Expression in High Yield of Pig α1β1 Na,K-ATPase and Inactive Mutants D369N and D807N in Saccharomyces cerevisiae

Studies of structure-function relationships in Na,K-ATPase require high yield expression of inactive mutations in cells without endogenous Na,K-ATPase activity. In this work we developed a host/vector system for expression of fully active pig Na,K-ATPase as well as the inactive mutations D369N and D807N at high levels in Saccharomyces cerevisiae. The α1- and β1-subunit cDNAs were inserted into a single 2-μm-based plasmid with a high and regulatable copy number and strong galactose-inducible promoters allowing for stoichiometric alterations of gene dosage. The protease-deficient host strain was engineered to express high levels of GAL4 transactivating protein, thereby causing a 10-fold increase in expression to 32,500 ± 3,000 [3H]ouabain sites/cell. In one bioreactor run 150-200 g of yeast were produced with 54 ± 5 μg of Na,K-pump protein/g of cells. Through purification in membrane bound form the activity of the recombinant Na,K-ATPase was increased to 42-50 pmol/mg of protein. The Na,K dependence of ATP hydrolysis and the molar activity (4,500-7,000 min) were close to those of native pig kidney Na,K-ATPase. Mutations to the phosphorylation site (D369N) or presumptive cation sites (D807N), both devoid of Na,K-ATPase activity, were expressed in the yeast membrane at the same α-subunit concentration and [3H]ouabain binding capacity as the wild type Na,K-ATPase. The high yield and absence of endogenous activity allowed assay of [3H]ATP binding at equilibrium, demonstrating a remarkable 18-fold increase in affinity for ATP in consequence of reducing the negative charge at the phosphorylation site (D369N).

transmembrane segments, has proven to pose special problems. The Na,K-pump has been expressed in the yeast Saccharomyces cerevisiae (1), COS cells (5), HeLa cells (6), NIH 3T3 cells (7), Xenopus oocytes (8), and in insect cells (9). The properties of isozymes of Na,K-ATPase have been analyzed (10,11) and important structure-function correlates have been established by work on mutants (5,7,12,13). However, it has been difficult to express Na,K-ATPase in large quantities and higher eucaryotic cell lines like HeLa (6) and COS cells (5) and Xenopus oocytes (8,14) express endogenous Na,K-ATPase of almost the same magnitude as the transfected activity. The ouabain selection methodology (5,6) does not allow analysis of the interesting mutants that are blocked in the reaction cycle. The expression level in baculovirus infected insect cells is high, but only a small fraction of the recombinant pumps are enzymatically active, and insect cells also possess endogenous Na,K-ATPase activity (9,11). A major advantage of yeast cells for expression studies is their lack of endogenous Na,K-ATPase activity. The work of Farley and co-workers (1,15,16) demonstrated that yeast cells are capable of expressing fully active Na,K-ATPase at the cell surface, but the expression is limited to levels of 2-4 pmol/mg of protein as determined by [ 3

H]ouabain binding.
In the present work the capacity of S. cerevisiae for the production of enzymatically active Na,K-pumps was characterized with respect to the dependence upon gene copy number, promoter strength, and growth medium composition. The ␣1and ␤1-subunit gene dosage was altered in parallel by insertion of their cDNAs into a single plasmid with a particularly high and regulatable copy number. The promoter activity was increased by the combination of a strong galactose regulated promoter and a host strain modified to express high levels of GAL4-transactivating protein (17). The plasmid was constructed to allow separation of the growth of the host cells from the phase of Na,K-ATPase expression. Computer-controlled bioreactors were used to increase the yield and growth experiments were performed to examine the influence of medium composition and induction time using [ 3 H]ouabain binding to assess the expression levels. Established methods were modified for partial purification of Na,K-ATPase from the yeast plasma membranes, in conditions where the endogenous H-ATPase was removed. The quantity, ligand binding, and enzymatic properties of the recombinant Na,K-ATPase were determined. The versatility of the expression system was examined by characterization of mutations to two side chains of the ␣-subunit, the D369N mutation of the phosphorylated side chain (7,18), and D807N at a presumptive site for cation binding (19). These mutants were devoid of Na,K-ATPase and potassium-dependent para-nitrophenyl phosphatase activities, but they could be expressed in the yeast membranes at the same ␣-subunit concentration and [ 3 H]ouabain binding capacity as the wild type Na,K-ATPase. The high yield of Na,K-ATPase from yeast and the absence of endogenous activity allowed assays not previously achieved for recombinant enzyme, such as [ 3 H]ATP binding at equilibrium.
Transformation of S. cerevisiae and E. coli-Yeast cells were transformed by electroporation according to Becker and Guarante (24). E. coli was transformed according to Nishimura et al. (25).
DNA Techniques-Isolation of plasmid DNA from E. coli was carried out as described by Del Sal et al. (26). Total DNA was isolated from S. cerevisiae as described by Rose et al. (27). Restriction endonucleases and DNA modifying enzymes were purchased from New England Biolabs, Boehringer Mannheim, and Promega and used according to the manufacturers' specifications. Common DNA manipulations were carried out as described by Sambrook et al. (28). Southern blots were developed using a biotinylated, nick-translated probe prepared as described by the manufacturer Tropix.
The Klenow polymerase-treated 941-bp NcoI-␤1-DraI fragment from pNK␤31 was ligated into XhoI-digested, Klenow polymerase-treated RS421. The resulting plasmid carries the NcoI-␤1-DraI fragment between the PMA1 (yeast H-ATPase) promoter and terminator region. The promoter-␤1-terminator region present on a HindIII fragment was cloned into the unique HindIII site in pEMBLyex4 in the correct orientation with respect to the CYC-GAL promoter. A 1,637-bp NruI-HpaI fragment was excised from the plasmid to eliminate one of three KpnI sites. The resulting plasmid was digested with KpnI and religated to remove the PMA1 promoter from the plasmid. The resulting ␤1 expression plasmid contains the pig ␤1 cDNA under control of the CYC-GAL promoter. The final ␣1␤1 expression plasmid, pPAP1466, carries the 5.0-kilobase NruI-CYC-GAL-␤1-HindIII fragment inserted into the NruI-HindIII-digested ␣1 expression plasmid.
Site-directed Mutagenesis-Site-directed mutagenesis was performed according to Ho et al. (35). The nucleotide sequences of the two D369N mutagenic primers were; 5Ј CAT CTG CTC AAA CAA AAC CGG 3Ј and 5Ј CCG GTT TTG TTT GAG CAG ATG 3Ј. The mismatched nucleotide changing Asp 369 to Asn 369 is underlined. A fragment containing the mutation was subsequently cloned into the expression plasmid pPAP1466. The nucleotide sequences of the two D807N mutagenic primers were; 5Ј GGA ACC ATG TTT GTG CCC AAG 3Ј and 5Ј CTT GGG CAC AAA CAT GGT TCC 3Ј. The mismatched nucleotide changing Asp 807 to Asn 807 is underlined. A restriction fragment containing the mutation was subsequently cloned into the expression plasmid pPAP1666. The DNA sequences of the entire PCR fragments were confirmed by dideoxy sequencing.
Induction of Na,K-Pump Synthesis-A single colony of transformed yeast cells was selectively propagated until saturation in 10 ml of glucose minimal medium supplemented with leucine (21). Aliquots of 5 ml were subsequently propagated in 100 ml of minimal medium lacking leucine. For small scale production, 10 ml of this preculture were used for inoculation of 3 liters of selective minimal medium containing 0.5% glucose and 2% lactate as carbon source and supplemented with all amino acids except leucine, tryptophan, and histidine. Galactose was added to a final concentration of 2% at A 450 ϭ 1.0. Routinely, cells were harvested 48 h after addition of galactose.
For large scale production, 1000 ml of preculture was used to inoculate 10 liters of minimal medium in an Applicon® fermentor equipped with an ADI 1030 Bio Controller. The culture was agitated at 158 rpm, and air was supplied through a 0.2-m filter. The carbon source was 5% glucose and 2% lactate, and the medium was supplemented with all amino acids except leucine, tryptophan, and histidine. During growth at 30°C in glucose, the pH of the medium was kept at 6.0 by computercontrolled addition of 1 M NaOH. The shift from growth on glucose to growth on lactate was monitored as an increase in pH of the growth medium and a decrease in growth rate. At this point galactose was added to a final concentration of 2%. Cells were harvested after 48 h.
Isolation of Yeast Membranes-Galactose-induced yeast cells were harvested at 5,800 g for 30 min and washed once in ice-cold water. Yeast cells were resuspended in ice-cold lysis buffer (25 mM imidazole, 1 mM EDTA, 1 mM EGTA, 10% (w/v) sucrose, pH 7.5) containing 1 mM PMSF, 1 g/ml chymostatin, 1 g/ml pepstatin, and 1 g/ml leupeptin. Cells were homogenized in an ice-cold Bead Beater (Biospec) for three times 1 min. The homogenate was centrifuged at 1000 ϫ g for 10 min at 4°C. The supernatant was further centrifuged at 3,000 ϫ g for 20 min at 4°C. The final supernatant was centrifuged at 100,000 ϫ g for 1.5 h at 4°C. The resulting crude membrane fraction was homogenized in a Teflon-glass Braun homogenizer in ice-cold lysis buffer containing the previously mentioned concentrations of protease inhibitors and kept at Ϫ80°C until use.
Gradient Membranes-Crude membranes were fractionated on step gradients consisting of 15 ml of 40% plus 15 ml of 15% (w/v) sucrose in lysis buffer with proteolysis inhibitors. A sample of 6 -8 ml containing 40 mg of protein was layered on top, and the gradient was centrifuged for 2 h at 50,000 rpm in the Ti-60 Beckman rotor. The membrane band at the 15/40% interface containing 20 -30% of sample protein in 6 ml was collected in a syringe, diluted 5-fold in lysis buffer, and collected by centrifugation at 2 h at 50,000 rpm.
Electrophoretic Analysis and Immunological Techniques-Protein samples were mixed with 2-fold concentrated loading buffer containing 1 mM PMSF and 1 g/ml chymostatin and incubated for 10 min at room temperature, prior to separation in 8% polyacrylamide gels (36). Western blotting was performed according to Kyhse-Andersen (37) using a polyclonal anti-␣-antibody (anti-KETYY) kindly provided by Jack Kyte, University of California. For quantitative determinations of Na,K-ATPase proteins, Western blots were digitalized on an HP-Scanjet IIP scanner, and densitometry was performed using the Cream software from KemEnTech, Copenhagen, Denmark.
Selective SDS Treatment of Crude Yeast Membranes and Purification of Na,K-ATPase-Yeast membranes were treated with SDS as before (38,39). The final concentration of the solution was 0.64 mg/ml SDS, 2 mg/ml protein, 3.0 mM Na 2 ATP, 25 mM Imidazole/HCl, 1 mM Na 2 EDTA, pH 7.5, and 1 mM PMSF, 1 g/ml chymostatin, 1 g/ml pepstatin, and 1 g/ml leupeptin to prevent proteolysis. The SDS-treated membranes were loaded on a step gradient containing two layers of 10 and 37% (w/v) sucrose in the 35-ml tubes of the Beckman Ti-60 angle rotor and separated by centrifugation for 2 h at 40,000 rpm. A band of 6 ml was collected from the interface of 10 and 37% sucrose through a needle mounted on a syringe, and Na,K-ATPase was collected as a pellet after 3-5 fold dilution and centrifugation at 40,000 rpm overnight.
For ATP binding experiments, gradient membranes were incubated at 2 mg of protein/ml in lysis buffer with 0.3 mg/ml SDS and proteolysis inhibitors for 30 min at 20°C. The mixture was centrifuged for 30 min at 70.000 rpm in the Beckman 100 A ultracentrifuge and resuspended in lysis buffer.
Equilibrium [  Equilibrium ATP Binding-Aliquots of 100 -200 g of SDS-treated gradient membranes were incubated at 0 -4°C in 1 ml containing either 10 mM NaCl or 10 mM KCl, 10 mM MOPS-Tris, pH 7.2, 10 mM EDTA-Tris, and [ 3 H]ATP (Amersham, specific activity: 36 Ci/mmol) and ATP-Tris to final concentrations of 0.2-300 nM. Bound and unbound [ 3 H]ATP were separated by centrifugation at 100,000 ϫ g for 30 min at 4°C. The supernatant was discharged and remaining buffer removed with a paper towel. Bound [ 3 H]ATP was determined by scintillation counting, and specific binding was calculated as binding in presence of NaCl minus binding in presence of KCl.
ATPase Assay-For assay of Na,K-ATPase activity 10-l aliquots containing 50 -100 g of protein were transferred to tubes containing 0.5 ml of 130 mM NaCl, 20 mM KCl, 3 mM MgCl 2 , 3 mM ATP, with or without 1 mM ouabain, 10 mM MOPS-Tris, pH 7.5, at 37°C. After incubation for 1, 3, 5, or 10 min at 37°C, the reaction was stopped with 1 ml of ice-cold 2% H 2 SO 4 , 0.1% ascorbic acid, 0.5% ammonium heptamolybdate, and 0.5% SDS. The tubes were left at room temperature for 15 min, and absorbance was measured at 690 nm as before (39). The Na,K-ATPase activity was calculated from the difference in absorbance for tubes with and without 1 mM ouabain.
For assay of vanadate sensitive H-ATPase activity (33), 10-l portions were transferred to test tubes containing 1 ml of 50 mM MES adjusted to pH 6.5 with Tris, 5 mM MgSO 4 , 50 mM KNO 3 (to inhibit vacuolar ATPase), 5 mM sodium azide (to inhibit mitochondrial ATPase) 0.2 mM ammonium molybdate (to inhibit acid phosphatase), and 2 mM Na 2 ATP without and with 1 mM NaVO 3 . After 10 min at 37°C, the reaction was stopped, and inorganic phosphate was measured as described above for Na,K-ATPase.
Molecular sieve high performance liquid chromatography was performed with a TSK 3000 SW (7.5 ϫ 300 mm) Toyo Soda gel filtration column with a TSK SW guard column (7.5 ϫ 75 mm) operated at flow rates of 0.2 ml/min using a Pharmacia Biotech Inc. solvent delivery system as before (40). Prior to solubilization, the membrane-bound Na,K-ATPase was sedimented at 100,000 rpm for 10 min in the Beckman Airfuge and resuspended in 300 mM potassium acetate, pH 6.0, 2 mM dithiothreitol, 2 mM EDTA-Tris. This suspension was mixed with equal volumes of C 12 E 8 , 10 mg/ml. The insoluble residue was removed by centrifugation for 10 min at 100,000 rpm in a Beckman Airfuge (40).

Development of a Host/Vector
System for Na,K-ATPase Expression in S. cerevisiae-In the 2-m-based plasmids constructed for expression of the pig Na,K-ATPase, pPAP1466 and pPAP1666, the cDNAs encoding the ␣1and ␤1-subunits are transcribed from identical CYC-GAL promoters to avoid nonstoichiometric transcription of the two subunit genes (Fig. 1). The CYC-GAL promotor is a strong galactose-inducible hybrid promoter containing the DNA sequences required for constitutive cytochrome c 1 (CYC1) expression and the upstream activating sequence from the GAL1-GAL10 intergenic promoter region (32). The 2-m origin of replication assures that the expression plasmid is stably maintained in the yeast cell. The expression plasmid carries two selective markers, URA3 and the poorly expressed allele of the LEU2 gene, leu2-d. A plasmid with two selective markers was favored for two reasons. First, manipulation of the plasmid copy number by altering the composition of the growth medium was facilitated, and second, homologous recombination between the identical CYC-GAL promoter regions transcribing the ␣1and ␤1-subunits, with loss of one of the subunit genes, was avoided. Selection for uracil autotrophy produces transformants with a plasmid copy number of around 20, while selection for both leucine and uracil autotrophy selects for growth of yeast cells with an extraordinary high copy number of around 200 (41).
The two expression plasmids differ in their E. coli origin of replication in that pPAP1466 is a pUC derivative (31) with a very high copy number, while pPAP1666 is a P15A-derived plasmid (34) with a 100-fold reduced copy number in E. coli compared with pPAP1466. The use of pPAP1666 with a low copy number in E. coli was necessary, as we were unable to clone various ␣1-subunit mutations into pPAP1466.
The yeast strain BJ5457 (20), lacking the PEP4 and PRB1 protease activities, was chosen as the basic host strain for the production of recombinant Na,K-ATPase in order to reduce the possibility of proteolytic degradation of the enzyme during synthesis and purification. The transactivating GAL4 protein is known to be limiting for expression controlled by galactoseregulated promoters (17). The GAL4 protein level can be increased by integrating a GAL10-GAL4 transcriptional fusion into the yeast chromosome (17). Therefore, we constructed the yeast-integrating plasmid pPAP1488 (Fig. 2), which was targeted to the trp1 locus of BJ5457 by homologous recombination generating strain PAP1500. Addition of galactose to PAP1500 transformed with pPAP1466 or pPAP1666 should initiate a cascade reaction leading to expression of GAL4 protein and Na,K-ATPase ␣1and ␤1-subunits. Growth Physiology and Kinetics of Na,K-ATPase Expression in Yeast-The physiological effect of expression of the Na,K-ATPase on growth of yeast cells was determined in a series of growth experiments with PAP1500(pPAP1466). Fig. 3 illustrates how the yeast growth phase was separated from the Na,K-ATPase expression phase. In the growth phase, the level of Na,K-ATPase expression is low due to the absence of galactose. Later in the growth phase, the cells were grown with lactate as carbon source to avoid repression by glucose of the CYC-GAL promoters during the galactose-induced phase of Na,K-ATPase synthesis. In Fig. 3 the cells were grown under inducing and noninducing conditions in media selecting for a high plasmid copy number. Growth of the transformed yeast cells ceased immediately after induction of Na,K-pump biosynthesis with galactose whether the cells were expressing wild type or the D369N and D369A mutations.
Integration of a GAL10-GAL4 transcriptional fusion in the yeast chromosome had a dramatic effect on the level of expression of Na,K-ATPase, in media selecting for the high plasmid copy number. Fig. 4 shows that induction of Na,K-ATPase was much faster and that the maximal number of [ 3 H]ouabain binding sites was about 10-fold higher in cells with the chromosomal fusion than in the yeast strain, with only a few copies of GAL4 protein/cell.
A number of experiments were performed to determine the influence of composition of the growth medium and induction time on the time course of appearance of high affinity [ 3 H]ouabain sites measured in crude yeast membranes. Fig. 5 shows that the highest density of [ 3 H]ouabain sites is seen in yeast cells, with a high copy number of the expression plasmid, and growing in synthetic minimal medium supplemented with all amino acids except leucine, tryptophan, and histidine. In medium without the supplement of amino acids, accumulation of Na,K-ATPase ceased after 48 h, and then the level decreased, suggesting that low levels of amino acids may be rate-limiting for synthesis of Na,K-ATPase.
Purification and Characterization of Na,K-ATPase from Crude Yeast Cell Membranes-Under optimum growth conditions in the computer-controlled bioreactor, the yield of yeast was 150 -200 g/10 liters of culture, and the concentration of [ 3 H]ouabain sites was 323 Ϯ 31 pmol/g cells (mean Ϯ S.E., n ϭ 5 preparations). This value corresponds to 32,500 Ϯ 3,000 sites/cell. The density of sites was the same in the intact trans-formed yeast cells as in spheroplasts after removal of the cell wall by lyticase (data not shown).
In crude yeast membranes, the binding capacity for [ 3 H]ouabain was 10 -15 pmol/mg of protein. For characterization of the recombinant enzyme, it was essential to reduce the endogenous H-ATPase activity and to purify the enzyme by removal of extraneous protein. Fig. 6 shows the SDS curve required for application of the purification scheme developed previously for purification of the Na,K-ATPase from kidney (38,39). In contrast to the 3-5-fold activation of Na,K-ATPase in kidney membranes after incubation with SDS, due to demasking of closed right-side-out vesicles, the activity of Na,K-ATPase or H-ATPase in yeast cell membranes increased only about 20%. At the membrane protein concentration of 2 mg/ml, an optimum concentration range for SDS was found (0.5-0.7 mg/ml) where most of the H-ATPase of the yeast cell membranes was inactivated, while [ 3 H]ouabain binding and Na,K-ATPase activity of the recombinant enzyme were preserved. The peak of [ 3 H]ouabain binding in crude yeast cell membranes was found at equilibrium densities in the range 1.15-1.2 g/ml  3. Separation of the yeast growth phase and the Na,Kpump production phase: without galactose induction (E and with addition of galactose 2% (q. Growth of PAP1500(pPAP1466) was in minimal medium at 30°C with 0.5% glucose or 2% lactate as carbon source. Yeast cells were growing exponentially with glucose as carbon source between A 450 ϭ 0.04 and A 450 ϭ 0.5 and exponentially with lactate as carbon source between A 450 ϭ 0.5 and A 450 ϭ 1.0. Induction of Na,K-pump biosynthesis with 2% galactose at A 450 ϭ 1.0 and higher is seen to arrest cell growth. Abbreviations: GLU, growth on glucose; LAC, growth on lactate; GAL, addition of galactose to a final concentration of 2%.

FIG. 4. The accumulation of [ 3 H]ouabain sites in the membranes of yeast strains without ( ) or with (q) a transcriptional GAL10-GAL4 fusion.
At the indicated times 80-ml aliquots of the growth medium were removed and crude cell membranes were prepared. Aliquots containing 200 g of protein were assayed for binding at 10 nM [ 3 H]ouabain as described under "Experimental Procedures." (42). However, after incubation with 0.64 mg/ml SDS, it was difficult to recover the recombinant Na,K-ATPase after centrifugation in discontinuous sucrose gradients, because the activity was distributed in a broad band at relatively low densities around 1.10 g/ml. The [ 3 H]ouabain binding data in Fig. 7 show a 3-5-fold enrichment of binding capacity from 10 -15 pmol/mg [ 3 H]ouabain binding in crude yeast membranes to 42-50 pmol/mg of protein in the partially purified preparation. Note that the dissociation constant was reduced from 21 Ϯ 2 nM to 11 Ϯ 2 nM upon incubation with SDS and fractionation, presumably because the detergent promotes the access of the ligands, Mg 2ϩ and vanadate, required for facilitation of [ 3 H]ouabain binding.
A higher yield, but a lower binding capacity (15-25 pmol/mg of protein, cf. Fig. 12), were obtained when the crude membranes were fractionated on sucrose gradients prior to SDS incubation. The gradient membranes at the interphase between 15 and 40% (w/v) sucrose were collected by centrifugation and subsequently incubated at 2 mg of protein/ml with SDS in a lower concentration, 0.5 mg/ml SDS in the presence of 3 mM ATP, or at 0.3 mg/ml SDS in the absence of ATP. Fig. 8 shows a comparison of Western blots of the recombinant Na,K-ATPase from this preparation and renal Na,K-ATPase. From this comparison, the purity in terms of ␣-subunit protein was 0.71% (7.1 g of ␣-subunit protein/mg of SDS-extracted yeast membrane protein). The purity in terms of [ 3 H]ouabain sites was 0.65%, corresponding to 44 pmol⅐mg Ϫ1 protein/6.8 nmol⅐mg Ϫ1 protein ϫ 100. The similarity of the two values suggests that the unit binding one molecule of [ 3 H]ouabain contains one recombinant ␣-subunit. The Na-and K-ion dependence of ATP splitting in Fig. 9 is identical to that observed before for preparations from pig kidney. The molecular activity of Na,K-ATPase in the preparation was 4,500 -7,000 P i ⅐min Ϫ1 per [ 3 H]ouabain site. This value is in the lower range of the molecular activity of the native pig kidney Na,K-ATPase (7,000 -8,000 P i ⅐min Ϫ1 ).
Chromatography of soluble Na,K-ATPase in C 12 E 8 on TSK 3000 SW size exclusion columns was shown to be a sensitive procedure for detecting the aggregation of ␣␤-units accompanying thermal denaturation of Na,K-ATPase (40). In the experiment in Fig. 10, it is seen that the [ 3 H]ouabain complex of soluble, wild type, recombinant Na,K-ATPase is stable during chromatography in C 12 E 8 at 2-4°C and that it elutes quantitatively at a volume (9.0 ml) corresponding to the elution volume (8.5 ml) of the [ 3 H]ouabain complex with the ␣␤-unit of kidney Na,K-ATPase. The elution volume of the D369N mutant ␣␤-unit was in the same range (9.0 ml). This provides evidence for organization of the recombinant, wild type, and mutant D369N, Na,K-ATPase in ␣␤-units with hydrodynamic properties similar to those of native renal Na,K-ATPase. There was no tendency for aggregation of the soluble protein to oligomers similar to the (␣-␣) aggregates demonstrated in solubilized preparations of the recombinant ␣-subunit from SF9 cells infected with baculovirus (43).
Expression of Recombinant Na,K-ATPases without Turnover-We constructed two mutants in the ␣1 pig subunit by site-directed mutagenesis to determine the versatility of the expression system. The D369N mutation exchanges the aspartic acid phosphorylated during the reaction cycle for an asparagine, while the D807N mutation exchanges an aspartic acid potentially involved in cation binding for an asparagine. As expected both D369N (7, 18) and D807N (19) were devoid of FIG. 5. Effect of composition of growth medium on accumulation of ouabain sites. Crude yeast membranes were isolated from strain PAP1500(pPAP1466) after induction with 2% galactose at time 0. Procedures for membrane preparation and [ 3 H]ouabain binding were as in Fig. 4. Cells were grown with 0.5% glucose and 2% lactate as carbon source prior to induction, as illustrated in Fig. 3. Abbreviations used: Ϫ leu Ϫ aa, growth in the absence of leucine and the presence of lysine; ϩ leu Ϫ aa, growth in the presence of leucine and lysine; Ϫleu ϩ aa, growth medium supplemented with all amino acids except leucine, tryptophan, and histidine. Na,K-ATPase and K-p-nitrophenyl phosphatase activity, and D369N was not phosphorylated from ATP (data not shown). Quantitative Western blotting (Fig. 8) showed that both mutations are expressed with the same concentration of ␣-subunit protein in the membrane as the wild type enzyme. Comparison with the purity estimated from [ 3 H]ouabain binding also shows that the stoichiometry is one ouabain site/␣-subunit for the two mutants. Data in Fig. 11 show that D369N and D807N had the same maximum binding capacity for [ 3 H]ouabain as the wild type enzyme. The apparent affinity for ouabain of D807N (K D ϭ 13 Ϯ 2 nM) was higher than for wild type Na,K-ATPase (K D ϭ 21 Ϯ 2 nM), while that of D369N (K D 273 Ϯ 13 nM) was much reduced. When the binding assay was conducted in the absence of vanadate, the mutant D369N and the wild type had dissociation constants in the same range (not shown) as an indication that D369N did not interact with vanadate. This agrees with the previous observation that D369N does not react with inorganic phosphate (7). mM KCl for D369N, reaching a steady state level at 3 mM KCl (data not shown). The explanation for this is that KCl stabilizes the E 2 K form of Na,K-ATPase with affinities for ATP (K ap 400 M) (2, 44) that are much higher than the range of [ 3 H]ATP concentrations (0.2-300 nM) that are used in the binding assay. Specific binding of ATP to the recombinant enzymes can thus be determined as the binding at 10 mM NaCl minus that in medium containing 10 mM KCl as in Fig. 12. This data, in combination with the titration with cations, confirm that the mutant D369N has retained the ability to form the alternative conformations E 1 Na in NaCl or E 2 K in KCl medium where specific [ 3 H]ATP binding is abolished.

Effects of Mutations to Asp
The data in Fig. 12 show that linear Scatchard plots were obtained for both wild type and the mutant D369N and that the mutation D369N led to a remarkable change in dissociation constant of the protein-ATP complex. The capacities for binding were in the same range as the capacities for [ 3 H]ouabain binding and the concentrations of ␣-subunit protein (cf. Fig. 8), suggesting that there was one high affinity binding site for ATP or ouabain per ␣␤-unit (cf. Ref. 39). However, the dissociation constant for wild type (109 Ϯ 11 nM) was 18-fold higher than for the D369N mutant (5.9 Ϯ 0.4 nM) (Fig. 12) and 32-fold higher than for the D369A mutant (3.4 Ϯ 0.3 nM, data not shown).

DISCUSSION
The concept behind the host/vector system for Na,K-ATPase expression in this study was to separate the phase of yeast cell growth with low basal expression from the expression phase, where the expression level of Na,K-ATPase in the yeast cell membranes increases to a maximum. In the latter phase, the expression system allowed for a high gene copy number and a high transcriptional activity due to the strong inducible promoter. Precise alteration of the ␣:␤ gene dose required that the ␣ and ␤ cDNAs be present on the same plasmid and be ex-pressed under control of identical promoters. A regulated increase in plasmid copy number in the Na,K-ATPase production phase was possible, due to the presence of a URA3 gene and the poorly expressed leu2-d gene on the expression plasmid. Selection for leucine autotrophy increased the plasmid copy number and simultaneously elevated the pump concentration 2-3-fold (Fig. 5). This change is smaller than the 8-fold increase in expression of schistosomal antigen P28-I observed upon changing from selecting for uracil to leucine autotrophy (45). However, these data are not directly comparable as the schistosomal antigen is a soluble protein, while Na,K-ATPase is inserted into the membrane. Transcriptional activity was increased by engineering the yeast host strain to express elevated concentrations of the GAL4-transactivating protein, causing a 10-fold increase in the expression level (Fig. 4). This is consistent with the finding of Schultz et al. (46) who reported a 10-fold increase in the expression level of the membrane bound Epstein-Barr virus gp350 protein after integration of a GAL10-GAL4 fusion into the genome of the yeast host strain. The high transcriptional activity may cause subsequent steps in the synthesis process to become rate-limiting and account for the 2-fold increase in expression level upon supplementing the yeast growth medium with amino acids (Fig. 5).
The combined effect of engineering these parameters was to increase the density of Na,K-pumps to 32,500 Ϯ 3,000 sites/cell or 54 Ϯ 5 g of Na,K-pump protein/g of yeast cell. In the crude membrane fraction from the yeast (10 -15 pmol/mg of protein) and in the SDS-extracted membranes (42-50 pmol/mg of protein), the pump density was higher than achieved previously in eucaryotic cells, cf. Table I. The recombinant enzyme is fully active, with one site for binding of ATP and ouabain per ␣-subunit and a range of molecular activities close to those of the native Na,K-ATPase of pig kidney. The relatively high activity and the lack of endogenous background allowed for analyses not previously achieved for recombinant Na,K-ATPase, such as ATP binding at equilibrium.
An important reason for separating the growth phase from the phase of Na,K-ATPase biosynthesis is the apparent toxicity of Na,K-ATPase protein synthesis. The evidence for this is the immediate arrest of cell growth and division following the galactose induction of Na,K-ATPase biosynthesis. This could be due to the activity of the Na,K-pump, but arrest of cell growth was independent of the enzymatic activity of the expressed pump as yeast strains expressing wild type or inactive mutations behaved identically. Also, the expression levels obtained in this work do not, in general, prohibit yeast cell growth. Much higher levels of constitutive heterologous expression has been described for soluble proteins (47). Also the endogenous GAL1 protein accumulates to 0.8% of total cell protein after galactose induction without affecting cell viability (48). It is therefore reasonable to propose that the toxicity of Na,K-ATPase protein synthesis is due to perturbation of the yeast membranes following the insertion of protein with hydrophobic membraneembedded sequences. Previous experiments with membrane proteins, OmpA (49) and ␣-hemagglutinin (50), demonstrated selection against the production of hydrophobic intramembrane segments in response to the poor tolerance by the host cell. Selection for low expressing variants has been observed in several cases, where constitutive promoters were used to drive expression of membrane proteins (51,52). In the present expression system, the selection pressure was avoided through the use of a strong inducible promoter, but the toxicity of the expression suggests that the capacity of the yeast membrane system may set the limit for the concentration of Na,K-ATPase that can be achieved in these cells.
The highest expression level of an ATPase in yeast has been described for isoforms of the Arabidopsis thaliana H-ATPase using a multicopy plasmid and the strong constitutive yeast PMA-1 (plasma membrane H-ATPase) promoter (33). This expression system directs the plant proton pump to the endoplasmic reticulum membranes, where it constitutes 45% of the endoplasmic reticulum proteins, but the plant proton pump may also complement the proton pump in the yeast plasma membrane (53), like the Neurospora H-ATPase (54). In our experience, insertion of ␣␤-unit cDNA in this plasmid did not lead to expression of significant amounts of Na,K-ATPase. 2 The constitutive PMA-1 host/vector system (33) also proved inefficient for expression of the SR-Ca-ATPase (55). The Na,K-ATPase expression system in yeast described by Horowitz et al.
(1) also used constitutive promoters and achieved only low expression levels, 0.05% of the plasma membrane protein, cf. Table I. It seems that the high expression level obtained by constitutive expression of the plant H-ATPase is the exception, while a high expression level for Na,K-ATPase and SR-Ca-ATPase is incompatible with constitutive promoters. A probable explanation could be that insertion of H-ATPase into yeast membranes is less prone to eliciting selection for low expressing variants than is expression of the Na,K-or Ca-ATPases. In agreement with these notions, the SR-Ca-ATPase has recently been expressed in a relatively high concentration, 0.3% in the yeast membranes, under control of a galactose-regulated promoter (55).
For the purification of the recombinant Na,K-ATPase, it was important to establish conditions where the plasma membrane H-ATPase could be removed. This enzyme constitutes background activity in the enzyme assay and [ 3 H]ATP binding experiments. Fortunately, the H-ATPase turned out to have a relatively high sensitivity to denaturation by incubation with SDS. Our data show that Na and K dependence of ATP hydrolysis, and ATP binding, of the recombinant enzyme are similar to the kinetics of the native renal Na,K-ATPase. In the best of our preparations, the molar activity was close to 7,000 P i /min, but the number varied in the range 4,500 -7,000 P i /min for reasons that are not yet fully understood. A variation in molecular activity is also apparent among the previous preparations from yeast, in the range from 792 to 8212 P i /min (1,15,56,57), cf. Table I. The relatively large scatter of the reported molecular activity among preparations from different host systems may reflect difficulties in determining site concentrations in preparations with relatively low activity. The highest expression in terms of ␣-subunit protein is achieved in baculovirus-infected Sf-9 cells (9), but only about 3% of the protein is active in terms of Na,K-ATPase activity and ligand binding, Table I. The insect cells (9,43), as well as the COS-1 (5) and HeLa (10,20,58) cells, also express endogenous Na,K-ATPase activity as denoted in Table I. The yeast expression system clearly does not distinguish active Na,K-ATPase from inactive mutant ␣␤-units with respect to biosynthesis and translocation to the cell membranes. The D369N and D807N mutations were expressed in yeast membranes at concentrations of ␣-subunit and [ 3 H]ouabain binding sites that are comparable with those of the wild type enzyme. For the D369N mutant it was further demonstrated that the concentration of protein expressed in the yeast membranes was equal to the concentration of [ 3 H]ATP sites. The ␣␤-units of the D369N mutant and recombinant Na,K-ATPase have the same hydrodynamic properties as purified Na,K-ATPase from kidney.
Mutation at the phosphorylation site of the negatively charged aspartic acid 369 for an asparagine residue caused a remarkable decrease in dissociation constant of the protein-ATP complex. This is most likely due to a true change in the binding constant of the E 1 conformation for ATP and not to a shift in conformational equilibrium. Our ATP binding data for D369N demonstrate that Na ϩ and K ϩ stabilize the E 1 Na and E 2 K conformations with widely different affinities to ATP. In addition, the D369N mutant has the same affinity for [ 3 H]ouabain as the wild type in the presence of Mg 2ϩ . Previous data also show that the [ 3 H]ouabain complex of the mutant responds to Na ϩ and K ϩ in the same manner as the wild type Na,K-ATPase (7). In the Ca-ATPase from SR, modification of either Asp 351 or the neighboring Lys 352 abolish phosphorylation (59). Our data show that reduction of the negative charge of Asp 369 greatly increases the affinity for ATP. Although the negatively charged Asp residue is essential for enzymatic turnover, it may thus reduce the affinity for ATP through electrostatic repulsion of the negatively charged ␥-phosphate present on the ATP molecule.
The extraordinary high affinities of D369N or D369A for ATP turn these mutants into powerful tools for future identification of residues involved in ATP binding to Na,K-ATPase and for studying the influence of Mg 2ϩ upon ATP binding. Given the high degree of homology of the amino acid sequence around the phosphorylation site (ICS D KTGTLT), it can be expected that the large increase in affinity for ATP upon reducing the negative charge on the phosphorylated side chain will be a general phenomenon for the P-type cation pumps, H,K-ATPase, Ca-ATPase, and H-ATPase.