|Home | About | Journals | Submit | Contact Us | Français|
The Golgi-localized Ca2+- and Mn2+-transporting ATPase Pmr1 is important for secretory pathway functions. Yeast mutants lacking Pmr1 show growth sensitivity to multiple drugs (amiodarone, wortmannin, sulfometuron methyl, and tunicamycin) and ions (Mn2+ and Ca2+). To find components that function within the same or parallel cellular pathways as Pmr1, we identified genes that shared multiple pmr1 phenotypes. These genes were enriched in functional categories of cellular transport and interaction with cellular environment, and predominantly localize to the endomembrane system. The vacuolar-type H+-transporting ATPase (V-ATPase), rather than other Ca2+ transporters, was found to most closely phenocopy pmr1Δ, including a shared sensitivity to Zn2+ and calcofluor white. However, we show that pmr1Δ mutants maintain normal vacuolar and prevacuolar pH and that the two transporters do not directly influence each other's activity. Together with a synthetic fitness defect of pmr1ΔvmaΔ double mutants, this suggests that Pmr1 and V-ATPase work in parallel toward a common function. Overlaying data sets of growth sensitivities with functional screens (carboxypeptidase secretion and Alcian Blue binding) revealed a common set of genes relating to Golgi function. We conclude that overlapping phenotypes with Pmr1 reveal Golgi-localized functions of the V-ATPase and emphasize the importance of calcium and proton transport in secretory/prevacuolar traffic.
A family of ATP-powered pumps drive the uphill transport of Ca2+ across membranes to maintain resting cytosolic Ca2+ at submicromolar levels as a prerequisite for ubiquitous and diverse signaling events essential to biology, such as sperm motility, fertilization, T-cell activation, synaptic vesicle fusion, and muscle contraction, to name a few. One such Ca2+-ATPase localizing to the Golgi is evolutionarily conserved from yeast to mammals, and it plays a critical role in secretory pathway functions, including protein processing, sorting, and glycosylation. A unique property of the secretory pathway Ca2+-ATPase (SPCA) is that it also transports Mn2+ with high affinity, serving the dual function of Mn2+ detoxification via exocytosis and providing Mn2+ for Golgi-localized enzymes such as Kex2 protease and mannosyltransferases (Ton et al., 2002 ). Disruption of one allele of ATP2C1, encoding human SPCA1, leads to Hailey Hailey disease (HHD), an ulcerative skin disorder characterized by disruption of desmosomal contacts in keratinocytes (Foggia and Hovnanian, 2004 ). Whereas the molecular etiology of the disease is not clear, it is thought that in HHD patients abnormally high cytosolic ion concentrations, abnormally low luminal ion concentrations, or both, could lead to defective expression, modification, and trafficking of desmosomal proteins. Homozygous null mutations in ATP2C1 seem to be inviable in mammals, although they are tolerated in lower eukaryotes, including fungi and Caenorhabditis elegans (Rudolph et al., 1989 ; Cho et al., 2005 ), where compensatory mechanisms presumably suffice to allow viability. A particularly tractable model for understanding such mechanisms is the Saccharomyces cerevisiae orthologue Pmr1 (for review, see Ton and Rao, 2004 ). The identification of diverse pmr1 mutant phenotypes in yeast has been invaluable in guiding studies on metazoan SPCA orthologues (e.g., see Cho et al., 2005 ; Ramos-Castaneda et al., 2005 ), although we cannot yet understand the complex network of gene interactions that lead to many of these phenotypes, including HHD in human. The use of ion-supplemented media (Durr et al., 1998 ) and ion-selective Pmr1 mutants (Mandal et al., 2000 ), provides the opportunity to distinguish between Ca2+- and Mn2+-specific phenotypes.
Conventional genome-wide approaches have identified global patterns of gene expression (transcriptome), physical interactions between gene products (proteome), gene interactions (e.g., synthetic lethal screens), and enzyme function (metabolome). By analogy, the phenome describes genome-wide phenotypic profiles, usually of growth. Herein, we analyze the yeast phenome to find pathways and genes relating to the cellular function of Pmr1. We found that any single phenotypic screen of the haploid yeast deletion collection elicited a broad response, with genes functioning in diverse pathways, making it difficult to discern the underlying gene network leading to the common phenotype with pmr1Δ. In contrast, the subset of genes sharing multiple, overlapping phenotypes with pmr1Δ was highly enriched in distinct functional categories and subcellular localization. Unexpectedly, this approach did not identify other Ca2+ transporters, pointing to a unique, nonredundant role for Golgi-localized ion homeostasis. Instead, mutants of the vacuolar-type H+-transporting ATPase (V-ATPase) were found to most closely phenocopy pmr1Δ. Although it is established that the V-ATPase is found in multiple compartments, including the Golgi (Manolson et al., 1994 ; Kawasaki-Nishi et al., 2001 ), it has not yet been possible to distinguish between organelle-specific functions, and most vma phenotypes in yeast are attributed to a vacuolar function by default. Our findings lead us to propose that shared phenotypes of vma mutants with pmr1Δ, including multidrug and ion sensitivity, are evidence for Golgi-localized functions of the V-ATPase. We suggest that this simple approach of overlapping multiple, distinct phenotypes can be applied to any gene to reveal new insights into cellular function.
A library of 4828 S. cerevisiae strains with deletion of each nonessential gene in a haploid background (BY4742; MATα) was purchased from Research Genetics (ResGen; Invitrogen, Carlsbad, CA). Each strain has a complete replacement of one open reading frame with the kanMX cassette. To make double knockouts, replacement of the PMR1::kanMX cassette with the NAT marker was made in the BY4741 background (MATa) as described previously (Tong et al., 2001 ), and the resulting strain mated with vma2Δ, vma5Δ or snf6Δ from the BY4742 collection. Diploids were selected on YPD plates supplemented with both 200 μg/ml kanamycin and 100 μg/ml clonNAT, sporulated, and dissected. Haploid double mutants resistant to both drugs were identified.
To screen for pmr1Δ phenotypes, 52 96-well plates representing the collection of mutant strains were thawed, and 5 μl of each culture was used to inoculate 200-μl seed cultures grown 18–36 h in synthetic complete (SC) medium at 30°C. Four microliters of each seed culture was then used to inoculate 96-well microtiter plates each containing 200 μl of SC medium with and without addition of either 1.5 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA), 10 mM MnCl2, or 10 μM amiodarone (AMD). Concentrations were chosen based on results from serial dilution experiments showing the greatest difference between wild type (WT) and pmr1Δ growth. On each microtiter plate, two wells were reserved for a WT culture and for growth medium alone, representing positive and negative controls, respectively. Growth was monitored after incubation for 19 h at 30°C by measuring A600 nm on a BMG FLUOstar Optima multimode plate reader with BMG FLUOstar Optima version 1.20 software (BMG Labtechnologies, Durham, NC). Immediately before recording, cultures were rapidly resuspended using an electromagnetic microtiter plate shaker (Union Scientific, Randallstown, MD), and all recordings were made at 30°C. A600 nm values were background subtracted and normalized to average WT growth (mean calculated from 52 separate cultures spanning all microtiter plates per screen). Strains showing <20% of WT growth (<1% of the collection) under control conditions (SC) were omitted from further analysis. Of the strains that were included, low A600 nm values of the seed cultures did not correlate with growth defects observed under experimental conditions. For each screen, conditional growth values were normalized to A600 nm values obtained under control conditions to estimate sensitivity or tolerance. Data were organized, plotted, and analyzed using Excel X (Microsoft, Redmond, WA), S-Plus 6.2 (Insightful, Seattle, WA), and SPSS 12.0 (SPSS, Chicago, IL) software.
Using the Munich Information center for Protein Sequences (MIPS) database (http://mips.gsf.de/projects/funcat), we acquired the annotated Functional Categorization (Fun Cat) and Cellular Location (Cell Loc) of genes identified in the phenocopy screens. Resulting data from all screens were integrated and visualized as network diagrams by using Osprey network visualization software http://biodata.mshri.on.ca/osprey/servlet/Index. Illustrations were created using Adobe Illustrator CS software (Adobe Systems, Mountain View, CA).
Total cellular accumulation of calcium was measured by growing yeast to logarithmic phase in SC media, followed by harvesting and resuspending in fresh SC media (~3 × 107 cells/ml) supplemented with tracer quantities of 45CaCl2 (26.1 μCi/ml). After 2 h of incubation at 30°C, in the absence or presence of the indicated amounts of drugs, cells were harvested rapidly by filtration on to nitrocellulose membrane filters (HAWP, 0.45 μm; Millipore, Billerica, MA), washed with ice-cold wash buffer (10 mM HEPES and 150 mM KCl), placed in scintillation vials, and processed for liquid scintillation counting by using CytoScint scintillation cocktail (MP Biomedicals, Aurora, OH).
Vacuolar accumulation of quinacrine was examined as described previously (Roberts et al., 1991 ). Briefly, ~3 × 107 log-phase yeast cells were harvested and resuspended in 500 μl of YPD buffered with 50 mM Na2HPO4, pH 7.6, containing 200 μM quinacrine. After incubation at room temperature for 5 min, cells were sedimented at 10,000 × g for 5 s, washed once with 500 μl of 2% glucose buffered with 50 mM Na2HPO4, pH 7.6, and resuspended in 100 μl of the same solution. Samples were applied to a microscope slide and viewed immediately in the fluorescence microscope by using a fluorescein filter. To stain with N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenyl-hexatrienyl) pyridinium dibromide (FM4-64), yeast cells were grown to logarithmic phase in YPD, and 16 μM FM4-64 (Invitrogen) was added from a stock solution in dimethyl sulfoxide (DMSO) for 30 min. Samples were applied to a microscope slide and viewed immediately in the fluorescence microscope by using a Texas Red filter.
The best-known growth phenotypes of pmr1 mutants are directly related to cellular Ca2+ or Mn2+ homeostasis, as expected for the loss of a Ca2+, Mn2+-transporting ATPase. Thus, pmr1Δ strains are hypersensitive to removal of divalent cations from the medium by the chelator BAPTA. We have shown that BAPTA sensitivity in pmr1 mutants corresponds to calcium starvation in the early part of the secretory pathway and that it can be complemented by heterologous expression of phylogenetically diverse Ca2+-ATPases of the endoplasmic reticulum (ER), plasma membrane, or Golgi subtypes (Ton et al., 2002 ). Likewise, hypersensitivity of pmr1 strains to high MnCl2 corresponds to accumulation of toxic levels of cellular Mn2+ and can be corrected specifically by Mn2+-transporting pumps unique to the Golgi subtype, including human SPCA1 and SPCA2. The recent observation that pmr1 mutants are sensitive to the antifungal agent amiodarone (Gupta et al., 2003 ) prompted us to search the database for other reports of drug sensitivities. We found that pmr1 was identified in genome-wide screens of the yeast deletion library for hypersensitivity to wortmannin, tunicamycin, and sulfometuron (Zewail et al., 2003 ; Parsons et al., 2004 ). These drugs target diverse and unrelated cellular pathways: wortmannin is an inhibitor of PI-4 kinase in yeast, tunicamycin blocks protein glycosylation, sulfometuron targets branched chain amino acid synthesis, and amiodarone, an antiarrhythmic agent, has been proposed to cause calcium-mediated cell death in yeast.
To evaluate whether the multiple drug-sensitive phenotypes associated with pmr1 null mutants arose from loss of Ca2+ or Mn2+ transport, or both, we made use of a previously described Pmr1 mutant that has an ion-selective defect. Mutant Q783A was shown to retain 45Ca transport and Ca2+-ATPase activity at nearly wild-type levels but was severely attenuated in all Mn2+-dependent functions examined previously (Mandal et al., 2000 ). We introduced Q783A mutant and wild-type PMR1 genes into yeast strain K616 (pmr1Δpmc1Δcnb1Δ; Cunningham and Fink, 1994 ), which shows robust ion-sensitive growth phenotypes, and we assessed drug sensitivity of the strains (Figure 1). As expected, wild-type Pmr1 improved growth, relative to the vector control, under all conditions tested. Mutant Q783A could not correct Mn2+ hypersensitivity of the host strain, but it improved growth in BAPTA similar to wild type (Figure 1, A and B), demonstrating the Mn2+-selective defect of this mutant. We show that expression of mutant Q783A complements drug hypersensitivity to wild-type (amiodarone; Figure 1C), or nearly wild-type levels (sulfometuron methyl, wortmannin, and tunicamycin; Figure 1, D–F), suggesting that Ca2+ transport plays a major role in mediating the requirement for Pmr1 in the cellular response to multiple drugs. We note that partial complementation by mutant Q783A in tunicamycin (Supplemental Figure 1) is consistent with a role for both Ca2+ and Mn2+ ions; thus, tunicamycin leads to accumulation of unfolded proteins, a form of cell stress known to require Ca2+ ions (Bonilla et al., 2002 ), however, tunicamycin toxicity may be exacerbated by depletion of Mn2+ ions from the Golgi, because endoglycosidases are Mn2+-requiring enzymes. Similarly, there may be overlapping requirements for Ca2+ and Mn2+ in other drug responses. A potential target, the ATM kinase, mutated in human ataxia telangiectasia, is a Mn2+-requiring enzyme that is sensitive to wortmannin (Chan et al., 2000 ). These observations underscore the importance of Pmr1-mediated ion homeostasis in multiple cellular pathways that impact drug sensitivity.
Calcium influx has been reported in response to various forms of cell stress (Bonilla et al., 2002 ; Matsumoto et al., 2002 ). Therefore, we postulated that drug hypersensitivity in pmr1 mutants might arise from defective Ca2+ homeostasis after drug-induced Ca2+ influx. Consistent with this hypothesis, we show that addition of amiodarone, wortmannin, or tunicamycin elicit substantive increase in 45Ca uptake in both wild-type and pmr1 mutants (Figure 2). We note that basal levels of 45Ca uptake in pmr1Δ yeast are higher than wild type in response to depletion of internal stores, as reported previously (Halachmi and Eilam, 1996 ; Locke et al., 2000 ). Although addition of sulfometuron did not elicit consistent increases in Ca2+ influx over the same period, increasing the drug concentration or time of incubation resulted in higher levels of Ca2+ influx (Yadav and Rao, unpublished data). We conclude that drug-induced Ca2+ influx could contribute, at least in part, to the Ca2+-related drug sensitivity of the pmr1 strain.
To find specific components and functional modules that operate within the same or parallel cellular pathways as Pmr1, we sought to identify genes that shared multiple pmr1 mutant phenotypes. Gene deletions resulting in hypersensitivity to wortmannin, tunicamycin, and sulfometuron have been reported, and a partial gene list for amiodarone hypersensitivity has been identified previously (Gupta et al., 2003 ; Zewail et al., 2003 ; Parsons et al., 2004 ). We therefore undertook genome-wide screens of the BY4742 haploid yeast deletion library (~4800 strains) to find additional mutants sensitive to 10 μM amiodarone as well as to 1.5 mM BAPTA and 10 mM MnCl2. Null mutant strains that were similar to, or more sensitive than pmr1Δ, in the growth response to each condition were identified and categorized according to function. Each test condition elicited a unique growth response of the deletion collection, characterized by enrichment of distinct functional categories. For example, genes involved in cell rescue and virulence were enriched 1.5-fold in the BAPTA- and amiodarone-sensitive screens (p = 0.025) but not in MnCl2, whereas a similar enrichment was seen for cell communication in BAPTA and MnCl2 screens but not in amiodarone. Certain functional categories showed significant, albeit modest, trends of overrepresentation in all six screens; these categories included cellular transport, protein fate, and interaction with the cellular environment (1.25- to 1.75-fold). We reasoned that if these categories were specifically related to Pmr1 function, then the trends observed in individual screens would be further enhanced in the set of genes that share multiple pmr1 phenotypes. We identified 118 of a total of 695 genes (17%) that shared at least two common growth sensitivities with pmr1Δ, and 34 of these (5% of total) shared three or more common phenotypes (Table 1). Graphical representation of functional categories (Figure 3A) in the set of genes sharing three or more phenotypes (≥3, outer ring) and two or more phenotypes (≥2, middle ring) relative to the yeast proteome (inside ring) reveals statistically high enrichment of categories that showed relatively modest changes in the individual phenotype sets. Thus, the ≥2 data set showed enrichment in the categories of protein fate by 1.33-fold (p = 1.35 × 10−5), cell transport by 1.75-fold (p = 9.2 × 10−10), and interaction with cellular environment by 2-fold (p = 5.56 × 10−8). The latter two functions showed even further enhancement, to 2- and 3.25-fold respectively, among genes in the ≥3 data set. Within broad functional categories, certain subcategories showed highly significant enrichment, e.g., the unfolded protein response and ER quality control (5 of a genome-wide pool of 69; p = 4.81 × 10−3) and biogenesis of the vacuole (5 of 44 proteins; p = 6.38 × 10−4).
Similarly, Figure 3B shows the subcellular distribution of gene products represented in the ≥2 and ≥3 phenocopy data sets, relative to the yeast proteome. Here, we observed prominent, statistically significant overrepresentation of proteins localizing to secretory and vacuolar pathways. This included the endoplasmic reticulum, which was enriched from 6% in the proteome to 11 and 15% (up to 2.5-fold), Golgi (up to 4-fold), transport vesicles (up to 4-fold) and vacuole (up to 8-fold). Thus, this analysis reveals that the major cellular pathways responsive to perturbations in ion homeostasis and drug toxicity localize to the endomembrane system.
Of the gene deletions displaying at least two pmr1 phenotypes (Table 1), only three shared all six drug and ion-related growth sensitivities with pmr1Δ: of these three phenotypes, HUR1 is a dubious open reading frame (ORF) encoded on the opposite strand to that of PMR1, so that hur1Δ essentially recapitulates pmr1Δ. Expression of HUR1 has not been detected at transcript or protein level, and it has no recognizable orthologues. Because reintroduction of PMR1 complements all six phenotypes (Figure 1), HUR1 represents a false positive of the screening strategy; however, it serves as an internal control and validates our findings. Deletion of VMA5, encoding a subunit of the vacuolar H+-ATPase complex, also shares all six growth phenotypes with pmr1. YKL118w is a dubious ORF that overlaps another subunit of the V-ATPase on the opposite strand, that of VMA12/VPH2. In all, 12 of the 18 known VMA subunits and assembly factors shared two or more knockout phenotypes with pmr1Δ: VMA5, VMA2, TFP3/VMA11, VMA13, VMA7, VMA4, VMA12, VMA8, VMA10, VMA22, VMA6, and CUP5/VMA3, making this the single-most significant functional module of the collective phenotypes. We note that the shared multidrug sensitivity of pmr1Δ and vmaΔ mutants was recently extended to DNA damaging agents, including cisplatin and hydroxyurea (Pan et al., 2006 ; Liao et al., 2007 ).
Remarkably, chromatin remodeling complexes were also abundantly represented (Table 1). This included three components of the SWI/SNF nucleosome-remodeling complex (HAF4/SNF5, SNF6, and SNF2) and five subunits of the SAGA/ADA histone acetyltransferase coactivator complex (GCN5/ADA4, SPT20/ADA5, ADA1, SPT3, and SPT7/GIT2). Two other knockouts would also result in disruption of SWI/SNF components: YJL175w overlaps with SWI13 on the opposite strand, and YLR322w is a dubious ORF that would disrupt SFH1, a SNF5 homolog. Also included in this functional category are SET3, a histone deacetylase, and NHP10, which is related to mammalian high-mobility group proteins and a likely component of the INO80 ATP-dependent chromatin remodeling machinery. Together with transcription factors (GAL11, SIN4, GCN4, OPI1, and RPN4), these are likely to represent cellular stress response pathways. There is evidence that the SAGA complex is required for the expression of roughly 10% of the genome that is involved in stress response (Huisinga and Pugh, 2004 ; van Voorst et al., 2006 ). Our findings indicate that although ion stress (Ca2+, Mn2+, or H+) and diverse cytotoxic drugs (amiodarone, wortmannin, sulfometuron, or tunicamycin) may target different pathways, they converge to elicit similar downstream cell survival responses.
Multiple genes in the ergosterol biosynthesis pathway, including ERG2, ERG3, ERG4, ERG6, and SAC1, encoding a lipid inositol phosphoinositide phosphatase, were found to share two or more knockout phenotypes with PMR1. These genes would be expected to impact membrane integrity and permeability not only by altering lipid composition but also by affecting the activity of ion and drug transporters as well as protein and vesicular trafficking pathways. Indeed, the latter were represented by genes directing traffic to or from compartments of the Golgi, endosomes, and vacuole (VPS45, VPS51, VPS52, VPS54, SNF8/VPS22, SNF7/VPS32, and COG5).
Notably missing from the list of gene mutations that phenocopy pmr1Δ were known Ca2+ transporters (PMC1, VCX1) and Ca2+ channels (CCH1, MID1, YVC1), pointing to a distinct, nonredundant cellular role of Pmr1. Although large, compensatory increases in expression of Pmc1, the vacuolar ATP-driven Ca2+ pump, have been reported (Marchi et al., 1999 ; Locke et al., 2000 ), pmr1 mutants retain distinct phenotypes. This may be because of the unusual ion selectivity of Pmr1, which transports both Mn2+ and Ca2+ ions, and more importantly, its unique localization in the Golgi.
Because the vacuolar ATPase has to traffic through the early secretory pathway and Golgi, where Pmr1 activity is important, we considered the possibility that the V-ATPase was dysfunctional in pmr1 mutants, resulting in overlapping vma and pmr1 phenotypes. The diagnostic test for loss of acidic vacuoles is the inability to grow at alkaline pH or to handle calcium stress at alkaline pH, as seen for vma5Δ (Figure 4, A and B). However, pmr1Δ was similar to wild type in the growth response to alkaline pH and high calcium (Figure 4, A and B), suggesting that vacuolar pH was normal in this mutant. In vma mutants, a consequence of defective acidification within a post-Golgi endosomal compartment is the failure to load the multicopper oxidase Fet3 with copper, resulting in iron starvation and growth sensitivity to the iron-specific chelating agent bathophenanthroline disulfonate (BPS) (Davis-Kaplan et al., 2004 ). We show that the pmr1 mutant is not hypersensitive to BPS, indicating that loss of the Golgi Ca2+, Mn2+ pump does not significantly affect the pH-dependent transport and availability of copper to Fet3 (Figure 4C). Furthermore, vacuoles of wild-type and pmr1Δ cells readily took up the fluorescent weak base quinacrine, which becomes protonated and trapped within acidic compartments (Figure 5, A and B). In contrast, the vma5Δ mutant failed to accumulate quinacrine, characteristic of defective vacuolar acidification (Figure 5C). However, quinacrine labeling did reveal aberrant vacuolar morphology in the pmr1 mutant, as has been reported previously (Kellermayer et al., 2003 ). This was more readily visualized using the vacuolar membrane dye FM4-64 (Figure 6), which showed the appearance of multiple vacuolar lobes or fragmented vacuolar compartments in pmr1Δ, suggesting a membrane fusion defect. Interestingly, addition of amiodarone caused a rapid fusion of these compartments, resulting in a single large vacuole within a few minutes (Figure 6C). We hypothesized that the amiodarone-induced calcium burst (Courchesne and Ozturk, 2003 ; Gupta et al., 2003 ) promoted membrane fusion of pmr1Δ vacuoles. Consistent with this idea, addition of 5 mM CaCl2 also promoted vacuolar fusion in the pmr1Δ mutant (Figure 6D). We conclude that despite the aberrant vacuolar morphology, the pmr1 mutant has normal vacuolar pH and pH-related vacuolar functions.
Alternatively, could the V-ATPase be required for normal localization or transport activity of Pmr1? We found that distribution of green fluorescent protein-tagged Pmr1 was similar in wild-type and vma mutant strains (Ton and Rao, unpublished data). Furthermore, concanamycin A, a specific inhibitor of V-ATPase activity, and protonophores do not inhibit 45Ca uptake by Pmr1 in purified Golgi vesicles (Sorin et al., 1997 ). Finally, detailed kinetic analyses of mammalian orthologues of Pmr1 confirm an absence of pH effects on ATPase activity (Dode et al., 2006 ). Thus, a proton gradient is not directly required for Pmr1 activity. These biochemical observations suggest that the ion transport activities of Pmr1 and V-ATPase do not directly influence each other but that they are required in parallel for the same cellular function.
Genes that share multiple knockout phenotypes with pmr1Δ may lie within the same pathway, or they may function in parallel pathways that achieve the same cellular outcome. A straightforward approach to differentiate between these two possibilities is to test for synthetic fitness defects in double null mutants. We show that disruption of both PMR1 and VMA genes leads to severe synthetic fitness defects. Although all spores derived from tetrads in these crosses were able to germinate, the double null mutants of PMR1 and either VMA2 or VMA5 were much smaller in size (Supplemental Figure 2) and grew poorly (Figure 7, A and B). In contrast, a double null mutant of PMR1 and SNF6, a component of the chromatin remodeling complex, showed no synthetic growth defect (Figure 7C). These genetic data corroborate our biochemical and phenotypic findings to show that Pmr1 and V-ATPase work in parallel toward a common cellular function.
As a complementary approach to growth fitness tests, we considered other function-based genome-wide screens that have identified a role for Pmr1. Bonangelino et al. (2002) identified pmr1Δ as one of 362 mutant strains (7.8% of the viable diploid deletion collection) that were impaired in the biosynthetic sorting of the vacuolar hydrolase carboxypeptidase Y (CPY), resulting in various levels of inappropriate secretion into the culture medium. This data set is enriched in vacuolar protein sorting (VPS) genes, as well as numerous other genes involved in regulating the actin cytoskeleton, glycosylation and other Golgi functions (Bonangelino et al., 2002 ). Durr et al. (1998) have demonstrated that CPY secretion in pmr1Δ yeast is a consequence of defective Ca2+ homeostasis, because addition of extracellular Ca2+, but not Mn2+, could correct this defect. We also took advantage of a genome-wide screen for defective mannosyl-phosphorylation in the outer layer of the yeast cell wall, which was performed by assaying the haploid deletion collection for decreased binding of the cationic dye Alcian Blue (Conde et al., 2003 ; Corbacho et al., 2005 ). Among the most severely attenuated of 198 low dye binding (ldb) mutants identified (4% of the collection), pmr1/ldb1 also showed the greatest reduction in size of secreted invertase, indicative of defective chain growth (Olivero et al., 2003 ). These phenotypes are likely to be a consequence of Mn2+ insufficiency within the Golgi of the pmr1 mutant, because endoglycosidases are Mn2+-requiring enzymes. Indeed, underglycosylation of invertase in pmr1 can be corrected by supplementing the growth medium with Mn2+ but not Ca2+ (Durr et al., 1998 ). Like the CPY screen, the ldb mutants fell into diverse cellular pathways; however, we note that secretory and vacuolar trafficking pathways within the cell were abundantly represented. Figure 8 is a Venn diagram showing the extent of overlap between the pool of genes sharing at least two, and up to six, pmr1 phenotypes and the functional screens for CPY secretion and low dye binding. Common to all three data sets were 22 genes: in addition to PMR1, this included 12 VMA genes, four genes involved in sorting or fusion of vesicles from or to the Golgi (VPS15, VPS45, VPS54, and COG5), and a Golgi-resident mannosyltransferase (HOC1). The relationship of the remaining genes to the observed phenotypes is currently not clear. For example, PER1 encodes an ER-localized protein that has been implicated in remodeling of GPI anchors and their association with raft-like domains, which could in turn impact on Golgi/secretory pathway function (Fujita et al., 2006 ). Overall, a majority of the common genes (18/22) encode proteins known to reside within or traffic to the Golgi membranes, including the V-ATPase that has a distinct, Golgi-localized isoform (Manolson et al., 1994 ; Kawasaki-Nishi et al., 2001 ). It is interesting to note that many members of this group also share a common synthetic lethality with endocytosis genes. For example, pmr1Δ, vma2Δ, vps15Δ, vps45Δ, and vps54Δ are individually inviable with end4/sla2Δ (Munn and Riezman, 1994 ; Conibear and Stevens, 2000 ). Together, we suggest that shared phenotypes of pmr1 and vma mutants, including multiple drug and ion sensitivity, derive from shared ion transport functions within the Golgi.
Based on this hypothesis, we asked whether pmr1 shared additional vma phenotypes that may be attributed to Golgi/secretory pathway function. We tested sensitivity to calcofluor white, an antimicrobial agent that binds to chitin on the cell wall. Altered sensitivity to calcofluor white relates to missorting of chitin synthetase and altered levels of chitin on the cell surface, and it is a known phenotype of vma mutants (Davis-Kaplan et al., 2004 ). Indeed, we found that like vma5Δ, pmr1Δ displayed increased sensitivity to calcofluor white (Figure 9A). Another vma phenotype also shared with pmr1Δ relates to zinc homeostasis. It has been argued that the acute Zn2+ sensitive phenotype of vma mutants is due to defective vacuolar sequestration and detoxification of this ion and that this phenotype may serve as a diagnostic test to distinguish vacuolar from pre-vacuolar functions of the V-ATPase. However, we show that pmr1Δ mutants are nearly as sensitive to Zn2+ as the vma5Δ mutant (Figure 9B). Given that pmr1 mutants have normal pHv, it is unlikely that Zn2+ sensitivity is associated with disruption of vacuolar sequestration of this ion in pmr1Δ cells. Therefore, we suggest that it may be due to disruption of a shared role in membrane trafficking within the Golgi or a prevacuolar compartment. We found that both calcofluor white sensitivity (Figure 9C) and Zn2+ sensitivity (Figure 9D) of pmr1Δ could be fully complemented by the Mn2+-defective Q783A mutant of Pmr1, indicating that these defects are related to Ca2+ homeostasis.
Our finding that vma mutants share synthetic fitness defect and multiple phenotypes with the pmr1 mutant adds one more connecting link to the complex network of gene interactions involved in cellular ion homeostasis (Figure 10). It is known that vma mutants have calcium-handling defects and are therefore sensitive to calcium-induced stress. The vacuole is the major reservoir of calcium in the yeast cell, most of it in a nonexchangeable form, presumably complexed with phosphates and other anions. However, if the role of the V-ATPase in Ca2+ homeostasis is solely to fuel vacuolar Ca2+ uptake and sequestration via H+-coupled transporters, it is indeed surprising that a mutant lacking Vcx1, the vacuolar H+/Ca2+ exchanger, does not share the multiple drug-sensitive growth phenotypes or trafficking defects described in this work. Moreover, vcx1 mutants show no synthetic fitness defect with pmr1 mutants (Cunningham and Fink, 1996 ). Although it cannot be excluded that there are other, as yet uncharacterized, H+/Ca2+ antiporters, Vcx1 has been shown to be the major transport pathway for low-affinity, high-capacity removal of cytosolic Ca2+ (Miseta et al., 1999 ; Forster and Kane, 2000 ). One explanation is that Vcx1 is redundant with Pmc1, the primary Ca2+ pump of vacuolar membrane; however, a pmc1Δvcx1Δ mutant is viable (Cunningham and Fink, 1996 ), although it has added sensitivity to high calcium stress. Double mutants of vma genes with pmc1Δ, pmr1Δ or calcineurin (cnb1Δ) show slow growth or no growth even in the absence of calcium stress (Forster and Kane, 2000 ; this work), consistent with additional prevacuolar/secretory functions of the V-ATPase, separate from a vacuolar role.
There is other evidence that a role for V-ATPase in Ca2+ homeostasis is unlikely to be restricted to vacuolar detoxification of excess calcium. Intriguingly, inactivation of vma genes in Neurospora crassa had dramatic effects on hyphal morphology by restricting hyphal elongation and stimulating branching (Bowman et al., 2000 ). The authors note that when they added the calcium reporter chlortetracycline to N. crassa, they saw brightly glowing hyphal tips in wild type but no fluorescence in vma null mutants. This points to a disruption of ion homeostasis along the secretory pathway in the vma mutants. Consistent with this idea, disruption of the N. crassa PMR1 gene, but not of other Ca2+ pumps and antiporters, resulted in a similar multibranched hyphal morphology (Bowman, B., and Bowman, E. J., personal communication). Many of the shared phenotypes between pmr1 and vma mutants described in this work are directly associated with disruption of the secretory pathway and prevacuolar trafficking pathways (such as calcofluor white sensitivity, low Alcian Blue binding, and CPY secretion). Furthermore, we show that multiple drug- and ion-sensitive growth phenotypes are predominantly associated with disruption of secretory and vacuolar biogenesis pathways (Figure 3).
In mammalian cells, there is experimental evidence for the trafficking and function of V-ATPase along secretory and endocytic pathways to and from the plasma membrane (Breton and Brown, 2007 ). Although yeast Vma has not been detected at the cell surface, vma mutants show markedly reduced rates of endocytosis (Perzov et al., 2002 ). Moreover, there is evidence for a distinct, Golgi-localized isoform of the 100-kDa a subunit of the V-ATPase, which contains the Stv1 subunit in place of Vph1, resulting in distinct functional and regulatory properties (Manolson et al., 1994 ; Kawasaki-Nishi et al., 2001 ). In contrast to the multiplicity of subunit isoforms in mammals, Stv1 and Vph1 represent the only V-ATPase subunit isoforms found in fungi, consistent with the importance of distinct Golgi and vacuolar functions of the V-ATPase. Because of the ability of these two subunit isoforms to partially compensate for each other, it has been difficult to separate Golgi-specific phenotypes of the V-ATPase from its vacuolar functions. The phenomic approach used in this study reveals distinct Vma functions that overlap with the Golgi Ca2+, Mn2+-ATPase Pmr1. We suggest that these shared phenotypes provide evidence for secretory pathway and prevacuolar functions of the V-ATPase. Given the excellent conservation of basic ion homeostasis mechanisms from yeast to human (for review, see Ton and Rao, 2005), our findings may be extrapolated to mammalian models. There are examples of defects in tissue-specific isoforms of V-ATPase or Ca2+-ATPase that are known or suspected to give rise to a similar disease phenotype, such as deafness and male sterility (Prasad et al., 2004 ; Breton and Brown, 2007 ). Calcium and proton homeostasis are also closely linked in normal and pathophysiological conditions; a case in point is osteopetrorickets (Kaplan et al., 1993 ), where defective acidification by V-ATPase leads to osteopetrosis, abnormally high body calcium, and paradoxically, poor calcium incorporation into bone. In conclusion, the overlap of calcium and proton homeostasis pathways, particularly in Golgi and prevacuolar traffic, revealed by a phenomics approach in yeast, may help in understanding and treatment of complex phenotypes associated with disease.
This work was supported by National Institutes of Health grant GM-62142 (to R.R.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E06-11-1049) on February 21, 2007.
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org).