Mutations in the Yeast KEX2 Gene Cause a Vma⁻-Like Phenotype: a Possible Role for the Kex2 Endoprotease in Vacuolar Acidification

Abstract

Mutants of Saccharomyces cerevisiae that lack vacuolar proton-translocating ATPase (V-ATPase) activity show a well-defined set of Vma⁻ (stands for vacuolar membrane ATPase activity) phenotypes that include pH-conditional growth, increased calcium sensitivity, and the inability to grow on nonfermentable carbon sources. By screening based on these phenotypes and the inability of vma mutants to accumulate the lysosomotropic dye quinacrine in their vacuoles, five new vma complementation groups (vma41 to vma45) were identified. The VMA45 gene was cloned by complementation of the pH-conditional growth of the vma45-1 mutant strain and shown to be allelic to the previously characterized KEX2 gene, which encodes a serine endoprotease localized to the late Golgi compartment. Both vma45-1 mutants and kex2 null mutants exhibit the full range of Vma⁻ growth phenotypes and show no vacuolar accumulation of quinacrine, indicating loss of vacuolar acidification in vivo. However, immunoprecipitation of the V-ATPase from both strains under nondenaturing conditions revealed no defect in assembly of the enzyme, vacuolar vesicles isolated from a kex2 null mutant showed levels of V-ATPase activity and proton pumping comparable to those of wild-type cells, and the V-ATPase complex purified from kex2 null mutants was structurally indistinguishable from that of wild-type cells. The results suggest that kex2 mutations exert an inhibitory effect on the V-ATPase in the intact cell but that the ATPase is present in the mutant strains in a fully assembled state, potentially capable of full enzymatic activity. This is the first time a mutation of this type has been identified.

A distinct class of proton-translocating ATPases, the vacuolar-type ATPases (V-ATPases), is responsible for acidifying the eukaryotic vacuolar network, including the vacuole or lysosome, Golgi apparatus, endosomes, clathrin-coated vesicles, and regulated secretory vesicles (9). The Saccharomyces cerevisiae vacuole is an acidic organelle functionally equivalent to the mammalian lysosome and the plant vacuole (50). It is involved in metabolite storage, macromolecular degradation, and calcium and amino acid homeostasis (2). The yeast V-ATPase is a multisubunit enzyme consisting of at least 12 different polypeptides encoded by the VMA genes (2, 27, 63; reference 15 and references therein). As in F₁F₀-ATPases, the enzyme is made up of two domains in the yeast V-ATPase: the peripheral sector (the V₁ sector), which contains the catalytic ATP-hydrolyzing domain and is peripherally associated with the cytoplasmic face of the vacuolar membrane, and the integral membrane sector (the V₀ sector), which contains the proton pore (27).

Many of the genes that encode the yeast V-ATPase subunits, including VPH1, STV1 (a functional homolog of VPH1), VMA1 to -8, -10, -11, and -13, have been cloned (4, 14, 22, 23, 38, 39, 44, 58, 63, 66, 69). In addition, four genes which are required for assembly of the V-ATPase but are not part of the final active complex (VMA12, VMA21, VMA22, and VPH6) have been cloned (17, 19–21). In total, 17 genes have been identified as essential for V-ATPase activity. Deletion of any of these genes leads to the loss of vacuolar acidification and a conditional lethality in the resulting vma (stands for vacuolar membrane H⁺-ATPase activity) mutants; the mutants fail to grow on media buffered to pH 7 or higher (43, 69), medium containing high concentrations of calcium (45), or medium containing a nonfermentable carbon source (45), but they retain the ability to grow on medium buffered to pH 5.0 (43, 69). Vma⁻ cells fail to accumulate the fluorescent weak base, quinacrine, in their vacuoles, indicating loss of vacuolar acidification (69).

V-ATPases are present in several distinct locations within a single cell, and it is not yet clear how the enzyme is targeted to different cellular locations or regulated such that different organelles are acidified to different degrees (reviewed in reference 10). It is possible that one or more of the subunits may be involved in targeting and/or regulation. Isoforms have been identified for two of the V-ATPase subunits, the 17-kDa proteolipid and the 100-kDa integral membrane subunit (38, 66), and there is evidence that the two 100-kDa subunit isoforms are localized to different cellular locations (38). It also appears that a small collection of nonsubunit proteins may play an essential role in assembling, regulating, and targeting V-ATPase (21, 27).

Three genetic screens have been used to identify gene products required for V-ATPase activity. The VPH1, VPH2 (which is the same as VMA12), and VPH6 genes were identified in a screen for vacuolar pH mutants (vph [39, 48]), the VMA11, VMA12, and VMA13 genes were identified in a screen for calcium-sensitive strains showing a petite phenotype (cls [21]), and the VMA5 and VMA21-23 genes were identified in a screen for failure to accumulate a colored adenine precursor in the vacuole (22). The vph screen did not isolate mutations in any of the previously identified VMA genes, the cls screen identified alleles of VMA1 and VMA3 in addition to five novel genes, and the vma screen isolated mutations in the VMA1 gene in addition to four novel genes (21, 22, 39, 48). The combined results from these screens indicate that the screening process is not yet saturated.

Using the Vma⁻ phenotypes described above, we designed a genetic screen to identify novel mutant yeast strains lacking vacuolar acidification in order to better understand the subunit composition, assembly, and function of V-ATPase. In this study, we describe the identification of five new complementation groups whose activities are required for vacuolar acidification by yeast V-ATPase. In particular, we show that the Kex2 endoprotease is required for activity of V-ATPase in vivo.

MATERIALS AND METHODS

Materials.

Restriction endonucleases were purchased from New England Biolabs or from Boehringer Mannheim, and Taq DNA polymerase was purchased from Boehringer Mannheim. Zymolyase 100T and Tran³⁵S-label were purchased from ICN. ³⁵S-dATP was purchased from DuPont-NEN. A 1-kb DNA ladder and prestained and ¹⁴C-labeled protein molecular mass standards were obtained from Life Technologies, Inc. Dithiobis(succinimidylpropionate) was purchased from Pierce. Glusulase was obtained from DuPont. Zwittergent 3-14 detergent was purchased from Calbiochem. All other reagents were purchased from Sigma.

Strains, media, and microbiological techniques.

Yeast strains used in this study and their genotypes are listed in Table 1. Yeast media were prepared as described by Sherman et al. (57). Buffered medium was prepared as described by Yamashiro et al. (69), except that 50 mM MES (2-[N-morpholino]ethanesulfonic acid) and 50 mM MOPS (2-[N-morpholino]propanesulfonic acid) were used to buffer YEPD (yeast extract–peptone–2% dextrose [pH 7.5]), containing 50 mM CaCl₂. Sporulation medium was prepared as described by Klapholz and Esposito (31) except that p-aminobenzoic acid was omitted from the supplement mix.

TABLE 1.

Yeast strains and genotypes

Strain name	Description	Reference or source
SF838-5A	MATα ura3-52 leu2-3,112 his4-519 ade6	60
SF838-1D	MATα ura3-52 leu2-3,112 his4-519 ade6 pep4-3	60
SF838-1D vma2Δ	MATα ura3-52 leu2-3,112 his4-519 ade6 vma2Δ::LEU2	69
SF838-1D vma3Δ	MATα ura3-52 leu2-3,112 his4-519 ade6 vma3Δ::URA3	29
CJRY20-3B	MATα ura3-52 leu2-3,112 his3-Δ300 ade2-101 lys2-801	53
CJRY20-4B	MATa ura3-52 leu2-3,112 his3-Δ300 ade2-101 lys2-801	53
MEY14	MATα ura3-52 leu2-3,112 his4-519 ade6 pep4-3 vma41-1	This study
MEY32	MATα ura3-52 leu2-3,112 his4-519 ade6 pep4-3 vma43-1	This study
MEY69	MATa ura3-52 leu2-3,112 his4-519 ade6 pep4-3 vma45-1	This study
YOY69-1Aα	MATα ura3-52 leu2-3,112 HIS3 HIS4 ade2 ade6 pep4-3 vma45-1	This study
YOY69-1Ca	MATa ura3-52 leu2-3,112 his3-Δ300 HIS4 ade6 pep4-3 vma45-1	This study
SF838-5A kex2-Δ1	MATa ura3-52 leu2-3,112 his4-519 ade6 kex2Δ::LEU2	This study
YOY69	MEY69 X CJRY20-3Bα	This study
YOY11	SF838-5Aa kex2Δ X MEY69-1Aα/pYO19	This study

Chemical mutagenesis of whole yeast cells.

Cells (5 × 10⁷; 5 optical density at 600 nm [OD₆₀₀] units) were taken from a culture of budding yeast strain SF838-1D grown to saturation (∼7.6 OD₆₀₀ units/ml) and resuspended in 1.0 ml of phosphate-buffered saline (PBS), and 30 μl of ethyl methanesulfonate was added as a chemical mutagen (33). The culture was incubated at 30°C for 1 h. After incubation, an equal volume of 10% sodium thiosulfate (Na₂S₂O₃) was added for 10 min to quench the mutagen and the cells were harvested by centrifugation. Cells were washed twice with 5% Na₂S₂O₃ to ensure effective quenching and removal of the mutagen and then transferred to 30 ml of YEPD, pH 5, and incubated at 30°C for 24 h to allow recovery. This treatment resulted in 50 to 60% killing of the cells.

Enrichment and selection for cells showing a Vma⁻ phenotype.

Cells (5 × 10⁷) were allowed to recover from mutagenesis and then incubated in 10 ml of YEP (yeast extract-peptone) containing 3% glycerol and 2% ethanol for 16 h to allow vma mutants to increase in density (46). Four OD₆₀₀ units of the culture was carefully layered on a 95% isosmotic Percoll solution, centrifuged at 30,000 × g for 10 min, and then collected in 1-ml fractions, beginning from the top of the gradient (47).

Screening for the presence of the Vma⁻ phenotype.

Samples (∼10% of total) from fractions 5 to 8 of a population of chemically mutagenized cells fractionated as described above were plated on YEPD, pH 5, at a concentration giving ∼500 to 700 colonies per plate and incubated at 30°C until colonies developed. Approximately 25,000 colonies were then replica plated sequentially onto (i) SD (synthetic minimal medium containing 2% dextrose), (ii) YEPD at pH 7.5, (iii) YEPD containing 100 mM CaCl₂, and (iv) YEPD at pH 5. Replica plates were incubated at 30°C for 24 h and then scored. Colonies that were unable to grow on plates 1, 2, and 3 but were able to grow on plate 4 were selected as Vma⁻, retested, and then further screened for complementation of known vma mutants and quinacrine accumulation.

Complementation testing.

Where necessary, yeast mating type switching was performed as described previously (18). In order to assign the new vma mutants to complementation groups, yeast cells of opposite mating types were patched together on YEPD (pH 5) plates and incubated for 10 to 14 h to allow mating to occur. The patches were then replica plated onto pH 7.5, pH 5, or CaCl₂-containing YEPD as described above. Complementation was indicated by production of a diploid able to grow on YEPD buffered to pH 7.5 or containing 100 mM CaCl₂.

The MEY69 mutant strain, containing the vma45-1 mutation, was backcrossed twice to wild-type haploid strain CJRY20-3B to help eliminate background mutations resulting from the mutagenesis. The resulting diploid (YOY69) was sporulated, and the Vma⁻ haploid spore (YOY69-1Ca) was used for cloning of the VMA45 gene and subsequent biochemical analysis.

Cloning and subcloning of the VMA45 gene.

The yeast genomic library constructed by Scott Houtteman, University of Chicago, was obtained from Saul Honigberg at Syracuse University. The library was made by cloning a yeast partial Sau3A genomic DNA into the BamHI site of YCp50 and has an average insert size of 9 kb. Yeast transformation was performed as described previously (49). The VMA45 gene was cloned by complementation of the pH-dependent growth phenotypes of yeast strain YOY69-1Ca. Transformants were plated on supplemented minimal medium lacking uracil (SD−ura) buffered to pH 7.5 in order to select transformants that had acquired a URA3-containing plasmid capable of complementing the Vma⁻ growth phenotype. Plasmids were isolated from transformants capable of growth on SD−ura (pH 7.5) plates as previously described (61) and retransformed into YOY69-1Ca to confirm the phenotype. The complementing fragment from plasmid pMEY69-1 was further isolated by subcloning fragments into pRS316 (59) as shown in Fig. 3 and checking for complementation. All DNA manipulations were done as described by Sambrook et al. (55).

FIG. 3.

(A) Restriction map of pMEY69-1 and various subclones. Restriction endonuclease sites are indicated. B, BamHI; C, ClaI; E, EcoRI; H, HindIII; Rv, EcoRV; S, SalI; X, XbaI. The fragments indicated were subcloned into pRS316 to give the plasmids indicated at the right and then tested for complementation of the Vma⁻ growth phenotypes of YOY69-1Ca. The hatched box represents the complementing subclone, and arrows indicate the direction and extent of sequence determination. The KEX2 gene is shown as stippled box. (B) Disruption of the KEX2 gene. A 1,056-bp AgeI-HpaI fragment within the KEX2 ORF was replaced with a 2.2-kb fragment containing the LEU2 gene.

Disruption of the KEX2 gene.

A null kex2 strain was constructed by the one-step allele replacement method (54). The 1,056-bp AgeI-HpaI fragment within the KEX2 open reading frame (ORF) in plasmid pYO19 (see Fig. 3B) was replaced by a 2.1-kb HpaI fragment containing the LEU2 gene. The resulting plasmid, pYO53, was digested with XhoI and NotI to release the LEU2-disrupted allele from the vector, and the linear DNA fragment generated was used to transform yeast strain SF838-5Aa. Leu⁺ transformants were selected, and disruption of the KEX2 locus was confirmed by PCR from chromosomal DNA with oligonucleotides 5′-CGACCACATATTATCTGTCCA-3′ and 5′-GGATTCTAATGTCTCTTCCGT-3′. Isolation of yeast DNA for PCR analysis was carried out as described by Nasmyth and Reed (41) except that DNA was treated with RNase A for 25 min at 37°C and 5 min at 65°C before the final ethanol precipitation.

DNA sequencing.

Plasmid DNA for sequencing was purified with a QIAprep-spin plasmid kit from Qiagen. Sequencing was done by the dideoxy chain termination method (56) with a Sequenase sequencing kit and Sequenase version 2.0 enzyme (United States Biochemical) and ³⁵S-dATP. Oligonucleotides corresponding to the T3 and T7 promoter sequences in pRS316 were used as primers.

Tetrad analysis.

Haploid yeast strain YOY69-1Aα carrying the KEX2 gene on a plasmid (pYO19) was mated with haploid strain SF838-5Aa kex2-Δ1 on YEPD (pH 5) plates. The resulting diploid strain (YOY11/pYO19) was selected on supplemented minimal medium lacking both uracil and leucine. Diploids pregrown on YEPD, pH 5, for 24 to 30 h were plated on sporulation medium and incubated at 30°C for 5 days. Tetrads were dissected on YEPD, pH 5, plates. Colonies of germinating spores became visible in less than 2 days. To select for loss of plasmid pYO19, Ura⁺ spores were grown nonselectively on YEPD, pH 5, after which uracil auxotrophs were identified.

Quinacrine vital staining.

Vacuolar accumulation of quinacrine was assessed as described by Roberts et al. (52). Once stained, cells were visualized within 10 min with a Zeiss Axioskop Routine immunofluorescence microscope. Cells were viewed under Nomarski optics to observe normal cell morphology and under a fluorescein isothiocyanate filter with a 100× objective to observe vacuolar staining with quinacrine.

Western blotting.

Whole-cell lysates and solubilized vacuolar membrane vesicles were prepared, and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed as previously described (29). Immunoblots were probed with monoclonal antibodies 10D7, 7D5, 13D11, and 7A2 and polyclonal anti-27-kDa-subunit antisera against, respectively, the 100-, 69-, 60-, 42-, and 27-kDa subunits of yeast V-ATPase (22, 29).

Purification of yeast V-ATPase.

Solubilization of vesicles and purification of V-ATPase were performed basically as described previously (27, 65) with the following modifications. Four hundred micrograms of solubilized vesicles were layered on a 12-ml 20 to 50% (vol/vol) glycerol gradient and centrifuged at 200,000 × g for 8 h in a Beckman Ti-75 rotor. Sixteen 700-μl fractions were collected and analyzed for ATPase activity in order to identify the peak fractions. Fractions were diluted 1:1 with water, and protein was precipitated by addition of an equal volume of 20% trichloroacetic acid. Precipitated proteins were solubilized in 50 μl of cracking buffer (50 mM Tris-HCl [pH 6.8], 1 mM EDTA, 8 M urea, 5% SDS, 5% β-mercaptoethanol), separated on an SDS–10% polyacrylamide gel, and detected by silver staining (67).

Proton pumping.

Proton pumping activity was measured by monitoring quinacrine fluorescence quenching in an SLM spectrofluorimeter as described previously (30) with minor modifications. Briefly, vacuolar membrane vesicles (30 to 40 μg) suspended in proton transport buffer (15 mM MES-Tris [pH 7.0], 4.8% glycerol) were added to proton transport buffer containing 7 μM quinacrine, 2.5 mM MgSO₄, 66 mM KCl, and 0.25 μM valinomycin in a total volume of 1 ml. The mixture was allowed to equilibrate for 2 min. Proton pumping was initiated by the addition of ATP to give a final concentration of 2.5 mM. Quinacrine quenching was monitored at room temperature for 10 min with excitation and emission wavelengths of 420 and 490 nm, respectively. Under our experimental conditions, the initial rate of fluorescence quenching was very rapid. As a result, it was difficult to measure a true initial rate, so the extents of quenching were compared after the first 15 s. Readings were normalized relative to the total change in quinacrine fluorescence in the presence or absence of vesicles.

Other methods.

Monoclonal antibody 13D11 (against the 60-kDa subunit) was used to coprecipitate the V-ATPase complex as previously described (26) with the following modifications. In order to optimize association of the 38-kDa protein (see below) with V-ATPase in kex2 mutants, 2% n-octylglucoside was substituted for 1% C₁₂E₉ and 2 mM (final concentration) benzamidine was added to the protease inhibitor mixture. As described previously, all samples contained 0.67 mM dithiobis(succinimydylpropionate) as a cross-linking agent. For pulse-chase experiments, spheroplasts were labeled for 5 min (pulse), unlabeled methionine and cysteine were each added to a final concentration of 50 μg/ml, and the incubation was continued for varied times (chase). For samples with no chase, then, unlabeled methionine and cysteine were added immediately before the spheroplasts were harvested. Isolation of vacuolar vesicles was performed as described previously (52). The protein concentration was determined by the method of Lowry et al. (37). ATPase activity was measured in a coupled enzyme assay as described previously (36).

RESULTS

Genetic screen for vma mutants.

Ohya et al. have demonstrated that vma mutant cells do not continue to divide in medium containing a nonfermentable carbon source even though they continue to synthesize protein and grow in size (46). As a result, these cells become substantially larger and denser than wild-type cells upon incubation in medium containing nonfermentable carbon sources such as glycerol and ethanol. This phenotype was used to develop an enrichment procedure for Vma⁻ strains based on density selection by Percoll density gradient centrifugation. As demonstrated in Fig. 1, two different Vma⁻ strains (SF838-1Dα vma2Δ::LEU2 and SF838-1Dα vma3Δ::URA3) could be separated from a mixture containing a ninefold excess of the congenic wild-type strain by this procedure. Wild-type cells remained near the top of the gradient, but both types of vma mutant cells peaked in fractions 5 to 8. Although other mutations in a population of mutagenized cells may also cause an increase in cell density, Fig. 1 indicates that density fractionation can potentially enrich for vma mutants in a mutagenized population.

FIG. 1.

Density selection for vma mutants. Wild-type yeast strain SF838-1Dα and two previously identified vma mutants (SF838-1Dα vma2Δ::LEU2 and SF838-1Dα vma3Δ::URA3) were cultured (separately) to log phase in YEPD, pH 5.0. Cells were then transferred to YEP containing 3% glycerol and 2% ethanol and incubated at 30°C for 16 h. Wild-type, vma2Δ, and vma3Δ cells were mixed together in the ratio 18:1:1, respectively, to give a total of 10⁸ cells. The mixed culture was washed once with size selection buffer (0.67 g of yeast nitrogen base per liter, 0.25 M sorbitol, 10 mM Tris-HCl [pH 7.5]) and resuspended in 500 μl of the same buffer. Two hundred microliters (4 × 10⁷ cells) was carefully layered on 10 ml of a 95% isosmotic Percoll solution and centrifuged at 30,000 × g for 10 min. The percentage of wild-type cells (▾) and vma mutant cells (□) present in each fraction was evaluated by plating a sample from each fraction on YEPD, pH 5.0, and then replica plating the samples to appropriate selective media in order to identify the percentages of Ura⁺ and Leu⁺ colonies (representing the vma mutants). Fractions were collected as 1-ml aliquots, beginning from the top of the gradient. The density gradient is indicated by the dashed line. Relative density of each fraction was measured with a handheld refractometer.

Haploid yeast strain SF838-1Dα was chemically mutagenized with ethyl methanesulfonate as described in Materials and Methods. The mutagenized cells were allowed to recover in YEPD, pH 5.0, and then transferred to medium containing the nonfermentable carbon sources glycerol and ethanol in order to allow any vma mutants to increase in density. The mutagenized cells were then fractionated on a Percoll gradient similar to that shown in Fig. 1, and fractions 5 to 8 were analyzed further for the presence of mutants exhibiting Vma⁻ growth phenotypes, including the inability to grow in medium buffered at pH 7 or above, in medium containing 100 mM CaCl₂, or in medium containing a nonfermentable carbon source. From an initial collection of over 25,000 colonies, 98 potential Vma⁻ strains were identified. After two more rounds of screening and loss of some mutants by reversion, 52 mutants were tested for complementation of existing Vma⁻ strains. Two vma1 alleles, one vma3 allele, one vma4 allele, and one vma12 allele were identified in the collection of mutants. The remaining mutants were tested for the ability to accumulate quinacrine in the vacuole as a measure of vacuolar acidification. All known vma mutants are unable to accumulate the fluorescent weak base quinacrine into their vacuoles due to loss of acidification (22, 69). Of the 47 mutants tested, five completely failed to accumulate quinacrine in their vacuoles and were selected for further studies. (A number of other mutants exhibited partially defective quinacrine uptake.) Complementation testing was performed by crossing each of the five mutants with the others and indicated that the five mutants represent five different complementation groups, which we designated vma41 to -45. Three of the new mutant strains (MEY69, MEY32, and MEY14, containing the vma45-1, vma43-1, and vma41-1 mutations, respectively) were backcrossed to CJRY20-3Bα or CJRY20-4Ba, and Vma⁻ spores obtained from sporulation of the resulting diploids, identified by the inability to grow on YEPD, pH 7.5 (Fig. 2), were selected for further characterization. In all cases, the Vma⁻ phenotype segregated 2:2.

FIG. 2.

Growth phenotypes of wild-type and vma cells. Cells were streaked on the medium indicated and incubated at 30°C for 2 days. The vma45 strain is YOY69-1Ca, and the vma41 and vma43 strains represent Vma⁻ spores derived from a single backcross of the original mutants. The wild-type strain is SF838-1Dα, and the vma12-1 strain represents a new allele of VMA12 identified in the screen described here.

In order to examine the stability of several known subunits of V-ATPase in the mutant strains, whole-cell protein extracts were prepared and analyzed by Western blotting with antibodies against five different V₁ (peripheral) subunits of V-ATPase, Vma1p, Vma2p, Vma4p, Vma5p, and Vma13p, and the V-ATPase assembly protein Vma12p (21, 22, 69). All these proteins appeared to be present at wild-type levels in the mutant cells (data not shown), indicating that the steady-state levels of these polypeptides are not affected in the new vma mutants.

Cloning of the VMA45 gene.

The wild-type VMA45 gene was cloned by complementation of the pH-dependent growth phenotype of the haploid yeast strain YOY69-1Ca containing the vma45-1 allele. YOY69-1Ca mutant cells were transformed with a yeast genomic library carried on the yeast low-copy-number plasmid YCp50 and plated directly on SD−ura buffered at pH 7.5 to select for those transformants bearing plasmids able to restore growth at pH 7.5 to the mutant cells. Of an estimated 35,000 Ura⁺ transformants, 10 were found to be Ura⁺ Vma⁺, and 6 of these contained the same plasmid, pMEY69-1. Of the remaining four transformants, three contained the same plasmid, pMEY69-5, and one contained plasmid pMEY69-7. Complementation for the vma phenotypes of the vma45-1 strain was plasmid dependent; pMEY69-1 fully complemented the growth phenotypes of the mutant strain, while plasmids pMEY69-5 and pMEY69-7 gave only partial complementation (data not shown).

Plasmid pMEY69-1, which gave the best complementation of the vma45-1 mutant growth phenotype, contained a 6.5-kb insert and was mapped as shown in Fig. 3A. Several different subclones were generated in the yeast shuttle vector pRS316 and tested for complementation. As shown in Fig. 3, a 3.3-kb EcoRI fragment (pYO19) of the insert was sufficient for complementation. A 480-bp region internal to this fragment was sequenced and used to search for homology to any sequences in the GenBank and EMBL databases. The sequence comparison (data not shown) indicated identity to the yeast KEX2 gene, which encodes a neutral serine protease localized to the late Golgi compartment (51), covering the entire sequenced region of VMA45. Comparison of the restriction maps of KEX2 and pYO19 also revealed a perfect match (Fig. 3A) but indicated that pYO19 lacked 300 bp from the C terminus of the KEX2 ORF. Previous reports of cloning of the KEX2 gene had also indicated that this fragment could fully complement a number of kex2 mutant phenotypes (24, 34). Restriction mapping of pMEY69-5 and pMEY69-7 indicated no overlap with pMEY69-1 or between the two plasmids.

A LEU2-disrupted copy of the KEX2 gene was constructed (Fig. 3B), and the chromosomal KEX2 locus was disrupted by a one-step gene replacement method (54). Disruption of the KEX2 gene was confirmed by PCR analysis of genomic DNA.

kex2-Δ1 mutants display Vma⁻ phenotypes.

Previous studies have indicated that kex2 mutants display pleiotropic phenotypes (5, 32, 34) but failed to indicate any link to vacuolar acidification. Similarly, previous studies of yeast V-ATPase did not suggest an interaction with the Kex2p endoprotease or any other protease. We therefore addressed first the issue of whether there was any overlap between the previously described growth phenotypes of kex2 and vma mutants. All vma mutants, except for vph1Δ and stv1Δ single mutants (38, 39), have been shown to grow more slowly than wild-type cells in unbuffered YEPD and to show no growth on YEPD buffered to pH 7.5, YEPD containing 4 mM ZnCl₂, or YEP containing glycerol and ethanol as the sole carbon sources (39, 43, 46, 69). As shown in Table 2, the Vma⁻ growth phenotypes of the vma45-1 and kex2-Δ1 mutants were identical to those of the vma3Δ mutant strain. kex2 null mutants have been reported to be cold sensitive (32, 40), so we tested whether vma mutants were also cold sensitive. After 72 h on unbuffered YEPD at 17°C, the vma3Δ, kex2-Δ1, and vma45-1 mutant strains exhibited no growth (Table 2). Interestingly, the cold-sensitive phenotype of the mutants proved to be pH dependent. Although all three mutants failed to grow on YEPD at 17°C, they grew quite well on YEPD, pH 5, plates at 17°C (Table 2). This pH-dependent cold sensitivity was observed for other vma mutants tested, strongly suggesting that the cold sensitivity of kex2 is a Vma⁻ phenotype. Two multicopy suppressors of the cold-sensitive phenotype and α-factor processing defect of kex2 null mutants, YAP3 and MKC7, have been identified (32). In order to determine whether these genes can also rescue the Vma⁻ growth phenotypes of kex2 mutants, kex2-Δ1 cells were transformed with the YAP3 or MKC7 gene on a multicopy plasmid (2μm) and the transformants were tested for pH-sensitive growth. The results are shown in Fig. 4. As described above, the kex2-Δ1 mutants could not grow on SD−ura plates buffered to pH 7.0 (Fig. 4, right plate, top left quadrant). This pH-dependent growth defect was fully complemented by the pYO19 plasmid (top right quadrant), suppressed quite well by 2μm-MKC7 (lower right quadrant), and weakly suppressed by 2μm-YAP3 (lower left quadrant). 2μm-MKC7 also partially suppressed the growth defect of kex2-Δ1 cells at pH 7.5, but the 2μm-YAP3 transformants could not grow under these conditions. These results indicate that YAP3 and MKC7 are multicopy suppressors of the Vma⁻ growth phenotypes of kex2 mutants. MKC7 appears to be a stronger suppressor of the pH-sensitive growth phenotype than YAP3; Komano and Fuller (32) found that MKC7 also suppressed the cold sensitivity of kex2 mutants more effectively than YAP3.

TABLE 2.

Overlap of known kex2 and vma phenotypes

Testa	Temp (°C)	Test resultb with:
		Wild-type cells	vma3Δ cells	kex2Δ cells	vma45-1 cells
Growth on:
YEPD	30	+++	+	+	+
YEPD, pH 5.0	30	+++	++	++	++
YEPD, pH 7.5	30	+++	−−	−−	−−
YEPD + 4 mM Zn2+	30	+++	−−	−−	−−
YEP-glycerol	30	+++	−−	−−	−−
YEPD	17	+++	−−	−−	−−
YEPD, pH 5.0	17	+++	++	++	++
YEPD, pH 7.5	17	+++	−−	−−	−−
Vacuole staining with quinacrine	25	+++	−−	−−	−−

^aCells were streaked on the indicated medium and incubated for 36 h (for 30°C samples) or 72 h (for 17°C samples). Quinacrine staining was performed as described in the legend to Fig. 5. 

^b+++, abundant growth; ++, moderate growth; +, poor growth; −−, no growth. 

FIG. 4.

Multicopy MKC7 and YAP3 suppress the growth defects of kex2-Δ1 mutants at pH 7.0. SF838-5Aa kex2Δ mutant cells were transformed with pYO19 (wild-type KEX2 on a low-copy-number plasmid), MKC7 on a 2μm plasmid, YAP3 on a 2μm plasmid, or YEp24 (the 2μm plasmid with no insert). Transformants were initially identified by growth on unbuffered SD−ura (pH approximately 5.7) and then streaked to unbuffered SD−ura (left plate) and grown for 3 days at 30°C or to SD−ura buffered to pH 7.0 (right plate) and grown for 5 days at 30°C. Growth of kex2-Δ1 cells transformed with the following plasmids is shown (clockwise from top): pYO19, 2μm-MKC7, 2μm-YAP3, and YEp24.

Our initial characterization of the vma45-1 mutant strain indicated that it was unable to accumulate quinacrine in the vacuole. In order to test whether kex2-Δ1 cells are able to acidify their vacuoles, we examined quinacrine uptake in these cells. Our results, shown in Fig. 5, indicate that, like the vma45-1 mutant strain YOY69-1Ca, kex2Δ cells do not accumulate quinacrine and thus appear to be defective in vacuolar acidification. Together, these results demonstrate that kex2Δ mutants behave as true vma mutants.

FIG. 5.

Loss of vacuolar acidification in kex2 mutants. Vacuolar acidification was assessed by quinacrine accumulation in the vacuole as described in Materials and Methods. Log-phase yeast cells were incubated in 500 μl of PBS, pH 7, containing 200 μM quinacrine for 5 min at 30°C. After being stained, cells were washed with 500 μl of PBS and resuspended in 100 μl of the same buffer. Cells were viewed with differential interference contrast optics for observation of normal cell morphology and by fluorescence microscopy with a fluorescein isothiocyanate filter for observation of vacuolar staining with quinacrine. Each monograph is a composite of three to four fields. The following strains were used: SF838-5Aa (wild type), SF838-5Aa kex2-Δ1 (kex2Δ), and YOY69-1Ca (vma45).

Both the Vma⁻ growth phenotypes and the loss of vacuolar quinacrine staining in the kex2-Δ1 mutant were fully rescued by the plasmids pMEY69-1 and pYO19. These plasmids restored normal, pH-independent growth to the vma45-1 mutant strain YOY69-1Ca but failed to fully restore quinacrine staining to this mutant (data not shown). Plasmids pMEY69-5 and pMEY69-7, which gave partial complementation of the vma45-1 growth phenotype, did not complement the kex2Δ mutant phenotypes. These results indicate that vma45-1 is probably not a null mutant and that the pMEY69-5 and pMEY69-7 plasmids behave as allele-specific suppressors.

VMA45 is allelic to KEX2.

To confirm that vma45-1 is indeed a mutant allele of KEX2, haploid strain MEY69-1Aα, carrying a plasmid-borne copy of the wild-type KEX2 gene (pYO19), was mated to SF838-5Aa kex2Δ::LEU2 to give diploid strain YOY11/pYO19. Since kex2 mutants are α-specifically sterile (34) and we had also observed that MEY69-1Aα appears to be sterile, it was necessary to introduce a plasmid-borne KEX2 gene into MEY69-1Aα before the mutants were mated. When the YOY11 diploid cells were cured of the pYO19 plasmid, they became unable to grow on YEPD (pH 7.5) plates, indicating that the kex2-Δ1 mutant is unable to complement the vma growth phenotypes of the vma45-1 mutant strain. When YOY11 diploid cells lacking the pYO19 plasmid were patched on sporulation medium and incubated at 30°C, no tetrads were detected even after 2 weeks, in agreement with previous results showing that kex2 homozygous diploids are deficient in sporulation (34). YOY11/pYO19 cells were able to sporulate, tetrads were dissected after 5 days on sporulation medium, and the spores were allowed to germinate on YEPD, pH 5. Tetrad analysis suggested a 4:0 segregation of the Vma⁻ phenotype, since all Ura⁻ spores are Vma⁻ and all Ura⁺ spores are Vma⁺. To confirm these results, four Ura⁺ spores from two different tetrads were cured of the plasmid. In all cases, Ura⁻ colonies became Vma⁻, confirming that the Vma⁻ phenotype segregates 4:0 in YOY11.

Characterization of V-ATPase from kex2-Δ1 cells.

Western blot analysis of whole-cell lysates showed that the steady-state levels of several V-ATPase subunits are not affected in the vma45-1 mutant (see above). Similar experiments were performed to determine the levels of several subunits of the V-ATPase in kex2-Δ1 cells. Immunoblot analysis revealed that the levels and apparent molecular masses of the 69-, 60-, 54-, 42-, and 27-kDa V₁ subunits were not altered in whole-cell lysates from the kex2-Δ1 cells relative to those of the wild type. Similarly, 25-kDa Vma12p, implicated in the assembly of V-ATPase, did not appear to be affected in these cells (data not shown).

A number of mutants that contain near-normal cellular levels of V-ATPase subunits show little or no assembly of the ATPase complex (6). We examined the assembly of V-ATPase in kex2-Δ1 mutant cells by immunoprecipitating the ATPase complex under nondenaturing conditions (6, 26). Cells were converted to spheroplasts and then biosynthetically labeled with Tran³⁵S-label for 5 min in order to examine the early steps in assembly of the complex or for 60 min in order to examine the final assembled complex. Immunoprecipitations were carried out with a monoclonal antibody against the 60-kDa V₁ subunit of ATPase (13D11). This antibody recognizes the 60-kDa V₁ subunit by itself, as part of a V₁ subcomplex or as part of a fully assembled V₁V₀ complex, and can therefore coprecipitate the whole V-ATPase complex under nondenaturing conditions. Coimmunoprecipitated proteins were separated by SDS-PAGE and detected by autoradiography. Comparison of immunoprecipitated proteins from the 60-min labeling experiments whose results are shown in Fig. 6B (60-min pulse, 0-min chase) indicates that all of the previously identified subunits of the yeast V-ATPase that are immunoprecipitated from the wild-type cells are also immunoprecipitated from the kex2-Δ1 cells, suggesting that the mutant cells do not have an assembly defect. Similarly, there does not appear to be any drastic difference in the kinetics of assembly for the two strains, based on the 5-min pulse and 0-min chase, 5-min pulse and 5-min chase, and 5-min pulse and 15-min chase samples (Fig. 6A). One detectable difference was the presence of an extra protein band of ∼38 kDa (Fig. 6) which is coprecipitated with the V-ATPase from kex2-Δ1 cells. This band had a relative mobility between those of the previously identified 42- and 36-kDa subunits of V-ATPase. The 38-kDa band is also coimmunoprecipitated with V-ATPase in vma45-1 mutant cells (data not shown). In addition, a protein with a relative molecular mass between 17 and 27 kDa also appears to be specifically coprecipitated from kex2Δ cells.

FIG. 6.

Assembly of V-ATPase in wild-type and kex2Δ mutant cells. Nondenaturing immunoprecipitation of V-ATPase from biosynthetically labeled yeast cells was performed as described previously (34). Monoclonal antibody 13D11 against the 60-kDa peripheral V₁ subunit was used for immunoprecipitation. Immunoprecipitated proteins were separated by SDS-PAGE and visualized by autoradiography. The positions of previously identified subunits of V-ATPase are indicated. The arrow indicates the 38-kDa band that is present in kex2Δ mutant strains but not in wild-type strains. Positions of protein molecular mass standards are indicated on the left. The strains used are the same as those used in the experiment shown in Fig. 5. (A) Steps in V-ATPase assembly (5-min pulse, varied chase times); (B) final assembled V-ATPase complex (60-min pulse, 0-min chase).

In order to address the vacuolar targeting and catalytic activity of V-ATPase in the kex2-Δ1 mutant cells, vacuolar vesicles were isolated from the kex2-Δ1 mutant cells and the congenic wild-type strain. The concanamycin A-sensitive ATPase activities in the isolated vacuolar membranes are compared in Table 3. Concanamycin A is a very potent and specific inhibitor of V-ATPases (8). As shown in Table 3, vacuolar vesicles from kex2-Δ1 cells have the same level of V-ATPase activity as that of the vesicles from wild-type cells in vitro, even though the growth phenotypes and lack of quinacrine accumulation strongly suggest that V-ATPase is not active in vivo. Proton pumping activity of vacuolar membrane vesicles was also measured by examining the Mg-ATP-dependent quenching of quinacrine fluorescence in the isolated vesicles (Table 3). The vesicles isolated from kex2-Δ1 cells showed an initial rate and extent of quinacrine quenching similar to or somewhat greater than those of wild-type vesicles (Table 3), indicating that ATP hydrolysis and proton pumping have not been uncoupled in the kex2-Δ1 mutant.

TABLE 3.

V-ATPase and proton pumping activities of vacuolar membrane vesicles isolated from wild-type and kex2-Δ1 cells

Strain	V-ATPase activity (μmol/min/mg of protein)a	% of wild-type activity	Proton pumping (relative level of quenching
			Rate in first 15 s (min−1)	Extent after 10 min
SF838-5A (wild type)	3.0 ± 0.2 (3)	100	7.2	2.1
SF838-5A kex2-Δ1	3.1 ± 0.5 (3)	103	7.3	2.4

^aVacuolar membrane vesicles were prepared as described in Materials and Methods, and the ATPase activity that is sensitive to 100 nM concanamycin A was determined. Activities are presented as means ± standard deviations (with the number of samples in parentheses). 

^bLevels of ATP-dependent proton pumping were determined by monitoring quinacrine fluorescence quenching. Activity is expressed as (arbitrary) fluorescence units/mg of protein, normalized to total fluorescence as described in Materials and Methods. 

We investigated whether there was a structural defect in V-ATPase in kex2-Δ1 mutant vacuoles (for example, altered subunit stoichiometry or the presence of extra inhibitory proteins) by two different approaches. First, the levels of various subunits in isolated vacuolar membrane vesicles were analyzed by Western blotting with antibodies against the 100-kDa integral membrane subunit and the 69-, 60-, 42-, and 27-kDa peripheral subunits of V-ATPase. As shown in Fig. 7A, all of the V-ATPase subunits monitored are present in kex2-Δ1 vacuolar membranes at levels similar to those of the wild type. Second, we solubilized vacuolar membrane vesicles and purified V-ATPase by glycerol gradient centrifugation (28, 65). ATPase activities fractionated to similar positions in gradients of the wild-type and mutant vacuolar membranes. Proteins from the fractions with peak V-ATPase activities were separated by SDS-PAGE and detected by silver staining. As shown in Fig. 7B, the subunit composition of the V-ATPase isolated from kex2-Δ1 mutant cells is indistinguishable from that of the V-ATPase from wild-type cells.

FIG. 7.

V-ATPase isolated from kex2-Δ1 vacuolar membranes is indistinguishable from that isolated from the wild type. (A) Vacuolar membrane vesicles were incubated in cracking buffer (50 mM Tris-HCl [pH 6.8], 8 M Urea, 1 mM EDTA, 5% SDS, 5% β-mercaptoethanol) for 20 min at 50°C for the 100-kDa V₀ subunit or 70°C for the remaining subunits. Fifteen micrograms (for detection of the 100-kDa subunit) or 3 μg (for detection of all other subunits) of vacuolar protein was loaded in each lane. Proteins were detected on Western blots with alkaline phosphatase-conjugated antibodies. (B) V-ATPase was purified from vacuolar membrane vesicles as described in Materials and Methods. Proteins in fractions with peak ATPase activities were separated on an SDS–10% polyacrylamide gel and visualized by silver staining. Positions of known V-ATPase subunits are indicated on the right. The asterisk indicates the position of an unidentified protein that consistently copurifies with V-ATPase activity. The positions of protein molecular weight standards (in thousands) are indicated on the left.

DISCUSSION

kex2 mutants exhibit Vma⁻ phenotypes in vivo but functional V-ATPases in vitro.

We have defined five new vma complementation groups based on the set of well-characterized Vma⁻ phenotypes used in previous screens by using a screening process that contains an initial step designed to enrich for vma mutants based on density differences following growth on a nonfermentable carbon source. The Vma⁻ growth defects, including pH-dependent growth (43, 69), Ca²⁺ sensitivity (46), Zn²⁺ sensitivity (3), and inability to utilize nonfermentable carbon sources (46), have proven to be highly diagnostic of defects in V-ATPase activity, particularly when these defects are combined with a loss of vacuolar acidification assessed with lysosomotropic dyes. With the exception of the vph1Δ mutant (38), mutants lacking each of the 13 cloned subunits of V-ATPase exhibit the full set of Vma⁻ growth defects as well as a loss of vacuolar acidification (references 2 and 15 and references therein; see also references 27 and 63). Several other mutants that exhibit Vma⁻ growth phenotypes have subsequently been shown to affect genes that do not encode subunits of V-ATPase but that are nevertheless essential for the assembly and/or activity of the enzyme (VMA12 or VPH2 [2, 21], VMA21 [19], VMA22 [20], and VPH6 [17]). Furthermore, analysis of point mutations causing partial defects in V-ATPase activity (35) has demonstrated that the onset of the pH-dependent growth phenotype requires the loss of at least 75% of V-ATPase activity at the vacuole. Therefore, the growth phenotypes and lack of quinacrine accumulation in the new mutants reported here, and particularly in the vma45-1 and kex2-Δ1 mutants, indicate that V-ATPase activity is seriously compromised in these mutants in vivo.

In this study, we have focused on the new vma45-1 allele generated in our mutant screen, demonstrated by a number of criteria that it is a mutant allele of the previously characterized KEX2 gene, and compared the genetic and biochemical characteristics of the vma45-1 strains with those of a kex2-Δ1 mutant strain. All of the data indicate that a functional Kex2 protein is essential for maintaining vacuolar acidification in vivo. In addition, an essential role for Kex2p in vacuolar acidification may help explain several of the previously reported pleiotropic phenotypes of kex2 mutants that could not be accounted for by the previously characterized functions of Kex2p (5, 32, 40). We demonstrate here that a vma3Δ mutant, which lacks the proteolipid subunit of V-ATPase (42), exhibits a similar cold sensitivity to kex2-Δ1 mutants and that this cold sensitivity is pH dependent. Komano and Fuller (32) have demonstrated that the growth arrest at 16°C of kex2Δ mutants is accompanied by aberrant cell morphologies, including formation of multiple buds, actin and chitin delocalization, and a substantial increase in cell volume. We have recently found that vma mutants exhibit similar morphological changes and alterations in actin and chitin delocalization when they are incubated at elevated pH (71), suggesting that the morphological defects of the kex2-Δ1 mutant in YEPD at 16°C may also be linked to its role in vacuolar acidification. In addition, kex2Δ strains have been shown to be hypersensitive to the drug quinidine, a weak base (5). Since quinidine appears to accumulate in acidic compartments in wild-type cells (5) and loss of vacuolar acidification would prevent this accumulation, one consequence of vacuolar acidification defects of kex2-Δ1 mutants might be to increase the effective concentration of quinidine in cytoplasm, resulting in hypersensitivity.

Despite the evidence that Kex2p is essential for normal vacuolar acidification, the data also present a paradox. All of the other mutants expressing a Vma⁻ phenotype that have been analyzed also show defects in V-ATPase activity in vitro, in purified vacuoles, and/or in the purified enzyme complex. In contrast, assembly of the V-ATPase complex did not seem to be affected significantly in kex2-Δ1 mutants (Fig. 6), isolated vacuolar vesicles from kex2Δ cells contained levels of V-ATPase and proton pumping activity comparable to those of vesicles from wild-type cells (Table 3), and the purified V-ATPase complex from a kex2-Δ1 mutant showed no obvious structural differences from the wild-type complex (Fig. 7). These results suggest that the effects of kex2 mutations on the V-ATPase are exerted in the intact cell and that the ATPase is present in the mutant strain in a fully assembled state, potentially capable of hydrolyzing ATP. This is the first time that a mutation which confers the characteristic Vma⁻ phenotypes in vivo but has no effect on the assembly or in vitro activity of V-ATPase has been identified. It is possible that a factor required for in vivo but not in vitro activity of V-ATPase is missing or defective in the kex2Δ mutant.

How might Kex2p regulate vacuolar acidification in vivo?

Based on the data shown here, it is possible that Kex2p plays a positive role in regulating vacuolar acidification or V-ATPase activity in vivo. It is not clear how V-ATPases are regulated in any system, and the available data indicate that multiple mechanisms are probably important in regulating activity (reviewed in reference 10). The yeast KEX2 gene product was first identified as the endopeptidase required for the processing of yeast prepro-α-factor and K1 killer toxin (25, 34). Kex2p is a highly conserved serine endoprotease localized to the late Golgi compartment (11, 51) that specifically cleaves proprotein substrates on the carboxyl sides of pairs of basic residues (preferentially -Lys-Arg- and -Arg-Arg-; reviewed in reference 12). Pleiotropic phenotypes of kex2 mutant strains suggest that other substrates have yet to be identified (5, 32, 40).

There are several potential models that may explain how Kex2p activity might be involved in regulation of vacuolar acidification in vivo without apparently affecting the in vitro activity of yeast V-ATPase. An explanation that is consistent with our data is that Kex2p might activate V-ATPase by processing a negative regulator of the enzyme (protein X). In such a model, protein X would be assembled initially as part of a V-ATPase (sub)complex and would inhibit any activity of the enzyme. It would then be removed in wild-type cells as a result of Kex2p processing in the late Golgi compartment. In this model, the V-ATPase would be activated in the Golgi apparatus; the site of activation would be the point at which Kex2p is localized and the earliest point at which organellar acidification is detected (1). In the absence of a functional Kex2p, V-ATPase would be assembled with protein X, protein X would not be processed, and the enzyme would be transported in a nonfunctional form to the vacuole. If protein X is loosely associated with the V-ATPase, our lysis conditions are not optimal for the association, or if other proteases active when cells are disrupted can replace Kex2p in processing, protein X might easily be lost during the process of cell lysis and vacuolar isolation, thus giving rise to a fully active V-ATPase in vitro.

This model is speculative, but several pieces of data lend it credibility. The experiment shown in Fig. 6, in which an extra 38-kDa protein band was seen to be associated with the V-ATPase from kex2-Δ1 cells, supports the model, although further experiments will be necessary to address directly whether this protein is a Kex2p substrate and a V-ATPase inhibitor. The protein of approximately 20 kDa that is coprecipitated from kex2Δ cells in Fig. 6 is also a candidate for an ATPase inhibitor, but this protein is also present in active, gradient-purified, V-ATPase from kex2Δ cells (Fig. 7), suggesting that it is not an inhibitor. Studies of V-ATPase from bovine chromaffin granules and kidney microsomes (62) indicated the presence of an accessory subunit (Ac45) in the V₀ sector of the enzyme, which appears to undergo posttranslational proteolytic processing. Ac45 was shown to be oriented with the bulk of the protein lying in the lumen of the vacuolar network (62), and the predicted lumenal domain contains a dibasic site (-Lys-Lys-) 60 residues from the C terminus that may potentially be recognized by a mammalian Kex2p homolog (mammalian Kex2p homologs cleave at -Lys-Lys- [7]). The Ac45 protein was shown to be present at an approximately 1:1 stoichiometry with the V-ATPase complex in chromaffin granules (62), but the specific effects of Ac45 association on activity of the chromaffin granule ATPase have not been examined. A search for proteins with sequence homology to Ac45 in the yeast genome database revealed no obvious candidates, but this finding does not eliminate the possibility that yeast cells have a functional homolog. Although the Ac45 protein has not been shown to be an inhibitor of V-ATPase, specific inhibitors of V-ATPases in bovine kidney microsomes have been identified (70). Zhang et al. have described a small (6.3-kDa) inhibitory protein isolated from a cytosolic fraction that probably acts as a dimer (70). Since this inhibitor is a cytosolic protein, it seems unlikely that it is a substrate for a Kex2p-related protease itself, but its action may be regulated indirectly by a Kex2p-like protease. A small, soluble inhibitor protein (IF₁) has also been identified in evolutionarily related F-type ATPases from different sources. IF₁ was shown to bind F₁F₀ in a 1:1 stoichiometry and completely inhibit enzyme activity (16).

Physiological implications of a role for Kex2p in regulation of vacuolar acidification.

This is the first report implicating a protease in regulation of vacuolar acidification and V-ATPase. It is difficult to demonstrate directly that the protease activity of Kex2p is essential for its role in vacuolar acidification, because Kex2p autocatalytic processing appears to be essential for its activity (13). However, Komano and Fuller have demonstrated that the cold-sensitive phenotypes of a kex2Δ mutant can be suppressed by multiple copies of two other proteases, Mkc7p and Yap3p, which are also capable of cleaving at clusters of basic residues, indicating that proteolysis of a single substrate or redundant substrates is essential for growth on unbuffered YEPD at 16°C (32). We have demonstrated that kex2-Δ1 cold sensitivity is pH dependent and is related to its Vma⁻ phenotypes. Furthermore, we show that the Vma⁻ growth defects of kex2 mutants can be suppressed by multiple copies of YAP3 or MKC7. Together, these results indicate that a substrate essential for vacuolar acidification requires proteolytic processing.

A role for Kex2p activity in regulation of vacuolar acidification makes physiological sense for several reasons. As described above, the localization of Kex2p is well-suited to activation of V-ATPase activity, since the late Golgi apparatus is the earliest compartment of the secretory pathway shown to have an acidic pH (1). In activating V-ATPase in mammalian cells, Kex2p activity, or the activities of related proteases in other cell types, might serve to enhance proteolytic activity toward other substrates as well. Yeast Kex2p has been shown to correctly process prohormones in mammalian cells (64), indicating that Kex2p activity is highly conserved in evolution. Mammalian prohormone processing is initiated in the late Golgi or trans-Golgi network, where Kex2p is localized, and dependent on the action of V-ATPase (68). Although the function of low pH in prohormone processing has not been fully characterized, it has been suggested that the processing enzymes may require either a low pH for optimum activity or a pH-dependent conformational change in the substrate for recognition and binding (68). In this context, Kex2p activity seems to be strategically poised to serve a function in promoting organelle acidification by V-ATPases, which in turn enhance Kex2p function toward other substrates.

Future studies will be directed toward further defining the functional relationships among yeast V-ATPase, vacuolar acidification, and the Kex2 endoprotease. To this end, we will be particularly interested in defining the Kex2p substrate that affects vacuolar acidification and in testing the hypothesis that this substrate inhibits the potentially active V-ATPase in kex2 mutant cells.