The bafilomycin/concanamycin binding site in subunit c of the V-ATPases from Neurospora crassa and Saccharomyces cerevisiae.

The vacuolar H+-ATPase is inhibited with high specificity and potency by bafilomycin and concanamycin, macrolide antibiotics with similar structures. We previously reported that mutation at three residues in subunit c of the vacuolar ATPase from Neurospora crassa conferred strong resistance to bafilomycin but little or no resistance to concanamycin (Bowman, B. J., and Bowman, E. J. (2002) J. Biol. Chem. 277, 3965-3972). We have identified additional mutated sites in subunit c that confer resistance to bafilomycin. Furthermore, by subjecting a resistant mutant to a second round of mutation we isolated strains with increased resistance to both bafilomycin and concanamycin. In all of these strains the second mutation is also in subunit c, suggesting it forms at least part of the concanamycin binding site. Site-directed mutagenesis of the gene encoding subunit c in Saccharomyces cerevisiae showed that single mutations in each of the residues identified in one of the double mutants of N. crassa conferred resistance to both bafilomycin and concanamycin. Mutations at the corresponding sites in the VMA11 and VMA16 genes of S. cerevisiae, which encode the c' and c" subunits, did not confer resistance to the drugs. In all, nine residues of subunit c have been implicated in drug binding. The positions of these residues support a model in which the drug binding site is a pocket formed by helices 1, 2, and 4. We hypothesize that the drugs inhibit by preventing the rotation of the c subunits.

Vacuolar proton-translocating ATPases (V-ATPases) 1 are large, complex enzymes responsible for acidification of many internal compartments in eukaryotic cells. They also occur on plasma membranes of specialized cells, where they acidify the surrounding milieu. Numerous physiological processes depend on the activity of V-ATPases, including protein processing, transport of metabolites, receptor-mediated endocytosis, neurotransmitter uptake, and apoptosis (1,2). V-ATPases are implicated as a contributing factor in multiple diseases such as osteoporosis (3), deafness (4), and cancer (5). Because of its vital role in so many processes and diseases, the V-ATPase is an attractive target for the development of therapeutic agents (6 -10).
The bafilomycins and concanamycins have been particularly important in elucidating the physiological role of V-ATPases and are under investigation as possible lead compounds for drug therapy directed toward this enzyme (9,11). These macrolide antibiotics, which are very similar in structure, are specific, highly potent inhibitors of all eukaryotic V-ATPases in vitro and in vivo (12)(13)(14). Despite the widespread use of these compounds, however, our understanding of how they inhibit is rudimentary. Early reports indicated that bafilomycin and concanamycin acted on the Vo segment of the enzyme (15)(16)(17)(18). Using a radiolabeled derivative of concanamycin, Huss et al. (19) identified subunit c as the site bound by the inhibitor in the Manduca sexta enzyme.
We have used a genetic approach to identify specific amino acid residues involved in binding the drugs. Eleven bafilomycin-resistant mutants (bfr strains) of Neurospora crassa were initially isolated, and all had mutations in subunit c (20). Among the 11 strains, three different residues were altered. A surprising outcome of the study was that mutations that conferred resistance to bafilomycin had little effect on inhibition by concanamycin. Assayed in vitro, the V-ATPase from the bfr mutant strains was 13-70-fold more resistant to bafilomycin than the wild type strain but less than 3-fold more resistant to concanamycin. Because bafilomycin and concanamycin have similar structures, we had expected them to bind the V-ATPase at the same site. The data suggested this might not be correct.
The consensus model of the structure of V-ATPases, which applies to yeasts, filamentous fungi, insects, and mammals, proposes 13 different subunits, several present in multiple copies (1). These are organized into two sectors, the V1 subunits (A-H), which can be removed from the membrane in soluble form, and the Vo subunits (a, c, cЈ, cЉ, and d), which are tightly associated with the membrane. Like the F-ATPase, the enzyme functions as a rotary motor (21,22). Within Vo, subunits c, cЈ, and cЉ constitute the rotor. Subunit c is present in multiple copies (4 -5); subunits cЈ and cЉ are present as a single copy each (23)(24)(25). In fungi all three proteolipids are essential components of each V-ATPase molecule. It is not yet clear if the V-ATPase in animal cells has a cЈ subunit (26,27).
Previously, we proposed a simple model to explain bafilomycin inhibition of the V-ATPase; binding occurs at the interface of the rotating c subunits and the stationary a subunit, blocking rotation (20). In the current paper we identify additional bfr mutations in N. crassa and describe strains with increased resistance to both bafilomycin and concanamycin. Using Saccharomyces cerevisiae we demonstrate that homologous mutations in subunit c confer resistance as in N. crassa and ask if mutations in the corresponding sites in subunits cЈ and cЉ can confer resistance to the macrolide antibiotics. The positions of the altered residues within subunit c lead us to a hypothesis for the folding of that polypeptide and a more detailed proposal for the bafilomycin and concanamycin binding sites.

EXPERIMENTAL PROCEDURES
N. crassa Strains; Growth of Cells-N. crassa strain 74A (or 74a) was used as the wild type. A and a refer to the two mating types of N. crassa. Bfr33 is a bafilomycin-resistant strain with the mutation T32I in subunit c of the V-ATPase. Strains were maintained on Vogel's medium N (a minimal medium salt solution at pH 5.8) supplemented with 2% sucrose. For experiments designed to test resistance to bafilomycin or concanamycin, this medium was modified by the addition of 20 mM HEPES buffer and was adjusted to pH 7.2 with NaOH. Bafilomycin or concanamycin from a 10 mM stock solution in dimethyl sulfoxide was added after autoclaving the medium. To compare the morphology of germinating spores growing in the presence of inhibitors, we photographed the colonies on agar plates with a digital camera mounted on a dissecting microscope.
S. cerevisiae Strains; Growth of Cells-The S. cerevisiae strains and plasmids used are summarized in Table I. They were maintained on YEPD medium or synthetic minimal medium supplemented with the appropriate amino acids. The dilution spotting growth assays were carried out by growing cells to saturation in liquid YEPD medium. Cell numbers were estimated by absorbance at 600 nm and diluted to 1 ϫ 10 7 cells/ml in water. Serial 5-fold dilutions of these cells in water (10-l aliquots) were pipetted onto YEPD plates adjusted to pH 7.5 with 50 mM HEPES/NaOH buffer and supplemented with bafilomycin or concanamycin. We incubated the plates at 30°C and photographed them after 2 and 3 days.
Mutagenesis and Selection of Mutant Strains in N. crassa-Two more bfr (bafilomycin resistant strains) were isolated as described previously (20). Briefly, we irradiated conidia from the wild type strain 74A with ultraviolet light and selected for growth on Vogel's minimal medium adjusted to pH 7.2 and supplemented with 1.0 M bafilomycin A1. Four bcr (bafilomycin-and concanamycin-resistant) strains were isolated by a similar procedure with the following differences. We started with strain bfr33a, which carries the T32I mutation in subunit c of the V-ATPase, and selected for growth on medium at pH 7.2 containing 0.5 M concanamycin C. Putative mutants were backcrossed to strain bfr33 al-3A to test for linkage of the new mutation to the vma-3 locus, which encodes subunit c. Al-3 (albino) is closely linked (ϳ5 map units) to vma-3.
Site-directed Mutagenesis of S. cerevisiae VMA3, VMA11, and VMA16 -Mutations were introduced into epitope-tagged versions of VMA3, VMA11, and VMA16 previously subcloned into the yeast CEN vector pRS316 using the QuikChange site-directed mutagenesis method (Stratagene). The following amino acid changes were made by the site-directed mutagenesis of each codon as indicated: Vma3p T32I (act 3 att), Vma3p F135L (ttt 3 ttg), Vma3p Y142N (tac 3 aac), Vma11p T38I (aca 3 ata), Vma11p F143L (ttc 3 ttg), Vma11p Y150N (tat 3 aat), Vma16p F106L (ttc 3 tta), and Vma16p Y113N (tac 3 aac). Using pLG47 as template, QuikChange mutagenesis was also used to generate Vma3p T32I (act 3 att), Vma3p I54F (att 3 ttt), and the double mutant T32I/I54F before integration into the yeast genome. These mutations were introduced into the genomic locus by single step gene replacement after transformation with the SacI/XhoI fragment of pLG160, -161, or -162 and selecting for Ura ϩ transformants. The presence of each mutation was confirmed by sequencing the plasmid insert using custom sequencing primers complementary to the 5Ј-flanking sequence and located ϳ100 bp upstream of the initiating codon. Sequence was generated by the DNA sequencing facility at the University of Oregon. Site-directed mutations in VMA3, VMA11, or VMA16 present on pRS316 vectors were transformed using standard yeast methods into their respective deletion strains and tested for the ability to complement the Vma Ϫ phenotype by growth on YEPD plus 100 mM CaCl 2 .
Isolation of Vacuolar Membranes, Analysis of ATPase Activity, and the Effects of Inhibitors-Vacuolar membranes were prepared from N. crassa as described previously (28) and modified (29). Vacuolar membranes were prepared from S. cerevisiae as described (30). Protein was assayed by the Lowry procedure (31). The assay mixture for N. crassa V-ATPase contained 5 mM Na 2 ATP, 5 mM MgSO 4 , 10 mM NH 4 Cl, 5 mM KN 3 , 0.1 mM Na 3 VO 4 , and 10 mM PIPES buffer, adjusted to pH 7.4 with Tris base. The assay mixture for S. cerevisiae V-ATPase contained 5 mM Na 2 ATP, 5 mM MgCl 2 , 10 mM NH 4 Cl, 5 mM KN 3 , 0.1 mM Na 3 VO 4 , and 25 mM MES buffer, adjusted to pH 6.9 with Tris base. V-ATPase activity was assayed by the Fiske-Subbarow colorimetric procedure as described (32). Activity assays were typically run for 25 min at 37°C. Bafilomycin and concanamycin were added to the V-ATPase assay mix from a 1 mM stock solution in dimethyl sulfoxide for the highest concentration of inhibitor being tested (typically 0.3 or 1.0 M); this solution was diluted serially in V-ATPase mix to obtain the desired lower concentrations of inhibitor. When K i values were compared for different strains, the titrations were carried out at the same time in the same assay mixture to compensate for instability of inhibitors when thawed several times. The reaction was initiated by the addition of vacuolar membranes.
Isolation and Sequencing of DNA-N. crassa genomic DNA was prepared from 20 mg of lyophilized mycelia, using the DNeasy Plant kit (Qiagen). The protein coding region of the vma-3 gene was amplified by the polymerase chain reaction. The 5Ј primer was CACGGCAATCTC-CAATTC, and the 3Ј primer was the reverse complement of GAACTCG-GTCTAGGTCTCC. Amplified DNA of 1080 bp was purified by separa-

RESULTS
Isolation of Two New bfr Strains-We previously described N. crassa strains mutated at three sites in the gene encoding subunit c of the V-ATPase. These mutations conferred resistance to bafilomycin in vivo and in vitro. In a further screen with mutagenesis of strain 74A and selection with bafilomycin we isolated two additional mutants whose properties were similar to those of the original bfr strains. V-ATPase activity was assayed in vitro for sensitivity to bafilomycin and concanamycin. With bafilomycin, we obtained K i values of 4.6 nM for 74A (wild type control), 94 nM for bfr34, and 17 nM for bfr314 (Fig.  1A). With concanamycin the K i values were 3.5 nM for 74A, 7.3 nM for bfr34, and 5.6 nM for bfr314 (Fig. 1B). Thus, the V-ATPase of the two mutant strains showed an increase in resistance to bafilomycin of 20-and 4-fold and a smaller increase in resistance to concanamycin of 2.1-and 1.6-fold. The mutations mapped to vma-3, and sequencing of the vma-3 gene identified the mutations as I39F in bfr34 and as L140F in bfr314 in subunit c.
Isolation of Strains with Significant Resistance to Concanamycin-Several years ago we mutagenized strain 74A of N. crassa and selected for growth on concanamycin at pH 7.2 (29). We obtained ϳ100 concanamycin-resistant strains. However, none of these grew vigorously in medium containing the inhibitor, and none were altered in a V-ATPase gene. We re-cently repeated this experiment and again failed to obtain any mutants altered in the V-ATPase. Reasoning that double mutants might confer strong resistance to concanamycin, we mutagenized the bfr33 strain, which has the T32I mutation in subunit c. This strain was chosen because it had the strongest resistance to the inhibitors of our original four bfr strains, a 70-fold increase in resistance to bafilomycin and a 3-fold increase in resistance to concanamycin. To select mutant strains, conidia of strain bfr33a were exposed to ultraviolet irradiation and spread on pH 7.2 plates containing 0.5 M concanamycin. Strain bfr33a cannot grow on this medium. After 2-3 days, many conidia had germinated and formed colonies. Selecting larger colonies with more normal morphology, we isolated 70 colonies on day 2 and 86 colonies on day 3. The isolates were designated bcr1-bcr156 (for bafilomycin/concanamycin resistant) and were initially maintained on the same medium with concanamycin.
All the isolates were backcrossed to strain bfr33A (T32I) to see if the resistance phenotype was heritable and to eliminate secondary mutations that may have been present in the primary isolate. Ninety-three of the strains produced ascospores that germinated on pH 7.2 medium containing 0.5 M concanamycin. However, most of the colonies grew poorly and had abnormal morphology. As in our previous genetic screens for drug-resistant mutants, most of the strains had mutations, not in the V-ATPase but in genes that were able to partially suppress the toxic effects of concanamycin. In the bcr strains occurrence of a second mutation in vma-3 strongly correlated with vigor of growth on the concanamycin plates. Approximately 10% of the strains grew vigorously and had nearly normal hyphal morphology, characterized by long straight hyphae with smaller side branches (Fig. 2, bcr 31, 69, 118, and 149). Most of the strains produced smaller colonies, often with dichotomous branching at the hyphal tips (Fig. 2, bcr 50, 37,19,134). In 8 of the 11 most vigorously growing strains sequencing of the vma-3 gene revealed both the original bfr mutation and a second mutation. Three strains were T32I/I54F mutants; one was a T32I/V55A mutant, two were T32I/M130I mutants, and two were T32I/L132F mutants (see below).
Three strains that grew reasonably well on concanamycin did not show genetic linkage to al-3, which lies within 5 map units of vma-3. In vitro assays of V-ATPase activity in the unlinked strains showed no increase in resistance to concanamycin. We identified one as a pma-1 mutant (bcr19) and did not pursue the others. The vma-3 gene was sequenced in nine other strains that grew weakly on medium with concanamycin. None of these had a second mutation in the vma-3 gene.
Assessment of V-ATPase Resistance to Concanamycin and Bafilomycin in Vitro and in Vivo-To test the effect of concanamycin and bafilomycin on the bcr strains with double mutations in subunit c, we chose one strain from each category, isolated vacuolar membranes, and assayed V-ATPase activity. As shown in Fig. 3 and Table II, the V-ATPases of the bcr strains showed increased resistance to both inhibitors. Compared with the original strain bfr33 (T32I), the V-ATPases of strains bcr31 (T32I/V55A) and bcr118 (T32I/M130I) were 2-fold resistant to concanamycin; the V-ATPase of bcr149 (T32I/ L132F) was 3-fold resistant, and that of bcr69 (T32I/I54F) was 12-fold resistant (Fig. 3A). Thus, the enzymes from the bcr strains were 6 -39-fold more resistant to concanamycin than the wild type enzyme. Three of the bcr mutants showed a further increase in resistance to bafilomycin in vitro as well. The V-ATPases of strains bcr118 (T32I/M130I), bcr149 (T32I/ L132F), and bcr69 (T32I/I54F) were, respectively, 3-, 4-, and 5-fold resistant to bafilomycin compared with bfr33 (T32I) or 37-325-fold resistant compared with 74A (WT) (Fig. 3B). Only bcr31 (T32I/V55A) showed no increase in resistance to bafilomycin (Fig. 3B).
The characteristic phenotype of S. cerevisiae and N. crassa cells lacking a functional V-ATPase is the inability to grow on medium at alkaline pH (33,34). Concanamycin prevents growth of wild type N. crassa on medium at pH 7.2, indicating that the V-ATPase is inhibited and inactive. At pH 5.8, concanamycin does not prevent growth of N. crassa, but it profoundly affects the morphology of the filamentous fungus, reducing the rate of hyphal extension and altering the pattern of hyphal branching. The mutations in the bcr strains partially suppressed this phenotype. In the presence of concanamycin, strains bcr31 (T32I/V55A), bcr69 (T32I/I54F), bcr118 (T32I/ M130I), and bcr149 (T32I/L132F) grew vigorously at pH 7.2 and had a nearly normal growth pattern, long straight hyphae with small side branches (Fig. 2). We did not see a direct correlation between the degree of resistance to concanamycin in vitro and in vivo. In vitro, the order of resistance of the V-ATPase was bcr69 (T32I/I54F) Ͼ bcr149 (T32I/L132F) Ͼ bcr118 (T32I/M130I) ϭ bcr31 (T32I/V55A) (Fig. 3 and Table II). In vivo, the strongest growers in the presence of concanamycin were bcr69 and bcr31 followed by bcr118 and bcr149 (Fig. 2). The mutations affected the activity of the enzyme as well as the binding of inhibitor. The specific activity of the V-ATPase in vacuolar membranes (average of two preparations) was 4.8 mol/min/mg for bcr31 (T32I/V55A), 2.3 mol/min/mg for bcr69 (T32I/I54F), 3.0 mol/min/mg for bcr118 (T32I/M130I), and 1.3 mol/min/mg for bcr149 (T32I/L132F). The relatively high specific activity of the bcr31 enzyme might explain its stronger growth when compared with bcr149.
Attempting to get more concanamycin-resistant mutants, we performed a similar screen for concanamycin-resistant double mutants with bfr21 (F136L) as the initial strain. The V-ATPase of bfr21 (F136L) is 13-fold resistant to bafilomycin and is not resistant to concanamycin in vitro as compared with the wild type enzyme. We found no concanamycin-resistant strains that  mapped to vma-3. Similarly, another attempt to obtain concanamycin-resistant mutants altered in the V-ATPase from strain 74A yielded no new mutants. In each of our mutant runs we have pursued several of the mutants that did not show linkage to vma-3, hoping to find a mutation in a different V-ATPase gene. Activity assays in vacuolar membranes have shown no increase in resistance of the V-ATPase to the antibiotics in these other strains.
Effect of bfr Mutations on V-ATPase of S. cerevisiae-The c, cЈ, and cЉ subunits of V-ATPases are highly conserved proteins. Comparison of the sequences of these proteins from N. crassa and S. cerevisiae shows 79% identity between c subunits, 64% identity between cЈ subunits, and 63% identity between cЉ subunits. Moreover, the three proteins of the V-ATPase have apparently evolved as gene duplications and fusions of an ancestral gene similar to the F-ATPase c subunit gene (35). Subunits c and cЈ with 4 membrane-spanning helices are nearly the same size (17 kDa). Subunit cЉ is a larger protein (23 kDa). The N terminus is predicted to have an additional membrane-spanning domain, and whereas conflicting results have been published (36, 37), our recent results clearly place the N terminus of the cЉ subunit in the cytosol and the C terminus in the lumen (38). 2 Therefore, cЉ (Vma16p) has five membrane-spanning helices, and helices 2 and 3 of cЉ align best with helices 3 and 4 of c and cЈ. Consistent with this arrangement, each of the three proteins contains a single membrane-embedded glutamate residue that is essential to proton transport, located in helix 4 in subunits c and cЈ and in helix 3 in subunit cЉ (24).
Taking advantage of the ease of doing site-directed mutagenesis and of the extensive information available on the Vo domain of the V-ATPase in S. cerevisiae, we asked if mutations corresponding to bfr mutations in N. crassa would have similar effects on the V-ATPase. Furthermore, would mutations in subunits cЈ and cЉ homologous to those in subunit c confer resistance to bafilomycin? We constructed the mutations in the plasmid borne versions of VMA3, VMA11, and VMA16 as indicated in Table I (see also "Experimental Procedures"). The mutated plasmids were then transformed into the appropriate yeast deletion strains and screened for their ability to complement the deleted gene. Homologous mutations for the three N. crassa sites in subunit c (Vma3p) are T32I, F135L, and Y142N in S. cerevisiae. Only the T32I mutation complemented the vma3 deletion strain (pLG110 in LGY101). Western blot analysis confirmed that the Vma3p in the F135L and Y142N mutant strains was stably present in the yeast cells (pLG112 and pLG113 in LGY101; data not shown). Localization of the nonfunctional Vma3p mutants by immunofluorescence indicated the proteins are localized to the membranes of the endoplasmic reticulum and not the vacuole (data not shown), consistent with their inability to assemble a functional V-ATPase. Noumi et al. (39) also observed that the Y142N mutation in VMA3 did not complement a vma3 deletion strain made in the W303 yeast strain background. The corresponding mutations in VMA11, which encodes subunit cЈ, were T38I, F143L, and Y150N. These mutants complemented the vma11 deletion strain (pLG114, pLG115, or pLG116 in RHA108). Similarly, the two vma16 (encodes subunit cЉ) mutant plasmids, F106L and Y113N, also made functional V-ATPases as shown by complementation of the vma16 deletion strain (pLG117 or pLG118 in LGY9).
We used a dilution plating assay to test for in vivo resistance of the yeast strains to bafilomycin. The plates contained YEPD medium adjusted to pH 7.5 supplemented with 0, 1, or 2 M bafilomycin. The vma3-T32I (pLG110 in LGY101) mutant showed strong resistance, growing almost as well on 2 M bafilomycin as in the absence of inhibitor (Fig. 4). Its control strain, the wild type VMA3 gene inserted on a plasmid and expressed in a vma3 yeast deletion strain, was strongly inhibited by the bafilomycin (pLG91 in LGY101). The vma11 deletion strain carrying plasmids with vma11 mutant genes showed no resistance to bafilomycin (pLG114, -115, or -116 in RHA108) and in fact grew less well than the control strain (pLG87 in RHA108). A similar result was observed for vma16. The wild type control (with VMA16 on a plasmid; pLG84 in LGY9) grew better than either of the two vma16 mutants on the 2 M bafilomycin plate (pLG117 or pLG118 in LGY9; Fig.  4). We conclude from the growth tests that the vma3-T32I mutant strain showed strong resistance to bafilomycin, similar to the results found in N. crassa. The vma11 and vma16 mutant strains did not appear resistant, suggesting that single mutations in these subunits did not confer resistance to bafilomycin.
In Vitro Analysis of S. cerevisiae Strains with V-ATPase Mutations-We next wanted to determine whether in vitro assays were consistent with the results obtained in vivo. We isolated vacuolar membranes from four strains and assayed the sensitivity of the V-ATPase to bafilomycin and concanamycin. As in the growth tests, the vma3-T32I strain exhibited strong resistance to bafilomycin, with a K i of 36.6 nM compared with a value of 2.2 nM for the wild type control. Comparable with results with the N. crassa bfr33 (T32I) strain, we saw only weak resistance to concanamycin, with a K i of 6.0 nM in the mutant compared with a K i of 2.2 nM in the WT control. Results with the homologous mutant in vma11 also agreed with the growth tests; we saw no increase in resistance to the inhibitors. For both the vma11-T38I strain and the WT VMA11 strain we obtained K i values of 3.8 nM for bafilomycin and 0.7 nM for concanamycin.
A Double Mutation in VMA3 Is Needed for Strong Resistance to Concanamycin-In N. crassa we were able to isolate concanamycin-resistant mutants altered in a V-ATPase gene only by selecting for double mutants. To determine the effect of each of the single mutants in one of the double mutant strains we constructed three versions of the VMA3 gene in S. cerevisiae, T32I/I54F, T32I, and I54F. The double mutant corresponds to N. crassa strain bcr69, described above. Because we had obtained somewhat low V-ATPase-specific activities in strains carrying mutations on centromeric plasmids, we decided to integrate the mutations into the chromosome for this test. Vacuolar membrane vesicles were prepared, and V-ATPase activity was assayed as before. The K i values for inhibition by bafilomycin were 3.8 nM for the 1D␣ WT control, 42 nM for the T32I mutant (LGY116), and 6.4 nM for the I54F mutant (LGY117; Fig. 5A). Inhibition by concanamycin gave K i values of 2.6 nM for the WT control, 7.7 nM for vma3-T32I, and 7.7 nM for vma3-I54F (Fig. 5B). We were unable to estimate K i values for the double mutant (LGY118) because of the very low activity of the V-ATPase, 0.02 mol/min/mg of protein. These three strains were also tested for growth at pH 7.5 using the same method as shown in Fig. 4. The 1D␣ WT control and the T32I mutant strain grew well at all five dilutions. The I54F grew more slowly at the highest dilution, whereas the T32I/I54F double mutant grew only at the lowest dilution. These data suggested that the double mutant had a low level of V-ATPase activity in vivo, limiting growth in alkaline medium.
We interpret these results as suggesting that the double mutant was necessary for strong resistance to concanamycin. The vma3-T32I mutation increased the resistance of the V-ATPase to bafilomycin and concanamycin ϳ10and 3-fold, respectively. Interestingly, the vma3-I54F mutant increased the resistance to bafilomycin only 1.7-fold but to concanamycin 3.0-fold, the first observation of a single mutant that conferred greater resistance to concanamycin than to bafilomycin. We could not directly answer the question of whether the double mutant had high resistance to the inhibitors because the V-ATPase either had intrinsically low activity or was highly unstable during the assay.

DISCUSSION
An analysis of four bafilomycin-resistant mutants of N. crassa had previously led us to propose that subunit c of the V-ATPase forms at least part of a highly conserved binding site for antibiotics. Unexpectedly, one of the mutations, F136L, conferred no resistance to concanamycin, and the other three had only 3-fold resistance compared with the wild type (20). Bafilomycin and concanamycin are very similar in structure, and both inhibit all V-ATPases tested with similar potency (12). The most significant difference is that the macrolide ring of bafilomycin has 15 carbons versus 17 in concanamycin. In the current study we isolated N. crassa strains with V-ATPase resistant to concanamycin by inducing second mutations in a bafilomycin-resistant strain. These mutations were all in subunit c, in the same general region as the mutations conferring strong resistance to bafilomycin. Thus, the two drugs appear to bind to the same part of the V-ATPase, but the importance of individual amino acids in the binding is different. Three of the strains with double mutations are even more resistant to bafilomycin, suggesting that essentially all residues that affect binding of concanamycin also affect bafilomycin binding; the converse is not true. The data are consistent with the idea that more amino acids participate in binding concanamycin and that mutations in single residues have smaller effects on the binding of concanamycin. Selection of more double mutant strains might be productive in identifying residues involved in binding concanamycin.
To determine the phenotype of each individual mutation in one of the doubly mutated strains (vma3-T23I/I54F) we changed specific residues in the c subunit of S. cerevisiae. Similar to the results obtained with N. crassa, the single mutation vma3-T32I significantly raised the K i for bafilomycin (11-18-fold) and moderately increased the K i for concanamycin (3-fold). The single mutation vma3-I54F raised the K i 2-fold for bafilomycin and 3-fold for concanamycin. In contrast to the phenotype of the comparable strain of N. crassa, the S. cerevisiae strain with both mutations grew very slowly at pH 7.5. A preparation of vacuolar membranes from this strain had insufficient V-ATPase activity to assess the resistance of the enzyme to inhibitors. The direct comparison of N. crassa and S. cerevisiae strains with mutations in the V-ATPase indicated that S. cerevisiae was less tolerant of several of the mutations. Strains of S. cerevisiae with the vma3-F136L or vma3-Y143N mutations did not assemble a functional V-ATPase, whereas the comparable strains of N. crassa had at least 50% of the wild type level of V-ATPase when assayed in vitro.
We used S. cerevisiae to test the effect of mutating the conserved residues in subunit cЈ and cЉ that were in the same positions as sites of mutation in subunit c. The strains of S. cerevisiae with mutations in subunits cЈ or cЉ did make functional V-ATPase as assessed by growth on alkaline medium. None of these strains exhibited resistance to bafilomycin or concanamycin. It should be noted that each of these strains had wild type c subunits, presumably with high affinity drug binding sites. Therefore, the results suggest that if a drug FIG. 5. Mutations in subunit c of S. cerevisiae confer resistance to bafilomycin and concanamycin in vitro. The effect of bafilomycin and concanamycin on V-ATPase activity in vacuolar membranes from S. cerevisiae was assayed at 37°C. Strains carrying either of the two mutations in Vma3p, T32I (LGY116) or I54F (LGY117), were tested. The double mutant T32I/I54F (LGY118) had too little V-ATPase activity to measure. Specific activities in the absence of inhibitor were 0.90 mol/min/mg (SF838 -1D␣ wild type), 0.48 mol/min/mg (T32I), and 0.39 mol/min/mg (I54F). A, half-maximal inhibition by bafilomycin was 3.8 nM for the wild type, 42 nM for the T32I mutant, and 6.4 for the I54F mutant. B, half-maximal inhibition by concanamycin was 2.6 nM for the wild type and 7.7 nM for both mutants.
binding site is present on subunits cЈ and cЉ or at the interface between these subunits and subunit c, it likely has a lower affinity than the sites on subunit c alone. To better assess the roles of subunits cЈ and cЉ it will be necessary to examine strains with mutations in all of the "c-type" subunits.
We have identified nine residues that might be part of the antibiotic binding site. None of the mutations caused a radical change in the chemical nature of the residue. In six mutations one hydrophobic side chain was swapped for another. This was to be expected because the experiments selected strains that retained a functional V-ATPase. In constructing a model of the bafilomycin/concanamycin binding site we assumed that the conformation of the c subunit was not dramatically altered and that most of the mutated residues were in the region where the drug binds. Because of the sequence similarity with subunit c of the F-ATPase, we also assumed that subunit c forms four transmembrane helices (Fig. 6).
For the F-ATPase, structures of subunit c have been determined using both NMR of the polypeptide in solvent (40,41) and x-ray diffraction of protein crystals (42). These results show most of the polypeptide as ␣ helix, including regions extending beyond the lipid bilayer. In the V-ATPase four or five c subunits plus the cЈ and cЉ subunits form the cylindrical rotor embedded in the membrane. Fig. 7 shows a model for the arrangement of individual helices of the c subunits, with helices 1 and 3 forming an interior ring and helices 2 and 4 on the exterior. The model is consistent with the crystal structure reported for the F-ATPase. Fig. 8 models the interface between two c subunits. Helices 1 and 3 would have to fit into a smaller space. Interestingly, these helices appear to have "flat" sides, i.e. regions with small side chains. In helix 3 four alanine residues cluster on one side, and five glycines cluster on the other. Helix 1 also has a cluster of glycine residues (Gly-27, Gly-31, Gly-38). Helix 4, with the putative proton-binding residue, Glu-138, is on the exterior where it can interact with the a subunit. Data from NMR determinations led to the hypothesis that the comparable helix in the F-ATPase undergoes a major conformational change, swiveling 120 degrees during proton binding and release (41,43).
Most of the residues implicated in binding bafilomycin/concanamycin can be clustered closely together in this model (Fig.  8). Leu-132, Phe-136, Leu-140 and Tyr-143 are clustered on 1 face of helix 4, opposite the proton binding residue Glu-138. Thr-32 and Ile-39 lie next to each other on helix 1, and Val-55 can be positioned in the same region. In the model bafilomycin and concanamycin bind in a pocket on the cytosolic half of the membrane formed by helices 1, 2, and 4 of subunit c. Two of the mutated residues, Ile-54 and Met-130, are farther away from the others in this model. Perhaps these residues are not in direct contact with bafilomycin or concanamycin, but the mutations cause a conformational change in the proposed bafilomycin/concanamycin binding pocket. Alternatively, the side For each of the transmembrane helices, the position of 18 residues is predicted, assuming an ␣-helical structure. A large arrow is placed by each residue mutated in bfr and bcr strains. The helices have been positioned to cluster these residues. chains may not be precisely in the position predicted by the ␣-helical model. Ile-54 lies next to Val-55 in the linear sequence, and Met-130 is near Leu-132. Thus, the side chains from these residues could be part of the binding site. Both Ile-54 and Met-130 may also be at the lipid interface (Fig. 6). We intend to use site-directed mutagenesis to test the accuracy of the model.
Harrison and co-workers (44 -46) proposed a model for the structure of c subunits in the V-ATPase based on data from cysteine replacement mutagenesis and analysis of lipid accessible sites. Their model differs significantly from the structure of yeast Fo (42). Helix 1 forms the inner core of Vo, whereas helix 4 with the proton binding site is positioned on the exterior. Helices 2 and 3 lie between helices 1 and 4. It is hard to fit the data from the bafilomycin/concanamycin mutations to this model because helices 1 and 4 are proposed to be distant from each other, with other helices inserted between them.
Characterization of the first three bafilomycin-resistant mutants led us to propose that the antibiotic inhibited the enzyme by preventing rotation of the c subunits. The positions of the new mutations support this hypothesis, suggesting that bafilomycin and concanamycin act like "a stone in the gears" of the rotary motor. The data strongly suggest that the antibiotics intercalate between helices of the c subunits. The stone could prevent the rotation of the cylinder of c subunits relative to the a subunit and/or block the large conformational change in helix 4 believed to be coupled to the rotation. Helix 4 is the most highly conserved part of subunit c, probably because it has a central role in proton transport. By targeting this region of the V-ATPase, organisms have evolved potent toxins, bafilomycin and concanamycin, which inhibit a vital enzyme in a broad range of organisms.