Excessive ATP hydrolysis in ischemic myocardium by mitochondrial F1F0-ATPase: effect of selective pharmacological inhibition of mitochondrial ATPase hydrolase activity

Mitochondrial F(1)F(0)-ATPase normally synthesizes ATP in the heart, but under ischemic conditions this enzyme paradoxically causes ATP hydrolysis. Nonselective inhibitors of this enzyme (aurovertin, oligomycin) inhibit ATP synthesis in normal tissue but also inhibit ATP hydrolysis in ischemic myocardium. We characterized the profile of aurovertin and oligomycin in ischemic and nonischemic rat myocardium and compared this with the profile of BMS-199264, which only inhibits F(1)F(0)-ATP hydrolase activity. In isolated rat hearts, aurovertin (1-10 microM) and oligomycin (10 microM), at concentrations inhibiting ATPase activity, reduced ATP concentration and contractile function in the nonischemic heart but significantly reduced the rate of ATP depletion during ischemia. They also inhibited recovery of reperfusion ATP and contractile function, consistent with nonselective F(1)F(0)-ATPase inhibitory activity, which suggests that upon reperfusion, the hydrolase activity switches back to ATP synthesis. BMS-199264 inhibits F(1)F(0) hydrolase activity in submitochondrial particles with no effect on ATP synthase activity. BMS-199264 (1-10 microM) conserved ATP in rat hearts during ischemia while having no effect on preischemic contractile function or ATP concentration. Reperfusion ATP levels were replenished faster and necrosis was reduced by BMS-199264. ATP hydrolase activity ex vivo was selectively inhibited by BMS-199264. Therefore, excessive ATP hydrolysis by F(1)F(0)-ATPase contributes to the decline in cardiac energy reserve during ischemia and selective inhibition of ATP hydrolase activity can protect ischemic myocardium.     

Improved purification for thermophilic F1F0 ATP synthase using n-dodecyl beta-D-maltoside

Here we report a fast, simple purification for thermophilic F1F0 ATP synthase (TF1F0) that utilizes a cocktail of stabilizing reagents and the detergent n-dodecyl beta-D-maltoside to yield enzyme with an ATPase activity of 41 micromol/min/mg, 2.5-fold higher than that previously reported. ATPase activity was 80% inhibited by the F0-reactive reagent dicyclohexylcarbodiimide, indicating that F1-F0 interactions were largely intact. To measure ATP-driven proton pumping activity, purified TF1F0 was incorporated into liposomes, and the ATP-induced change in internal pH was measured using the fluorescent probe pyranine. In the presence of valinomycin, a maximum ATP-driven deltapH of 0.8 units was obtained. To measure ATP synthesis activity, TF1F0 was incorporated into liposomes with the light-dependent proton pump bacteriorhodopsin. Proteoliposomes were illuminated to generate an electrochemical gradient, after which ADP and inorganic phosphate were added to initiate ATP synthesis. A steady state ATP synthesis activity of 490 nmol/min/mg was achieved after an initial approximately 30-min lag phase.     

Probing conformations of the beta subunit of F0F1-ATP synthase in catalysis

A subcomplex of F0F1-ATP synthase (F0F1), alpha3beta3gamma, was shown to undergo the conformation(s) during ATP hydrolysis in which two of the three beta subunits have the "Closed" conformation simultaneously (CC conformation) [S.P. Tsunoda, E. Muneyuki, T. Amano, M. Yoshida, H. Noji, Cross-linking of two beta subunits in the closed conformation in F1-ATPase, J. Biol. Chem. 274 (1999) 5701-5706]. This was examined by the inter-subunit disulfide cross-linking between two mutant beta(I386C)s that was formed readily only when the enzyme was in the CC conformation. Here, we adopted the same method for the holoenzyme F0F1 from Bacillus PS3 and found that the CC conformation was generated during ATP hydrolysis but barely during ATP synthesis. The experiments using F0F1 with the epsilon subunit lacking C-terminal helices further suggest that this difference is related to dynamic nature of the epsilon subunit and that ATP synthesis is accelerated when it takes the pathway involving the CC conformation.     

Mutagenesis Studies of the F1F0 ATP Synthase b Subunit Membrane Domain

A homodimer of b subunits constitutes the peripheral stalk linking the F₁ and F₀ sectors of the Escherichia coli ATP synthase. Each b subunit has a single-membrane domain. The constraints on the membrane domain have been studied by systematic mutagenesis. Replacement of a segment proximal to the cytoplasmic side of the membrane had minimal impact on F₁F₀ ATP synthase. However, multiple substitutions on the periplasmic side resulted in defects in assembly of the enzyme complex. These mutants had insufficient oxidative phosphorylation to support growth, and biochemical studies showed little F₁F₀ ATPase and no detectable ATP-driven proton pumping activity. Expression of the b _N2A,T6A,Q10A subunit was also oxidative phosphorylation deficient, but the b _N2A,T6A,Q10A protein was incorporated into an F₁F₀ complex. Single amino acid substitutions had minimal reductions in F₁F₀ ATP synthase function. The evidence suggests that the b subunit membrane domain has several sites of interaction contributing to assembly of F₀, and that these interactions are strongest on the periplasmic side of the bilayer.     

F1F0 ATP synthase subunit c is a substrate of the novel YidC pathway for membrane protein biogenesis

The Escherichia coli YidC protein belongs to the Oxa1 family of membrane proteins that have been suggested to facilitate the insertion and assembly of membrane proteins either in cooperation with the Sec translocase or as a separate entity. Recently, we have shown that depletion of YidC causes a specific defect in the functional assembly of F1F0 ATP synthase and cytochrome o oxidase. We now demonstrate that the insertion of in vitro-synthesized F1F0 ATP synthase subunit c (F0c) into inner membrane vesicles requires YidC. Insertion is independent of the proton motive force, and proteoliposomes containing only YidC catalyze the membrane insertion of F0c in its native transmembrane topology whereupon it assembles into large oligomers. Co-reconstituted SecYEG has no significant effect on the insertion efficiency. Remarkably, signal recognition particle and its membrane-bound receptor FtsY are not required for the membrane insertion of F0c. In conclusion, a novel membrane protein insertion pathway in E. coli is described in which YidC plays an exclusive role.     

Characterization of the relationship between ADP- and epsilon-induced inhibition in cyanobacterial F1-ATPase

The ATPase activity of chloroplast and bacterial F(1)-ATPase is strongly inhibited by both the endogenous inhibitor ε and tightly bound ADP. Although the physiological significance of these inhibitory mechanisms is not very well known for the membrane-bound F(0)F(1), these are very likely to be important in avoiding the futile ATP hydrolysis reaction and ensuring efficient ATP synthesis in vivo. In a previous study using the α(3)β(3)γ complex of F(1) obtained from the thermophilic cyanobacteria, Thermosynechococcus elongatus BP-1, we succeeded in determining the discrete stop position, ～80° forward from the pause position for ATP binding, caused by ε-induced inhibition (ε-inhibition) during γ rotation (Konno, H., Murakami-Fuse, T., Fujii, F., Koyama, F., Ueoka-Nakanishi, H., Pack, C. G., Kinjo, M., and Hisabori, T. (2006) EMBO J. 25, 4596-4604). Because γ in ADP-inhibited F(1) also pauses at the same position, ADP-induced inhibition (ADP-inhibition) was assumed to be linked to ε-inhibition. However, ADP-inhibition and ε-inhibition should be independent phenomena from each other because the ATPase core complex, α(3)β(3)γ, also lapses into the ADP-inhibition state. By way of thorough biophysical and biochemical analyses, we determined that the ε subunit inhibition mechanism does not directly correlate with ADP-inhibition. We suggest here that the cyanobacterial ATP synthase ε subunit carries out an important regulatory role in acting as an independent "braking system" for the physiologically unfavorable ATP hydrolysis reaction.     

Effects of mutations in the beta subunit hinge domain on ATP synthase F1 sector rotation: interaction between Ser 174 and Ile 163

A complex of γ, ε, and c subunits rotates in ATP synthase (F_oF₁) coupling with proton transport. Replacement of βSer174 by Phe in β-sheet4 of the β subunit (βS174F) caused slow γ subunit revolution of the F₁ sector, consistent with the decreased ATPase activity [M. Nakanishi-Matsui, S. Kashiwagi, T. Ubukata, A. Iwamoto-Kihara, Y. Wada, M. Futai, Rotational catalysis of Escherichia coli ATP synthase F1 sector. Stochastic fluctuation and a key domain of the β subunit, J. Biol. Chem. 282 (2007) 20698-20704]. Modeling of the domain including β-sheet4 and α-helixB predicted that the mutant βPhe174 residue undergoes strong and weak hydrophobic interactions with βIle163 and βIle166, respectively. Supporting this prediction, the replacement of βIle163 in α-helixB by Ala partially suppressed the βS174F mutation: in the double mutant, the revolution speed and ATPase activity recovered to about half of the levels in the wild-type. Replacement of βIle166 by Ala lowered the revolution speed and ATPase activity to the same levels as in βS174F. Consistent with the weak hydrophobic interaction, βIle166 to Ala mutation did not suppress βS174F. Importance of the hinge domain [phosphate-binding loop (P-loop)/α-helixB/loop/β-sheet4, βPhe148-βGly186] as to driving rotational catalysis is discussed.     

Myocardial ischemic preconditioning and mitochondrial F1F0-ATPase activity

A short period of ischemia followed by reperfusion (ischemic preconditioning) is known to trigger mechanisms that contribute to the prevention of ATP depletion. In ischemic conditions, most of the ATP hydrolysis can be attributed to mitochondrial F₁F₀-ATPase (ATP synthase). The purpose of the present study was to examine the effect of myocardial ischemic preconditioning on the kinetics of ATP hydrolysis by F₁F₀-ATPase. Preconditioning was accomplished by three 3-min periods of global ischemia separated by 3 min of reperfusion. Steady state ATP hydrolysis rates in both control and preconditioned mitochondria were not significantly different. This suggests that a large influence of the enzyme on the preconditioning mechanism may be excluded. However, the time required by the reaction to reach the steady state rate was increased in the preconditioned group before sustained ischemia, and it was even more enhanced in the first 5 min of reperfusion (101 ± 3.0 sec in preconditioned vs. 83.4 ± 4.4 sec in controls, p 0.05). These results suggest that this transient increase in activation time may contribute to the cardioprotection by slowing the ATP depletion in the very critical early phase of post-ischemic reperfusion.     

Modulation at a Distance of Proton Conductance through the Saccharomyces cerevisiae Mitochondrial F1F0-ATP Synthase by Variants of the Oligomycin Sensitivity-Conferring Protein Containing Substitutions near the C-Terminus

We have sought to elucidate how the oligomycin sensitivity-conferring protein (OSCP) of the mitochondrial F₁F₀-ATP synthase (mtATPase) can influence proton channel function. Variants of OSCP, from the yeast Saccharomyces cerevisiae, having amino acid substitutions at a strictly conserved residue (Gly166) were expressed in place of normal OSCP. Cells expressing the OSCP variants were able to grow on nonfermentable substrates, albeit with some increase in generation time. Moreover, these strains exhibited increased sensitivity to oligomycin, suggestive of modification in functional interactions between the F₁ and F₀ sectors mediated by OSCP. Bioenergetic analysis of mitochondria from cells expressing OSCP variants indicated an increased respiratory rate under conditions of no net ATP synthesis. Using specific inhibitors of mtATPase, in conjunction with measurement of changes in mitochondrial transmembrane potential, it was revealed that this increased respiratory rate was a result of increased proton flux through the F₀ sector. This proton conductance, which is not coupled to phosphorylation, is exquisitely sensitive to inhibition by oligomycin. Nevertheless, the oxidative phosphorylation capacity of these mitochondria from cells expressing OSCP variants was no different to that of the control. These results suggest that the incorporation of OSCP variants into functional ATP synthase complexes can display effects in the control of proton flux through the F₀ sector, most likely mediated through altered protein—protein contacts within the enzyme complex. This conclusion is supported by data indicating impaired stability of solubilized mtATPase complexes that is not, however, reflected in the assembly of functional enzyme complexes in vivo. Given a location for OSCP atop the F₁-₃₃ hexamer that is distant from the proton channel, then the modulation of proton flux by OSCP must occur at a distance. We consider how subtle conformational changes in OSCP may be transmitted to F₀.     

Hypothyroidism Leads to a Decreased Expression of Mitochondrial F0F1-ATP Synthase in Rat Liver

In liver mitochondria isolated from hypothyroid rats, the rate of ATP synthesis is lower than in mitochondria from normal rats. Oligomycin-sensitive ATP hydrolase activity and passive proton permeability were significantly lower in submitochondrial particles from hypothyroid rats compared to those isolated from normal rats. In mitochondria from hypothyroid rats, the changes in catalytic activities of F₀F₁-ATP synthase are accompanied by a decrease in the amount of immunodetected -F₁, F₀1-PVP, and OSCP subunits of the complex. Northern blot hybridization shows a decrease in the relative cytosolic content of mRNA for -F₁ subunit in liver of hypothyroid rats. Administration of 3,5,3-triodo-L-thyronine to the hypothyroid rats tends to remedy the functional and structural defects of F₀F₁-ATP synthase observed in the hypothyroid rats. The results obtained indicate that hypothyroidism leads to a decreased expression of F₀F₁-ATP synthase complex in liver mitochondria and this contributes to the decrease of the efficiency of oxidative phosphorylation.     

The native F0F1-inhibitor protein complex from beef heart mitochondria and its reconstitution in liposomes

A functional F₀F₁ ATP synthase that contains the endogenous inhibitor protein (F₀F₁I) was isolated by the use of two combined techniques [Adolfsen, R., McClung, J.A., and Moudrianakis, E. N. (1975).Biochemistry 14, 1727–1735; Dreyfus, G., Celis, H., and Ramirez, J. (1984).Anal. Biochem. 142, 215–220]. The preparation is composed of 18 subunits as judged by SDS-PAGE. A steady-state kinetic analysis of the latent ATP synthase complex at various concentrations of ATP showed aV _max of 1.28mol min^–1 mg^–1, whereas theV _max of the complex without the inhibitor was 8.3mol min^–1 mg^–1. In contrast, theK _m for Mg-ATP of F₀F₁ I was 148M, comparable to theK _m value of 142M of the F₀F₁ complex devoid of IF₁. The hydrolytic activity of the F₀F₁I increased severalfold by incubation at 60C at pH 6.8, reaching a maximal ATPase activity of 9.5mol min^–1 mg^–1; at pH 9.0 a rapid increase in the specific activity of hydrolysis was followed by a sharp drop in activity. The latent ATP synthase was reconstituted into liposomes by means of a column filtration method. The proteoliposomes showed ATP-Pi exchange activity which responded to phosphate concentration and was sensitive to energy transfer inhibitors like oligomycin and the uncouplerp-trifluoromethoxyphenylhydrazone.     

Role of alphaPhe-291 residue in the phosphate-binding subdomain of catalytic sites of Escherichia coli ATP synthase

The role of αPhe-291 residue in phosphate binding by Escherichia coli F₁F₀-ATP synthase was examined. X-ray structures of bovine mitochondrial enzyme suggest that this residue resides in close proximity to the conserved βR246 residue. Herein, we show that mutations αF291D and αF291E in E. coli reduce the ATPase activity of F₁F₀ membranes by 350-fold. Yet, significant oxidative phosphorylation activity is retained. In contrast to wild-type, ATPase activities of mutants were not inhibited by MgADP-azide, MgADP-fluoroaluminate, or MgADP-fluoroscandium. Whereas, 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (NBD-Cl) inhibited wild-type ATPase essentially completely, ATPase in mutants was inhibited maximally by ∼75%, although reaction still occurred at residue βTyr-297, proximal to αPhe-291 in the phosphate-binding pocket. Inhibition characteristics supported the conclusion that NBD-Cl reacts in βE (empty) catalytic sites, as shown previously by X-ray structure analysis. Phosphate protected against NBD-Cl inhibition in wild-type but not in mutants. In addition, our data suggest that the interaction of αPhe-291 with phosphate during ATP hydrolysis or synthesis may be distinct.     

Activation and Deactivation of F0F1-ATPase in Yeast Mitochrondia

The regulation of membrane-bound proton F₀F₁ATPase by the protonmotive force and nucleotides was studied in yeastmitochondria. Activation occurred in whole mitochondria and the ATPaseactivity was measured just after disrupting the membranes with Triton X-100.Deactivation occurred either in whole mitochondria uncoupled with FCCP, or indisrupted membranes. No effect of Triton X-100 on the ATPase was observed,except a slow reactivation observed only in the absence of MgADP. BothAMPPNP and ATP increased the ATPase deactivation rate, thus indicating thatoccupancy of nucleotidic sites by ATP is more decisive than catalyticturnover for this process. ADP was found to stimulate the energy-dependentATPase activation. ATPase deactivated at the same rate in uncoupled anddisrupted mitochondria. This suggests that deactivation is not controlled byrebinding of some soluble factor, like IF1, but rather by the conversion ofthe F₁.IF1 complex into an inactive form.     

F0F1 ATP synthase of Streptomycetes: Modulation of activity and oligomycin resistance by protein Ser/Thr kinases

Inverted membrane vesicles of Gram-positive actinobacteria Streptomyces fradiae, S. lividans, and S. avermitilis have been prepared and membrane-bound F₀F₁ ATP synthase has been biochemically characterized. It has been shown that the ATPase activity of membrane-bound F₀F₁ complex is Mg²⁺-dependent and moderately stimulated by high concentrations of Ca²⁺ ions (10–20 mM). The ATPase activity is inhibited by N,N′-dicyclohexylcarbodiimide and oligomycin A, typical F₀F₁ ATPase inhibitors that react with the membrane-bound F₀ complex. The assay of biochemical properties of the F₀F₁ ATPases of Streptomycetes in all cases showed the presence of ATPase populations highly susceptible and insensitive to oligomycin A. The in vitro labeling and inhibitory assay showed that the inverted phospholipid vesicles of S. fradiae contained active membrane-bound Ser/Thr protein kinase(s) phosphorylating the proteins of the F₀F₁ complex. Inhibition of phosphorylation leads to decrease of the ATPase activity and increase of its susceptibility to oligomycin. The in vivo assay confirmed the enhancement of actinobacteria cell sensitivity to oligomycin after inhibition of endogenous phosphorylation. The sequencing of the S. fradiae genes encoding oligomycin-binding A and C subunits of F₀F₁ ATP synthase revealed their close phylogenetic relation to the genes of S. lividans and S. avermitilis.     

Coupling Proton Movement to ATP Synthesis in the Chloroplast ATP Synthase

The chloroplast F₀F₁-ATP synthase-ATPase is a tiny rotary motor responsible for coupling ATP synthesis and hydrolysis to the light-driven electrochemical proton gradient. Reversible oxidation/reduction of a dithiol, located within a special regulatory domain of the γ subunit of the chloroplast F₁ enzyme, switches the enzyme between an inactive and an active state. This regulatory mechanism is unique to the ATP synthases of higher plants and its physiological significance lies in preventing nonproductive depletion of essential ATP pools in the dark. The three-dimensional structure of the chloroplast F₁ gamma subunit has not yet been solved. To examine the mechanism of dithiol regulation, a model of the chloroplast gamma subunit was obtained through segmental homology modeling based on the known structures of the mitochondrial and bacterial γ subunits, together with de novo construction of the unknown regulatory domain. The model has provided considerable insight into how the dithiol might modulate catalytic function. This has, in turn, suggested a mechanism by which rotation of subunits in F₀, the transmembrane proton channel portion of the enzyme, can be coupled, via the ε subunit, to rotation of the γ subunit of F₁ to achieve the 120° (or 90°+30°) stepping action that is characteristic of F₁ γ subunit rotation.     

Relevance of Divalent Cations to ATP-Driven Proton Pumping in Beef Heart Mitochondrial F0F1-ATPase

The ATP hydrolysis rate and the ATP hydrolysis-linked proton translocation by the F₀F₁-ATPase of beef heart submitochondrial particles were examined in the presence of several divalent metal cations. All Me–ATP complexes tested sustained ATP hydrolysis, although to a different extent. However, only Mg- and Mn-ATP-dependent hydrolysis could sustain a high level of proton pumping activity, as determined by acridine fluorescence quenching. Moreover, the K _m of the Me-ATP hydrolysis-induced proton pumping activity was very similar to the K _m value of Me-ATP hydrolysis. Both oligomycin and DCCD caused the full recovery of the fluorescence, providing clear evidence for the association of Mg-ATP hydrolysis with proton translocation through the F₀F₁-ATPase complex. In contrast, with other Me-ATP complexes, including Ca-ATP as substrate, the proton pumping activity was undetectable, implicating an uncoupling nature for these substrates. Attempts to demonstrate the involvement of the subunit of the enzyme in the coupling mechanism failed, suggesting that the participation of at least the N-terminal segment of the subunit in the coupling mechanism of the mitochondrial enzyme is unlikely.     

Modulation of nucleotide specificity of thermophilic F(o)F(1)-ATP Synthase by epsilon-subunit

The C-terminal two α-helices of the ε-subunit of thermophilic Bacillus F(o)F(1)-ATP synthase (TF(o)F(1)) 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 TF(o)F(1) expressed in Escherichia coli, we compared TF(o)F(1) with up- and down-state ε in the NTP (ATP, GTP, UTP, and CTP) synthesis reactions. TF(o)F(1) with the up-state ε was achieved by inclusion of hexokinase in the assay and TF(o)F(1) with the down-state ε was represented by εΔc-TF(o)F(1), in which ε lacks C-terminal helices and hence cannot adopt the up-state under any conditions. The results indicate that TF(o)F(1) with the down-state ε synthesizes GTP at the same rate of ATP, whereas TF(o)F(1) with the up-state ε synthesizes GTP at a half-rate. Though rates are slow, TF(o)F(1) with the down-state ε even catalyzes UTP and CTP synthesis. Authentic TF(o)F(1) 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-TF(o)F(1) suggests similar modulation of nucleotide specificity in NTP hydrolysis. Thus, depending on its conformation, ε-subunit modulates substrate specificity of TF(o)F(1).     

Met23Lys mutation in subunit gamma of FOF1-ATP synthase from Rhodobacter capsulatus impairs the activation of ATP hydrolysis by protonmotive force

H⁺-F_OF₁-ATP synthase couples proton flow through its membrane portion, F_O, to the synthesis of ATP in its headpiece, F₁. Upon reversal of the reaction the enzyme functions as a proton pumping ATPase. Even in the simplest bacterial enzyme the ATPase activity is regulated by several mechanisms, involving inhibition by MgADP, conformational transitions of the ε subunit, and activation by protonmotive force. Here we report that the Met23Lys mutation in the γ subunit of the Rhodobacter capsulatus ATP synthase significantly impaired the activation of ATP hydrolysis by protonmotive force. The impairment in the mutant was due to faster enzyme deactivation that was particularly evident at low ATP/ADP ratio. We suggest that the electrostatic interaction of the introduced γLys23 with the DELSEED region of subunit β stabilized the ADP-inhibited state of the enzyme by hindering the rotation of subunit γ rotation which is necessary for the activation.     

Properties of Bound Inorganic Phosphate on Bovine Mitochondrial F1F0-ATP Synthase

Beef-heart mitochondrial F₁F₀-ATP synthase contained six molecules of bound inorganic phosphate (P_i). This phosphate exchanged completely with exogenous ³²P_i when the enzyme was exposed to 30% (v/v) dimethyl sulfoxide (DMSO) and then returned to a DMSO-free buffer (Beharry and Bragg 2001). Only two molecules were replaced by ³²P_i when the enzyme was not pretreated with DMSO. These two molecules of ³²P_i were not displaced from the enzyme by the treatment with 1 mM ATP. Similarly, two molecules of bound ³²P_i remained on the DMSO-pretreated enzyme following addition of ATP, that is, four molecules of ³²P_i were displaced by ATP. The ATP-resistant ³²P_i was removed from the enzyme by pyrophosphate. It is proposed that these molecules of ³²P_i are bound at an unfilled adenine nucleotide-binding noncatalytic site on the enzyme. Brief exposure of the enzyme loaded with two molecules of ³²P_i to DMSO, followed by removal of the DMSO, resulted in the loss of the bound ³²P_i and in the formation of two molecules of bound ATP from exogenous ADP. A third catalytic site on the enzyme was occupied by ATP, which could undergo a P_i ATP exchange reaction with bound P_i The presence of two catalytic sites containing bound P_i is consistent with the X-ray crystallographic structure of F₁ (Bianchet, et al., 1998). Thus, five of the six molecules of bound P_i were accounted for. Three molecules of bound P_i were at catalytic sites and participated in ATP synthesis or P_i ATP exchange. Two other molecules of bound P_i were present at a noncatalytic adenine nucleotide-binding site. The location and role of the remaining molecule of bound P_i remains to be established. We were unable to demonstrate, using chemical modification of sulfhydryl groups by iodoacetic acid, any gross difference in the conformation of F₁F₀ in DMSO-containing compared with DMSO-free buffers.