Purification and properties of a cytosolic V1-ATPase.

The native V1 complex of the tobacco hornworm vacuolar type ATPase (V-ATPase) was purified from cytosolic extracts of molting larval midgut. It consisted of the established V-ATPase subunits A, B, and E along with the 14-kDa subunit F and the novel 13-kDa subunit G. The final amount of purified V1 complex made up an unexpectedly high 2% of the total cytosolic protein, with a yield of ∼0.4 mg/g of tissue. An equally high amount of cytosolic V1 complex was obtained from starving intermolt larvae. By contrast, the cytosolic V1 pool was reduced drastically in feeding intermolt larvae or in larvae that had been refed after starvation. The activity of the membrane-bound V-ATPase holoenzyme was inversely related to the size of the cytosolic V1 pool, suggesting that the insect plasma membrane V-ATPase is regulated by reversible disassembly of the V1 complex as a function of the feeding condition of the larvae. Like F1-ATPases, the purified V1 complex exhibited Ca2+-dependent ATPase activity and, in the presence of 25% methanol, exhibited Mg2+-dependent ATPase activity. Therefore, we designate the native V1 complex, V1-ATPase. Both enzyme activities were completely inhibited by micromolar N-ethylmaleimide. In contrast to the Ca2+-dependent V1-ATPase activity, the Mg2+/methanol-dependent V1-ATPase activity did not decrease with the incubation time and thus was not inhibited by ADP. Methanol appears to induce a conformational change of the V1 complex, leading to enzymatic properties of the V1-ATPase that are similar to those of the membrane-bound V-ATPase holoenzyme. This is the first time that a native and enzymatically active V1 complex has been purified from the cytosol.

The native V 1 complex of the tobacco hornworm vacuolar type ATPase (V-ATPase) was purified from cytosolic extracts of molting larval midgut. It consisted of the established V-ATPase subunits A, B, and E along with the 14-kDa subunit F and the novel 13-kDa subunit G. The final amount of purified V 1 complex made up an unexpectedly high 2% of the total cytosolic protein, with a yield of ϳ0.4 mg/g of tissue. An equally high amount of cytosolic V 1 complex was obtained from starving intermolt larvae. By contrast, the cytosolic V 1 pool was reduced drastically in feeding intermolt larvae or in larvae that had been refed after starvation. The activity of the membrane-bound V-ATPase holoenzyme was inversely related to the size of the cytosolic V 1 pool, suggesting that the insect plasma membrane V-ATPase is regulated by reversible disassembly of the V 1 complex as a function of the feeding condition of the larvae. Like F 1 -ATPases, the purified V 1 complex exhibited Ca 2؉ -dependent ATPase activity and, in the presence of 25% methanol, exhibited Mg 2؉ -dependent ATPase activity. Therefore, we designate the native V 1 complex, V 1 -ATPase. Both enzyme activities were completely inhibited by micromolar N-ethylmaleimide. In contrast to the Ca 2؉ -dependent V 1 -ATPase activity, the Mg 2؉ /methanoldependent V 1 -ATPase activity did not decrease with the incubation time and thus was not inhibited by ADP. Methanol appears to induce a conformational change of the V 1 complex, leading to enzymatic properties of the V 1 -ATPase that are similar to those of the membranebound V-ATPase holoenzyme. This is the first time that a native and enzymatically active V 1 complex has been purified from the cytosol.
Vacuolar type ATPases (V-ATPases) 1 are proton-translocating enzymes that occur in endomembranes of all eukaryotes and in plasma membranes of many eukaryotes (for review, see Ref. 1). The evolution of V-ATPases appears to be related to that of F-ATPases, the F 1 F o -ATP synthases of mitochondrial, chloroplast, and eubacterial phosphorylating membranes (2). Like their sister F-ATPases, they are multisubunit heteromeric proteins composed of two structural domains, a membranespanning complex and a peripheral catalytic complex. The latter forms ball and stalk structures, portasomes, that in electron micrographs look remarkably similar to the corresponding structures of F-ATPases (3). As an analogy to F-ATPases, the peripheral domain of V-ATPases has been designated the V 1 complex and the membrane-spanning domain has been designated the V o complex (4). Like its F 1 counterpart, the V 1 complex consists of several subunits. At least three V 1 subunits are common to all V-ATPases, subunits A and B that are homologous to the F 1 subunits ␤ and ␣, respectively, and subunit E. Like the corresponding F 1 subunits, subunits A and B occur in three copies/complex, whereas subunit E occurs in only one copy (5). cDNAs encoding subunits A, B, and E have been cloned and sequenced from various organisms. Two further V 1 subunits have been designated C (6) and D (7), but they may not occur in every V-ATPase. Last but not least, two V-ATPase subunits that appear to be members of the insect V 1 complex were cloned from the tobacco hornworm, Manduca sexta, and were designated subunits F (8) and G (9). Since these subunits were also cloned from the evolutionarily distant yeast (10 -12), they appear to be general V-ATPase subunits.
The peripheral complex of F-ATPases can be isolated from the holoenzyme and is called F 1 -ATPase because of its catalytic activity. It has been investigated thoroughly, culminating in the resolution of its atomic structure (13). By contrast, many structural and functional properties of the catalytic V 1 complex are unknown. The most extensively investigated V 1 complex is the V c complex of the clathrin-coated vesicle V-ATPase. It was obtained by in vitro association of V-ATPase subcomplexes purified by ammonium sulfate precipitation and density gradient centrifugation after treatment of the V-ATPase with urea (14) or by the addition of recombinant V 1 subunits to these subcomplexes (15)(16)(17)(18). V 1 complexes from other V-ATPase preparations have been obtained by treatment of the membrane-bound V-ATPase with chaotropic salts followed by reassociation of the dissociated V 1 subunits by dialysis (9,19); however, this procedure yields low amounts of V 1 complexes.
In contrast to F 1 complexes that are always associated with membranes, V 1 complexes also occur in a soluble form in the cytosol as was shown for yeast and for a bovine kidney epithelial cell line (20,21). These cytosolic V 1 complexes were detected by immunoprecipitation, but they were not quantified or purified in greater amounts. We expected to detect considerable quantities of cytosolic V 1 complexes in the midgut of molting tobacco hornworms since Sumner et al. (22) had demonstrated that the insect plasma membrane V-ATPase is downregulated during the molt by detachment of V 1 subunits from the membrane. The tobacco hornworm V-ATPase may serve as a valuable model for V-ATPases in general since it is well characterized (23) and all five of its known V 1 subunits have been cloned and sequenced (8, 9, 24 -26). In this paper, we show that the midgut cytosolic extract from molting tobacco hornworms is a rich source for the purification of enzymatically active V 1 complexes. We present evidence that the size of the cytosolic V 1 pool depends on the feeding condition of the M. sexta larvae.

EXPERIMENTAL PROCEDURES
Insects-Larvae of M. sexta (Lepidoptera, Sphingidae) were reared under long day conditions (16 h of light) at 27°C on a synthetic diet modified according to Bell and Joachim (27). Experiments were carried out on larvae that were molting from the fourth to fifth instar (molting larvae) and on fifth instar larvae after the molt (intermolt larvae). Molting larvae were in stages E or F (approximately 20 h after the entry into the molt (22)) and weighed about 1.5 g, whereas intermolt larvae weighed about 5 g.
Purification of the Cytosolic V 1 Complex-Whole midguts were dissected from larvae that had been kept on ice for about 15 min. Approximately 1 g of midgut tissue (wet weight) was homogenized in 5 ml of an ice-cold buffer consisting of 300 mM mannitol, 5 mM Na-EDTA, 50 mM NaCl, and 17 mM Tris-HCl (pH 7.5) using an Ultraturrax homogenizer (IKA, Germany) at 20,500 rpm for 60 s. After centrifugation at 14,500 ϫ g for 4 min at 4°C, the supernatant was supplemented with Pefabloc SC (BIOMOL) to a final concentration of 5 mM and was centrifuged again at 186,000 ϫ g for 30 min at 4°C. The supernatant was filtrated through a 0.2-m filter using a 5-ml syringe. The volume was adjusted to 5 ml, and an equal volume of ice-cold saturated ammonium sulfate was slowly added while stirring on ice. The precipitation of cytosolic proteins was allowed for 15 min on ice. After centrifugation at 14,500 ϫ g for 5 min at 4°C, the precipitate was dissolved in 1 ml of a buffer consisting of 16 mM Tris-HCl, 0.32 mM EDTA, and 50 mM NaCl (pH 8.1). Aliquots of 0.5 ml each were layered onto a discontinuous sucrose density gradient (2 ml of 40%, 1 ml of 30%, 1 ml of 20%, and 0.8 ml of 10% sucrose (w/v) dissolved in the same buffer that now contained 9.6 mM 2-mercaptoethanol) and centrifuged in a vertical rotor at 220,000 ϫ g for 1 h at 4°C. The 20% sucrose fraction containing the V 1 complex was chromatographed on a Mono Q anion exchange column (Pharmacia Biotech Inc.) using an elution buffer composed of 0.05-0.4 M NaCl (linear gradient) and 20 mM Tris-HCl (pH 8.1). Fast protein liquid chromatography (FPLC) was performed for 30 min at a flow rate of 1 ml/min. The V 1 complex, represented by two peaks, was collected in two fractions (about 0.26 and 0.3 M NaCl). Both fractions were concentrated to final volumes of approximately 120 l using Centricon 100 microconcentrators, and 100 l of each were loaded onto a Superdex 200 HR 10/30 gel chromatography column (Pharmacia) for further purification. FPLC was performed at a flow rate of 0.5 ml/min, using a running buffer composed of 150 mM NaCl and 20 mM Tris-HCl (pH 8.1). The V 1 complex was found in the fraction containing proteins of approximately 450 kDa (using ferritin as standard).
ATPase Activity-Assays were performed at 30°C and had a final volume of 80 l. Cytosolic V 1 -ATPase activity was measured in the presence of 50 mM Tris-MOPS (pH 8.1), 20 mM KCl, 4 -10 mM NaCl, 3 mM 2-mercaptoethanol, 1 mM ATP, and 3 mM CaCl 2 or 1 mM MgCl 2 and 25% methanol instead of CaCl 2 . The protein content was 3-5 g in assays of Ca 2ϩ -dependent ATPase activity and 0.5-1 g in assays of Mg 2ϩ /methanol-dependent ATPase activity. For assays of the membrane-bound holoenzyme activity, partially purified goblet cell apical membranes were used (28). Incubation mixtures had a protein content of 10 g and consisted of 50 mM Tris-MOPS (pH 8.1), 20 mM KCl, 1 mM ATP, 1 mM MgCl 2 , 0.1 mM vanadate, 0.5 mM azide, and 3 mM 2-mercaptoethanol. All further conditions, including the determination of inorganic phosphate, were as described previously (29).
Other Methods-Isolation and purification of the V-ATPase from M. sexta midgut goblet cell apical membranes, protein determination with Amido Black, standard SDS-polyacrylamide gel electrophoresis, Western blotting on nitrocellulose membranes (BA85), and immunostaining were performed as described previously (8, 28 -30). (22) recently demonstrated that V 1 subunits were removed from the membrane during molt. Although it was not clear whether the V 1 complex remained complete or whether the V 1 subunits disintegrated, there was some evidence for the first alternative since cytosolic V 1 pools had already been described in yeast and in a kidney cell line (20,21). Therefore, we chose the midgut cytosol of molting M. sexta larvae as a putative source for the purification of the V 1 complex and established an isolation protocol. The result of each purification step was monitored by SDS-gel electrophoresis, and the fractions containing the main part of the V 1 subunits were used for the next purification step. In brief, proteins were precipitated with ammonium sulfate, and the precipitate was dissolved and size-fractionated by density gradient centrifugation on a sucrose step gradient. The 20% sucrose fraction was collected, and proteins were separated by FPLC on a Mono Q ion exchange column applying a linear NaCl gradient from 0.05 to 0.4 M. V 1 subunits were found in two fractions containing distinct peaks at approximately 0.26 and 0.3 M NaCl, respectively ( Fig. 1). The proteins in both fractions were subjected separately to further size fractionation by gel chromatography. In both cases, protein exhibiting an apparent molecular mass of ϳ450 kDa could be detected in a sharp peak (Fig. 1, inset). The amount of protein in the peak obtained from ion exchange fraction 2 (peak 2) was 1.5-5 times higher than that from fraction 1 (peak 1). After SDS-gel electrophoresis of both peaks, comparison of their protein band patterns with that of the purified M. sexta V-ATPase holoenzyme (28), as well as staining of Western blots with an antiserum to the V-ATPase holoenzyme, revealed that both peaks contained the V 1 complex (Fig. 2). There were almost no proteins except the V 1 subunits, and no major difference could be detected between the peaks. In both cases, the V 1 complex was composed of the established subunits A, B, and E along with the 14-kDa subunit F (8) and the novel 13-kDa subunit G (9). The amount of V 1 complex obtained from the cytosol was unexpectedly high. The final amounts of purified V 1 complex from both fractions made up 1.64 Ϯ 0.23% of the total cytosolic protein (mean Ϯ S.D., five independent preparations). Thus, approximately 0.4 mg of V 1 complex could be purified from 1 g of midgut tissue within about 7 h (Fig. 5). This is the first time that a native V 1 complex has been purified from the cytosol in such high quantity.

Purification of the Cytosolic V 1 Complex from Molting Larvae-Sumner et al.
Ca 2ϩ -dependent Enzyme Activity of the Cytosolic V 1 Complex-In the presence of 3 mM Ca 2ϩ , the M. sexta cytosolic V 1 complex from peak 2 hydrolyzed 2.4 Ϯ 0.2 mol of ATP/mg of protein/5 min (mean Ϯ S.D., five independent preparations), whereas the activity of the peak 1 complex was more than 3 times lower (0.7 Ϯ 0.2 mol of ATP/mg of protein/5 min (mean Ϯ S.D., five independent preparations)). Since the V 1 complexes from both peaks had the same subunit composition, we concluded that peak 1 contained a partially inactivated form of the V 1 complex. Therefore, the following experiments were performed with the V 1 complex from peak 2.
Ca 2ϩ -dependent ATPase activity decreased with increasing incubation time; and after 5 min, the activity had dropped to approximately 35% of the 1-min value, indicating increasing product inhibition (not shown). The use of the ATP regenerating system employing pyruvate kinase was impossible since pyruvate kinase itself is dependent on Mg 2ϩ and since even 0.1 mM Mg 2ϩ inhibited the Ca 2ϩ -dependent ATPase activity of the V 1 complex almost completely. Therefore, we worked in most cases at two different incubation times (2 and 5 min) and verified identical enzyme properties at both times. 0.1 mM vanadate, 0.5 M azide, or 1 M folimycin did not affect the ATPase activity (not shown), but N-ethylmaleimide did inhibit it with an IC 50 of approximately 0.1 M (Fig. 3). ATP was the preferred substrate of the V 1 enzyme, followed by GTP (Table  I). Unlike ATPase activity, GTPase activity was not inhibited with increasing incubation times. UTP and CTP were not effective substrates (Table I), and ADP or p-nitrophenyl phosphate was not hydrolyzed (not shown). Neither ATPase nor GTPase activity of the V 1 complex was detected when Ca 2ϩ was replaced by Mg 2ϩ . Both 30 and 0.3 mM Ca 2ϩ led to a significant decrease in ATPase activity, and at 30 M Ca 2ϩ , almost no activity could be measured. Thus it is very unlikely that the V 1 complex is a functional Ca 2ϩ -dependent ATPase in vivo. On the other hand, its Ca 2ϩ -dependent ATPase activity proves that the purified V 1 complex is a functional enzyme. Therefore, we suggest the term V 1 -ATPase for the native V 1 complex.
Mg 2ϩ -dependent Enzyme Activity of the Cytosolic V 1 -ATPase in the Presence of Methanol-Organic solvents, especially methanol, are known to be effective enhancers of Mg 2ϩ -dependent enzyme activity of F 1 -ATPases from bacteria, chloroplasts, and mitochondria (31)(32)(33)(34). Ca 2ϩ -dependent enzyme activity of the M. sexta cytosolic V 1 -ATPase is strongly reminiscent of the characteristics of the purified F 1 -ATPase from Bacillus firmus (31) and of the CF 1 -ATPase from spinach chloroplasts (35). Consequently, the effect of methanol on the activity of the M. sexta V 1 -ATPase was investigated. The presence of 25% methanol in the incubation mixture caused a dramatic change in the enzymatic properties of the V 1 -ATPase. It exhibited a high Mg 2ϩ -dependent specific ATPase activity of 1.8 Ϯ 0.5 units/mg of protein (mean Ϯ S.D., five independent preparations), whereas the Ca 2ϩ -dependent ATPase activity was reduced approximately 10-fold. In contrast to the assays performed in the absence of methanol, the enzyme activity did not change with incubation time, suggesting that product inhibition was no longer occurring. ATP was the preferred substrate, GTP was still a reasonably good substrate, UTP and CTP were not effective substrates (Table I), and p-nitrophenyl phosphate was not hydrolyzed (not shown). Like the Ca 2ϩ -dependent ATPase activity, Mg 2ϩ -dependent activity was sensitive to N-ethylmaleimide with an IC 50 of approximately 0.1 M. This is the first time that Mg 2ϩ -dependent enzyme activity of a native V 1 -ATPase has been demonstrated.
Cytosolic V 1 -ATPase in Intermolt Larvae-Since the cytosol of midguts from molting larvae contains a large pool of V 1 -ATPase, it was tempting to speculate that this pool consisted mainly of V 1 complexes that had dissociated from the plasma membrane during molt. Consequently, the midgut cytosol of intermolt larvae should contain a significantly lower amount of V 1 -ATPase. Therefore, the cytosolic V 1 -ATPase from feeding larvae was purified by the same method as that used for midguts from molting larvae. The V 1 -ATPase again eluted in two peaks from the anion exchange column, but the yield with respect to the tissue wet weight or to the total protein content in the cytosolic extract was consistently about 70% lower than that obtained with midguts from molting larvae (0.56 Ϯ 0.04%; mean Ϯ S.D., three independent preparations). The specific ATPase activity in the presence of Ca 2ϩ without methanol, or of Mg 2ϩ with methanol, was in the same range as the V 1 -ATPase activity from molting larvae (not shown). SDS-gel electrophoresis revealed no difference between the V 1 complexes obtained from the molting and intermolt larvae. These results were in line with the hypothesis that a large part of the V 1 -ATPase pool in the cytoplasm of molting larvae is formed by the dissociation of intact V 1 complexes from the plasma membrane V-ATPase, leaving the V o parts in the membrane.
The physiological condition of starved intermolt larvae should be similar to that of molting larvae since both suffer from a lack of food. Therefore, we purified the cytosolic V 1 -ATPase from starving intermolt larvae and found that the amount of V 1 complex obtained after 17-19 h of starvation was as high as that obtained from molting larvae (approximately   (Fig. 5) or 1.68 Ϯ 0.24% (mean Ϯ S.D., three independent preparations) of total cytosolic protein). As expected, ATPase activity tests and SDS-gel electrophoresis revealed no differences between the V 1 complex preparations from the molting and feeding intermolt larvae (not shown). In line with the high amount of cytosolic V 1 -ATPase, Coomassie Blue stainings of SDS gels from partially purified goblet cell apical membranes showed that the amount of V 1 subunits was reduced (Fig. 4). Consequently, the membrane-bound V-ATPase activity was reduced drastically by about 70% (Fig.  5). When larvae were allowed to feed for 2 h after starvation, the amount of cytosolic V 1 -ATPase again approached the low level found in feeding intermolt larvae (0.67 Ϯ 0.20% total cytosolic protein; mean Ϯ S.D., three independent preparations), and membrane-bound V-ATPase activity recovered concomitantly (Fig. 5). Dissociation of the V 1 complex from the membrane V o part appears to be a method for down-regulation of V-ATPase activity not only during the molt but also during starvation and, most likely, under other physiological conditions. DISCUSSION We describe, for the first time, the purification of the native catalytic V 1 complex of a V-ATPase from the cytosol. This V 1 -ATPase could be isolated in considerable amounts and purity. The high yield (ϳ0.4 mg/g of tissue) is on the same order of magnitude as that of the preparation of chloroplast F 1 -ATPase from spinach leaves (36). Since the V 1 -ATPase occurred in vivo in the cytosol, it could be isolated without disintegration of the V-ATPase holoenzyme. As shown before for the reconstituted V c complex of the V-ATPase from clathrin-coated vesicles, the native insect V 1 complex exhibited Ca 2ϩ -dependent ATPase activity. For the first time, Mg 2ϩ -dependent enzyme activity of a native V 1 -ATPase has now been demonstrated under conditions similar to those used for studies of enzyme activity in F 1 -ATPases.
Composition of the Cytosolic V 1 -ATPase-Comparison of the native cytosolic V 1 complex of the M. sexta V-ATPase with the peripheral V 1 complexes from other sources of the V-ATPase holoenzyme (19 -21, 37-39) revealed its unique composition. In addition to the established V 1 subunits A, B, and E, the 14-kDa subunit F (8) as well as the novel 13-kDa subunit G (9) clearly appeared to be members of the V 1 complex. Subunits in the range of 40 and 32 kDa, corresponding to subunits C and D, do not occur in the V1 complex, suggesting that these subunits are unnecessary for either Ca 2ϩ -dependent ATPase activity or for the methanol-induced Mg 2ϩ -dependent ATPase activity of the complex. The lack of these two subunits was not unexpected since, even for the holoenzyme, there is no unequivocal indication that subunits C and D occur in the insect V-ATPase. Regarding subunit C, our results appear to be in conflict with those reported for the coated vesicle V 1 domain, which appears to exhibit a 20-fold reduced Ca 2ϩ -dependent ATPase activity in the absence of subunit C (15). Our results are, however, in agreement with other studies suggesting that significant V-ATPase activity can be observed in the absence of subunit C (19).
The weak bands in the range of 30 and 62 kDa may not represent components of the V 1 complex since they were very faint in Coomassie Blue stainings of SDS electrophoresis gels and since they copurified with the V 1 complex in varying amounts. However, as in the V-ATPase holoenzyme, we reproducibly found a weakly stained band in the range of 56 kDa, just below the strongly stained subunit B (Fig. 2). Multiple bands in the 56-kDa range are reminiscent of V-ATPase preparations from bovine kidney microsomes and brush border membranes. In these cases, more than one 56-kDa band was detected in two-dimensional polyacrylamide gels (40,41), apparently corresponding to isoforms of the B subunit. Although cDNAs encoding the different isoforms of the bovine B subunit have already been cloned (42,43), there is no genetic evidence for the existence of B-subunit isoforms in M. sexta.
In contrast to the cytosolic V 1 complex, the V 1 complex obtained after KI treatment of goblet cell apical membranes (KI-V 1 complex) did not contain the 14-kDa subunit F ( Fig. 6; see Lepier et al. (9)). However, when Malpighian tubule brush border membranes were stripped with KI in the presence of excess recombinant subunit F (8), the subunit F incorporated (during dialysis to remove the KI) into the nascent KI-V 1 complex (Fig. 6). Thus, subunit F is able to bind to the V 1 complex, but it appears to be not as strictly associated with the V 1 complex as the other four components. This conclusion is in line with our earlier suggestion that the 14-kDa subunit is located between the V 1 and V o parts of the enzyme (8). It is also in line with the results of Graham et al. (11), who reported that the 14-kDa subunit of the yeast V-ATPase, Vma7p, is involved in the assembly and stability of the V o complex.
Regulation of V-ATPase Activity by Disassembly and Reassembly of the V 1 V o Complexes-By producing a voltage in excess of 240 mV across the goblet cell apical membrane, the H ϩ -translocating V-ATPase in the tobacco hornworm midgut energizes electrophoretic K ϩ /2H ϩ antiport and thus net active K ϩ transport (see Ref. 44). The resulting K ϩ electrochemical potential drives all secondary transport processes across the midgut epithelium, including the absorption of amino acids and the regulation of the high pH in the midgut lumen (45). Maintaining active K ϩ transport requires an enormous amount of energy, consuming at least 10% of the total ATP production of the tobacco hornworm (46). Therefore, one would expect for reasons of economy that K ϩ transport should be down-regulated in those cases where less energy is needed. Indeed, Sumner et al. (22) detected that active K ϩ transport was downregulated during molt, and they showed that down-regulation was due to the detachment of the peripheral V 1 subunits from the plasma membrane V-ATPase. Since preliminary studies using Northern blots had indicated no significant change in mRNA concentration for peripheral subunits during the molt, they suggested that the V 1 complexes remain intact and are reassociated with the apical membrane at the end of the molt.
Our results support this hypothesis. While the membranebound V-ATPase activity decreases approximately 6-fold during molt (22), the amount of cytosolic V 1 complex and the Ca 2ϩ -dependent cytosolic V 1 -ATPase activity concomitantly increase more than 3-fold (Fig. 5). The disassembly of the V 1 V o complexes may reflect a down-regulation of the plasma membrane V-ATPase that, during molt, is not used for the energization of secondary active transport processes since the larva does not feed. However, the molting midgut epithelium is rather complex since new goblet and columnar cells, both deriving from undifferentiated stem cells, intercalate between the mature differentiated goblet and columnar cells (47). Therefore, we cannot exclude the possibility that all or a part of the V 1 subunits that detach from the apical membranes of mature goblet cells are degraded instead of remaining intact in integral cytosolic V 1 complexes. We also cannot exclude the possibility that the increased pool of cytosolic V 1 complexes during molt is produced by an up-regulation of V 1 complex biosynthesis not only in the newly emerging but also in the mature goblet cells. On the other hand, the high energy costs for an up-regulation argue against this interpretation because they would counteract the energy saving attained by down-regulation of active K ϩ transport during molt.
To evaluate this disassembly-reassembly hypothesis in the absence of the complex rearrangement of cells during the molt, we studied starving intermolt tobacco hornworms because neither molting nor starving larvae eat, and thus both suffer from a lack of food. We obtained cytosolic V 1 complexes from starving intermolt larvae in amounts similar to those from molting larvae. In line with this result, V 1 subunits detached from the goblet cell apical membranes during starvation, and the membrane-bound V-ATPase activity was reduced more than 3-fold. Although we cannot completely rule out that new protein biosynthesis would start with refeeding, it is tempting to assume that the V 1 complexes cycled back to the goblet cell apical membrane since the cytosolic V 1 pool decreased more than 3-fold and the membrane-bound V-ATPase activity increased concomitantly when the larvae were refed for only 2 h. This result is like that obtained in yeast, where V 1 complexes fall off the vacuolar membrane upon glucose deprivation and reassociate with membrane V 0 complexes upon restoration of glucose in the absence of protein biosynthesis (48). Perhaps the disassembly and reassembly of the V-ATPase, found in yeast and evidenced in both molting and starving tobacco hornworms, are general features of V-ATPase regulation, being the responses to a drop in energy (food) supply. They provide a second mechanism of V-ATPase regulation that contrasts with the insertion and removal of holoenzyme-containing vesicles in response to acid load in kidney intercalated cells (49).
Enzymatic Properties of the M. sexta V 1 Complex-Enzyme activity of V 1 complexes isolated by in vitro treatment has been found in the clathrin-coated vesicle V-ATPase and the insect plasma membrane V-ATPase (14,50). In both cases, the V 1 complex exhibited Ca 2ϩ -dependent ATPase activity but was silent in the presence of Mg 2ϩ . In contrast to these eukaryotic V 1 complexes, the prokaryotic V 1 complex of Enterococcus hirae exhibited, like F 1 -ATPases, Mg 2ϩ -dependent enzyme activity after it had been released from the membrane by EDTA extraction (51).
The native cytosolic V 1 -ATPase of M. sexta displayed differential patterns of activity, depending on the presence or absence of methanol. In the absence of methanol, the V 1 -ATPase did not accept Mg 2ϩ and was dependent on Ca 2ϩ , and Ca 2ϩdependent activity was inhibited by low Mg 2ϩ concentrations (0.1 mM). Furthermore, product ADP was inhibitory, whereas product GDP did not affect V 1 -ATPase activity. In the presence of methanol, the V 1 -ATPase preferred Mg 2ϩ over Ca 2ϩ and was no longer inhibited by ADP. The high methanol concentration evidently caused a conformational change in the V 1 -ATPase leading, perhaps due to the more hydrophobic environment, to properties which were similar to those of the membrane-bound holoenzyme (52). This interpretation is supported by the finding that methanol had no effect on the membrane-bound V-ATPase activity. 2 Thus the V 1 complex might occur in at least two different conformations, a cytosolic state and a membraneattached state. This speculation is in line with the observation that V 1 subunits can easily be stripped from the membrane by chaotropic salts, indicating that binding of the V 1 complex to the V o complex may be stabilized by hydrophobic interactions.
Similarity of the M. sexta V 1 -ATPase to F 1 -ATPases-Although the specific activity of the V 1 -ATPase is about 1 order of magnitude lower than that of F 1 -ATPases (e.g., Refs. 32 and 2 R. Grä f, and H. Wieczorek, unpublished results. FIG. 6. Affinity of the recombinant 14-kDa subunit F to the KI-V 1 complex. SDS-polyacrylamide electrophoresis gel stained with Coomassie Blue is shown. First lane, approximately 3 g of KI-V 1 complex with associated recombinant 14-kDa subunit F. 1 mg of recombinant fusion protein (subunit F with maltose-binding protein (8)) was digested with 15 g of Factor Xa protease (New England Biolabs Inc.) for 24 h as described (26) and added to malpighian tubule brush border membranes (ϳ1 mg of membrane protein (9)). The suspension containing 10 mM dithiothreitol was incubated for 3 h on ice before chaotropic treatment with KI and purification of the KI-V 1 complex were performed. Second lane, control of approximately 5 g of purified KI-V 1 complex without recombinant subunit F added (9). 34), the effects of Ca 2ϩ , Mg 2ϩ , and methanol on the catalytic properties of the M. sexta V 1 -ATPase are reminiscent of their effects on F 1 -ATPases. For example, the CF 1 -ATPase from spinach and the F 1 -ATPase from B. firmus both prefer Ca 2ϩ over Mg 2ϩ in the absence of methanol. By contrast, they both prefer Mg 2ϩ over Ca 2ϩ in the presence of methanol as do the respective membrane-bound F 1 F o holoenzymes (31,32). In the bovine heart mitochondrial F 1 -ATPase, the presence of methanol leads, as in the M. sexta V 1 -ATPase, to a total loss of inhibition by ADP (34).
The evolutionary relationship of F-and V-ATPases is based on the amino acid sequence similarity of the proton-translocating proteolipids and of the F-ATPase subunits ␣ and ␤ with the V-ATPase subunits B and A, respectively (53). Furthermore, the novel V 1 subunit G of M. sexta and the Vma10p protein of the yeast V-ATPase were found to exhibit amino acid sequence similarity to subunit ␤ of bacterial F-ATPases (9,12). Moreover, F-and V-ATPases are also similar in high resolution electron micrographs (54). The similarity in enzymatic properties of the M. sexta V 1 -ATPase and F 1 -ATPases adds a further argument for a close structural and functional relationship between these two enzyme families. The ultrastructure of the bacterial F 1 -ATPase was recently resolved by crytallographic analysis (13). Since sufficient amounts of pure V 1 -ATPase are easily available now, the crystallization of V 1 -ATPase may be possible, employing procedures similar to those that have proven to be successful for the bacterial F 1 -ATPase.