Vacuolar-type proton pumps in insect epithelia

Insects exhibit a unique alkali metal ion pump which actively transports,depending on the species and/or the physiological circumstances, K⁺and/or Na⁺ across epithelia. The pump was initially detected in Malpighian tubules from various insect species(Ramsay, 1953) where it was suggested to actively transport K⁺ from the hemolymph into the tubule lumen. Subsequently, the alkali metal ion pump was (1) established in other ion-transporting epithelia such as salivary glands, labial glands,midgut and sensory sensilla, (2) shown to be located in the apical membranes of epithelial cells, and (3) demonstrated to be electrogenic and energized by the hydrolysis of ATP (for a review, see Harvey, 1982).

One of the longstanding model systems for the investigation of active transepithelial K⁺ transport is the larval midgut of the tobacco hornworm Manduca sexta. Over the years, Bill Harvey and his coworkers as well as David Moffett and Alan Koch have provided many important and useful insights into the function of the midgut K⁺ pump. Moira Cioffi and Mike Wolfersberger from the Harvey laboratory were the first to explore the biochemical identity of the pump. They identified a K⁺-stimulated ATPase activity in partially purified plasma membranes and then in highly purified apical membranes of goblet cells(Wolfersberger et al., 1982; Harvey et al., 1983). One of us (H.W.) joined the Harvey laboratory for a collaboration that led to the unequivocal identification and characterization of the K⁺-stimulated ATPase activity(Wieczorek et al., 1986),thanks to the availability of (1) highly purified goblet cell apical membranes(Cioffi and Wolfersberger,1983) and (2) a very sensitive method for measuring inorganic phosphate as the product of ATP hydrolysis(Wieczorek, 1982).

Back in Germany, H.W. together with Helmut Schweikl detergent-solubilized and purified the ATPase from apical membranes of goblet cells and found it to be a vacuolar-type ATPase on the basis of the molecular mass of the holoenzyme, its subunit composition, its sensitivity to N-ethlymaleimide and its insensitivity to azide and to vanadate(Schweikl et al., 1989). The result was rather baffling because, at that time, V-ATPases were known to be proton pumps of intracellular organelles. The only V-ATPase which had been found in the plasma membrane of cells was that of the vertebrate urinary bladder (Al Awqati, 1978).

In experiments using membrane vesicles, the tobacco hornworm V-ATPase was shown to transport H⁺, and not alkali metal ion(Wieczorek et al., 1989). What transported K⁺ in these membrane vesicles turned out to be a K⁺/H⁺ antiporter that these membrane vesicles also housed (Wieczorek et al.,1989; Wieczorek et al.,1991). Thus the molecular correlate of the midgut alkali metal pump emerged as an electrogenic H⁺-transporting V-ATPase that energizes K⁺/H⁺ antiport in the same membrane. Subsequent studies elucidated the electrophoretic stoichiometry of K⁺/2H⁺ (Azuma et al.,1995). The discovery of K⁺/2H⁺ antiport explained why in the earlier studies a K⁺-stimulated ATPase activity had been measured. When protons are transported by the V-ATPase into vesicles, a voltage as well as a pH difference develop across the membrane that counteract further H⁺ transport, thereby decreasing V-ATPase activity. However, in the presence of K⁺/2H⁺ antiport,H⁺ can leave the vesicle, thereby diminishing the voltage and pH difference and maintaining pump activity.

The new paradigm, a H⁺-transporting V-ATPase forming a partnership with a K⁺/nH⁺ antiporter, was quickly accepted by the scientific community and transferred to other ion-transporting epithelia such as Malpighian tubules and salivary glands. Today, the V-ATPase has been demonstrated in numerous ion-transporting insect epithelia (Harvey and Wieczorek,1997; Harvey et al.,1998). The molecular evidence for the antiporter, however, has been elusive. Recently genes that may encode the antiporter have been identified (Day et al., 2008; Rheault et al., 2007).

While the V-ATPases were isolated from cell organelles, they also occur in many animal plasma membranes where they energize diverse membrane transport systems (for a review, see Wieczorek et al., 1999). The V-ATPase from goblet cell apical membranes of the tobacco hornworm midgut was the first vacuolar-type ATPase shown to energize secondary active transport in a eukaryotic plasma membrane(Fig. 1). In this review we will first focus on the V-ATPase in the midgut of the tobacco hornworm. We then will review the V-ATPase in Drosophila Malpighian tubules where the power of genomic and post-genomic approaches has been convincingly demonstrated. Finally we will turn to an integrated physiological approach in Malpighian tubules of mosquitoes that reveals the V-ATPase as the energizer of both transcellular and paracellular transport pathways.

V-ATPases are ubiquitous and highly conserved proton pumps that acidify specific organelles such as endosomes, lysosomes or secretory vesicles in every eukaryotic cell, but also are found in the plasma membrane of many animal cell types where they are involved either in pH homeostasis or in membrane energization (Nishi and Forgac,2002; Wieczorek et al.,1999; Beyenbach and Wieczorek,2006). Like the evolutionarily related F-ATPases of mitochondria,chloroplasts and bacteria, V-ATPases are heteromultimeric proteins comprising two functional parts, a peripheral catalytic V₁ complex and a membrane-bound, proton-conducting V_O complex(Fig. 2A). In the midgut of the tobacco hornworm, the V₁ complex of the plasma membrane V-ATPase contains the eight different subunits A to H, whereas the V_Ocomplex consists of the four different subunits a, c, d and e(Merzendorfer et al., 2000). Thus the insect V-ATPase is, in most of its facets, a typical representative of V-ATPases in general. Even though the yeast V-ATPase has been the main source for insights into the structure, function and regulation of this important family of ion pumps, research on the V-ATPase from tobacco hornworm midgut delivered significant findings of general interest. In the following we will concentrate on these aspects.

The V₁ subunit F was first cloned from tobacco hornworm midgut(Gräf et al., 1994). It was shown to be a constitutive V-ATPase subunit because monospecific antibodies inhibited both ATPase activity and ATP-dependent proton transport. The association of subunit F with the V₁ complex was confirmed by stripping it from the membrane by treatment with the chaotropic agent KI. The 14 kDa subunit F appears to be not as strictly associated with the V₁ complex as the other V₁ subunits because subunit F is absent from the V₁ complex after KI stripping(Gräf et al., 1996). Furthermore, subunit F appears to have a coupling role between the V₁ and V_O parts of the enzyme because neither the assembly of V₁ with V_O nor the assembly of the V_O complex itself takes place in the absence of subunit F(Graham et al., 1994). Most recent cryo-electron microscopy data from the tobacco hornworm holoenzyme also locate subunit F at a position consistent with V₁V_Olinkage (Muench et al.,2009).

The V₁ subunit G was initially cloned from yeast and interpreted as a member of the V_O complex, based on cold inactivation results,the properties of the null mutant, and its similarity to the bacterial F_O subunit b (Supekováet al., 1995). However, treatment of midgut goblet cell apical membranes and of Malpighian tubule brush border membranes from tobacco hornworm with chaotropic iodide as well as cold inactivation led to the conclusion that subunit G is a peripheral V-ATPase subunit(Lepier et al., 1996; Gräf et al., 1996). The finding that subunit G stimulates the ATPase activity of the reassembled V₁ complex supported the designation as a V₁ subunit(Xie, 1996). Structural and further biochemical studies also unequivocally allocate subunit G to the V₁ complex (Ohira et al.,2006; Zhang et al.,2008; Muench et al.,2009).

Subunit e was first cloned as M9.2, an extremely hydrophobic 9.2 kDa protein present in bovine chromaffin granule membranes(Ludwig et al., 1998). Because it showed sequence and structural similarity to Vma21p, a yeast protein required for the assembly of the V-ATPase, it was unclear whether it was a component of the mature enzyme. Studies with the tobacco hornworm V-ATPase revealed that this subunit indeed is a member of the V_O complex,and that it is a constitutive subunit of the mature holoenzyme(Merzendorfer et al., 1999). Meanwhile subunit e was also described as a member of the yeast V-ATPase and shown to be essential for enzyme activity(Sambade and Kane, 2004). Unlike the yeast and bovine subunit e, the insect subunit e from midgut and Malpighian tubules of the tobacco hornworm is highly glycosylated, with half of the 20 kDa molecular mass consisting of sugar residues(Merzendorfer et al., 1999). Whether this is typical for plasma membrane V-ATPases in general remains an open question.

In his excellent textbook `The Vital Force: A Study of Bioenergetics' Frank Harold provides a figure showing the schematic and hypothetical structure of the F-ATPase with the following caveat(Harold, 1986): `After several authors, none of whom should be held responsible'. That was the situation for V-ATPases as well until several papers appeared illuminating the picture further in the late nineties (Boekema et al., 1997; Boekema et al.,1998; Wilkens et al.,1999). The first examination of the quaternary structure of the isolated V₁ complex, performed by small-angle X-ray scattering of the tobacco hornworm V₁, revealed homologies to but also differences from the bacterial F₁ complex(Svergun et al., 1998). Electron microscopy of the negatively stained tobacco hornworm V₁complex supported the hexagonal arrangement of the A and B subunits together with a seventh mass either centrally or asymmetrically to the hexamer(Radermacher et al., 1999). The analysis of the three-dimensional structure of the V₁ complex at a resolution of 18 Å revealed further internal features of this protein (Radermacher et al.,2001). Moreover, the most detailed image of the whole V-ATPase was obtained by cryo-electron microscopy of the tobacco hornworm enzyme at a resolution of 17 Å (Muench et al.,2009) that is approximately 2-fold higher than previous cryo-EM models of either the V-ATPase or the F-ATPase holoenzyme. This unprecedented resolution of the holoenzyme uncovered a third stalk, indicating that there are three (and not just two) peripheral connections (stators) between the V₁ and the V_O complexes. Each stalk most probably consists of the V₁ subunits E and G (see Fig. 2).

The plecomacrolides bafilomycin and concanamycin were the first specific and highly potent inhibitors of V-ATPases to be detected, with IC₅₀values at nanomolar concentrations (Bowman et al., 1988; Dröse et al., 1993). In Manduca the IC₅₀ values were shown to be approximately 10 nmol l^–1(Huss et al., 2002). Despite intensive investigation of the structure–activity relationship the location of the binding site in the enzyme remained elusive for a long time. Studies with the chromaffin granule V-ATPase suggested that the membrane-bound V_O complex lodged the binding site(Hanada et al., 1990), and later studies with the bovine clathrin-coated vesicle V-ATPase reinforced this assumption, pointing to the V_O subunit a as the target(Crider et al., 1994; Zhang et al., 1994). Conversely, indirect evidence suggested the V_O subunit c as a candidate (Rautiala et al.,1993). Finally, mutational analysis of the V-ATPase from Neurospora crassa (Bowman and Bowman, 2002) as well as radiolabeling studies of the insect V-ATPase with a semi-synthetic derivative of concanamycin(Huss et al., 2002) proved that subunit c indeed contains the biggest part of the plecomacrolide binding site; a minor contribution of the V_O subunit a could be demonstrated by site-directed mutagenesis(Wang et al., 2005). The availability of a high resolution structure of the bacterial Na⁺–V_O ring from Enterococcus hirae, in combination with the present data from mutational analysis, revealed the location of the binding pocket in the luminal half of subunit c(Bowman et al., 2006; Murata et al., 2005).

More recently it was shown, using the tobacco hornworm midgut V-ATPase as a target, that the myxobacterial antibiotics archazolid and apicularen are further highly efficient and specific novel inhibitors of V-ATPases, with IC₅₀ values of approximately 20 nmol l^–1(Huss et al., 2005). Archazolid, in spite of its different structure compared with the plecomacrolides, probably has a similar mode of inhibition and binding site in the V-ATPase as it prevents radiolabeling of subunit c with the concanamycin derivative. Compared with this, the benzolacton enamides apicularen and salicylihalamide appear to exhibit a manner of inhibition which is different from that of the plecomacrolides as, on the one hand, they did not compete with concanamycin for its binding site and, on the other hand, they are the first V-ATPase inhibitors which discriminate between various species as they do not inhibit fungal V-ATPases. For a more comprehensive outline of V-ATPase inhibitors we refer to the recent review by Huss and Wieczorek(Huss and Wieczorek,2009).

The reversible disassembly of the V₁ complex as an in vivo control mechanism of V-ATPase activity was first demonstrated in the midgut of moulting tobacco hornworms. In the midgut, more than 10% of the animal's total ATP production is utilized by the H⁺-translocating V-ATPase to generate a voltage of more than 200 mV across the apical membrane of goblet cells which drives secondary K⁺/2H⁺ antiport and thus net active K⁺ secretion(Wieczorek and Harvey, 1995). The resulting K⁺ electrochemical potential energizes secondary transport processes across the midgut epithelium such as the uptake of amino acids by K⁺/amino acid symport. When moulting or starving, the caterpillars do not feed and thus there is no need for secondary transport. One would therefore expect for reasons of economy that active transepithelial K⁺ transport is diminished if not altogether stopped. Indeed, this was found to happen during moult due to dissociation of the peripheral V₁ subunits from the membrane V_O complex(Sumner et al., 1995). The V₁ subunits appeared to dissociate as a whole(Fig. 2B), as the cytosolic concentration of the V₁ complex was doubled during moult(Gräf et al., 1996). Subunit C seemed to be released separately to the cytoplasm(Fig. 2B), because the purified V₁ complex lacked most of it(Merzendorfer et al., 2000). The reverse process, namely the reassembly of the V₁V_Oholoenzyme, evidently does not require biosynthesis of new V₁subunits, because it was not influenced by inhibition of translation in moulting larvae with cycloheximide(Jäger and Klein, 1996). As expected, the V₁ complex dissociates reversibly from the V_O complex not only during moult but also during starvation,indicating that this type of V-ATPase regulation is a response to a drop in energy supply (Gräf et al.,1996). In vitro experiments with the isolated V₁V_O holoenzyme clearly showed that the dissociation is dependent on the ATP/ADP ratio, which represents the energy load of the cell(Huss and Wieczorek,2007).

Shortly after detecting the reversible disassembly of the V₁V_O complex in the tobacco hornworm, it was also observed in the yeast Saccharomyces cerevisiae(Kane, 1995). In cells grown in the presence of glucose, a rich carbon source, disassembly was induced by brief glucose deprivation. Upon restoration of glucose, rapid reassembly was observed, without the need for new subunit biosynthesis. Reversible disassembly was afterwards also detected in mammalian cells(Trombetta et al., 2003; Sautin et al., 2005). The observation of the reversible assembly of V₁ and V_Ocomplexes in three evolutionarily distant organisms (insects, vertebrates and yeast) indicates an ancient mechanism for regulating V-ATPase that has been conserved.

The regulation of V-ATPase activity by reversible disassembly has been best characterized in yeast. Apparently disassembly and reassembly are separately controlled processes, because disassembly but not reassembly requires intact microtubules (Xu and Forgac,2001), and because reassembly but not disassembly is aided by the heterotrimeric protein complex RAVE (Seol et al., 2001; Smardon et al.,2002). The glycolytic enzyme aldolase which binds V-ATPase subunits may play a role in stabilizing the assembled holoenzyme because overexpressing aldolase blocks V-ATPase disassembly upon removal of glucose(Lu et al., 2007). Recently it was shown that yeast cells with constitutively active protein kinase A (PKA)do not respond to glucose depletion by disassembly of the V-ATPase(Bond and Forgac, 2008). This means that either activated PKA causes the V-ATPase to (re)assemble or it prevents its disassembly.

In salivary glands of the blowfly Calliphora vicina serotonin induces a cAMP-mediated reversible assembly of the V-ATPase resulting in increased V-ATPase-driven proton transport(Dames et al., 2006). It could be shown that PKA is the target of cAMP, implying that there must be at least one protein in the signaling pathway leading to V-ATPase (re)assembly that is phosphorylated by PKA (Rein et al.,2008). One of the candidates may be the V₁ subunit C which is, in the tobacco hornworm V-ATPase, the only subunit that can be phosphorylated by PKA (Voss et al.,2007) (see also Fig. 3). Subunit C can be phosphorylated as a single recombinant protein as well as when embedded in the V₁ complex. In contrast it cannot be phosphorylated as part of the V₁V_O holoenzyme. Both the phosphorylated and the unphosphorylated forms of subunit C are able to reassociate with the V₁ complex from which subunit C had been removed before. In salivary glands of the blowfly a membrane-permeable cAMP analog as well as serotonin cause phosphorylation of subunit C but this is abolished by the PKA inhibitor H-89. These data are in line with the assumption that subunit C is a target for PKA-mediated phosphorylation and that this phosphorylation may be a regulatory switch for the (re)assembly of the V-ATPase. Indeed, subunit C is unique among V-ATPase subunits as it is released into the cytosol upon dissociation of the holoenzyme into its V₁ and V_O complexes(Kane, 2000; Merzendorfer et al., 2000). Subunit C is an elongated molecule (Drory et al., 2004) that is proposed to form part of a stator linking the V₁ with the V_O complex(Muench et al., 2009). Moreover, subunit C binds to actin filaments and this interaction may be involved in stabilizing the proton pump in its assembled state (see below)(Vitavska et al., 2003; Vitavska et al., 2005). These properties make subunit C suited to the control of V-ATPase (re)assembly state and to the mediation of the relevant cellular signals.

The interaction of the V-ATPase with the actin cytoskeleton was first demonstrated in mouse osteoclasts (Lee et al., 1999). Immunocytochemical and biochemical studies with a mouse bone marrow culture as well as with the bovine kidney V-ATPase suggested that the V-ATPase not only co-localizes with the actin cytoskeleton in vivo but also binds directly to actin filaments via its subunit B in vitro (Holliday et al.,2000).

In the tobacco hornworm, the co-localization of the V-ATPase with actin filaments was demonstrated at the apical membrane of midgut goblet cells as well as at the apical brush border membranes of epithelial cells from Malpighian tubules and salivary glands(Vitavska et al., 2003). In midgut goblet cells, no co-localization was observed under conditions where the V₁ complex detaches from the apical membrane. In co-pelleting assays, both the V₁ complex and the V₁V_Oholoenzyme bound to actin filaments in a concentration-dependent manner. Overlay blots revealed that not only subunit B but also subunit C binds to F-actin. Experiments with the recombinant subunit C also confirmed its interaction with F-actin under native conditions, which may explain why the holoenzyme, fully equipped with subunit C, exhibits a higher affinity for F-actin than the V₁ complex, which contains subunit C only in substoichiometric amounts. When the V₁ complex was reconstituted with the recombinant subunit C, enhanced binding to F-actin was observed. Incubation of F-actin with subunit C led to cross-linking of actin filaments,forming bundles of varying thickness(Vitavska et al., 2005). Subunit C was also shown to bind to monomeric G-actin with high affinity in the nanomolar range, and no difference could be found between ATP–G-actin or ADP–G-actin.

What may be the biological function of actin binding by subunit C? The goblet cell apical membrane of the tobacco hornworm midgut is densely packed with several thousand copies of V-ATPase. Therefore one may assume an actin-anchoring function, especially as spectrin, which usually links membrane proteins to the actin cytoskeleton, is absent in the apical sphere of goblet cells (Bonfanti et al., 1992). The high concentration of up to 10 μmol l^–1 free subunit C in the midgut cytoplasm allows us to extend this view. As subunit C binds non-preferentially to actin filaments and not only to their plus or minus ends, it could be an ideal candidate for the organization of actin dynamics in the apical cytoskeletal web.

The activity of the V-ATPase in Malpighian tubules of Aedes aegypti appears to correlate with the state and distribution of the actin cytoskeleton (Karas et al.,2005). Treatment of tubules with cAMP or blood feeding of adult mosquitos led not only to an increase of fluid secretion, indicating V-ATPase activation, but also to the redistribution and enrichment of actin at the apical membranes of principal cells (Karas et al., 2005). Accordingly, the protein associations described above for the tobacco hornworm may be vital for the secretory response to the blood meal in mosquito Malpighian tubules.

The importance of the fruitfly Drosophila as a model organism for developmental and genetic research is amply documented by several Nobel prizes awarded to researchers working with Drosophila, beginning with the award to Thomas Hunt Morgan in 1933. Less well known are the essential contributions of the little fly to the field of V-ATPase research. Soon after the discovery that a V-ATPase energizes the midgut epithelium of the tobacco hornworm (Schweikl et al.,1989; Wieczorek et al.,1989) it was shown in Drosophila hydei that a V-ATPase also plays a central role in fluid secretion in the insect kidney, the Malpighian tubules (Bertram et al.,1991). There and in subsequent studies it was confirmed that the cooperation of a V-ATPase and a cation/H⁺ exchanger is essential for the regulation of intracellular pH, the secretion of K⁺ and Na⁺ and the formation of fluid by the Malpighian tubules(Bertram et al., 1991; Bertram and Wessing, 1994; Wessing et al., 1993). Since then, candidates of the K⁺/H⁺ antiporter have been suggested in Drosophila and mosquitoes(Day et al., 2008; Rheault et al., 2007).

Not entirely unexpected, the first animal with a knockout of a V-ATPase was Drosophila melanogaster after Davies and her colleagues were able to assign the vha55 gene encoding subunit B of the V₁ complex to SzA, a lethal locus already described in 1979(Davies et al., 1996; Gausz et al., 1979). Using the LacZ reporter gene of a P-element insertion in vha55, they could show an elevated expression of subunit B exactly in tissues where a high number of V-ATPases in the plasma membrane is expected, namely in oviducts,antennal palps, rectum and Malpighian tubules. The expression of vha55 in the Malpighian tubules matched the nuclei of principal cells whereas stellate cells showed no significant expression(Davies et al., 1996; Dow et al., 1997). The observation that other V-ATPase subunits in the Malpighian tubules are also solely allocated in principal cells(Sözen et al., 1997) was in line with earlier results of intracellular measurements from Ae. aegypti indicating cation transport in the tubule epithelium as a unique property of principal cells (Beyenbach and Petzel, 1987). An overall picture of ion transport processes in the Malpighian tubules will be drawn in the paragraph on Ae. aegyti(see below).

The use of 700 P{GAL4} enhancer trap lines has revealed a high complexity and diversification of genetic and functional regions along the length of Malpighian tubules in Drosophila(Sözen et al., 1997) and opened new vistas for investigating cellular functional diversity. A big advantage of Drosophila is its sequenced genome and hence the availability of microarrays which allows a fast access to gene transcripts and the corresponding proteins in the context of defined physiological conditions. In a microarray approach that compared the transcriptome of Malpighian tubules with that of the whole adult fly, a surprisingly new set of players besides the common ones needed for fluid secretion (see Fig. 5) appeared on the scene,allowing us to draw an overall picture of what is necessary to fulfil the manifold tasks of Malpighian tubules as the insect kidney(Wang et al., 2004). Supported by these data a genome-wide survey of the V-ATPase genes was performed using in situ hybridization and reporter gene expression(Allan et al., 2005). The results revealed that 33 genes encode V-ATPase subunits, and that 13 of these 33 genes code for the plasma membrane V-ATPase. Furthermore, the clear translucent tubule phenotype in Drosophila embryos, a general feature of lethal plasma membrane V-ATPase alleles, was identified. This phenotype has been suggested to be the consequence of the failure to sufficiently acidify the lumen of the Malpighian tubules; because above a certain pH excreted urate cannot be precipitated to uric acid crystals, the tubule lumen remains translucent (Allan et al.,2005; Dow, 1999; Gausz et al., 1979). This finding is of particular medical interest as there are also human renal disorders due to mutations in kidney-specific isoforms of V-ATPase subunits;therefore the Malpighian tubules of Drosophila may serve as a model for human diseases (Dow, 2009; Karet et al., 1999; Smith et al., 2000).

In Malpighian tubules of the yellow fever mosquito, Ae. aegypti,the V-ATPase drives the transepithelial secretion of NaCl, KCl and possibly other solutes. Located at the apical membrane of principal cells(Fig. 4) it energizes not only that membrane but also the basolateral membrane and the paracellular pathway(Beyenbach, 2001).

The V-ATPase translocates protons across the apical membrane of Malpighian tubules, thereby producing a H⁺ electrochemical potential across this membrane. In considering how this proton-motive force can drive, for example, the transepithelial secretion of cations and anions, it is useful to examine separately chemical and electrical potentials.

A V-ATPase located at the apical membrane is expected to acidify the tubule lumen. However, the measured concentration of H⁺ in the lumen (pH 7.2) is lower than that in the cell (pH 7.0)(Petzel et al., 1999). Clearly, the chemical H⁺ potential difference has the wrong orientation to drive K⁺ or Na⁺ into the tubule lumen via cation/H⁺ exchange(Fig. 5).

Next to the high apical membrane voltage generated by the V-ATPase, there is another consequence to the activity of the proton pump. Because the V-ATPase translocates H⁺ across the apical membrane without exchange of a cation (or cotransport of an anion), the transport of H⁺ constitutes a current across the apical membrane that must return to the cytoplasmic face of the V-ATPase(Fig. 5). If the pump current were not allowed to return, the voltage across the apical membrane would quickly rise to the electron-motive force of the V-ATPase. In this case, the ATPase would stop pumping, because the energy of ATP hydrolysis would reach the maximum H⁺ electrochemical potential across the apical membrane. Thus, normal conditions of transepithelial secretion of NaCl and KCl require that apical membrane current (carried by H⁺) returns to the cytoplasmic face of the V-ATPase. Some of this current may be carried by H⁺ itself via the hypothetical cation exchanger/nH⁺ discussed above(Fig. 5). Current may also return through stellate cells, where current across the apical membrane is carried by Cl^– passing from cell to lumen viaCl^– channels that have been identified(O'Connor and Beyenbach,2001). The current would then presumably pass into principal cells carried by HCO₃^– diffusing from principal cells to stellate cells via gap junctions(Fig. 5) that recently have been identified in Aedes Malpighian tubules(Weng et al., 2008). A second route for the return of current is through the paracellular septate junction and then through ion channels in the basolateral membrane of principal cells. Here, current passing through the septate junction from tubule lumen to hemolymph is carried by Cl^– passing in the opposite direction(Fig. 5). From the hemolymph current reaches the cytoplasmic face of the V-ATPase viaK⁺ channels that dominate the basolateral membrane conductance of principal cells (Fig. 5)(Beyenbach and Masia, 2002). Ion channels in the basolateral membrane of stellate cells have not yet been identified.

The effects of two diuretic hormones on tubule function illustrate the importance of the intraepithelial currents generated by the V-ATPase. One hormone is the mosquito natriuretic peptide (MNP) which the laboratory of Coast identified as Anoga-DH₃₁(Coast et al., 2005). Anoga-DH₃₁ selectively increases the rates of transepithelial secretion of NaCl and water, but not of KCl. The Na⁺ selectivity stems from the activation of Na⁺ channels in the basolateral membrane of principal cells via cAMP, thereby allowing the return current to be carried by Na⁺ over K⁺(Beyenbach, 2003a; Sawyer and Beyenbach, 1985). In contrast, peptides of the kinin family of diuretic hormones trigger a sharp decrease in the paracellular resistance to produce a near transepithelial short circuit (Beyenbach,2003b; Pannabecker et al.,1993). The drop in paracellular resistance enhances the paracellular secretion of Cl^– with the effect of increasing the transepithelial secretion of both NaCl and KCl and consequently water(Yu and Beyenbach, 2001; Yu and Beyenbach, 2004).

The above analysis reveals that the primary energizer of epithelial currents and transepithelial electrolyte secretion is the electrical potential generated by the V-ATPase, at least in Malpighian tubules of the yellow fever mosquito.

The Krogh principle is central to Comparative Physiology(Krogh, 1929): `For a large number of problems there will be some animal of choice, or a few such animals, on which it can be most conveniently studied'. The converse is also true, where an unusual model system leads to fruitful avenues of investigation. The scientific history of the insect V-ATPase is a case in point, where interest in intestinal transport mechanisms in the world champion of growth, the larval Manduca sexta, led to the discovery of a V-ATPase embedded in a plasma membrane and not a vacuolar or subcellular membrane. Moreover, it was found that the V-ATPase and not the Na⁺/K⁺-ATPase energized secondary active transport not only in the plasma membrane where the proton pump is located but also at other sites in the epithelium. The natural abundance of the V-ATPase in the larval Manduca midgut facilitated its purification and opened the possibility of studying its structure, function and regulation.

Studies in Manduca have revealed a novel mechanism for regulating transport activity: the remarkable dissociation/re-association of V_O and V₁ complexes in ways that couple the availability of nutrients to the assembly and activation of the proton pump. The disassembly of the V-ATPase may reflect the endocrinology of moulting in Manduca sexta, and it may reflect the absence of glucose in yeast. Signaling pathways to and from the proton pump are largely unknown as the first examples of phosphorylation of some subunits are now being identified. But phosphorylation is just one of many mechanisms for post-translational modification of proteins. Accordingly, other mechanisms for regulating pump activity are likely to emerge. These mechanisms may also focus on the assembled, intact proton pump, i.e. on rates of ATP hydrolysis, on changes in the ATP/H⁺ transport-coupling ratio, or on the mobility of the rotor within the plasma membrane. Interactions of the V-ATPase with scaffolding and cytoskeletal proteins (that often mediate signal transduction)are likely to influence the transport activity of the holoenzyme, but these interactions have not yet been explored. Finally, the V-ATPase appears to be an ideal model for investigating the effects of voltage on an electrogenic pump.

It is puzzling why the presently known pharmaceutical inhibitors of the V-ATPase such as bafilomycin, concanamycin, archazolid and apicularen all target the V_O complex, while specific inhibitors of the V₁ complex have not yet been found. Do macrolides identify particularly vulnerable part(s) of the proton pump? If so, could endogenous agents that mimic plecomacrolides account for diseases associated with defects of the V-ATPase? Clearly, the discovery of additional inhibitors and activators of the V-ATPase will figure importantly in completing the molecular and mechanistic models of this proton pump.

The power of genetics and its post-genomic biology is unquestioned. Genetic and molecular approaches illuminate molecular processes in great detail. Thus,as we near the first decade of the 21st century, we have a decent structural model of the V-ATPase, and we know some of its functional properties. The structural complexity signals functional complexity beyond the roles of proteins functioning as mechanical devices of stator and rotor. While molecular and genetic studies will continue to reach the deepest bases for understanding physiology, in vivo experiments – as difficult as they are and as uncertain as they can be – will continue to differentiate between genetic/molecular potential and physiological reality in the living cell.