If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
These reflections present a perspective of how I and my graduate students and postdoctoral fellows, over a span of many years, arrived at the concept that ATP is made by an unusual rotational catalysis of the ATP synthase. A recent sketch of the structure of this remarkable enzyme is given in Fig. 1. Such a depiction is the culmination of the efforts of many investigators.
Except for a few instances, the mention of important advances in information about the ATP synthase and in related areas of biochemistry is included without specific references. The objective of this contribution is not to provide a review of the field and to recognize priorities for contributions but to note how various advances impacted on studies by my group.
1Except for a few instances, the mention of important advances in information about the ATP synthase and in related areas of biochemistry is included without specific references. The objective of this contribution is not to provide a review of the field and to recognize priorities for contributions but to note how various advances impacted on studies by my group.
The two portions of the enzyme are the membrane-imbedded F0 and the attached F1 that has three catalytic sites, principally on the large β subunits. ATP is formed when protons pass through the F0, driving the rotation of the ring-shaped cluster ofc subunits and the attached ε and γ subunits. Other subunits attached to outer portions of the F0 and F1 served as a stator. The internal rotary movement of the γ subunit is coupled to sequential changes in the conformation of the catalytic sites. During ATP synthesis these conformational changes promote the binding of ADP and Pi, the formation of tightly bound ATP, and the release of ATP.
Revealing the mechanism of the ATP synthase became a major research goal in the latter part of my long career. This paper recalls how my career developed as related to the remarkable progress in biochemical knowledge. It presents the background and results of fruitful, as well as mistaken, approaches that were explored.
The Early Years
Born and educated through college in Utah, at the age of 21 I entered graduate school in the Department of Biochemistry at the University of Wisconsin in the fall of 1939. The biochemical research and teaching there were excellent. Not until years later did I appreciate all that is necessary to create such a fine scientific environment.
I had had no previous courses or research experience in biochemistry and was uncertain about my career choice. By the end of my first year of graduate study the fascination of biochemical understanding and the addictive effect of experimental attempts to uncover new knowledge had firmly launched me toward a career in biochemical research. The Department of Biochemistry at Wisconsin was at the forefront of research in nutrition and metabolism. Recent achievements included the identification of nicotinic acid as a vitamin, the irradiation of milk to produce vitamin D, the discovery of a vitamin K antagonist (dicoumarin), and the discovery of lipoic acid as a growth factor for bacteria. At that time incoming graduate students were assigned to a mentor professor. Both Henry Lardy, from South Dakota, and I joined the group of Professor Paul Phillips whose major interest was in dairy cattle nutrition. Evidence had been obtained that vitamin C might help prevent reproductive difficulties in cattle, and one of my assignments was to find if vitamin C might ameliorate the reproductive failure that occurred in rats with vitamin E deficiency. No benefits of vitamin C were noted, but the rats also showed the striking muscular dystrophy characteristic of vitamin E deficiency. Exploration, together with Henry Lardy, of the possible cause of this dystrophy led me into study of ATP-related enzymes. Henry is still active in an exceptionally distinguished career that has included major contributions to the understanding of oxidative phosphorylation.
The milieu at Wisconsin (meetings where students and staff discussed recent research papers, frequent research seminars, and class instructions) introduced me to the wonder of enzyme catalysis. A prominent event was a symposium on respiratory enzymes at which the outstanding biochemists Meyerhof, Cori, Ochoa, Lipmann, Kalckar and others contributed (
). From this and other sources I learned that ATP and phosphorylations were central to the capture and use of energy derived from foodstuffs.
Perhaps defective formation of ATP might underlie the muscle dystrophy in my vitamin E-deficient rats. One approach was to measure the ability of muscle extracts to make phosphocreatine during glycolysis. No definitive defect from vitamin E deficiency was found, but in the course of these experiments, I noted a stimulation of the transfer of phosphate from 3-phosphoglycerate to creatine by K+ ions. This was traced to a requirement of K+ for transfer of the phosphoryl group from 2-phosphoenolpyruvate to ADP. The discovery of the K+ activation of pyruvate kinase was the first demonstration of a K+ requirement for an enzyme reaction. The two Journal of Biological Chemistry publications reporting this were the best of several from my graduate studies (
), ATP was discovered by Lohmann in 1927, and oxidative phosphorylation was first demonstrated by Engelhardt and Liubimova in 1932. These salient contributions at that time seemed far from recent to me, and discoveries such as that of cell-free fermentation by Buchner made about 40 years earlier were relegated to the distant past—science after the escape from the Middle Ages. Now, from my present perspective, research of 30 years ago still seems fairly recent and vibrant. Time seems to go much faster, but it is I who has changed while a unit of time has retained its constant value.
An Introduction to Properties of Proteins
Some 20 amino acids linked in peptide bonds can yield proteins with a truly remarkable diversity of structural properties and the ability for specific combination and catalysis. The versatility of proteins is arguably the most important property of matter that has made life possible. Little was known about protein structure when I was a graduate student. As stated in a 1946 textbook of biochemistry (
): “Since the protein molecule is often built up of hundreds, even thousands, of these amino acids, the problem of protein structure is one of almost insuperable difficulty.” In the following years, to be an observer as the wondrous properties of proteins have been revealed is one of the finest rewards provided by my profession.
My appreciation of protein structure and function arose in 1943 when I joined a small group at Stanford University that was supervised by Murray Luck, founder of the Annual Review of Biochemistry. Our nation was at war, and Luck's group was asked if they could find how concentrated solutions of human serum albumin, used primarily for the treatment of shock in wounded soldiers, could be heated to inactivate pathogens without denaturing the albumin. The group found that low concentrations of long chain fatty acids or other non-polar anions such as acetyltryptophan would satisfactorily stabilize the albumin. Albumin preparations used militarily and commercially are still stabilized with small concentrations ofN-acetyltryptophan. As part of these studies, I noted that when albumin solutions were exposed to urea or guanidine hydrochloride, the large viscosity increase accompanying denaturation could be reversed by fatty acid addition; a specific combination was markedly influencing the folding of the protein (
). My interest in protein structure was firmly initiated.
Early Studies at Minnesota
In my 17 years at the University of Minnesota, I studied a wide variety of biochemical problems, including such items as the chemistry of α-tocopherol oxidation products, possible formation of antibodies by a refolding of denatured γ-globulins, sulfhydryl groups and enzyme catalysis, and the free energy of hydrolysis of ATP. Mostly my interests have concerned enzymes, and over the years names of some 25 different enzymes have appeared in titles of my publications.
An unsettled problem from my graduate studies at Wisconsin was clarified by the demonstration that mitochondria from the muscles of vitamin E-deficient rats performed oxidative phosphorylation as well as those from as normal muscle (
) independently reported that during oxidative phosphorylation oxygen uptake was decreased by the lack of phosphate acceptors. Such respiratory control was the basis for the later development, by Britton Chance and others, of the extensive use of an oxygen electrode to replace the cumbersome Warburg manometric method for measuring rates of oxygen uptake during oxidative phosphorylation.
More importantly, stimulated by the pioneering studies of Mildred Cohn (
), we initiated studies using the heavy oxygen isotope,18O, for probing phosphorylation reactions. As noted in later sections, insights into ATP synthase catalysis by my group were crucially dependent upon the use of 18O. The18O isotope and mass spectrometer facilities were made available by physics professor Alfred Nier (a benefit of a research university and a cooperative faculty). Over the years we and others have modified and improved techniques for 18O measurements. Yet studies with 18O remain more laborious than many approaches and have not been widely used. The lack of familiarity with the 18O measurements probably added to the reluctance of the field to accept our concepts, as they were later developed in the 1970s.
In our early studies with 18O we demonstrated that in the glyceraldehyde-3-phosphate dehydrogenase reaction an oxygen from inorganic phosphate appears in the carboxyl group of the 3-phosphoglycerate formed (
). The phosphorylation accompanying this oxidative step of glycolysis was a prominent basis for the widely adopted paradigm that a phosphorylated intermediate was likely formed during the oxidative phosphorylation of the respiratory chain.
In related experiments my group showed that the enzymic catalyses for formation of phosphocreatine from 3-phosphoglycerate occurred with the retention of all 3 oxygens of the phosphoryl group. Thus such phosphoryl transfers do not involve any steps giving exchange of Pi oxygens with water (
Our initial studies of oxidative phosphorylation with 18O revealed an important characteristic of the oxidative phosphorylation process. We incubated mitochondria with Pi labeled with both 18O and 32P and unlabeled ATP in the presence or absence of substrates or of oxidation inhibitors. We were surprised to discover that, in addition to the strikingly rapid exchange of Pi oxygens with water, a quite rapid Pi ↔ ATP exchange was occurring (
). The reactions of oxidative phosphorylation appeared to be dynamically reversible. The reversibility continued even when electron carriers were inhibited or nearly fully reduced. This gave evidence for formation of some type of energized compound or state, independent of oxidation-reduction reactions that allowed the ready reversal of the reaction sequence. We thought this likely was some type of chemical intermediate; the idea of an electrochemical gradient across a coupling membrane was far from our thoughts.
Possibilities arose of pursuing interesting aspects of enzyme catalysis not related to ATP formation. For example, in 1955 while on a Guggenheim fellowship for study in Sweden with Nobelist Hugo Theorell, I noted a previously overlooked shift in the fluorescence of NADH upon binding by a dehydrogenase (
). This gave a new basis for measuring combinations of NADH with enzymes. However, the problem was not as interesting as the studies of oxidative phosphorylation that I was also pursuing in the laboratories of Olov Lindberg and Lars Ernster at the Wenner Gren Institute. In an experiment conducted in part in Sweden18O was used to demonstrate that the terminal bridge oxygen in ATP formed by oxidative phosphorylation came from ADP, not Pi. This and some other research were reported at an International Union of Biochemistry symposium in Japan (
). At that time I was a bit pessimistic about gaining a satisfactory insight into how oxidative phosphorylation occurs. In my contribution I stated: “Our basic knowledge of the chemistry involved does not appear adequate for the task, and the problem is likely to be with us for some time. Researchers who undertake indirect approaches to the problem should do so with recognition that their experiments cannot give final answers, and may not even point the way to final solutions.” In retrospect, the pessimism seems appropriate.
During the next several years we undertook experiments looking for intermediates in oxidative phosphorylation, particularly by making use of 32P as a tracer. We learned that radioactively induced reactions of phosphorus compounds with highly labeled32Pi could give rise to radioactive impurities that stick to mitochondrial components but that did not behave like intermediates. Most of my publications during this period were from some worthwhile investigations with other enzymes; one needs to keep research funding available. One of my favorite sayings is that most of what you accomplish in research is the coal that you mine while looking for diamonds.
Some of our studies concerned patterns of isotope exchanges at equilibrium with glutamine synthetase using 18O,32P, and 14C. It soon became apparent that covalent bond cleaving and formation may not be rate-limiting in enzyme-catalyzed exchanges. Somewhat surprisingly, adequate rate equations governing exchange reactions of enzymes were mostly lacking. I spent a fair effort in a pioneering development of appropriate relationships (
). To some reviewers these relationships were unexpected, and there is an interesting story not told here about what I needed to do to get the publication accepted. Various applications were made by my group. For example, data with glutamine synthetase revealed that the binding of ATP and glutamate was random, and such subtleties as a spatial selectivity of transfer of only one oxygen of the glutamate γ-carboxyl group to phosphate when glutamate and glutamine are readily interconverted at the catalytic site (
). We did not pursue such observations until about a decade later when we belatedly recognized their potential relationship to the mechanism of oxidative phosphorylation.
The Phosphohistidine Story
In 1961 it seemed that our searches with 32P had hit pay dirt. We reported that under oxidative phosphorylation conditions a brief exposure to 32Pi and solubilization of the mitochondria with concentrated urea and detergent gave a non-dialyzable 32P-labeled substance. The rate of its formation from Pi or ATP, the disappearance in a cold Pi chase, and the effect of inhibitors and reaction conditions were consistent with its being an intermediate in oxidative phosphorylation. Our interest was heightened when my capable associates identified the substance as a phosphorylated histidine residue in a protein (
). This was the first recognition of a phosphohistidine in biochemical systems. The ability to form the bound phosphohistidine in soluble preparations from mitochondria encouraged the possibility that we could characterize details of the formation process. As the research developed, I became overly enthusiastic in regarding the phosphorylated protein as an intermediate of oxidative phosphorylation (
). In retrospect, I should have been more cautious. It was at this stage that my laboratory group moved to UCLA where we joined the Biochemistry Division of the Chemistry Department.
Our continued studies showed that dialyzable substances from mitochondria could modulate the bound phosphohistidine formation, and this led to the recognition that CoA and succinate were particularly effective. We had overlooked the substrate level phosphorylation accompanying the citric acid cycle. We became aware that a Ph.D. thesis at Illinois by Upper (
) had reported evidence of formation of a phosphoenzyme with the Escherichia coli succinyl-CoA synthetase and that such formation had been suggested earlier from catalysis of an ADP ↔ ATP exchange by the synthetase (
). Our further studies showed that the phosphorylated protein we had detected was indeed an intermediate in the formation of nucleoside triphosphate (ATP or GTP depending on enzyme source) from Pi by succinyl-CoA synthetase (
). Our bound phosphohistidine was clearly not an intermediate in oxidative phosphorylation. In Olympic analogy, we were reaching for a gold but were fortunate to have obtained a bronze.
Another Decade with Little Essential Progress
At this stage I felt that perhaps I could do more for science by accepting an opportunity to become the initial Director of the Molecular Biology Institute at UCLA. Fortunately this did not prevent reasonable continuation of laboratory studies, although I was not encouraged about the progress we and others were making toward elucidation of the major problem of how cells captured energy from oxidations to make ATP. At that time I of course did not know that a decade later we would be fortunate in developing a new concept for oxidative and photosynthetic phosphorylation.
Meanwhile my group pursued some worthwhile studies with other enzymes and continued a few probes of ATP synthesis that were useful but did not yield or point to breakthroughs. In a more sensitive search for the labeling of unidentified components with 32P, a small amount of rapidly labeled lipid fraction was detected (
). However, this labeling was found to continually increase with time, not an expected characteristic of an intermediate. The independence of oxygen exchanges from oxidation-reduction reactions was more firmly established (
). Most of them came from water and substrates, undoubtedly by exchange patterns we had been investigating. Other studies gave a welcome observation that laid the base for the later extensive use of chloroplasts by my group; under appropriate conditions chloroplasts catalyzed rapid Pi ↔ ATP, Pi ↔ HOH, and ATP ↔ HOH exchanges. Like oxidative phosphorylation, photophosphorylation was dynamically reversible and its mechanism could be probed by oxygen exchange measurements (
). From later developments the lack of the exchanges noted previously was likely because of the unusual and particularly strong Mg-ADP inhibition of chloroplast ATPase activity that can occur in the dark but is readily reversed by protonmotive force.
In an attempt to gain more insight about energy coupling we also conducted some studies on active transport by E. coli. We obtained convincing evidence that a common energized state or intermediate could drive transport or ATP synthesis (
) and others. However, unlike Harold, we were reluctant at that time to regard that the energized state was a protonmotive force. We were not alone in this reluctance. The field was active, and frequently reviewed. The 1967 (
) reviews in the Annual Review of Biochemistry on electron transport and phosphorylation gave brief and generally negative assessments of Mitchell's proposal that protonmotive force drove ATP synthesis. My hesitation in accepting this proposal came from the lack of a satisfying explanation as to how proton migration could drive ATP formation.
The mechanism of the ATP synthase remained unclear. As noted in the reviews mentioned above, there were a plethora of hypothetical compounds and reactions suggested for participation in ATP formation. A possibility consistent with our various experiments was that an energized state, not involving oxidation-reduction reactions, was used to drive a reaction in which an oxygen from Pi formed water as ADP was phosphorylated to yield ATP. We and others wondered if in some manner energy captured in conformational changes of proteins was involved. Remarkable advances in recognizing the versatility of protein structure were occurring. The x-ray structure of hemoglobin and other proteins and the allosteric properties of enzymes suggested the energy requirements for ATP formation might be accommodated in conformational changes of proteins. But we still had no clear idea about how the conformational changes might function.
A New Concept
In 1972, the first part of what I later called the binding change mechanism of ATP synthesis came from further considerations of past data, not new experimental findings. While attending a seminar that I did not understand, some puzzling aspects of oxygen exchange measurements were occupying my mind. Our thoughts had been that the major use of captured energy was to make the covalent structure of ATP. The realization struck me that past data could be explained if the major use of energy was not to form the ATP but to release a tightly bound ATP from the enzyme. Reversible formation of bound ATP at a catalytic site could explain why the exchange of Pi oxygens was less sensitive to uncouplers than net oxidative phosphorylation. For me it was a rare moment of insight, like suddenly reaching a summit on a mountain climb and seeing a beautiful valley spread below. All enzymes have the capacity for ready reaction reversal at catalytic sites and to bind both products and reactants. The reversal of the hydrolysis of ATP by the ATP synthase is no more remarkable than the reversal of simple hydrolyses by many enzymes, except that with the ATP synthase the product ATP is tightly bound. An additional step or steps must intervene for ATP release. This could logically be an energy-requiring conformational change of the catalytic site.
Richard Cross had joined our laboratory as a postdoctoral fellow. At UCLA he further documented the uncoupler-insensitive oxygen exchange and other aspects. We submitted a paper, “On a New Concept for Energy Coupling in Oxidative Phosphorylation Based on a Molecular Explanation of the Oxygen Exchange Reactions,” to the Journal of Biological Chemistry for consideration. The publication was declined; at that stage our evidence was not strongly convincing. However, the concept remained appealing. I had recently been elected to the National Academy, and the paper was published in Proceedings of the National Academy of Sciences as the first paper I sponsored for the journal (
The presence of multiple binding sites for ADP and ATP on the isolated F1-ATPase and the ATP synthase had been recognized by Slater's group and others. During catalytic turnover some of these nucleotides exchanged with medium nucleotides, and Slater and associates had also suggested the possibility that energy-requiring release of bound ATP might occur in oxidative phosphorylation (
). The estimated −ΔG0 of the binding of ATP from our and other data was 12–13 kcal/mol. A corresponding tight binding was anticipated for the ATP synthase. In related experiments, Bagshaw and Trentham had recently shown that the apparent ΔG for the hydrolysis of the bound ATP to bound ADP and Pi was only about −1.3 kcal/mol (
); the equilibrium was not far from unity. In a subsequent cooperative study with these investigators an exchange of phosphate oxygens of bound ATP with water was demonstrated to accompany the ATP hydrolysis by myosin (
). Later observations characterized how the combination of myosin with actin promotes the release of the tightly bound ATP, a conformational transition analogous to that proposed for the ATP synthase (
At this time I contributed a chapter on “Conformational Coupling in Biological Energy Transductions” in which the possibility that changes with ATP synthase were driven by protonmotive force was recognized. However, my preferred view was still that the conformational changes were driven by some type of interaction with oxidation-reduction enzymes (
), and in ensuing years he and others continued to present evidence and win converts. By the early 1970s even holdouts like myself were beginning to see the light. It seems probable that the role of protonmotive force would not have been recognized for a long time without Mitchell's contributions.
If proton translocation were coupled to ATP synthesis, I felt it would be accomplished indirectly by protein-linked conformational changes. In contrast, Mitchell proposed that the translocated protons reached the catalytic site and participated directly in the removal of a water molecule. I found his 1974 proposal in FEBS Letters (
Over the years Peter and I had extensive correspondence and shared a mutual respect. Although we were looking at essentially the same mechanism we tended to present different pictures of our views. Too often in science there is rancor between those who disagree. An important lesson that I have learned is that more will be accomplished if one can maintain cordial relations in an exchange of interpretations.
By the mid-seventies other investigators had provided much welcomed information about the ATP synthase that was quite relevant to mechanism studies. Hatefi and others in David Green's laboratory had shown that the mitochondrial inner membrane could be fractionated to yield separate complexes of the respiratory chain components and the ATP synthase. They (and particularly Racker
Racker's contributions were outstanding. One of his former associates, Geoffrey Schatz, has provided a splendid memoir of Racker's career (Memoirs, National Academy of Sciences, 1996, available on the internet from www.nationalacademies.org).
and associates) had separated and characterized the F1-ATPase. The knobs visible in electron micrographs of mitochondrial membranes were identified with the F1-ATPase, connected by a stalk to the membrane portion of the synthase. A similar ATPase had been found in a wide variety of organisms. The ATPase was known to have two or three copies of major α and β subunits and single copies of other smaller subunits. The unusual subunit stoichiometry and observations in a number of laboratories that modification of one β subunit per enzyme essentially stopped catalysis raised intriguing questions about mechanism. The portion of the synthase imbedded in the membrane, F0, was recognized as being involved in proton transport. The addition of F1-ATPase to F0 preparations could restore oxidative phosphorylation or photophosphorylation. Either proton gradients or membrane potential sufficed to drive ATP formation. Beechey had shown that a buried carboxyl group on a small hydrophobic subunit of F0, present in multiple copies, readily reacted with dicyclohexylcarbodiimide (DCCD)3 and that this blocked oxidative phosphorylation.
Although information about the ATPase was becoming extensive, how proton translocation could be coupled to ATP formation remained poorly understood. We were encouraged some by the concept that energy-linked binding changes were involved. Fortunately, at this time we obtained evidence for an unusual catalytic site cooperativity displayed by the ATP synthase and the isolated F1-ATPase. There was a feeling in my research group that some important secrets about the ATP synthase were being revealed. This created an ambience that stimulated research efforts. Such occasions are an all too infrequent reward of basic research. They help soften the disappointments of the many experiments that yield little or no helpful information.
Alternating Site Participation
Many enzymes have more that one catalytic site, suggesting the possibility of a catalytic cooperativity between sites such that catalytic events at one site are promoted by substrate binding at another site. With most multicatalytic site enzymes, limited or no cooperativity has been observed. In contrast, we found that the ATP synthase showed a nearly complete dependence of continued catalytic steps at one site on the presence of substrate(s) at a second site. This was the first enzyme for which such a striking behavior had been discovered, adding to our interest in the phenomenon.
Our discovery arose from researches by Jan Rosing, a postdoctoral fellow with exceptional experimental skills from Slater's group, and Celik Kayalar, a gifted graduate student. They were symbiotically productive. We devised methods for estimating oxygen exchanges by submitochondrial particles that accompany: (a) the binding, exchange, and return to the medium of Pi; (b) the binding, exchange, and return to the medium of ATP; (c) the binding of Pi, intermediate exchange, and the release of ATP formed; and (d) the binding of ATP, intermediate exchange, and the release of the Pi formed. These measurements with 18O were accompanied by measurement of the Pi ↔ ATP exchange with 32Pi. The exchange patterns gave evidence that besides promoting ATP release, energy input also increased competent Pi binding. More importantly, the measurements yielded exchange patterns that Kayalar proposed could be explained if the binding of a substrate at one site was necessary for the release of a product from another site.
Whether two or three catalytic sites per enzyme were present was not known at that time. We proposed alternating behavior of two sites, although it was recognized that the results would also be compatible with sequential participation of three sites (
). During net ATP formation or hydrolysis, sites were considered to proceed sequentially through the steps of binding, interconversion of reactants, and release so that at any one time each catalytic site was at a different stage of the catalysis. The concept seemed attractive, but more evaluation was needed.
David Hackney, a talented postdoctoral fellow from Dan Koshland's laboratory, had joined our group and initiated his excellent experimental and theoretical studies of the oxygen exchanges. We were proposing that Pi and ADP can bind and reversibly form bound ATP but that ATP cannot be released until Pi and ADP bind to an additional site. If dynamic reversal of ATP formation at a catalytic site continued in the absence of net reaction, then reductions in the concentration of Pi or ADP should increase the amount of intermediate oxygen exchange per ATP made. We were encouraged by a report from a former postdoctoral fellow of our group, Robert Mitchell, that he and his colleagues observed increased intermediate oxygen exchange accompanying ATP hydrolysis by submitochondrial particles when ATP concentration was lowered (
) that when ATP was exposed to chloroplasts in the light, the ATP showed nearly complete exchange of its oxygens before being released. This is as expected if low ADP concentration in the medium prevented the release of the ATP and many reversals occurred before its release.
Hackney observed that during net oxidative phosphorylation as either ADP or Pi concentration was decreased, there was a marked increase in water oxygen incorporation into each ATP formed (
). Additional observations made it unlikely that some type of enzyme heterogeneity or hysteresis could explain the exchange patterns. It deserves emphasis that these experiments were performed with submitochondrial particles during net ATP synthesis, giving them relevance to the actual oxidative phosphorylation process.
An interesting possibility was that catalytic site cooperativity might also be found with the isolated F1-ATPase. Several years earlier, Ef Racker brought some of his purified F1-ATPase to our laboratory to find if his enzyme would catalyze an intermediate Pi ↔ HOH exchange. We tested this at millimolar concentrations of ATP and found that the Pi formed contained only close to the one water oxygen necessary for the hydrolysis. Now, however, with our evidence for cooperativity, it was evident that if reversible ATP formation could occur in the absence of protonmotive force and if participation of alternating sites was necessary, then the extent of intermediate Pi ↔ HOH exchange with each Pi released should increase as ATP concentrations are lowered. This was found to be so (
). They termed this “uni-site catalysis,” and their results added considerably to the acceptance by others of alternating site participation. In these studies theKd for ATP binding to one site of the F1-ATPase was shown to be near 10−12m (
), a contribution that warrants wider recognition. In a slow reaction, needing relatively high Pi concentration, a tightly bound ADP became phosphorylated. Other findings made it probable that this was at the same site as the ADP that was rapidly released in the acid-base transition of thylakoid membranes and thus that this site was likely where covalent bond formation occurred during photophosphorylation.
In addition, results of various investigators established that chemical modification of only one catalytic site effectively stopped catalysis and that each of the three catalytic sites had a different capacity for derivatization. Such behavior agreed with the concept that during catalysis all three catalytic sites were in different conformations and proceeded sequentially through the conformations.
The Basis of 18O Exchange
Our studies with 18O are interpreted on the basis that the exchange results from a reversal of the formation of bound ATP from bound ADP and Pi. As covered in the Appendix of a review there is strong support for this interpretation (
), as well as that of myosin ATPase as mentioned above, result from reversible formation of a phosphorylated enzyme or enzyme-bound pyrophosphate or ATP, respectively.
Probes of Initial Reaction Rates
Other evaluations of our postulates were needed. Rapid mixing and quenching techniques yielded essential information. One objective was to find if a tightly bound ADP on the chloroplast ATP synthase might react with medium Pi to form ATP in the first turnover of the enzyme. We used rapid mixing in an acid-base transition of chloroplast thylakoid membranes, as introduced by Jagendorf and colleagues, to start ATP synthesis in a few milliseconds. We found that the tightly bound ADP was not directly phosphorylated but was rapidly released to the medium and that the first ATP formed came from medium Pi and ADP (
). As substantiated in later experiments, the tightly bound ADP in such chloroplast membranes prior to release is tightly bound at a catalytic site without Pi.
The demonstration that exposure to protonmotive force caused the release of a tightly bound ADP from a catalytic site without phosphorylation had important implications for later developments. The tightly bound ADP in the presence of Mg2+ causes potent inhibition of ATPase activity of the ATP synthase and F1-ATPase. Thus such inhibition in the intact synthase is readily and quickly overcome by protonmotive force. When a step of rotational catalysis occurs, the binding site with the tight ADP is opened as if it had an ATP present, while another site is binding ADP and Pi. The properties of the tightly bound ADP also aided interpretation of Walker's 1994 x-ray structure of the major portion of the F1-ATPase, in which one β subunit has a tightly bound ADP and Mg2+ present (
Our rapid mixing experiments verified that medium ADP was rapidly bound and phosphorylated as if no phosphorylated intermediates were involved. They provided evidence that during photophosphorylation, in addition to a transitorily bound ATP, about one bound Pi and one bound ADP per enzyme are present and committed to ATP synthesis (
). Such results harmonize with the alternating site model with more than one catalytic site having bound reactants, as required if a tight site is already filled and substrates must initially bind at another site.
Research Conferences and Binding Change Mechanism
Research conferences are important to scientific progress because concepts can be freely discussed, and the publication of proceedings often allows inclusion of material not suited for the usual journals. For example, in my contribution to a 1979 conference honoring Ef Racker, I summarized our concepts and considered how to name our suggested mechanism. A name seemed desirable for ease of discussion and to identify the concept in the field. My contribution entitled “The Binding Change Mechanism for ATP Synthesis” was the first publication in which this nomenclature was used (
The binding change mechanism at that time included the following concepts. The first compound made from Pi is ATP itself (no intermediates); a principal requirement of energy is not for the formation but for the release of ATP; energy input also promotes the competent binding of Pi and the sequential participation of catalytic sites so that binding of substrate(s) at one site is necessary for release of product(s) from another site. Two years later, another and even more novel concept of the binding change mechanism was developed, namely the proposal of rotational catalysis. The suggestion that rotation of internal subunit(s) drives the binding changes for catalysis was first published in reports from 1981 and 1983 conferences at the University of Wisconsin (
In the 1970s highly enriched 18O was available, mass spectrometry techniques for 18O analysis had improved, and Mildred Cohn had introduced an NMR method for measuring18O in phosphate compounds. David Hackney developed theoretical aspects of 18O measurements relevant to observed distributions of 18O isotopomers of Piwith 0 to 4 18O atoms per Pi or 0 to 318O atoms per ATP molecule. Measurement of the presence of18O in ATP formed by photophosphorylation showed a pronounced increase in 18O loss at lower ADP and Pi concentrations (
). More importantly, the distribution of 18O isotopomers corresponded to that statistically expected if all the ATP were produced by the same catalytic pathway. This eliminated the possibility that substrate modulation arose from heterogeneity of the enzyme used and made modulation by control sites unlikely. We now regarded the catalytic site cooperativity of ATP synthase to be reasonably well established.
Companion studies with the F1-ATPase showed that when highly 18O-labeled ATP was hydrolyzed by F1-ATPase at different ATP concentrations, the distribution of 18O isotopomers was as expected for a single catalytic pathway (
). At appropriate labeling and substrate concentration ranges, the distribution patterns provided a sensitive test for more than one catalytic pathway. A statistically homogeneous distribution meant that every substrate that reacted faced the same possibilities of proceeding through the same reaction steps. This means that rate constants governing the binding and release of substrate(s), their reversible interconversion, and the release of product were the same. To me, the power of this type of 18O use is unusual and indeed a bit awesome.
By now essential contributions of other investigators, including Kagawa and associates in their fine studies with the F1-ATPase from thermophilic bacteria, had established the presence of three catalytic sites with circular distribution of alternating large α and β subunits around a central core. Catalytic sites were regarded as largely on the β subunits, with the core representing the γ and possibly other small subunits. Observations in McCarty's laboratory demonstrated that modifications of –SH groups on the γ subunit markedly affected catalytic capacity of the chloroplast enzyme (
). The capacity of the FO component for DCCD-sensitive proton transport had been established. These and other findings strengthened our view that conformational changes in the FO were in some manner transmitted through the stalk to the catalytic sites on the β subunits to drive the binding changes for ATP synthesis.
Catalytic sites on multisubunit enzymes can be very sensitive to conformational changes in adjacent subunits. Changes in the γ subunit markedly modulated catalysis. How could all three β subunits have identical interactions with the γ subunit? Occurrence of tripartite symmetry of the γ subunit seemed unlikely. The evidence that all three sites conducted catalysis identically was compelling to me. The more I puzzled about these aspects, the more it seemed that there was only one satisfactory answer. This is that the internal asymmetric core, composed of γ and any other tightly associated minor subunits, would need to move rotationally with respect to the outer ring of catalytic subunits. Such movement would allow identical interactions with β subunits as the rotation drove the sequential conformational changes of catalytic sites. When I first presented this concept to my research group, their acceptance was initially quite reserved (they knew all too well that I could be wrong). With further consideration, they became interested and supportive. Much remained to be explored, and some experimental approaches are summarized in the next few sections.
Modulation of Oxygen Exchanges by ATP Concentration
The modulation by ATP concentration of the 18O exchange by the mitochondrial F1-ATPase was more carefully documented (
). Various wild type and mutant E. coliF1-ATPase likewise showed increased exchange of the Pi formed with lower ATP concentrations. However, the distribution of 18O isotopomers with the E. colienzyme revealed more than one reaction pathway, apparently arising in part from the degree of dissociation of the inhibitory ε subunit (
). A question had been raised about whether the F1-ATPase from a thermophile showed catalytic cooperativity because uni-site catalysis was not readily apparent. A cooperative experiment disclosed the expected modulation of the oxygen exchange but at a higher range of ATP concentration (
). These various results meant that the increase in the extent of oxygen exchange with each Pi formed (which occurs with a decrease in the ATP concentration) is likely a general property of all F1-ATPases and supports the probability that all ATP synthases share a common mechanism.
The ATPase of vacuolar membranes has been noted to have a composition resembling that of the ATP synthase. We felt that it should show similar oxygen exchange properties, and measurements demonstrated that this was so (
We devised methods to measure bound reactants during steady-state ATP synthesis. A hexokinase accessibility method gave a measure of bound ATP, and a rapid dilution of medium32Pi gave a measure of bound Picommitted to form ATP. Measurements during photophosphorylation showed that even at lower substrate concentrations the total of catalytic site-bound ATP and committed Pi was greater than one per enzyme, as anticipated if the proposed catalytic site cooperativity was occurring. During photophosphorylation, 32Pirapidly labeled catalytic ATP, and then the medium [32P]ATP formed was incorporated much more slowly into the non-catalytic sites (
). Such results helped explain labeling patterns we and others had observed and supported our concepts of tightly bound ATP as an intermediate and of catalytic site cooperativity.
The Insidious MgADP Inhibition
Occasionally in biochemical research one encounters a property of a system that seems designed to confuse and thwart the researcher. Such is the case with the inhibition by Mg2+, which is dependent on the presence of an ADP bound without Pi at a catalytic site. Clarification of this unusual role of a tight ADP was necessary for an adequate understanding of the proposed binding change mechanism. The F1-ATPase as conventionally isolated usually has a considerable portion with tight ADP present. In 1975, Moyle and Mitchell reported that mitochondrial F1-ATPase was slowly inactivated by Mg2+ (
). Subsequent studies in our and other laboratories revealed characteristics of the inhibition. F1-ATPases with tightly bound ADP when exposed to Mg2+ shows little or no initial activity upon ATP addition. Added ATP promotes slow release of the inhibitory ADP from a catalytic site as an increase to a steady-state rate is attained. At steady state, a slow interconversion of active and inactive forms continues. The bound ADP required for inhibition may arise from the cleavage of bound ATP or from medium ADP depending upon reaction conditions. Pi and various anions activate by promoting release of the ADP. The inhibitory ADP is at a catalytic site, not at a regulatory site as had been suggested.
Another important result of our continued probing was the recognition that, under some conditions, the presence of ATP at a certain non-catalytic site is necessary for the onset of activity of the chloroplast F1-ATPase (
). This was the first recognized function for a non-catalytic bound ATP. The action was found to result from acceleration of the release of the inhibitory ADP from catalytic sites that follows the addition of medium ATP (
). With the mitochondrial F1-ATPase, ATP binding to the non-catalytic site could also accelerate the onset of the Mg-ADP inhibition. Upon addition of ATP and Mg2+ to the mitochondrial enzyme, an initial burst of activity declines to a slow rate as the Mg2+-induced inhibition sets in; then the rate increases to a steady state as the non-catalytic sites slowly bind ATP (
From the above it is apparent that complicated rate patterns may be found. It is probable that with all F1-ATPases, and even under favorable conditions, a fair portion of the enzyme may be in the inhibited form. Many reported and planned experiments may be undermined by an unrecognized occurrence of the Mg-ADP inhibition. A procedure for estimating the portion of the enzyme in the inhibited form, as developed by Murataliev (
As mentioned earlier, with the intact ATP synthase the inhibitory MgADP is quickly removed by exposure to protonmotive force. This is akin to the removal of inhibitory imido-ATP that blocks hydrolysis but not synthesis. Chloroplast fragments show a light-activated ATPase that can be maintained by ATP cleavage. For unknown reasons, the activity continues even at higher Mg2+ concentrations that would readily result in inhibition of the separated F1-ATPase (
By the mid-1980s, the sequence of the ATP synthase subunits was becoming available. An ATP derivative, 2-azido-ATP, which serves as a good substrate and upon photolysis becomes covalently attached, was described in Lardy's laboratory. We embarked on studies to find the number and clarify the location of ATP and ADP binding sites on the F1-ATPases. That the ADP needed for the Mg2+inhibition was bound at a catalytic site was readily confirmed. The 2-azido-ATP or ADP at catalytic or non-catalytic sites (known to be principally on the β or α subunits, respectively) labeled specific tyrosines not far apart on the β subunit. The sites were thus near subunit interfaces. Sites with similar conserved sequences were noted with the mitochondrial, chloroplast, and E. coli enzymes (
and earlier references) had six nucleotide binding sites remained in question. Our data with the 2-azido nucleotides supported the probability that they, like the mitochondrial enzyme, had six total nucleotide sites (
). Such results added to the already recognized similarity of structure and mechanism of the enzyme from different sources.
We were somewhat surprised to find that derivatization by 2-azido nucleotides of some catalytic or of non-catalytic sites of the chloroplast F1-ATPase gave rise to multiple catalytic pathways. Measurements of the distribution of 18O isotopomers formed revealed that partially modified enzymes retained some activity that still showed modulation of oxygen exchange by ATP concentration. With more extensive derivatization, the native catalytic pathway disappeared, and two weak, but independent, pathways were noted. Clearly some remaining catalytic sites retained weak activity independent from what neighboring sites are doing (
Our experiments developing the binding change mechanism had been performed with isolated F1-ATPases or fragmented membranes. The characteristics of the exchange reactions of the ATP synthase under conditions where mitochondria were capable of a high rate of tightly coupled oxidative phosphorylation were not known. To gain such information, we undertook cooperative experiments with the research group of K. LaNoue, using the 32P and18O labels (
The tightly coupled mitochondria were incubated with oxidizable substrates. When low ADP limited net ATP synthesis, all catalytic steps continued rapidly as concentrations of Pi, ADP, and ATP remained unchanged. The expected rapid exchange of medium Pi with medium ATP with 32P was observed. However, this rate was only about a fourth of the rate of interconversion of bound Pi and ATP at the catalytic site as measured by 18O. Strikingly, the interconversion rate remained high even when the membrane potential was reduced considerably by dinitrophenol addition. The addition of ADP and a hexokinase and glucose trap resulted in rapid formation of glucose 6-phosphate. The distribution of 18O isotopomers in the glucose 6-phosphate showed a single reaction pathway, even when some uncoupler was added. The rapid reversal of bound ATP formation continued, so that about two reversals at the catalytic site occurred for each ATP released to the medium. Even though this reversal was occurring and some medium Pi was being formed from bound ATP, the overall reversal to form medium Pi from medium ATP ceased. This can be explained by the lack of import of ATP by the ADP-ATP translocase. During rapid ATP synthesis, unlike the rapid reaction reversal that is occurring at the catalytic site of the ATP synthase, with high ADP in the medium the translocase is a one-way street.
Another important confirmation of alternating site participation came from these experiments with intact mitochondria. In the dynamic state with no added ADP the forward and reverse rates of all steps are equal, so that there is an equal chance that medium Pi that has formed bound ATP will be released to the medium as Pi or as ATP. If release of the bound ATP can only occur when another ADP and Pi bind, then the rate that medium Pi forms bound ATP will be twice the rate that it forms medium ATP. A ratio of about 2 to 1 for these rates was found when the overall rates were varied up to 10-fold by changes in reactant concentrations or temperature. This provides evidence that alternating site participation is occurring under conditions where rapid oxidative phosphorylation is possible.
One aspect of these experiments may be pointing to an important and unrecognized property of the ATP synthase in its native environment. When the membrane potential was reduced by the addition of some uncoupler while ATP was present not only did net ATP hydrolysis occur but also the rate of reaction reversal at the catalytic site remained high. This was reminiscent of earlier observations with submitochondrial particles that the oxygen exchange accompanying ATP hydrolysis was much greater at higher ATP concentrations than with the separated F1-ATPase and was relatively insensitive to uncouplers. In some manner, the capacity for rapid interconversion of bound reactants is retained better with the native membrane-bound synthase. I would still like to know how this is accomplished.
Site Filling and Catalysis, an Unfinished Story
It is well recognized that when only one catalytic site on the F1-ATPase binds ATP or ADP and Pi, a slow interconversion of the substrates occurs. What remains uncertain at the time of this writing is whether a second or a second and a third site must bind ATP for the rapid release of ADP and Pi to occur. For over two decades since slow uni-site catalysis and alternating participation of catalytic sites was recognized, my group and most others felt that the binding of ATP to a second site sufficed for rapid catalytic turnover during net ATP hydrolysis. Because of binding affinities, it was recognized that all three catalytic sites would be filled at millimolar concentrations of ATP, approximating physiological conditions. I proposed that ADP during net synthesis, or ATP during net hydrolysis, entered the catalytic cycle by binding to different sites as indicated in Fig. 2. The critical need under conditions favoring rapid synthesis was regarded as the presence of interconverting, tightly bound substrates at Site 2 of Fig. 2 and ADP and Pi at Site 1. Conversely, under conditions for rapid hydrolysis the critical need was the presence of ATP at Site 3. In each case one rotational step would change the site to the tight conformation where covalent catalysis could occur. For both net synthesis and hydrolysis, the presence or lack of ADP or ATP at a third site was regarded as having a minor influence on the rates.
Several years ago, Senior and Weber and colleagues introduced a fluorometric method for estimating the amount of bound nucleotides at catalytic sites. They replaced a tyrosine at catalytic sites with tryptophan and replaced tryptophans in other locations of theE. coli F1-ATPase (Ref.
and earlier references). The binding of nucleotides at the catalytic sites quenched the fluorescence of the tryptophan and allowed an estimation of the number of catalytic sites filled. They and Allison's group with the thermophilic F1-ATPase (Ref.
and earlier references) found that three sites appeared to be filled with nucleotide as near maximal velocity was reached with an increase in the ATP concentration. They then assumed that the binding of ATP to a second and to a third site was necessary for rapid net hydrolysis. This I believe will prove to be incorrect. Instead, it still seems likely that although three sites may become filled (probably mostly with ADP), the essential need for rapid hydrolysis is the binding of ATP to a second site. In other words, bi-site activation probably occurs along with tri-site filling. These issues are considered in more detail elsewhere (
As the research journey proceeded, we attempted some evaluations of whether rotational catalysis indeed occurs. We found that when the chloroplast F1-ATPase was reacted first with 2-azido-ATP and then with [14C]DCCD, two different β units were derivatized. Thus the DCCD does not label a subunit with tightly bound nucleotide present. The DCCD-labeled enzyme retained weak catalytic activity. This made it possible for us to find if catalysis changed the conformations of the β subunits that determine their chemical reactivity, as would be expected if rotational movement of the γ relative to the β subunit had occurred. When the enzyme was first reacted with DCCD, allowed to perform catalysis with 2-azido-ATP, and then the azido-ATP photolyzed, the subunits were randomly labeled (
). To us, this made rotational catalysis likely, but the weak catalytic activity of the DCCD-modified enzyme detracted from the result.
In another approach a bifunctional cross-linking agent that reacted with lysine NH2 groups, and that had a central cleavable –S–S– linkage, was used. About three cross-links, mostly between the γ and δ subunits and the α subunit, caused loss of two-thirds of the activity, and the activity was recovered when the disulfide bonds were cleaved (
). This at first appeared to be a to be a definitive finding. However, examination of their paper suggested some possible experimental uncertainties. Also, possibly the derivatization had uncoupled Ca2+-activated hydrolysis from rotation. A more likely possibility was that the long –CH2– chains in their cross-linkers may have allowed sufficient freedom of movements to not be restrictive. The catalytically induced movements of β subunits that we had observed with the 2-azido experiments (
) still seemed valid. I thus did not abandon the concept of rotation and looked forward to better evaluations. The concept remained controversial in the field.
It was becoming clear that structural data could provide the base for critical assessment of rotational catalysis. I was aware that Walker's group was attempting to obtain suitable crystals for x-ray analysis. In the meantime, it seemed that my group might accomplish more by studies that were under way with the 2-azido derivatives and by trying to define the location and function of bound nucleotides. As these and related studies progressed I prepared a comprehensive 1993 review of the status of research on how ATP is made under the title “The Binding Change Mechanism for ATP Synthase—Some Probabilities and Possibilities” (
). The literature at that time was regarded as giving strong support to the concepts proposed in the binding change mechanism, with the exception that rotational catalysis was regarded as likely but definitely not established.
Fortunately progress continued in Walker's laboratory (
). Receipt of an advance copy of the report from Walker was an occasion for gratifying emotion. The reported structure showed that the three β subunits were indeed in different conformations, and one had poor nucleotide affinity. The γ subunit was centrally located with structural associations consistent with its rotation driving sequential conformational changes of the β subunits. The authors interpreted their data as strongly supporting the binding change mechanism. Other x-ray studies interpreted as inconsistent with the binding change mechanism (
The availability of high resolution structural data made more critical assessments of rotational catalysis possible. Richard Cross, my former postdoctoral associate, noted residues in the γ and β subunits that were closely adjacent. When these were replaced with cysteines by mutagenesis good catalytic activity was retained. Oxidation of the –SH groups to form a disulfide cross-link blocked catalytic capacity, which was regained when the disulfide was reduced. With the disulfide linkage present, two β subunits that were not cross-linked were replaced by β subunits from an enzyme labeled with radioactivity during growth of the E. coli. Cleavage of the disulfide and catalysis resulted in randomization of the position of the γ subunit relative to the labeled β subunits (
). Such results and other related findings were considered by the field to establish the occurrence of rotational catalysis.
In the spring of 1997, a stunning visual conformation of rotational catalysis came from the laboratories of Yoshida and Kinosita in Japan. In a novel experimental approach, they attached a long actin side chain with a fluorescent label to the γ subunit and, through inserted histidine residues, attached the modified enzyme to solid support. Upon hydrolysis of ATP, the rotational movement of γ was observable in the microscope (
). Important characteristics of the catalysis were shown and are still being studied. I remember the thrill when I saw the rotation from a VCR recording that Yoshida kindly sent me. The dramatic experiment has gained wide recognition and removed nearly all doubt about the existence of rotational catalysis. Independently, Junge and associates developed a sophisticated fluorescence polarization technique that showed rotation accompanying ATP cleavage (
). This progress added to the near certainty of rotational catalysis. I was able to include references to their papers in press in a contribution entitled “The ATP Synthase—A Splendid Molecular Machine” that appeared in the 1997 Annual Review of Biochemistry(
By 1994, my research laboratories were essentially closed. ATP synthase and bioenergetics and enzymology had yielded center stage to biochemistry related to genetics and development. Postdoctoral fellows were no longer seeking my laboratory. It was over 50 years since I had received my Ph.D. My mental capacities seemed to have slipped more than my physical capacities. I felt it would be difficult to continue to be at the research forefront and that unfunded U. S. Public Health Service applicants might make better use of funds than I, so I gave 2 years of committed research support back to the U. S. Public Health Service and bought a summer home in a mountain valley in Wyoming with accompanying golf and tennis facilities.
In October of 1997, while on my way to back to my winter haven in a home I love in the hills above UCLA, I received that electrifying early morning call informing me that I had been chosen to share the 1997 Nobel Prize in Chemistry. This of course was personally very gratifying. It increased my stature with my grandchildren. More importantly, it was a recognition of the many fine investigators that had contributed to the unveiling of the ATP synthase mechanism. It gave additional meaning to the careers of postdoctoral fellows and graduate students of my group without whom there would have been no prize, and it has made my life since more vital and interesting. I recommend that if you are going to receive a Nobel prize, have it come late in your career when you no longer have the rewards that come from participation with a fine research group. Also, then you do not need to answer the question, “What research are you going to do now?”
Some Final Comments
The contributions of many scientists were essential for the gaining of our present insights into the ATP synthase catalysis. They should share in the satisfaction that comes from knowing much about how this important biological catalyst appears to operate. As the scientific enterprise grows ever larger, many fine contributors do not receive the appreciation they deserve. They should share in a pride for what has been achieved.
The scientific accomplishment of the Twentieth Century that I admire most is the revealing of the multifaceted capabilities of proteins and of their structures that make these capabilities possible. Perhaps I am a bit more infatuated with enzymes than some. I do not know any enzyme that I could not learn to love, although I will admit that some are more attractive than others.
To all who created our stable and prosperous country and its research universities, which made a career such as mine possible, and mostly to my colleagues, I give my thanks.
Symposium on Respiratory Enzymes (1942) University of Wisconsin Press, Madison, WI.