Proliferation of Intracellular Structures upon Overexpression of the PMAB ATPase in Saccharomyces cerevisiae*

The PMAB gene is a presumed isogene of the PMAl gene, encoding the major yeast plasma membrane H+-ATPase. When controlled by its own promoter, PMAB in multiple copies does not complement a deficient PMAl gene. Under the control of the PMAl promoter, however, and expressed on a centromeric plasmid in yeast strains specially designed for stable expression, the PMAB gene replaces the PMAl gene to some extent, allowing growth on standard medium but not on acidic media. Plasma membranes of cells expressing only the PMAB enzyme display low ATPase activity correlating with low amounts of PMAB protein. This low activity is maintained throughout growth and does not increase when overexpression is favored by in- creased gene dosage. Immunoelectron microscopy reveals a dramatic proliferation of intracellular structures (probably endoplasmic reticulum) in which overexpressed PMA2 protein accumulates. Overex- pression of PMAl ATPase causes a similar phenome-non, but quantitative effects are lower compared to PMA2. These results indicate that the PMA2 gene encodes a functional plasma membrane H+-ATPase and suggest a specific control of the intracellular traffic of plasma membrane ATPase.


B-1348 Louuain-La-Neuue, Belgium
The PMAB gene is a presumed isogene of the P M A l gene, encoding the major yeast plasma membrane H+-ATPase. When controlled by its own promoter, PMAB in multiple copies does not complement a deficient P M A l gene. Under the control of the P M A l promoter, however, and expressed on a centromeric plasmid in yeast strains specially designed for stable expression, the PMAB gene replaces the P M A l gene to some extent, allowing growth on standard medium but not on acidic media. Plasma membranes of cells expressing only the PMAB enzyme display low ATPase activity correlating with low amounts of PMAB protein. This low activity is maintained throughout growth and does not increase when overexpression is favored by increased gene dosage. Immunoelectron microscopy reveals a dramatic proliferation of intracellular structures (probably endoplasmic reticulum) in which overexpressed PMA2 protein accumulates. Overexpression of PMAl ATPase causes a similar phenomenon, but quantitative effects are lower compared to PMA2. These results indicate that the PMA2 gene encodes a functional plasma membrane H+-ATPase and suggest a specific control of the intracellular traffic of plasma membrane ATPase.
For many years, the only P-type ion-motive ATPase (Pedersen and Carafoli, 1987) identified and characterized in yeast was the plasma membrane H'-ATPase. This ATPase extrudes protons from the cell through the plasma membrane, thereby creating an electrochemical gradient which constitutes the driving force for nutrient import (reviewed in Slayman (1981), in Scarborough (1984), in Bowman and Bowman (1986), and in ). The corresponding gene, called PMAl, has been sequenced in the case of Saccharomyces cereuisiae, Neurospora crassa, Schizosaccharomyces pombe, Zygosaccharomyces rouxii, and Candida albic a m (see Wach et al. (1992)). It is essential to yeast viability (Serrano et al., 1986).
The simple picture of yeast cells containing only one Ptype H+-ATPase has been complicated by the discovery of the PMAZ gene in S. cereuisiae. The sequence of the protein * This work was supported in part by grants from the Services de la Politique Scientifique: Action Sciences de la Vie and the Fonds National de la Recherche Scientifique, Belgium. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisenent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
3 Supported by a fellowship from the Institut pour I'Encouragement de la Recherche dans 1'Industrie et l'Agriculture, Belgium.
§ To whom correspondence should be addressed. Tel.:  encoded by the PMA2 gene is 90% identical to the PMAl gene product sequence, suggesting that the PMA2 protein is also an Hf-ATPase. However, it is expressed at a much lower level than PMAl. Consistently, the PMAQ gene is not only dispensable for haploid growth but its disruption fails to produce a clear phenotype . Strikingly, a second PMA gene has also been found in S. pombe. This pma2 gene seems to share similar properties with its S. cereuisiae homonym (Ghislain and Goffeau, 1991).
The apparent duplication of H'-ATPase genes in yeasts raises important questions. What are their functional specificities? Where are these ATPases located? What are the mechanistic and metabolic consequences of the observed differences in amino acid sequence?
To address these questions, we have engineered a set of new S. cereuisiae strains for overexpressing the PMAZ gene in the absence of the PMAl gene and vice versa. These strains are of general value in the study of pma mutations which in other strains can be repaired by recombination with wild-type PMA copies (Harris et al., 1991). ' We here show that the PMAB ATPase can functionally replace the PMAl ATPase during normal mitotic growth. Our results suggest that the plasma membrane is probably the true physiological location of the PMA2 protein. We reveal, however, major differences in the intracellular traffic of PMAl and PMAB proteins. A companion study (Supply et al., 1993) focuses on the compared enzymatic properties of the plasma membrane-bound PMAB andPMAl ATPases.

EXPERIMENTAL PROCEDURES
Chemicals-5-FOA2 was purchased from PCR incorporated (Gainesville, FL); NazATP from Sigma; protease inhibitors from Boehringer Mannheim GmbH. All other compounds were of analytical grade. Molecular mass markers were provided by Pharmacia LKB Biotechnology Inc.
Enzymes-Exonuclease I11 was from Life Technologies Inc.; mung bean nuclease from Biolabs. All other enzymes were of the highest purity commercially available.
Media-Unless otherwise stated, yeast strains were grown at 30 "C on YD rich medium containing 2% glucose, 2% yeast extract (KAT, Ohly, Hamburg, Germany) or on minimal medium containing 2% glucose, 0.7% yeast nitrogen base without amino acids (Difco) supplemented when necessary with the required nutrients. Solid media additionally contained 2% agar (Difco). The sporulation and 5-FOA media were prepared as in Treco (1989).
Site-directed Mutagenesis-The PMAl promoter was modified by standard techniques described in Sambrook et al. (1989). We used A. Wach, P. Supply, and A. Goffeau, manuscript submitted for publication. the mutagenic oligonucleotide 5'-CAATATCGATATGAC-3' corresponding to positions -10 to +4 with respect to the ATG start codon of the PMAl gene (Serrano et al., 1986;Capieaux et al., 1989). The guanidine at position -3 replaces an adenine of the wild-type sequence. Plasmid pECPTZ-PMA1, usedto produce the single-stranded template, was obtained by inserting a 4.65-kb HindIII-XbaI fragment containing PMAl into the Hind111 and XbaI sites of PTZ19R.3 The AG transition in mutated plasmid pPSPMAlC created a unique ClaI site in the promoter region of PMAl.
Nuclease Deletion-Removal of the promoter from the PMAS gene was carried out as instructed in the pBluescript Exo/Mung DNA Sequencing System manual (Stratagene). PlasmidpPTZlBR-PMA2L ) contains a 5.07-kb HindIII fragment, encompassing the PMA2 gene, inserted into the HindIII site of vector PTZ18R with the PMAZ promoter proximal to the polylinker. This plasmid was restricted first with SphI, which cut the polylinker and left an exonuclease 111-resistant 3' overhang, then with BssHII, which cut the PMAB promoter at position -551 from the start ATG, leaving an exonuclease 111-sensitive 5' overhang. To remove most of the remaining 551 bp, the restricted DNA was digested with exonuclease 111 and mung bean nuclease. After ligation and amplification, three clones were selected: pPTZ-PMA2-15, -21, and -22. The size of the remaining sequence upstream from the start ATG was determined for each clone by sequencing, being 30,10, and 24 bp, respectively.
Construction of the Overexpression Plasmids-To construct the plasmids pPS15-P12-15, -21, and -22 (PPM~,::PMAS, CEN, LEU2, see Fig. lA), the mutagenized plasmid pPSPMAlC was digested with CZQI and XbaI to remove the region downstream from the promoter in the 4.65-kb fragment of PMAl, then treated with Klenow DNA polymerase and religated. Then the promoter, extracted with HindIII (blunted with Klenow) and BamHI, was inserted into the XbaI (blunted with Klenow) and BamHI sites of the polylinker of pRS315 (Sikorski and Hieter, 1989), resulting in plasmid pPS15-Pphl~~. On the other hand, pPTZ-PMA2-15, -21, and -22 were digested with BamHI and HindIII to excise the PMA2 gene lacking its promoter.
RNA extraction was carried out according to Schmitt et al. (1990) and the Northern experiment according to .
Plasma Membrane Preparations-Plasma membranes were prepared as in Dufour et al. (1988) with the following modifications. After overnight growth in 1 liter of 5.8% glucose, 2% KAT yeast extract (to a density of 200-300 lo6 cells/ml), the cells were harvested and washed with ice-cold water. The pellet was divided in two equal parts. One part was resuspended in 2 volumes of 250 mM glucose, the other in 2 volumes of 250 mM sorbitol, then both were incubated for 15 min at 30 "C. After centrifugation, the pellets were resuspended in either 250 mM glucose, 1 mM MgC12, 50 mM imidazole adjusted to pH 7.5 with NaOH, 1 mM phenylmethylsulfonyl fluoride, 5 mM dithiothreitol, or the same medium containing 250 mM sorbitol instead of 250 mM glucose. The next steps (cell disruption, crude membranes and plasma membranes preparations) were unmodified. When stated, plasma membranes were treated with Triton X-100 as described in Dufour et al. (1988), except that glycerol was added (0.5%, final concentration). At the final step, the proteins were resuspended (21 mg/ml) in 10 mM imidazole, pH 7.5 (HCl), 1 mM MgCl,. They were then divided into aliquots, frozen in liquid nitrogen, and stored at -80 "C. One freeze/thaw cycle did not modify the ATPase activity.
To measure the plasma membrane ATPase activity of cells at different growth stages (see Table III), 20-liter fermentations were carried out with YPS14 or YPS12-1 in 5.8% glucose and 2% KAT yeast extract. At various cell densities, a variable volume of culture was harvested so as to collect 5-7 g of cells. The pellet was washed once with 5 volumes of ice-cold water and resuspended in 250 mM sorbitol, 10 mM MgCL, 5% glycerol, 50 mM imidazole adjusted to pH 7.5 with NaOH and immediately frozen in liquid nitrogen. After thawing, an antiprotease mixture containing 1 mM phenylmethylsulfonyl fluoride and chymostatin, aprotinin, pepstatin, E-64, leupeptin, antipain at the respective concentrations of 0.05, 0.2, 0.1, 0.7, 0.05, 0.25 mg/ml was added. The cells were immediately disrupted and their plasma membranes prepared as described above.
Protein concentrations were determined by the method of Lowry et al. (1951).
ATPase assays were performed at 35 "C in a medium containing 6 mM ATP, 9 mM M F (3 mM free M e ) , 50 mM MES-NaOH (pH 6.0), and 10 mM sodium azide, as described in Dufour et al. (1988), with the following modifications: the reaction was started by addition of 50 pl of enzyme in 450 p1 of reaction medium concentrated 10/9. After 5, 10,15, and 20 min, 100-pl aliquots were mixed with 1% SDS to stop the reaction. Specific activities were calculated from the slopes of Pi release versus time.
Protein Gels and Western Imrnunodetection-Plasma membrane proteins were analyzed on a Tricine-SDS-PAGE polyacrylamide gel as in Schigger and von Jagow (1987). The gels were fixed and stained as in Navarre et al. (1992).
For immunodetection, the proteins were separated on a standard SDS-PAGE polyacrylamide gel (Laemmli, 1970) and blotted onto a nitrocellulose membrane by means of the Bio-Rad Trans-BlotTM system. The blotted proteins were incubated first with antibodies raised against purified plasma membrane H+-ATPase? then with %labeled protein A, as described in Harlow and Lane (1988).
Immunoelectron Microscopy-Cells were harvested at a density of 25-40 10' cells/ml, washed with PBS (20 mM PO!-, 140 mM NaCI, pH 7.21, and fixed overnight at 4 "C as a suspension in PBS containing 4% paraformaldehyde and 0.5% glutaraldehyde. The cells were treated according to Van Tuinen and Riezman (1987) with 1% NaIO, and further with NHICl. After inclusion in low melting point agarose, the cells were dehydrated with graded ethanol solutions and embedded in LR white (The London Resin Co. Ltd, Cardiff, UK). Ultrathin sections were labelled with polyclonal rabbit antibodies (raised against presumed cytoplasmic regions of PMAl ATPase6) and protein A-gold (10 nm in diameter) as described by Roth et al. (1978). The antibodies used cross-react with the PMAZ protein (not shown).

RESULTS
Overexpression by Increased Dosage of the PMAB Gene-The PMAB gene under the control of its own promoter was inserted into a multicopy 2p vector, yielding plasmid pPSUP2 ( Fig. lA). Besides the URA3 marker, this plasmid also bears the ku2-d marker which can increase the copy number of the 2p plasmid up to 300 when selected (Beggs, 1978). By selection for leucine auxotrophy, plasmid pPSUP2 was introduced into diploid strain MG1 containing one URA3-disrupted genomic copy of the PMAl gene and one intact genomic copy (Table  I). In 18 dissected tetrads, only one or two viable spores were obtained per tetrad. As pPSUP2 is not highly unstable (data not shown), this indicates that multiple copies of PMAB cannot compensate for disruption of the essential PMAI gene.
The following observations indicate that the presence of PMAI on the multicopy vector is toxic to the cell while the presence of PMAB on such a vector is not. First, YPS1-5A is transformed to the Ura+ phenotype at the same rate whether the transforming plasmid is pPSUP2 or the vector alone. When pPSUPlL is used, however, the transformation fre- quency is some 50 times lower. Furthermore, cells transformed with pPSUPlL grow slowly, unlike pPSUP2-harboring cells. Finally, though the plasmid remains intact in the transformants, Ura+ cells transformed with the PMAl -bearing plasmid fail to grow on minimal leucine-free medium, contrary to cells transformed with pPSUP2 or the unaltered vector (data not shown). We conclude that for the PMAl-bearing plasmid, the highest copy number tolerated by yeast cells is below that required for sufficient expression of the partially defective leu2-d marker.
pmal-1 mutants cannot grow on a medium supplemented with 0.2 M acetate at pH 5.0 (Van Dyck et dl., 1990), probably because the mutant ATPase is not active enough to reject the protons brought into the cytoplasm by permeant protonated acetate molecules (McCusker et al., 1987). Fig. 2A shows that pPSUP2 does not rescue pmal-1 cells from their sensitivity to such acidic conditions.
When the pmal-1 strain is transformed with pPSUPlL ( Fig. 3), the ATPase activity of crude membrane preparations exhibits higher vanadate sensitivity. Plasmid pPSUP2 also increases sensitivity to vanadate, but to a much lesser extent. From the results presented above, it appears that the PMA2 gene borne on a high copy number plasmid fails to complement PMAl defects both in vivo and in vitro.
Overexpression of the PMA2 Gene with the PMAl Promoter-Three fusion hybrids were constructed in a centrom- . Cells were harvested a t a density of about 30 X lo6 cells/ml (the two former strains were grown in minimal medium, the latter in minimal medium plus leucine). No glucose or sorbitol incubation was carried out prior to cell disruption. The "Van(adate) sen.s(itive)" activity is the difference between the activities measured without ("Tot(a1)" activity) and with 20 p~ vanadate. eric CEN, LEU2 vector (pPS15-P12-n, Fig. 1B) so as to place the PMA2 coding region under the control of the PMAl promoter. The only difference between the three fused sequences is the length of PMA2 promoter sequence left between the PMAl promoter and the.PMA2 start codon (see "Experimental Procedures").
The plasmid shuffling method (Boeke et al., 1987) was used to test the ability of the fusion hybrids to complement PMAl gene inactivation. To do this, diploid strain YPS2-2 was constructed, in which the coding region of one genomic PMAl copy is completely replaced by the HIS3 marker and the coding region of one PMA2 genomic copy is completely deleted and replaced by the TRPl marker.
YPS2-2 was first transformed with pPS16-P1 (PMA1, CEN, URA3). After sporulation, the Ura+ His' TI$ spore YPS9-7A was selected. As expected, pPS16-P1 is 100% stable under conditions which are non-selective for the auxotrophic marker. Strain YPS9-7A was transformed with each of the three different pPS15-Pl2 plasmids (PPMAI::PMA2, CEN, LEU2). The resulting transformants are resistant to 5-FOA (Fig. 4A), indicating that they have shuffled out the URA3 plasmid pPS16-P1. The loss of pPS16-P1 and the integrity of the complementing plasmid were confirmed by Southern blot analysis (data not shown). The complementing plasmid was conserved under non-selective conditions. Besides, pPS15-P12-21 was directly introduced into diploid strain YPS2-2. After sporulation and dissection, complete tetrads were obtained (Fig. 4B, upperpanel). The HIS3 marker for the PMAl deletion was found to cosegregate with the plasmid-borne LEU2 marker (Fig. 4B, lower panel). This indicates that the PMA2 gene, when controlled by the PMAl promoter, is sufficiently overexpressed to complement a PMAl deletion.
Purification of Plasma Membrane-bound PMA2 ATPase-Haploid cells deleted of their genomic PMAl and PMAS gene copies and complemented by overexpression of PMAS (strain YPS12-1) or by PMAl (strain YPS14-4) were grown, washed with water, and incubated with either glucose or sorbitol. Glucose stimulates plasma membrane ATPase activity (Serrano, 1983) while sorbitol does not.6 In the glucose medium, cells expressing PMA2 did acidify the external medium but this acidification, averaged over four experiments, amounted to only 2.6 & 0.3 pH units, as opposed to 3.4 & 0.2 pH units for PMAl -expressing control cells.
Plasma membranes of incubated PMA2-expressing cells were isolated. Their plasma membrane ATPase activity was stimulated by glucose, but only about 1.1-to 1.4-fold as opposed to 2-fold for PMAl-expressing cells (Table 11). With or without stimulation, the activity was lower in membranes prepared from PMA2-expressing cells than in control membranes, being two times lower after sorbitol incubation and three times lower after glucose incubation. After Triton X-100 treatment, PMA2 ATPase activity was three times lower than PMAl ATPase activity after sorbitol incubation and five times lower after glucose incubation.
The plasma membrane protein profile obtained after Triton X-100 treatment was examined on a Coomassie Blue-stained SDS-PAGE gel. As compared with the control strain YPS14-4, the plasma membranes of YPS12-1 no longer contain the -100-kDa PMAl subunit. Instead, there is a -105-110-kDa protein (predicted molecular mass of PMA2 = 102 kDa) not present in the control (Fig. 5A, compare lanes 1 and 2 with  lanes 3 and 4). The latter cross-reacts with antibodies raised against purified plasma membrane H'-ATPase (Fig. 5 B ) . Glucose stimulation does not affect the abundance of PMA2 or PMAl protein (Fig. 5, A  The 15,000 X g 40-min pellet. 4 ) . This is in keeping with the known mechanism of glucose activation which modifies the kinetic properties of the ATPase but not its synthesis (Serrano, 1983). On the other hand, the overexpressed PMA2 ATPase appears less abundant than PMAl in plasma membranes, as measured by Coomassie Blue staining (Fig. 5 A ) (or silver staining, not shown). The low PMA2 ATPase activity in plasma membranes thus correlates with a lower enzyme concentration. This might account a t least partially for the low immunoreactivity of the PMA2-containing plasma membranes with the anti-PMAl antibody (Fig. 5B). Similar results were obtained with plasma membranes before Triton X-100 treatment (not shown).

lanes 3 and
The presence of quickly migrating proteolipids associated with the plasma membrane ATPase (Navarre et dl., 1992) is not affected by substituting the PMAB ATPase for the PMAl ATPase (Fig. 5A).
The low plasma membrane ATPase activity displayed in uitro by cells expressing PMA2 affects but slightly their growth in standard medium. Doubling times of 1.8 h and 1.6 h were measured in rich medium for strain YPS12-1 and control strain YPS14, respectively. However, in contrast with PMAl-expressing controls, cells expressing the PMA2 ATPase fail to grow on a medium containing 0.2 M acetate at pH 5.0 (Fig. 2B) and grow poorly at pH 3.0 (not shown).
Plasma membranes were purified from cells harvested around 10,50, and 220 X lo6 cells/ml. In each of these samples, the PMA2 plasma membrane ATPase activity was three to five times lower than the PMAl ATPase activity measured in similar experiments (Table 111).  2 and 4 ) or glucose (lanes I   and 3), were electrophoresed, fixed, and stained with Serva blue G. B, immunoautoradiogram of a protein transfer blot with polyclonal antibodies raised against purified S. cerevisiae plasma membrane ATPase. The same plasma membrane proteins as in ( A ) (0.5 pg) were run on a standard SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and tested for reaction with anti-ATPase antibodies. growth stages ATP hydrolysis rates were measured as described in Table 11. Plasma membranes were purified in parallel from 5-7-g cell aliquots of YPS12-1 (PMA2) and of YPS14-4 (PMAI), previously harvested at the indicated cell densities and frozen, as described under "Experimental Procedures."

ATPase expressed
Cell density Plasma membrane ATPase activity  Table I 1 (cell densities of 225 X lo6 and 217 X lo6 for YPS12-1 and 14-4, respectively) but harvested separately and frozen, as described under "Experimental Procedures." Approximately 10 pg of RNA were loaded onto a formaldehyde gel, electrophoresed, and transferred to a nitrocellulose membrane. A BamHI-EagI DNA fragment spanning 0.95 kb upstream from the PMAl start ATG was used to probe the transcript leaders of PMAI (233 bases, Capieaux et a[. (1989)) and of the p p~~, : : p M A 2 fusion. As an internal control for standardizing RNA levels, a 1.1-kbp RamHI-Hind111 DNA fragment spanning the actin gene was used to probe the actin transcript.

Effects of Gene Dosage and Morphological Consequences of
Overexpression-Total RNA was extracted from YPS12-1 and YPS14 and probed with a PMAl promoter DNA fragment. The hybrid messenger produced by fusion between the PMAl promoter and the PMA2 coding region was not less abundant than the control PMAI mRNA (Fig. 6).
We tested the effect of gene dosage on PMA2 plasma membrane ATPase activity. A fragment obtained by fusing the PMAl promoter with the PMAZ coding and terminator regions was subcloned into the multicopy vector described above. The resulting plasmid pPSUP12 (PPM~~::PMAP, 2p, URA3, leu2-d, Fig. lA), and the control plasmid pPSUPlL (PMAl, 2p, URA3, leu2-d) were then introduced separately into diploid strain YPS2-2 and clones were selected for uracil auxotrophy. After sporulation, complete tetrads were obtained in both cases. Interestingly, in both cases, the Ura+ segregants failed to grow on leucine-free minimal medium, indicating

Effect of gene dosage on PMA2 and PMA I plasma membrane
ATPase activity ATP hydrolysis rates were measured as described in Table 11. Plasma membranes were purified in parallel from YPS12-1OD ([CEN, that expression of PMA2 under the control of the PMAl promoter is tolerated only as long as the plasmid copy number remains too low to allow leu2-d selection. This suggests that an excess of either PMAl or PMA2 ATPase is toxic to the cell. Ura+ His' Trp+ haploids, deleted of both genomic PMA copies and complemented by pPSUP12 (YPS22-9C) or pPSUPlL (YPS3-2A), were selected. We determined their plasma membrane ATPase activity and that of YPS12-1OD control cells (expressing PMAZ on the CEN-LEU2 plasmid pPS15-P12-21) after incubation with glucose or sorbitol (Table IV). The PMAl plasma membrane ATPase activity of YPS3-2A (multicopy , Table IV) and YPS14-4 (centromeric ,  Table 11) is similar. On the other hand, a comparison of YPS22-9C and YPS12-1OD shows that overexpression of the PMA2 gene on a multicopy vector rather than a centromeric vector does not increase the PMA2 plasma membrane ATPase activity, which thus remains low. Consistently, no appreciable variation in concentration of ATPase was observed as the plasma membrane fraction from YPS22-9C and YPS12-1OD or from YPS3-2A and YPS14-4 were compared (data not shown).

P~M A I : : P M A~] ) , YPS22-9C ([2/~, PPMAI.':PMA~]), and YPS 3-2A ([2p,
The cellular morphology of various strains containing the double genomic deletion and expressing either the PMAl or the PMA2 ATPase was examined by electron microscopy. PMA ATPases tagged with antibodies raised against the cytoplasmic domains of the PMAl protein (gift of E. Capieaux) were gold-labeled. Cells expressing the PMAl ATPase from the centromeric plasmid look normal and almost all the gold particles are located, as expected, in the plasma membranes (Fig. 7A). When cells of a strain expressing the PMA2 ATPase from the centromeric plasmid are examined, this "normal" phenotype is found in the majority of cell sections. However, about 10% of the sections display a proliferation of intracellular membranes, which are extensively gold-labeled (Fig. 7B), indicating a local concentration of PMA2 ATPase. The proportion of sections exhibiting this phenotype increases dramatically to about 40% when one looks a t cells expressing the PMA2 ATPase from the multicopy system (Fig. 7C). Interestingly, about 10% of sections of cells expressing the PMAl ATPase from the multicopy vector display a similar proliferation of gold-labeled structures (Fig. 70). The proliferated membranes probably derive from ER. Indeed, connections between these membranes and the nuclear cisternum were observed (Fig. 7E). Moreover, the proliferated structures both from PMA1' or PMA2 overexpressing cells (Fig. 7F) were labeled by antibodies against KAR2 (gift from M. D. Rose), which is a lumenal ER protein (Rose et al., 1989).

DISCUSSION
Yeast strains YPS2-2 (diploid) and YPS9-7A (haploid) are deleted of the entire coding regions of one genomic copy of PMAl and one genomic copy of PMA2. These strains present several advantages. First, the double deletion should increase the stability after complementation of a non-lethal pma mutant copy introduced on a plasmid. Indeed, several pmal mutations are repaired by recombination with another copy of PMAl or PMA2, when present (Harris et al., 1991). Second, when successfully complemented, these strains yield haploids which express the desired ATPase without any interference of the wild-type PMAl or PMA2 ATPases. Third, with haploid strain YPS9-7A) it is possible to use the plasmid shuffling method, faster and easier than spore segregation analysis, for the complementation tests.
Is PMAB a Plasma Membrane ATPase?-By introducing into these strains the PMAB gene controlled by the PMAl promoter, we have shown that the PMAS gene can functionally replace the PMAl gene. Consistently, the presence of PMAS protein in the plasma membrane was demonstrated in vitro, by SDS-polyacrylamide gel analysis and Western blot analysis of the plasma membrane fraction, and in situ, by immunoelectron microscopy.
However, overexpressed PMA2 ATPase is not exclusively located in the plasma membrane. Our immunoelectron micrographs show that overexpression of the PMA2 ATPase causes proliferation of intracellular membranes in which overexpressed PMAB protein accumulates. However, there are several reasons to believe that this intracellular location is not the protein's true physiological location but more likely a general effect produced by overexpression of membrane proteins. First, overexpression (by gene dosage) of PMAl ATPase, a known plasma membrane protein, leads to similar, albeit lesser, accumulation in intracellular membranes. Second, the proliferation of membranes and the quantity of PMAS ATPase present in these structures both depend on the expression level, as inferred from our comparison of overexpression from centromeric and multicopy vectors. Third, high expression in S. cerevisiae of a plasma membrane H+-ATPase from a plant (Villalba et al., 1992)' or of membrane proteins of the ER, such as indigenous 3-hydroxy-3methylglutaryl-CoA reductases (Wright et al., 1988) or P-450 cytochromes from Candida maltosa (Schunck et al., 1991), also leads to proliferation of intracellular membranes with the highly expressed proteins accumulating therein.
Our data further show that the activity and the amount of PMA2 (or PMA1) ATPase in the plasma membrane do not increase when the ATPase is expressed from a multicopy plasmid rather than from a centromeric plasmid. This is in agreement with the observations of Eraso et al. (1987), who reported only a slight increase in the amount of the plasma membrane ATPase when the PMAl gene was expressed from a multicopy plasmid. This suggests that the plasma membrane localization of the PMAS enzyme is independent of the overexpression level and thus probably represents the true physiological localization of the PMAZ ATPase.
IS the PMAB ATPase an H+-ATPase2"Functional replacement of the PMAl ATPase with the PMA2 ATPase gives strong support to the assertion that PMA2 is an H+-ATPase, a A. de Kerchove, P. Supply, and A. Goffeau, unpublished results. as already suggested by the high identity of the PMAl and PMA2 amino acid sequences . Direct in vivo evidence is provided by the observation that cells expressing only the PMAZ ATPase still acidify the external medium in response to glucose addition. Indeed, the low final pH (<4) reached in the acidification measurements indicates that this acidification is not only due to excretion and dissociation of weak acid metabolites (mainly carbonic acid, pKl = 6.4) but also to proton pumping (Ulaszewski et al., 1987b). There is generally a positive correlation between proton efflux and PMAl plasma membrane ATPase activity (Serrano, 1983;Ulaszewski et al., 1987b;Cid et al., 1987;Portillo and Serrano, 1989). A similar correlation is observed for PMA2; the lower ATPase activity corresponding with a lesser acidification of the medium. Intracellular Traffic Control of the Amount of Plasma Membrane ATPase-Why is the ATPase activity of plasma membranes containing the PMA2 enzyme lower than that of plasma membranes containing the PMAl enzyme? Growth effects can be ruled out, since the lower activity is observed at various points of the growth curve. On the other hand, two results indicate that the low in vitro activity of PMA2 ATPase in plasma membranes probably reflects its low in vivo activity.
Indeed, cells expressing the PMAB ATPase acidify the medium less than do control cells. Moreover, PMAP-expressing cells fail to grow at pH 5.0 with acetate or at pH 3.0, conditions requiring high proton-pumping activity in order to maintain the intracellular pH. We show here that this lower ATPase activity is due to there being less PMA2 enzyme in the plasma membranes of cells expressing PMA2 than PMAl in the membranes of PMAl-expressing cells, even when both proteins are produced from genes controlled by a same promoter and borne by a same vector. Underexpression due to instability or defective translation of the PMA2 hybrid transcript can be ruled out for the following reasons: 1) the mRNA amount produced by PMAS under the control of the PMAl promoter is similar to that of the original PMAl gene, and 2) increasing the dosage of the hybrid PMAB gene, theoretically a means of bypassing transcriptional or translational failings by multiplying the expression units, fails to increase PMA2 plasma membrane ATPase activity. Eraso et al. (1987), examining the effects of PMAl gene dosage, found that the amount of plasma membrane ATPase is controlled at different levels. First, copy number of the PMAl-bearing plasmid is reduced compared to that of the original multicopy vector. A similar mechanism might partially explain that the PMAS plasma membrane ATPase activity does not vary whether the PMAB gene is overexpressed from a multicopy or a centromeric vector. But what would explain the low PMA2 activity when the hybrid gene is expressed from a centromeric plasmid? Posttranslational control was the second mechanism evidenced by Eraso et al. (1987), a mechanism preventing the PMAl plasma membrane ATPase from increasing in proportion to its mRNA. The authors proposed a control at the level of protein synthesis and/or degradation. Actually, immuno electron microscopy experiments reported here rather suggests posttranslational "traffic control" of overexpressed PMAl and PMAS: part of the enzyme produced is trapped in proliferating intracellular membranes, probably derived from ER, and would thus be unable to reach the plasma membrane. This trapping phenom- enon affects the PMA2 ATPase to a greater extent than the PMAl ATPase, which may explain the former protein's lesser abundance/activity in the plasma membrane.
The plasma membrane PMAl ATPase is transported to the cell surface via the secretory pathway (Brada and Schekman, 1988). During this intracellular transport, phosphorylation of the newly synthesized ATPase occurs in multiple steps (Chang and Slayman, 1991). Retention of overexpressed PMAl or PMAB protein in ER-derived membranes may suggest that their secretion is controlled at an early step in the transport process.
Eraso et al. (1987) suggest that the toxicity of overexpressed plasma membrane ATPase could be due to structural constraints in the membrane, limiting the amount of inserted ATPase. If so, it could be advantageous for the cell to control secretion: the proliferated membrane structures could be a dead end or a waiting room for PMAl or PMAB ATPase in excess.
Specificity of Intracellular Retention-Our results indicate that when both proteins are expressed at equivalent level in separate cells, the proportion of protein targeted to the plasma membrane is higher for the PMAl than for the PMAB ATPase. Furthermore, when a plant plasma membrane ATPase is expressed in yeast, nearly no heterologous protein is detected in the plasma membrane (Villalba et al., 1992). This suggests that the yeast PMAl H+-ATPase is specifically favored in the secretion process.
Specificity in the yeast intracellular transport process has recently been evidenced by Ljungdahl et al. (1992). These authors have shown that the inactivation of an ER integral membrane protein, SHR3, causes a specific block of amino acid permeases in the ER whereas transport of PMAl plasma membrane ATPase, secretory proteins and vacuolar proteins is not affected. Ljungdahl et al. (1992) suggested that SHR3 might function as a permease-specific "adapter" molecule interacting with components of the general secretory pathway or as catalyzer of specific folding or translocation reactions, necessary for transport of permeases from ER. The differences in intracellular secretion process of PMA1, PMA2, and plant H'-ATPases may reflect the existence of a SHR3analog in the secretory pathway, specifically required for targeting of plasma membrane ATPase. This analog could preferentially interact with a structural domain of the yeast constitutive PMAl H'-ATPase. Differences of conservation of this domain in PMAP and plant H+-ATPases could explain why the former is partially targeted to plasma membrane whereas the latter is nearly totally blocked in the ER.

Failure of Overexpression by Increased Dosage of the PMAB Gene under Control of Its Own
Promoter-Multiple copies of the complete PMAB gene failed to complement PMAl defects in vivo or in vitro. Moreover, the toxicity associated with increased PMAl gene dosage was not observed with a similar dosage of PMA2. As positive results were obtained with the PMA2 gene controlled by the PMAl promoter, we may conclude a posteriori that the complete PMA2 gene is probably insufficiently expressed, even when present in high copy number. The S. pombe pma2 gene likewise functionally replaces the pmal gene when controlled by the adh promoter but not when controlled by its own promoter (Ghislain and Goffeau, 1991). This is consistent with the extreme weakness of the PMA2 promoter, observed by Schlesser et al. (1988); under normal conditions. PMA2 ATPase might possibly be expressed only under conditions requiring its specific catalytic properties, which are described in a companion study (Supply et al., 1993).