Structure and Regulation of Insect Plasma Membrane H⁺ V-ATPase

Like the blind men who described the elephant after each touched a single part, specialists tend to describe the H⁺-translocating V-ATPase (V-ATPase) from their own vantage point. The yeast geneticist sees the enzyme as a composite of subunits, each a separate gene product; ‘knocking out’ any single gene is lethal to yeast that are cultured in normal, pH neutral, medium (e.g. Nelson et al., 2000). The mammalian cell biologist sees the enzyme as an acidifier of endomembrane-bounded vacuoles and vesicles; key cysteine residues are responsible for redox regulation (e.g. Forgac, 2000). The insect physiologist sees the enzyme as an energizer of plasma membranes; the transmembrane voltage it generates drives nutrient uptake and fluid secretion, in some cases alkalizing the gut lumen (e.g. Harvey et al., 1998). Yet the biochemist and molecular biologist describe the V-ATPase as being remarkably conserved; its V₁ sector appears to consist of subunits A, B, C, D, E, F, G and H and its V_o sector to consist at least of subunits a, d and c, in organisms ranging from yeast to insects and mammals (Forgac, 1998). Throughout the last decade, we have largely been able to combine the results from such diverse vantage points to arrive at a unified view of the structure and function of the V-ATPase. In this review, we focus on the plasma membrane V-ATPase from the midgut of larval Manduca sexta, commonly called the tobacco hornworm. This V-ATPase occurs in the apical cell membranes of goblet cells, one of the two principal cell types in this epithelium, and energizes all transepithelial secondary transport processes in the midgut, including the absorption of amino acids and the regulation of the high luminal pH (Fig. 1). We focus on four aspects of the insect plasma membrane V-ATPase that are currently in the centre of our interest.

First, we will introduce the insect plasma membrane V₁ complex, emphasizing recent work on its quaternary structure. The first construction of the V₁ structural envelope by small-angle X-ray scattering and the direct visualization of the hexagonal V₁ headpiece by electron microscopy will be described as significant steps towards an understanding of the overall structure of the enzyme. The prospect that certain subunits may be encoded by more than one gene and the prospect of protein isoforms will be discussed. Next, we will introduce the V_o subunits, emphasizing the cloning of the novel M9.7 subunit and progress towards identification of the M100 subunit; cloning its cDNA brings us close to completing the primary structure of the insect V_o sector. Mechanisms for regulating V-ATPase biosynthesis and activity will be considered next. The discovery of dsRNA in caterpillar tissues that are expressing anti-sense RNA and the presence of ecdysterone-responsive elements in V-ATPase genes paves the way for molecular analysis of the hormonally regulated development of the caterpillars. Finally, the disassembly of the V₁ from the V_o sector, when feeding ceases during moulting or upon starvation, and its reassembly in the absence of protein synthesis as feeding resumes will be considered together with other putative factors controlling enzyme activity and ion transport.

Two major advances towards an elucidation of the quaternary structure of V-ATPases have been made during the past 2 years. First, a model-independent approach, which is based upon the multipole expansion method using spherical harmonics, has been developed to study the three-dimensional structure of macromolecules from solution X-ray scattering (Stuhrmann, 1970; Svergun et al., 1998). Application of this approach has led to a low-resolution (3.6 nm) structure of the V₁ ATPase from M. sexta (Svergun et al., 1998). Second, image processing of electron micrographs of negatively stained V-ATPase from Chlostridium fervidus (Boekema et al., 1997, 1998) and V₁ ATPase from M. sexta (Radermacher et al., 1999) has yielded two-dimensional structures at a resolution of 1.8 nm and 2.4 nm, respectively. The V₁ complex is asymmetric, with a headpiece approximately 14.5 nm in diameter and a stalk with a length of approximately 11 nm (Svergun et al., 1998) that accounts for linking catalytic site events in the headpiece with ion pumping through the V_o portion (Fig. 2). When comparing the independently identified structures (Boekema et al., 1998; Svergun et al., 1998; Radermacher et al., 1999), a pseudohexagonal arrangement of six masses can be visualized in the headpiece. These masses may be interpreted as three copies each of the major subunits A and B arranged in an alternating manner. A stalk, which extends perpendicular to the centre of the hexameric unit, was visualized from the X-ray data (Svergun et al., 1998). It was also observed by electron microscopy as a seventh density either centrally or asymmetrically to the hexamer (Fig. 3; Radermacher et al., 1999). The asymmetrical placement of this density may be caused by small tilts of the V₁ ATPase perpendicular to its hexagonal projection. Since a preferred orientation of the V₁ ATPase from M. sexta is preserved, it is now feasible to compute a three-dimensional reconstruction from electron micrographs of these images in order to visualize functionally important features of the V₁ complex.

To obtain further insight into the quaternary structure, high-resolution studies based on X-ray analysis of three-dimensional crystals of the V₁ complex will be necessary. The V₁ ATPase of M. sexta midgut is obtainable in purity and amount that are sufficient for this purpose. A second precondition is the complete amino acid sequence of all eight V₁ subunits. Sequences of subunits A, B, E, F and G have been obtained by cDNA cloning (see Wieczorek et al., 1999b). Southern and/or northern blot analysis favours the assumption that subunits A, E, F and G may be encoded by single genes expressing single protein isoforms, at least in the midgut. However, in western blots using monospecific antibodies against different regions of the B subunit, we have detected differential immunoreactions, indicating that subunit B occurs in two isoforms (Huss et al., 1999); we are presently clarifying which isoform belongs to the cytosolic V₁ complex derived from the plasma membrane. Primary sequences of subunits C, D and H have recently been obtained by cloning cDNAs from our M. sexta midgut library (H. Merzendorfer, B. Jacobmeier, X.-F. Zhao and H. Wieczorek, unpublished results).

Information regarding the primary structures of these subunits appears to be keeping pace with the development of V₁ crystals capable of diffracting X-rays.

The structural organization of the membrane-embedded V_o complex of V-ATPases is far less well understood than that of the V₁ complex. In M. sexta, the V_o complex, as isolated from the V_o-enriched midgut goblet cell apical membranes of starving fifth-instar larvae (Merzendorfer et al., 1999a), appears to consist of at least the four subunits M100, M40, M17 and M9.7 (Fig. 2) for which primary structures have been derived from cDNA sequencing (Wieczorek et al., 1999b; and see below). Subunits M40 and M17 fit into the canon of sequences obtained from other V-ATPase sources, whereas subunits M9.7 and M100 deserve special attention.

The novel subunit M9.7 was detected in the M. sexta midgut quite recently and identified as a constitutive part of both the mature V₁V_o holoenzyme and of the mature V_o complex (Merzendorfer et al., 1999b). On the basis of N-terminal amino acid sequencing, cDNA cloning and sequencing led to an open reading frame encoding for a protein of 9.7 kDa. The apparent molecular mass of 20 kDa that was observed by sodium dodecylsulphate (SDS)–polyacrylamide gel electrophoresis was shown to be due to extensive glycosylation of the subunit. The sequence showed appreciable amino acid identity to the novel 9.2 kDa membrane-associated protein from bovine chromaffin granules (Ludwig et al., 1998). Smaller but significant similarities to the yeast protein vma21p became evident, too (Hill and Stevens, 1994). However, H_vma21p is not part of the mature yeast V-ATPase, being confined to the endoplasmic reticulum, where it is involved in assembly processes during biosynthesis of the V-ATPase holoenzyme. Partially purified endoplasmic reticulum membranes from M. sexta midgut, unlike those from yeast, contained only small amounts of M9.7. In contrast, highly purified goblet cell apical membranes contained M9.7 in stoichiometric amounts with respect to the other V-ATPase subunits. These results demonstrate that insect M9.7 (and probably bovine M9.2 as well) is a new member of the mature V-ATPase and that it resides in the V_o complex. We term it subunit e.

For years, we have not been able to identify subunit M100 unequivocally in the plasma membrane V-ATPase of M. sexta. Since M100 appears to be very sensitive to proteolysis (Adachi et al., 1990), its absence in the holoenzyme isolated from the midgut of M. sexta could, in principle, be due to degradation during the purification procedure. However, holoenzyme purification in the presence of a protease inhibitor cocktail (0.07 µmol l⁻¹ pepstatin A, 8 µmol l⁻¹ antipain, 0.7 µmol l⁻¹ trypsin inhibitor, 0.1 mmol l⁻¹ benzethonium chloride, 1 mmol l⁻¹ benzamidine, 2 mmol l⁻¹ iodoacetamide, 10 µmol l⁻¹ leupeptin and 5 mmol l⁻¹ EDTA) provided no hint of proteolysis (U. Klein, A. Lepier and H. Wieczorek, unpublished results). Recently, by making a slight change in the purification protocol, we have been able to obtain purified holoenzyme or V_o complex containing subunit M100 (Merzendorfer et al., 1999a); its identity was confirmed by immunostaining with polyclonal antibodies to the 115 kDa V-ATPase subunit of bovine chromaffin granules (kindly provided by D. Apps, Edinburgh, UK). Even more recently, we have partially cloned and sequenced cDNA encoding the subunit. We used a probe derived from a clone encoding the putative Drosophila melanogaster M100 protein (vha100-2, kindly provided by J. A. T. Dow, Glasgow, UK); the M. sexta cDNA shows significant similarity to isoform a1 of the 116 kDa V-ATPase subunit of bovine clathrin-coated vesicles. Northern blot analysis, using mRNA from M. sexta midgut and an RNA probe derived from in vitro transcription of the M100 cDNA, demonstrated a single labelled band (H. Merzendorfer, S. Reineke, A. Wülfing and H. Wieczorek, unpublished results). Taken together, our results suggest that subunit M100 is a constituent part of the tobacco hornworm V-ATPase.

The performance of a physiological task that involves plasma membrane V-ATPase activity may be modulated at various levels of very different complexity, from protein biosynthesis to regulatory modifications of any of the V-ATPase subunits (see Fig. 4). Besides, any manoeuvre relaxing the proton-motive force built up by a V-ATPase will in turn influence the net V-ATPase performance, resulting in very variable functions such as shifts in compartmental pH, powering of secondary cation/anion transepithelial transport, transmitter pile-up, removal of heavy metal ions or other concerted mechanisms that depend on V-ATPase performance. Here, we shall focus on the insect plasma membrane V-ATPase and hitherto known modes of regulation, but we will also mention other examples. Specifically, we address epithelial transport energization to highlight those roles of the insect V-ATPase that differ from more conventional functions, such as pH regulation or transmitter storage.

Little is known about the control of the biosynthesis of V-ATPase subunits. In many tissues or cells, such as in Neurospora crassa, V-ATPases are thought to act as housekeeping enzymes in the endomembrane system, with no apparent control of their expression. In accord with this view, promoters of several V-ATPase genes isolated from N. crassa are similar to those of known mammalian housekeeping genes (Wechser and Bowman, 1995). However, V-ATPase biosynthesis appears to be strictly regulated in cells in which V-ATPases play special roles in addition to their ubiquitous acidification of intracellular organelles. Thus, in the slime mould Dictyostelium discoideum, the gene encoding V-ATPase subunit B may be controlled during development at the level of transcription, since the respective amounts of mRNA are increased during growth and decreased during starvation (Bracco et al., 1997). Transcriptional regulation of V-ATPase genes has been demonstrated in more detail for the mammalian haematopoetic system; during monocyte to macrophage differentiation, the transcription of the gene encoding isoform 2 of subunit B appears to be up-regulated by a cyclic-AMP-dependent signal transduction pathway (Lee et al., 1995, 1997).

In goblet cell apical membranes of the tobacco hornworm, V-ATPase activity is down-regulated during moulting and starvation by the reversible dissociation of the V₁V_o holoenzyme (Sumner et al., 1995, Gräf et al., 1996; see below). Northern blots revealed that inactivation of enzyme activity is paralleled by decreased transcript levels for every V-ATPase subunit for which we had identified a respective cDNA in our M. sexta larval midgut library (S. Reineke, H. Merzendorfer and H. Wieczorek, unpublished results). Thus, coordinated reduction of transcript amounts may reflect a transcriptional control mechanism that minimizes V-ATPase biosynthesis during periods of enzyme inactivation and therefore helps to save cellular energy. Since haemolymph titres of ecdysteroids reach high levels during moulting (Bollenbacher et al., 1981), the observed decline in transcript levels for V-ATPase subunits may be mediated by an ecdysteroid-responsive element in the promoter of the corresponding genes. This suggestion is supported by the analysis of three 5′ regions of the genes mvaB, mvaG and mvaM40 encoding the M. sexta V-ATPase subunits B, G and M40, respectively (Merzendorfer, 1998). Sequence analysis revealed a consensus region for a putative ecdysterone-responsive element (EcRE; Luo et al., 1991), which is present in all three promoters investigated. The presumed regulatory function of ecdysterone was additionally corroborated by reporter gene assays performed with a series of plasmids containing different 5′ regions of mvB (Merzendorfer, 1998). These experiments demonstrated an inhibitory effect of ecdysterone on the transcriptional activity of the mvB promoter.

Comparison of the overall promoter structures revealed striking similarities between mvB and mvG (Merzendorfer, 1998). They exhibited a similar GC distribution pattern and, in addition, possessed canonical TATA boxes and cyclic-AMP-responsive elements (CREs). Both elements appear to be functional units of the mvB promoter, because reporter assays showed that the TATA box is necessary for basal activity and that transcription is activated by 8-chlorophenylthiocyclic AMP, indicating a protein kinase A (PKA)-dependent signal transduction pathway. In contrast to the 5′ regions of the V₁ genes, the promoter of the gene encoding the V_o subunit M40 exhibited a different GC distribution and lacked apparent TATA boxes and CREs. Overall, this promoter resembled housekeeping genes known from other organisms. Thus, differences in promoter structures may indicate that gene expression of V₁ and V_o subunits is regulated, at least partly, in distinct ways.

The finding of antisense RNA in the larval midgut that is complementary to the spliced transcript of the V_o subunit M40 (Merzendorfer et al., 1997b) may indicate a novel regulatory mechanism of protein biosynthesis. The function of endogenous antisense transcripts in the tobacco hornworm midgut, as in most other organs of eukaryotes in which they have been found, is largely unknown (Vanhee-Brossollet and Vaquero, 1998). However, earlier results from immunocytochemistry and in situ hybridization may throw some light on the reasons for the expression of antisense RNA. The apical membranes of goblet cells are the only ones in which large amounts of V-ATPase protein are found, whereas other plasma membranes, as well as endomembranes in both goblet and columnar cells, contain little or no enzyme (Klein et al., 1991). In contrast, in situ hybridization using midgut cryosections clearly demonstrated high transcript levels of RNA encoding subunits A and M17 in the columnar cells (Jäger et al., 1996). In addition to translational control mechanisms mediated by protein/RNA interactions via the 5′ or 3′ untranslated region, this discrepancy may also be explained by the action of endogenously expressed antisense RNA, as has been suggested for the lin-4/lin-14 transcripts of Caenorhabditis elegans (Lee et al., 1993). However, expression of antisense RNA may also affect the stability of the corresponding sense transcript by RNA/RNA hybrid formation and its subsequent degradation by cellular dsRNAases (Nellen and Lichtenstein, 1993). Since we have detected a significant fraction of dsRNA in larval midgut cells using nuclease protection assays (Merzendorfer, 1998), this mechanism may also play a role in the midgut of larval M. sexta.

Massive solute and/or water fluxes occur in the intestine, urinary system and glandular tissues of insects. There, specialized cell types, e.g. goblet cells in the midgut or principal cells in Malpighian tubules, provide a specific environment for optimal plasma membrane V-ATPase function. In contrast to several vertebrate urinary epithelia (see Wieczorek et al., 1999a) which, when stimulated, shuttle V-ATPase via exo/endocytotic pathways between vesicular compartments and the plasma membrane, the insect plasma membrane V-ATPase seems to be a ‘constitutive’ membrane enzyme. In a certain sense, the goblet cell cavity is like a vacuole: it first appears during the moult in differentiating stem cells by a unique, but irreversible, fusion of a large endosome with the apical membrane (Baldwin and Hakim, 1991). What we have here is a shuttling of V-ATPase which is ‘frozen’ in the exocytotic state.

V-ATPase holoenzyme in midgut plasma membrane responds to energy demand by reversible shuttling of V₁ complexes that dissociate and reassociate with the membrane V_o complex, rather than by a reversible membrane shuttle. During moulting or starvation, the amount of V-ATPase holoenzyme in the goblet cell apical membrane of the tobacco hornworm midgut decreases while the concentration of cytosolic V₁ complexes increases from less than 1 % of the total cytosolic protein to the unexpectedly high level of nearly 2 % (Sumner et al., 1995; Gräf et al., 1996). Parallel findings in yeast, where a rapid dissociation of V₁ complexes from V_o complexes occurred upon deprivation of glucose (Kane, 1995), indicated that ‘dissociation/reassociation’ may be a general regulatory mechanism for V-ATPases (see Kane and Parra, 2000). Although not shown directly, a decrease in glucose level may also be an important factor in the tobacco hornworm, since haemolymph glucose levels fall to one-third of the original concentration within 30 min after the caterpillars stop feeding and to less than 5 % within the first 10 h of starvation (Gies et al., 1988). Cytosolic V₁ complexes lose their ability to hydrolyze ATP in the presence of Mg²⁺ (Gräf et al., 1996; Kane and Parra, 2000) but continue hydrolyzing ATP in the presence of millimolar levels of Ca²⁺; therefore, under physiological conditions, the complexes appear to be silent, enabling the cell to save energy. When glucose level is restored in yeast or when feeding resumes in the tobacco hornworm, the V₁V_o holoenzyme is reassembled (Kane, 1995; Sumner et al., 1995; Gräf et al., 1996). As in yeast (Parra and Kane, 1996, 1998), the reduction in cytosolic pH that follows the resumption of metabolism in the tobacco hornworm could be a factor in V₁V_o assembly.

Whether specific signalling molecules induce V₁V_o disassembly and whether reassembly is dependent on a reduction in pH, alone or in combination with signalling molecules, is not known for insects but may be the outcome of other, short-term signalling second-messenger cascades (see below). In tissues such as insect midgut or Malpighian tubules, V₁V_o disassembly will immediately curtail all events that are secondarily powered by the V-ATPase-dependent proton-motive force. These events include active electrogenic K⁺ transport, luminal alkalization by the midgut and the powerful fluid secretion mechanisms in insect Malpighian tubules. In contrast, up-regulation of assembled V-ATPases may not only restore the pre-disassembly conditions, but may also lead to a new setpoint, with less disassembled and more assembled V-ATPases. For instance, one could imagine that the dramatic increase in urinary fluid secretion rates as Malpighian tubules respond to hormonal signals (Dow et al., 1998) may be due, at least partly, to such a process.

As in other V-ATPases, enzyme activity in caterpillar midgut could also be regulated by the oxidation of cysteine via disulphide bond formation (Merzendorfer et al., 1997a; Stevens and Forgac, 1997). The V-ATPases of clathrin-coated vesicles or yeast vacuoles are maintained in a partially inactive state by cysteine oxidation in the vicinity of the ATP-binding site of subunit A. The formation of such disulphide bonds in an oxidizing environment could mediate short-or long-term deactivation of the V-ATPase. Significantly, mitochondria-rich epithelial cells are known for their enormous density of plasma membrane V-ATPase (Brown and Breton, 1996), which may be maintained in a fully stimulated state since abundant mitochondria deplete oxygen locally and thus establish optimal reducing conditions for full V-ATPase activity (Harvey and Wieczorek, 1997). Of course, even in well-oxygenated regions, cytosolic molecules might specifically attack susceptible SH groups or disulphide bonds and therefore induce interconversion of the V-ATPase between the inactive and active states. S-nitrosogluthathione, a reagent generating nitric oxide, has been shown to inhibit bovine clathrin-coated vesicle V-ATPase through disulphide bond formation between cysteine residues at the catalytic site (Forgac, 1999; but see below for nitric oxide as a messenger for cyclic GMP generation resulting in V-ATPase stimulation). Moreover, it has been proposed that the block of V-ATPase by nitrate may, in fact, be due to oxidation of the critical cysteine residue in subunit A (Dschida and Bowman, 1995).

V-ATPases are highly electrogenic, establishing a proton-motive force consisting of transmembrane voltages that exceed 240 mV or transmembrane pH gradients that exceed 4 pH units (W. R. Harvey, 1992). In insects, both extremes are realized, a high voltage across the apical membrane of goblet cells in the midgut (Dow, 1992) and, conversely, highly variable pH gradients across the apical membrane of principal cells in Malpighian tubules (Bertram and Wessing, 1994). These differences must, of course, be a consequence of the transporters and channels of cells in different tissues responding to the electrical or chemical component of the proton-motive force to different degrees and in different ways. Partial electrical short-circuiting of the V-ATPase by the parallel operation of electrogenic K⁺/2H⁺ antiport has been observed in vesicles derived from the midgut goblet cell apical membrane (Wieczorek et al., 1989; Azuma et al., 1995). However, the default condition seems to be a short-circuit of the V-ATPase voltage by an anion conductance that permits Cl⁻, for example, to leave the cell and enter the apical medium. Such short-circuiting has been found in isolated goblet cell apical membrane vesicles maintained in high Cl⁻ concentrations (Wieczorek et al., 1989) but not in the cells in situ (Zeiske, 1992), where the Cl⁻ concentration is low (Dow et al., 1984). The implication is that a Cl⁻-activated Cl⁻ conductance is present in the goblet cell apical membrane. A large, lumen-alkaline, pH gradient does develop across the adjacent columnar cell apical membrane, but it opposes (and balances) the V-ATPase-generated voltage so that the net proton-motive force across that membrane is approximately zero (Fig. 1). In Malpighian tubules, it appears that K⁺/H⁺ antiport is electroneutral, and the V-ATPase-generated voltage is short-circuited by apical anion pathways (O’Donnell et al., 1996).

5-Hydroxytryptamine has long been known to stimulate changes in cell Ca²⁺ and cyclic AMP levels in insect salivary glands, resulting in increased salivation (Prince and Berridge, 1972). Clearer evidence has been obtained for the stimulation of V-ATPase by cyclic AMP or cyclic GMP in Malpighian tubules of Drosophila melanogaster by Dow and Maddrell and in moth ovarian follicles by Telfer and associates (for a review, see Harvey et al., 1998). In contrast, Ca²⁺ acting as a second messenger seems to open anion channels (O’Donnell et al., 1996). Phosphorylation by cyclic-nucleotide-dependent protein kinases A or G has so far not been detected in V-ATPases. Nevertheless, the evidence obtained from experiments with Drosophila melanogaster Malpighian tubules is promising: in response to diuretic hormones, cyclic AMP appears to stimulate ion transport via the V-ATPase, whereas a cardioaccelerating hormone generates cyclic GMP via a G-protein/nitric oxide reaction chain (O’Donnell et al., 1996).

Epithelial cells in which the plasma membrane V-ATPase is driving ion transport generally exhibit densely packed proton pumps in quasi-crystalline, two-dimensional arrays in their plasma membranes (Brown and Breton, 1996). If, as in the midgut, the apical proton pump is deeply buried in the goblet cavity, direct access (e.g. by a patch electrode) will be impossible. However, if the V-ATPase operates as a current source with randomly chopped conductance pulses, noise analysis of the transepithelial current, such as that performed on goblet cell ion channels (Zeiske et al., 1986), would seem to be the method of choice. The feasibility of this approach for monitoring electrogenic proton pump activity was first demonstrated by B. J. Harvey (1992) in mitochondria-rich cells of frog skin: strictly defined conditions enabled parameters such as single-pump current and pump area density to be analysed.

We currently use this method to study V-ATPase regulation in the insect midgut in situ. This requires that competing ionic current sources should be silenced by the appropriate selection of solutes and/or pharmacological inhibitors. The midgut is mounted in an Ussing chamber and, after basolateral permeabilization using amphotericin B, the resulting pseudo-mono (apical) membrane preparation also allows appropriate adjustment to the now accessible cytosolic medium (Schirmanns and Zeiske, 1994a,b). Under conditions in which the V-ATPase is the sole active current source, noise analysis should allow microscopic pump parameters to be measured as a function of various elements of second-messenger cascades and the other manoeuvres described above that modify proton pump activity.

This work was supported by the Deutsche Forschungsgemeinschaft (Wi 698 and SFB 431) and the National Institutes of Health Grant AI 22444.