Biogenesis of the yeast vacuolar H⁺-ATPase

The yeast vacuolar H⁺-ATPase closely resembles the V-ATPases from other fungi, plants and animals, both in its overall structure and in the sequences of the subunit genes that have been cloned (reviewed in Kane and Stevens, 1992). The yeast enzyme has been partially purified by density gradient centrifugation of solubilized yeast vacuoles and the fraction exhibiting ATPase activity contains eight polypeptides of relative molecular masses (Af_r) 100, 69, 60, 42, 36, 32, 27 and 17×10³ (Uchida et al. 1985; Kane et al. 1989b). The same collection of polypeptides was immunoprecipitated by a monoclonal antibody recognizing the 69×10³M_r subunit, suggesting that all of these polypeptides are part of the active complex (Kane et al. 1989b). Further support for this subunit composition has been provided by genetic studies, which indicate that mutagenesis of any one of the genes for the 100, 69, 60, 42, 27 and 17×10³M_r subunits can totally abolish

ATPase activity in isolated vacuolar vesicles and elicit a set of phenotypes which has been associated with loss of vacuolar H⁺-ATPase activity (Manolson et al. 1992; Kane et al. 1990; Hirata et al. 1990; Yamashiro et al. 1990; Nelson and Nelson, 1990; Beltran et al. 1992; Foury, 1990). Cloning of the genes for the 36 and 32×10³M_r polypeptides present in the glycerol gradient has not yet been reported.

Biochemical studies have revealed that the yeast vacuolar H⁺-ATPase is composed of both peripheral and integral membrane subunits. The 69, 60 and 42×10³M_r subunits can be stripped from the vacuolar membrane by potassium nitrate, indicating that they are peripheral membrane proteins (Kane et al. 1989b). The 27×10³M_r subunit also appears to be a peripheral membrane protein based on its predicted amino acid sequence (Foury, 1990). The 100 and 17×10³M_r subunits are integral membrane proteins that remain in the membrane through either potassium nitrate or alkaline sodium carbonate treatment (Kane et al. 1992). Both the nitrate-stripping data and analysis of mutants lacking one peripheral subunit (described below) suggest a structural association between the 69, 60 and 42×10³M_r subunits (Kane et al. 1989b, 1992; Noumi et al. 1991; Beltran et al. 1992). These three subunits along with other peripheral subunits appear to form a peripheral complex, called the V₁ sector of the vacuolar H⁺-ATPase by analogy with the Fi sector of the F₁F₀. ATPases (Bowman et al. 1989; Puopolo and Forgac, 1990). The V₁ sector of the vacuolar H⁺-ATPase contains the catalytic sites for ATP hydrolysis (Uchida et al. 1988) but, unlike the F₁-ATPase, the V₁ sector has no activity when dissociated from the membrane subunits in yeast or in any other system that has been studied (Kane et al. 1989b; Bowman et al. 1989; Moriyama and Nelson, 1989; Adachi et al. 1990). The 100 and 17×10³M_r subunits have been proposed to form the V_o membrane sector of the enzyme that is thought to contain the proton pore, based on labeling of the 17×10³M_r subunit with dicyclohexyl carbodiimide (Uchida et al. 1985), indirect evidence suggesting a structural association between the 100 and 17×10³M_r subunits (Kane et al. 1992) and by analogy with the FiF₀-ATPases. The current working model for the structure of the yeast vacuolar H⁺-ATPase is shown in Fig. 1.

A variety of methods have been developed that exploit homologous recombination in yeast cells in order to replace the chromosomal copy of a wild-type gene with a copy of the cloned gene that has been mutagenized in vitro (reviewed in Rothstein, 1991). These techniques have been used to construct a variety of mutant yeast strains that fail to produce one subunit of the vacuolar H⁺-ATPase. Studies of mutant cells lacking one of the subunits of the vacuolar H⁺-ATPase have not only provided support for the biochemically determined subunit composition, but have also generated a number of insights about the biosynthesis and assembly of the enzyme complex. The genes for the 100, 69, 60, 42, 27 and 17×10³M_r subunits (called VPH1, TFPUVMA1, VAT2/VMA2, VATC, VMA4 and VMA3, respectively) have been identified and cloned (Manolson et al. 1992; Shih et al. 1988; Nelson et al. 1989; Beltran et al. 1992; Foury, 1990; Nelson and Nelson, 1989) and the corresponding mutant strains have been constructed. Deletions in any of these subunit genes yield a well-defined set of phenotypes, which includes a complete loss of vacuolar acidification, absence of all ATPase activity in isolated vacuoles and failure to grow in media buffered to neutral pH (Nelson and Nelson, 1990; Yamashiro et al. 1990; Ohya et al. 1991). Lack of one subunit does not necessarily affect the synthesis or stability of the other subunits (see below).

Deletion of any of several of the peripheral subunits, including the 69, 60, 42 and 27×10³M_r subunits, has a very similar effect on the other subunits of the vacuolar ATPase. In general, the other peripheral and integral membrane subunits for which a means of detection is available (including the 100, 69, 60, 42 and 27×10³M_r subunits) are present in whole-cell lysates of the mutant cells at nearly the same levels as in wildtype cells (Kane et al. 1992; Beltran et al. 1992; M. N. Ho and T. H. Stevens, personal communication). Isolated vacuolar membranes from the mutant cells contain none of the peripheral subunits when one of the peripheral subunits is missing, suggesting that, although these subunits are synthesized in the mutant cells, they are not properly assembled and/or targeted to the vacuolar membrane (Kane et al. 1992; Beltran et al. 1992; M. N. Ho and T. H. Stevens, personal communication). In contrast, both the 17 and 100×10³M_r subunits are present in the vacuolar membranes of these mutant cells, indicating that they are not dependent on the peripheral subunits for targeting to the vacuole (Kane et al. 1992; Noumi et al. 1991; Umemoto et al. 1990). These results have been confirmed by immunofluorescence microscopy using antibodies recognizing the 60 and 100×10³M_r subunits. In wild-type cells, the 60×10³M_r subunit appears as a ring around the vacuolar membrane. However, in cells lacking the 69×10³M_r subunit, the 60×10³M_r subunit appears to be diffusely distributed in the cytoplasm, while the antibody recognizing the 100×10³M_r subunit stains the vacuolar membrane (Kane et al. 1992).

Deletion of the 17×10³M_r subunit gene (VMA3), which encodes an integral membrane subunit of the ATPase, has somewhat different effects on assembly and targeting of the other subunits. The peripheral subunits, including the 69, 60 and 42X KPA/r subunits, are present in the cells at near-normal levels, but are not present in isolated vacuolar vesicles (Umemoto et al. 1990; Noumi et al. 1991; Kane et al. 1992). Antibodies against the 69×10³M_T subunit (Umemoto et al. 1990) or the 60×10³M_r subunit (Kane et al. 1992) appear to stain the cytoplasm when viewed by immunofluorescence microscopy. The 100×10³M_r subunit, another integral membrane subunit, is present at reduced levels (about 10% of wild-type) in whole-cell lysates from vma3Δ mutants, but it has not yet been established whether changes in the rate of synthesis or the stability of this subunit cause the reduced cellular levels. Isolated vacuolar vesicles from vma3Δ cells contain no 10()×10³M_r subunit, suggesting that even the small amount of the 100×10³M_r subunit that is synthesized in the mutants is not transported to the vacuole (Kane et al. 1992). Therefore, the 17×10³M_r subunit appears to be necessary for the vacuolar targeting of both the 100×10³M_rintegral membrane subunit and the peripheral subunits of the ATPase.

The yeast mutants lacking one ATPase subunit have been analyzed in more detail to determine whether the remaining subunits that do not reach the vacuolar membrane are still partially assembled. This type of analysis has previously yielded information about the steps involved in assembly of other multisubunit proteins because assembly can often proceed in the mutant cells up to the point where the missing subunit is required (Crivellone et al. 1988; Manolios et al. 1991). Wild-type and mutant cells lacking one subunit were subjected to high-speed centrifugation in order to separate the ATPase subunits into an insoluble fraction (containing membrane-bound proteins and large protein aggregates) and a soluble fraction. Even in wild-type cells, the 69, 60, 42 and 27×10³M_r subunits were found partially in the soluble fraction, while the 100×10³M_r subunit of the ATPase and another vacuolar membrane protein, dipeptidyl aminopeptidase B, were found entirely in the insoluble fraction. The distribution of the subunits between the soluble and insoluble fractions varied in the different mutant cells, with tfpl Δ and vat2Δ mutant cells exhibiting the smallest proportion of the peripheral subunits in the soluble fraction and the vma3A cells exhibiting a large proportion in the soluble fraction (R. D. Doherty and P. M. Kane, unpublished data). Further fractionation of the soluble fractions by glycerol density gradient centrifugation showed that all of the strains contained the peripheral subunits at least partially in a low-density fraction, corresponding to a relative molecular mass of less than 100×10³M_r when compared to molecular mass standards run in parallel. The subunits in this fraction are probably unassembled or associated to only a very limited degree. In the vma3 Δ and wild-type cells, the 69, 60, 42 and 27×10³M_r subunits are also found in a second region of the gradient of much higher density, corresponding to a relative molecular mass of roughly 400×10³ compared with the molecular mass standards. The subunits fractionating in this region of the gradient may be associated into partially assembled sub-complexes and the predicted size of these complexes is large enough to constitute most or all of the V i sector (R. D. Doherty and P. M. Kane, unpublished data). The absence of the high-density peak in the tfplΔand vma4Δ mutants suggests that partial assembly into this complex cannot occur in the absence of the 69 or the 27×10³M_r subunits. The high proportion of the peripheral subunits in the insoluble fraction from tfpl Δ and vat2Δ cells may be attributable to insoluble protein aggregates resulting from a failure to assemble, since the initial analysis of these mutants (described above) provided little evidence for association with the integral membrane subunits. A model for assembly of the peripheral subunits into the V₁ sector, which includes the results of the experiments with mutant cell lines, is presented in the cartoon in Fig. 2.

The experiments described above focus on the subunit composition of the final, active form of the vacuolar H⁺-ATPase and the role played by those subunits in assembly of the complex. However, it has become increasingly clear that even complexes that can selfassemble in vitro often utilize other cellular proteins to assist their assembly in vivo (reviewed in Ellis and van der Vies, 1991; Gething and Sambrook, 1992). The general class of ‘assembly factors’ covers a range of functions, encompassing both chaperone proteins, which assist a wide variety of proteins in attaining tertiary or quaternary structure, and very specific assembly factors, which appear to act on a single protein complex. Candidates for both types of assembly factor have been identified for the yeast vacuolar H⁺-ATPase.

The VPS3 and VPS6 gene products appear to influence the assembly of the yeast vacuolar H⁺-ATPase, but have pleiotropic effects on the cell, suggesting that they may act on a number of proteins unrelated to the ATPase or acidification. A number of vps3 and vps6 mutants were initially identified as part of a large set of mutants defective in sorting of soluble vacuolar proteases to the vacuole (vps=vacuolar protein sorting; Rothman and Stevens, 1986; Rothman et al. 1989a). The vps3 and vps6 mutants were subsequently found to be distinct from the rest of the collection in that they failed to accumulate the lysosomotropic amine quinacrine, indicating that they had an acidification defect as well as the protein-sorting defect (Rothman et al. 1989b). Further analysis of these mutants revealed that the specific activity of the vacuolar ATPase in isolated vacuoles was greatly decreased (6–15% of wild-type activity) and the levels of the 69 and 60×10³M_r subunits in vacuolar membranes were comparably diminished, even though these subunits were present at near-normal levels in whole-cell lysates from the mutant cells (Rothman et al. 1989b). However, comparison of the phenotypes of a vps3Δ mutant and a vat2Δ mutant revealed significant differences, suggesting that the VPS3 gene product plays a number of cellular roles in addition to its potential role in ATPase assembly (Raymond et al. 1990). For example, the dramatic defect in protein sorting seen in the vps3 mutants appeared to be distinct from the defect in acidification (Raymond et al. 1990). Similarly, the vacuolar pH of a pepl2Δ mutant (PEP12 is the same gene as VPS6;Rothman et al. 1989a) appeared to be lower than that of mutants lacking an ATPase subunit, indicating a less severe acidification defect (Preston et al. 1989), but the effects of the mutation on other cellular functions, including vacuolar protein sorting and zymogen activation, were more severe than those seen in the vacuolar ATPase mutants (Yamashiro et al. 1990). Thus, both the VPS3 and VPS6 gene products appear to affect a number of cellular functions, only one of which is assembly of the vacuolar H⁺-ATPase. It is also entirely possible that some of the even more general yeast chaperone proteins (Gething and Sambrook, 1992) are involved in vacuolar H⁺-ATPase assembly, but there is no direct evidence of this at present.

In contrast, a number of other gene products have now been identified that appear to be specifically necessary for ATPase activity even though they are not part of the final ATPase complex and these proteins are good candidates for the highly specific assembly factors described above. Mutations in the VMA12 and VMA13 genes yield a set of phenotypes identical to that of the tfplΔ, vat 2 Δ and vma3Δ mutants (Ohya et al. 1991). However, the predicted sizes and sequences of the VMA12 and VMA13 gene products do not correspond to those of any of the previously identified polypeptides in the glycerolgradient-purified vacuolar H⁺-ATPase, indicating that even though these proteins are indispensable for vacuolar ATPase activity they may not be part of the active complex (Anraku et al. 1992; Kane and Stevens, 1992). Another gene, VMA11, has been identified from a mutant strain exhibiting the characteristic set of phenotypes for vacuolar H⁺-ATPase mutants (Ohya et al. 1991). The VMA11 gene is highly homologous with VMA3 and may encode a second proteolipid (17×10³M_r) subunit of the ATPase (Umemoto et al. 1991 ). However, the protein sequence obtained for the chloroform methanol-extracted proteolipid from vacuolar membranes corresponded to the predicted sequence of the VMA3 gene (Anraku et al. 1991), so it is not yet clear whether the VMA11 gene product is present in vacuolar membranes or the active vacuolar H⁺-ATPase complex. The exact functions of the VMA11–13 gene products are still under investigation.

No multisubunit protein of the yeast vacuolar membrane has been extensively studied previously, so the requirements for assembly and vacuolar targeting of the yeast vacuolar H⁺-ATPase are largely unknown. However, the proposed structure for the yeast vacuolar H⁺-ATPase shown in Fig. 1 has immediate implications for its biogenesis and the experiments with mutants described above provide further clues about these processes.

The biogenesis of the peripheral subunits, which are on the cytoplasmic face of the vacuolar membrane, may bear more resemblance to the synthesis and transport of cytoplasmic proteins than to those of vacuolar proteins. Two types of experimental evidence indicate that the peripheral subunits never enter the secretory pathway. First, the sequences of cloned genes of the 69, 60, 42 and 27×10³M_r subunits contain no evidence of signal sequences or potential transmembrane domains. Second, there are several potential sites for N-linked glycosylation in the predicted amino acid sequence of the 60 and 69×10³M_r subunits, but neither subunit receives N-linked glycosylation (P. M. Kane and T. H. Stevens, unpublished data). These results suggest that the peripheral subunits that make up the V₁ sector remain in the cytosol prior to attachment to the integral membrane subunits and leave open the possibility that they may partially or completely assemble into the V₁ sector before they become attached to the V_o sector. The presence of high-molecular-mass complexes in the supernatants from vma3Δ and wild-type cells provides experimental support for this possibility.

The structure shown in Fig. 1 suggests that the integral membrane subunits may behave like other integral membrane proteins of the yeast vacuole. The fact that the integral membrane subunits can reach the vacuole independently of assembly with the peripheral subunits also suggests that they are targeted as vacuolar membrane proteins. The transport of several integral membrane proteins, including dipeptidylaminopeptidase B (Roberts et al. 1989), alkaline phosphatase (Klionsky and Emr, 1989) and the vacuolar glycoprotein vgp72 (Nishikawa et al. 1990), to the vacuole has been well defined. All of these proteins follow the initial stages of the secretory pathway in their transit to the vacuole. They enter the endoplasmic reticulum and travel to the Golgi apparatus along with proteins destined for secretion and are separated from these proteins by some type of sorting step in the Golgi apparatus. If the integral membrane proteins of the vacuolar H⁺-ATPase also follow this pathway, then the failure of the 100×10³M_r subunit to reach the vacuole in the absence of the 17×10³M_r subunit could be caused by a requirement for the Vo sector to assemble in the endoplasmic reticulum. A large number of membrane protein complexes that travel through the secretory pathway appear to assemble in the endoplasmic reticulum (Hurtley and Helenius, 1989). It is still very unclear at what point the peripheral subunits become associated with the integral membrane subunits, but the point of association between the two sectors of the ATPase could be very important in determining which organelles are acidified (see below).

The overall structural resemblance of the vacuolar H⁺-ATPases to the F₁F₀-ATPases has frequently been noted and suggests that there might also be similarities in the assembly of the two types of complexes (Nelson and Taiz, 1989). There is evidence that the F₁ portion of both the E. coli and the mitochondrial F₁F₀-ATPase can assemble in the absence of the membrane sector (Klionsky and Simoni, 1985; Schatz, 1968). In E. coli, the F₀ subunits also appear to be able to assemble in the absence of the F| subunits (Aris et al. 1985). There is evidence that the F₀ and F₁ portions of the yeast enzyme each have associated assembly factors ( Ackerman and Tzagoloff, 1990a, b). However, the membrane topology and organellar location of the V-type and F-type ATPases indicate that other aspects of their biogenesis may be quite different (Fig. 3). Subunits of the chloroplast or mitochondrial enzymes that are encoded by nuclear genes are synthesized in the cytoplasm with an appropriate organelle targeting signal and individually imported into the organelle, where they are assembled into the enzyme complex. In some cases, including the yeast mitochondrial F₁F₀-ATPase, assembly of the full complex also requires association of subunits encoded by the organellar genome with nuclearly encoded and imported subunits. Although assembly of the F₁F₀-ATPase is still under study, it is clear that subunit import may be an important regulatory step in the assembly pathway (Burns and Lewin, 1986). In this respect, the vacuolar H⁺-ATPases of eukaryotic cells resemble the prokaryotic F₁F₀-ATPase more closely than they do the mitochondrial or chloroplast enzymes (see Fig. 3), because the subunits of the V₁ sector may be immediately available for assembly after synthesis.

One of the central questions surrounding the acidification of intracellular compartments is the issue of how different organelles of the biosynthetic and endocytic pathways, including the lysosome (or vacuole), endosomes, the Golgi apparatus, clathrin-coated vesicles and secretory vesicles, can be acidified to different extents (Mellman et al. 1986). One means of regulating organelle pH is the regulation of other ion permeabilities of the organelle membrane (Fuchs et al. 1989). However, it is impossible to dissociate the question of how the V-ATPases are assembled in eukaryotic cells from questions about the distribution of acidic compartments and the regulation of acidification. Evidence that the V₁ and V₀ sectors of the yeast vacuolar H⁺-ATPase may be assembled and targeted to the vacuole independently suggests that the stage at which the V₁ sector becomes attached to the V₀ sector could be an important factor in determining what intracellular compartments are acidified. For example, if V₁ attached to V₀ in the Golgi apparatus, then the active complex could travel from the Golgi apparatus to the vacuole and both compartments might be acidified to the extent of their occupancy by the vacuolar ATPase. Alternatively, identical V₁ sectors, assembled in the cytoplasm, could attach to multiple organelle-specific V₀ sectors, which regulate acidification of the compartments where they reside. The existence of two potential proteolipid genes (VMA3 and VMA11;Umemoto et al. 1991) and two homologous genes that could encode 100× 10³M_r subunits (VPH1 and STVT, Manolson et al. 1992) fits well into this model. (To date, only one gene for each of the peripheral subunits has been discovered in yeast.) It is clear that experiments aimed at understanding the assembly of vacuolar H⁺-ATPase complexes in yeast and other eukaryotes may not only yield insights into the assembly of multisubunit complexes, but also have much broader implications for cell physiology.

This work was supported by a grant from the Howard Hughes Medical Institute to the College of William and Mary and a National Science Foundation Presidential Young Investigator Award to P.M.K.