Energy Transduction by Ion Currents?

TOC

Energy Transduction by Ion Currents?

Chapter 19

FRANKLIN M. HAROLD and PETER C. MALONEY

PERSPECTIVE

The Coupling Problem

In the world of physics, events take a predictable course. Water flows downhill, warm bodies cool, electric batteries discharge, paint peels off the wall; things fall apart. The formal expression of our common experience is the second law of thermodynamics, which states that the spontaneous direction of natural processes tends toward the degradation of energy, the loss of potential, and the dissipation of order. It is quite otherwise in biology, yet the ability of living organisms to swim and fly, to accumulate nutrients, and to generate molecular and anatomical order does not contravene the second law. Biological work functions are always performed at the expense of energy drawn from the environment and are accompanied by an overall increase of entropy.

The phrase "energy coupling" embodies one of the fundamental characteristics of living organisms and defines the central issue in bioenergetics, at least in its cellular and molecular dimensions. How do cells capture the potential energy inherent in a ray of light or a pinch of sugar and harness it to the execution of the many kinds of work required to stay alive and multiply? That is the coupling problem. It is a problem, not of thermodynamics, but of mechanism. Bacterial physiology has contributed prodigiously to our present understanding of biological energetics and holds the promise of extending this insight to the molecular level.

The first general solution to the coupling problem was attained over 50 years ago by Fritz Lipmann (139), who recognized that ATP and a few related phosphorylated metabolites serve as carriers of biological energy. In a nutshell, the function of the great highways of metabolism—respiration, photosynthesis, and fermentation—is to drive the endergonic synthesis of ATP from ADP and P_i. ATP, in turn, is the immediate energy donor for the performance of work; ATP is itself a reactant in the processes that are energized by its concurrent breakdown back to ADP and P_i. It is inaccurate but irresistible to envisage the free energy of metabolism being transiently "conserved" in the phosphoryl bonds of ATP. This clear and simple framework made bioenergetics comprehensible, and for many purposes it remains adequate today. It also focused attention on the next generation of problems, namely, the mechanism of ATP production by respiration and photosynthesis.

From Energized Membrane to Proton Potential

By 1960, the broad outlines of oxidative and photosynthetic phosphorylation had been worked out. It was known that in mitochondria and aerobic bacteria, electrons pass from reduced donors such as NADH and succinate to oxygen via a cascade of electron carriers that includes flavins, quinones, and cytochromes. The overall sequence of events had been established, and it was appreciated that the electron transfer chain is firmly associated with lipid membranes: the inner membrane of mitochondria and the plasma membrane of bacteria. The pathway of electron transport in photosynthesis was less definite, though a cyclic as well as a noncyclic route had been recognized. Everyone understood that the electron transport chains represent the exergonic limbs of a coupled process, whose endergonic limb is the phosphorylation of ADP to ATP. The enzyme that catalyzes this reaction had also been identified. Membranes that mediate energy transduction display a powerful ATPase activity which, in the intact organelle or cell, works in reverse, producing ATP and conserving the free energy released by electron transport. The mystery centered on the mechanism by which the two limbs are coupled.

Biochemists, fresh from the triumphant elucidation of the mechanism of ATP generation by glycolysis, hoped to isolate energy-rich intermediates linking electron transport to the ATP synthase, and during the 1950s and 1960s they devoted much effort to the hunt. Hopes for success would rise now and again, only to be dashed as one proposed intermediate after another was disqualified. But it must not be forgotten that much of the modern understanding of membrane energetics has its roots in the work of this period. Three discoveries are especially pertinent to what follows. First, the coupling between electron transport and ATP is reversible: NADH oxidation can drive ATP generation, hydrolysis of ATP can drive NAD⁺ reduction. Second, the linkage between the two can be disrupted by an array of simple substances, including 2,4-dinitrophenol, which became known as "uncouplers." Third, the functions of the elusive intermediate(s) went beyond the transfer of energy from respiratory chain to ATP synthase (55, 218). For instance, mitochondria take up substrates and Ca²⁺ ions in an energy-dependent manner, and the immediate energy donor proved to be that same elusive intermediate. Investigators came to speak of an ill-defined "energized state" of coupling membranes, commonly designated by Lipmann’s squiggle, ∼ X, and more than one suspected that their persistent failure to assign the squiggle a chemical identity betrayed, not the lack of technical skill, but a flawed conceptual framework (200, 233).

The contemporary resurgence of bioenergetics can be traced directly to Peter Mitchell’s publication of a radical alternative, one he called the "chemiosmotic hypothesis" (165). Figure 1 summarizes the basic principles. The electron transport chain, he proposed, is not plastered onto the membrane but arranged across it such that as electrons wend their way to oxygen, protons are translocated from one surface to the other. The membrane forms a closed vesicle and is relatively impermeable to protons. In consequence, proton translocation generates an electrical potential across the membrane, interior negative, and in time a pH difference may also arise, interior alkaline. Protons at the external surface therefore find themselves at a higher electrochemical potential than those in the interior. They are subject to a "pull," derived from both the pH gradient and the electrical potential, which Mitchell dubbed the proton motive force; it drives protons back across the membrane, down the electrochemical gradient established by respiration. The ATP synthase provides a pathway that allows the protons to traverse the membrane, and it is so articulated as to couple the flux of protons downhill to the uphill formation of ATP from ADP and P_i. In Mitchell’s view, the coupling of electron transport to ATP synthesis is effected, not by energy-rich chemical intermediates but by the circulation of protons. The immediate energy donor is Δp, the proton motive force, nowadays referred to as the proton potential; and the energized state of the membrane is nothing but the proton potential across the membrane.

Fig. 1Chemiosmotic energy coupling. Proton extrusion by the respiratory chain generates a proton potential, cytoplasm electronegative and alkaline. Protons return to the cytoplasm via the F0F1-ATPase and by symport with a substrate (S). Arrowheads point in the physiological direction, but all the reactions are in principle reversible.

Source:

It was apparent to Mitchell from the beginning that a proton circulation might support not only ATP production but other kinds of work as well. Figure 1 illustrates a case in point, the accumulation of a substrate, S, against a concentration gradient. Cohen and Monod had earlier characterized one of the first bacterial transport systems, the lactose permease (36). Cells possessing the permease could accumulate β-galactosides to an internal concentration 1,000-fold higher than that of the medium; the energy source was unknown, presumably ATP. Mitchell (166) proposed an alternative. In his view, the permease protein has two substrates, a sugar molecule and a proton, whose fluxes are obligatorily coupled as the diagram shows: cotransport, or symport. Accumulation of the sugar is achieved by the electrochemical driving force upon the proton. In principle, any membrane-linked function can be coupled to the proton circulation by an appropriate molecular device that allows passage to protons and harnesses their downhill flow to the performance of work.

Simple and elegant, but is this in fact how membranes work? Mitchell’s penchant for neologisms ("chemiosmotic" refers to reactions, including the one catalyzed by the ATP synthase, that have both a chemical and an osmotic aspect such that either one may drive the other) was but one of the causes of the controversy that ensued. Vectorial reactions, endowed with an intrinsic direction in space; energy transfer mediated by a current instead of a proper bond; and the crucial role of topological integrity—none of these sat well with biochemists who clung to the metaphor of the cell as a bag of enzymes and who suspected that subcellular compartments were little more than an excuse. A more comprehensive presentation (167) did not settle the issue; the debate over principles as well as technicalities continued well into the 1970s, subsiding only with award of the 1978 Nobel Prize in chemistry to Peter Mitchell. By then, the chemiosmotic hypothesis had been as closely and rigorously scrutinized as any proposition in biology (77, 79, 170, 178).

Something more should perhaps be said here about this extraordinary man, who passed away in 1992 at age 72. A graduate of Cambridge in its heyday, Mitchell was teaching in the Zoology Department of the University of Edinburgh when he formulated the chemiosmotic hypothesis. Shortly thereafter, ill health compelled him to seek a milder climate. He purchased Glynn House, a decaying manor in Cornwall, and, acting as his own architect, transformed it into a modern laboratory cum family residence. For the next quarter of a century, Mitchell and his lifelong colleague Jennifer Moyle served as codirectors of the Glynn Research Laboratory. In this peaceful rural setting, whence the view ranges over green meadows and woodland to a gleam of the Channel coast, they wrought a revolution in bioenergetics. And everyone with an interest in the subject would sooner or later make the five-hour train journey from London to discuss data, speculations, and the universe at large; to enjoy the liberal hospitality of Helen and Peter Mitchell; and to savor a different way of living the life of science. Mitchell was an original and penetrating investigator who tossed off ideas like firecrackers; and with his traveled mind, his deeply felt social concerns, and his unfailing fortitude, he inspired affection no less than respect. To have played some part in the intellectual drama that revolved around him has been one of the chief rewards of our own scientific labors.

Remembering Key Experiments

The noisiest battles of the chemiosmotic wars were fought over mitochondria and chloroplasts, but microbiological terrain also saw significant action. The issue here was not phosphorylation but energy coupling to active transport. The dispute is worth recalling, for it illustrates the kind of arguments that eventually led to the near-universal acceptance of the chemiosmotic hypothesis by the microbiological community.

The tale begins, not with Escherichia coli but with the fermentative bacterium Streptococcus faecalis (recently renamed Enterococcus hirae). Enterococci live by glycolysis. They have no respiratory chain and do not carry out oxidative phosphorylation, but Abrams et al. found that they possess a membrane-bound ATPase quite like that of mitochondria (2). The function of that ATPase was not at all obvious, since these bacteria produce ATP by substrate-level phosphorylation. Light broke when Harold and Baarda (82) discovered that a series of compounds known to uncouple oxidative phosphorylation also dissociated streptococcal glycolysis from the accumulation of K⁺ ions and other metabolites. Moreover, these uncouplers facilitated the diffusion of protons across the plasma membrane, just as Mitchell and Moyle (172) had found for mitochondria. The clear inference was that the uptake of metabolites was energized not by ATP itself but by the proton motive force. This conclusion was quickly extended to E. coli. Thanks to a pioneering study by Winkler and Wilson (240), it was known that energy is not required for the transport of β-galactosides across the plasma membrane but only for their accumulation. Pavlasova and Harold (191) then showed that in anaerobic cells, uncouplers inhibited neither the lactose permease nor the production of ATP but did abolish galactoside accumulation. The results lent strong support to a variation on the chemiosmotic hypothesis, one that Mitchell had anticipated: in glycolyzing organisms, the ATPase pumps protons outward, generating a proton potential that supports the uptake of nutrients (Fig. 2).

Fig. 2Streptococcus "chemiosmoticus." In enterococci, and in many other fermenting bacteria, the F0F1-ATPase extrudes protons and generates the proton potential. Protons return by diverse pathways, two of which are shown.

Source:

Now the pace quickened. Streptococci generate a pH gradient, apparently in consequence of proton extrusion by the ATPase (85). A group of Soviet investigators (15) introduced the use of lipid-soluble ions to measure the electrical potential of small cells, and this technique allowed Harold and Papineau (83, 84) to demonstrate that proton extrusion by the streptococcal ATPase generates the expected potential, interior negative. An especially compelling set of experiments made use of the ionophorous antibiotic valinomycin to facilitate the diffusion of potassium ions. When valinomycin is added to K⁺-replete, starved cells suspended in buffer devoid of K⁺, efflux of the cation generates an artificial membrane potential independent of any source of metabolic energy. A proton potential so induced supported the uptake of sugars and amino acids (12, 122); reversed the normal direction of the ATPase to generate ATP (153); and even supported rotation of the flagellar motor (157). The evidence as it stood in 1972 was presented by Harold in a frankly evangelistic review (75) that argued Mitchell’s case in the microbial context and earned its author a red pin on the world map in Mitchell’s study.

In the meantime, H. R. Kaback had introduced the use of sealed membrane vesicles of E. coli as a major new tool for research on biological energy coupling (103, 104). Vesicles sharpened the issue because they lack cytoplasmic metabolites and enzymes but possess a functional respiratory chain. When supplied with an oxidizable substrate, respiring vesicles were shown to accumulate β-galactosides, amino acids, and other substrates. Since vesicles neither produce ATP nor respond to added ATP, energy coupling must take place at the level of the membrane itself. Kaback and his associates argued vehemently against the chemiosmotic explanation and in favor of a direct linkage between the respiratory chain and diverse transport proteins (104, 106, 140). In fact, most of their observations were in principle compatible with both hypotheses, but their failure to observe accumulation of the lipid-soluble cation dibenzyldimethyl ammonium (DDA⁺) by respiring vesicles seemed to prove the absence of an electrical potential (interior negative). On the face of it, the chemiosmotic hypothesis had failed a crucial test.

Often in science, issues of great moment turn on niggling technical details. To use DDA⁺ uptake as a measure of the electrical potential, one must supply trace amounts of a lipid-soluble anion (83, 84, 138), apparently to diminish electrostatic repulsion at the membrane surface which drastically reduces the rate of cation uptake. Kaback et al. missed that point; when K. Altendorf and H. Hirata, then postdoctoral assistants in Harold’s laboratory, supplied the anion, the vesicles readily accumulated DDA⁺, demonstrating that they did after all generate the electrical potential predicted by the chemiosmotic hypothesis. No doubt we should have warned Kaback, but we did not; and so the issue came to a head at a meeting organized by the New York Academy of Sciences in 1973. As I (F.M.H.) listened to Kaback’s presentation, I realized that The Lord had delivered him into my hands, and when my turn came I smote him with glee (76). Publication in full (6, 91, 92) persuaded most of the remaining holdouts. Kaback himself joined the chemiosmoticists soon afterward and made his conversion the cornerstone of an enormously productive research program (105).

In the intact cell, the proton circulation is cryptic: any protons extruded across the membrane are immediately drawn back by the proton potential and are therefore all but invisible. To validate the chemiosmotic hypothesis, it was necessary to devise procedures to open the circuit, and by the end of the decade this task had been largely completed. Among the findings that underpin today’s consensus are the demonstration that the respiratory chain and the F₀F₁-ATPase do indeed translocate protons (89, 133, 146, 147, 237); the demonstration that these reactions possess intrinsic polarity, which can be displayed by comparing right-side-out with everted membrane vesicles (7); and verification of the low proton permeability of the plasma membrane (148, 201, 208). Various transport processes were shown to entail the coupled translocation of protons together with the substrate, either in the same direction (symport) or in the opposite direction (antiport) (180, 235, 236, 237a). Quantitative measurements of the membrane potential, pH gradient, and stoichiometries of pumps and porters were well in hand and were providing internally consistent results (37, 118, 119, 120, 122, 150, 154, 239, 245). Consensus was celebrated by a spate of review articles (73, 77, 78, 170, 202); they marked a change in the intellectual weather, the end of the cellular phase of energetics and the inception of the molecular one. Doubts persist on some aspects of energetics, and alternative views still find champions (123). But we would note that no alternative proposed to date has been nearly as successful in unifying the mass of data as the chemiosmotic hypothesis; and in what follows, we shall take its validity for granted.

QUANTITATIVE ASPECTS OF CHEMIOSMOTIC COUPLING

Magnitude of the Proton Potential

The chemiosmotic viewpoint prevailed in the end because it made sense but also because it was firmly grounded in thermodynamics and therefore offered quantitative predictions for opponents to challenge and for supporters to verify. Systematic treatments will be found in books by Harold (79), Cramer and Knaff (42), Nicholls and Ferguson (178), and Skulachev (217), not to mention Mitchell’s own presentations (167, 168, 169). Here we illustrate the quantitative side of the subject with reference to E. coli.

Imagine a system consisting of two compartments separated by a membrane (e.g., a cell and its environs) and a proton-translocating reaction that pumps H⁺ electrogenically from one compartment to the other (from inside to outside [Fig. 1]). Under ideal circumstances, the process will proceed to equilibrium and "stall" when the electrochemical potential of protons equals that of the driving reaction (at that point, the rate of proton backflow equals the rate of extrusion and therefore net proton transport ceases). At equilibrium,

(where Δ G is the free energy change per mole,

is the difference in electrochemical potential between protons on the outside and those on the inside, and n is the number of protons translocated by the reaction). Suppose the reaction in question is the oxidation of NADH by oxygen, with a Δ G (under physiological conditions) of –212 kJ/mol (–50.5 kcal/mol), and that oxidation results in the translocation of 6H⁺ per electron pair (see the following section); then the maximal that can be generated is –212/6 or –35 kJ/mol. It is convenient and customary to express electrochemical potentials in electrical units, by dividing by Faraday’s constant, F; we thus obtain a proton potential, Δ p = –35/0.0965 = –360 mV. This is the highest potential attainable; the actual potential will be lower, as discussed below.

The electrochemical potential of protons, which is the thermodynamic force that pulls protons back into the cell, is the sum of a concentration term and an electrical term.

= F Δψ + 2.3RT log [H⁺]_in/[H⁺]_out (kJ or kcal)

In this equation

is defined as above, Δψ is the difference in electrical potential, and R, T, and F have their usual meanings. Dividing both sides of the equation by F and replacing the logarithmic term by the pH difference across the membrane, we arrive at the more familiar form:

/F = Δ p = Δψ – (2.3RT/F)ΔpH (mV) (2)

Δ p ≈ Δψ – 60ΔpH (mV)

By convention, the electrical potential of the medium is defined as zero, and the sign of any other phase reflects the work done in moving a positive charge across the barrier; this makes the cytoplasm negative under most conditions. Likewise, ΔpH is defined as pH_i – pH_o. In E. coli, Δ p is therefore normally a negative quantity. In the example under discussion, the maximal Δ p is –360 mV; assuming a ΔpH of 0.5 unit (or –30 mV), we would expect the maximal membrane potential to be –330 mV.

Table 1 shows some actual measurements of Δψ and ΔpH for E. coli. The methods used are reasonably well standardized and have been reviewed in detail (121). Suffice it here to mention that Δψ is measured by the accumulation of radioactive, lipid-soluble ions (cations in the present instance, anions when Δψ is positive); ΔpH is inferred from the uptake of lipid-soluble weak acids (e.g., benzoate), or of bases when pH_i is more acidic than pH_o. Nuclear magnetic resonance (NMR) spectroscopy provides the most accurate measure of pH_i presently available. In all cases, corrections have to be applied to the data, which limit the accuracy of the determination. Nevertheless, there is no doubt that the observed proton potentials (–200 mV at most) are substantially lower than the thermodynamic maximum.

Table 1Electrochemical proton gradient of E. colia

The difference between observation and expectation can be attributed to a number of causes (aside from the uncertainties in the data themselves), of which we would note three. First, a cell does not operate at equilibrium but represents, at best, a steady state. Transport processes, including the ATPase and an array of porters, mediate the return of protons to the cytoplasm and, in so doing, diminish the steady-state proton potential. Second, the plasma membrane has a significant basal permeability to protons; it is small by comparison with the carrier-mediated backflow of protons, about 1% (148), but there is reason to believe that it rises sharply at high proton potentials (see below). Finally, note that early measurements drastically underestimated the stoichiometry of proton extrusion by mitochondria, which is now thought to be 10 to 12 protons for the oxidation of NADH (90). If the true stoichiometry of respiratory proton extrusion in bacteria were of the same order, the measured potentials would be in good accord with expectations from thermodynamics. But it is also quite possible that E. coli translocates only 6H⁺/O and thus dissipates (or "wastes") part of the free energy potentially available to the cells. We shall return to this subject below.

Composition of the Proton Potential

The relationship between ΔG and

is fixed by thermodynamics, but no such rules govern the proportions of Δψ and ΔpH. Those vary with the physiological conditions, as a function of diverse transport systems that let ions move across the membrane and thus interconvert Δψ and ΔpH. The existence of such transport systems was one of Mitchell’s original postulates (165, 167, 169).

The principles are conveniently illustrated by the ionophores (Fig. 3), which enable the investigator to manipulate ΔpH and Δψ almost at will (16, 81). Valinomycin facilitates the movement of K⁺ ions across lipid membranes. When valinomycin is added to cells respiring in K⁺-rich medium, K⁺ enters in exchange for protons expelled by the respiratory chain; Δψ thereby diminishes while pH_i rises. Conversely, valinomycin can be used to generate a transient Δψ in nonmetabolizing cells, as mentioned above. The antibiotic nigericin, which mediates electroneutral antiport of K⁺ for H⁺, has other useful attributes: in the presence of K⁺ ions, nigericin collapses ΔpH, allowing Δψ to rise. Also, the combination of valinomycin and nigericin mediates net movement of protons, just as proton conductors do, and short-circuits the proton current. E. coli, like other bacteria, contains an array of endogenous transport systems that mediate ion exchanges of this kind. Their functions, and in particular their role in the regulation of cytoplasmic pH, are considered below.

Fig. 3Ionophores. Valinomycin catalyzes K+ translocation, nigericin catalyzes the exchange of K+ for H+, and proton conductors ("uncouplers") catalyze the translocation of protons. All these fluxes are reversible. Arrowheads identify the direction of net ion movements expected in a cell that possesses a metabolic proton pump.

Source:

Physiological Work

How much work can a given proton potential do? Work can always be defined as the product of a capacity factor and an intensity factor (electric potential and charge, for example). In physiology, capacity is proportional to the number of protons extruded, i.e., to the current. Estimating the gradient that can be achieved is more subtle.

Consider the proton/sugar symporter depicted in Fig. 1. Assuming that the substrate does not undergo metabolism, that the membrane does not leak, and that the stoichiometry is one proton per molecule of substrate, then at equilibrium

–

/F = –Δ p ≈ 60 log[S]_in/[S]_out (mV) (3)

In other words, a proton potential of –120 mV could support a substrate concentration gradient of 100-fold, and a potential of –180 mV could support a gradient of 1,000-fold. These gradients fall well within the physiological range, and the accumulation of [¹⁴C]lactose by E. coli has, in fact, been used to measure the proton potential (119). It is important to note that the gradient attained is strongly dependent on the proton stoichiometry: with 2H⁺/S, –180 mV could support a gradient of 10⁶-fold!

In oxidative phosphorylation, the proton potential generated by the respiratory chain poises the equilibrium of the ATPase reaction. How effective will this be? Let us write the full reaction catalyzed by the ATPase, thus:

From standard considerations, and neglecting both the participation of water and the appearance of scalar protons as a result of the differing pKs of the nucleotides, one can write an equilibrium constant for the reaction:

(where the square brackets indicate concentrations and the curly brackets give electrochemical activities). The equation can be made more useful by using natural logarithms, multiplying both sides by RT, and rearranging, to give

nRT ln{H⁺}_in/{H⁺}_out = –RT ln K _eq – RT ln [ATP]/[ADP][P_i]

Terms to the right of the equality sign may be replaced by the expression Δ G_p, the "phosphorylation potential," or Gibbs free energy made available during ATP hydrolysis. In the same way, the term to the left may be replaced by its equivalent,

. Accordingly, at equilibrium

Clearly, the electrochemical gradient that can be generated by ATP hydrolysis is set by the proton stoichiometry of the reaction (see also equation 1) and, likewise, that stoichiometry determines the "poise" of the ATP/ADP couple that can be achieved by any given

. If we take to be near –200 mV (199) (equivalent to –19.3 kJ/mol or –4.6 kcal/mol) and n = 3, we calculate that the cells might achieve a Δ G_p of –58 kJ/mol (–13.8 kcal/mol). The observed value of –48 kJ/mol (–11.5 kcal/mol) for respiring cells (119, 149, 150) is thus entirely consistent with expectations. Note that if the stoichiometry were 2H⁺/ATP, the proton potential of –200 mV would not be sufficient to poise the ATPase at the observed phosphorylation potential.

Considerations of Scale

A bacterial cell is a minute object engaged in intense metabolic activity; several aspects of energy coupling by ion currents are best appreciated by expressing them numerically. For this purpose, let us imagine a canonical bacterial cell with a radius of 0.5 μm. In that case, the cell surface area is π × 10^–8 cm² and the cell volume is π/6 × 10^–15 liters.

We should note first that the internal pH is quite tightly controlled at pH 7.5 (see below); it follows that, on average, there are only 10 free protons per cell! Since a single cell contains many thousands of H⁺-coupled pumps or carriers, one may legitimately wonder about the physiological role of H⁺ ions per se. Escape from this apparent paradox requires us to recall that biological substances sample their world on a timescale of milliseconds but that the acid-base dissociation and association reactions of aqueous buffer groups have a picosecond timescale. As a result, enzymatically mediated disappearance or appearance of H⁺ may be readily compensated for by environmental chemistry and the large number of buffer molecules present in a cell. Moreover, in an aqueous medium, the availability of protons to a pump or carrier depends not only on free H⁺ concentration but also (more importantly) on the interaction of water with protein-bound H⁺ donors and acceptors, i.e., on the pKs of relevant H⁺-binding sites. There is no lack of residues with pKs between 6 and 9 in any biological system. Similar considerations must apply to any other ion at such low free concentration in bacteria (e.g., Ca²⁺).

A second consequence of scale deals with the time required to establish or dissipate an ionic gradient. To estimate the magnitude of the proton current generated by a single cell, let us take the maximal respiratory rate to be 12 mmol of O₂ per g (dry weight) per h and the stoichiometry of NADH oxidation to be 6H⁺/O. Since the dry weight of a single cell is 7.5 × 10^–13 g (assuming 1.5 g/ml of cell volume), a little arithmetic brings one to the conclusion that a cell of E. coli pumps about 2 × 10⁷ H⁺ ions per s. Maloney (151), in the first edition of this book, estimated proton extrusion by the F₀F₁-ATPase at a somewhat smaller value, 3 × 10⁶ H⁺ ions per s, but with a revised estimate of F₀F₁ turnover (241), one calculates a comparable number (8 × 10⁶ H⁺ ions per s per cell) for the ATPase.

At this maximal rate, how long would it take to polarize the membrane to its maximum physiological value of –150 to –200 mV? Bacterial membranes, like other biological membranes, have an electrical capacitance of about 10^–6 F/cm². Given this capacitance, one can calculate that to establish a membrane potential of –200 mV over a membrane surface of π × 10^–8 cm², one needs to transfer only 12π × 10³ charges, or about 40,000 H⁺ ions. As little as 5 to 10 ms would suffice to do that job. By contrast, if the pumps supporting the steady state were to stop suddenly, dissipation of the potential gradient by passive proton leakage would take much longer, at least 1 s. Note also that while the Δψ component of the proton potential responds quickly to fluctuations in metabolic rate, the ΔpH component changes but slowly. From the buffering capacity of 50 μmol of H⁺ per pH unit per g of cells (24), one can calculate that expulsion of the 4 × 10⁴ protons per cell needed to charge membrane capacitance would elevate cytosolic pH by only 0.015 unit. Obviously, significant pH changes require a concurrent flux of other ions to compensate for the electrical potential generated by proton extrusion.

Let us note one final consequence of scale. Recent studies suggest that there are stretch-activated ion channels in bacterial plasma membranes, much like those found in eukaryotic cells (20, 159, 222). The trouble is that channels pass ions very quickly, 10⁶ to 10⁹ ions per s, and the small size of bacteria places them at risk of uncoupling if even a single moderately active channel were to open for only part of a second (at 0.15 M K⁺, there are after all only 5 × 10⁷ K⁺ ions in our canonical cell). In fact, just such an uncoupling is thought to be the basis for the toxicity of certain colicins (143, 206). It will be very important to reconcile the evidence for channels with these warning flags hoisted by common sense (see chapter 74 of this volume for more discussion).

A COMPREHENSIBLE CELL

Proton Circulation of E. coli

We now shift focus from the validation of a general principle of energetics to the workings of one particular organism. That calls for no long leap. The microbial world exhibits numerous variations on the theme of energy coupling by ion currents, some of them quite bizarre: light-driven chloride pumps, sodium currents energized by respiration, and proton potentials elicited by anion exchange (80, 152). But E. coli happens to correspond almost exactly to Mitchell’s original vision, and hence the insights and predictions of his hypothesis can be directly applied to questions of cellular physiology.

Figure 4 depicts E. coli as the membrane physiologist sees it. The central feature of the cell is not the genome but the plasma membrane. Energy transduction and work are very largely the business of the plasma membrane; the outer membrane, which has other responsibilities, has been omitted for clarity. The phospholipid bilayer—closed, fluid, and self-sealing—serves as the main barrier to the diffusion of ions and polar molecules; it makes each cell a unit of metabolism and energy transduction. A variety of proteins span the membrane, most of which serve as specific conduits for the passage of selected molecules, ions, or chemical groups from one surface to the other. Some of these translocations are intimately linked to a concurrent chemical reaction, notably the hydrolysis of ATP and oxidation-reduction. We refer to these as primary transport processes and to their catalysts as osmoenzymes. Other translocations entail no exchanges of covalent bonds, only the movement of the substrate(s) down their electrochemical gradient; we designate these as secondary transport processes. What orchestrates the hubbub of individual transport catalysts into a coherent, even purposeful, whole is the global circulation of protons across the plasma membrane. Certain osmoenzymes, particularly the respiratory chain and the F₀F₁-ATPase, draw upon the free energy of the reactions that they catalyze to pump protons outward across the plasma membrane. These reactions can generate the proton potential,

, and represent the supply side of the membrane economy. Secondary transport systems represent the demand side: they "consume" the proton potential and use the free energy inherent in that potential to carry out useful tasks such as nutrient accumulation and motility. E. coli also generates an ancillary circulation of sodium ions and a more specialized anion current as well. Let us look at the component elements in a little more detail, keeping in mind that each derives biological meaning from its membership in a larger ensemble, the whole cell.

Fig. 4Energy transduction by the proton circulation in E. coli. The diagram illustrates generation of a proton current by vectorial electrogenic H+ extrusion through the respiratory chain and completion of the current loop by several paths that perform useful work: F0F1-ATPase (oxidative phosphorylation), pyridine nucleotide transhydrogenase, the flagellar motor, and several kinds of proton-linked transport systems. Also shown is the generation of a secondary sodium current by Na+/H+ antiport and an example of sodium-linked substrate transport. S, substrate.

Source: Reprinted from reference 78 with permission.

Supply-Side Economics

In aerobic cells, the prime mover of the metabolic engine is the respiratory chain. From the biochemist’s perspective, the function of this cascade of redox carriers is to catalyze the exergonic oxidation of diverse substrates: lactate, NADH, succinate, ascorbate, and others. To the physiologist, however, the respiratory chain is a device that translocates protons electrogenically out of the cytoplasm, in order that the free energy of oxidation be conserved in the free energy of the proton potential. E. coli also produces a number of anaerobic redox chains, with nitrate or fumarate as terminal electron acceptors; these also translocate protons outward. The characteristics and sequence of the redox carriers are the subject of chapter 17 of this volume (also see earlier reviews by Poole and Ingledew [193] and by Anraku [10]); our concern here is with the nature of proton translocation.

Figure 5 shows the core of the aerobic chain as it is presently understood, arranged in the form of two redox loops. The principle of the redox loop was an important element in Mitchell’s original formulation of the chemiosmotic hypothesis, presenting a clear example of "vectorial" metabolism. Mitchell (167) envisaged oxidation chains strung across the barrier, such that hydrogen carriers (flavines, quinones) would be reduced at the inner surface. The reduced carrier would then "diffuse" to the outer surface, where protons and electrons separate; the electrons would pass across the membrane by an electron carrier (cytochromes, nonheme iron) while protons would be released into the medium. Note that electrons traverse the membrane as such, but protons do not. It is the consumption of protons by chemical reactions at the cytoplasmic surface and their liberation by chemical reactions at the exterior surface that would bring about the effective translocation of protons across the membrane. The same principle can be seen in the construction of the proton motive cycles discussed at the end of this chapter.

Fig. 5Redox loops and proton pump in E. coli. A hypothetical arrangement of part of the main respiratory chain, under conditions of oxygen sufficiency. The diagram includes two redox loops and a proton-translocating cytochrome oxidase o (198). Oxidation of NADH involves both loops; oxidation of lactate and succinate involves only the second loop.

Source: Modified from reference 79 with permission.

Just how protons pass across the membrane in the course of redox reactions remains uncertain. Redox loops are consistent with what is known of the topology of the respiratory chain, much of it inferred from the meticulous studies of Garland and his colleagues on the nitrate reductase pathway in E. coli (102). Redox loops are also consistent with the finding that respiring cells and membrane vesicles extrude two protons per electron pair when oxidizing lactate or succinate, four when the substrate is NADH (102, 161). More recent measurements, however, indicate that the situation is more complex. E. coli is known to produce two terminal oxidases, cytochrome o in air and cytochrome d at low oxygen tension. Cytochrome d transfers electrons to the inner surface of the plasma membrane where oxygen is reduced, serving as the electron-carrying limb of a redox loop. Cytochrome o does likewise, but in addition it pumps one proton per electron out of the cell (198), as the cytochrome oxidases of mitochondria and some other bacteria do. It appears, then, that the terminal oxidase translocates 4H⁺/2e^– in air but only 2H⁺/2e^– under reduced oxygen tension. Note that proton transfer by cytochrome oxidase o cannot be represented by a redox loop but must be envisaged as direct ("conformational") transport of protons as such. Quite probably, both redox loops and true proton transport contribute to the generation of the proton current.

Figure 6 illustrates the kind of measurements on which Puustinen et al. (198) based their conclusion, and the quantitation of proton translocation in general. Briefly, spheroplasts of mutant strains containing either cytochrome o or cytochrome d were suspended, under anaerobic conditions, in very lightly buffered medium supplemented with KCl and valinomycin. A pulse of oxygen (water saturated with air) was injected at the arrow to initiate oxidation of the substrate (quinol). Thanks to the presence of K⁺ ions and the ionophore, the membrane potential induced by proton extrusion is instantly dissipated, allowing the protons to be registered as a sharp drop in the pH of the suspension. The subsequent slow alkalinization represents diffusive leakage of protons back into the vesicles, whose lumen has become alkaline.

Fig. 6Proton translocation by E. coli spheroplasts. Oxidation of the substrate, ubiquinol, was initiated by injection of air-saturated water. Valinomycin, KCl, and KSCN were present to prevent development of a membrane potential. (A) Strain GV-102, which expresses only cytochrome d; H+/e? = 1.0. Note that proton extrusion was virtually abolished when nigericin was also present; this is because valinomycin and nigericin together catalyze net H+ flow. (B) Strain GO-103, which expresses only cytochrome o; in this case, H+/e? was 1.8.

Source: Redrawn from reference 198 with permission.

The number of protons expelled by the respiratory chain as a whole depends, then, both on the substrate and on environmental conditions. For E. coli, estimates based on studies with the individual respiratory complexes run as high as 8H⁺/2e^– for the oxidation of NADH in air. Evidently, the organisms do not conserve all the energy potentially available, probably because thermodynamic efficiency is less advantageous than rapid turnover of the respiratory chain.

The other major osmoenzyme of the plasma membrane is the proton-translocating F₀F₁-ATPase, or ATP synthase. This remarkable enzyme, whose true nature had first to be grasped by an extraordinary leap of the imagination (165, 167), spans the membrane and couples the chemical equilibration of adenine nucleotides to the transport of protons across the barrier. In anaerobic cells provided with a fermentable substrate (usually glucose), glycolysis supplies the bulk of the ATP. The ATPase then operates in the hydrolytic direction, extruding protons and generating a proton potential across the plasma membrane (Fig. 2). In aerobic cells, the proton potential is a product of electron transport (Fig. 5) and is somewhat larger (more negative) than that generated by glycolysis (Table 1). The ATPase now operates as a synthase, allowing protons to return to the cytoplasm and capturing the free energy of this downhill flux to drive ATP synthesis (oxidative phosphorylation). The F_0F1-ATPase thus occupies a unique and central position in cellular energetics: it mediates the reversible interconversion of the two cellular energy currencies, ATP and

The F₀F₁-ATPase is an uncommonly complex assemblage, almost rivaling the ribosome. The prefix refers to its two domains, first recognized by the late Ephraim Racker in the mitochondrial enzyme 30 years ago. The headpiece, F₁, is visible in electron micrographs as a 10-nm knob that projects from the membrane into the cytosol; F₁ bears the catalytic sites. F₀ spans the membrane, forming a proton channel or carrier that connects the external surface to the headpiece. In the absence of adenine nucleotides, the channel is blocked; only when catalysis takes place do protons pass through the holoenzyme, with a fixed stoichiometry and with a direction linked to that of the chemical reaction.

How do we know that the F₀F₁-ATPase transports protons as a direct consequence of the catalytic process? The most direct evidence comes from the kind of experiment illustrated in Fig. 7A. When everted (i.e., inside-out) membrane vesicles were allowed to hydrolyze ATP, the pH of the suspension rose while the vesicle lumen turned more acidic. Proton translocation was blocked both by proton-conducting uncouplers and by N,N '-dicyclohexylcarbodiimide (DCCD), an inhibitor of the ATPase (89, 237). Parallel results were obtained with purified ATPase inlaid into liposomes, with the catalytic site facing the medium (108, 175, 215). Figure 7B depicts an indirect but more impressive result. If the ATPase translocates protons, an applied proton potential should poise the equilibrium of the reaction, as discussed in the preceding section. The experiment documents that starved cells synthesize ATP in response to an electrical potential, artificially imposed with the aid of valinomycin. A membrane potential, a pH gradient, or a combination of the two were equally effective, and ATP synthesis was prevented by uncouplers and by DCCD (147, 239). Here again, analogous experiments have been carried out with purified F₀F₁-ATPase reconstituted into liposomes.

Fig. 7Indicators of proton translocation by the F0F1-ATPase of E. coli. (A) Transient alkalinization of the medium during ATP hydrolysis by everted membrane vesicles. (a) Vesicles were suspended in anaerobic, lightly buffered medium, pH 6.13, in the presence of valinomycin, KCl, and the proton conductor CCCP. Upon addition of ATP, proton production begins immediately. (b) As for curve (a), but in the absence of CCCP. Note the transient alkalinization upon ATP addition, due to transport of protons into the vesicles. Reprinted from reference 89 with permission of Academic Press, Inc. (B) ATP synthesis in response to an artificial membrane potential. Cells of a parent strain and of an ATPase-negative mutant were starved to reduce the basal ATP level nearly to zero, then resuspended in potassium-deficient buffer with valinomycin. Transient ATP accumulation was seen in the parent cells but not in mutant cells.

Source: Reprinted from reference 153 with permission.

How many protons are translocated in each catalytic cycle? In principle, there are two entirely dissimilar experimental approaches to this question. One is to measure proton translocation and ATP hydrolysis concurrently in experiments such as that in Fig. 7A. Pioneering measurements of this kind, made on mitochondrial membrane vesicles by Mitchell and Moyle (173), indicated a stoichiometry of 2H⁺/ATP. This value was in agreement with Mitchell’s theoretical expectations, and appeared at the time to be consistent with the stoichiometry inferred from the second method. This calls for measurements of

and of the concentrations of ATP, ADP, and P_i in the steady state; equation 4 can then be used to calculate n. More recently, however, measurements of both kinds but with improved methodology have converged on a stoichiometry of 3H⁺/ATP (119, 120, 150). Similar values have been obtained with mitochondria, chloroplasts, and several bacteria, and 3H⁺/ATP is now generally accepted. This stoichiometry is in accord with the threefold symmetry of the headpiece and with our growing understanding of the reaction mechanism, which are discussed below.

The place of the F₀F₁-ATPase in the cellular economy is nicely illustrated by a series of E. coli mutants, first described by F. Gibson, G. Cox, and their colleagues in 1971, which grew on fermentable substrates but failed to grow on oxidizable ones (succinate, for example) (30). These mutants were called unc, for uncoupled, and proved to lack a functional ATP synthase. In consequence, they were unable to generate ATP by oxidative phosphorylation even though redox reactions were unimpaired. When techniques were devised to measure internal pH and membrane potential, it was found that the wild type can generate a proton potential at the expense of either respiration or ATP hydrolysis. unc mutants could still utilize respiration for that purpose, but not ATP hydrolysis (40, 67). Mutants of this kind contributed notably to clarification of the uses to which cells put their two energy currencies, ATP and . Incidentally, despite the physiological importance of the F₀F₁-ATPase, it does not serve as a metabolic control point: modulation of expression of the unc (atp) operon has minor effects on growth rate or yield (101).

Uses of

Turning now to the demand side, we note (Fig. 4) diverse molecular devices that perform useful work at the expense of the proton potential. Most of these are "secondary," in the sense that they entail no covalent bond exchanges, but a few are "primary" because they use to drive a chemical reaction. Conceptually speaking, the simplest are the various porters, membrane transport proteins that couple the downhill flux of protons into the cell to the uphill flux of some metabolite inward or outward. Following Mitchell’s terminology (168), which has been generally adopted by microbiologists, we distinguish three kinds of porters (see chapter 74 of this volume). (i) Uniporters transport a substrate down its electrochemical gradient, unlinked to any ion. The glycerol uniporter of E. coli is one example, but there are not many. (ii) Symport is exemplified by the β-galactoside "permease," discussed in the preceding section. It is solidly established that this porter mediates the movement of two substrates in the same direction (sym), a molecule of sugar plus one proton (105, 236). The obligatory coupling between the two fluxes ensures that the electrochemical driving force upon the proton powers the concurrent accumulation of the sugar. Another case in point is the accumulation of K⁺ ions; this is believed to occur by symport with H⁺, effectively making potassium divalent (17, 18). (iii) Antiporters are so articulated that the flux of protons into the cell powers the efflux of some substrate from the cytoplasm. The most important of these is the Na⁺/H⁺ antiporter, which expels sodium ions (45, 188, 216, 237a). Extrusion of Na⁺ in the face of the membrane potential and of the ion’s concentration gradient establishes an electrochemical potential gradient of Na⁺ ions, which supports a sodium circulation ancillary to that of protons. The sodium circulation and other special circuits will be taken up below; the characteristics of bacterial transport systems are the subject of chapters 74 through 76 of this volume.

A very different work function energized by the proton circulation is the flagellar motor. The operation and molecular design of the world’s most ancient rotary engine are discussed in detail in chapter 10 of this volume. Suffice it here to note the early discovery (130) that the energy source for E. coli motility is not ATP but what was then still referred to as the "energized membrane." Just a few years later, investigators working with several other bacteria demonstrated that the motor is in fact driven by the proton current (68, 157, 160). In all cases, the experiments hinged on showing that an imposed proton potential could energize the motility of starved cells, at least briefly. It is estimated that each rotation of the motor requires the passage of 10³ protons.

The plasma membrane of E. coli, like that of many other bacteria and also of mitochondria, contains an enzyme that reversibly transfers 2H between NADH and NADP⁺. In membrane fragments, the equilibrium constant of the reaction is not far from unity, but when a proton potential of the usual polarity is present, the equilibrium shifts sharply in the direction of reduction of NADP⁺ to NADPH. Early on, Mitchell identified the NADH/NADP⁺ transhydrogenase as yet another proton-coupled osmoenzyme. Passage of one proton, or possibly less than one, is stoichiometrically linked to the transfer of 2H from NADH to NADP⁺ (23, 100). The function of the transhydrogenase is believed to be the production of NADPH, the chief reducing agent for biosynthesis, by cells growing on oxidizable substrates. Somewhat surprisingly, mutants lacking the transhydrogenase grow perfectly well, suggesting that the transhydrogenase is but one of several sources of cellular reducing power; it may play a special role in glutamine biosynthesis (74, 137).

The foregoing list does not exhaust the functions of the proton circulation. The secretion of membrane proteins is known to require a proton potential, though the reason is still unclear (19, 50). Additional functions may well be discovered in the future, especially in the area of homeostatic regulation, but the path of discovery is beset with pitfalls! Ionophore effects can be particularly misleading, as these reagents perturb the regulation of cytoplasmic pH.

Ancillary Ion Circulations

In E. coli, and indeed in most of the common bacteria, the chief link between metabolism and work is the proton circulation. But protons are not the only coupling ions available to cells, and we know many cases of work functions that are coupled to a flux of sodium ions or of anions. In E. coli, these are ancillary to the proton circulation in the sense that the proton potential generates the gradients of sodium or of anions (Fig. 4). However, we may yet discover that E. coli, like many other bacteria, can under conditions of metabolic stress generate a sodium current with the aid of a primary sodium pump without relying on the proton potential (13, 45, 216).

The link between the circulation of H⁺ and that of Na⁺ is the sodium/proton antiporter, anticipated by Mitchell (167) and first documented by West and Mitchell (273a) and by Harold and Papineau (84). All the evidence suggests that in E. coli, a secondary antiporter maintains the cytoplasmic sodium concentration at 1/10 of that of the medium. Two distinct porters have been identified (34, 45, 209), one of which, NhaA, appears to have a stoichiometry of 1Na⁺/2H⁺. External Na⁺ ions are thus subject to a sodium motive force quite analogous to the proton motive force (equation 2). E. coli contains sodium-linked symporters for the uptake of proline, melibiose, and glutamate (see chapter 74), but the critical function of the Na⁺/H⁺ antiporter is likely to be the regulation of cytoplasmic pH, which will be taken up below.

Chemiosmotic energy coupling commonly relies on cations, but there is no reason in principle why anions or uncharged molecules may not serve as coupling agents for special purposes. Examples from chloroplasts and mitochondria have been known for some time and have now been found in E. coli as well (152). For instance, the uptake of glucose 6-phosphate, originally interpreted as a case of proton symport, is actually effected by electroneutral exchange of the ester for intracellular P_i. The nature and significance of anion exchanges will be discussed more fully in chapter 74.

Finally, we would briefly mention one additional current, not fully established but all the more tantalizing. It is well known that in eukaryotic cells, signals are often carried across the plasma membrane by a flux of calcium ions. Recent reports from Adler’s laboratory suggest that, contrary to conventional wisdom, something of this kind may be true of bacteria as well. In E. coli, an influx of Ca²⁺ ions elicits tumbling and may thus play a role in chemotaxis (226, 227). If this is correct, Ca²⁺ ions probably flow down an electrochemical gradient generated by a calcium/proton antiporter or by a primary calcium pump.

Homeostasis

From the standpoint of chemiosmotic energy transduction, the pH gradient and the membrane potential are additive and interconvertible components of the proton motive form (equation 2). But normal cellular operations demand that the cytosolic pH be kept within relatively narrow limits for optimal results. E. coli, for example, maintains a cytoplasmic pH of 7.6 to 7.8 over a wide range of external pH values (187, 189). Figure 8 shows some typical data (119). In cells growing in acidic media, ΔpH is large and makes a major contribution to the proton potential. In alkaline media, Δψ dominates the proton potential and ΔpH diminishes, turning negative above pH 7·5. Evidently, E. coli is not much concerned about Δψ and ΔpH but keeps the internal pH nearly constant. The same conclusion is drawn from experiments in which the external pH was abruptly shifted by the addition of acid or of alkali. The cytoplasmic pH responded as expected but returned to near neutrality within a few minutes.

Fig. 8The proton potential of growing cells of E. coli varies with the pH of the medium. The membrane potential was estimated from the uptake of triphenylphosphonium ion, and ?pH was estimated from the distribution of benzoate. The values of ?p and pHi were calculated from these data.

Source: Reprinted from reference 119 with permission.

This is a clear example of cellular homeostasis. E. coli apparently has a set pH_i that it "endeavors" to maintain in the face of perturbations, both metabolic and environmental. In this instance, something is even known of the underlying mechanisms (24, 209). Acidification of the cytoplasm stimulates proton extrusion; uptake of K⁺ compensates for the buildup of Δψ, allowing the cytosolic pH to rise. Alkalinization of the cytoplasm involves at least two mechanisms. One is Na⁺/H⁺ antiport, serving now not so much to expel Na⁺ but to admitH⁺ ions in a controlled manner that avoids uncoupling the proton current. This function of the antiporter has been documented by studies on mutants defective in the synthesis of one or both of the antiporter proteins (209). It implies that Na⁺ ions should be required for growth in alkaline media, and McMorrow et al. (162) found that to be the case. But there is also evidence for a sodium-independent mode of pH regulation, which has been attributed to a K⁺/H⁺ antiporter. Somewhat surprisingly, examination of this literature leaves one with a strong sense that much remains to be learned about the how and the why of cytoplasmic pH regulation. Indeed, we consider this to be one of the important unresolved issues in this field.

The regulatory perspective may also be the appropriate one for reflection on the function of the transhydrogenase. Growing cells of E. coli poise the NADPH/NADP⁺ ratio near unity, while that of NADH/NAD⁺ is only 0.1 (9). The difference is no doubt related to the biological function of NADPH, which serves as the major source of reducing equivalents for biosynthesis. The transhydrogenase, which catalyzes the transfer of 2H between the pyridine nucleotides, can thus mediate both the oxidation of NADPH and its formation, but its true function may be best described as homeostatic. The mitochondrial enzyme, at least, is poised to provide a protective buffer against the dissipation of either cellular redox power or the mitochondrial energy supply (93). Perhaps the E. coli enzyme likewise serves as one strand in a network of parallel redox reactions that collectively ensure a stable supply of reducing power in an ever-changing metabolic environment.

E. coli, like other bacteria and cells in general, accumulates K⁺ ions and excludes Na⁺. Just why they do this has never been quite resolved, perhaps because there is no single simple answer that would satisfy our penchant for linear thinking. It is generally agreed that K⁺ ions make more "compatible" solutes than Na⁺ ions, in the sense that the former are less disruptive to cellular macromolecules and their associated water shells (238); and it has been known for 30 years that protein synthesis, in particular, requires high concentrations of K⁺ ions (141). Extrusion of Na⁺ ions is, of course, the basis of any sodium circulation and a favorite element in pH regulation. But there is also a body of evidence to suggest that high [Na⁺]_i is intrinsically inhibitory to growth, though specific targets have not been identified (86, 183, 188). Finally, the relatively massive reservoirs of intracellular K⁺ and extracellular Na⁺ constitute a kind of energy storage that can, if necessary, power motility and transport (49). Is buffering of the proton potential the chief reason for K⁺ accumulation? Presumably not, but stabilization of the

may well be a fringe benefit that makes the difference under some conditions.

Finally, consider the regulation of the membrane potential or of the proton motive force. As far as is presently known, neither parameter is subject to control (Fig. 8). But then, what happens if cells are suspended in buffer with a good supply of metabolic energy but no useful work to do? The answer must be sought in mechanisms that dissipate the excess energy as heat. Futile cycling of K⁺ ions, a glutamine cycle, and a polyphosphate cycle have all been proposed, but the simplest mechanism would be a proton cycle analogous to that employed by the mitochondria of adipose tissue to generate heat (79). Taylor and Jackson (225), working with Rhodospirillum rubrum, have provided evidence for the existence of a controlled proton channel that opens whenever Δψ exceeds a threshold, thus short-circuiting the proton current. E. coli may well do likewise: a controlled proton leak would at least explain how it comes about that the measured proton potential is so much lower than expected from thermodynamics. Such energy dissipation is not necessarily wasteful; on the contrary, by encouraging rapid turnover of the respiratory chain, it should maximize power output and accelerate biomass production.

With the proton circulation woven into so many facets of the bacterial economy, one might expect it to be an obligatory feature. Somewhat surprisingly, that is not the case. Harold and van Brunt (86) showed almost two decades ago that enterococci can grow in the presence of the powerful cation conductor gramicidin, with Δ p, Δψ, and ΔpH all set at zero, provided allowance is made for their disabilities. The cells required a medium of high K⁺ concentration and low Na⁺ concentration; a pH not far from 7; and high concentrations of extracellular amino acids. But as long as they were not stressed, their growth rate and morphology were normal, showing that the proton circulation is not needed to produce the fabric of the bacterial cell. The same apparently holds for E. coli cells (127); they could even cope with osmotic stress despite an uncoupled proton circulation (184). All this probably tells us something fundamental. The proton circulation is required for the cell to keep its milieu interieur constant and different from its environment. In the laboratory, we can arrange matters so that this function is dispensable; however, in the real world it is essential.

F₀F₁AND OTHER ION MOTIVE PUMPS

In the rest of this chapter, we focus mainly on transport systems linked to ATP, especially the F₀F₁-ATPase. At the end, however, we broaden the perspective and introduce some newly discovered and unorthodox mechanisms used to establish a proton potential.

Nomenclature: P, F, and V

Only a few systems directly couple ion transport to the scalar events of intermediary metabolism. Included in this select group are the devices associated with electron transport (see chapter 17) and several Na⁺-translocating decarboxylases (46), but apart from these, the main ion pumps in the microbial world, as elsewhere, are linked to the hydrolysis or synthesis of ATP.

We recognize two major classes of ATP-linked ion motive pumps. On the one hand, there are the P-type ATPases (formerly E₁E₂), so named because they are phosphorylated during their reaction cycle; these are discussed below, but only briefly. The second major class contains the F₀F₁-ATPases, of which there are two distinct subgroups: F and V types. The former (F-F₀F₁) is composed of the enzymes, all related, that take part in oxidative and photosynthetic phosphorylations in eubacteria and their descendants, the mitochondria and chloroplasts. The latter subgroup (V-F₀F₁) includes a similarly coherent collection of enzymes located mainly in the archaebacteria and in the endomembrane or vacuolar system of the eukaryotic cell (177). F- and V-type ATPases show an obvious evolutionary kinship, and reconstruction of this relationship suggests that their divergence correlates with appearance of the eubacterial lineage (126, 177; also see reference 155). As a general rule, then, F-type ATPases are found in the eubacteria and their descendants, while V-type ATPases populate archaebacterial and eukaryotic domains. Of course, there are exceptions. For example, Kakinuma and coworkers have described the ntp gene cluster in the gram-positive eubacterium E. hirae. Sequence analysis makes it clear that ntp encodes a Na⁺-translocating V-ATPase that is expressed when the normal F-ATPase is inoperative (111, 223, 224). This discovery suggests we might uncover new principles of regulation by studying how E. hirae coordinates the activities of two ion motive ATPases, each with its own chemical specificity.

P-Type ATPases

P-type ATPases hydrolyze ATP to move one or more cations inward or outward or in exchange (H⁺, Na⁺,K⁺, Ca²⁺, Mg²⁺, Cu²⁺, and Cd²⁺ are known substrates). The main player in the reaction is the ca. 100-kDa catalytic subunit, usually designated α, which, as part of the enzymatic reaction cycle, is phosphorylated at a conserved aspartyl residue. Ion transport therefore reflects the cyclic transformation between phosphorylated and dephosphorylated forms of the enzyme, driven by formation and destruction of the high-energy acylphosphate bond (briefly reviewed by Stokes and Nakamoto [221]). P-ATPases are of simple composition relative to the F- or V-ATPases, most often with one (α) or two (αβ) subunits and never with more than three (αβγ). Thus, with its single α subunit, the Mg²⁺ ATPase of Salmonella typhimurium is the more typical case (219), while the Kdp (K⁺)^-ATPase of E. coli is the one known example of a trimeric P-ATPase (54). In all these cases, the catalytic (α) subunit is an integral membrane protein with 8 or 10 transmembrane segments (the latter is more often cited). The β subunit, when present, has only a single transmembrane segment, so it is usually assumed that α, which contains the conserved aspartate and the nucleotide-binding domains, also carries determinants of ionic specificity and selectivity, an idea supported by the properties of chimeras of the eukaryotic Na⁺/K⁺- and Ca²⁺-ATPases(142).

All P-type ATPases share extensive amino acid sequence homology, extending well beyond that found at the active site containing the phosphorylated aspartyl residue, and this has prompted computational efforts to establish sequence relationships. That exercise defines four large clusters (56) composed of, respectively, Ca²⁺- and Mg²⁺-motive ATPases (cluster 1); K⁺/H⁺- and K⁺/Na⁺- exchange pumps (cluster 2); electrogenic H⁺-ATPases of fungi and plants (cluster 3); and H⁺-, Cd²⁺-, and K⁺-ATPases, as well as the Cu²⁺-ATPases whose defects are associated with Menkes (228) and Wilson’s (29) diseases (cluster 4).

F₀F₁—the Major Proton Pump in E. coli

The main bacterial ion pump is the H⁺-translocating F₀F₁-ATPase, which, because of its central role in energy transductions, is the subject of a number of useful reviews covering various aspects of structure and function (32, 44, 58, 109, 149, 151, 171, 210, 221). The biochemical description of F-F₀F₁ is essentially complete, and we have come to appreciate that even in its simplest version (the E. coli enzyme) it is extraordinarily complex. F₀F₁-ATPase has at least eight distinct subunits, several in multiple copy, so a coordinated operation of nearly two dozen separate elements is required to couple the flux of H⁺ to ATP synthesis or hydrolysis. Until recently, the task of understanding these relationships was undertaken without detailed knowledge of structure. Fortunately, this situation has changed. In the fall of 1994, Abrahams et al. presented a 2.8-Å (0.28-μm) structure of F₁ from bovine heart mitochondria (1). Even though this gives us but one snapshot of part of a complex protein that adopts several conformations, it is clear that the style of study in the next decade will be qualitatively different and enormously more satisfying from that which has come before.

Sector Organization.

The overall architecture of bacterial F₀F₁ is shown in Fig. 9; a schematic is shown next to it. The enzyme has two main domains, as noted above—the F₀ and F₁ sectors—each a distinct biochemical entity. F₁ has five subunits, organized as an α ₃ β ₃ γδε oligomer, with an aggregate molecular mass of 382 kDa. F₀, integral to the membrane, has three subunits (ab ₂c_10±1) and a mass of about 150 kDa. Separation of F₁ from F₀ is paralleled by a functional division that reveals their individual properties. Thus, solubilized F₁ hydrolyzes ATP, while membranes stripped of F₁ have greatly elevated H⁺ permeability. The disengagement is reversible, and rebinding of F₁ restores the normal coordination of the two half-reactions. The central biochemical issue is therefore the following: how are these partial reactions "coupled" so that H⁺ flux through F₀ is stoichiometrically associated with ATP synthesis or hydrolysis? This question has been asked for three decades, but when it comes to offering explicit answers of how "coupling" occurs, we are nearly as much in the dark now as we were then.

Fig. 9Two views of F0F1. (Left) Side view of E. coli F0F1, obtained by averaging many images from cryoelectron microscopy. The roughly globular F1 sector is linked to the membrane-bound F0 sector by a 45-Å (4.5-nm) stalk. Reprinted from reference 32 with permission. (Right) Sector organization of F0F1. At pH 7 and physiological ionic strength, the dynamic equilibrium between F0, F1, and F0F1 very strongly favors the fully assembled complex (see the text). Each of the separated sectors has its characteristic activity, as indicated at the right. In the diagram at the top right, F1 has been rotated so that the surface interacting with F0 faces out of the page.

Source: Reprinted from reference 151 with permission.

Physiological Position.

With few exceptions, F₀F₁ is found on all bacterial membranes. In obligate or facultative aerobes, it functions to synthesize ATP during oxidative phosphorylations, while under anaerobic conditions F₀F₁ itself initiates the H⁺ circulation. Our acceptance of this unusual dual role follows from several observations. Validation of the chemiosmotic hypothesis came in part from demonstrations that electron transport generated membrane potential and a pH gradient (6, 70). The companion experiments showed that F₀F₁ mediated ATP synthesis in response to imposed electrical and/or chemical gradients comparable in size to those found in vivo (Fig. 7) (153, 239) and that such synthesis was linked to H⁺ entry (146, 147). Correspondingly, it was shown that F₀F₁ linked ATP hydrolysis to H⁺ translocation (Fig. 7) (89, 237) and that this reaction sustained a membrane potential or a pH gradient (82, 83, 84). Such early observations proved that F₀F₁ could take part in a proton circulation, and subsequent work addressed two quantitative issues (see below) approached most easily by using the intact cell. The first group of experiments dealt with the value of stoichiometry (nH⁺/ATP); the second asked whether ATP synthesis occurs at equal rates when driven by thermodynamically equivalent electrical or chemical gradients. This latter topic is relevant to the function of F₀ and its relation to a theoretical structure known as a "proton well."

Stoichiometry of Coupling.

In studies of chemiosmotic proteins, questions about the stoichiometry of ion coupling are central to mechanistic proposals. In part for this reason, such questions were a significant and often controversial topic with regard to F₀F₁. But these issues are now settled, and the work with bacterial systems is as good as any available, perhaps better.

The best estimates of stoichiometry, nH⁺/ATP, derive from experiments based on the thermodynamic relationships noted above:

n Δ p ≥ Δ G_p/F when H⁺ movements drive ATP synthesis, or (5a)

n Δ p ≤ Δ G_p/F when ATP hydrolysis supports H⁺ transport (5b)

where equation 5a refers to respiring cells and equation 5b describes the anaerobic state. Kashket (119, 120) applied both relationships to E. coli, in experiments in which independent measurements of both Δ p and Δ G_p were available. For respiring cells, stoichiometry fell near 3H+/ATP (n ≥ 2.7); similar findings were made for anaerobic cells (n ≤ 3.1). Maloney (150) examined the same parameters in the anaerobe Streptococcus lactis (now Lactococcus lactis), concluding that F₀F₁ operation used ca. 3H⁺/ATP (relationship 5b) over the entire range of Δ p and Δ G_p/F to be expected under physiological conditions. These approaches have also been used to characterize the mitochondrial and chloroplast ATPases, and there, too, stoichiometry is best understood as near 3H⁺/ATP (21, 196). Among all these measurements, we would stress that those in E. coli (119, 120) have special importance, because the value of the proton motive force was established by two very different methods. In one case, Δ p was calculated, as is customary, from the distribution of indirect probes that separately estimate membrane potential and pH gradient. The alternative experiments used lactose accumulation via the H⁺/lactose symporter (LacY) to infer Δ p as a single quantity (Table 1). That the two approaches agree so closely is strong evidence that Δ p was accurately measured, a reassurance difficult to obtain in other systems.

While it is agreed that F-F₀F₁ has a net stoichiometry near 3H⁺/ATP, note that fractional stoichiometry is feasible, at least in principle. A complete reaction cycle for the F-type ATPase requires synthesis (hydrolysis) of three molecules of ATP (see below), so that a molecular stoichiometry would approximate 9H⁺/3ATP per turnover. Since it is unlikely that current techniques (above) can distinguish between predictions made by stoichiometries of 8, 9, or 10H⁺/3ATP, the true value will probably be inferred from other kinds of information. For example, the number of protons translocated per turnover is probably the same as the number of subunits c within F₀ (1H⁺ per subunit c) (58, 59). Subunit c stoichiometry is presently set at 10±1/F₀ for F-type F₀F₁ enzymes (58), and when this value is known with certainty, we will also probably know the H⁺/ATP stoichiometry. (In the eukaryotic V-F₀F₁, the subunit c analog has a lower copy number, so the two ATPase families may have different stoichiometries.) Presently, then, the main contribution of experiments showing stoichiometry as ca. 3H⁺/ATP (above) has been to rule out models that require 2H⁺/ATP (6H⁺/3ATP) (summarized by Maloney [149] and Mitchell [171]).

Genetic and Biochemical Organization of F₀F₁

unc Operon.

As noted above, the genetic study of F₀F₁ began in 1971 with the isolation of uncoupled (unc) mutants (30). These findings were the start of elegant genetic experiments (see references 67 and 210) which located the unc operon near 83 min on the E. coli linkage map and used F'-mediated complementation to partition various mutations into the cluster of unc genes (47, 67) (Fig. 10). Correspondence between specific genes and F₀F₁ polypeptides was established first for the major F₁ subunits (51, 212), but in short order gene-polypeptide relationships were completed (28, 48, 72). The unc operon contains nine genes, uncIBEFHAGDC, which encode (in order) an operator-proximal protein of unknown function (UncI) followed by the eight F₀F₁ subunits: a, c, b, δ, α, δ, β, and ε. Mutants defective in any one of the F₀F₁ subunits are now available.

Fig. 10Genetic organization of the unc operon. The unc operon lies at 82.5 min on the E. coli chromosome (clockwise to the left). Sector, gene, and subunit designations are shown. The number of amino acid residues in each polypeptide is indicated.

Source: Reprinted from reference 151 with permission.

The aim of such work had been to apply genetics as an adjunct to biochemical study, and the necessary intermediate goal of defining the unc operon nucleotide sequence was achieved by independent work in three laboratories (65, 66, 113, 114, 115, 116, 144, 179, 204). This analysis immediately highlighted several issues, none as yet resolved. (i) For example, it is of striking but uncertain significance that gene order segregates genes encoding F₀ subunits (abc) from those encoding F₁ subunits (αβγδε) (230), here and in most other organisms. (ii) More puzzling is that the first gene, uncI, encodes a protein of unknown function. On the basis of amino acid content (66) and physical behavior (28, 207), we know that UncI is a membrane protein, expressed at low copy number (205), but we find no role for UncI in either the biochemical functioning, assembly (in vivo or in vitro), or expression of F₀F₁ (28, 229). (iii) Note also that the stoichiometry of F₀F₁ subunits is highly variable—some in single copy, some in multiple copy—but the unc operon is transcribed as a single polycistronic mRNA. Is gene expression regulated so as to produce the corresponding ratios of each subunit? We do know that a differential gene expression exaggerates synthesis of subunit c (10 ± 1 copies per mol of F₀) (28), supporting the idea (27) that mRNA secondary structure can regulate translation efficiency. And codon usage correlates roughly with tRNAs of appropriately major or minor abundance (113, 114), again suggesting a simple mechanism for regulating gene expression. (iv) Finally, because the F₀ and F₁ sectors can be assembled independently of one another (11, 128), one wonders if their individual activities are controlled so as to avoid unproductive ATP hydrolysis or uncoupled proton entry (39, 67). Little direct information addresses this theoretical problem, but we do know that K _eq for the reaction F₀F₁ ↔ F₀ + F₁ is about 25 nM (61) and that the reaction is completed rather quickly (minutes) at the high ionic strength expected of the in vivo event. As a result, there may be no need to regulate sector association, except perhaps for rapidly growing cells. And if one seeks only to regulate F₀ (the H⁺ leak), it may be necessary only to modestly underproduce subunit c in relation to its final stoichiometry, lowering levels of unoccupied F₀ because of an elevation in the pool of free F₁.

F₁ Sector.

F₁, contributing about 70% of total F₀F₁ mass, appears in electron micrographs as a roughly spherical complex of ca. 100 Å (10 nm) in diameter (Fig. 9) (summarized in references 1 and 32). It has five different subunits, organized as an α ₃ β ₃ γδε oligomer (107, 210). This subunit organization is also found for F₁ in chloroplasts and mitochondria. (In mitochondrial F₁, the smallest subunit [ε] corresponds to the bacterial δ protein, while the oligomycin-sensitivity-conferring protein, OSCP, occupies the next larger [δ] position.) Topological relationships among the F₁ subunits have not yet been completely clarified, but X-ray crystallography has produced striking images of the enzyme from bovine heart (1, 1a) and rat liver (22) and cryoelectron microscopy has yielded equally dramatic views of the bacterial form (32). The most detailed view comes from the bovine heart F₁, which shows the nucleotide-binding subunits (α and β) arranged in alternating fashion around a long central core, much like segments in an orange (1). At the center of this encircling ring of αβ units, the C-terminal portion of the γ subunit forms a single 90-Å (9-nm) α-helix passing through the entire F₁ ensemble. Beginning about halfway down to the F₀ surface, this helix interacts with a second α-helix, arising from the γ N terminus, forming a coiled coil that extends to form a protuberance (1) contributing to the stalk structure seen in cryoelectron microscopy (Fig. 9) (32). Even though the positions of subunits δ and ε are not resolved, it is apparent that these and other studies are establishing the structures that define the various conformation states that accompany F₁ function (1, 32). It is on this kind of scaffold that one must now superimpose a mechanism of coupling.

This crystallography (1) validates the early biochemical efforts of Kagawa’s laboratory, which had recombined, in all possible combinations, the F₁ subunits of the thermophile PS3. This approach identified α and β as the subunits most relevant to nucleotide interactions (107, 108), and subsequent efforts in E. coli showed αβγ as the minimal unit of ATP hydrolysis (52, 53). That α and β are each important was further reinforced by the finding that F₁ function is blocked when a single α or a single β is defective (134, 182). It is also clear that F₁ has six nucleotide- binding sites, one associated with each of the major subunits (1, 210). The study of E. coli, in vitro and in vivo, shows that nucleotide bound to α does not participate in energy-linked reactions but is stable during the course of many turnovers (145). For this reason, α may bind nucleotides for regulatory or structural reasons, in contrast to the catalytic site on the β subunit. This reasoning is now reinforced by the crystal structure, which offers reasons why sites on β could behave catalytically while those on α would not (1). Further evidence pointing to the catalytic role of β comes from the analysis of E. coli uncD (β) mutants, some of which show altered metal (Ca²⁺ or Mg²⁺) selectivity (99, 112, 117, 181, 213).

Crystallography (1) has also verified that nucleotide binding in F₁ occurs with the aid of two short motifs often found in nucleotide-binding proteins (231). These motifs, which Walker and his colleagues had identified by examining the E. coli sequence, are located at residues 149 to 156 (Walker A) and 230 to 242 (Walker B) in subunit β (similar motifs are found in α). The A motif coordinates with the terminal phosphates on ADP and ATP; the B motif may help coordinate the associated Mg²⁺. Much site-directed mutagenesis confirms the impact of the Walker motifs (99, 186, 190, 214), and for a large set of such mutants, Senior and his colleagues can show that the entire range of phenotype is accommodated by a single kinetic and thermodynamic framework (5a, 210; revised in reference 211).

In E. coli, subunit γ is required to reconstitute the minimal hydrolytic unit (αβγ), and for this reason it had been assumed that γ provides a frame around which α and β can organize themselves. This is seen quite vividly in bovine heart F₁ (1), and the crystallographic finding is again reflected in the behavior of certain bacterial mutants. The latter work shows that cells lacking the gene for γ fail to express the major F₁ subunits (α + β), although the αβγ triad can be formed by cells lacking genes for either of the remaining F₁ subunits (61, 117, 128). One might recall that Kagawa had suggested that as well as acting to organize F₁, γ might play a role in "coupling," since H⁺ movement is suppressed when the three minor F₁ subunits (γδε but not δε) are bound to F₀ (243). The idea that γ acts to monitor H⁺ movement (107, 109) is further supported by the finding of mutants of γ in which H⁺ flux is dissociated from ATP hydrolysis (98). As well, cryoelectron microscopy shows that γ alters its position relative to β according to whether ADP or ATP is present at the catalytic site, an asymmetry also evident in the F₁ crystal structure (1). The negative effect of γ-β cross-links, but the null effect of this modification in the absence of cross-linking, also implies that these subunits move relative to one another during catalysis (3, 4). Perhaps most striking, these latter experiments place the N terminus of γ near the catalytic site in subunit β (32), a deduction consistent with the crystal structure (1).

Other than a demonstration that it is essential for proper attachment of the αβγ core to F₀ (53, 62), there is little concrete information concerning the role of subunit δ. On the other hand, there is considerable success in the analysis of subunit ε. It is quite clear that ATP synthesis (hydrolysis) coupled to H⁺ movements requires the participation of ε. It has also been suggested that ε functions as does the mitochondrial F₁ inhibitor protein, since binding of ε to ε-depleted F₁ will somewhat inhibit ATPase activity (220; see also reference 129). While this may be so, more recent work points to a different, more interesting role for ε. Just as there is repositioning of γ (see above), there is a shift of ε relative to the major subunits, according to the nucleotide present on β. In one case (ATP), ε lies close to β; in the other case (ADP), ε lies nearer to α (summarized in reference 32). The facts that γ and ε are each in the central core of F₁ and that each changes position relative to the surrounding αβ oligomer suggest an interdependent set of conformational changes. It will be especially exciting, therefore, to follow up on the recent crystallization of a γε complex by Cox et al. (38).

In sum, catalysis occurs on the β subunit, and whether product (ATP) or reactants (ADP + P_i) are bound seems to determine the conformational state of a γε doublet, placed central to an encircling αβ oligomer. Since it is likely that the ε subunit in some way "monitors" events in F₀ (see below), it appears that an understanding of the limiting conformational states of F₁ are close at hand. The challenge is now to link this understanding to the phenomenon of "coupling."

F₀ Sector.

F₀ is less well understood than F₁. We do know that bacterial F₀ has three subunits (a, b, and c, in order of decreasing mass), that the active complex has a stoichiometry of ab _2c10?1, and that all three subunits are required for the H⁺ permeability pathway (11, 60, 88). On the other hand, the arrangement of these subunits relative to one another is still not clear. Presumably, subunits c form an oligomeric structure. But is this a ring encircling an ab ₂ complex, much as the three αβ pairs are arranged around γδε? Or do F₀ subunits form independent rather than interdependent structures? Both arrangements have been discussed previously (44, 59).

On the basis of predictions from the primary sequence, the a subunit has multiple transmembrane α-helices, but the exact number is still debated; an even number (six or eight) is favored by the results of fusions with alkaline phosphatase (135, 136; also see reference 59), but this is not universally accepted (44). It is agreed, however, that certain residues in subunit a, particularly R-210, are absolutely required for F₀ function. The reason is not entirely clear, and two general possibilities have been suggested. For example, R-210 might itself participate in proton conduction, perhaps as part of a "proton relay" involving other subunit a residues, such as E-219 and H-245, which are likely to interact with each other (31). The appeal of this idea had been strengthened by the possibility that similar residues function as a proton relay in LacY (33, 197), but since the latter is unlikely (chapter 74), and since F₀ also conducts Na⁺ in some species (see below), an alternative role for residues such as R-210 should be considered (see below).

The b protein is conveniently arranged for study—this polypeptide has a short (30-residue) hydrophobic N-terminal tail, localized in the membrane (95), and a long (130-residue) hydrophilic body, expected to be largely α-helical. This hydrophilic C terminus extends into the aqueous cytoplasmic phase, where one presumes it contributes to the stalk connecting F₀ and F₁ and possibly extends into F₁ itself. Certainly, this hydrophilic extension is important to F₁ binding, since protease applied at the cytoplasmic surface of F₁-stripped membranes readily cleaves subunit b and abolishes F₁ rebinding, without affecting H⁺ conduction (88, 94, 192). At a minimum, then, b has an evident structural role.

Of F₀ subunits, the smallest, subunit c, has received the most attention, largely in hopes that catalysis of H⁺ conduction might be largely assigned to this single polypeptide. Speculation on this point began with the observation that low concentrations of DCCD blocked H⁺ flux through F₀ and at the same time inhibited ATP synthesis (or hydrolysis) by F₀F₁, all without affecting the properties of isolated F₁. This prompted the suggestion that F₀ might behave as a H⁺ carrier or channel (see reference 169). Subsequently, the site of DCCD action in E. coli F₀ was shown to be subunit c (57) (also known as the "proteolipid" subunit because of its solubility in organic solvents), and the DCCD-reactive residue was identified as Asp-61 (see reference 58); a glutamate at this position is the target for DCCD in other examples of subunit c. The relevance of D-61 to proton flux has been strengthened by two new findings: (i) that DCCD modification of but a single subunit c can block F₀ function (87) and (ii) that H⁺ movement is also blocked when D-61 is replaced by residues that cannot undergo protonation-deprotonation (D61G, D61N, or D61Q) (96, 97; see also references 58 and 59).

Many different examples of subunit c have been sequenced, and in each case one predicts a hairpin organization in which two membrane-embedded α-helices (158) are separated by a short hydrophilic stretch exposed to cytoplasmic F₁ (58, 97, 176). These inferences are now supported by NMR studies which directly analyze subunit c solubilized in mixtures of chloroform-methanol-water (63, 64; summarized in reference 59). To be sure, solubilized c is monomeric, not oligomeric (c _10±1), but reactivity with DCCD is retained, and solubilized c from a mutant resistant to DCCD shows reduced reactivity of D-61 in vitro. This suggests that at least some features of the isolated subunit c reflect the in vivo condition. Moreover, aside from verifying the predicted hairpin structure, the NMR analysis explains the unexpected finding that F₀ function is retained in a double mutant in which aspartate has been moved from the C-terminal to the N-terminal helix (A24D/D61G) (164). It appears that retention of a hairpin arrangement by monomeric subunit c reflects strong interactions between the N- and C-terminal helices, suggesting they act as a unit (59). If so, one might imagine that while the positioning of D-61 is crucial to function, mutations placing aspartate in the same region might also preserve function. Indeed, NMR shows that A-24 and D-61 lie directly across from one another in the intact structure (64; also see reference 59), so preservation of function in the double mutant can be rationalized. All this evokes a simple picture in which a protonatable group in the center of the membrane (e.g., D-61) plays an obligatory role in F₀ function. The hypothesis that D-61 acts as the primary proton acceptor in F₀ is becoming more and more attractive but remains circumstantial even now.

What of interactions between the various F₀ subunits and between F₀ and F₁? Information on this point is incomplete but provocative enough to raise expectations. There are two important observations, each dealing with contacts made by subunit c as judged by the behavior of second-site suppressors. On the one hand, Fillingame and colleagues isolated suppressors that enhance functioning of the A24D/D61G double mutant (see above). The suppressors usually map to subunit a, most often in positions defining one face of transmembrane helix V (cited in reference 59). These studies suggest, not surprisingly, that subunits a and c interact and also point to the likely points of direct, physical contact. Equally striking, the required R-210 of subunit a (see above) also lies on this face of helix V, strengthening the idea (59) that R-210 might form transitory associations with D-61 of subunit c, stabilizing the latter in its anionic form prior to protonation of subunit c during H⁺ transit. The other significant finding is that second-site mutations in F₁ subunit ε can suppress the effects of a null mutation in the cytoplasmic, polar loop of subunit c (244). Whatever physical interpretation is placed on this finding, it seems that subunit ε must "monitor" the structural status of subunits c. And inasmuch as the conformational state of subunit ε also monitors the catalytic site on F₁ subunit β (above), a chain of probable causal links is forming.

Proton Well and Mechanisms of Coupling.

We believe biophysics will eventually describe the various conformational states adopted by F₀F₁. The next step will be to relate these physical structures to kinetic behavior. In this case, one must address both F₀ and F₁, for just as each has a distinct biochemical phenotype, each has a characteristic kinetic signature. In effect, the F₀ sector acts as a transducer, enabling the separate parts of Δ p (ΔΨ and –60ΔpH) to transmit the same signal. F₁, on the other hand, utilizes this common signal to drive the release of ATP in a reaction of extraordinary cooperativity among its three catalytic sites.

The transduction within F₀ is best framed by discussion of a theoretical structure known as the "proton well," an ingenious device introduced by Mitchell (see references 169 and 171) in the context of a "direct" coupling mechanism. At issue is how a chemical reaction (ATP synthesis) might be structured so that it can be driven by thermodynamic forces whose physical manifestations are so different—by an electrical potential difference (Δψ), a chemical potential difference (–60ΔpH), or some unpredictable mixture of the two (Δ p). This problem arises in bacteria as they regulate internal pH, since both membrane potential and pH gradient change as the external pH varies. By the same token, Δ p is composed largely of a membrane potential in mitochondria, but a pH gradient in chloroplasts, yet the same enzyme makes ATP in these two very different settings. A proton well solves this problem by allowing electrical and chemical driving forces to be expressed in the same way at a molecular level—as an increased "local acidity" in the active site of the enzyme (reviewed in reference 149).

Relevance of the proton well to direct coupling mechanisms arises in the following way. The stereochemistry of F₁-mediated ATP hydrolysis (232) points to the synthetic step as an in-line displacement, during which progress from the intermediate state [shown in brackets] can be represented as follows (and see the brief discussion in reference 1)

. . . .⇔ [ADPO^– →PO₃ →O^–] + 2H⁺ ⇔ ADPOPO3 + H₂O

Briefly, in a direct model (see references 149 and 171) the 2H⁺ provided by F₀ serve as acceptors of the terminal oxygen from phosphate [→O^–], yielding the products, ATP, and water. In this case, F₀ must make H⁺ available as a reactant even when the net driving force is composed entirely of an electrical component and when the external proton concentration is relatively low. How is the proton concentration raised within the enzyme? By imposing ion selectivity on F₀, so that H⁺ is drawn inward along the electric field (negative on the F₁ side) without an accompanying anion. As a result, proton reactivity at the bottom of the "well," after the electric field has been crossed, corresponds to the reactivity of protons at the concentration expected when Δ p is expressed solely as an equivalent pH gradient (149, 169). In effect, the proton well transforms the electric part of Δ p into its corresponding chemical component and thereby provides protons as chemical reactants at the required concentration. In the context of ATPase mechanism, the protons required for ATP synthesis are those which move through F₀, hence the idea of "direct" coupling.

In fact, kinetic evidence is consistent with the view that F₀ behaves as a proton well (124, 149), and the idea that ion binding to an active site may occur within the electric field is now an accepted part of membrane biophysics, whether in discussions of ion channels (242), ATPases (132, 149), or the flagellar motor (125). On the other hand, it is equally clear that a direct coupling mechanism for F₀F₁ is unlikely, despite the elegance of its conception, for any of several reasons. For example, the simplest of direct coupling mechanisms requires a stoichiometry of2H⁺/ATP, but the evidence shows stoichiometry as 3H⁺/ATP (see above). Second, it is now apparent that energy (i.e., Δ p) is not required for the synthetic act, which occurs readily on the surface of F₁; instead, energy is utilized at a later step, to accelerate dissociation of ATP and binding of new reactants, the so-called "binding change" mechanism (25, 26). Third, consider the distances involved (Fig. 9). Nucleotide binding sites on F₁ lie 50 to 100 Å (5 to 10 nm) away from F₀ (see reference 32), so it seems that ADP is not sensibly positioned to accept protons coming from F₀. Fourth, and most convincingly, recent experiments with hybrid versions of F₀F₁ argue against such an instructive role on the part of F₀. Specifically, coupling of ATP synthesis to Na⁺ movements is observed when the catalytic portion of E. coli F₁ is appropriately combined with the F₀ sector from a Na⁺-translocating variant found in Propionigenium modestum (110, 131) (in this chimera, the δ subunit and part of α are also from P. modestum). This shows that any ionic specificity on the part of F₁ (e.g., H⁺) is separate and apart from that of F₀ (H⁺, Na⁺, etc.). To summarize, then, the distinctive feature of direct coupling models is the literal link between transported H⁺ and the chemical reaction, and this appears highly unlikely for F₀F₁. As a result, we must address various "indirect" models in which ion (H⁺, Na⁺, etc.) movement drives changes of protein conformation.

F₁ shows equally unusual kinetic properties, as first made evident by studies of the mitochondrial enzyme (43, 69, 71). Equivalent studies of bacterial F₁ are now available, largely as a result of the efforts of Senior and his collaborators (210, 211, 234). At low substrate concentration ([ATP] < [F₁]), ATP is bound very much more tightly and hydrolyzed very much more slowly (by factors of 10⁶ to 10⁸) than expected on the basis of ATPase activity measured at high substrate levels ([ATP] > [F₁]). Modeling exercises indicate three "levels" of activity for the mitochondrial enzyme, presumably corresponding to participation (per mole of F₁) of 1, 2, and 3 catalytic units (so-called uni-, bi- and tri-site behavior, respectively); by contrast, only uni-site and tri-site behavior is clearly seen for bacterial F₁ (234). More important, assays of ATP binding and hydrolysis and of ADP and P_i release show that the reaction . . . ATP ↔ ADP + P_i . . . is nearly at equilibrium on the surface of F₁, corroborating the early finding that the F₁-ATPase reaction is reversible at low substrate levels (26, 35). Finally, kinetic modeling predicts that the basis of rate acceleration (cooperativity) in going from uni- to tri-site behavior involves a dramatic (≥10⁶-fold) acceleration of product release from one site as the next one is occupied by substrates (ADP + P_i) (43, 210, 211). Thus, increased turnover requires a give-and-take in which substrate occupancy at one site at the same time promotes product release at another, and vice versa.

These findings with F₁ have had two significant effects. On the one hand, direct models for ATP synthesis are less attractive than before, since for suitably arranged conditions F₁ quite obviously mediates ATP synthesis (albeit slowly) with no immediate participation of H⁺. Accordingly, the role of H⁺ (Δ p) is now assigned to driving conformational changes that lead to release of preformed ATP, as had been suggested earlier (25, 26). A second, more pronounced change in thinking centers on imagining ways in which three catalytic units might be sampled in sequence during a completed turnover. One way to account for this (and explain cooperativity) is to invoke a movement of subunits relative to one another. A decade-old speculation (41) involves a literal rotation of a complex involving a subunit b dimer and the minor subunits (γδε) of F₁ within two encircling arrays—the one in F₀, the other in F₁ (Fig. 11). Thus, in F₀, the short hydrophobic arms of the b dimer would fall inside a ring of subunits c, so that rotation could be driven by the sequential electrostatic attractions between the two subunits during proton transfer. At the same time, in cytoplasmic F₁ the asymmetric γδε core would be arranged so as to present a different surface to each of three αβ pairs. Thus (pace Wankel), a full rotation would allow each of three otherwise identical catalytic sites to assume an attribute (substrate binding, bond formation, product release, etc.) required in the binding-change mechanism (25, 26). In the same spirit, but with an emphasis on F₁, others (171) imagine that rotation of the central core induces sympathetic reverse rotations of the entire F₁ αβ ensemble, so that efficient operation of one catalytic site could demand suitable operation of all others. These early speculations are consistent with the recent analysis of F₁ structure (1), which verifies that each catalytic site is in a suitably different environment, but there is little direct evidence to support rotating models. On the other hand, such thinking is provocative. F₀F₁ is sufficiently complex to be considered a machine, and this is best emphasized by explicit analogy to electrical motors and chemical engines. At the very least, it brings unexpected excitement to the next phases of study.

Fig. 11Rotating model of F0F1. The central rotating shaft is a composite of a subunit b dimer and the minor F1 subunits (???). The b dimer interacts with an encircling array of subunits c. Subunit a (not shown) may mediate this interaction (see the text). Within F1, the asymmetric ??? triad comes into sequential contact with each of three surrounding ?? pairs, generating three different nucleotide-binding sites.

Source: Reprinted from reference 151 with permission.

ION MOTIVE METABOLIC CYCLES

OxlT

In bioenergetics and membrane biology, the major contemporary emphasis is a pursuit of protein structure. This is an admirable goal, but we should not now forget how important it is to understand the connections between these structures at the level of physiology. The microbial world offers instructive examples of this latter point in the form of OxlT and similar transporters that take part in "proton motive metabolic cycles." Here, an unusual combination of vectorial and scalar elements is used to construct a virtual (rather than a literal) proton pump.

The gram-negative anaerobe Oxalobacter formigenes derives metabolic energy from the decarboxylation of oxalate ( ^–OOC-COO^–) (5). In this and other cells that degrade oxalate or other dicarboxylates, catabolism begins with transport of substrate into the cell, followed by a decarboxylation that generates CO₂ and the corresponding monocarboxylate, formate (HCOO^–) in the case of O. formigenes. Because O. formigenes has a limited capacity to metabolize formate, there is near stoichiometric transformation of oxalate into carbon dioxide and formate (1:1:0.9), suggesting that generation of metabolic energy might be associated with the transport or decarboxylation reactions themselves. In fact, some cells have a decarboxylase that itself pumps Na⁺ outward, initiating a Na⁺ circulation (46). However, this is not feasible for O. formigenes, in which oxalate degradation is mediated by a soluble oxalyl-coenzyme A decarboxylase (14). Accordingly, studies with O. formigenes focused on the possibility that generation of metabolic energy was in some way related to the movement of oxalate itself.

In experiments based on the reconstitution of oxalate transport in liposomes, Anantharam et al. (8) showed that membranes of O. formigenes catalyze the electrogenic exchange of divalent oxalate (precursor) for monovalent formate (product). This activity was later assigned to a single membrane protein, termed OxlT (for oxalate transporter) (203). The significance of these findings is outlined in Fig. 12, which illustrates how this exchange carrier, OxlT, along with the cytosolic decarboxylase, can function as an entirely new kind of proton pump (8, 203). Note that there is inward transport of a divalent anion (^–OOC-COO^–), but outward movement of a monovalent anion (HCOO^–). This means that each completed transport cycle brings in a single negative charge. And coincident with this charge transfer, there is disappearance of a single internal proton, as decarboxylation transforms precursor into product. Together, these vectorial and scalar events establish the thermodynamic equivalent of an outwardly directed proton pump, one whose stoichiometry is 1H⁺ per cycle. Three cycles would extrude 3H⁺ and establish a proton motive gradient sufficient to drive ATP synthesis by "decarboxylative phosphorylation."

Fig. 12Generation and utilization of a proton motive force in O. formigenes. The origin of a proton motive force in O. formigenes is based on the electrogenic exchange of divalent oxalate and monovalent formate. A membrane potential (negative inside) is maintained by this electrogenic exchange, while a pH gradient (alkaline inside) arises by consumption of protons during intracellular decarboxylation. The completed cycle (influx ? decarboxylation ? efflux) constitutes a H+ pump with a stoichiometry of 1H+ per turnover.

Source: Reprinted from reference 8 with permission.

Proton Motive Metabolic Cycles

With this prototype in mind (Fig. 12), it has been possible to search for equivalent examples in other cells, and in recent years a number of likely proton motive decarboxylation cycles have been identified (Table 2) (see also reference 194). In several cases, the experimental evidence strongly supports the idea (174, 185, 195, 203), although in other instances the evidence is incomplete (Table 2). Even so, the present status is sufficient to establish the phenomenon as significant to microbial systems. In particular, it may be worth considering these cycles in connection with the acid-inducible amino acid decarboxylases of E. coli (see reference 163). Proton motive decarboxylation cycles may be especially suited to survival under acidic conditions, since these pumps are not limited by the thermodynamic restraints usually associated with ion transport. For example, in the case of F₀F₁, the maximal proton motive gradient arising from ATP hydrolysis is set by the stoichiometry of coupling and the value of the phosphorylation potential (equation 5b); stoichiometry is fixed, nucleotide and phosphate levels must be compatible with metabolism and its control, and so the size of the proton potential maintained by F₀F₁ is (typically) no more than 2.5 pH units or its equivalent, 150 mV (150). By contrast, during a decarboxylation-linked proton motive cycle, the dissemination of gaseous CO₂ renders the overall reaction functionally irreversible. As a result, the size of the proton potential may be regulated by expression of the scalar decarboxylase, whose level would determine the rate of transformation of precursor into products. Accordingly, such metabolic cycles might function, in principle, as "tunable" proton pumps, sustaining the proton potential as low or as high as circumstances demand.

Table 2Likely proton motive metabolic cyclesa

The proton motive metabolic cycles so far known (Table 2) strongly resemble the prototype in O. formigenes (Fig. 12), but this is not required on formal grounds. Instead, there are many ways to achieve the final result (203), since these cycles require only a one-for-one stoichiometry between entry of negative charge (or exit of positive charge) and consumption of internal protons (or external hydroxyl anions). Note as well that the presently known cycles are rather simple (Table 2). Again, this is not required. And to illustrate the larger scale, it has been suggested (156, 203) one might look to methanogenesis, which builds on compounds such as acetate, formate, or bicarbonate. If these precursors enter as anions, bringing in their single negative charge (say, by a uniporter or a channel, such as the E. coli FocA protein [see chapter 74]), the ensemble of cytoplasmic reactions leading to methane production can be viewed as the scalar portion of an indirect proton pump—for each anion that moves inward, the metabolic pathway for methanogenesis ensures that a single internal proton is consumed. A final comment seems warranted in this more general setting. Although metabolic proton motive cycles seem restricted to bacterial systems, the possibility of wider distribution should not be discounted too early. An important lesson of recent years is that membranes are much the same everywhere. At the very least, as OxlT demonstrates, we might look forward to other emergent functions as new and unexpected organizational patterns impinge on membrane biology.

ACKNOWLEDGMENTS

It is a pleasure to thank Robert Fillingame, Peter Hinkle, John Ingledew, and Marten Wikström for helpful advice on proton stoichiometries. We also thank Roderick Capaldi for permission to reprint the cryoelectron micrograph in Fig. 9. Original research from our laboratories has been supported in part by grants from the Public Health Service (AI-03568 to F.M.H., GM24195 to P.C.M.) and from the National Science Foundation (DCB 90–17130 to F.M.H., MCB 92–20823 to P.C.M.).

† Dedicated to the memory of Peter Mitchell (1920–1992), who first made chemical sense of biological energetics and explained it to the rest of us.

References

1. Abrahams, J. P., A. G. W. Leslie, R. Lutter, and J. E. Walker. 1994. Structure at 2.8 Å resolution of F1-ATPase from bovine heart mitochondria. Nature (London) 370:621–628.

1a. Abrahams, J. P., R. Lutter, R. J. Todd, M. J. van Raaij, A. G. W. Leslie, and J. E. Walker. 1993. Inherent asymmetry of the structure of F₁-ATPase from bovine heart mitochondria at 6.5 Å resolution. EMBO J. 12:1775–1780.

2. Abrams, A., P. McNamara, and F. B. Johnson. 1960. Adenosine triphosphatase in isolated bacterial cell membranes. J. Biol. Chem. 235:3659–3662.

3. Aggeler, R., S. X. Cai, J. F. Keana, T. Koike, and R. A. Capaldi. 1993. The γ subunit of the Escherichia coli F₁-ATPase can be cross-linked near the glycine-rich loop region of a β subunit when ADP + Mg²⁺ occupies catalytic sites but not when ATP + Mg²⁺ is bound. J. Biol. Chem. 268:20831–20837.

4. Aggeler, R., and R. A. Capaldi. 1992. Cross-linking of the γ subunit of the Escherichia coli ATPase (ECF₁) via cysteines introduced by site-directred mutagenesis. J. Biol. Chem. 267:21355–21359.

5. Allison, M., K. A. Dawson, W. R. Mayberry, and J. G. Foss. 1985. Oxalobacter formigenes gen. nov., sp. nov.: oxalate-degrading anaerobes that inhabit the gastrointestinal tract. Arch. Microbiol. 141:1–7.

5a. al-Shawi, M. K., D. Parsonage, and A. E. Senior. 1990. Thermodynamic analyses of the catalytic pathway of F₁-ATPase from Escherichia coli. Implications regarding the nature of energy coupling by F₁-ATPase. J. Biol. Chem. 265:4402–4410.

6. Altendorf, K., H. Hirata, and F. M. Harold. 1975. Accumulation of lipid-soluble ions and of rubidium as indicators of the electrical potential in membrane vesicles of Escherichia coli. J. Biol. Chem. 250:1405–1412.

7. Altendorf, K., and L. A. Stachelin. 1974. Orientation of membrane vesicles from Escherichia coli as detected by freeze-cleave electron microscopy. J. Bacteriol. 117:888–899.

8. Anantharam, V., M. J. Allison, and P. C. Maloney. 1989. Oxalate:formate exchange. The basis for energy coupling in Oxalobacter. J. Biol. Chem. 264:7244–7250.

9. Andersen, K. B., and K. von Meyenburg. 1977. Changes of nicotinamide adenine nucleotides and energy charge as regulatory parameters for growth of E. coli. J. Biol. Chem. 252:4151–4156.

10. Anraku, Y. 1988. Bacterial electron transport chains. Annu. Rev. Biochem. 57:101–132.

11. Aris, J. P., D. J. Klionsky, and R. D. Simoni. 1985. The F₀ subunits of Escherichia coli F₀F₁-ATP synthase are sufficient to form a functional proton pore. J. Biol. Chem. 260:11207–11215.

12. Asghar, S. S., E. Levin, and F. M. Harold. 1973. Accumulation of neutral amino acids by Streptococcus faecalis: energy coupling by a proton motive force. J. Biol. Chem. 248:5225–5233.

13. Avetisyan, A. V., A. V. Bogachev, R. A. Murtasina, and V. P. Skulachev. 1993. ATP-driven Na⁺ transport and Na⁺-dependent ATP synthesis in Escherichia coli grown at low . FEBS Lett. 317:267–270.

14. Baetz, A. L., and M. J. Allison. 1989. Purification and charactrerizatoin of oxalyl-CoA-decarboxylase from Oxalobacter formigenes. J. Bacteriol. 171:2605–2608.

15. Bakeeva, L. E., L. L. Grinius, A. A. Jasaitis, V. V. Kuliene, D. O. Levitsky, E. A. Liberman, I. I. Severina, and V. P. Skulachev. 1970. Conversion of biomembrane-produced energy into electric form. II. Intact mitochondria. Biochim. Biophys. Acta 216:13–21.

16. Bakker, E. P. 1979. Ionophore antibiotics, p. 67–97. In F. E. Hahn (ed.), Antibiotics, vol. V/1. Mechanisms of Antibacterial Agents. Springer-Verlag KG, Berlin.

17. Bakker, E. P. 1993. Cell K⁺ and K⁺ transport systems in prokaryotes, p. 205–224. In E. Bakker (ed.), Alkali Cation Transport Systems in Prokaryotes. CRC Press, Inc., Boca Raton, Fla.

18. Bakker, E. P., and F. M. Harold. 1980. Energy coupling to potassium transport in Streptococcus faecalis: interplay of ATP and the protonmotive force. J. Biol. Chem. 255:433–441.

19. Bakker, E. P., and L. L. Randall. 1984. The requirements of energy during the export of α-lactamase in Escherichia coli is fulfilled by the total protonmotive force. EMBO J. 3:895–901.

20. Berrier, C., A. Coulombe, I. Szabo, M. Zoratti, and A. Ghazi. 1992. Gadolinium ion inhibits loss of metabolites induced by osmotic shock and large stretch-activated channels in bacteria. Eur. J. Biochem. 206:559–565.

21. Berry, E. A., and P. C. Hinkle. 1983. Measurement of the electrochemical proton gradient in submitochondrial particles. J. Biol. Chem. 258:1474–1486.

22. Bianchet, M., X. Ysern, J. Hullihen, P. L. Pedersen, and L. M. Amzel. 1991. Mitochondrial ATP synthase. Quaternary structure of the F₁ moiety at 3.6 Å determined by X-ray diffraction analysis. J. Biol. Chem. 266:21197–21201.

23. Bizouarn, T., and J. B. Jackson. 1993. The ratio of protons translocated/hydride ion equivalent transferred by nicotinamide nucleotide transhydrogenase in chromatophores from Rhodospirillum rubrum. Eur. J. Biochem. 217:763–770.

24. Booth, I. R. 1985. Regulation of cytoplasmic pH in bacteria. Microbiol. Rev. 49:359–378.

25. Boyer, P. D. 1979. The binding-change mechanism of ATP synthesis, p. 461–479. In C. P. Lee, G. Schatz, and L. Ernster (ed.), Membrane Bioenergetics. Addison-Wesley, Reading, Mass.

26. Boyer, P. D., R. L. Cross, and W. Momsen. 1973. A new concept for energy coupling in oxidative phosphorylation based on a molecular explanation of the oxygen exchange reactions. Proc. Natl. Acad. Sci. USA 70:2837–2839.

27. Brusiliow, W. S. A., D. J. Klionsky, and R. D. Simoni. 1982. Differential polypeptide synthesis of the proton-translocating ATPase of Escherichia coli. J. Bacteriol. 151:1363–1371.

28. Brusilow, W. S. A., A. C. G. Porter, and R. D. Simoni. 1983. Cloning and expression of uncI, the first gene in the unc operon of Escherichia coli. J. Bacteriol. 155:1265–1270.

29. Bull, P. C., G. R. Thomas, J. M. Rommens, J. R. Forbes, and D. W. Cox. 1993. The Wilson disease gene is a putative copper transporting P-type ATPase similar to the Menkes gene. Nat. Genet. 5:327–337.

30. Butlin, J. D., G. D. Cox, and F. Gibson. 1971. Oxidative phosphorylation in Escherichia coli K12. Mutations affecting magnesium ion- or calcium ion-stimulated adenosine triphosphatase. Biochem. J. 124:75–81.

31. Cain, B. D., and R. D. Simoni. 1988. Interaction between Glu-219 and His-245 within the a subunit of the F₁F₀-ATPase in Escherichia coli. J. Biol. Chem. 263:6606–6612.

32. Capaldi, R. A., R. Aggeler, P. Turina, and S. Wilkens. 1994. Coupling between catalytic sites and the proton channel in F₁F₀-type ATPases. Trends Biochem. Sci. 19:284–289.

33. Carrasco, N., L. M. Antes, M. S. Poonian, and H. R. Kaback. 1986. Lac permease of Escherichia coli: histidine-322 and glutamic acid-325 may be components of a charge-relay system. Biochemistry 25:4486–4488.

34. Castle, A. M., R. M. Macnab, and R. G. Shulman. 1986. Coupling between the sodium and proton gradients in respiring Escherichia coli cells measured by ²³Na and ³¹P nuclear magnetic resonance. J. Biol. Chem. 261:7797–7806.

35. Choate, G. L., R. L. Hutton, and P. D. Boyer. 1979. Occurrence and significance of oxygen exchange reactions catalyzed by mitochondrial adenosine triphosphatase preparations. J. Biol. Chem. 254:286–290.

36. Cohen, G. N., and J. Monod. 1957. Bacterial permeases. Bacteriol. Rev. 21:169–194.

37. Collins, S. H., and W. A. Hamilton. 1976. Magnitude of the proton motive force in respiring Staphylococcus aureus and Escherichia coli. J. Bacteriol. 126:1224–1231.

38. Cox, G. B., B. A. Cromer, J. M. Guss, I. Harvey, P. D. Jeffrey, R. G. Solomon, and D. C. Webb. 1993. Formation in vivo, purification and crystallization of a complex of the γ and ε subunits of the F₀F₁-ATPase of Escherichia coli. J. Mol. Biol. 229:1159–1162.

39. Cox, G. B., J. A. Downie, L. Langman, A. E. Senior, G. Ash, D. R. H. Fayle, and F. Gibson. 1981. Assembly of the adenosine triphosphatase complex in Escherichia coli: assembly of F₀ is dependent on the formation of specific F₁ subunits. J. Bacteriol. 148:30–42.

40. Cox, G. B., and F. Gibson. 1974. Studies on electron transport and energy-linked reactions using mutants of Escherichia coli. Biochim. Biophys. Acta 346:1–25.

41. Cox, G. B., D. A. Jans, A. L. Fimmel, F. Gibson, and L. Hatch. 1984. Hypothesis. The Mechanism of ATP synthase. Conformational change by rotation of the b-subunit. Biochim. Biophys. Acta 768:201–208.

42. Cramer, W. A., and D. B. Knaff. 1991. Energy Transduction in Biological Membranes. Springer-Verlag, New York.

43. Cross, R. L., C. Grubmeyer, and H. S. Penefsky. 1982. Mechanism of ATP hydrolysis by beef heart mitochondrial ATPase. Rate enhancements resulting from cooperative interactions between multiple catalytic sites. J. Biol. Chem. 257:12101–12105.

44. Deckers-Hebestreit, G., and K. Altendorf. 1992. The F₀ complex of the proton-translocating ATPase of Escherichia coli. J. Exp. Biol. 172:451–459.

45. Dibrov, P. A. 1991. The role of sodium ion transport in Escherichia coli energetics. Biochim. Biophys. Acta 1056:209–224.

46. Dimroth, P. 1990. Mechanisms of sodium transport in bacteria. Philos. Trans. R. Soc. London Ser. B 326:465–477.

47. Downie, J. A., F. Gibson, and G. B. Cox. 1979. Membrane adenosine triphosphatases of prokaryotic cells. Annu. Rev. Biochem. 48:103–131.

48. Downie, J. A., L. Langman, G. B. Cox, C. Yanofsky, and F. Gibson. 1980. Subunits of the adenosine triphosphatase complex translated in vitro from the Escherchia coli unc operon. J. Bacteriol. 143:8–17.

49. Drachev, A. L., V. S. Markin, and V. P. Skulachev. 1985. -buffering by Na⁺ and K⁺ gradients in bacteria. Model and experimental systems. Biochim. Biophys. Acta 811:197–215.

50. Driessen, A. J. M., and W. Wickner. 1991. Proton transfer is rate-limiting for translocation of precursor proteins by the Escherichia coli translocase. Proc. Natl. Acad. Sci. USA 88:2471–2475.

51. Dunn, S. D. 1978. Identification of the altered subunit in the inactive F₁ ATPase of an Escherichia coli uncA mutant. Biochem. Biophys. Res. Commun. 82:596–602.

52. Dunn, S. D., and M. Futai. 1980. Reconstitution of a functional coupling factor from the isolated subunits of Escherichia coli F₁ ATPase. J. Biol. Chem. 255:113–118.

53. Dunn, S. D., and L. A. Heppel. 1981. Properties and functions of the subunits of the Escherichia coli coupling factor ATPase. Arch. Biochem. Biophys. 210:421–436.

54. Epstein, W. 1992. Kdp, a bacterial P-type ATPase whose expression and activity are regulated by turgor pressure. Acta Physiol. Scand. Suppl. 607:193–199.

55. Ernster, L., and C.-P. Lee. 1964. Biological oxidoreductions. Annu. Rev. Biochem. 33:729–788.

56. Fagan, M. J., and M. H. Saier, Jr. 1994. P-type ATPases of eukaryotes and bacteria: sequence comparisons and construction of phylogenetic trees. J. Mol. Evol. 38:57–99.

57. Fillingame, R. H. 1976. Purification of the carbodiimide-reactive protein component of the ATP energy-transducing system of Escherichia coli. J. Biol. Chem. 251:6630–6673.

58. Fillingame, R. H. 1990. Molecular mechanics of ATP synthesis by F₁F₀-type H⁺-transporting ATP synthases, p. 345–391. In T. A. Krulwich (ed.), The Bacteria: a Treatise on Structure and Function. Academic Press, Inc., New York.

59. Fillingame, R. H., M. E. Garvin, D. Fraga, and Y. Zhang. 1993. Correlations of structure and function in H⁺ translocating subunit c of F₁F₀ ATP synthase. Ann. N. Y. Acad. Sci. 671:323–334.

60. Foster, D. L., and R. H. Fillingame. 1982. Stoichiometry of subunits in the H⁺-ATPase complex of Escherichia coli. J. Biol. Chem. 257:2009–2015.

61. Friedl, P., J. Hoppe, R. P. Gunsalus, O. Michelsen, K. von Meyenburg, and H. U. Schairer. 1983. Membrane integration and function of the three F₀ subunits of the ATP synthase of Escherichia coli K12. EMBO J. 2:99–103.

62. Futai, M., and H. Kanazawa. 1983. Structure and function of proton-translocating adenoside triphosphatase (F₀F₁): biochemical and molecular biological approaches. Microbiol. Rev. 47:285–312.

63. Garvin, M. E., and R. H. Fillingame. 1993. Helical structure and folding of subunit c of F₁F₀ ATP synthase: ¹H NMR resonance assignments and NOE analysis. Biochemistry 32:12167–12177.

64. Garvin, M. E., and R. H. Fillingame. 1994. Hairpin folding of subunit c of F₁F₀ ATP synthase: ¹H distance measurements to nitroxide-derivatized aspartyl-61. Biochemistry 33:665–674.

65. Gay, N. J., and J. E. Walker. 1981. The atp operon: nucleotide sequence of the region encoding the α-subunit of Escherichia coli ATP-synthase. Nucleic Acids Res. 9:2187–2194.

66. Gay, N. J., and J. E. Walker. 1981. The atp operon: nucleotide sequence of the promoter and the genes for the membrane proteins, and the δ subunit of Escherichia coli ATP-synthase. Nucleic Acids Res. 9:3919–3926.

67. Gibson, F. 1983. Biochemical and genetic studies on the assembly and function of the F₀F₁ adenosine triphosphatase of Escherichia coli. Biochem. Soc. Trans. 11:229–240.

68. Glagolev, A. N., and V. P. Skulachev. 1978. The proton pump is a molecular engine of motile bacteria. Nature (London) 272:280–282.

69. Gresser, M. J., J. A. Myers, and P. D. Boyer. 1982. Catalytic site cooperativity of beef heart mitochondrial F₁ adenosine triphosphatase. Correlations of initial velocity, bound intermediate, and oxygen exchange measurements with an alternating three-site model. J. Biol. Chem. 257:12030–12038.

70. Griniuviene, B., V. Chimieliauskaite, and L. Grinius. 1974. Energy-linked transport of permeant ions in Escherichia coli cells: evidence for membrane potential generation by proton pumps. Biochem. Biophys. Res. Commun. 56:206–213.

71. Grubmeyer, C., R. L. Cross, and H. S. Penefsky. 1982. Mechanism of ATP hydrolysis by beef heart mitochondrial ATPase. Rate constants for elementary steps in catalysis at a single site. J. Biol. Chem. 257:12092–12100.

72. Gunsalus, R. P., W. S. A. Brusilow, and R. D. Simoni. 1982. Gene order and gene-polypeptide relationships of the proton-translocating ATPase operon (unc) of Escherichia coli. Proc. Natl. Acad. Sci. USA 79:320–324.

73. Haddock, B. A., and W. A. Hamilton (ed.). 1977. Microbial energetics. Symp. Soc. Gen. Microbiol. 27:1–442.

74. Hanson, R. L., and C. Rose. 1980. Effects of an insertion mutation in a locus affecting pyridine nucleotide transhydrogenase (pnt: Tn5) on the growth of Escherichia coli. J. Bacteriol. 141:401–404.

75. Harold, F. M. 1972. Conservation and transformation of energy by bacterial membranes. Bacteriol. Rev. 36:172–230.

76. Harold, F. M. 1974. Chemiosmotic interpretation of active transport in bacteria. Ann. N. Y. Acad. Sci. 227:297–311.

77. Harold, F. M. 1977. Membranes and energy transduction in bacteria. Curr. Top. Bioenerg. 6:84–151.

78. Harold, F. M. 1977. Ion currents and physiological functions in microorganisms. Annu. Rev. Microbiol. 31:181–203.

79. Harold, F. M. 1986. The Vital Force: a Study of Bioenergetics. W. H. Freeman & Co., New York.

80. Harold, F. M. 1991. Biochemical topology: from vectorial metabolism to morphogenesis. Biosci. Rep. 6:347–385.

81. Harold, F. M., K. H. Altendorf, and H. Hirata. 1974. Probing membrane transport mechanisms with ionophores. Ann. N. Y. Acad. Sci. 235:149–160.

82. Harold, F. M., and J. R. Baarda. 1968. Inhibition of membrane transport in Streptococcus faecalis by uncouplers of oxidative phosphorylation and its relationship to proton conduction. J. Bacteriol. 96:2025–2034.

83. Harold, F. M., and D. Papineau. 1972. Cation transport and electrogenesis by Streptococcus faecalis. I. The membrane potential. J. Membr. Biol. 8:27–44.

84. Harold, F. M., and D. Papineau. 1972. Cation transport and electrogenesis by Streptococcus faecalis. II. Proton and sodium extrusion. J. Membr. Biol. 8:45–62.

85. Harold, F. M., E. Pavlasova, and J. R. Baarda. 1970. A transmembrane pH gradient in Streptococcus faecalis: origin and dissipation by proton conductors and N, N'-dicyclohexylcarbodiimide. Biochim. Biophys. Acta 196:235–244.

86. Harold, F. M., and J. van Brunt. 1977. Circulation of H⁺ and K⁺ across the plasma membrane is not obligatory for bacterial growth. Science 197:372–373.

87. Hermolin, J., and R. H. Fillingame. 1989. H⁺-ATPase activity of Escherichia coli F₁F₀ is blocked after reaction of dicyclohexylcarbodiimide with a single proteolipid (subunit c) of the F₀ complex. J. Biol. Chem. 264:3896–3903.

88. Hermolin, J., J. Gallant, and R. H. Fillingame. 1983. Topology, organization, and function of the psi subunit in the F₀ sector of the H⁺-ATPase of Escherichia coli. J. Biol. Chem. 258:14550–14555.

89. Hertzberg, E. L., and P. C. Hinkle. 1974. Oxidative phosphorylation and proton translocation in membrane vesicles prepared from Escherichia coli. Biochem. Biophys. Res. Commun. 58:178–184.

90. Hinkle, P. C., M. A. Kumar, A. Resetar, and D. L. Harris. 1991. Mechanistic stoichiometry of mitochondrial oxidative phosphorylation. Biochemistry 30:3576–3582.

91. Hirata, H., K. Altendorf, and F. M. Harold. 1973. Role of an electrical potential in the coupling of metabolic energey to active transport by membrane vesicles of Escherichia coli. Proc. Natl. Acad. Sci. USA 70:1804–1808.

92. Hirata, H., K. Altendorf, and F. M. Harold. 1974. Energy coupling in membrane vesicles of Escherichia coli: accumulation of metabolites in response to an electrical potential. J. Biol. Chem. 249:2939–2945.

93. Hoek, J. B., and J. Rydstrom. 1988. Physiological roles of nicotinamide nucleotide transhydrogenase. Biochem. J. 254:1–10.

94. Hoppe, J., P. Friedl, H. U. Schairer, W. Sebald, K. von Meyenburg, and B. B. Jorgensen. 1983. The topology of the proton translocating F₀ component of the ATP synthase from E. coli K12: studies with proteases. EMBO J. 2:105–110.

95. Hoppe, J., C. Montecucco, and P. Friedl. 1983. Labeling of subunit b of the ATP synthase from Escherichia coli with a photoreactive phospholipid analogue. J. Biol. Chem. 258:2882–2885.

96. Hoppe, J., H. U. Schairer, and W. Sebald. 1980. The proteolipid of a mutant ATPase from Escherichia coli defective in H⁺-conduction contains a glycine instead of the carbodiimide-reactive aspartyl residue. FEBS Lett. 109:107–111.

97. Hoppe, J., and W. Sebald. 1984. The proton conducting F₀-part of bacterial ATP synthases. Biochim. Biophys. Acta 768:1–27.

98. Iwamoto, A., J. Miki, M. Maeda, and M. Futai. 1990. H⁺-ATPase γ subunit of Escherichia coli. Role of the conserved carboxyl-terminal region. J. Biol. Chem. 265:5043–5048.

99. Iwamoto, A., H. Omote, H. Hanada, N. Tomioka, A. Itai, M. Maeda, and M. Futai. 1991. Mutations in Ser174 and the glycine-rich sequence (Gly149, Gly150 and Thr156) in the β subunit of Escherichia coli H⁺-ATPase. J. Biol. Chem. 266:16350–16355.

100. Jackson, J. B. 1991. The proton-translocating nicotinamide adenine dinucleotide transhydrogenase. J. Bioenerg. Biomembr. 23:715–741.

101. Jensen, P. R., O. Michelsen, and H. V. Westerhoff. 1993. Control analysis of the dependence of Escherichia coli physiology on the H⁺-ATPase. Proc. Natl. Acad. Sci. USA 90:8068–8072.

102. Jones, R. W., A. Lamont, and P. B. Garland. 1980. The mechanism of proton translocation driven by the respiratory nitrate reductase of Escherichia coli. Biochem. J. 190:79–94.

103. Kaback, H. R. 1970. Transport. Annu. Rev. Biochem. 39:561–598.

104. Kaback, H. R. 1972. Transport across isolated bacterial cytoplasmic membranes. Biochim. Biophys. Acta 265:367–416.

105. Kaback, H. R. 1986. Active transport in Escherichia coli: passage to permease. Annu. Rev. Biophys. 15:279–319.

106. Kaback, H. R., and E. M. Barnes, Jr. 1971. Mechanisms of active transport in isolated membrane vesicles. II. The mechanism of energy coupling between d-lactic dehydrogenase and β-galactoside transport in membrane preparations from Escherichia coli. J. Biol. Chem. 246:5523–5531.

107. Kagawa, Y. 1978. Reconstitution of the energy transformer, gate and channel, subunit reassembly, crystalline ATPase and ATP synthesis. Biochim. Biophys. Acta 505:45–93.

108. Kagawa, Y. 1980. Energy-transducing proteins in thermophilic biomembranes. J. Membr. Biol. 55:1–8.

109. Kagawa, Y. 1984. Proton motive ATP synthesis, p. 149–186. In L. Ernster (ed.), Bioenergetics. Elsevier Science Publishers BV, Amsterdam.

110. Kaim, G., and P. Dimroth. 1993. Formation of a functionally acitve sodium-translocating hybrid F₁F₀ ATPase in Escherichia coli by homologous recombination. Eur. J. Biochem. 218:937–944.

111. Kakinuma, Y., S. Kakinuma, K. Takase, K. Igarashi, and I. Yamoto. 1993. A gene encoding the 16-kDa proteolipid subunit of Enterococcus hirae Na⁺-ATPase complex. Biochem. Biophys. Res. Commun. 195:1063–1069.

112. Kanazawa, H., Y. Horiuchi, M. Takagi, Y. Ishino, and M. Futai. 1980. Coupling factor F₁ ATPase with defective β subunit from a mutant of Escherichia coli. J. Biochem. 88:695–703.

113. Kanazawa, H., T. Kayano, T. Kiyasu, and M. Futai. 1982. Nucleotide sequence of the genes for β and ε subunits of proton-translocating ATPase from Escherichia coli. Biochem. Biophys. Res. Commun. 105:1257–1264.

114. Kanazawa, H., T. Kayano, K. Mabuchi, and M. Futai. 1981. Nucleotide sequence of the genes coding for α, β and γ subunits of the proton-translocating ATPase of Escherichia coli. Biochem. Biophys. Res. Comm. 103:604–612.

115. Kanazawa, H., K. Mabuchi, T. Kayano, T. Noumi, T. Sekiya, and M. Futai. 1981. Nucleotide sequence of the genes for F₀ components of the proton-translocating ATPase from Escherichia coli: prediction of the primary structure of F₀ subunits. Biochem. Biophys. Res. Commun. 103:613–620.

116. Kanazawa, H., K. Mabuchi, T. Kayano, F. Tamura, and M. Futai. 1981. Nucleotide sequence of gene coding for dicyclohexylcarbodiimide-binding protein and the α subunit of proton-translocating ATPase of Escherichia coli. Biochem. Biophys. Res. Commun. 100:219–225.

117. Kanazawa, H., T. Noumi, N. Oka, and M. Futai. 1983. Intracistronic mapping of the defective site and the biochemical properties of β subunit mutants of Escherichia coli H⁺-ATPase: correlation of structural domains with functions of the β subunit. Arch. Biochem. Biophys. 227:596–608.

118. Kashket, E. R. 1981. Proton motive force in growing Streptococcus lactis and Staphylococcus aureus cells under aerobic and anaerobic conditions. J. Bacteriol. 146:369–376.

119. Kashket, E. R. 1982. Stoichiometry of the H⁺-ATPase of growing and resting aerobic Escherichia coli. Biochemistry 21:5534–5538.

120. Kashket, E. R. 1983. Stoichiometry of the H⁺-ATPase of Escherichia coli cells during anaerobic growth. FEBS Lett. 154:343–346.

121. Kashket, E. R. 1985. The proton motive force in bacteria: a critical assessment of methods. Annu. Rev. Microbiol. 39:219–242.

122. Kashket, E. R., and T. H. Wilson 1973. Proton-coupled accumulation of galactoside in Streptococcus lactis 7962. Proc. Natl. Acad. Sci. USA 70:2866–2869.

123. Kell, D. B. 1992. The protonmotive force as an intermediate in electron transport-linked phosphorylation: problems and prospects. Curr. Top. Cell Regul. 33:279–289.

124. Khan, S., and H. C. Berg. 1983. Isotope and thermal effects in chemiosmotic coupling to the membrane ATPase of Streptococcus. J. Biol. Chem. 258:6709–6712.

125. Khan, S., and H. C. Berg. 1983. Isotope and thermal effects in chemiosmotic coupling to the flagellar motor of Streptococcus. Cell 32:913–919.

126. Kibak, H., L. Taiz, T. Starke, P. Bernasconi, and J. P. Gogarten. 1992. Evolution of structure and function of V-ATPases. J. Bioenerg. Biomembr. 24:415–424.

127. Kinoshita, N., T. Unemoto, and H. Kobayashi. 1984. Proton motive force is not obligatory for growth of Escherichia coli. J. Bacteriol. 160:1074–1077.

128. Klionsky, D. J., and R. D. Simoni. 1985. Assembly of a functional F₁ of the proton-translocating ATPase of Escherichia coli J. Biol. Chem. 260:11200–11206.

129. Klionsky, D. J., W. S. A. Brusilow, and R. D. Simoni. 1984. In vivo evidence for the role of the ε subunit as an inhibitor of the proton-translocating ATPase of Escherichia coli. J. Bacteriol. 160:l055–1060.

130. Larsen, S. H., J. Adler, J. J. Gargus, and R. W. Hogg. 1974. Chemomechanical coupling without ATP: the source of energy for motility and chemotaxis in bacteria. Proc. Natl. Acad. Sci. USA 71:1239–1243.

131. Laubinger, W., G. Deckers-Hebestreit, K. Altendorf, and P. Dimroth. 1990. A hybrid adenosinetriphosphatase composed of F₁ of Escherichia coli and F₀ of Propionigenium modestum is a functional sodium ion pump. Biochemistry 29:5458–5463.

132. Lauger, P. 1979. A channel mechanism for electrogenic ion pumps. Biochim. Biophys. Acta 552:143–161.

133. Lawford, H. G., and B. A. Haddock. 1973. Respiration-driven proton translocation in Escherichia coli. Biochem. J. 136:217–220.

134. Lee, R. S., S. Wilke-Mounts, and A. E. Senior. 1992. F₁-ATPase with cysteine instead of serine at residue 373 of the α subunit. Arch. Biochem. Biophys. 297:334–339.

135. Lewis, M. J., J. A. Chang, and R. D. Simoni. 1990. A topological analysis of subunit a from the Escherichia coli F₁F₀ synthatse predicts eight transmembrane segments. J. Biol. Chem. 265:10541–10550.

136. Lewis, M. J., and R. D. Simoni. 1992. Deletions in hydrophobic domains of subunit a from the Escherichia coli F₁F₀-ATP synthase interfere with membrane insertion or F₀ assembly. J. Biol. Chem. 267:3482–3489.

137. Liang, A., and R. L. Houghton. 1981. Coregulation of oxidized nicotinamide adenine dinucleotide (phosphate) transhydrogenase and glutamate dehydrogenase activities in enteric bactreria during nitrogen limitation. J. Bacteriol. 146:997–1002.

138. Liberman, E. A., and V. P. Skulachev. 1970. Conversion of biomembrane-produced energy into electric form. IV. General discussion. Biochim. Biophys. Acta 216:30–42.

139. Lipmann, F. 1941. Metabolic generation and utilization of phosphate bond energy. Adv. Enzymol. 1:99–162.

140. Lombardi, F. J., J. P. Reeves, and H. R. Kaback. 1973. Mechanisms of active transport in isolated membrane vesicles. XIII. Valinomycin-induced rubidium transport. J. Biol. Chem. 248:3551–3565.

141. Lubin, M., and H. L. Ennis. 1963. On the role of intracellular potassium in protein synthesis. Biochim. Biophys. Acta 80:614–631.

142. Luckie, D. B., V. Lemas, K. L. Boyd, D. M. Fambrough, and K. Takeyasu. 1992. Molecular dissection of functional domains of the E1E2-ATPase using sodium and calcium pump chimeric molecules. Biophys. J. 62:220–226.

143. Luria, S. E., and J. L. Suit. 1987. Colicins and col plasmids, p. 1615–1624. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, D.C.

144. Mabuchi, K., H. Kanazawa, T. Kayano, and M. Futai. 1981. Nucleotide sequence of the gene coding for the δ subunit of proton-translocating ATPase of Escherichia coli. Biochem. Biophys. Res. Commun. 102:172–179.

145. Maeda, M., H. Kobayashi, M. Futai, and Y. Anraku. 1977. Studies on the turnovers in vivo of adenoside di- and triphosphates in a coupling factor of Escherichia coli. J. Biochem. 82:311–314.

146. Maloney, P. C. 1977. Obligatory coupling between proton entry and the synthesis of adenosine 5'-triphosphate in Streptococcus lactis. J. Bacteriol. 132:564–575.

147. Maloney, P. C. 1978. Coupling between H⁺ entry and ATP formation in Escherichia coli. Biochem. Biophys. Res. Commun. 83:1496–1501.

148. Maloney, P. C. 1979. Membrane H⁺ conductance of Streptococcus lactis. J. Bacteriol. 140:197–205.

149. Maloney, P. C. 1982. Energy coupling to ATP synthesis by the proton-translocating ATPase. J. Membr. Biol. 67:1–12.

150. Maloney, P. C. 1983. Relationship between phosphorylation potential and electrochemical H⁺ gradient during glycolysis in Streptococcus lactis. J. Bacteriol. 153:1461–1470.

151. Maloney, P. C. 1987. Coupling to an energized membrane: role of ion-motive gradients in the transduction of metabolic energy, p. 222–243. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, D.C.

152. Maloney, P. C., S. V. Ambudkar, V. Anantharam, L. A. Sonna, and A. Varadhachary. 1990. Anion-exchange mechanisms in bacteria. Microbiol. Rev. 54:1–17.

153. Maloney, P. C., E. R. Kashket, and T. H. Wilson. 1974. A protonmotive force drives ATP synthesis in bacteria. Proc. Natl. Acad. Sci. USA 71:3896–3900.

154. Maloney, P. C., E. R. Kashket, and T. H. Wilson. 1975. Methods for studying transport in bacteria, p. 1–49. In E. D. Korn (ed.), Methods in Membrane Biology, vol. 5. Plenum Press, New York.

155. Maloney, P. C., and T. H. Wilson. 1985. Evolution of ion pumps. BioScience 35:43–48.

156. Maloney, P. C., and T. H. Wilson. 1993. The evolution of membrane carriers, p. 147–160. In L. Reuss, J. M. Russell, Jr., and M. L. Jennings (ed.), Molecular Biology and Function of Carrier Proteins. Rockefeller University Press, NewYork.

157. Manson, M. D., P. Tedesco, H. C. Berg, F. M. Harold, and C. van der Drift. 1977. A protonmotive force drives bacterial flagella. Proc. Natl. Acad. Sci. USA 74:3060–3064.

158. Mao, D., E. Wachter, and B. A. Wallace. 1982. Folding of the mitochondrial proton adenosinetriphosphatase proteolipid channel in phospholipid vesicles. Biochemistry 21:4960–4968.

159. Martinac, B., M. Buechner, A. H. Delcour, and J. Adler. 1987. Pressure-sensitive ion channel in Escherichia coli. Proc. Natl. Acad. Sci. USA 84:2297–2301.

160. Matsumura, S., J.-I. Shioi, and Y. Imae. 1977. Motility in Bacillus subtilis is driven by an artificial proton-motive force. FEBS Lett. 82:187–190.

161. Matsushita, K., and H. R. Kaback. 1986. d-Lactate oxidation and generation of the proton electrochemical gradient in membrane vesicles from Escherichia coli GR19N and in proteoliposomes reconstituted with purified d-lactate dehydrogenase and cytochrome-o oxidase. Biochemistry 25:2321–2327.

162. McMorrow, I., H. A. Shuman, D. Sze, D. M. Wilson, and T. H. Wilson. 1989. Sodium/proton antiport is required for growth of Escherichia coli at alkaline pH. Biochim. Biophys. Acta 981:21–26.

163. Meng, S.-Y., and G. N. Bennet. 1992. Nucleotide sequence of the Escherichia coli cad operon: a system for neutralization of low extracellular pH. J. Bacteriol. 174:2659–2669.

164. Miller, M. J., M. Oldenburg, and R. H. Fillingame. 1990. The essential carboxyl group in subunit c of the F₁F₀ ATP synthase can be moved and H⁺-translocating function retained. Proc. Natl. Acad. Sci. USA 87:4900–4904.

165. Mitchell, P. 1961. Coupling of phosphorylation to electron and hydrogen transfer by a chemi-osmotic type of mechanism. Nature (London) 191:144–148.

166. Mitchell, P. 1963. Molecule, group and electron translocation through natural membranes. Biochem. Soc. Symp. 22:142–169.

167. Mitchell, P. 1966. Chemiosmotic coupling in oxidative and photosynthetic phosphorylation. Biol. Rev. Camb. Philos. Soc. 41:445–502.

168. Mitchell, P. 1967. Translocations through natural membranes. Adv. Enzymol. 29:33–87.

169. Mitchell, P. 1969. Chemiosmotic coupling and energy transduction. Theor. Exp. Biophys. 2:159–216.

170. Mitchell, P. 1979. David Keilin’s respiratory chain concept and its chemiosmotic consequences. Science 206:1148–1159.

171. Mitchell, P. 1985. Molecular mechanics of protonmotive F₀F₁ ATPases. Rolling well and turnstile hypothesis. FEBS Lett. 182:1–7.

172. Mitchell, P., and J. Moyle. 1967. Acid-base titration across the membrane system of rat-liver mitochondria: catalysis by uncouplers. Biochem. J. 104:588–600.

173. Mitchell, P., and J. Moyle. 1968. Proton translocation coupled to ATP hydrolysis in rat-liver mitochondria. Eur. J. Biochem. 4:530–539.

174. Molenaar, D., J. S. Bosscher, B. ten Brink, A. J. M. Driessen, and W. N. Konings. 1993. Generation of a proton motive force by histidine decarboxylation and electrogenic histidine/histamine antiport in Lactobacillus buchneri. J. Bacteriol. 175:2864–2879.

175. Moriyama, Y., A. Iwamoto, H. Hanada, M. Maeda, and M. Futai. 1991. One-step purification of Escherichia coli H⁺-ATPase (F₀F₁) and its reconstitution into liposomes with neurotransmitter transporters. J. Biol. Chem. 296:22141–22146.

176. Mosher, M. E., L. K. White, J. Hermolin, and R. H. Fillingame. 1985. H⁺-ATPase of Escherichia coli. An uncE mutation impairing coupling between F₁ and F₀ but not F₀-mediated H⁺ translocation. J. Biol. Chem. 260:4807–4814.

177. Nelson, N. 1992. Organellar proton-ATPases. Curr. Opin. Cell Biol. 4:654–660.

178. Nicholls, D. G., and S. J. Ferguson. 1992. Bioenergetics 2. Academic Press Ltd., London.

179. Nielson, J., F. G. Hansen, J. Hoppe, P. J. Friedl, and K. von Meyenburg. 1981. The nucleotide sequence of the atp genes coding for the F₀ subunits a, b, c and the F₁ subunit δ of the membrane bound ATP synthase of Escherichia coli. Mol. Gen. Genet. 184:33–39.

180. Niven, D. F., and W. A. Hamilton. 1974. Mechanisms of energy coupling in the transport of amino acids by Staphylococcus aureus. Eur. J. Biochem. 44:517–522.

181. Noumi, T., M. E. Mosher, S. Natori, M. Futai, and H. Kanazawa. 1984. A phenylalanine for serine substitution of the β subunit of Escherichia coli F₁-ATPase affects dependence of its activity on divalent cations. J. Biol. Chem. 259:10071–10075.

182. Noumi, T., M. Taniai, H. Kanazawa, and M. Futai. 1986. Replacement of arginine 246 by histidine in the β subunit of Escherichia coli H⁺-ATPase resulted in loss of multi-site ATPase activity. J. Biol. Chem. 261:9196–9201.

183. Ohyama, T., R. Imaizumi, K. Igarashi, and H. Kobayashi. 1992. Escherichia coli is able to grow with negligible sodium ion activity at alkaline pH. J. Bacteriol. 174:7743–7749.

184. Ohyama, T., S. Mugikura, M. Nishikawa, K. Igarashi, and H. Kobayashi. 1992. Osmotic adaptation of Escherichia coli with a negligible proton motive force in the presence of carbonyl cyanide m-chlorophenylhydrazone. J. Bacteriol. 174:2922–2928.

185. Olsen, E. B., J. B. Russell, and T. Henick-Kling. 1991. Electrogenic l-malate transport by Lactobacillus plantarum: a basis for energy derivation for malolactic fermentation. J. Bacteriol. 173:6199–6206.

186. Omote, H., M. Maeda, and M. Futai. 1992. Effects of mutations of conserved Lys-155 and Thr-156 residues in the phosphate-binding glycine-rich sequence of the F₁-ATPase β subunit of Escherichia coli. J. Biol. Chem. 267:20571–20576.

187. Padan, E., and S. Schuldiner. 1987. Intracellular pH and membrane potential as regulators in the prokaryotic cell. J. Membr. Biol. 95:189–198.

188. Padan, E., and S. Schuldiner. 1993. Na⁺ transport systems in prokaryotes, p. 3–24. In E. P. Bakker (ed.), Alkali Cation Transport Systems in Prokaryotes. CRC Press, Inc., Boca Raton, Fla.

189. Padan, E., D. Zilberstein, and H. Rottenberg. 1976. The proton electrochemical gradient in Escherichia coli. Eur. J. Biochem. 63:533–541.

190. Parsonage, D., M. K. al-Shawi, and A. E. Senior. 1988. Directed mutations of the strongly conserved lysine 155 in the catalytic nucleotide-binding domain of β-subunit of F₁-ATPase from Escherichia coli. J. Biol. Chem. 263:4740–4744.

191. Pavlasova, E., and F. M. Harold. 1969. Energy coupling in the transport of β-galactosides by Escherichia coli: effect of proton conductors. J. Bacteriol. 98:198–204.

192. Perlin, D. S., D. N. Cox, and A. E. Senior. 1983. Integration of F₁ and the membrane sector of the proton-ATPase of Escherichia coli. Role of subunit "b" (uncF protein). J. Biol. Chem. 258:9793–9800.

193. Poole, R. K., and W. J. Ingledew. 1987. Pathways of electrons to oxygen, p. 170–200. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, D.C.

194. Poolman, B., and W. N. Konings. 1993. Secondary solute transport in bacteria. Biochim. Biophys. Acta 1183:5–39.

195. Poolman, B., D. Molenaar, E. J. Smid, T. Ubbink, T. Abee, P. P. Renault, and W. N. Konings. 1991. Malolactic fermentation: electrogenic malate uptake and malate/lactate antiport generate metabolic energy. J. Bacteriol. 173:6030–6037.

196. Portis, A. R., Jr., and R. E. McCarty. 1976. Quantitative relationships between phosphorylation, electron flow, and internal hydrogen ion concentrations in spinach chloroplasts. J. Biol. Chem. 251:1610–1617.

197. Puttner, I. B., H. K. Sarkar, M. S. Poonian, and H. R. Kaback. 1986. Lac permease of Escherichia coli: histidine-205 and histidine-322 play different roles in lactose/H⁺ symport. Biochemistry 25:4483–4485.

198. Puustinen, A., M. Finel, T. Haltia, R. B. Gennis, and M. Wikström. 1991. Properties of the two terminal oxidases of Escherichia coli. Biochemistry 30:3936–3942.

199. Ramos, S., and H. R. Kaback. 1977. The relationship between the electrochemical proton gradient and active transport in Escherichia coli membrane vesicles. Biochemistry 16:854–859.

200. Robertson, R. N. 1970. Protons, Electrons, Phosphorylation and Active Transport. Cambridge University Press, London.

201. Rosen, B. 1973. β-Galactoside transport and proton movements in an adenosine triphosphatase-deficient mutant of Escherichia coli. Biochem. Biophys. Res. Commun. 53:1289–1296.

202. Rosen, B. P., and E. R. Kashket. 1978. Energetics of active transport, p. 559–620. In B. P. Rosen (ed.), Bacterial Transport. Marcel Dekker, Inc., New York.

203. Ruan, Z.-S., V. Anantharam, I. T. Crawford, S. V. Ambudkar, S. Y. Rhee, M. J. Allison, and P. C. Maloney. 1992. Identification, purification, and reconstitution of OxlT, the oxalate:formate antiport protein of Oxalobacter formigenes. J. Biol. Chem. 267:10537–10543.

203a. Salema, M., B. Poolman, J. S. Lolkema, M. C. Dias, and W. N. Konings. 1994. Uniport of monoanionic L-malate in membrane vesicles from Leuconostoc oenos. Eur. J. Biochem. 225:289–295.

204. Sarastre, M., N. J. Gay, A. Eberle, M. J. Runswick, and J. E. Walker. 1981. The atp operon: nucleotide sequence of the genes for the δ, β, and ε subunits of Escherichia coli ATP synthase. Nucleic Acids Res. 9:5287–5296.

205. Schaefer, E. M., D. Hartz, L. Gold, and R. D. Simoni. 1989. Risosome-binding sites and RNA-processing sites in the transcript of the Escherichia coli unc operon. J. Bacteriol. 171:3901–3908.

206. Schein, S. J., B. L. Kagan, and A. Finkelstein. 1978. Colicin K acts by forming voltage-dependent channels in phospholipid bilayer membranes. Nature (London) 276:159–163.

207. Schneppe, B., G. Deckers-Hebestreit, and K. Altendorf. 1990. Overproduction and purificaion of the uncI gene product of the ATP synthase of Escherichia coli. J. Biol. Chem. 265:389–395.

208. Scholes, P., and P. Mitchell. 1970. Respiration-driven proton translocation in Micrococcus denitrificans. J. Bioenerg. 1:309–323.

209. Schuldiner, S., and E. Padan. 1993. Na⁺/H⁺ antiporters in Escherichia coli, p. 25–51. In E. P. Bakker (ed.), Alkali Cation Transport Systems in Prokaryotes. CRC Press, Inc., Boca Raton, Fla.

210. Senior, A. E. 1990. The proton-translocating ATPase of Escherichia coli. Annu. Rev. Biophys. Chem. 19:7–41.

211. Senior, A. E., and M. K. al-Shawi. 1992. Further examination of seventeen mutations in Escherichia coli F₁-ATPase β subunit. J. Biol. Chem. 267:21471–21478.

212. Senior, A. E., D. R. H. Fayle, J. A. Downie, F. Gibson, and G. B. Cox. 1979. Properties of membranes from mutant strains of Escherichia coli in which the β subunit of the adenosine triphosphatase is abnormal. Biochem. J. 180:111–118.

213. Senior, A. E., L. Langman, G. B. Cox, and F. Gibson. 1983. Oxidative phosphorylation in Escherichia coli. Characterization of mutant strains in which F₁-ATPase contains abnormal β subunits. Biochem. J. 210:395–403.

214. Senior, A. E., S. Wilke-Mounts, and M. K. al-Shawi. 1993. Lysine 155 in the β-subunit is a catalytic residue of Escherichia coli F₁ ATPase. J. Biol. Chem. 268:6989–6994.

215. Shin, K., R. K. Nakamoto, M. Maeda, and M. Futai. 1992. F₀F₁-ATPase α subunit mutations perturb the coupling between catalysis and transport. J. Biol. Chem. 267:20835–20839.

216. Skulachev, V. P. 1985. Membrane-linked energy transductions. Bioenergetic functions of sodium: H⁺ is not unique as a coupling ion. Eur. J. Biochem. 151:199–208.

217. Skulachev, V. P. 1988. Membrane Bioenergetics. Springer-Verlag KG, Berlin.

218. Slater, E. C. 1971. The coupling between energy-yielding and energy-utilizing reactions in mitochondria. Q. Rev. Biophys. 4:35–72.

219. Smith, D. L., T. Tao, and M. E. Maguire. 1993. Membrane topology of a P-type ATPase. The MgtB magnesium transport protein of Salmonella typhimurium. J. Biol. Chem. 268:22469–22479.

220. Sternweis, P. C., and J. B. Smith. 1980. Characterization of the inhibitory (ε) subunit of the proton-translocating adenosine triphosphatase from Escherichia coli. Biochemistry 19:526–531.

221. Stokes, D. L., and R. K. Nakamoto. 1994. Structures of P-type and F-type ion pumps. Curr. Opin. Struct. Biol. 4:197–203.

222. Szabo, I., V. Petronelli, and M. Zoratti. 1993. A patch-clamp investigation of the Streptococcus faecalis cell membrane. J. Membr. Biol. 131:203–218.

223. Takase, K., S. Kakinuma, I. Yamoto, K. Konishi, K. Igarashi, and Y. Kakinuma. 1994. Sequencing and characterization of the ntp gene cluster for vacuolar-type Na⁺-translocating ATPase of Enterococcus hirae. J. Biol. Chem. 269:11037–11044.

224. Takase, K., I. Yamoto, and Y. Kakinuma. 1993. Cloning and sequencing of the genes coding for the A and B subunits of vacuolar-type Na⁺-ATPase from Enterococcus hirae. Coexistence of vacuolar- and F₀F₁-type ATPases in one bacterial cell. J. Biol. Chem. 268:11610–11616.

225. Taylor, M. A., and J. B. Jackson. 1987. Adaptive changes in membrane conductance in response to changes in specific growth rate in continuous cultures of phototrophic bacteria under conditions of energy sufficiency. Biochim. Biophys. Acta 893:242–255.

226. Tisa, L. S., and J. Adler. 1992. Calcium ions are involved in Escherichia coli chemotaxis. Proc. Natl. Acad. Sci. USA 89:11804–11808.

227. Tisa, L. S., B. M. Olivera, and J. Adler. 1993. Inhibition of Escherichia coli chemotaxis by ω-conotoxin, a calcium ion channel blocker. J. Bacteriol. 175:1235–1238.

228. Volpe, C., B. Levinson, S. Whitney, S. Packman, and J. Gitschier. 1993. Isolation of a candidate gene for Menkes disease and evidence that it encodes a copper-transporting ATPase. Nat. Genet. 3:7–13.

229. von Meyenburg, K., B. B. Jorgensen, J. Nielsen, and F. G. Hansen. 1982. Promoters of the atp operon coding for the membrane-bound ATP synthase of Escherichia coli mapped by Tn 10 insertion mutation. Mol. Gen. Genet. 188:240–248.

230. Walker, J. E., M. Saraste, and N. J. Gay. 1984. The unc operon. Nucleotide sequence, regulation and structure of ATP synthase. Biochim. Biophys. Acta 768:164–200.

231. Walker, J. E., M. Saraste, M. J. Runswick, and N. J. Gay. 1982. Distantly related sequences of α and β-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and and common nucleotide binding fold. EMBO J. 1:945–951.

232. Webb, M. R., C. Grubmeyer, H. S. Penefsky, and D. R. Trentham. 1980. The stereochemical course of phosphoric residue transfer catalyzed by beef heart mitochondrial ATPase. J. Biol. Chem. 255:11637–11639.

233. Weber, B. H. 1991. Glynn and the conceptual development of the chemiosmotic theory: a retrospective and prospective view. Biosci. Rep. 6:577–617.

234. Weber, J., S. Wilke-Mounts, R. S. Lee, E. Grell, and A. E. Senior. 1993. Specific placement of tryptophan in the catalytic sites of Escherichia coli F₁-ATPase provides a direct probe of nucleotide binding: maximal ATP hydrolysis occurs with three sites occupied. J. Biol. Chem. 268:20126–20133.

235. West, I. C., and P. Mitchell. 1972. Proton-coupled β-galactoside translocation in non-metabolizing Escherichia coli. J. Bioenerg. 3:445–462.

236. West, I. C., and P. Mitchell. 1973. Stoichiometry of lactose-proton symport across the plasma membrane of Escherichia coli. Biochem. J. 132:587–592.

237. West, I. C., and P. Mitchell. 1974. The proton-translocating ATPase of Escherichia coli. FEBS Lett. 40:1–4.

237a. West, I. C., and P. Mitchell. 1974. Proton/sodium ion antiport in Escherichia coli. Biochem. J. 44:87–90.

238. Wiggins, P. M. 1990. Role of water in some biological processes. Microbiol. Rev. 54:432–449.

239. Wilson, D. M., J. F. Alderete, P. C. Maloney, and T. H. Wilson. 1976. Proton motive force as the source of energy for adenosine 5'-triphosphate synthesis in Escherichia coli. J. Bacteriol. 126:327–337.

240. Winkler, H. H., and T. H. Wilson. 1966. The role of energy coupling in the transport of β-galactosides by Escherichia coli. J. Biol. Chem. 241:2200–2211.

241. Wise, J. G. 1990. Site-directed mutagenesis of the conserved β subunit tyrosine 331 of Escherichia coli ATP synthase yields catalytically active enzymes. J. Biol. Chem. 265:10403–10409.

242. Woodhull, A. 1973. Ionic blockade of sodium channels in nerve. J. Gen. Physiol. 61:687–708.

243. Yoshida, M., H. Okamaoto, N. Sone, H. Hirata, and Y. Kagawa. 1977. Reconstitution of thermostable ATPase capable of energy coupling from its purified subunits. Proc. Natl. Acad. Sci. USA 74:936–940.

244. Zhang, Y., M. Oldenburg, and R. H. Fillingame. 1994. Suppressor mutations in F_{1 subunit ε} recouple ATP-driven H⁺ translocation in uncoupled Q42E subunit c mutant of Escherichia coli F₁F₀ ATP synthase. J. Biol. Chem. 269:10221–10224.

245. Zilberstein, D., S. Schuldiner, and E. Padan. 1979. Proton electrochemical gradient in Escherichia coli cells and its relation to active transport of lactose. Biochemistry 18:669–673.