2F₁F_o-ATP synthase from native membranes (B) was measured with increasing amounts of LDAO at 45°C using the ATP-regenerating assay described in Materials and Methods. The reaction was initiated with 2 mM ATP. The basal ATPase activities (no LDAO addition) of TA2F₁( ${varepsilon}$ ^WT) and TA2F₁F_o was 0.90 units mg protein^–1 and 0.97 units mg protein^–1, respectively.

To determine the effect of ${varepsilon}$ on ATP hydrolysis activity, TA2F₁( ${varepsilon}$ ^WT)was compared to TA2F₁( ${varepsilon}$ ^C). At low concentrations of ATP (50 µM),TA2F₁( ${varepsilon}$ ^WT) exhibited very low levels of ATP hydrolysis with aspecific activity of 0.12 units mg protein^–1 (Fig. 4A).In contrast, TA2F₁( ${varepsilon}$ ^C) exhibited levels of ATPase activity thatwere approximately sevenfold higher (0.85 units mg protein^–1)(Fig. 4A). At higher concentrations of ATP (2 mM), TA2F₁( ${varepsilon}$ ^WT)had a rate of ATP hydrolysis activity of 1.2 units mg protein^–1and the ATP hydrolysis activity of TA2F₁( ${varepsilon}$ ^C) was 4.3 units mgprotein^–1 (Fig. 4B). To identify the residues in the C-terminaldomain of the ${varepsilon}$ subunit important in this process, we changedthe cluster of basic residues at positions Arg¹¹⁶, His¹¹⁷, Lys¹¹⁸,and Arg¹¹⁹, as well as Arg¹²³ and Arg¹²⁷, to alanine residues(Fig. 1). In a mutant in which four of the basic residues (viz.,Arg¹¹⁶, His¹¹⁷, Lys¹¹⁸, and Arg¹¹⁹) were changed to alanines[i.e., TA2F₁( ${varepsilon}$ ^4A)], ATP hydrolysis activity was still blockedat 50 µM ATP, but at 2 mM the ATPase activity was similarto TA2F₁( ${varepsilon}$ ^C) (Fig. 4C and D). The TA2F₁( ${varepsilon}$ ^6A) mutant in which allsix basic residues were changed to alanines exhibited levelsof ATP hydrolysis activity that were equal to that of TA2F1( ${varepsilon}$ ^C)at both low and high concentrations of ATP (Fig. 4C and D).The kinetics of ATP hydrolysis for TA2F₁( ${varepsilon}$ ^WT), TA2F₁( ${varepsilon}$ ^C), andTA2F₁( ${varepsilon}$ ^6A) were determined (Fig. 4E). With increasing concentrationsof ATP, TA2F₁( ${varepsilon}$ ^WT) hydrolyzed ATP at a rate that was almost proportionalto the external ATP concentration, and the kinetic analysesusing the Lineweaver-Burk equation indicated that the apparentK_m and V_max were 0.52 mM and 1.27 units mg protein^–1,respectively. In contrast, TA2F₁( ${varepsilon}$ ^C) and TA2F₁( ${varepsilon}$ ^6A) showed highrates of ATP hydrolysis at low ATP concentrations (Fig. 4E).For TA2F₁( ${varepsilon}$ ^C), the apparent K_m and V_max were 0.20 mM and 3.03units mg protein^–1, respectively. Similar values wereobtained for TA2F₁( ${varepsilon}$ ^6A) (i.e., the K_m and V_max were 0.17 mM and3.11 units mg protein^–1, respectively). Based on the observationthat TA2F₁( ${varepsilon}$ ^6A) behaved like TA2F₁( ${varepsilon}$ ^C), we decided to use TA2F₁( ${varepsilon}$ ^6A)for further characterization studies. When the rates of ATPhydrolysis for TA2F₁( ${varepsilon}$ ^WT) and TA2F₁( ${varepsilon}$ ^6A) were measured at increasingconcentrations of ATP in the presence of 0.05% LDAO, typicalsaturation kinetics were observed (Fig. 4F). The V_max of TA2F₁( ${varepsilon}$ ^6A)was significantly higher (twofold) than that of TA2F₁( ${varepsilon}$ ^WT) (Fig.4F), but the affinity constants were in similar ranges (i.e.,between 0.2 to 0.3 mM ATP).

View larger version (24K):
[in this window]
[in a new window]
FIG. 4. Effect of ${varepsilon}$ on ATPase activity of recombinant TA2F₁ complexes. (A and B) ATP hydrolysis by TA2F₁( ${varepsilon}$ ^WT) and TA2F₁( ${varepsilon}$ ^C) complexes was measured with the ATP-regenerating assay, as described in Materials and Methods, by monitoring the continuous change in absorbance at 340 nm (OD₃₄₀). The reaction was initiated by the addition of ATP at the times indicated by arrows. ATP concentrations were 50 µM (A) and 2 mM (B). (C and D) ATPase activity of TA2F₁ complexes and TA2F₁F_o-ATP synthase using 50 µM ATP (C) and 2 mM ATP (D). (E and F) ATPase activity of F₁( ${varepsilon}$ ^WT) ( ${square}$ ), F₁( ${varepsilon}$ ^C) ( ${triangleup}$ ), and F₁( ${varepsilon}$ ^6A) (•) with increasing amounts of ATP without LDAO (E) and with 0.05% LDAO (F).

The above results demonstrate that the C-terminal arm of the ${varepsilon}$ subunit inhibits ATP hydrolysis activity of TA2F₁, and themost likely mechanism for this inhibition is a strong electrostaticinteraction between the cluster of basic residues in the second ${alpha}$ -helix of ${varepsilon}$ and another subunit in the F₁ complex. The othersubunit in this interaction remains to be identified, but, onthe basis of previous studies, the DELSEED region of the ßsubunit seems a likely possibility (12). Studies with the chloroplastATP synthase have shown that the C-terminal domain of the ${varepsilon}$ subunitis a regulator of ATP hydrolysis activity both in CF₁ and CF₁F_o,and most of this inhibitory activity can be attributed to thelast 45 C-terminal amino acids (38, 39). While the ${varepsilon}$ subunitis a regulator of ATP hydrolysis activity, ATP synthesis experimentsrevealed that the C-terminal domain of the ${varepsilon}$ subunit is not requiredfor photophosphorylation (39).

Lowry and Frasch (33) have highlighted that other subunits ofthe soluble E. coli F₁-ATP synthase (ECF₁) play a role in regulatingATP hydrolysis activity, namely, the salt bridges between theresidues ßD372 and ${gamma}$ K9. We noted some variation inthe apparent K_m values for ATP of the TA2F₁( ${varepsilon}$ ^WT) and TA2F₁( ${varepsilon}$ ^6A)complexes, suggesting a potential role for ${varepsilon}$ in modulating theK_m of the ß subunits for ATP. ${varepsilon}$ has been proposed toact as a sensor of intracellular ATP in vivo for some bacterialspecies (26). This could be mediated through an effect whereby ${varepsilon}$ converts the enzyme from a form with a high affinity for bindingATP to one where the affinity is greatly reduced. Under physiologicalconditions within the cell, ATP hydrolysis could simply be impairedby the lowered affinity of the enzyme for ATP. ATP hydrolysiswould therefore be observed only if the intracellular ATP concentrationincreased to very high levels. In strain TA2.A1, the intracellularATP content of exponential-phase cells is in the range 0.5 to3.5 mM and this drops to below 0.1 mM in stationary-phase cells(41). A role for ${varepsilon}$ in modulating the affinity of the ßsubunits for Mg-nucleotide complexes has been reported for ECF₁(56). In ECF₁, ${varepsilon}$ alters the binding affinity of high-affinitycatalytic nucleotide binding site 1 for MgADP and MgATP. Theaffinity for MgADP and MgATP is 10-fold higher in an ${varepsilon}$ -repleteECF₁ compared to a ${varepsilon}$ -depleted form of the enzyme (56). The authorssuggest that a potential explanation for this difference couldbe the reduced off rate for the Mg-nucleotide complex in thepresence of ${varepsilon}$ and hypothesize that the ${varepsilon}$ subunit decelerates aconformational change step that is necessary to release ligandsfrom the high-affinity binding site.

Proteolytic digestion of TA2F₁ complexes. The above results indicate that alanine substitutions in theC-terminal arm of the ${varepsilon}$ subunit prevent inhibition of ATPaseactivity in TA2F₁. To determine if these alanine substitutionsinduced a change in the conformation of the C-terminal arm,we studied the effect of protease treatment on TA2F₁, TA2F₁( ${varepsilon}$ ^6A),and TA2F₁F_o (Fig. 5). When TA2F₁F_o was treated either with chymotrypsinor trypsin, low ATP hydrolysis activities were observed (i.e.,0.32 and 0.25 units mg protein^–1, respectively), indicatingthat the enzyme was still blocked in ATP hydrolysis activity.SDS-PAGE revealed some proteolytic digestion products in boththe chymotrypsin- and trypsin-treated TA2F₁F_o enzyme when comparedto the untreated control; however, the ${varepsilon}$ subunit was unaffected(Fig. 5A). Extensive proteolytic digestion was noted in TA2F₁( ${varepsilon}$ ^WT),but no proteolysis of the ${varepsilon}$ subunit was discernible (Fig. 5B,lanes 2 and 3), and the ATPase activity of the protease-digestedTA2F₁( ${varepsilon}$ ^WT) remained low (i.e., 0.58 to 0.9 units mg protein^–1).In contrast, the ${varepsilon}$ subunit in TA2F₁( ${varepsilon}$ ^6A) was completely degradedby trypsin (Fig. 5C, lane 3) but not by chymotrypsin (Fig. 5C,lane 2). In neither case did protease treatment further activateATP hydrolysis in TA2F₁( ${varepsilon}$ ^6A) (2.8 to 2.92 units mg protein^–1)compared with the untreated control (3.8 units mg protein^–1).These data imply that the conformation of the ${varepsilon}$ subunit in bothTA2F₁ and TA2F₁F_o is protected by proteolytic enzymes comparedto TA2F₁( ${varepsilon}$ ^6A), where ${varepsilon}$ is degraded. Since no further activationof ATP hydrolysis in TA2F₁( ${varepsilon}$ ^6A) was observed, despite extensiveproteolytic digestion, the latent ATPase activity of the enzymeseems to be controlled by the ${varepsilon}$ subunit, at least in terms ofprotein-protein interactions. The conformation of ${varepsilon}$ in strainTA2F₁ would appear to be different to that in ECF₁ since, undersimilar experimental conditions, trypsin was shown to activateATPase activity by the removal of ${varepsilon}$ from the ECF₁ complex (32,49). High-resolution structural data have indicated that the ${varepsilon}$ subunit can exist in two conformations with different arrangementsof the two ${alpha}$ -helices. In the "up-state form" the two ${alpha}$ -helicesextend into the catalytic head of the F₁. In the "down-state"or retracted form the helices lie in a plane parallel to theoligomeric c ring (10, 43, 52). A third conformation, wherethe ${alpha}$ -helices are fully extended, has been suggested by cross-linkingstudies (51). Biochemical data have accumulated that indicatethat the ${varepsilon}$ subunit adopts the extended form in the absence ofadded nucleotide or in the presence of ADP, in which ATPaseactivity is inhibited, and that ATP counteracts ADP by favoringthe retracted form through the binding of ATP to the catalyticß subunit, which is a noninhibitory form (19, 51).The blockage in ATP hydrolysis activity of the TA2F₁F_o-ATP synthase,even under conditions of increasing ATP, suggests that underphysiological conditions the enzyme never works in the ATPasedirection, and thus it is tempting to speculate that the ${varepsilon}$ subunit ${alpha}$ -helices are fully extended, resulting in ATP hydrolysis inhibitionand protection from trypsin.

View larger version (49K):
[in this window]
[in a new window]
FIG. 5. Effect of protease treatment on recombinant TA2F₁ complexes and TA2F₁F_o-ATP synthase. TA2F₁F_o (A), TA2F₁( ${varepsilon}$ ^WT) (B), and TA2F₁( ${varepsilon}$ ^6A) (C) complexes were treated with chymotrypsin (lanes 2) and trypsin (lanes 3) as described in Materials and Methods, and samples containing 50 µg of protein were electrophoresed on a 13% SDS-polyacrylamide gel. Lanes 1 contained the untreated subunit complexes.

Johnson and McCarty (23) have provided evidence that the chloroplastATP synthase undergoes a conformational shift during its activationby the ${Delta}$ µ_H⁺ in light conditions, which specifically inducesa change in the conformation of the C-terminal domain of the ${varepsilon}$ subunit from a protected to a solvent-exposed state. Studieson the orientation of the ${varepsilon}$ subunit within CF₁ demonstrate thatit shares a common structure and positioning within the ATPsynthase to that of the E. coli ${varepsilon}$ subunit (4, 22).

Proton pumping of reconstituted TA2F₁ complexes with F₁-stripped membranes. To determine what effect the alanine substitutions in ${varepsilon}$ had onthe entire TA2F₁F_o-ATP synthase, purified TA2F₁( ${varepsilon}$ ^WT) and TA2F₁( ${varepsilon}$ ^6A)were reconstituted with F₁-stripped native membranes of Bacillussp. strain TA2.A1. Native membranes exhibited approximately4.4 units mg protein^–1 of ATP hydrolysis activity in thepresence of 0.5% LDAO. After F₁ was stripped off the membranes,only 13% of the original LDAO-stimulated ATPase activity couldbe detected, indicating that the procedure removed up to 87%of the F₁ moiety (data not shown).

After reconstitution of TA2F₁( ${varepsilon}$ ^WT) to F_o membranes, proteinswere visualized by SDS-PAGE (Fig. 6). Even when stripped membraneswere loaded in excess onto the SDS-PAGE gel, no F₁ subunitswere visible (Fig. 6, lane 3). These are clearly visible innative membranes that are overloaded on the same gel (Fig. 6,lane 5). When TA2F₁( ${varepsilon}$ ^WT) was reconstituted with F₁-stripped F_omembranes [i.e., TA2F_oF₁( ${varepsilon}$ ^WT)], all F₁ subunits are clearly visible,indicating that reconstitution of F₁ to F_o had indeed takenplace (Fig. 6, lane 2).

View larger version (123K):
[in this window]
[in a new window]
FIG. 6. Analysis of reconstituted TA2F₁( ${varepsilon}$ ^WT) complex with F₁-stripped F_o membranes by 13% SDS-polyacrylamide gel electrophoresis. Lane 1, 50 µg of purified F₁( ${varepsilon}$ ^WT) complex; lane 2, 50 µg of reconstituted TA2F_oF₁( ${varepsilon}$ ^WT) membranes; lane 3, 75 µg of F₁-stripped membranes; lane 4, 50 µg of purified TA2F₁F_o-ATP synthase; lane 5, 50 µg of native membranes.

When proton transport was measured by the quenching of ACMAfluorescence in native membranes, only very low levels of protonpumping (fluorescence quenching) could be detected (Fig. 7A),which is consistent with the latent ATPase activity of thisenzyme in strain TA2.A1. The residual ATP-dependent proton pumpingin native membranes was dissipated by the addition of the protonophoreCCCP (Fig. 7A). Importantly, no proton pumping could be detectedin the stripped membranes (Fig. 7B). When proton pumping measurementswere performed with reconstituted TA2F_oF₁( ${varepsilon}$ ^WT), ACMA fluorescencewas quenched to some extent by the addition of ATP, and thiswas dissipated by CCCP addition (Fig. 7C). Identical experimentswere performed with TA2F_oF₁( ${varepsilon}$ ^6A) and TA2F_oF₁( ${varepsilon}$ ^C) membranes (Fig.7D and E). Under these conditions, proton pumping was fasterthan with TA2F_oF₁( ${varepsilon}$ ^WT) and the quenching of ACMA fluorescencewas again reversed by CCCP. While TA2F_oF₁( ${varepsilon}$ ^C) membranes showeda greater level of proton pumping when compared to TA2F_oF₁( ${varepsilon}$ ^WT),the level was lower than that of TA2F_oF₁( ${varepsilon}$ ^6A) despite both mutantforms having similar ATP hydrolysis activities. ATP-dependentproton pumping of TA2F_oF₁( ${varepsilon}$ ^WT), TA2F_oF₁( ${varepsilon}$ ^6A), and TA2F_oF₁( ${varepsilon}$ ^C) wassensitive to DCCD (N,N'-dicyclohexylcarbodiimide) (data notshown). ATPase activity was determined in TA2F_oF₁( ${varepsilon}$ ^WT), TA2F_oF₁( ${varepsilon}$ ^6A),and TA2F_oF₁( ${varepsilon}$ ^C), and the ATP hydrolysis rates at high ATP concentration(i.e., 2 mM) (Fig. 7F) agree with those observed in the correspondingTA2F₁ complexes (Fig. 4D).

Cipriano et al. (5) have put forward a model that proposes thatthe ${varepsilon}$ subunit plays an important role in keeping the ATP synthaseefficient by preventing uncoupled ATP hydrolysis. This is furthersupported by single-molecule experiments with F₁-ATPase fromthe thermophilic Bacillus sp. strain PS3 which demonstrate unequivocallythat the ${varepsilon}$ subunit is essential for coupling catalysis with mechanicalrotation in the ATP synthesis direction (44). Dunn and coworkersshowed that if the E. coli ATP synthase is blocked in rotationby the fusion of a 20-kDa protein to the N terminus of ${varepsilon}$ , ATPaseactivity is not eliminated (6). Moreover, the residual ATPaseactivity of this enzyme is doubled if the C-terminal domainis truncated. In the wild-type enzyme, truncation of the C-terminaldomain resulted in an increased rate of ATP hydrolysis activitybut a decreased rate of proton pumping, indicating uncoupledactivity (6). Rondelez et al. (44) report that ${varepsilon}$ is dispensablefor ATP hydrolysis activity using ${alpha}$ ₃ß₃ ${gamma}$ subcomplexeswhere the mechanochemical coupling efficiency was three ATPshydrolyzed per turn either in the presence or absence of ${varepsilon}$ . However,these studies did not measure proton-pumping activity, and thereforeit is not known whether ATP hydrolysis was coupled to protontranslocation. Our data suggest that an activation of ATPaseactivity in TA2F₁( ${varepsilon}$ ^6A) does correlate with increased proton pumping.However, in TA2F₁( ${varepsilon}$ ^C), we observed a 50% decrease in proton pumping,suggesting that, in this form, ATP hydrolysis was partiallyuncoupled from proton-pumping activity.

We have previously demonstrated that the F₁F_o-ATP synthase ofthe thermoalkaliphile Bacillus sp. strain TA2.A1 is blockedin ATP hydrolysis activity, and this is intrinsic to the F₁moiety (7). The results of the present study demonstrate thatthe ${varepsilon}$ subunit plays an important role in blocking ATP hydrolysisactivity of both TA2F₁ and TA2F₁F_o. This inhibition of ATPaseactivity was attributed to a number of basic residues locatedin the C-terminal arm of the ${varepsilon}$ subunit (viz., Arg¹¹⁶, His¹¹⁷,Lys¹¹⁸, Arg¹¹⁹, Arg¹²³, and Arg¹²⁷). As shown in Fig. 1, variousmembers of the genus Bacillus (alkaliphilic and nonalkaliphilic)show a higher proportion of basic residues in the C-terminalarm of the ${varepsilon}$ subunit when compared to E. coli. We hypothesizethat this serves an important regulatory function when cellsbecome either limited for oxygen (decrease in ${Delta}$ µ_H⁺) orgrow at high pH (low ${Delta}$ µ_H⁺ due to an inverted pH gradient).Notably, both conditions are prevalent in the lifestyle of alkaliphilicbacilli. Under these conditions, there is a need to overcomethe thermodynamic problem of uncontrolled ATPase activity, andkinetically these bacteria have solved this problem by blockingATP hydrolysis activity (7, 14, 15, 17). We have recently performedside-by-side experiments with inverted membrane vesicles ofall Bacillus species listed in Fig. 1 and report that they allshow various degrees of latent ATPase activity, but importantlyall were able to synthesize ATP at high rates (B. Seale, S.Keis, and G. M. Cook, unpublished data). For alkaliphilic bacteria,the blockage in the ATP hydrolysis direction, but not in theATP synthesis (proton import) direction, is further magnifiedby the requirement for intracellular pH homeostasis, both inthe retention and acquisition of protons to prevent intracellularalkalinization. The role of the ATP synthase in alkali adaptationhas been reported for E. coli, where the enzyme has been shownto be up-regulated at pH 8.7 as a mechanism of proton import(34). Conversely, systems involved in proton export were down-regulatedin E. coli at high pH (e.g., cytochrome o and NADH dehydrogenasesI and II).

The precise molecular mechanism whereby the enzyme can blockrotation in one direction but rotates in the ATP synthesis directionremains unknown but clearly involves the ${varepsilon}$ subunit and otherprotein partners. Notwithstanding this important role, the ${varepsilon}$ subunit alone is not sufficient for ATP synthase function athigh pH. Wang and coworkers have identified the presence ofamino acid motifs that are unique to the c and a subunits ofalkaliphilic bacilli (55). Studies by this group with the F₁F_o-ATPsynthase from the facultative alkaliphile B. pseudofirmus OF4have demonstrated that these alkaliphilic motifs are criticalfor growth on nonfermentable carbon sources and ATP synthesisat pH 10.5 but not pH 7.5 (55).

If the ATP synthase of strain TA2.A1 is specifically adaptedto work only in the ATP synthesis direction, then one couldspeculate that the C-terminal arm of ${varepsilon}$ is obligatory for growthunder oxidative phosphorylation conditions at high pH. To ourknowledge, the role of ${varepsilon}$ on the energetics of growth of E. colion nonfermentable carbon sources has not been assessed, andtherefore the physiological role of ${varepsilon}$ in vivo remains unknown.Future research will focus on developing a genetic system tomake in vivo mutations in the ${varepsilon}$ subunit of strain TA2.A1 andassess the effect on growth at high pH.

ACKNOWLEDGMENTS

We thank M. Yoshida at Tokyo Institute of Technology for providingus plasmid pTrc99A and for helpful discussions regarding reconstitutionstudies. We thank S. Dunn for critical reading of the manuscript.

This work and S.K. were supported by a Marsden grant from theRoyal Society of New Zealand and by the Swiss National ScienceFoundation and Research Commission of ETH Zürich. G.M.C.acknowledges the financial support of the EidgenössischeTechnische Hochschule during his sabbatical in Zürich.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Immunology, Otago School of Medical Sciences, University of Otago, P.O. Box 56, Dunedin, New Zealand. Phone: 64 3 4797722. Fax: 64 3 4798540. E-mail: greg.cook{at}stonebow.otago.ac.nz.

These authors contributed equally to the experimental work describedin this paper.

REFERENCES

Top