Effects of Human a3 and a4 Mutations That Result in Osteopetrosis and Distal Renal Tubular Acidosis on Yeast V-ATPase Expression and Activity*

V-ATPases are multimeric proton pumps. The 100-kDa “a” subunit is encoded by four isoforms (a1–a4) in mammals and two (Vph1p and Stv1p) in yeast. a3 is enriched in osteoclasts and is essential for bone resorption, whereas a4 is expressed in the distal nephron and acidifies urine. Mutations in human a3 and a4 result in osteopetrosis and distal renal tubular acidosis, respectively. Human a3 (G405R and R444L) and a4 (P524L and G820R) mutations were recreated in the yeast ortholog Vph1p, a3 (G424R and R462L), and a4 (W520L and G812R). Mutations in a3 resulted in wild type vacuolar acidification and growth on media containing 4 mm ZnCl2, 200 mm CaCl2, or buffered to pH 7.5 with V-ATPase hydrolytic and pumping activity decreased by 30–35%. Immunoblots confirmed wild type levels for V-ATPase a, A, and B subunits on vacuolar membranes. a4 G812R resulted in defective growth on selective media with V-ATPase hydrolytic and pumping activity decreased by 83–85% yet with wild type levels of a, A, and B subunits on vacuolar membranes. The a4 W520L mutation had defective growth on selective media with no detectable V-ATPase activity and reduced expression of a, A, and B subunits. The a4 W520L mutation phenotypes were dominant negative, as overexpression of wild type yeast a isoforms, Vph1p, or Stv1p, did not restore growth. However, deletion of endoplasmic reticulum assembly factors (Vma12p, Vma21p, and Vma22p) partially restored a and B expression. That a4 W520L affects both Vo and V1 subunits is a unique phenotype for any V-ATPase subunit mutation and supports the concerted pathway for V-ATPase assembly in vivo.

of ATP hydrolysis to proton transport across membranes. Intracellular V-ATPases are found in compartments such as clathrin-coated vesicles, Golgi, endosomes, lysosomes, secretory vesicles, and the central vacuoles of yeast as reviewed previously (1). V-ATPases are also present in the plasma membranes of specialized cells such as osteoclasts, renal intercalated cells, spermatids, neutrophils, and macrophages, where they function in such processes as bone resorption, renal acidification, pH homeostasis, and coupled transport (2)(3)(4)(5)(6)(7).
V-ATPases are complexes composed of at least 13 different subunits. These subunits are organized into two domains, a cytoplasmic V 1 domain that hydrolyzes ATP and an integral membrane V o domain that translocates protons across membranes. The V 1 is composed of eight subunits, A-H, with three copies of the nucleotide binding subunits A and B, and possibly two copies of subunit E, and two copies of subunit G (8,9). The yeast V o is composed of six different subunits, a, c, cЈ, cЉ, d, and e, with four copies of subunit c (10 -13). Subunits a, c, cЈ, and cЉ are thought to be responsible for proton translocation, but the functions of subunits d (14) and e (10) are unknown. Proton translocation through V o is driven by rotational catalysis of V 1 (15).
Yeast V-ATPases fail to assemble when any of the genes that encode subunits are deleted, except for subunits H and cЉ (16,17). Previous studies have revealed that deletion of any V o subunit, except for the cЉ subunit (16), results in the loss of all V o subunits from the vacuole (18). The loss of any V 1 subunit, with the exception of subunit H (17), leads to an absence of all V 1 subunits at the vacuole (18). Without subunit H, the assembled V-ATPase is not active (17,19), and loss of the cЉ subunit results in uncoupling of enzymatic activity (16).
V o domain assembly depends on the presence of three assembly factors, Vma12p, Vma21p, and Vma22p (20 -23). Phenotypes of cells that lack Vma12p, Vma21p, or Vma22p resemble those lacking a V o structural subunit, in that V o does not assemble properly, V o subunits are not found at the vacuole, and Vph1p is rapidly degraded (22,24). Two of these proteins, Vma12p and Vma22p, form a complex that binds transiently to Vph1p and aids in Vph1p assembly and maturation (21-23, 25, 26). A recent study has shown that Vma21p is required for the assembly of the V o domain (27).
An important question is how V-ATPase can function in the wide variety of locations and physiological processes listed above. In yeast, V-ATPases have two organelle-specific isoforms of the "a" subunit, Vph1p and Stv1p (28). Vph1p localizes to the vacuole, whereas Stv1p localizes to the Golgi (28,29). In humans, four different a isoforms have been identified, a1 to a4, which are expressed in a tissue-and organelle-specific manner (30,31). The V 1 domain also contains tissue-specific subunit isoforms, including B, C, E, and G (2,(31)(32)(33). Mutations in the a3, a4, and B1 subunits are known to lead to human disease (32, 34 -37). There are no known human diseases associated with mutations in a1 or a2. The a3 isoform is highly enriched in osteoclasts (38), and mutations within a3 result in autosomal recessive osteopetroses (35)(36)(37)39). The a4 isoform is specific to renal intercalated cells of the kidney (31), and mutations within the a4 gene lead to distal renal tubule acidosis (dRTA) (5,34,40).
The bone disease osteopetrosis demonstrates the essential function of V-ATPases in bone resorption. The osteopetroses are a group of heritable conditions characterized by defects in osteoclast bone resorption. Osteoclasts resorb bone by first creating a sealed zone, the resorption lacuna, between the osteoclast ruffled border membrane and the bone surface. The process of V-ATPase-mediated proton transport into resorption lacunae (41) is an essential component of bone remodeling.
The importance of V-ATPase in renal proton secretion is highlighted by inherited dRTA. V-ATPases in the plasma membranes of renal intercalated cells of the distal nephron pump protons from the blood into the urine, an essential step in the removal of metabolic acid. dRTA is characterized by impaired renal acid secretion, resulting in metabolic acidosis. Autosomal recessive dRTA involves mutations to the a4 or the B1 subunit isoforms of V-ATPase. Mutations within the V-ATPase B1 and a4 genes in some cases also result in sensorineural deafness (34,42).
In humans, 26 mutations have been identified in a3 that result in autosomal recessive osteopetrosis (36,37,43,44), and 21 mutations have been identified in a4 that result in dRTA (5,40). The majority of these mutations result in frameshifts, abnormal splicing, and insertion of stop codons and as such are uninformative, except to further illustrate the essential roles of a3 and a4. However, missense and small deletion mutations were identified that could pinpoint critical domains. Two missense mutations, G405R and R444L, were identified in the V-ATPase a3 subunit isoform that account for all the defects in nine unrelated families in Costa Rica (36,45). We have recreated these a3 mutations in the yeast a subunit ortholog, Vph1p, as G424R and R462L. Of the 21 a4 mutations, only three were missense mutations. We recreated two of them, P524L and G820R, in Vph1p as W520L and G812R. Characterizing the effect of these missense mutations can identify critical domains within a3 and a4 that are essential for assembly, targeting, or retention and retrieval of V-ATPases to and from the plasma membrane. Technically, it is difficult to characterize these mutations in humans. The only biochemical information is from fibroblast and lymphoblast cell lines from patients with frameshift and abnormal splicing mutations, and those studies confirm the expected null phenotype (37,44).
Yeast V-ATPases are an attractive model for the study of the biochemistry of a3 and a4 mutations because the a subunit is remarkably conserved across species. The subunit sequences of human a3 and a4 isoforms have ϳ55% similarity to the yeast ortholog, Vph1p. Characterizing the mechanistic outcome of a3 and a4 mutations in yeast could reveal critical amino acids involved in V-ATPase assembly, targeting, or activity. Here we have recreated four missense mutations, two from a3 and two from a4 mutations in the yeast Vph1p subunit, and report on their respective V-ATPase assembly and activity phenotypes.

EXPERIMENTAL PROCEDURES
Materials-Escherichia coli and yeast culture media were purchased from Difco. General chemicals and protease inhibitors were purchased from Sigma. Restriction endonucleases, T4 DNA ligase, and other molecular biology reagents were from Fermentas Life Sciences (Burlington, Canada). Zymolyase 100T was obtained from Seikagaka Corp. (Rockville, MD). The monoclonal antibodies, 8B1-F3 against the yeast V-ATPase 69-kDa A subunit, 13D-11 against the yeast V-ATPase 60-kDa B subunit, and 10D7 against the 100-kDa a subunit, were purchased from Molecular Probes, Inc. (Eugene, OR). A polyclonal serum against Vma12p was the kind gift from Dr. Tom H. Stevens (University of Oregon), and a serum against Vma22p was the kind gift from Dr. Antony A. Cooper (University of Missouri). Standard YPD medium was formulated as 20 g of Difco peptone, 10 g of yeast extract, and 20 g of D-glucose/ liter, with the pH adjusted to 5.8.
Strains and Plasmids-For strains and plasmids, see Table 1.
For VPH1 mutation G812R, first round PCR primers were MO30 and MO155 with MO154 and T7. Primers MO30 and T7 were used for second round PCR. The PCR product was cut with EcoRI and NotI and cloned into pRS316 from MM322 cut with EcoRI and NotI. For VPH1 mutation R462L, first round PCR primers were MO23 and MO157 with T7 and MO156. Primers MO23 and T7 were used for second round PCR. The PCR product was cut with SacI and cloned into pRS316 from MM623 cut with SacI. For VPH1 mutation W520L, first round PCR primers were MO23 and MO161 with T7 and MO160. Primers MO23 and T7 were used for second round PCR. The PCR product was cut with SacI and cloned into pRS316 from MM623 cut with SacI. For VPH1 mutation G424R, first round PCR primers were MO23 and MO159 with T7 and MO158. Primers MO23 and T7 were used for second round PCR. The PCR product was cut with SacI and cloned into pRS316 from MM623 cut with SacI. DNA sequencing confirmed the mutations in these plasmids.
Vph1p_W520L was overexpressed by inserting the SalI-NotI fragment of Vph1p_W520L into pRS424 cut with SalI and NotI. Inserting the SalI-NotI fragment of MM623 pRS316 ϩ Vph1p into pRS424 and pRS426 cut with SalI and NotI created overexpressed VPH1. STV1 (similar to VPH1Ј (a homolog of VPH1)) was overexpressed using MM273 pRS426.
Preparation of Whole Cell Extracts-Yeast whole cell extracts were prepared by growing yeast cells to stationary phase, washing ϳ5 ϫ 10 9 cells twice with water, and chilling the pellet for 30 min on ice. Just before homogenization the pellet was resuspended in 1 ml of 100 mM Tris-HCl, pH 7.5, containing 1 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, and 1 mg/ml pepstatin. Eppendorf tubes (2 ml) were filled with zirconiumcoated 0.5-mm diameter glass beads to four-fifths of the final volume and chilled to Ϫ80°C. Resuspended cells were added to the chilled beads and then placed in a bead-breaker (Bio101 Savant FastPrep FP120) set at 5.5 speed for 30 s, in a 4°C cold room. Cell homogenates were centrifuged in a Beckman GS-6R centrifuge at 15,000 ϫ g for 10 min at 4°C to remove whole cell debris. The supernatants (yeast whole cell extract) were denatured for SDS-PAGE in 63 mM Tris-HCl, pH 6.8, containing 1% (w/v) SDS, 0.6 mM 2-mercaptoethanol, and 5% (w/v) glycerol for 3 min at 70°C.
Recreation of Human a3 and a4 Mutations in the Yeast Ortholog, Vph1p-See Fig. 1

for information.
Scoring Growth Phenotypes-Exponentially growing cells were suspended at 5 ϫ 10 7 cells/ml in 1 well of a standard 96-well plate. Five serial dilutions were made from the original stock into adjacent wells of the 96-well plate such that the final well held 1 ϫ 10 5 cells/ml with each well containing 130 l. A 48-prong inoculating manifold (Dan-Kar Corp., Reading, MA) was used to transfer cells to solid medium. Each prong of the inoculating manifold deposits ϳ10 l; thus the cell number per spot in the serial dilution ranged from ϳ5 ϫ 10 5 to 1 ϫ 10 3 . Growth phenotypes were scored after incubating the plates at 30°C for 5 days on uracil-minus media supplemented with 200 mM CaCl 2 , 4 mM ZnCl 2 , or buffered to pH 7.5 with 50 mM Tris-MES.

Effects of Human "a" Mutations on V-ATPase Function
enized with 10 strokes of a tightly fitting pestle in a Dounce homogenizer. The homogenate was overlaid onto a 30-ml 10 -30% discontinuous sucrose gradient and centrifuged in the same ultracentrifuge listed above, using an SW-20 (Beckman) rotor at 100,000 ϫ g, 4°C for 2 h. Material at the 10 -30% interface was collected, diluted 10-fold with overlay medium, and centrifuged for 45 min at 100,000 ϫ g at 4°C. The pellets were resuspended in 1 ml of overlay medium and stored at Ϫ80°C until used. ATPase and Proton Translocating Activities-Concanamycin A-sensitive ATPase activities were measured using a coupled spectrophotometric assay (48). ATP-dependent proton

Effects of Human "a" Mutations on V-ATPase Function
translocation was measured as a percentage of the maximum fluorescence quench obtained with wild type vacuoles by using acridine orange in the presence or absence of 150 nM concanamycin A with a Hitachi F-2500 fluorescence spectrophotometer with excitation at 493 nm and emission at 545 nm. ATP was added at a final concentration of 5 mM to start transport (49).
Quinacrine Labeling-Cells grown to exponential phase in YPD or selective medium were washed and then incubated for 5 min in 200 mM quinacrine as described previously (48). The cells were then washed twice and viewed within 15 min of quinacrine labeling.

RESULTS
To investigate the effect of human a3 and a4 mutations that lead to recessive malignant osteopetrosis and dRTA, we recreated four missense mutations as follows: two a3 and two a4 in conserved residues of the yeast ortholog Vph1p. We selected a3 and a4 missense mutations to further characterize the roles of these residues in the basic function of the a subunit in V-ATPases. Recreating these point mutations in yeast Vph1p was possible because of the high degree of conservation between mammalian and fungal a subunits (see Fig. 1). As seen in Fig. 1, the human a3 mutations, G405R and R444L, and the a4 mutation, G820R, all reside within highly conserved regions. Identification of these a3 and a4 mutations was clear, as the glycine and arginine residues are perfectly conserved among all four mammalian a subunit isoforms and the yeast Vph1p. The assignment of the a4 mutation, P524L, was not as apparent because the proline was not conserved within Vph1p. Nevertheless amino acids both upstream and downstream of a4 Pro-524 are perfectly conserved, making our alignment in this region unambiguous and indicating that the a4 Pro-524 is a tryptophan residue in Vph1p. As proline and tryptophan are both nonpolar amino acids, the change (P524L in a4 and W520L in Vph1p) is considered to be conservative.

The a3 Osteopetrosis-inducing Mutations in Yeast (G424R and R462L) Had No Effect on V-ATPase Assembly yet Had
Decreased Activity-Growth of yeast cells on media supplemented with cations, or buffered at a high pH, is dependent upon V-ATPase activity. As seen in Table 2, growth of yeast containing the a3 R462L and G424R mutations ranged from 75 to 100% compared with wild type on selective media supple-mented with 4 mM ZnCl 2 , 200 mM CaCl 2 , or buffered to pH 7.5. Despite the slight growth defect, the yeast vacuoles appear to be acidified as shown by the accumulation of quinacrine within the vacuoles (Fig. 2). Considering that vmaϪ strains can be rescued with mutant V-ATPases that only have 20% wild type activity (50), it was necessary to measure the hydrolytic and proton pumping activity from purified vacuolar vesicles from each mutant. Fig. 3 shows that the G424R mutation retained 74% of ATP-dependent proton pumping and 67% ATPase activity, whereas the R462L mutation demonstrated 64% of proton pumping and 62% of ATPase activity relative to wild type. Furthermore, immunoblots of these membranes reveal that both a3 mutations had wild type levels of the V o subunit Vph1p (a ortholog) and V 1 subunits Vma1p (A ortholog) and Vma2p (B ortholog), indicating that neither mutation affected the assembly or targeting of the V-ATPase complex to the vacuole (Fig.  4). These results suggest that the human a3 Gly-405 and Arg-444 residues are not essential to V-ATPase assembly and FIGURE 2. Effects of the a3 (G424R and R462L) and the a4 (W520L and G812R) mutations on vacuolar acidification. Exponentially growing cells transformed with the single copy plasmid pRS316 with wild type (VPH1), without wild type (⌬vph1), or VPH1 bearing the a3 mutations, G424R and R462L, and the a4 mutations, W520L and G812R, were washed and incubated for 5 min in 200 mM quinacrine, washed again, and photographed as described under "Experimental Procedures." A 300-ms exposure was used for all images. The images shown are representative of three separate assays for uptake of quinacrine.

TABLE 2 Effects of a3 (G424R and R462L) and a4 (W520L and G812R) mutations on V-ATPase activity as assayed by growth on selective media
Exponentially growing cells transformed with the single copy plasmid pRS316, with wild type VPH1 (VPH1), without wild type (⌬vph1), or with VPH1 bearing the a3 mutations G424R and R462L, or the a4 mutations, W520L and G812R, were assayed for growth on uracil-minus plates with or without 200 mM CaCl 2 , 4 mM ZnCl 2 , or buffered to pH 7.5, and incubated as described under "Experimental Procedures." Three separate growth assays were conducted that gave similar results. Growth was scored relative to the VPH1 strain as follows: ϩϩϩϩ, 100 to 75% growth relative to VPH1; ϩϩϩ, 75 to 50% growth relative to VPH1; ϩϩ, 50 to 25% growth relative to VPH1; ϩ, 25 to 0% growth relative to VPH1; Ϫ, no detectable growth.

Effects of Human "a" Mutations on V-ATPase Function
reduce ATP hydrolytic and proton pumping activity by ϳ30 -35%. The a4 Mutation G812R Affects V-ATPase Activity but Not Assembly or Vacuolar Targeting-The G812R mutation resulted in 25-50% growth compared with wild type on selective media supplemented with 4 mM ZnCl 2 , 200 mM CaCl 2 , or buffered to pH 7.5 ( Table 2). It also resulted in reduced accumulation of quinacrine in the vacuole (Fig. 2) and only 17% of proton pumping and 14% of ATPase activity relative to wild type (Fig. 3). Fig. 4 shows expression of V o subunit Vph1p (a) and V 1 subunits Vma1p (A) and Vma2p (B) on the vacuolar membrane at wild type levels.
These results indicate that the human a4 Gly-820 residue is not required for assembly or targeting of V-ATPases to the vacuole but is critical for V-ATPase activity.
The a4 W520L Mutation Results in Decreased Expression of Both V o and V 1 Subunits-The W520L mutation resulted in no detectable V-ATPase activity as assayed by growth on selective media supplemented with 4 mM ZnCl 2 , 200 mM CaCl 2 , and 25-0% growth on media buffered to pH 7.5 ( Table 2). Furthermore, there was no detectable accumulation of quinacrine in the yeast vacuole (Fig. 2). No measurable proton pumping or ATPase activity was detected on purified vacuolar vesicles (Fig.  3) as well as barely detectable levels of Vph1p, Vma1p, and Vma2p on vacuolar membranes (Fig. 4).
To ask whether the W520L mutation affected vacuolar targeting or complex assembly, whole cell extracts were immunoblotted. Surprisingly, the results for W520L differed not only from ⌬vph1 but from any other published VPH1 mutation; not only was Vph1p expression compromised but the V 1 subunits, Vma1p and Vma2p, were barely detectable, even when 4-fold more protein was immunoblotted (Fig. 5A, 80 g for W520L compared with 20 g for VPH1 and ⌬vph1). Overexpressing VPH1 in the presence of W520L did not rescue Vma1p or Vma2p expression (Fig. 5B), suggesting that the W520L mutation had a dominant effect.
The a4 W520L Mutation Is Dominant-Single or multicopy plasmids bearing Vph1p_W520L, when expressed in ⌬vph1 or wild type (VPH1) yeast strains, eliminated V-ATPase activity as assayed by growth on 4 mM ZnCl 2 (Table 3). Previously, we showed that overexpressing one "a" isoform (Stv1p) could compensate for the absence of the other (Vph1p) (28). To this end, we asked whether we could rescue the W520L phenotype by overexpressing Vph1p or Stv1p.  . Effects of a3 (G424R and R462L) and a4 (W520L and G812R) mutations on concanamycin A-sensitive ATP hydrolytic activity and ATP-dependent proton translocation. Vacuolar membrane vesicles were isolated from ⌬vph1 yeast strain (MM53) transformed with single copy plasmid pRS316 with wild type (VPH1), without wild type (⌬vph1), or VPH1 bearing a3 mutations G424R and R462L, and a4 mutations G812R and W520L. Concanamycin A-sensitive ATP hydrolytic activity and proton transport were measured using 10-g aliquots of vesicles as described under "Experimental Procedures." The results are combined from two individual membrane purifications performed for each strain, with ATP hydrolytic activity and proton transport performed in duplicate. Wild-type activity was defined as 100%. The specific activity of vesicles from yeast expressing the wild type VPH1 plasmid was 3.7 mol of ATP/min/mg protein with ϳ80% inhibition of hydrolytic activity and 100% of proton translocation with 150 nM of concanamycin A.  assembly in the endoplasmic reticulum (ER). To test this, we asked whether the ER V-ATPase assembly factors Vma12p, Vma21p, and Vma22p were affected by the Vph1p_W520L mutation or, alternatively, whether they could affect the Vph1p_W520L phenotype. Immunoblots of whole cell extracts obtained from wild type, ⌬vph1, and Vph1p_W520L transformed into a ⌬vph1 strain all had similar levels of Vma12p and Vma22p (Fig. 6), indicating that the W520L mutation did not affect the stability of these ER assembly factors. We next asked whether deletions of the assembly factors could affect the Vph1p_W520L phenotype.

Dominant W520L Phenotypes Are Partially Rescued by Disrupting the ER V-ATPase Assembly Factors Vma12p, Vma21p, and
Vma22p-Yeast lacking assembly proteins Vma12p, Vma21p, or Vma22p have phenotypes similar to those seen in yeast missing V o subunits. In both cases, the V o does not assemble; the Vph1p is degraded in the ER, and V 1 subunits are unaffected (22,24,51). We speculated that in the absence of the assembly factors, Vph1p_W520L would be unable to associate with V 1 subunits and thus degradation of the Vma1p (A) and Vma2p (B) subunits would be attenuated. As hypothesized, expressing Vph1p_W520L in strains bearing full disruptions of VMA12 and VMA22 partially restored expression levels of Vma2p (B) (Fig. 7A). Interestingly, the stability of Vph1p_W520L was also restored to levels similar to those expected in ⌬vma12 or ⌬vma22 strains. Expressing Vph1p_W520L in ⌬vma21 also partially restored Vph1p_W520L expression but was consistently not as effective as ⌬vma12 or ⌬vma22 (Fig. 7B). This result is in agreement with the model where the V-ATPase complex interacts with Vma21p downstream of the Vma12p and Vma22p interactions (27).

DISCUSSION
Mutations in human a3 and a4 result in osteopetrosis and dRTA, respectively. We hypothesize that these human mutations are highlighting critical amino acids within the "a" subunits that are essential for V-ATPase assembly, activity, or targeting. The assignment for three of the four mutations between the human "a" subunits and the yeast ortholog Vph1p was straightforward (Fig. 1). Assigning human Pro-524 to yeast Trp-520 was speculative. Although the residue itself was not perfectly conserved, the alignment in this region appears unambiguous due to identity both upstream and downstream of this residue. This led us to believe that the mammalian a4 proline residue was conservatively altered to tryptophan, another nonpolar residue, in the yeast Vph1p. Although assigning a4 Pro-524 to Vph1p Trp-520 is controversial, the Vph1p_W520L mutation resulted in the most interesting phenotype.
The Human a3 Mutations G405R and R444L Are Not Critical for Folding, Targeting, or Assembly of Yeast V-ATPases-In humans, the osteopetrotic G405R and R444L point mutations are phenotypically identical to full a3 deletions (35,52). Also, the human Gly-405 and Arg-444 residues are perfectly conserved among canine, bovine, murine, and yeast a subunits, further suggesting a critical role for these residues. Considering all this, we expected these missense mutations, when recreated in yeast, to have a profound phenotype. Unexpectedly, recreating the human a3 mutations in yeast resulted in only a slight impairment of growth on selective media. V-ATPase assembly on the vacuolar membrane and vacuolar acidification were not noticeably affected, although proton pumping and ATP hydrolysis were reduced by 30 -35%.
Considering that Gly-405 and Arg-444 are critical in human osteoclasts but only slightly affect activity in yeast, one could conclude that yeast is not a good model for studying human mutations. Alternatively, one could hypothesize that the difference between the yeast phenotype and the null phenotype in humans is highlighting an essential osteoclast function for  VPH1, ⌬vph1, ⌬vma12, ⌬vma22, and Vph1p_W520L transformed into ⌬vph1 were separated by SDS-PAGE and immunoblotted with polyclonal sera against Vma12p and Vma22p. Three independent immunoblots revealed similar results.

TABLE 3
The Vph1p_W520L mutation is dominant ⌬vph1 (MM57) (top) and wild type (MM51) (bottom) yeast strains were transformed with the single copy (pRS3 series, CEN-ARS) or multicopy (pRS4 series, 2 m) plasmids containing Vph1p_W520L, Vph1p, or Stv1p and assayed for growth on YPD and YPD supplemented with 4 mM ZnCl 2 as described under "Experimental Procedures." The growth assays were conducted three times, all with similar results. Growth was scored relative to the VPH1 strain as follows: ϩϩϩ, equivalent to wild type; ϩϩ, Ͼ75% wild type; ϩ, Ͻ25% wild type; Ϫ, no growth.

Effects of Human "a" Mutations on V-ATPase Function
these two residues distinct from V-ATPase assembly or activity. For example, these point mutations may affect targeting to and from the osteoclast-ruffled border during the resorptive cycle. A mammalian trafficking defect would likely not affect Vph1p targeting to the yeast vacuole, particularly because the default pathway for integral membrane proteins in yeast is to the vacuolar membrane (53). Although the a3 isoform has been shown to be highly enriched in the osteoclast plasma membrane (38), it is currently thought that an actin-binding domain in the V 1 B2 subunit (54) controls recycling from the osteoclast-ruffled border. As we have shown that the 50-kDa N-terminal domain of the a subunit interacts with V 1 subunits (55), one can speculate that conformational changes in a could affect the availability of the actin-binding sites in B2. Further elucidation of mechanisms involving these a3 mutations clearly requires a mammalian osteoclast culture system. Specific to yeast, the results here suggest that Gly-424 is not critical for assembly. Nevertheless, another study has shown that the adjacent residue Gly-425, when mutated to an asparagine, did in fact reduce assembly (47).
The Human a4 Mutation G820R Is Required for Activity of V-ATPases-In contrast to the recreated a3 mutations, recreating the human dRTA a4 mutation, G820R (G812R in yeast Vph1p), resulted in severely defective growth on selective media and barely detectable vacuolar acidification with 17% of ATP-dependent proton pumping and 13% ATPase activity relative to wild type. Nevertheless, there were wild type expression levels of V-ATPase subunits Vph1p (a), Vma1p (A), and Vma2p (B) on the vacuolar membrane, indicating that folding, assembly, and targeting were unimpaired. These results suggest that Gly-820 is a key residue for activity of all V-ATPases. Gly-812 is found within the C-terminal tail domain of Vph1p, a region for which orientation within the vacuolar lumen (57) or cytoplasm (23,58) is still in dispute. Previous studies have demonstrated that the C-terminal region is essential for V-ATPase assembly (59). Our present results, together with the results of Leng et al. (59), suggest that the C-terminal tail domain of the "a" subunit has an essential function in V-ATPase assembly and activity.
The Human a4 Recessive Mutation P524L, When Recreated in Yeast, Results in a Dominant Mutation That Affects the Stability of Both V o and V 1 Subunits-The P524L mutation, when recreated in yeast Vph1p as W520L, affects activity, stability, and expression of all V-ATPase subunits. Vph1p bearing the W520L mutation was similar to ⌬vph1 with respect to inability to grow on selective plates, absence of vacuolar acidification, no detectable V-ATPase activity, and absence of V 1 subunits on the vacuolar membrane. Surprisingly, when subunit levels in whole cell extracts were examined, Vph1p_W520L had a more deleterious effect than the complete absence of Vph1p (⌬vph1). Not only was the expression of Vph1p_W520L reduced but also expression of the two V 1 subunits, Vma1p (A) and Vma2p (B), was attenuated. This is the first observed instance of a mutation within a V o subunit affecting the stability of a V 1 subunit, and it challenges the theory that the two complexes assemble independently in vivo.
To test if overexpression of one "a" isoform could complement mutations to the other "a" isoform, with respect to W520L, we asked whether single copy expression or overexpression of Vph1p or Stv1p could complement the Vph1p_W520L phenotype. We found that when Vph1p_W520L was expressed from a single copy plasmid in a wild type strain, it was necessary for Vph1p or Stv1p to be overexpressed to achieve even partial complementation with respect to growth on selective media. These results illustrate the dominant nature of the W520L mutation. This is in contrast to the recessive nature of the human a4 P524L mutation (5), suggesting that the biochemical basis for the resulting phenotype may differ between yeast and human.
The Dominant Negative Vph1p_W520L Phenotypes Are Partially Rescued in the Absence of the ER Assembly Factors, Vma12p, Vma21p, and Vma22p-A common reason for dominant negative phenotypes for mutant proteins present in a single copy in a complex (such as the a subunit) is that the mutant protein has a higher affinity for a crucial interacting factor and inappropriately sequesters it from wild type proteins (60). The dominant negative nature of Vph1p_W520L could result from its sequestering the V-ATPase assembly proteins Vma12p, Vma21p, and Vma22p. To this end, we asked whether Vph1p_W520L affected the ER V-ATPase assembly factors or, alternatively, whether they could affect the Vph1p_W520L phenotype.
Immunoblots of whole cell extracts revealed wild type levels of both Vma12p and Vma22p in the presence or absence of the Vph1p_W520L mutation (Fig. 6), indicating that this mutation does not affect the steady state stability of these two V-ATPase FIGURE 7. The W520L phenotype is partially rescued by disrupting the genes encoding the ER V-ATPase assembly factors Vma12p, Vma21p, and Vma22p. A, whole cell extracts of 40 g of protein for Vph1p_W520L transformed into ⌬vph1 (W520L in ⌬vph1), ⌬vma12 (W520L in ⌬vma12), and ⌬vma22 (W520L in ⌬vma22) and 20 g from VPH1 transformed into ⌬vph1 (VPH1) were separated on 12% polyacrylamide gels and immunoblotted with monoclonal antibodies against Vph1p (10D7), Vma2p (13D-11), and polyclonal sera against Vma12p. B, whole cell extracts (40 g for Vph1p_W520L transformed into ⌬vph1 (W520L in ⌬vph1), ⌬vma21 (W520L in ⌬vma21), and ⌬vma22 (W520L in ⌬vma22), and 20 g from VPH1 transformed into ⌬vph1 (VPH1)) were separated on 12% polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with monoclonal antibodies against Vph1p (10D7), Vma2p (13D-11), and polyclonal sera against Vma22p. Three independent immunoblots revealed similar results. assembly factors. In contrast, deletion of either Vma12p or Vma22p partially restored Vph1p_W520L and Vma2p expression (Fig. 7). The similar results arising from an absence of Vma12p and Vma22p (⌬vma12 and ⌬vma22) are consistent with these assembly factors forming a single complex in the ER that transiently binds to Vph1p (with a half-life of ϳ5 min) to aid in its assembly and maturation (25). We speculate that the dominant negative phenotypes arising from the Vph1p_W520L mutation result from aberrant, possibly inappropriately long interactions with the Vma12p-Vma22p complex. A tight association between Vph1p_W520L and Vma12p/Vma22p could exclude wild type Vph1p or Stv1p, explaining their inability to complement, even when overexpressed (Table 3 and Fig. 5).
Deleting Vma21p also partially restored Vph1p_W520L and Vma2p expression, but not as effectively as ⌬vma12 and ⌬vma22, with respect to Vph1p_W520L expression itself (Fig.  7). This is in agreement with a model where Vma21p acts downstream of Vma12p and Vma22p by escorting the V o complex from the ER to the Golgi (27).
Vph1p_W520L Provides Evidence for the Concerted Pathway for V-ATPase Assembly-The precise mechanisms by which V-ATPases assemble are still controversial with evidence suggesting two different possibilities. Mutational analysis and in vitro assays have shown that preassembled V o and V 1 domains can combine to form one complex in a process called independent assembly. Support for independent assembly includes the findings that the assembled V o domain can be found at the vacuole in the absence of the V 1 domain, whereas free V 1 domains can be found in the cytoplasm and not at the vacuole (61). In contrast, in vivo pulse-chase experiments have revealed early interactions between V o and V 1 subunits, specifically Vph1p (a) and Vma1p (A), suggesting that subunits are added in a stepwise fashion to form a single complex in a concerted assembly process (56).
Previously, all deletions and mutations of V o subunits did not affect V 1 subunit expression, stability, or assembly, supporting the independent pathway model reviewed in Ref. 1. The Vph1p_W520L phenotype is unique in that it affects the expression of V 1 subunits Vma1p (A) and Vma2p (B). We speculated that the W520L mutation affects how Vph1p_W520L interacts with the Vma12p-Vma22p complex within the ER, possibly by inappropriately extending the interaction. The degradation of Vma1p (A) and Vma2p (B) suggests that these V 1 subunits also associate with this early ER complex and, if true, would be further evidence of a concerted pathway.
In summary, the osteopetrotic a3 mutations, when recreated in yeast, retain 65-70% of wild type activity with wild type expression of V-ATPase subunits. In contrast, the human a4 dRTA mutation G820R, when recreated in yeast, resulted in an inactive but assembled complex within the vacuole membrane suggesting that this conserved residue is essential for activity, but not assembly, of all V-ATPases. The human a4 dRTA P524L mutation in yeast resulted in a dominant negative phenotype with degradation of a, A, and B subunits. Deletion of the ER assembly factors partially rescued this phenotype, whereas overexpressing yeast a isoforms Vph1p or Stv1p were ineffective. We hypothesize that this a4 mutation results in prolonged interactions with the ER assembly factors, sequestering these assembly factors from wild type a subunits and leading to the degradation of V 1 subunits. To test this hypothesis, we plan to delineate precisely when and where Vph1p_W520L, Vma1p (A) and Vma2p (B) are being degraded, through pulse-chase experiments and by using various sec mutants blocked in ER-Golgi transport.