Animal Plasma Membrane Energization by Chemiosmotic H⁺ V-ATPases

Our primary purpose in this review is to argue that plasma membranes in many animal epithelial cells are energized by proton-motive forces generated by H⁺ V-ATPases. After proton-motive forces have been defined, the isolation and identification of H⁺ V-ATPases in Manduca sexta goblet cell apical membranes will be discussed in relation to the cloning and sequencing of most of the subunits. Noting evidence that the so-called K⁺ pump is composed of an H⁺ V-ATPase, K⁺/2H⁺ antiporter in the goblet cell apical membrane (GCAM), we will refute a claim that the V-ATPase transports K⁺ rather than H⁺. Many animal plasma membranes that are energized by proton-motive forces will be described. The divergent evolution of V-ATPases from F-ATP synthases will be discussed in relation to Michael Forgac’s hypothesis that reducing conditions favor H⁺ V-ATPase activation. The suspicion that either active or inactive H⁺ V-ATPases are present in apical plasma membranes of nearly all animal epithelial cells will be voiced. A hypothesis that mitochondria act as ‘reducing agents’ will be used to explain why active H⁺ V-ATPases are found so frequently in so-called ‘mitochondria-rich’ cells. Finally, voltage-driven K⁺/2H⁺ antiport and animo acid:K⁺ symport will be discussed.

The constancy of intracellular constituents, like that of extracellular fluids (Bernard, 1878/1879), implies a balance between energy-trapping and energy-utilizing mechanisms. The concentration of ATP, the key constituent, is held constant within cells by a balance between ATP synthesis and ATP utilization. Two ways to make ATP have evolved in nature – chemiosmotic coupling of ATP synthesis to proton electrochemical gradients and covalent coupling of ATP synthesis to chemical reactions via phosphorylated intermediates. Chemiosmotically coupled ATP synthesis is catalyzed by H⁺ F-ATP synthases, such as those in the oxidative phosphorylation pathway of mitochondria (see Fillingame, 1997). Covalently coupled ATP synthesis is catalyzed by phosphoenzymes, such as those in the glycolytic pathway. Two ions have evolved as primary membrane energizers – protons and sodium ions. ATP hydrolysis (ΔG_ATP/m) via H⁺ F-ATPase, H⁺ V-ATPase and H⁺ P-ATPase yields a proton electrochemical gradient ⁠. Expressed in volts, is called the proton-motive force, Δp (Mitchell, 1961). At 25 °C. Δp≈60 mV(ΔpH)+ΔΨ. Similarly, ATP hydrolysis via the Na⁺/K⁺ P-ATPase yields a Na⁺ electrochemical gradient (see below for definitions of variables and Appendix for further discussion of transport nomenclature).

Just as continuous chains of electrons carry current rapidly through metals, continuous chains of protons, hydrogen-bonded to water, carry current rapidly through aqueous solutions. Because ion-translocating channels are water-lined, this continuity results in protons having nearly an order of magnitude higher mobility than any other ion in channels (Hille, 1992). It is therefore not surprising that protons are widespread carriers of electrical current through biological membranes. The concentration of protons is seldom greater than 10⁻⁴ mol l⁻¹ in cells, so proton currents do not contribute significantly to membrane conductances or cellular osmotic balance. Proton electrochemical gradients are generated by P-ATPases that operate with a phosphorylated intermediate, e.g. the H⁺ P-ATPase of yeast, fungal and plant plasma membranes and the K⁺/H⁺ P-ATPase of gastric mucosa. Proton electrochemical gradients are also generated chemiosmotically, i.e. without any chemical intermediate (Mitchell, 1961) by H⁺ V-ATPases. In rare cases, Na⁺ gradients are generated chemiosmotically by Na⁺ V-ATPases (e.g. Kakinuma and Igarashi, 1989, 1994).

Proton electrochemical gradients and ATP are interconvertible currencies – proton electrochemical gradients drive ATP synthesis and ATP hydrolysis drives the formation of proton electrochemical gradients (Harold, 1986). This remarkable interconversion between ATP and proton electrochemical gradients is found throughout nature. In bacteria, proton electrochemical gradients, which are generated across the plasma membrane by electron-transport-dependent proton pumps, drive the H⁺ F-ATP synthase as well as various antiporters and symporters (Fig. 1). In yeast and fungi, an H⁺ P-ATPase generates a proton electrochemical gradient across the plasma membrane, whereas an H⁺ V-ATPase energizes the vacuolar membrane (Anraku et al. 1989; Fig. 2). The situation is similar in plants, where the proton-motive force across the plasma membrane is generated by an H⁺ P-ATPase and the tonoplast membrane is energized by an H⁺ V-ATPase (Fig. 3). Finally, animal endomembranes are energized by an H⁺ V-ATPase. It is only in animal plasma membranes that proton currrents are thought to be unimportant – few people believe that ATP hydrolysis sets up proton electrochemical gradients across animal plasma membranes and that these gradients drive secondary symporters and antiporters, just as they do in all the rest of the living world (Fig. 4), even though proton-linked antiporters, e.g. the Na⁺/H⁺ exchanger, are common in animal plasma membranes.

Na⁺, not H⁺, is widely believed to energize animal plasma membranes. The fragile animal plasma membrane is osmotically vulnerable; water would be expected to enter the hypertonic cytoplasm, causing the cells to swell and burst. However, the Na⁺/K⁺ P-ATPase pumps Na⁺ out of cells, incidently maintaining a Na⁺ chemical gradient with [Na⁺] high outside and low inside. Simultaneously, the Na⁺/K⁺ P-ATPase pumps K⁺ into cells. Cell membranes have K⁺ channels through which K⁺ diffuses out faster than any other ion and, almost instantaneously, an inside-negative voltage balances the K⁺ gradient. The cell membrane is relatively impermeable to Na⁺, so both this voltage and the Na⁺ chemical gradient remain poised to drive Na⁺ back into cells, thereby doing transport work (Dean, 1941). For example, glucose is driven into cells via the glucose:Na⁺ symporter (cotransporter) and Ca²⁺ is driven out of cells via the Ca²⁺/Na⁺ antiporter (exchanger) (Hediger, 1994).

Numerous alternatives notwithstanding, the global bioenergetic scheme is simple and elegant. In chemiosmotic systems of bacteria, yeast and plants, proton electrochemical gradients drive ATP synthesis and ATP hydrolysis forms proton electrochemical gradients for transport work (Harold, 1986). In covalent systems of these organisms, ATP synthesis via phosphointermediates (e.g. during glycolysis) and P-type ATPases form H⁺ and Na⁺ gradients for transport work. Animal plasma membranes strike the sole dissonant chord. Na⁺-linked covalent energization, not H⁺-coupled chemiosmotic energization, is dominant in animal plasma membranes. The Na⁺ paradigm is rock solid. Dean’s (1941) Na⁺ pump, Hodgkin and Huxley’s (1952) resting and action potentials and Skou’s (1957) isolated Na⁺/K⁺-ATPase, together with thousands of studies on vertebrate nerve, muscle, kidney, intestine and red blood cells and other animal cells and tissues, turned the Na⁺ paradigm into the Na⁺ dogma. But, are there no exceptions to this dogma, no remnants of the elsewhere-prominant, proton-centered scheme, in animal plasma membranes?

Yes, there were exceptions to the Na⁺ paradigm from the outset. Insect Malpighian tubules are energized by K⁺, not Na⁺ (Ramsay, 1953). The isolated lepidopteran midgut, leached of its small amount of Na⁺, can still pump K⁺ for hours (Harvey and Nedergaard, 1964). Insect salivary glands (Berridge et al. 1976) and Malpighian tubules (for a review, see Maddrell, 1991) use the K⁺ pump to drive fluid secretion. Insect sensory sensilla use the K⁺ pump to generate the receptor potential (for a review, see Thurm and Küppers, 1980). K⁺ pump particles, portasomes (Harvey, 1980), are visible in electron micrographs of apical membranes in orthopteran rectum (Gupta and Berridge, 1966) and lepidopteran midgut (Anderson and Harvey, 1966); they are remarkably like the particles studding the membranes of the proton-secreting epithelia of vertebrates (Brown et al. 1987). The insect K⁺ pump generates a 240 mV gradient that depends moment to moment upon oxygen uptake (Wood et al. 1969; Dow and Peacock, 1989). The K⁺ pump is insensitive to ouabain – the tissue does not even bind this cardiac glycoside (Jungreis and Vaughan, 1977). Using the presence of portasomes as an assay, the K⁺-pump-containing goblet cell apical membrane (GCAM) was isolated (Cioffi and Wolfersberger, 1983; for a review, see Harvey et al. 1983) and K⁺-stimulated ATPase activity was localized overwhelmingly in vesicles from GCAM (Wieczorek et al. 1986). Clearly, Na⁺ is replaced by K⁺ in these insect epithelia.

But do protons play any role? The 10 000-fold proton gradient across M. sexta midgut GCAM (Dow, 1984) was found to be balanced by a voltage of −240 mV (Dow and Peacock, 1989), leaving no net proton electrochemical gradient; could the voltage that sustains this pH gradient be generated by a perverse proton pump, whose output side is alkaline? Wieczorek and coworkers provided an astonishing ‘yes’ to that question. When the K⁺-stimulated ATPase in purified GCAM was solubilized, it turned out to be a V-ATPase (Schweikl et al. 1989). This conclusion, based on the ATP stimulation of GCAM vesicle acidification as well as on the substrate and inhibitor specificity and the migration of its subunits on gels, has been confirmed overwhelmingly by the high identity of its sequenced subunits with those of other V-ATPases (for reviews, see Wieczorek, 1992; Merzendorfer et al. 1997). But perhaps the enzyme is a K⁺ V-ATPase; a Na⁺ V-ATPase has been identified in Enterococcus hirai (e.g. Kakinuma and Igarashi, 1994), so a V-ATPase that pumps K⁺ rather than H⁺ would not be completely out of the question. Harvey et al. (1981, 1983) had long ago postulated such an enzyme, on the basis of electrical, structural and thermodynamic resemblances between the K⁺ pump and mitochondrial F-ATPase. The GCAM enzyme turned out to be an H⁺ V-ATPase (Schweikl et al. 1989); moreover, it turned out to be the primary part of the K⁺ pump; the secondary part was a novel K⁺/nH⁺ antiporter (Wieczorek et al. 1991; for a review, see Lepier et al. 1994). The key finding in this astounding development was that acidified GCAM vesicles are alkalized upon addition of K⁺ to ATP-free preparations. Moreover, the ATPase activity could be inhibited independently of the antiporter by bafilomycin and the antiporter could be inhibited independently of the ATPase by amiloride. Immunocytochemical labelling demonstrated that the primary part of the ‘K⁺ pump’ in other insect tissues was also a V-ATPase (Klein, 1992).

Recently, Küppers and Bunse (1996) have re-invoked the K⁺ V-ATPase hypothesis after showing that the receptor current in cockroach sensory sensilla is not affected by the antiporter inhibitors amiloride and harmaline. Such resistance to antiporter inhibitors might be explained by the failure of inhibitors to reach their binding site on the antiporter at effective concentrations in vivo. They suggested that mutations might transform a proton-translocating V-ATPase into a K⁺-translocating V-ATPase. Such a transformation is highly unlikely because Dow et al. (1992) found that the M. sexta proteolipid is virtually identical with known H⁺ V-ATPase proteolipids. To alter the Escherichia coli F-ATPase so that Li⁺ inhibited H⁺ transport required the mutation of four residues (Zhang and Fillingame, 1995); even after this extensive change, Na⁺ did not inhibit H⁺ transport and the transport of neither Li⁺ nor Na⁺ was demonstrated. Finally, Küppers and Bunse (1996) asserted that driving K⁺/nH⁺ antiport with the voltage generated by an adjacent H⁺ V-ATPase violates the Second Law of Thermodynamics.

The list of animal plasma membranes that are energized by protons is long and growing rapidly (Table 1). On the basis of the physiological, biochemical and molecular evidence reviewed above, the lepidopteran midgut is energized by protons. Based mainly upon labelling with fluorescent antibodies to V-ATPase subunits and/or bafilomycin sensitivity, not only insect midguts (Schweikl et al. 1989) but also salivary glands (Just and Walz, 1994), rectum (U. Klein, M. Timme, F. J. S. Novak, A. Lepier, W. R. Harvey and H. Wieczorek, in preparation), Malpighian tubules (Zhang et al. 1994) and sensory sensilla (Klein and Zimmermann, 1991) are all energized by protons. Based mainly upon physiology, protein biochemistry, molecular biology and immunocytochemistry, vertebrate kidney and urinary bladder (for a review, see Gluck et al. 1992), epididymis (Brown et al. 1992; for a review, see Brown and Breton, 1996), phagocytes (for a review, see Grinstein et al. 1992), osteoclasts (Vaananen et al. 1990; for a review, see Chatterjee et al. 1992) and even frog skin (for a review, see Ehrenfeld and Klein, 1997) all contain specialized cells with plasma membranes energized by H⁺ V-ATPases. In fact, a pattern is emerging – it is usually the apical plasma membrane of epithelial cells that is energized by protons – just as it is usually the basolateral plasma membrane that is energized by sodium ions. Apparently, animal plasma membranes are not exceptions to the proton paradigm – there is ample proton coupling in animal plasma membranes as well as in plasma membranes of other living organisms. The regulation of V-ATPases is reviewed by Merzendorfer et al. (1997) and by Dow et al. (1997).

The proton energization of specific apical plasma membranes presents both clinical and industrial opportunities. Although some cancer cell plasma membranes are energized by H⁺ V-ATPases, it probably is not worthwhile to focus on the nearly ubiquitous V-ATPase itself for clinical intervention. More promising are specific channels and H⁺-linked porters in these proton-energized membranes. Also promising are the special conditions, such as extreme acidity or alkalinity surrounding the membranes, that can be exploited to activate targeted control agents.

It is relatively easy to obtain pharmacological and cytological evidence that a membrane is energized by an H⁺ V-ATPase. The specific V-ATPase inhibitors bafilomycin and concanamycin (for a review, see Drose and Altendorf, 1997) are commercially available. Fluorescent antibodies to V-ATPase subunits can be obtained for immunocytochemistry (Klein, 1992). Finally, the V₁ domains of the V₁V_o ATPase are likely to appear as portasomes, which are visible in electron micrographs (for a review, see Harvey, 1992).

What is the mechanism by which H⁺ V-ATPases are targeted to the plasma membrane in some cases and to endomembranes, such as lysosomes, in other cases? Differential targeting would seem to imply that plasma membrane and endomembrane H⁺ V-ATPases differ in structure within a single cell (Merzendorfer et al. 1997). An alternative explanation might be that a single form of the ATPase is carried by membrane cycling between plasma membranes and endomembranes (e.g. Lacoste et al. 1993). This explanation implies that H⁺ V-ATPases are present in plasma membranes as well as in endomembranes of all eukaryotic cells. The presence of an H⁺ V-ATPase is implied in any plasma membrane that undergoes endocytosis and produces early endosomes which contain H⁺ V-ATPases. This recycling hypothesis is illustrated in Fig. 5 (modified from Forgac, 1992; Mellman, 1992). The hypothesis is complicated at its inception by the argument of Mellman’s group (Fuchs et al. 1994) that acidifying H⁺ V-ATPases are not likely to be present in early endosomes and the finding by Sabolic et al. (1992) that early endosomes are labeled only by antibodies to the B subunit. However, M. Forgac (personal communication) has countered with evidence that early endosomes are also labelled by antibodies to the A subunit.

If early endosomes contain H⁺ V-ATPase, even in an inactive form, then the following hypothesis for differential targeting is viable (Fig. 5). In the ‘constitutive exocytotic’ pathway (green arrows) simplified from Forgac (1992), secretory vesicles with H⁺ V-ATPases and Cl⁻ channels bud off from the trans-Golgi. They fuse with the apical plasma membrane and discharge their contents to the exterior, leaving behind H⁺ V-ATPases and Cl⁻ channels in the plasma membrane. As ligands bind to receptors on the plasma membrane, clathrin-coated pits form and the inactive H⁺ V-ATPases, closed Cl⁻ channels and ligand–receptor complexes are pinched off as non-acidified, clathrin-coated vesicles which soon lose their coat to become early endosomes. Binding of V-ATPase to the vesicles is thought to be mediated by a 50 kDa subunit of the AP-2 adaptin complex (Myers and Forgac, 1993). The early endosomes fuse with the membrane of the compartment for uncoupling receptor and ligand (CURL) where the V-ATPase is activated, the Cl⁻ channels open and provide CURL with additional acidifying capacity to release the ligands from receptors. Finally, vesicles, containing ATPases and channels, bud off from CURL and return to the apical plasma membrane. This pathway implies that at least small numbers of V-ATPase and Cl⁻ channels are always present in apical plasma membranes, although they may be inactive there. The existence of pinocytotic vesicles containing H⁺ V-ATPase would also imply the constitutive presence of this enzyme in the plasma membrane. However, a new question arises: what turns on V-ATPases in the cell interior and turns them off in the plasma membrane?

Feng and Forgac (1994) have focused attention on Cys254 of the catalytic subunit A in coated-vesicle H⁺ V-ATPase. This residue is in the consensus sequence, GXGKTV, that is found in many nucleotide-binding proteins (Walker et al. 1984) and is near the triphosphate portion of ATP in the tertiary structure of the enzyme. They propose a novel mechanism for regulation of vacuolar acidification in which Cys254 is oxidized and forms a disulfide bond with Cys532 that inactivates the enzyme. Similar ‘redox modulation’ has been proposed for the human glutamate transporter (Trotti and Hediger, 1996). We propose that oxidizing conditions are likely around many plasma membranes and that the enzyme would be inactivated there. However, in the cell interior, where reducing conditions are maintained by agents such as glutathione, the critical cysteine residues would be reduced, the constraining disulfide bond would be broken and the enzyme would be activated (see also Merzendorfer et al. 1997). The implication that if H⁺-V-ATPase-containing endosomes bud off from animal plasma membrane then V-ATPases must be ubiquitous, albeit inactive, constituents of animal plasma membranes has so far been overlooked.

The presence of ubiquitous, constitutive plasma membrane V-ATPases is consistent with current views regarding their evolution (Nelson, 1989). On the basis of an analysis of the distribution of highly conserved sequences in H⁺ A-ATPases of present-day archaebacteria, H⁺ F-ATP synthases from mitochondria and H⁺ V-ATPases from yeast vacuoles, Nelson and Taiz (1989), Ihara et al. (1992), Bakker-Grunwald (1992) and others propose that all three present-day enzymes derive from an A-ATPase-like common ancester.

The key event in this divergence of H⁺ V-ATPases from H⁺ F-ATP synthases is thought to be the introduction of oxygen into the atmosphere by cyanobacteria. H⁺ V-ATPases, with their inactivating disulfide bridge involving Cys254 of the ATP binding site, could not function in the oxidizing environment at the plasma membrane. Mutants lacking this regulatory cysteine are thought to have initiated the H⁺ F-ATP synthase line, starting with plasma membrane H⁺ F-ATP synthases of bacterial plasma membranes. Thus, H⁺ V-ATPases would be selected against in bacteria where, indeed, they are seldom, if ever, found. But in all eukaryotes the friendly, reducing atmosphere of the cytoplasm would allow H⁺ V-ATPases to remain active and survive. Finally, since there would be a selective advantage in retaining the secretory pathway that targets them to the plasma membranes, H⁺ V-ATPases would remain there as constitutive but inactive components, to be activated upon demand.

The plasma membranes of macrophages and the ‘apical’ plasma membranes of their cousins the osteoclasts probably reside under conditions as reducing as those deep within cells – it is not surprising that their plasma membrane H⁺ V-ATPases are active. The apical plasma membranes of many epithelial cells likewise probably face reducing conditions, e.g. the lumen of insect Malpighian tubules. Particularly striking are the many ‘mitochondria-rich’ cells that are now known to contain active H⁺ V-ATPases on their apical plasma membranes, e.g. the columnar cells of insect Malpighian tubules (Beams et al. 1955), the goblet cells of lepidopteran larvae (Anderson and Harvey, 1966) and the trichogen and tormogen cells of insect sensory sensilla (for a review, see Thurm and Küppers, 1980). Mitochondria-rich cells have been described in many renal tissues and even in frog skin. Most recently, they have been described in epididymis (Breton et al. 1996; for a review, see Brown et al. 1997). All of these mitochondria-rich cells are heavily studded with portasomes (Harvey, 1980), the visible manifestation of the V₁ domain of H⁺ V-ATPases, on their apical membranes (for a review, see Brown and Breton, 1996). The usual explanation for the mitochondria-richness of these cells is that it provides abundant ATP for the dense arrays of H⁺ V-ATPases. However, in goblet cells of posterior caterpillar midgut, mitochondria are not adjacent to the apical membrane (Cioffi, 1979), yet this region transports K⁺ as well as the anterior regions (Cioffi and Harvey, 1981), where mitochondria and plasma membrane are exquisitely intimate.

An alternative explanation for the mitochondria-richness may be the capacity of cytochrome oxidase to remove oxygen from the environment of nearby plasma membrane V-ATPases. Chance (1957) has shown that this enzyme has such a high affinity for oxygen that it remains fully oxidized at oxygen concentrations as low as 4 μmol l⁻¹ (Fig. 6). This property means that mitochondria can continue undiminished ATP production while creating the oxygen-poor environment that would allow plasma membrane H⁺ V-ATPases to remain active. The intimacy of H⁺-F-ATP-synthase-containing mitochondrial inner membranes and H⁺-V-ATPase-containing plasma membranes reaches shocking proportions in lepidopteran midgut (Cioffi, 1979) and in insect Malpighian tubules where mitochondria move into microvilli upon activation (Bradley and Satir, 1981). These two tissues hold world records, respectively, for the magnitutide of the short-circuit current (2 mA cm⁻²; for a review, see Harvey and Zerahn, 1972) and fluid secretion rate (one-third of the cell volume per minute; for a review, see Maddrell, 1991).

If two aqueous compartments are separated by a membrane that is ideally impermeable to all ions, including protons, and if an H⁺ V-ATPase is inserted into the membrane, then the hydrolysis of added ATP would merely separate H⁺ from A⁻, where A⁻ is an anion, across the membrane. There would be no net H⁺ or A⁻ flux and therefore no pH change in the output compartment. Assuming ideal coupling between pump and ATP hydrolysis, ΔΨ was estimated to be 240 mV for the lepidopteran midgut, using data supplied by Mandel et al. (1980); this value was calculated from the relationship, ΔΨ=ΔG_ATP/2F=RT/2F(lnK_e/Q), where RT/2F≈30 mV at 25 °C and K_e/Q is the products-to-reactants concentration ratio at equilibrium/actual conditions (Harvey, 1992). In the presence of a Cl⁻ channel, Cl⁻ would follow the H⁺ and the output compartment would be acidified, as in most endomembranes. In lepidopteran midgut, a K⁺/2H⁺ antiporter is present in the same membrane as the H⁺ V-ATPase. ΔΨ drives the pumped proton along with a second proton back into the cell in exchange for a K⁺ (Azuma et al. 1995). In vivo the K⁺ activity is held steady at approximately 140 mmol l⁻¹ on both sides of the membrane, the [Na⁺] being low and also equal on both sides (Dow and Harvey, 1988); the entire energy from the ATP hydrolysis appears as a measured transmembrane voltage of approximately 240 mV (Dow and Peacock, 1989); thus, coupling between ΔG_ATP/2F and ΔΨ must be almost ideal, as it is in other chemiosmotically coupled processes.

Giordana et al. (1982, 1989) identified six amino acid:K⁺ symport systems in brush-border membrane vesicles in larval midgut from Philosamia cynthia with specificities for neutral amino acids, L-proline, glycine, L-lysine, glutamic acid and D-alanine. In vesicles from larval M. sexta, three types of amino acid:K⁺ symporter have been identified. (1) A broad-spectrum, zwitterionic amino acid:K⁺ symporter, system B (Hennigan et al. 1993), was found that resembled the one in P. cynthia. A cooperative effort, between Italian and American groups, is under way to clone system B by expressing M. sexta poly(A) RNA in Xenopus laevis (Castagna et al. 1997). (2) A K⁺ symporter that accepts either proline or glycine was found in M. sexta (Bader et al. 1995) rather than the two separate systems of P. cynthia. (3) A cationic lysine, arginine:K⁺ symporter, system R⁺ (Liu and Harvey, 1996) is probably identical with the ‘L-lysine’ system of P. cynthia, but lysine uptake was non-competitively inhibited by arginine in M. sexta, as it was in P. cynthia, so the system probably transports only cationic arginine in vivo. System R⁺ has a substrate spectrum almost identical to that of the mammalian cationic amino acid uniporter, y⁺, so probes constructed from y⁺ cDNA (MacLeod et al. 1992) are being used in an attempt to clone R⁺ (D. Feldman, B. Stevens and W. R. Harvey, unpublished results). No specific glutamate:K⁺ symporter was found in M. sexta; although glutamate was accumulated in response to a K⁺ gradient, countertransport accumulation of glutamate was not demonstrable (Xie et al. 1994). The voltage alone is thought to drive all of these amino acid:K⁺ symporters; thus, the V_max of the symporters for cation is increased with voltage (Giordana and Parenti, 1994; Parthasarathy and Harvey, 1994). Similarly, the voltage alone is thought to drive K⁺/2H⁺ antiport across the goblet cell apical membrane in vivo (Wieczorek et al. 1991).

Although K⁺/2H⁺ antiport from the midgut lumen in vivo would be difficult because [H⁺]_lumen is 10⁻¹¹ mol l⁻¹, antiport from the goblet cavity would be easier because [H⁺]_cavity≈10⁻⁷ mol l⁻¹. How can such a low concentration of H⁺ drive antiport? The answer is that the antiport is driven by the approximately −240 mV voltage. The effect of voltage is such that every 60 mV at approximately 25 °C approximates a tenfold increase in the effective concentration of the ionic species. Although this explanation is correct thermodynamically, the mind searches for an intuitive explanation of how a voltage can drive transport. Alan Hodgkin offered a suggestion, which is consistent with statistical mechanics diffusion theory – diffusion is like fleas hopping and electrodiffusion is like fleas hopping in a breeze. Thus, fleas that were hopping in a gentle breeze of 0.0l6 km h⁻¹ with no voltage would be blasted with a hurricane of 160 km h⁻¹ (=100 m.p.h.) at a voltage of 240 mV. In other words, the probability of particles hitting the positive side of the membrane would be 10 000 times greater with a voltage of 240 mV than with no voltage, and the effective [H⁺]_cavity would increase to 10⁻³ mol l⁻¹. In the M. sexta midgut, the high voltage can drive the pump–antiporter couple to an alkalization of 3 pH units (Azuma et al. 1995), producing a lumen pH of 11, if the cell pH is 8. In Malpighian tubular lumen and sensory sensillar lymph, where in each case pH≈7, the V-ATPase–antiporter couple would also not face an impossible task. However, the mosquito larval midgut presents a challenge; the lumen pH is approximately 12 (Dadd, 1975) and neither a goblet cavity nor a large ΔΨ has yet been reported.

As immunocytochemistry and molecular biology allow increasingly sophisticated studies on plasma membranes and endomembranes of epithelial cells, the classical, Hodgkin–Huxley analysis of membrane function becomes increasingly limited. In nerve and muscle, miniscule ion fluxes lead to 100 mV changes in membrane potential, whereas in rapidly transporting epithelia, massive ion fluxes are accomplished at constant voltage. Hodgkin and Huxley could view a neuron as a two-compartment system, inside and outside, and needed only to consider two ions, Na⁺ and K⁺. By contrast, even the simplest epithelium consists of at least a basal extracellular space, the cell and the apical compartment, as well as several intracellular compartments. Nevertheless, even a complex epithelium is much simpler than, say, the circuit in a modern-day computer or television set. Martin (1992) suggested that the same type of circuit analysis that has been useful in designing computers and televisions would be useful in analysing epithelia. Key aspects of amino acid:K⁺ symport and K⁺/nH⁺ antiport were simulated using reasonable values from experimental data (Martin and Harvey, 1994). What contribution this adaptation of circuit theory to complex biological systems will make remains to be determined.

ATP hydrolysis via H⁺-coupled ATPases yields proton electrochemical gradients; thus, ΔG_ATP/m=RTln([H⁺_i]/ ⁠, where ΔG_ATP/m is the free energy of ATP hydrolysis divided by the number of protons transported per ATP. Dividing by F, to express the value in volts, and noting that z_H+=+1, yields Mitchell (1961) called this last term, the proton-motive force or Δp. At 25 °C, ΔG_ATP/F≈60 mV(ΔpH)+ΔΨ. For example, across the goblet cell apical membrane (GCAM) in Manduca sexta larval midgut, ΔG_ATP/F≈500 mV and ΔG_ATP/mF≈−250 mV (Mandel et al. 1980). Surprisingly, there is little or no pH gradient across GCAM (Chao et al. 1991), so 60 mV(ΔpH)≈0, but ΔΨ was found to be approximately −240 mV (Dow and Peacock, 1989). So, in this case, ATP hydrolysis is tightly coupled to the generation of a transmembrane voltage.

When one approaches the broader literature that includes bacteria, yeast and fungi, plants and insects, one feels like Alice when told that ‘a word means exactly what I say it means’. Moreover, transporter designations are confusing. Who would guess that the H⁺-ATPase of plant apical membranes is a proton-translocating ATPase with a phosphorylated enzyme in its catalytic cycle and that the V-ATPase of the vacuole is a proton-translocating ATPase that acts with no such phosphoenzyme? A more ‘user-friendly’ designation scheme is suggested in Table 2. In particular, we follow a suggestion by Peter Maloney (personal communication) that cotransported (symported) solutes should be separated by a colon (:), thus glucose:Na⁺ symporter, and that exchanged (antiported) solutes should be separated by a slash (/), thus Na⁺/H⁺ antiporter. We also suggest that the convention ‘driver/driven’ be used if possible. Thus, glucose:Na⁺ symport denotes that glucose uptake is driven by Na⁺ and K⁺/H⁺ antiport denotes that K⁺ secretion is driven by H⁺. Since both types of porter operate in either direction, depending upon electrochemical gradients, a great deal of judgement is still involved.

This review was supported in part by research grants from the National Institutes of Health (AI-22444 and AI 30464) to W.R.H. and the Deutsche Forschungsgemeinschaft (Wi 698) and EEC (SCI*-CT90-0480) to H.W. We thank Daniel J. Harvey for assistance with the art work.