Structure of F1-ATPases

F₁-ATPases are large multimeric proteins that can be isolated from the membrane bound system that catalyzes the phosphorylation of ADP by inorganic phosphate in bacteria, plants, and mitochondria. They can be visualized in electron micrographs of the inner mitochondrial membranes where they appear as large protruding spheres 90 Å in diameter. The purified F₁-ATPases have a molecular weight of 320,000 to 400,000 daltons and are composed of five non-identical subunits (, , , and ). The stoichiometry of these subunits in the complex is still unknown but compositions of the type ₃₃ and ₂₂₂₂₂ were found to be consistent with some of the available experimental data. This review discusses the recent data and the experimental approaches utilized for the structural characterization of F₁-ATPases.     

The Structural and Functional Connection between the Catalytic and Proton Translocating Sectors of the Mitochondrial F 1 F 0 -ATP Synthase

The structural and functional connection between the peripheral catalytic F₁ sector and theproton-translocating membrane sector F₀ of the mitochondrial ATP synthase is reviewed. Theobservations examined show that the N-terminus of subunit , the carboxy-terminal and centralregion of F₀I-PVP(b), OSCP, and part of subunit d constitute a continuous structure, the lateralstalk, which connects the peripheries of F₁ to F₀ and surrounds the central element of thestalk, constituted by subunits and . The ATPase inhibitor protein (IF₁) binds at one sideof the F₁F₀ connection. The carboxy-terminal segment of IF₁ apparently binds to OSCP. The42L-58K segment of IF₁, which is per se the most active domain of the protein, binds at thesurface of one of the three / pairs of F₁, thus preventing the cyclic interconversion of thecatalytic sites required for ATP hydrolysis.     

ATP Synthases in the Year 2000: Evolving Views about the Structures of These Remarkable Enzyme Complexes

This introductory article briefly summarizes how our views about the structural features ofATP synthases (F₀F₁) have evolved over the past 30 years and also reviews some of our currentviews in the year 2000 about the structures of these remarkably unique enzyme complexes.Suffice it to say that as we approach the end of the first year of this new millinium, we canbe conservatively confident that we have a reasonably good grasp of the overall low-resolutionstructural features of ATP synthases. Electron microscopy techniques, combined with the toolsof biochemistry, molecular biology, and immunology, have played the leading role here byidentifying the headpiece, basepiece, central stalk, side stalk, cap, and in the mitochondrialenzyme, the collar around the central stalk. We can be reasonably confident also that we havea fairly good grasp of much of the high-resolution structural features of both the F₁ moietycomprised of fives subunit types (, , , , and ) and parts of the F₀ moiety comprised ofeither three (E. coli) or at least ten (mitochondria) subunit types. This information acquiredin several different laboratories, either by X-ray crystallography or NMR spectroscopy, includesdetails about the active site and subunit relationships. Moreover, it is consistent with recentlyreported data that the F₁ moiety may be an ATP driven motor, which, during ATP synthesis,is driven in reverse by the electrochemical proton gradient generated by the electron transportchain. The real structural challenges of the future are to acquire at high resolution completeATP synthase complexes representative of different stages of the catalytic cycle during ATPsynthesis and representative also of key regulatory states.     

Identification of subunits required for the catalytic activity of the F1-ATPase

F₁() complexes containing equimolar ratios of the and subunits have been shown to function as active ATPases, whereas individually isolated and subunits show no real ATPase activity. These results indicate that the single-copy subunits are not required for F₁-ATPase activity. The minimal F₁()-core complexes exhibit, however, lower rates and some different properties from those of their parent whole F₁ or ₃₃ complexes. It is therefore concluded that for obtaining a full spectrum of the characteristic functional properties of an F₁-ATPase the presence of the F₁- subunit is also required. The implications of these findings on the subunit location of both catalytic and noncatalytic nucleotide binding sites is discussed.     

Evolutionary relationship between Enterobacteriaceae: comparison of the ATP synthases (F1F0) of Escherichia coli and Klebsiella pneumoniae

The ATP synthase complex of Klebsiella pneumoniae (KF₁F₀) has been purified and characterized. SDS-gel electrophoresis of the purified F₁F₀ complexes revealed an identical subunit pattern for E. coli (EF₁F₀) and K. pneumoniae. Antibodies raised against EF₁ complex and purified EF₀ subunits recognized the corresponding polypeptides of EF₁F₀ and KF₁F₀ in immunoblot analysis. Protease digestion of the individual subunits generated an identical cleavage pattern for subunits , , , , a, and c of both enzymes. Only for subunit different cleavage products were obtained. The isolated subunit c of both organisms showed only a slight deviation in the amino acid composition. These data suggest that extensive homologies exist in primary and secondary structure of both ATP synthase complexes reflecting a close phylogenetic relationship between the two enterobacteric tribes.Abbreviations ACMA 9-amino-6-chloro-2-methoxyacridine - DCCD N,N-dicyclohexylcarbodiimide - FITC fluorescein isothiocyanate - SDS sodium dodecyl sulfate - TTFB 4,5,6,7-tetrachloro-2-trifluoromethylbenzimidazole     

Structural and Functional Features of the Escherichia coli F 1 -ATPase

The structural organization and overall dimensions of the Escherichia coli F₁-ATPase in solutionhas been analyzed by synchroton X-ray scattering. Using an independent ab initio approach,the low-resolution shape of the hydrated enzyme was determined at 3.2 nm resolution. Theshape permitted unequivocal identification of the volume occupied by the _{3 3} complex ofthe atomic model of the ECF₁-ATPase. The position of the ^ and subunits were found byinteractive fitting of the solution scattering data and by cross-linking studies. Laser-inducedcovalent incorporation of 2-azido-ATP established a direct relationship between nucleotidebinding affinity and the different interactions between the stalk subunits and with the threecatalytic subunits () of the F₁-ATPase. Mutants of the ECF₁-ATPase with the introductionof Trp-for-Tyr replacement in the catalytic site of the complex made it possible to monitorthe activated state for ATP synthesis (ATP conformation) in which the and subunits arein close proximity to the subunits and the ADP conformation, with the stalk subunits arelinked to the subunit.     

Unisite Hydrolysis of [γ32P]ATP by Soluble Mitochondrial F1ATPase and Its Release by Excess ADP and ATP. Effect of Trifluoperazine

Some of the characteristics of unisite hydrolysis of [³²P]ATP as well as the changes that occur on the transition to multisite catalysis were further studied. It was found that a fraction of [³²P]ATP bound at the catalytic sites of F₁ under unisite conditions undergoes both hydrolysis and release induced by medium nucleotides upon addition of millimolar concentrations of ADP or ATP. The fraction of [³²P]ATP that undergoes release is similar to the fraction that undergoes hydrolytic cleavage, indicating that the rates of the release and hydrolytic reactions of bound [³²P]ATP are in the same range. As part of studies on the mechanisms through which trifluoperazine inhibits ATP hydrolysis, its effect on unisite hydrolysis of [³²P]ATP was also studied. Trifluoperazine diminishes the rate of unisite hydrolysis by 30–40%. The inhibition is accompanied by a nearly tenfold increase in the ratio of [³²P]ATP/³²Pi bound at the catalytic site and a 50% diminution in the rate of ³²Pi release from the enzyme into the media. Trifluoperazine also induces heterogeneity of the three catalytic sites of F₁ in the sense that in a fraction of F₁ molecules, the high-affinity catalytic site has a turnover rate lower than the other two. Trifluoperazine does not modify the release of previously bound [³²P]ATP induced by medium nucleotides. The latter indicates that hindrances in the release of Pi do not necesarily accompany alterations in the release of ATP even though both species lie in the same site.     

Structure of theEscherichia coli ATP synthase and role of the γ and ε subunits in coupling catalytic site and proton channeling functions

The structure of theEscherichia coli ATP synthase has been studied by electron microscopy and a model developed in which the and subunits of the F₁ part are arranged hexagonally (in top view) alternating with one another and surrounding a central cavity of around 35 Å at its widest point. The and subunits are interdigitated in side view for around 60 Å of the 90 Å length of the molecule. The F₁ narrows and has three-fold symmetry at the end furthest from the F₀ part. The F₁ is linked to F₀ by a stalk approximately 45 Å long and 25–30 Å in diameter. The F₀ part is mostly buried in the lipid bilayer. The subunit provides a domain that extends into the central cavity of the F₁ part. The and subunits are in a different conformation when ATP+Mg²⁺ are present in catalytic sites than when ATP+EDTA are present. This is consistent with these two small subunits switching conformations as a function of whether or not phosphate is bound to the enzyme at the position of the phosphate of ATP. We suggest that this switching is the key to the coupling of catalytic site events with proton translocation in the F₀ part of the complex.     

ATP synthases: insights into their motor functions from sequence and structural analyses

ATP synthases are motor complexes comprised of F₀ and F₁ parts that couple the proton gradient across the membrane to the synthesis of ATP by rotary catalysis. Although a great deal of information has been accumulated regarding the structure and function of ATP synthases, their motor functions are not fully understood. For this reason, we performed the alignments and analyses of the protein sequences comprising the core of the ATP synthase motor complex, and examined carefully the locations of the conserved residues in the subunit structures of ATP synthases. A summary of the findings from this bioinformatic study is as follows. First, we found that four conserved regions in the sequence of subunit are clustered into three patches in its structure. The interactions of these conserved patches with the and subunits are likely to be critical for energy coupling and catalytic activity of the ATP synthase. Second, we located a four-residue cluster at the N-terminal domain of mitochondrial OSCP or bacterial (or chloroplast) subunit which may be critical for the binding of these subunits to F₁. Third, from the localizations of conserved residues in the subunits comprising the rotors of ATP synthases, we suggest that the conserved interaction site at the interface of subunit c and (mitochondria) or (bacteria and chloroplasts) may be important for connecting the rotor of F₁ to the rotor of F₀. Finally, we found the sequence of mitochondrial subunit b to be highly conserved, significantly longer than bacterial subunit b, and to contain a shorter dimerization domain than that of the bacterial protein. It is suggested that the different properties of mitochondrial subunit b may be necessary for interaction with other proteins, e.g., the supernumerary subunits.     

Chemical modification of active sites in relation to the catalytic mechanism of F1

Recent studies of chemically modified F₁-ATPases have provided new information that requires a revision of our thinking on their catalytic mechanism. One of the subunits in F₁-ATPase is distinguishable from the other two both structurally and functionally. The catalytic site and regulatory site of the same subunit are probably sufficiently close to each other, and the interaction between the various catalytic and regulatory sites are probably sufficiently strong to raise the uni-site rate of ATP hydrolysis by several orders of magnitude to that of promoted (multi-site) ATP hydrolysis. Although all three subunits in F₁ possess weak uni-site ATPase activity, only one of them () catalyzes promoted ATP hydrolysis. But all three subunits catalyze ATP synthesis driven by the proton flux. Internal rotation of the ₃₃ or ₃ moiety relative to the remainder of the F₀F₁ complex did not occur during oxidative phosphorylation by reconstituted submitochondrial particles.     

ATP Synthases in the Year 2000: Defining the Different Levels of Mechanism and Getting a Grip on Each

ATP synthases are unusually complex molecules, which fractionate most readily into two major units, one a water soluble unit called F₁ and the other a detergent soluble unit called F₀. In almost all known species the F₁ unit consists of 5 subunit types in the stoichiometric ratio ₃₃ while the F₀ unit contains 3 subunit types (a, b, and c) in E. coli, and at least 10 subunit types (a, b, c, and others) in higher animals. It is now believed by many investigators that during the synthesis of ATP, protons derived from an electrochemical gradient generated by an electron transport chain are directed through the F₀ unit in such a way as to drive the rotation of the single subunit, which extends from an oligomeric ring of at least 10 c subunits in F₀ through the center of F₁. It is further believed by many that the rotating subunit, by interacting sequentially with the 3 pairs of F₁ (360° cycle) in the presence of ADP, P_i, and Mg⁺⁺, brings about via power strokes conformational/binding changes in these subunits that promote the synthesis of ATP and its release on each pair. In support of these views, studies in several laboratories either suggest or demonstrate that F₀ consists in part of a proton gradient driven motor while F₁ consists of an ATP hydrolysis driven motor, and that the subunit does rotate during F₁ function. Therefore, current implications are that during ATP synthesis the former motor drives the latter in reverse via the subunit. This would suggest that the process of understanding the mechanism of ATP synthases can be subdivided into three major levels, which include elucidating those chemical and/or biophysical events involved in (1) inducing rotation of the subunit, (2) coupling rotation of this subunit to conformational/binding changes in each of the 3 pairs, and (3) forming ATP and water (from ADP, P_i, and Mg⁺⁺) and then releasing these products from each of the 3 catalytic sites. Significantly, it is at the final level of mechanism where the bond breaking/making events of ATP synthesis occur in the transition state, with the former two levels of mechanism setting the stage for this critical payoff event. Nevertheless, in order to get a better grip in this new century on how ATP synthases make ATP and then release it, we must take on the difficult challenge of elucidating each of the three levels of mechanism.     

The photosynthetic F1-α3β3 and α1β1 catalytic core complexes

Minimal photosynthetic catalytic F₁() core complexes, containing equimolar ratios of the and subunits, were isolated from membrane-bound spinach chloroplast CF₁ and Rhodospirillum rubrum chromatophore RrF₁. A CF₁-₃₃ hexamer and RrF₁-₁₁ dimer, which were purified from the respective F₁() complexes, exhibit lower rates and different properties from their parent F₁-ATPases. Most interesting is their complete resistance to inhibition by the general F₁ inhibitor azide and the specific CF₁ inhibitor tentoxin. These inhibitors were earlier reported to inhibit multisite, but not unisite, catalysis in all sensitive F₁-ATPases and were therefore suggested to block catalytic site cooperativity. The absence of this typical property of all F₁-ATPases in the ₁₁ dimer is consistant with the view that the dimer contains only a single catalytic site. The ₃₃ hexamer contains however all F₁ catalytic sites. Therefore the observation that CF₁-₃₃ can bind tentoxin and is stimulated by it suggests that the F₁ subunit, which is required for obtaining inhibition by tentoxin as well as azide, plays an important role in the cooperative interactions between the F₁-catalytic sites.Abbreviations CF₀F₁ chloroplast F₀F₁ - CF₁ chloroplast F₁ - CF₁ chloroplast F₁ subunit - CF₁ chloroplast F₁ subunit - CF₁() a complex containing equal amounts of the CF₁ and subunits - MF₁ mitochondrial F₁ - RrF₀F₁ Rhodospirillum rubrum F₀F₁ - RrF₁ R. rubrum F₁ - RrF₁ R. rubrum F₁ subunit - RrF₁ R. rubrum F₁ subunit - RrF₁() a complex containing equal amounts of the RrF₁ and subunits - Rubisco Ribulose-1,5-bisphosphate carboxylase - TF₁ thermophilic bacterium PS3 F₁     

Topography of the subunits of Micrococcus lysodeikticus F1-ATPase

Summary The combined use of proteolytic digestion and lactoperoxidase catalyzed labelling with [¹²⁵I] applied to membrane-bound or soluble pure F₁-ATPase from Micrococcus lysodeikticus has allowed us to establish the topography of its , , and subunits within the protein molecule and with respect to the plane of the membrane.The subunit is most externally located to the membrane bilayer looking towards the cytoplasmic face, a position consistent with its proposed catalytic role. The and subunits lie in an intermediate layer between the subunits and the membrane, in which the subunit occupies a central position within the F₁-ATPase molecule in contact with the subunit. The subunit appears to be tightly bound to the F₀ component of the ATPase complex, probably buried in the membrane bilayer. A molecular arrangement of M. lysodeikticus ATPase is proposed that, taking into account the subunit stoichiometry _{3 3 2 2} (MW 420 000), accommodates the role assigned to each subunit and most, if not all, the known properties of this bacterial energy-transducing protein.     

The oligomycin sensitivity conferring protein (OSCP) of beef heart mitochondria: Studies of its binding to F1 and its function

The binding of oligomycin sensitivity conferring protein (OSCP) to soluble beef-heart mitochondrial ATPase (F₁) has been investigated. OSCP forms a stable complex with F₁, and the F₁ · OSCP complex is capable of restoring oligomycin- and DCCD-sensitive ATPase activity to F₁- and OSCP-depleted submitochondrial particles. The F₁ · OSCP complex retains 50% of its ATPase activity upon cold exposure while free F₁ is inactivated by 90% or more. Both free F₁ and the F₁ · OSCP complex release upon cold exposure a part—probably 1 out of 3—of their subunits; whether subunits are also lost is uncertain. The cold-treated F₁ · OSCP complex is still capable of restoring oligomycin- and DCCD-sensitive ATPase activity to F₁- and OSCP-depleted particles. OSCP also protects F₁ against modification of its subunit by mild trypsin treatment. This finding together with the earlier demonstration that trypsin-modified F₁ cannot bind OSCP indicates that OSCP binds to the subunit of F₁ and that F₁ contains three binding sites for OSCP. The results are discussed in relation to the possible role of OSCP in the interaction of F₁ with the membrane sector of the mitochondrial ATPase system.Abbreviations DCCD N,N-dicyclohexylcarbodiimide - OSCP oligomycin sensitivity conferring protein - SDS sodium dodecylsulfate This paper is dedicated to the memory of David E. Green—scholar, pioneer, visionary.     

Subunit movement during catalysis by F1-F0-ATP synthases

The catalytic portion (F₁) of ATP synthases have the subunit composition ₃, ₃, , , . This composition imparts structural asymmetry to the entire complex that results in differences in nucleotide binding affinity among the six binding sites. Evidence that two or more sites participate in catalysis, alternating their properties, led to the notion that the interactions of individual pairs with the small subunits must change as binding site properties alternate. A rotation of the subunit within the ₃₃ hexamer has been proposed as a means of alternating the properties of catalytic sites. Evidence argues that the rotation of the complete subunit during ATP hydrolysis is not mandatory for activity. The subunit of chloroplast F₁ may be cleaved into three large fragments that remain bound to F₁. This cleavage enhances ATPase activity without loss of evidence of site-site interactions. Complexes of ₃₃ have been shown to have significant ATPase activity in the absence of . Mg²⁺ATP affects the interaction of with the different subunits, and induces other changes in F₁, but whether these changes are induced by catalysis, or are fast enough to be involved in the catalytic turnover of the enzyme has not been established. Likewise, changes in structure and in binding site properties induced in thylakoid membrane bound CF₁ by formation of an electrochemical proton gradient may activate the enzyme rather than be apart of catalysis. Mechanisms other than rotary catalysis should be considered.     

Inhibitory Mg-ADP—Fluoroaluminate Complexes Bound to Catalytic Sites of F1-ATPases: Are They Ground-State or Transition-State Analogs?

Schemes are proposed for coupling sequential opening and closing the three catalytic sites of F₁ to rotation of the subunit during ATP synthesis and hydrolysis catalyzed by the F_oF₁-ATP synthase. A prominent feature of the proposed mechanisms is that the transition state during ATP synthesis is formed when a catalytic site is in the process of closing and that the transition state during ATP hydrolysis is formed when a catalytic site is in the process of opening. The unusual kinetics of formation of Mg-ADP—fluoroaluminate complexes in one or two catalytic sites of nucleotide-depleted MF₁ and wild-type and mutant ₃₃ subcomplexes of TF₁ are also reviewed. From these considerations, it is concluded that Mg-ADP—fluoroaluminate complexes formed at catalytic sites of isolated F₁-ATPases or F₁ in membrane-bound F_oF₁ are ground-state analogs.     

Substructure of F1-ATPase (BF1 factor) from Micrococcus lysodeikticus

Summary Dimethyl suberimidate and dithiobis (succinimidyl propionate) have been used to explore the nearest neighbor relationship of the subunits (, , and by decreasing molecular weight) of F₁-ATPase or BF₁ factor of Micrococcus lysodeikticus. Cross-linking with the two diimido esters inhibited the ATPase activity but this inhibition never exceeded 50% of the initial value. The cross-linking pattern of this BF₁ factor, as revealed by sodium dodecyl sulfate gel electrophoresis, shows a relative low proportion of high molecular weight aggregates which move slowly than the heaviest subunit (). They are resolved as three components of molecular weights 200,000, 130,000 and 100,000 in 5% acrylamide gels, plus an additional component (mol. wt 80,000) identified in 10% acrylamide gels. The other aggregate bands represent cross-linking products of the smaller subunits ( and ) that may travel to the conventional position of the heavier subunits.The subunit composition of the aggregate bands has been determined through the reversion of dithiobis (succinimidyl propionate) cross-linking of the BF₁ factor by dithiothreitol and analysis in second dimension by gel electrophoresis. The results indicate that subunit can cross-link with itself and with each of the other subunits except . The subunit is also able to cross-link with itself and with the other subunits although to a minor extent than , and that ₂ aggregates are present. These results represent a specific pattern of cross-linking for this BF₁ factor as compared to other F₁ coupling factors. It suggests a certain asymmetry in the spatial organization of the major subunits of M. lysodeikticus F₁-ATPase where the subunit must play a central role. A subunit stoichiometry _{3 3 2 2} is proposed for whole F₁-ATPase which leads to a molecular weight 440,000 consistent with the 430,000 value estimated by sedimentation equilibrium at low speed. A tentative structural model of M. lysodeikticus BF₁ factor is derived from these data. The significance of the results in relation to the possible generalization of the molecular architecture of F₁ factors is discussed.     

Intracellular sites of the synthesis of sweet potato mitochondrial F1ATPase subunits

Summary Five subunits (-, -, -, - and -subunits) of the six -and -subunits) in the F₁ portion (F₁ATPase) of sweet potato (Ipomoea batatas) mitochondrial adenosine triphosphatase were isolated by an electrophoretic method. The - and -subunits were not distinguishable immunologically but showed completely different tryptic peptide maps, indicating that they were different molecular species. In vitro protein synthesis with isolated sweet potato root mitochondria produced only the -subunit when analyzed with anti-sweet potato F₁ATPase antibody reacting with all the subunits except the -subunit. Sweet potato root poly(A)⁺RNA directed the synthesis of six polypeptides which were immunoprecipitated by the antibody: two of them immunologically related to the -subunit and the others to the - and -subunits. We conclude that the -subunit of the F₁ATPase is synthesized only in the mitochondria and the -, - and -subunits are in the cytoplasm.     

GABA A -receptors: Structural requirements and sites of gene expression in mammalian brain

GABA_A-receptors, the major synaptic targets for the neutotransmitter GABA, are gated chloride channels. By their allosteric drug-induced modulation they serve as molecular control elements through which the levels of anxiety, vigilance, muscle tension and epileptiform activity can be regulated. Despite their functional prominence, the structural requirements of fully functional GABA_A-receptors are still elusive. Expression of cDNAs coding for the ₁- and ₁-subunits of rat brain yielded GABA-gated chloride channels which were modulated by barbiturates but displayed only agonistic responses to ligands of the benzodiazepine receptor. GABA_A-receptors with fully functional benzodiazepine receptor sites were formed when the ₁- and ₁-subunits were coexpressed with the ₂-subunit of rat brain. These receptors, however, failed to show cooperativity of GABA in gating the channel. In order to determine the subunit repertoire available for receptor assembly in different neuronal populations in vivo, the sites of subunit gene expression were (₁, ₂, ₃, ₅, ₆, ₁, ₂, ₃, ₂) mapped by in situ hybridization histochemistry in brain sections. The mRNAs of the ₁-, ₁- and ₂-subunits were co-localized e.g. in mitral cells of olfactory bulb, pyramidal cells of hippocampus as well as granule cells of dentate gyrus and cerebellum. The lack of colocalization in various other brain areas points to an extensive receptor heterogeneity. The presence of multiple GABA_A-receptors in brain may contribute to synaptic plasticity, differential responsiveness of neurons to GABA and to variations in drug profiles.Special issue dedicated to Dr. Erminio Costa     

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.