Concanamycin A, the Specific Inhibitor of V-ATPases, Binds to the Vo Subunit c*

Vacuolar-type ATPase (V-ATPase) purified from the midgut of the tobacco hornworm Manduca sexta is inhibited 50% by 10 nm of the plecomacrolide concanamycin A, the specific inhibitor of V-ATPases. To determine the binding site(s) of that antibiotic in the enzyme complex, labeling with the semisynthetic 9-O-[p-(trifluoroethyldiazirinyl)-benzoyl]-21,23-dideoxy-23-[125I]iodo-concanolide A (J-concanolide A) was performed, which still inhibits the V-ATPase 50% at a concentration of 15–20 μm. Upon treatment with UV light, a highly reactive carbene is generated from this concanamycin derivative, resulting in the formation of a covalent bond to the enzyme. In addition, the radioactive tracer 125I makes the detection of the labeled subunit(s) feasible. Treatment of the V1/Vo holoenzyme, the Vo complex, and the V-ATPase containing goblet cell apical membranes with concanolide resulted in the labeling of only the proteolipid, subunit c, of the proton translocating Vo complex. Binding of J-concanolide A to subunit c was prevented in a concentration-dependent manner by concanamycin A, indicating that labeling was specific. Binding was also prevented by the plecomacrolides bafilomycin A1 and B1, respectively, but not by the benzolactone enamide salicylihalamide, a member of a novel class of V-ATPase inhibitors.

V-ATPases in eukaryotes appear to be exclusive proton pumps that energize intracellular membranes of all cells as well as plasma membranes in a variety of animal cells (1). V-ATPases play crucial roles for many cellular processes and may also be involved in diseases such as osteoporosis and cancer (2).
More than a decade ago, bafilomycin A 1 , a plecomacrolide antibiotic containing a 16-membered lactone ring, was reported to be a specific inhibitor of V-ATPases with a K i value of less than 10 nM (3). A structure-activity study showed that the concanamycins, 18-membered plecomacrolides, exhibit 10-fold lower K i values and thus are better and more specific inhibitors of V-ATPases than the bafilomycins (4). Kinetic studies of the bovine chromaffin granule V-ATPase provided the first indirect evidence that bafilomycin A 1 interacts with the hydrophobic V o complex (5). Direct evidence came from two independent approaches, both using the bovine-coated vesicle V-ATPase (6,7). First, a latent acid-activated proton conductance of proteoliposomes containing the reconstituted V o complex could be blocked by 1 nM bafilomycin A 1 (6). Second, reconstituted vesicles containing the V o complex added to clathrin-coated vesicles could protect ATP-dependent proton transport from bafilomycin inhibition (7). Since considerable protection was also achieved by the reconstituted 100-kDa subunit a alone, it was concluded that the binding site for bafilomycin may reside in this subunit.
Affinity chromatography of chicken osteoclast V-ATPase using bafilomycin C 1 -coupled cellulose led to the purification of the holoenzyme (8). In the presence of 1 mM DCCD the V-ATPase did not bind to the column, indicating that the binding site for bafilomycin is near the binding site(s) for DCCD and may thus be located in the proteolipid, subunit c. However, because DCCD, especially at the rather high concentration of 1 mM, may bind to sites different from that in the proteolipid and may bind not only to carboxyl groups but also to sulfhydryl groups or to tyrosine residues (9), and since the 100-kDa subunit a was not clearly resolved in this study, the bafilomycin binding subunit was still not identified. Strong hints in favor of subunit c representing the bafilomycin binding subunit were provided by recent results from analysis of Neurospora crassa strains, indicating that those with point mutations in subunit c show a higher tolerance to bafilomycin (I 50 80 -400 nM) compared with the wild type (I 50 6.3 nM); interestingly, the tolerance to concanamycin was not changed simultaneously (10).
Based on extensive screening studies of systematically modified derivatives of concanamycin, a probe was designed that still inhibits V-ATPase activity and which generates upon UV-irradiation a covalent bond between the inhibitor and the V-ATPase (11,12). In the present report we show for the first time that this derivative of concanamycin A, 9-O-[p-(trifluoroethyldiazirinyl)-benzoyl]-21,23-dideoxy-23-[ 125 I]iodo-concanol-ide A (J-concanolide A), binds only to subunit c of the V-ATPase from tobacco hornworm (M. sexta) midgut.

EXPERIMENTAL PROCEDURES
Purification of the V 1 /V o ATPase and the V o Complex-As many as 20 whole M. sexta midguts were dissected from feeding fifth instar larvae and homogenized in ice-cold buffer A (5 mM Tris-HCl, 250 mM sucrose, 5 mM Pefabloc SC (Biomol), 5 mM Na-EDTA, pH 8.1). After centrifugation of the crude homogenate in a fixed-angle rotor at 12,000 ϫ g max for 5 min at 4°C, the pellet was resuspended in buffer A and centrifuged again. The resulting pellet was resuspended in buffer A and centrifuged in a fixed-angle rotor at 233,000 ϫ g max for 30 min at 4°C to yield a final pellet that was solubilized in buffer B (16 mM Tris-HCl, 0.32 mM EDTA, 9.6 mM 2-mercaptoethanol and 0.1% C 12 E 10 , pH 8.1) at 4°C with a detergent to protein ratio of 2:1. After centrifugation in a fixed-angle rotor at 233,000 ϫ g max for 1 h at 4°C, the supernatant was layered on a discontinuous sucrose density gradient (3 ml of 40%, 2 ml of 30%, 2 ml of 20%, and 1.6 ml of 10% sucrose (w/v) dissolved in buffer B with 200 mM KCl and a lower detergent concentration of 0.01% C 12 E 10 ) and centrifuged in a vertical rotor at 339,000 ϫ g max for 90 min at 4°C. The 30% sucrose fraction was diluted 4-fold with buffer C (20 mM Tris-HCl, 9.6 mM 2-mercaptoethanol, 0.01% C 12 E 10 , pH 8.1) containing 50 mM NaCl and subjected to anion exchange chromatography using a Mono Q (Amersham Biosciences) column. The V 1 /V o ATPase was eluted in a linear salt gradient (50 -400 mM NaCl dissolved in buffer C) at NaCl concentrations between 250 and 280 mM. The last step in the purification protocol was gel permeation chromatography on a Superdex 200 column (Amersham Biosciences), which was performed in buffer C containing 150 mM NaCl. The final yield was ϳ1 mg of V-ATPase per 20 midguts.
For purification of the V o complex, midguts of larvae starved for 16 -20 h were prepared, homogenized, and centrifuged twice as described above. The second pellet was resuspended in a buffer consisting of 5 mM Tris-HCl, 0.8 M KI, 5 mM Na-EDTA, and 5 mM Pefabloc SC (pH 8.1) and incubated for 30 min on ice. Then the sample was diluted with 16 mM Tris-HCl, 0.32 mM EDTA, 9.6 mM 2-mercaptoethanol (pH 8.1) to a KI concentration of 40 mM and centrifuged at 233,000 ϫ g max for 30 min at 4°C. The pellet was resuspended in the dilution buffer and centrifuged again under the same conditions. The final pellet was solubilized and purified as described above for the V 1 /V o ATPase, except for the use of 20% instead of the 30% sucrose fraction after rate zonal centrifugation. The final yield of the preparation was ϳ0.5 mg of V o complex per 20 midguts.
Antibiotics-J-concanolide A as well as its 125 I-labeled form were synthesized as described elsewhere (11). Concanamycin A, bafilomycin A 1 , and bafilomycin B 1 were isolated according to published procedures (12). Salicylihalamide was a generous gift from M. R. Boyd (National Cancer Institute, MD, USA). To avoid freeze thaw cycles, which have significant influence on the stability of the substances, aliquots of stock solutions in dimethyl sulfoxide were stored at Ϫ70°C and thawed only once immediately before use. The actual concentrations of the stock solutions were determined spectrophotometrically.
ATPase Assays-ATPase assays with a final volume of 160 l and a pH of 8.1 contained 3-4 g of protein, 50 mM Tris-Mops, 3 mM 2-mercaptoethanol, 1 mM MgCl 2 , 0.1 mM sodium orthovanadate, 0.5 mM sodium azide, 20 mM KCl, 0.003% C 12 E 10 , 20 mM NaCl, and 3 mM Tris-HCl. After 5 min of preincubation at 30°C with or without inhibitors, 1 mM Tris-ATP was added, and after an additional 2 min the reaction was stopped by placing the reaction tubes in liquid nitrogen. Determination of the produced inorganic phosphate followed the protocol of Wieczorek et al. (13).
Labeling-20 g of the samples were incubated with varying concentrations of J-concanolide A in a volume of 30 -40 l for 1 h on ice or 3 min at room temperature and irradiated for 3 min with UV light (366 nm) on ice. After irradiation, 7.5-10 l of 5-fold sample buffer (14) was added. The mixture was heated for 45 s at 95°C or 30 min at 40°C, cooled on ice, and subjected to SDS-PAGE (10 -15% T, 3.3% C; (15)) or to Tricine-SDS-PAGE (16.5% T, 3% C separating gel and 10% T, 3% C spacer gel (16)), followed by Coomassie staining. The gels were either dried on Whatman paper or sealed in plastic wrap before they were exposed to a phosphoscreen for as many as 72 h and analyzed with the aid of a phosphorimager (Molecular Dynamics).
Mass Spectrometric Analysis-After SDS-PAGE of the purified V 1 /V o ATPase, the 100-kDa and the 17-kDa Coomassie-stained bands, respectively, were prepared for mass spectrometry in a procedure adapted from Shevchenko et al. (17) and Zhang et al. (18). Briefly, the proteins were digested by trypsin or chymotrypsin in the gel, and the peptides were mapped using MALDI-MS (TofSpec-2E, Micromass Ltd., Manchester, UK).
Other Procedures-Fifth instar larvae of M. sexta (Lepidoptera, Sphingidae), weighing 6 -8 g, were reared under long-day conditions (16 h of light) at 27°C using a synthetic diet modified according to Bell et al. (19). The preparation of highly purified goblet cell apical membranes followed published protocols (13,20). SDS-PAGE, Western blotting on nitrocellulose membranes, and immunostaining were performed as described previously (13,21).

Revised Purification Protocol Reveals Subunit a in the Insect
V-ATPase-The unequivocal assignment of an inhibitor such as concanamycin or bafilomycin to a special V-ATPase subunit requires experimental evidence for the existence of all constitutive subunits in the V-ATPase preparation used. Although nearly all subunits of the insect V-ATPase have been cloned and sequenced (22,23), we had so far not been able to identify the 100-kDa subunit a of the insect V-ATPase unequivocally (24 -26). Because this subunit on the one hand appeared to be very sensitive to proteolysis (27), and on the other hand was proposed to be a candidate for bafilomycin binding (7), we repeatedly modified purification protocols to recover a putative subunit a, also in the insect V-ATPase. Degradation during the process of purification was evidently not responsible for our lack of evidence because even with complex cocktails of protease inhibitors we were not able to resolve subunit a (28).
Purification of the M. sexta V-ATPase according to a new protocol as compared with previous procedures (Ref. 24, see under "Experimental Procedures") resulted, mainly because of the presence of 200 mM KCl during zonal centrifugation of the detergent-solubilized protein in a sucrose gradient, for the first time in the detection of a 100-kDa subunit in an insect V-ATPase preparation (Fig. 1, lane 3). The 100-kDa band reacted in immunoblots with polyclonal antibodies to the 116-kDa subunit a of the V-ATPase from bovine chromaffin granules (Ref. 29, not shown, but see below the results for the V o complex), suggesting the existence of a 100-kDa subunit a, also in the M. sexta midgut V-ATPase. Respective evidence had already been provided recently by the demonstration of at least two genes encoding a V-ATPase subunit a in the midgut (23). Definitive proof that the 100-kDa band represented the V-ATPase subunit a was finally obtained from MALDI-MS. A number of tryptic peptides were found in the map of the protein ( Fig. 2A), providing 36% sequence coverage (compare Ref. 23 with GenBank TM accession no. AJ249390).
Purification of the V o complex was facilitated by the fact that goblet cell apical membranes of molting or starving M. sexta larvae lose most of their V 1 complexes by dissociation from the membrane, whereas the V o complexes remain membrane bound (30,31). Further treatment of the membranes with chaotropic iodide (32) supported the dissociation of V 1 complexes, leaving virtually V 1 -free membranes. According to silver staining of SDS gels, the V o complex appeared to consist of five major protein bands with apparent molecular masses of 100, 40, 26, 20, and 17 kDa, respectively (Fig. 1, lane 4). Four polypeptides could be identified as established V o subunits by staining Western blots with specific antibodies against the V o subunits a, d, and e, respectively (Fig. 1, lanes 6 -8) or by DCCD-labeling of subunit c (Fig. 1, lane 5). The fifth major band with an apparent molecular mass of ϳ26 kDa was N-terminally sequenced and appeared to be a dimer of subunit c (data not shown).
Plecomacrolide Antibiotics Bind to Subunit c of the Insect V-ATPase-J-concanolide A (Fig. 3) was synthesized as described elsewhere (11). Two features of this potential inhibitor appeared to be essential for the identification of the plecomacrolide binding subunit(s). First, the p-(trifluoroethyldiazirinyl)-benzoyl group at position C9 generates, after treatment with UV light, a highly reactive carbene, which itself can be inserted into covalent bonds, especially C-H bonds. Second, the 125 I at position C23 of the hemiketal ring could be used as a radioactive tracer, thus making detection of the labeled subunit by autoradiography possible.
Before performing labeling experiments we tested the inhibitory efficacy of J-concanolide A on the activity of the V-ATPase holoenzyme. As shown in Fig. 4, J-concanolide A inhibited the purified V-ATPase half-maximally at a concentration of 15-20 M, equivalent to an I 50 value of ϳ0.6 mol per mg of protein. This value was 1000 times higher than that for the plecomacrolides concanamycin A and bafilomycin A 1 and B 1 , respectively, and the benzolactone enamide salicylihalamide. All of these substances inhibited the M. sexta V-ATPase at submicromolar concentrations, with I 50 values of ϳ0.5 nmol per mg and with half-maximal concentrations of 10 nM. Although J-concanolide A was three orders of magnitude less inhibitory, it still was an effective inhibitor with an I 50 value comparable to that of the Escherichia coli Kdp-ATPase for bafilomycins (4).
UV illumination of the purified M. sexta V-ATPase in the presence of J-concanolide A led, after SDS-PAGE, Coomassie staining, and autoradiography of the gel to the detection of 125 I at the exact position of the V-ATPase subunit c (Fig. 5A). The identity of the band as subunit c was proven by MALDI-MS. Analysis of tryptic and chymotryptic peptides resulted in a sequence coverage of 76% with subunit c (Fig. 2B, compare Ref. 33 with GenBank TM accession no. X65051). However, two distinct masses of 1523,9 m/z and 1739,9 m/z, respectively, were not covered by the known sequence of subunit c. These frag- ments might derive from an isoform of the c subunit, c', which is present in V-ATPases of other species (2) but so far could not be identified in M. sexta. On the one hand, genomic Southern blots had suggested that there is only one single gene for subunit c in M. sexta, but on the other hand, Northern blots had revealed two transcripts (33). The latter finding may be interpreted as a hint in favor of a c' isoform. Nevertheless, labeling of the band representing subunit c and/or a putative isoform was specific because it was dependent on the concentration of J-concanolide A (Fig. 5A) and because labeling with 10 M J-concanolide A was prevented by preincubation with concentrations of concanamycin A ranging from 1 to 300 M (Fig. 5B). Fig. 5B shows that a concentration of 1 M concanamycin was not enough to displace J-concanolide A. At first glance this may appear counterintuitive, because at a concentration of 0.1 M the V-ATPase activity was already inhibited nearly 100%. However, one has to keep in mind that already the binding of one molecule of concanamycin A to one of the probably six subunits c in a V o complex may lead to complete inhibition of enzyme activity, whereas six molecules of concanamycin A would be needed for complete displacement of J-concanolide A. To verify the correlation between inhibition and labeling by J-concanolide A, we cut out the Coomassie-stained bands of subunit c from the SDS-PAGE gel and measured their radioactivity in a ␥-counter. Indeed, a concentration-dependent correlation between bound radioactivity and inhibition of the enzyme activity was found (Fig. 6). Labeling of the holoenzyme subunit c evidently was specific for plecomacrolides, because it was also prevented by bafilomycin A 1 and B 1 , respectively (Fig.  5C). This finding implied that the binding site for both families of plecomacrolide antibiotics, concanamycin and the bafilomycins, in general resides in subunit c. At present, we cannot  1, 3, and 5). Autoradiographies of gels after exposition to a phosphoscreen (lanes 2, 4, and 6). Evidently during electrophoresis the hydrophobic J-concanolide A entered the gel together with the SDS of the running buffer. 3rd-9th lane, with UV-illumination. Mainly the J-concanolide A bound to protein entered the gel. Most of the unbound but photolyzed J-concanolide A, perhaps because of formation of cross-linked products with, for example, the non-ionic detergent C 12 E 10 had no electrophoretic mobility and therefore remained in the stacking gel (data not shown). B, autoradiography of the gel after exposition to a phosphoscreen. Samples were preincubated for 1 h on ice without concanamycin A (control) or with the indicated concentrations of concanamycin A. J-concanolide A was then added to give a final concentration of 10 M. The mixture was incubated for 1 h on ice and treated afterward with UV light. C, autoradiography of the gel after exposition to a phosphoscreen. Samples were preincubated for 1 h on ice with the indicated concentrations of bafilomycin A 1 (Baf A1), bafilomycin B 1 (Baf B1), or salicylihalamide (Salicyl), respectively. J-concanolide A was then added to give a final concentration of 10 M. The mixture was incubated for 1 h on ice and treated afterward with UV light. Control with preincubation, but without effectors. explain the discrepancy between this result and the results obtained with N. crassa mutants, which exhibited a higher tolerance to bafilomycin but not to concanamycin (10). Because labeling was not impaired by salicylihalamide, we conclude that the site(s) and mechanism of inhibition for benzolactone enamides may be different from that for plecomacrolides.
In another approach we used, in addition to the purified holoenzyme, the purified V o complex and highly purified goblet cell apical membranes in which the V-ATPase is the predominant protein and, moreover, resides in its native lipid surrounding (Fig. 7). Clear labeling of subunit c was again obtained for the V o complex as well as for the goblet cell apical membrane.
Taken together, we provided for the first time direct proof that subunit c, which forms the major part of the proton translocating V o complex of V-ATPases, carries the binding site for plecomacrolide antibiotics. Experiments more precisely defining the site of covalent modification are in progress.