Structure-Function Relationships in Membrane Segment 5 of the Yeast Pma1 H+-ATPase*

Membrane segment 5 (M5) is thought to play a direct role in cation transport by the sarcoplasmic reticulum Ca2+-ATPase and the Na+,K+-ATPase of animal cells. In this study, we have examined M5 of the yeast plasma membrane H+-ATPase by alanine-scanning mutagenesis. Mutant enzymes were expressed behind an inducible heat-shock promoter in yeast secretory vesicles as described previously (Nakamoto, R. K., Rao, R., and Slayman, C. W. (1991) J. Biol. Chem. 266, 7940–7949). Three substitutions (R695A, H701A, and L706A) led to misfolding of the H+-ATPase as evidenced by extreme sensitivity to trypsin; the altered proteins were arrested in biogenesis, and the mutations behaved genetically as dominant lethals. The remaining mutants reached the secretory vesicles in sufficient amounts to be characterized in detail. One of them (Y691A) had no detectable ATPase activity and appeared, based on trypsinolysis in the presence and absence of ligands, to be blocked in the E1-to-E2 step of the reaction cycle. Alanine substitution at an adjacent position (V692A) had substantial ATPase activity (54%), but was likewise affected in the E1-to-E2 step, as evidenced by shifts in its apparent affinity for ATP, H+, and orthovanadate. Among the mutants that were sufficiently active to be assayed for ATP-dependent H+ transport by acridine orange fluorescence quenching, none showed an appreciable defect in the coupling of transport to ATP hydrolysis. The only residue for which the data pointed to a possible role in cation liganding was Ser-699, where removal of the hydroxyl group (S699A and S699C) led to a modest acid shift in the pH dependence of the ATPase. This change was substantially smaller than the 13–30-fold decrease in K+affinity seen in corresponding mutants of the Na+,K+-ATPase (Arguello, J. M., and Lingrel, J. B (1995) J. Biol. Chem. 270, 22764–22771). Taken together, the results do not give firm evidence for a transport site in M5 of the yeast H+-ATPase, but indicate a critical role for this membrane segment in protein folding and in the conformational changes that accompany the reaction cycle. It is therefore worth noting that the mutationally sensitive residues lie along one face of a putative α-helix.

brane H ϩ -ATPase. This enzyme is encoded by the PMA1 gene, accounts for 10% of plasma membrane protein, and splits as much as one-quarter of the ATP produced by the cell (reviewed in Ref. 15). It generates the proton electrochemical gradient that underlies nutrient uptake and, consistent with its key physiological role, is essential for cell viability. While the mechanism of proton transport by the yeast ATPase has not yet been studied in detail, work on a very closely related pump (the Pma1 ATPase of the filamentous fungus Neurospora crassa) provides evidence for a simple stoichiometry of 1 H ϩ translocated per ATP split (16).
In this study, we have carried out alanine-scanning mutagenesis along the full length of M5 in the yeast ATPase. The results have identified five amino acid residues that play a significant role in the reaction cycle, along with three others that are required for proper protein folding and transit through the secretory pathway.
Mutagenesis-Mutagenesis (20) was performed on a 519-base pair BglII-SalI restriction fragment subcloned into a modified Bluescript plasmid (Stratagene, La Jolla, CA). Following DNA sequencing, the BglII-SalI fragment carrying the mutation was moved into plasmid pPMA1.2 (18). The 3.8-kilobase HindIII-SacI fragment, which contains the entire pma1 coding region, was cloned into the yeast expression vector YCp2HSE (18), placing the mutant allele under heat-shock control. Plasmids were then transformed into yeast according to the method of Ito et al. (21).
Isolation of Secretory Vesicles and Measurement of Expressed ATPase-Transformed SY4 cells were grown to mid-exponential phase (A 600 ϳ 1) at 23°C in supplemented minimal medium containing 2% galactose, shifted to medium containing 2% glucose for 3 h, and then heat-shocked at 39°C for an additional 2 h. The cells were harvested and washed, and the secretory vesicles were isolated as described previously (22). To determine the level of expressed Pma1 protein relative to a wild-type control, secretory vesicles (5-20 g) were subjected to SDS-polyacrylamide gel electrophoresis and immunoblotted (18), followed by PhosphorImager (Molecular Dynamics) analysis; typically, the analysis was carried out at two protein concentrations within the linear range, and the expression level was calculated from the average of the two determinations.
ATPase Activity-Unless otherwise noted, ATP hydrolysis was assayed at 30°C in 0.5 ml of 50 mM MES/Tris, pH 5.7, 5 mM KN 3 , 5 mM Na 2 ATP, 10 mM MgCl 2 , and an ATP-regenerating system (5 mM phosphoenolpyruvate and 50 g/ml pyruvate kinase). The reaction was stopped after 20 -40 min, and the release of inorganic phosphate from ATP was determined by the method of Fiske and SubbaRow (23). Specific activity was calculated as the difference between ATP hydrolysis measured in the absence and presence of 100 M sodium orthovanadate, an inhibitor of P-type ATPases. For determination of K m values, the concentration of Na 2 ATP was varied between 0.15 and 7.5 mM, with MgCl 2 always in excess of ATP by 5 mM. Actual concentrations of MgATP were calculated as described previously (24). To determine the effects of pH on hydrolysis, the pH of the assay mixture was adjusted to values between 5.2 and 7.5 with Tris base.
Fluorescence Quenching-ATP-dependent proton transport was determined by measuring the initial rate of acridine orange fluorescence quenching as described by Ambesi et al. (25). The specific initial rate of fluorescence quenching for each mutant was adjusted for ATPase expression and is reported as a percent of the wild-type rate.
Metabolic Labeling and Immunoprecipitation-To measure the synthesis of mutant ATPases that were unable to reach the secretory vesicles, SY4 cells were shifted from galactose medium at 23°C to glucose medium at 39°C as described above and then metabolically labeled with [ 35 S]methionine (26). Total membranes were isolated and immunoprecipitated with anti-Pma1 antibody (26), and after SDS-polyacrylamide gel electrophoresis, the gels were fixed, incubated in 1 M sodium salicylate (30 min at 23°C), dried, and exposed to Hyperfilm-MP (Amersham Pharmacia Biotech).
Trypsinolysis-Limited trypsinolysis was performed on both isolated secretory vesicles and 35 S-labeled yeast total membranes. Vesicles or membranes were diluted into 1 mM EGTA/Tris, pH 7.5; centrifuged at 100,000 ϫ g for 35 min; and suspended at 0.5 mg/ml in 20 mM Tris-HCl, pH 7.0, and 5 mM MgCl 2 . Following preincubation in the absence or presence of 100 M orthovanadate, 10 mM MgADP, or 10 mM MgATP at 30°C for 5 min, tosylphenylalanyl chloromethyl ketone-treated trypsin was added (trypsin/protein ratio of 1:4 for secretory vesicles or 1:20 for total membranes), and the incubation was continued for 0.5-20 min. The reaction was terminated by the addition of 1 mM diisopropyl fluorophosphate. Reaction products were analyzed either by immunoblotting (secretory vesicles) or by immunoprecipitation and fluorography (total membranes).
Protein Determination-Protein concentrations were determined by a modification of the method of Lowry et al. (27) as described by Ambesi et al. (22) using bovine serum albumin as a standard.

RESULTS
Expression and ATP Hydrolysis-In this study, alaninescanning mutagenesis was used to examine the structural and functional role of amino acids in M5 of the yeast plasma membrane H ϩ -ATPase. Residues Ser-690 to Leu-713 were included based on hydropathy analysis of the Pma1 protein sequence (reviewed in Ref. 15). All but two of the residues were replaced with alanine, whereas alanines at positions 697 and 711 were replaced with serine. The mutant alleles were transformed into yeast strain SY4, expressed under control of an inducible heatshock promoter, and secretory vesicles were isolated and characterized with respect to expression and ATP hydrolysis.
Most substitutions allowed reasonable amounts of the ATPase to reach the secretory vesicles, but there were three cases (R695A, H701A, and L706A) in which little or no Pma1 protein could be detected in the vesicles ( Fig. 1A and Table I, part A). Immunoprecipitation from 35 S-labeled total membranes revealed that each of these mutant proteins was synthesized, but became arrested in an earlier compartment of the secretory pathway, presumably the endoplasmic reticulum (Fig. 1B).
The remaining 21 mutant ATPases reached the secretory vesicles in amounts ranging from 39 to 99% of the wild-type control (Table I, part A). When these mutants were assayed for their ability to hydrolyze ATP, 17 of them had activities of 37% or better after correction for the level of expression in the secretory vesicle preparations. Three showed significant reductions in activity: Y694A to 22%, W709A to 21%, and S699A to 8% of the wild-type control. One mutant ATPase (Y691A) was expressed well in secretory vesicles (92%), but appeared to be completely inactive. Y691A and S699A, along with R695A, H701A, and L706A, which were blocked in biogenesis, were later examined for structural defects by limited trypsinolysis, as described below.
Kinetic Properties-The mutant ATPases were next assayed for vanadate sensitivity, MgATP dependence, and the effect of pH on the rate of ATP hydrolysis. In all but one case, the kinetic properties of the mutants proved to be essentially normal (Table II, part A). The exception was V692A, which displayed an increased K i for vanadate (11 M compared with 1.8 M for the wild-type control), a decreased K m for MgATP (0.1 mM compared with 1.1 mM for the wild type), and a relatively alkaline pH optimum (pH 6.4 compared with pH 5.7 for the wild type). Because Val-692 is presumably buried in the membrane, it is unlikely to contribute in a direct way to the vanadate-and MgATP-binding sites. Rather, the increased K i , decreased K m , and altered pH optimum can more reasonably be accounted for by a slowing of the E 1 P-to-E 2 P conformational change; as a result, the ATPase accumulates in E 1 , which has a relatively high affinity for ATP and protons and a relatively low affinity for orthovanadate. This idea is supported by the fact that mutations at three positions in M4 lead to a similar set of kinetic changes (25).
ATP-dependent Proton Transport-Given the fact that membrane segment 5 is believed to play a direct role in cation translocation in the Ca 2ϩ -and Na ϩ ,K ϩ -ATPases, it was of particular interest to explore the proton-pumping ability of the yeast M5 mutants. For this purpose, secretory vesicle preparations were assayed for ATP-dependent quenching of acridine orange fluorescence. In seven of the mutants (Y691A, Y694A, R695A, S699A, H701A, L706A, and W709A), ATPase activities were below the limit at which associated proton pumping could have been reliably detected by the acridine orange assay. With one exception, all of the remaining mutants showed a reasonable correlation between the initial rate of ATP-dependent quenching and the rate of ATP hydrolysis (Table I, part A). In V692A, the rate of acridine orange quenching (123% of the wild-type control) appeared to exceed the rate of ATP hydrolysis (54% of the wild-type control). However, separate measurements indicated that the apparent discrepancy could be accounted for by the pH difference between the hydrolysis assay and the quenching assay, together with the above-mentioned alkaline shift in pH optimum. Thus, at pH 6.7, the V692A enzyme split ATP at 119% of the wild-type rate, completely consistent with its relative rate of acridine orange quenching at the same pH (123%; see Table I, part A). There was no evidence for abnormal ATP-dependent proton transport in any of the other 16 mutants that were studied (Table I,

part A).
Further Study of Tyr-694, Ser-699, and Glu-703-Based on comparison with other P-type ATPases, it seemed useful to make additional amino acid replacements at three positions: Tyr-694, Ser-699, and Glu-703. In the case of Tyr-694, Andersen (28) has reported that mutation of the corresponding residue to glycine leads to uncoupling of the SERCA Ca 2ϩ -ATPase; the other two residues are believed to play a role in cation binding in the Na ϩ ,K ϩ -and SERCA Ca 2ϩ -ATPases, a Calculations were made from yields of mutant and wild-type H ϩ -ATPase protein/mg of total secretory vesicle protein, as determined by quantitative immunoblotting (see "Experimental Procedures").
b Vanadate-sensitive ATP hydrolysis was measured as described under "Experimental Procedures." One unit is defined as 1 mol of P i /min. c The initial rate of fluorescence quenching (proton transport) was determined as previously described (25). One unit is defined as 1% of total fluorescence quenching/min. d ND, not determined. Corrections for expression have not been made for mutants with measured ATPase activities below 3% of the wild-type value, but have been made for mutants with measured activities between 3 and 10%. In both cases, proton transport could not be detected by the acridine orange assay.
At Tyr-694 of the yeast ATPase, substitution by Gly gave better expression (Y694G, 64%; Table I, part B) than substitution by Ala (Y694A, 40%; Table I, part A). Interestingly, the Y694G enzyme was extremely resistant to orthovanadate, with a K i of 60 M (Table II, part B). This large increase in K i , with only minor changes in K m and pH optimum, suggests that the mutation has a relatively specific effect on vanadate binding, raising the possibility that the cytoplasmic end of M5 may somehow interact with the vanadate-binding pocket. Y694G also showed an apparent difference between the rate of ATP hydrolysis (18% before correction for the level of expression in secretory vesicles, 28% after correction) and the rate of acridine orange fluorescence quenching (8% before correction, 12% after correction). Taken at face value, this difference could indicate partial uncoupling, but the rates are below the limit at which a detailed analysis is possible.
In the case of Ser-699, both the Cys and Thr mutants were well expressed, but S699T was considerably more active (96% hydrolysis, 100% transport) than S699C (16% hydrolysis, 12% transport). In both cases, the pH profile for ATP hydrolysis was examined to see whether there was any evidence for a decrease in the apparent affinity of the ATPase for protons, similar to the decrease in K ϩ affinity reported for the corresponding mutants of the Na ϩ ,K ϩ -ATPase (14). As shown in Fig. 2, the pH curve for S699C did indeed shift in the acid direction, but only modestly (ϳ0.2 pH units). A similar shift of ϳ0.2 pH units was seen in S699A (data not shown), although in this case, the ATPase activity was very low. By contrast, the more conservative substitution in S699T had no effect on the pH profile.
Finally, although the substitution of Glu-703 by Asp (56% expression, 39% hydrolysis, 40% transport) was less well tolerated than the substitution by Ala (95% expression, 105% hydrolysis, 92% transport), there was sufficient activity in both cases to carry out a quantitative study of H ϩ pumping over a range of MgATP concentrations. The results are illustrated in Fig. 3. As reported earlier (25), the relationship between the initial rate of acridine orange fluorescence quenching and the rate of hydrolysis was approximately linear in the wild-type control. Significantly, the data points for E703A and E703D fell along the same straight line, consistent with normal coupling between H ϩ pumping and ATP hydrolysis. Conformational Analysis by Limited Trypsinolysis-As described above, three mutations in M5 led to a virtually complete block in Pma1 biogenesis: R695A, H701A, and L706A. Because it seemed possible that these proteins might be structurally abnormal, they were examined by limited tryptic digestion. Total membranes were isolated from 35 S-labeled cells expressing either mutant or wild-type ATPase and incubated at a trypsin/protein ratio of 1:20 for 0, 1, or 5 min (Fig. 4). Under these conditions, the 100-kDa wild-type enzyme underwent minor proteolytic cleavage, yielding a 97-kDa band, whereas the three mutants were completely degraded in Ͻ1 min. Thus, alanine substitutions of Arg-695, His-701, and Leu-706 appear to cause severe misfolding, interfering with the ability of the newly synthesized ATPase to move through the secretory pathway.
Two other mutants were studied by limited trypsinolysis: Y691A and S699A. Both were well expressed in secretory vesicles (92 and 90%, respectively), but Y691A had no detectable ATPase activity, and S699A had very low activity (8%). To examine the folding of the mutant proteins, vesicles were incubated at a trypsin/protein ratio of 1:4 in the absence of ligands (Fig. 5A). Under the conditions of this experiment, the 100-kDa wild-type ATPase was rapidly converted to the 97-kDa form, which in turn was digested to smaller fragments. Significantly, Y691A and S699A were no more trypsin-sensitive than the wild-type enzyme (Fig. 5A), consistent with their ability to undergo normal trafficking. However, a difference between the two mutants became apparent in the experiment of Fig. 5B, which explored the ability of ligands to protect against tryptic degradation. MgADP and MgATP (which should hold the enzyme in the E 1 conformation) protected 100-, 97-, and 62-kDa protein fragments in the wild-type and both mutant ATPases, whereas vanadate (which should pull the enzyme into the E 2 conformation) protected 97-and 80-kDa fragments in the wild- type and S699A ATPases, but not in Y691A. Thus, like its neighbor (V692A), Y691A appears to be defective in the E 1to-E 2 conformational change.
Coexpression of M5 Mutants with Wild-type PMA1-Wach et al. (29) have previously shown that Asp, Gln, and Arg substitutions of His-701 behave as dominant lethal mutations. In a similar series of experiments, we subcloned H701A, R695A, and L706A into plasmid YCplac33 (30) and placed the mutant alleles under GAL1 control. Each of the plasmids was then transformed into yeast strain NY605, which carries a wild-type PMA1 gene on the chromosome. When the resulting cells were plated on glucose-containing medium, where only the constitutive wild-type gene could be expressed, all of them grew normally. By contrast, on galactose-containing medium, where the mutant gene was also expressed, no growth was observed for cells transformed with H701A, R695A, or L706A. This dominant lethal phenotype indicates that the abnormal proteins interfere with the processing of coexpressed wild-type ATPase, as has previously been shown for H701D, H701Q, H701R, and a number of other PMA1 mutations (26,29,(31)(32)(33). DISCUSSION In analyzing the data from this study, the sequence alignment of Fig. 6 can serve as a useful guide. It illustrates a modest degree of evolutionary conservation along membrane segment 5 of the P-ATPases and highlights six residues that are discussed in detail below.
Tyr-694 lies near the cytoplasmic boundary of M5 and is present in all known P-type H ϩ pumps from fungi, algae, higher plants, and protozoans. Other members of the family have a similar amino acid residue at this position (usually Tyr or Phe; occasionally Trp, Met, or Leu), except for the Cta3 Ca 2ϩ -ATPase of Schizosaccharomyces pombe, where there is a His. It is now clear that Tyr-694 plays a key role in the reaction cycle of the yeast Pma1 H ϩ -ATPase since ATP hydrolysis decreased substantially in Y694A and Y694G, and ATP-dependent H ϩ transport was barely detectable. In the mammalian SERCA ATPase, as mentioned above, mutation of the corresponding Tyr to Gly led to the uncoupling of Ca 2ϩ transport from ATP hydrolysis (28); and in the Na ϩ ,K ϩ -ATPase, the same substitution led to inactivation of the enzyme (34). Thus, there appears to be a nearly uniform requirement for a bulky, hydrophobic amino acid at the position of Tyr-694.
Ser-699 is another residue displaying significant, although not complete, evolutionary conservation (Fig. 6). All of the known P-type H ϩ -ATPases contain either Ser or Thr at this position, and there is a Ser only one residue away in the mammalian Na ϩ ,K ϩ -and SERCA Ca 2ϩ -ATPases, the MgtB ATPase of Salmonella typhimurium, and the Cta3 ATPase of S. pombe. Clear evidence has been put forward for the functional importance of this Ser in the Na ϩ ,K ϩ -ATPase, where mutation to Ala or Cys produced a severalfold decrease in the V max for ATP hydrolysis and a 13-31-fold increase in the K1 ⁄2 for K ϩ ; as a result, cells expressing the altered pump required higher than normal extracellular K ϩ for growth (14). By contrast, Ser-to-Ala and Ser-to-Cys mutations had little effect on either the activity or the Ca 2ϩ affinity of the SERCA ATPase (35). In this study, yeast Pma1 ATPase containing the S699A mutation traveled efficiently to the secretory vesicles, and based on limited trypsinolysis, it was well folded and able to bind MgATP, MgADP, and orthovanadate; however, it had exceedingly low ATPase activity and no detectable ATP-dependent proton pumping. The S699C and S699T enzymes were more active, allowing them to be studied in greater detail. S699T appeared normal in every respect, but S699C displayed a modest acid shift in the pH dependence of ATP hydrolysis (0.2 units). While the shift does not necessarily reflect a lowered affinity for H ϩ at a cation-liganding site in the translocation pathway, this is clearly one possibility. If so, the effect is substantially smaller than the change seen in the Na ϩ ,K ϩ -ATPase (14), where recent work by Blostein et al. (36) has provided direct evidence that Ser-775 is either a cation-liganding residue or part of a gating structure close to the liganding sites. The difference between the two enzymes may be related to the immediate downstream sequences, which are LHLE in the case of the H ϩ -ATPase and NIPE in the Na ϩ ,K ϩ -ATPase. As emphasized in a recent study of the NHE-1 Na ϩ /H ϩ exchanger by Counillon et al. (37), a proline residue leaves the backbone carbonyl at position Ϫ4 without a hydrogen donor and introduces a kink that disrupts hydrogen bonding between position Ϫ3 (Ser-775) and position ϩ1. Thus, the presence of a Pro in the Na ϩ ,K ϩ -ATPase would free up backbone carbonyls that might contribute to cation binding; it would also increase the local flexibility of the polypeptide chain, which in turn could influence conformational interactions with other parts of the protein.
Glu-703 is a third residue that deserves careful attention. In the mammalian SERCA ATPase, replacement of the corresponding glutamate (Glu-771) by an uncharged amino acid destroyed Ca 2ϩ occlusion (6) and Ca 2ϩ transport (3,4), consistent with the idea that a side chain oxygen is needed at this position to ligand Ca 2ϩ . Measurements of phosphorylation from ATP and P i as a function of Ca 2ϩ concentration soon led to the idea that Glu-771 contributes to the deeper (more luminal) of two sequentially filled Ca 2ϩ -binding sites (7,38). In the Na ϩ ,K ϩ -ATPase, the corresponding glutamate (either Glu-779 or Glu-781, depending upon the species) plays a less critical role since substitution by Ala, Asp, Gln, or Arg failed to disrupt function (10 -12). However, the Ala mutant did display a severalfold decrease in the apparent affinity for K ϩ and Na ϩ (11,12), along with the disappearance of voltage dependence in patch clamp experiments, suggesting that this residue may line the cation access channel (13). In the yeast Pma1 H ϩ -ATPase, on the other hand, there is no demonstrable role for Glu-703. Replacement by an uncharged amino acid (E703A) had little or no effect on biogenesis, ATPase activity, or the rate of ATP-dependent proton pumping. Furthermore, a detailed look at the relationship between H ϩ pumping and ATP hydrolysis in E703A and E703D gave no evidence for a defect in coupling. It may therefore be significant that Glu-703 is replaced by Val or Cys in the P-type H ϩ -ATPases from Leishmania donovani, Dunaliella bioculata, and Arabidopsis thaliana and by Ala in two Ca 2ϩ -ATPases (yeast Pmr1 and mammalian PMCA); it is also replaced by an uncharged amino acid (Gly or Met) in the heavy metal-transporting ATPases (Fig. 6).
Finally, Arg-695, His-701, and Leu-706 are at least structurally important for the Pma1 ATPase since replacement by Ala led to trypsin sensitivity (misfolding), failure to move to the secretory vesicles, and a dominant lethal phenotype. In the case of His-701, this result corroborates earlier work by Wach et al. (29), who showed that H701R, H701Q, and H701E behave as dominant lethal mutations. Furthermore, although Arg-695, His-701, and Leu-706 are poorly conserved among the P-type ATPases, position 706 is occupied by Ile in the SERCA ATPase, where mutation to Ala has been reported to produce a nonfunctional enzyme (35).
To place these and the rest of the data in a useful context, Fig. 7 provides a helical wheel diagram for M5 of the yeast H ϩ -ATPase. It is perhaps significant that the structurally and functionally important amino acids identified in this study are seen to lie along one face of the putative ␣-helix. Included are three residues required for proper ATPase folding and biogenesis (Arg-695, His-701, and Leu-706), three residues at which Ala substitutions inactivate the ATPase (Tyr-694, Trp-709, and Ser-699), and two residues at which mutations lead to complex changes in ligand binding and/or the equilibrium between E 1 and E 2 conformations (Tyr-691, based on the inability of vanadate to protect against trypsinolysis, and Val-692, based on simultaneous shifts in several kinetic parameters).
Taken together, although the results of this study give no firm evidence for a transport site in M5 of the yeast ATPase, they do indicate that M5 plays a significant role in the reaction cycle. As progress is made toward solving the structure of the closely related Neurospora H ϩ -ATPase (40), the mutational data should provide helpful insight into structure-function relationships.