Molecular genetics of the yeast vacuolar H⁺-ATPase

The fungal vacuole is an acidic compartment which plays essential roles in metabolic storage and in cytosolic ion and pH homeostasis. In addition, it functions in endolytic macromolecular degradation in a manner similar to that occurring in phagocytotic animal lysosomes (Figs 1 and 2: Anraku, 1987a,b;Anraku et al. 1989; Klionsky et al. 1990). During the last 10 years, it has become known that a new, distinct class of H⁺-pumping ATPase, a V-ATPase, exists ubiquitously in vacuo-lysosomal and endomembranous organelles, including fungal and plant vacuoles, animal lysosomes, coated vesicles, Golgi bodies, chromaffin granules and synaptic membrane vesicles (for recent reviews, see Anraku et al. 1989, 1991a, 1992; Forgac, 1989; Nelson and Taiz, 1989; Stone et al. 1989). The V-ATPase is also present in plasma membranes of vertebrate renal tissues (Gluck, 1992), osteoclasts (Chatterjee et al. 1992) and insect gastrointestinal and sensory epithelia (Harvey, 1992; Klein, 1992; Wieczorek, 1992).

Biochemical studies of yeast vacuoles originated with the work of Ohsumi and Anraku (1981), who established a simple method for separating intact vacuoles of high purity from Saccharomyces cerevisiae. Kakinuma et al. (1981) found that a preparation of vacuolar membrane vesicles with a right-side-out orientation had an unmasked Mg²⁺-ATPase activity with an optimal pH of 7.0. The activity was sensitive to dicyclohexylcarbodiimide (DCCD) and was stimulated threefold by the protonophore uncoupler SF6847 and 1.5-fold by the H⁺/K⁺ antiporter nigericin. ATP-hydrolysis-dependent uptake of protons into vacuolar membrane vesicles has been demonstrated directly by the change in quenching of 9-aminoacridine and quinacrine fluorescence (Kakinuma et al. 1981; Ohsumi and Anraku, 1981). The electrochemical potential difference of protons across the vacuolar membrane generated upon ATP hydrolysis was determined to be 180 mV, consisting of a proton gradient of 1.7 pH units, interior acid, and of a membrane potential of 75 mV, interior positive (Kakinuma et al. 1981).

Studies from our laboratory have shown that the vacuolar membrane of yeast is equipped with two distinct Cl⁻ transport systems, each of which contributes to the formation of a chemical gradient of protons across the vacuolar membrane by shunting the membrane potential generated by the H⁺-ATPase (Fig. 3: Anraku et al. 1989, 1992; Wada et al. 1992a). Vacuolar acidification is a prerequisite for operation of amino acid/H⁺antiporters (Ohsumi and Anraku, 1981; Sato et al. 1984a,b), a Ca²⁺/H⁺antiporter (Ohsumi and Anraku, 1983) and a K⁺ channel (Wada et al. 1987; Tanifuji et al. 1988). In cases where this ability to acidify is lost, vacuolar protein transport and nonspecific fluidphase endocytosis are markedly affected (Klionsky et al. 1990; Mellman et al. 1986; Umemotoeia/. 1990; Yamashiro et al. 1990).

Vacuolar H⁺-ATPases are large multimeric enzymes with a functional relative molecular mass (M_r) of about 500×10³ (Bowman et al. 1986; Hirata et al. 1989) and contain at least nine subunits (Adachi et al. 1990; Arai et al. 1988; Bowman et al. 1989; Kane et al. 1989; Moriyama and Futai, 1990; Moriyama and Nelson, 1987a,b; Parry et al. 1989; Xie and Stone, 1986). The enzymes are sensitive to bafilomycin Ai (Bowman et al. 1988b; Umemoto et al. 1990; Yoshimori et al. 1991). The proposed reaction mechanism (Hirata et al. 1989; Uchida et al. 1988) is similar to that for mitochondrial and bacterial FiFo-ATPases (see Futai et al. 1988, 1992).

Taiz and his coworkers first cloned and sequenced a cDNA encoding the carrot 69x10³M_r polypeptide, a catalytic subunit of the enzyme (Zimniak et al. 1988). Then, Bowman et al. (1988a,c) reported isolation and sequencing of two genes from Neurospora crassa, vmal and vma2, which they designated for vacuolar membrane ATPase. These earlier contributions have provided breakthroughs for molecular biological and genetic studies of V-ATPases.

This article addresses genetic and molecular biological views of the yeast vacuolar H⁺-ATPase, emphasizing the manipulation of genetic screening for mutations with defective vacuolar acidification and, hence, of the VMA genes that affect expression of the enzyme activity in Saccharomyces cerevisiae.

Initially, the yeast vacuolar H⁺-ATPase in S. cerevisiae was partially purified and characterized as a three-subunit enzyme (Uchida et al. 1985). Kane et al. (1989) examined the original method of purification more carefully and demonstrated that the fraction with the highest specific activity included eight polypeptides with apparent M_r values of 100, 69, 60, 42, 36, 32, 27 and 17×10³. They also showed that a monoclonal antibody raised against the 69×10³M_r polypeptide immunoprecipitated this eight-subunit enzyme, suggesting that all eight polypeptides are good candidates for being subunits of the enzyme.

Based on information from the peptide and nucleotide sequences of respective subunits and cDNAs encoding the peptides of plant and mammalian counterparts, several yeast VMA genes have been cloned and sequenced. VMA1 (Hirata et al. 1990), VMA2 (Anraku et al. 1991a; Ohya et al. 1991; Yamashiro et al. 1990), VMA3 (Nelson and Nelson, 1989; Umemoto et al. 1990) and VMA5 (Beltrán et al. 1992) were cloned and characterized on the basis of partial amino acid sequences determined with the purified 67, 57, 16 and 42×10³M_r subunits, respectively. The sequence of VMA2 (Nelson et al. 1989) was determined by using a synthetic oligonucleotide derived from the counterpart cDNA (Manolson et al. 1988). VMA4 was accidentally discovered and characterized during a sequence study of MIP1 (Foury, 1990).

Anraku and coworkers (Hirata et al. 1990; Ohya et al. 1991; Umemoto et al. 1990) have studied growth phenotypes of the chromosomal VMA1-, VMA2-and VMA3-disrupted mutants. The three mutants can grow well in YPD medium (2% Bacto-yeast extract, 2% polypeptone and 2% glucose: Ohya et al. 1991), indicating that each VMA gene is not indispensable for growth. However, they all show a Pet^–cls phenotype (Ohya et al. 1991): the vma null mutants cannot grow on a YPD plate containing 100 mmol l⁻¹ CaCh and on YP plates (2% Bacto-yeast extract, 2% polypeptone and 2% agar: Ohya et al. 1991) containing nonfermentable carbon sources such as 3% glycerol and 2% succinate. The Pet⁻ phenotype was unexpected and difficult to explain at this stage of study, but the calcium-sensitive cls phenotype could be logically understood because the three VMA disruptants have defects of vacuolar H⁺-ATPase activity, ATP-dependent Ca²⁺ uptake into isolated vacuoles and vacuolar acidification in vivo (Ohya et al. 1991).

In parallel with these studies, Ohya et al. (1986) have isolated 30 Ca²⁺-sensitive (cls) mutants of 5. cerevisiae, each with a single recessive chromosomal mutation, and classified them into 18 complementation groups with four subtypes based on their calcium contents and Ca²⁺ uptake activities. Of these four subtypes, type IV mutants (cls7^–clsl 1), which all have normal calcium contents but show increased initial rates of Ca²⁺ uptake, are a pet mutant and this Pet⁻ phenotype co-segregates with the Cls⁻ phenotype (Ohya et al. 1986). A genetic study was planned to determine whether vma mutations are allelic to some of the Pet^–cls mutations. The results of complementation analysis between vmal^–vma3 and cls7^–clsll mutants demonstrated that vma1 and vma3 do not complement cls8 and cls7, respectively, and that vma2 complements all five cls mutants, indicating that VMA1 and VMA3 are identical with CLS8 and CLS7, respectively. The vma2 mutation is not involved in the cls mutations tested (Ohya et al. 1991). Vacuolar membrane vesicles were prepared from the five mutants; DCCD-sensitive ATPase activity and ATP-dependent activity for Ca²⁺ uptake were not detected in these vesicles (Ohya et al. 1991). Based on these genetic and cell biological data, it was concluded that the Pet^–cls mutants were ascribable to vma defects. Thus, CLS9, CLS10 and CLS11 are a family of VMA genes and are designated henceforth VMA11, VMA 12 and VMAJ3, respectively (Ohya et al. 1991).

In yeast cells growing in YPD medium, the cytosolic free Ca²⁺ concentration ([Ca²⁺]_i) is critically regulated at about 150–180 nmol l⁻¹ (lida et al. 1990a,b). Measurements of [Ca²⁺]_i in individual cells of the Pet^–cls mutants yielded a mean value of 900–1100 nmol l⁻¹, as a primary consequence of the vma mutation (Ohya et al. 1991). Thus, the sixfold increase in [Ca²⁺]_i may trigger serious metabolic perturbation and is injurious to growth of yeast cells (Anraku et al. 1991 b; Galons et al. 1990). Unlike the majority of previously isolated pet mutants (Tzagoloff and Dieckmann, 1990), however, vma (Pet⁻cls) mutants show no detectable mitochondrial defects (Ohya et al. 1991). The vma mutants show pH-conditional growth phenotypes (Umemoto et al. 1991). These pH-conditional growth phenotypes have become known (Beltrán et al. 1992; Noumi et al. 1991; Yamashiro et al. 1990) and can be used for selecting new genes of the VMA family.

By 1991, nine VMA genes had been identified from the yeast S. cerevisiae (Table 1). Table 1 also lists the subunits encoded by the respective genes, with their names revised according to the proposals of Anraku et al. (1992) and of Nelson and Taiz (1989). Subunits designated by an italic capital letter are polypeptides that are peripheral in nature and are a counterpart of Fi of the ATP synthase, whereas those designated by an italic lower case are subunits that are integral in nature and are a counterpart of the F_o sector. All the candidate subunits detected biochemically and immunochemically (Kane et al. 1989) are listed for reference.

VMA1 was isolated from a yeast genomic DNA library (Yoshihisa and Anraku, 1989) by hybridization with a 39-mer oligonucleotide probe corresponding to the 13 amino acid sequence in the purified 67×10³M_r subunit (Hirata et al. 1990). The nucleotide sequence of the gene predicts a polypeptide of 1071 amino acids (118 635 Da), which is much larger than the mature form of the 67×10³M_r subunit in the vacuolar membrane. N-and C-terminal regions of the deduced sequence (residues 1–284 and 739–1071) are very similar to those of the catalytic subunits of vacuolar H⁺-ATPases from Daucus carota (69×10³M_r) (Zimniak et al. 1988) and Neurospora crassa (67×10³A_r) (Bowman et al. 1988c). Alignment of the deduced sequence of yeast VMA1 with these two sequences also revealed that it contains a nonhomologous insert of 454 amino acids (residues 285–738), which shows no detectable sequence similarities to any known ATPase subunits (Hirata et al. 1990). None of the six tryptic peptides determined with the purified subunit is located in this internal region (Anraku et al. 1991a; Hirata et al. 1990).

The VMA1 gene does not have any splicing consensus sequence for nuclear-coded genes (Langford and Gallwitz, 1983). However, the nonhomologous region may be excised by a mechanism similar to mitochondrial mRNA splicing (Lazowska et al. 1989). Northern blotting analysis was carried out with two DNA probes: probe 1 from the homologous region of the VMA1 gene and probe 2 from the nonhomologous insert. Each probe detected only a single RNA species of 3.5 kb in both poly(A)+ and total RNA fractions (Hirata et al. 1990), which is consistent with the whole length of the VMA1 open reading frame (3213 bases). This 3.5-kb species was not observed in the RNA fraction from the null vmal cells. Thus, it is concluded that the transcript of VMA1 is not spliced and that a novel processing mechanism, which may involve a post-translational excision of the integral region followed by peptide ligation, operates on the yeast VMA1 product (Hirata et al. 1991). Recently, Kane et al. (1990) have shown that yeast cells carrying VMA1 under control of the inducible GAL10 promoter express a 119×10³M_r polypeptide of the unprocessed VMA1 gene product in galactose medium and that the precursor undergoes post-translational cleavage and splicing to yield the mature 67×10³M_r subunit A and a 50×10³M_r polypeptide.

Assuming that the whole stretch of the nonhomologous insert (residues 285–738) is removed from the VMA1 product, a molecular mass of 67 722 Da is calculated for the mature subunit consisting of 617 amino acids. This is in good agreement with the value for the relative molecular mass of 67×10³M_r estimated by SDS–polyacrylamide gel electrophoresis (Hirata et al. 1990). Thus, the deduced primary sequence of yeast Vmalp is very similar to those of the Neurospora crassa (Bowman et al. 1988c) and Daucus carota (Zimniak et al. 1988) counterparts: About 73 and 60%, respectively, of the residues are identical with the fungal and plant sequences.

Vmalp is the catalytic subunit of the enzyme complex (Uchida et al. 1988) and localizes to the cytoplasmic side of the vacuolar membrane (Fig. 4). Consistent with this biochemical evidence, the deduced primary sequence of Vmalp shows about 25% sequence identity over 400 residues with β subunits of F₁F₀-ATPases (Hirata et al. 1990). Vmalp has consensus sequences for the nucleotide-binding domain proposed by Walker et al. (1982) and contains conserved amino acid residues that have proved to be important for ATP hydrolysis (Futai et al. 1989), suggesting that the catalytic subunits from the two classes of ATPases share similar structures and mechanisms of ATP hydrolysis.

Based on the nucleotide sequence information of the 57×10³M_r subunit of Arabidopsis thaliana vacuolar H⁺-ATPase (Manolson et al. 1988), a cDNA clone encoding a counterpart subunit in yeast has been isolated (Nelson et al. 1989; Yamashiro et al. 1990). The predicted amino acid sequence deduced from the nucleotide sequence proved to contain all the four peptides that were determined with the purified 57×10³M_r subunit from S. cerevisiae (Anraku et al. 1991a; Hirata et al. 1990). Independent of these studies, Ohya et al. (1991) isolated the VMA2 gene and showed by Western blotting analysis that the null vma2 strain has no immunoreactive 57×10³M_r subunit in the cell lysate.

The nucleotide sequence of VMA2 predicts a polypeptide of 517 amino acids (57 749 Da). Comparison of sequence homology (Yamashiro et al. 1990) revealed extensive sequence identities of 82, 74, 54, 58 and 74%, respectively, to the 60 × 10³M_r subunits from Neurospora crassa (Bowman et al. 1988a), Arabidopsis thaliana (Manolson et al. 1988), Sulfolobus acidcaldarius (Denda et al. 1988a,b), Methanosarcia barkeri (Inatomi et al. 1989) and human endomembrane (Südhof et al. 1989).

Vma2p seems to be present in an equimolar amount with Vmalp in purified enzymes from yeast (Uchida et al. 1985; Kane et al. 1989). Vacuoles isolated from the vma2 cells showed no vacuolar H⁺-ATPase activity and no vacuolar acidification ability (Ohya et al. 1991; Yamashiro et al. 1990), so this major subunit is essential for the expression of enzyme activity, probably functioning as a regulatory component (Hirata et al. 1989).

Two independent strategies were adopted for cloning the VMA3 gene. For a hybridization probe of VMA3 from a yeast genomic DNA library, Nelson and Nelson (1989) synthesized a 105-mer oligonucleotide based upon 35 amino acids of the C terminus of the 17×10³M_r proteolipid from bovine chromaffin granules (Mandel et al. 1988) and isolated two positive clones by dot blots and Southern hybridization. Umemoto et al. (1990) isolated and characterized one positive clone, using a 43-mer oligonucleotide probe that was synthesized based upon the determination of the N-terminal 17 amino acids with the purified 16×10³M_r proteolipid from yeast vacuoles. Nucleotide sequencing of all the candidates revealed that they contain a single open reading frame encoding a hydrophobic polypeptide of 160 amino acids (16350Da) (Nelson and Nelson, 1989; Umemoto et al. 1990). VMA3 (CLS7) has been mapped on the left arm of chromosome V in S. cerevisiae (Ohya et al. 1986).

The predicted amino acid sequence of the VMA3 gene product shows extensive sequence identity (64%) to the 17×10³M_r proteolipid from bovine chromaffin granules (Mandel et al. 1988), but is less homologous (30% identity) to the proteolipid from Sulfolobus acidocaldarius (Denda et al. 1989). The amino acid sequence of the N-terminal half of Vma3p (residues 1–78) was found to be 23% identical to that of the C-terminal half (residues 79-160) (Umemoto et al. 1990). The C-terminal half of yeast Vma3p showed significant homology (about 35% identity) to SxliPAfr proteolipids of spinach chloroplasts, yeast mitochondria, bovine mitochondria and cyanobacterium Synechococcus (Cozens and Walker, 1987; Sebald and Hoppe, 1981). Homology of the N-terminal half was less marked and showed about 27% identity to 8×10³M_r proteolipids of thermophilic bacterium PS3 and Bacillus megaterium (Brusilow et al. 1989; Sebald and Hoppe, 1981). This suggests that the yeast VMA3 gene is a duplicated and diverged form of the genes encoding SxKPAfp proteolipids of the F_o sectors in F₁F₀-ATPases (Nelson and Nelson, 1989).

Subunit c in the partially purified yeast enzyme bound DCCD (Uchida et al. 1985), suggesting that it may function as a part of a channel for proton translocation in the H⁺-ATPase complex (Kakinuma et al. 1981). Hydropathy analysis predicts that Vma3p contains four membrane-spanning domains (Nelson and Nelson, 1989; Umemoto et al 1990: Glu-137 exists in the fourth domain, which has been reported to be the conserved DCCD-binding site in various proteolipids of the F₁F₀-ATPases.

Foury (1990) discovered VMA4 while characterizing the M1P1 gene that encodes the catalytic subunit of the yeast mitochondrial DNA polymerase (Foury, 1989). The VMA4 open reading frame (699 bases) was determined; it predicted a hydrophilic polypeptide of molecular mass 26.6 kDa. The deduced amino acid sequence shows 34% identity to the 31×10³M_r subunit of the V-ATPase from kidney microsomes (Hirsch et al. 1988). VMA4 and MIP1 were found to be located on chromosome XV and the initiation sites of their mRNAs are only separated by about 185 bp (Foury, 1989, 1990). Vma4p is a peripheral 27×10³M_r subunit of the enzyme complex (Table 1). The function of the subunit is not known yet.

Beltrán et al. (1992) isolated the 42×10³M_r subunit from purified yeast vacuolar H⁺-ATPase and determined its partial amino acid sequence. Based on this peptide information, an oligonucleotide was designed for screening clones containing VMA5 from a yeast genomic DNA library. The nucleotide sequence of VMA5 predicts a polypeptide of 373 amino acids (42 287 Da). The protein is hydrophilic in nature with a neutral isoelectric point of 7.03. The predicted amino acid sequence contains the sequence of the 20 amino acids determined and shows 39% identity to the bovine counterpart in 311 overlapping amino acids. Vma5p is a peripheral subunit of the enzyme complex and is liberated from the vacuolar membrane by sodium carbonate treatment (Kane et al. 1992). The function of the 42×10³M_r subunit C is not known yet.

VMA11 was isolated from a yeast genomic DNA library by complementation of the vmall mutation (Umemoto et al. 1991): a haploid strain NUY30 (vmall leu2) was transformed with the DNA library on YEpl3, and five colonies that grew on YP–glycerol plates were isolated from about 12000 Leu⁺ transformants. These five positive transformants could also grow on a YPD plate containing 100 mmol l⁻¹ CaCh-Two plasmids were recovered after the second round of transformation followed by tests of plasmid loss. The restriction maps of the two plasmids show that both inserts contain the same DNA fragment. The 1.8-kb EcoRV-Spel fragment (pNUVA366) that complements the vmall mutation as a minimal essential region was restricted after testing a series of deletions of the inserts constructed from the 11-kb original isolate pNUVA35O, confirming that this complementing activity was not due to extragenic suppression by integrative mapping with this clone (Umemoto et al. 1991). The nucleotide sequence of pNUVA366 shows that the authentic VMA11 gene encodes a hydrophobic polypeptide of 164 amino acids (17 037 Da).

The nucleotide sequence of the VMA11 gene contains a nine-base repeat, AGCTGCCAT, at positions 72–80 and 99–107; these sequences were not present in a reported sequence of the TFP3 gene (Shih et al. 1990) encoding a hydrophobic protein of 10×10³M_r. The deduced amino acid sequence predicts a surprising coincidence in amino acid composition with Vma3p, showing extensive sequence identity (56.7% in 150 amino acids) to Vma3p (Umemoto et al. 1991). R. Hirata and Y. Anraku (unpublished observations) demonstrated that Vmal 1p is located in the vacuolar membrane. Based on this finding and its extensive homology with Vma3p, the VMA11 gene product was designated as subunit c′ (Table 1).

The disruption of either one of the VMA3 and VMA11 genes causes loss of vacuolar acidification (Fig. 5) and leads to defective assembly of subunits A, B and c of the H⁺-ATPase (Umemoto et al. 1991), suggesting that the functions of the two genes are independent. To confirm this point genetically, they constructed plasmids harboring each gene on multicopy vector pYO325 and used them for analysis of multicopy suppression. Results indicated that VMA11 and VMA3 on multicopy plasmids do not suppress null mutations of vma3 and vmall, respectively. Thus, the two genes do not share functions, but function independently. Vmal 1p may be a second species of DCCD-binding proteolipid from the yeast vacuole because the deduced amino acid sequence predicts that a conserved glutamic acid residue of the DCCD-binding site in proteolipids of the F₁F_o-ATPase and the V-ATPase is present in the sequence (Umemoto et al. 1991).

Vacuolar membrane vesicles prepared from the VMA11-disrupted cells had lost Vma3p completely, and neither Vma1p nor Vma2p assembled on the membranes, although these two peripheral subunits were synthesized normally and were present in the total cell extract. These results suggest that the function of Vmal lp is a prerequisite for assembly of subunit c and then subunits A and B on the vacuolar membrane (Umemoto et al. 1991).

It has been proposed that VMA6 encodes a 36 ×10³M_r subunit of the H⁺-ATPase in yeast (C. M. Bauerle, M. N. Ho, M. A. Lindorfer and T. H. Stevens, personal communication). VMA 12 and VMA13 have been shown to be indispensable genes for expression of the enzyme activity (Ohya et al. 1991). The nucleotide sequences of VMA12 (N. Umemoto, R. Hirata, Y. Ohya and Y. Anraku, unpublished data) and VMA13 (R. Hirata, N. Umemoto, Y. Ohya and Y. Anraku, unpublished data) have been determined; the sequences predict that Vmal2p is a 25 ×10³M_r integral polypeptide with two membrane-spanning domains and that Vmal3p is a 54x10³M_r hydrophilic polypeptide with low homology for the y subunit of Sulfolobus acidocaldarius (Denda et al. 1990). Vmal3p seems to be a counterpart of the 54×10³M_r subunit detected in the vacuolar H⁺-ATPase from Beta vulgarius (Parry et al. 1989).

Studies from our laboratory have demonstrated that the yeast vacuole is the center for regulation of ionic homeostasis in the cytosol (Anraku et al. 1989, 1991a,b, 1992). Even if a family of the VMA genes is all present and normal, the large volume of a central vacuole is needed physiologically to confer on the organelle a high capacity for maintenance of homeostatic levels of cytosolic free Ca²⁺ and basic amino acids (Kitamoto et al. 1988a,b’, Ohsumi et al. 1988) and for compartmentation of a number of vacuolar proteases (Banta et al. 1990; Wada et al. 1990). Wada et al. (1990, 1992b) have developed several genetic methods for isolating yeast mutants defective in vacuolar morphogenesis and they have identified genes involved in the acquisition of large vacuoles. Interestingly, several mutations in the VAM genes (vam1, vam5, vam8 and vam9; for vacuolar morphology), which result in complete loss of central vacuoles (Wada et al. 1992b), show a Ca²⁺-sensitive phenotype of type-I cls mutation (Ohya et al. 1986,1991) and are allelic to the respective vps, pep and end mutations for low vacuolar peptidases and missorting of carboxypeptidase Y. These results suggest that a recessive mutation on a single chromosomal gene can cause pleiotropic defects in vacuolar lytic function and vacuolar morphogenesis (Wada et al. 1992b).

This article summarizes the present status of genetic information on how many VMA genes are required for full expression and regulation of the yeast vacuolar H⁺-ATPase. Nine VMA genes have proved to be essential for expression of the enzyme activity. The structure and function of the VMA gene products are discussed, in addition to the phenotypes of the null vma mutations.

The enzyme is a large hetero-oligomeric complex with at least nine subunits but the nature of subunit composition and function awaits further elucidation. Molecular biological issues regarding the biogenesis of this holoenzyme and the vacuolar morphogenesis remain to be studied. Of great current interest is the mechanism by which this typical V-type H⁺-ATPase can accomplish vacuolar acidification and control homeodynamic chemiosmosis in a eukaryotic cell system.

The original work from our laboratory described in this article was carried out in collaboration with Drs Y. Ohsumi, Y. Kakinuma, E. Uchida and N. Umemoto and other coworkers, whose names appear in the references. This study was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science and Culture of Japan and a grant from the Human Frontier Science Program Organization, Strasbourg, France.