Expression of Na+,K+-ATPase in Pichia pastoris: analysis of wild type and D369N mutant proteins by Fe2+-catalyzed oxidative cleavage and molecular modeling.

Na+,K+-ATPase (pig alpha1,beta1) has been expressed in the methylotrophic yeast Pichia pastoris. A protease-deficient strain was used, recombinant clones were screened for multicopy genomic integrants, and protein expression, and time and temperature of methanol induction were optimized. A 3-liter culture provides 300-500 mg of membrane protein with ouabain binding capacity of 30-50 pmol mg-1. Turnover numbers of recombinant and renal Na+,K+-ATPase are similar, as are specific chymotryptic cleavages. Wild type (WT) and a D369N mutant have been analyzed by Fe2+- and ATP-Fe2+-catalyzed oxidative cleavage, described for renal Na+,K+-ATPase. Cleavage of the D369N mutant provides strong evidence for two Fe2+ sites: site 1 composed of residues in P and A cytoplasmic domains, and site 2 near trans-membrane segments M3/M1. The D369N mutation suppresses cleavages at site 1, which appears to be a normal Mg2+ site in E2 conformations. The results suggest a possible role of the charge of Asp369 on the E1 <--> E2 conformational equilibrium. 5'-Adenylyl-beta,gamma-imidodi-phosphate(AMP-PNP)-Fe2+-catalyzed cleavage of the D369N mutant produces fragments in P (712VNDS) and N (near 440VAGDA) domains, described for WT, but only at high AMP-PNP-Fe2+ concentrations, and a new fragment in the P domain (near 367CSDKTGT) resulting from cleavage. Thus, the mutation distorts the active site. A molecular dynamic simulation of ATP-Mg2+ binding to WT and D351N structures of Ca2+-ATPase (analogous to Asp369 of Na+,K+-ATPase) supplies possible explanations for the new cleavage and for a high ATP affinity, which was observed previously for the mutant. The Asn351 structure with bound ATP-Mg2+ may resemble the transition state of the WT poised for phosphorylation.

The Na ϩ ,K ϩ -ATPase utilizes the free energy of hydrolysis of ATP to actively transport three intracellular Na ϩ ions and two extracellular K ϩ ions in opposite directions across animal cell membranes. The Na ϩ ,K ϩ -ATPase is a member of the P-type family of cation pumps. The kinetic mechanism of Na ϩ ,K ϩ -ATPase, as of other P-type pumps, involves a phosphoenzyme intermediate and is now largely understood (1,2). As pointed out by Jencks (3), strict cation and substrate specificities of the phosphorylation and dephosphorylation reactions, and tight coupling of the E 1 7 E 2 conformational changes to cation movements are the essential features of all P-type ion pump mechanisms. These central questions of the energy transduction mechanism of P-type pumps can now be posed in structural terms (4) because of availability of molecular structures of the sarcoplasmic reticulum Ca 2ϩ -ATPase for both E 1 2Ca 2ϩ and E 2 conformations (5,6).
The Ca 2ϩ -ATPase molecule consists of head, stalk, and membrane sectors (5). There are 10 transmembrane segments in the membrane domain with two Ca 2ϩ ions ligated approximately in the center of the bilayer and between transmembrane segments M4, M5, M6, and M8 in the E 1 2Ca 2ϩ conformation. The stalk sector consists of the cytoplasmic extensions of the transmembrane helices, particularly S5 and S4. The cytoplasmic sector consists of three domains, nucleotide binding (N), 1 phosphorylating (P), and anchor or actuator domain (A). Comparison of the crystal structure an E 1 2Ca 2ϩ and E 2 conformations shows that in the E 1 conformations the N, P, and A domains are separate, whereas in E 2 conformations the domains are gathered together, moving essentially as rigid bodies (5,6). Movement of the A domain toward P and N domains in the E 1 3 E 2 transition is associated with a bending of S5 that entails complex movements of several transmembrane segments. This changes the ligation of the occluded Ca 2ϩ ions within the transmembrane segments allowing them to dissociate within the sarcoplasmic reticulum. Whereas this general paradigm clearly applies to the other P-type pumps, not all features are explained by the crystal structures. As one example, phosphorylation by ATP requires close proximity of the nucleotide binding N and phosphorylation P domains, but this is not observed in the E 1 2Ca 2ϩ (Protein Data Bank code 1EUL) structure. In addition, of course, there are the specific features of other ion pumps particularly the cation selectivities, ␤ subunits in the case of Na ϩ ,K ϩ -ATPase or H ϩ ,K ϩ -ATPase (7), or FXYD proteins, which are accessory subunits of Na ϩ ,K ϩ -ATPase (8).
For analysis of conformational transitions and ligand binding on renal Na ϩ ,K ϩ -ATPase or gastric H ϩ ,K ϩ -ATPase specific oxidative cleavage catalyzed by bound Fe 2ϩ or the ATP-Fe 2ϩ complex has turned out to be very informative (see Refs. 9 -13 and a recent review in Ref. 14). Recently this technique has also been applied to Ca 2ϩ -ATPase (15) and H ϩ -ATPase (16). The cleavages provide information on the spatial organization surrounding the bound Fe 2ϩ , or ATP-Fe 2ϩ (14). One major prediction from the cleavages catalyzed by bound Fe 2ϩ ions is that E 1 7 E 2 conformational changes are associated with large movements in the cytoplasm domains, N, P, and A domains interacting in E 2 states but separate in the E 1 state (14). This is fully consistent with the difference observed between the crystal structure of Ca 2ϩ -ATPase in the E 1 2Ca 2ϩ conformation (Protein Data Bank code ID 1EUL) or an E 2 conformation in the presence of vandadate/Mg 2ϩ (Protein Data Bank code 1KJU) (Refs. 5 and 17, see also Ref. 14) for a discussion of differences between the 1KJU structure and the new E 2 structure 1IWO). The Fe 2ϩ in ATP-Fe 2ϩ substitutes for Mg 2ϩ ions in catalyzing phosphorylation and ATPase activity (18,19). Thus cleavages mediated by ATP-Fe 2ϩ provide information on ligation of Fe 2ϩ or Mg 2ϩ ions as well as N, P, and A domain proximity. The high selectivity of the ATP-Fe 2ϩ -mediated cleavages, as well as recent observations of similar cleavages mediated by a fluoresecein-DTPA-Fe 2ϩ complex (13), suggest that cleavages occur only next to residues that bind the Fe 2ϩ directly. Thus, overall, the cleavages have led to a picture of N, P, and A domain interactions and ATP-Mg 2ϩ or Mg 2ϩ ions binding in the different conformational states of the catalytic cycle (summarized in Refs. 4, 13, and 14). In another recent development the masses of cleavage fragments have been measured by matrix-assisted laser desorption ionization timeof-flight mass spectrometry (14,20). With the exception of two fragments with exactly known N termini ( 214 ESE and 712 VNDS), other fragments could not be sequenced, and the N termini were designated previously as near XXXX by comparison with known proteolytic fragments. The analysis by mass spectrometry has now greatly improved the assignments, so that the designation of all N termini of fragments as near XXXX is accurate to within 2-4 residues. Whereas the Ca 2ϩ -ATPase structures and modeling based on the structures are invaluable for testing the validity of interpretations, cleavage experiments also lead to inferences on ATP and Mg 2ϩ binding and N to P domain interactions not available from the crystal structures. Thus, it is highly desirable to investigate the predicted roles of specific residues by mutational analysis. As one relevant example, mutations of residues in the conserved 708 TGDGVNDS sequence in the P domain have already shown that Asp 710 is a Mg 2ϩ binding residue (21).
We describe here expression of Na ϩ ,K ϩ -ATPase in the methylotrophic yeast Pichia pastoris and analysis of the wild type protein and the D369N mutant by Fe 2ϩ -catalyzed oxidative cleavages. The ␣1 and ␤1 subunit of cDNA, under control of the AOX1 promoter, are incorporated into the alcohol oxidase chromosomal locus (AOX1), and protein synthesis is induced with methanol (see Ref. 22 for a general review). P. pastoris can grow to high cell densities and produce high levels of secreted soluble proteins. As a precedent for expression of an integral membrane protein in P. pastoris, the P-glycoprotein (MDR3) has been expressed in large quantities and purified for biochemical and structural work (23,24). Thus, expression of functional Na ϩ ,K ϩ -ATPase in P. pastoris at significant levels appeared to be an attractive possibility. Na ϩ ,K ϩ -ATPase has been expressed in many cell types, and used extensively for structure-function analysis, but of most relevance to the present work wild type and mutant Na ϩ ,K ϩ -ATPase proteins have been expressed in Saccharomyces cerevisae and interactions of the pump ligands, ATP, Na ϩ , K ϩ , Mg 2ϩ , and ouabain characterized in detail, particularly by direct binding assays (21,(25)(26)(27)(28)(29). On the other hand, analysis of structural organization and conformational changes by proteolytic cleavage or metalcatalyzed oxidative cleavage has not been described.
As a first step in analysis of mutants by oxidative cleavage we have chosen to examine the active site mutant, D369N. Previously, the D369N and D369A mutants of pig ␣1␤1 Na ϩ ,K ϩ -ATPase complex were expressed in S. cerevisae and analyzed in detail at the biochemical level (26). The mutant proteins are, of course, inactive because of the inability to undergo phosphorylation, but they show two striking properties. First, the ATP binding affinity of the D369N and D369A mutants is much higher than that of the WT protein. A similar effect of D351N, A and T mutants of Ca 2ϩ -ATPase has also been described (30). Second, the conformational equilibrium E 2 7 7 E 1 is significantly poised toward the E 2 forms compared with the WT. However, the structural consequences of the mutations have not been investigated. In view of the central role of the active site aspartate, evidence on changes in structural organization caused by the D369N mutation could provide important information on the energy transduction mechanism.

Materials
Escherichia coli HB101 (hsdS20 (r B Ϫ m B Ϫ), recA13, ara-14, proA2, lacY1, galK2, rpsL20 (Smr r ), xyl-5, mtl-1, supE44, Ϫ /F Ϫ ) was used for propagation and preparation of various plasmid constructs. P. pastoris strains GS115 (his4) and SMD1165 (his4, prb1) were used for transformation with the expression vector pHIL-D2(␣/␤). YNB medium (without amino acids, with ammonium sulfate) was obtained from Difco. Construction of Expression Vector, pHIL-D2(␣/␤) (Fig. 1) cDNAs encoding porcine ␣1 (accession number X03938) in plasmid pGEM and porcine ␤1 (accession number X04635) in plasmid pBR322 were provided by P. L. Jorgenesen (Copenhagen University). Plasmid pGEM1(␣1) was digested with XbaI, HindIII, and DraI, and treated by Klenow. The XbaI-HindIII fragment containing ␣1 was then ligated with the large fragment released from the E. coli-P. pastoris shuttle vector, pHIL-D2 after EcoRI, Klenow, and alkaline phosphatase treatment. The right orientation of the cloned fragment under the AOX1 promoter was determined by restriction analysis. Next, the ␤1 open reading frame and flanking regions (in plasmid pBR322) were amplified by PCR using two synthetic oligonucleotides: sense, 5Ј-cccgaattcgctgctgacccgccaccatggcccg and antisense 5Ј-atagaattcgtgtaggtccctacgtatgacag containing EcoRI sites (underlines). The PCR fragment was digested by EcoRI and ligated to EcoRI-treated plasmid pHIL-D2 to produce pHIL-D2␤1. A large part of the ␤1 gene was then replaced (using SacII and SnaBI) by the original gene, and the remaining PCR-generated region was sequenced. To construct a vector encoding both genes (␣1 and ␤1), an additional selection marker has been introduced; the kan R gene that confers resistance to kanamycin in bacteria and to G418 in yeast cells. The bacterial plasmid, pCV3 (kan R ) 2 prepared in E. coli SCS110damdcm-(Stratagene) was digested with StuI and a fragment containing kan R was ligated to the SnaBI-treated plasmid pHIL-D2␤1 to produce plasmid pHIL-D2␤1(kan R ). Next, the P AOX1 /␤1/kan R fragment was excised from pHIL-D2␤1(kan R ) by NotI and ClaI (partial digestion). The appropriate NotI-ClaI fragment (3,670 bp) was treated by Klenow and ligated to plasmid pHIL-D2␣1 that was partially digested with NaeI and dephosphorylated to produce the final expression vector pHIL-D2(␣/␤) (15,332 bp). This plasmid was used to mutate Asp 369 in the ␣1 gene (D369N) by oligonucleotide-directed, site-specific mutagenesis using the overlap-extension PCR method (31). The PCR product was sequenced.

DNA Dot Blot Screens
Mut s colonies were grown overnight at 30°C in 200 l of YPD medium in 96-well plates. 5-l Aliquots were then spotted onto a nitrocellulose filter (0.45 m, Millipore) placed on the surface of a fresh YPD plate. After evaporation of the liquid, the plates were inverted and incubated at 30°C for 24 h. Cell lysis and DNA fixation were done according to Ref. 32. A StuI-SacII fragment of the ␣1 porcine cDNA (accession number X03938) was amplified by PCR with two deoxyoligonucleotides: sense 5Ј-gatgccaaggcctgcgtggtcc, and antisense 5Ј-gccccgcggcaggaagatctcttgtagtaggtttccttctc. The PCR product was used to prepare a radioactive probe by random priming using a hexanucleotide mixture (Roche catalog number 1-277-081). The filter was incubated overnight with the probe at 42°C in a hybridization solution containing 50% formamide, 5 ϫ Denhardt's solution, 2 ϫ SSC (ϫ20 stock contains 3 M NaCl, 0.3 M sodium citrate, pH 7.0), 1% SDS, and 100 g/ml denaturated salmon sperm DNA.

Cell Growth and Induction of Protein Synthesis
Routinely yeast cultures were propagated in YPD medium. Screening Clones for Expression-5 ml of fresh BMG broth (0.1% glycerol) was inoculated with a candidate clone and incubated for 20 -24 h with vigorous shaking at 25°C, until the glycerol was exhausted and the A 600 was 2-4. Then, 0.5% methanol was added daily for 5 days. Cells were pelleted and membranes were prepared as described below. Cells were disrupted with 0.5-mm glass beads by shaking for 1 h in 1.5-ml Eppendorf tubes placed in a multitube shaker device VXR IKA basic IKA-Vibrax TM set to 2000 rpm.
Optimization of Expression-5 ml of BMG broth (1% glycerol) were inoculated with the clone of interest and incubated with shaking at 30°C for 24 h. Cells were then diluted 1000-fold into 50 -200 ml of BMG (1% glycerol) and grown to the desired A 600 , usually 2-4 units. Cells were centrifuged and re-suspended in fresh BMM broth; cells were grown for 5 days with a daily addition of methanol (0.5%) at the desired temperature.
Scaling Up of Cultures-3-Liter cultures were grown in 8-or 3-liter Bellco Spinner TM flasks with magnetic stirring and temperature was maintained at 25°C using a thermostatically controlled water bath. Sterile compressed air was supplied via a 0.2-m filter and bubbled into the culture via a 10-m porous filter. 3 liters of BMG broth (0.2% glycerol) was inoculated with 0.2 liter of a starting culture and grown for 24 h with maximal agitation of 400 rpm and airflow set to 0.1-0.2 liters h Ϫ1 . After the cell growth had stopped (at an A 600 of ϳ 2-4 units), expression of the Na ϩ ,K ϩ -ATPase was induced by adding 0.5% methanol daily for 6 days.

Membrane Preparations
P. pastoris cells were collected by centrifugation and the pellet was resuspended in an ice-cold buffer containing 1.4 M sorbitol, 10 mM MOPS/Tris-HCl, pH 7.2, 1 mM EDTA, plus protease inhibitors (1 mM PMSF, 10 g/ml leupeptin, 10 g/ml pepstatin and 10 g/ml chymotrypsin), at 1 g of cells/10 ml of buffer. The cells were mixed with an equal volume of glass beads (0.5 mm) and disrupted by 8 ϫ 1-min cycles using a Glass Bead Beater (Biospec Products, Inc.). The cooling jacket of the bead beater was filled with 50% glycerol at Ϫ20°C. Unbroken cells and heavy membranes were removed by centrifugation at 10,000 ϫ g for 10 min and light membranes were collected at 100,000 ϫ g for 1 h. The pellet was suspended in 10 mM MOPS/Tris-HCl, pH 7.2, 1 mM EDTA, 25% glycerol with protease inhibitors (as above), and stored at Ϫ80°C. Renal Na ϩ ,K ϩ -ATPase was prepared as described in Ref. 33.

SDS-PAGE and Western Blots
12-25 g of yeast membranes were mixed with 5-fold concentrated sample buffer containing 1 mM PMSF, and separated on 7.5 or 10% Tricine SDS-PAGE (34). Proteins were blotted to a polyvinylidene difluoride membrane and immunostained with anti-Lys 1012 -Tyr 1016 (anti-KETYY antibodies, 1:3,000) that recognize the C terminus of the ␣ subunit, or with anti-␤ antibodies raised against the extracellular domain of the ␤ subunit (35). Blots were visualized by enhanced chemiluminescence (ECL Plus kit) using anti-rabbit IgG horseradish peroxidase-conjugate as instructed (Amersham Biosciences). Immunoblots were scanned with an imaging densitometer (GS-690, Bio-Rad) and analyzed using the Multi-analyst software (Bio-Rad).

Biochemical Assays
Ouabain binding to membranes or to whole cells was done using [ 3 H]ouabain essentially as described in Refs. 25 and 26. In assays of ouabain binding to membranes, at the end of incubation 3 ml of ice-cold 10 mM Tris-HCl was added, the samples were filtered on Whatman GF/C filters and washed twice with the same buffer. Na ϩ ,K ϩ -ATPase activities were assayed using [␥-32 P]ATP as described in Ref. 33. Prior to both ouabain binding and ATPase assays, membranes (2 mg/ml) were incubated with 0.3 mg/ml SDS at room temperature for 30 min and then centrifuged. This procedure inactivates ouabain-insensitive ATPase and demasks any Na ϩ ,K ϩ -ATPase molecules in closed vesicles (26).

Chymotryptic Cleavage (see Refs. 10 and 36)
Renal Na ϩ ,K ϩ -ATPase or membranes containing the recombinant enzyme were centrifuged, and re-suspended in a medium containing 10 mM Tris-HCl, pH 7.4, and either 20 mM NaCl or 20 mM RbCl, at a concentration of 0.1 mg of protein/ml. The membranes were incubated for 10 min at 37°C with 5 g/ml ␣-chymotrypsin. The digestion was arrested by addition of 140 mM KCl plus 1 mM PMSF, and centrifugation. The pellets were re-suspended in 10 mM Tris-HCl, pH 7.4, and 1 mM PMSF. Cleavage fragments were visualized by immunoblotting using the anti-KETYY antibody.

Oxidative Cleavage Reactions
Recombinant membranes were centrifuged and re-suspended (0.5 mg/ml) in media containing 20 mM MES (Tris), pH 6.5, with or without 130 mM NaCl or RbCl, or ATP, AMP-PNP, or other ligands as indicated in the figure legends. To 25 l of the membrane suspension, freshly prepared solutions of 10 mM ascorbate, and different concentrations of FeSO 4 were added, then 10 mM H 2 O 2 was added to a total volume of 50 l, and the suspension was incubated at 0°C for 15 min. To arrest the reaction, 25 l of a solution of EDTA, desferrioxamine mesylate (Desferal), and PMSF were added to final concentrations of 30, 3, and 1 mM, respectively. After 5 min at room temperature, 25 l of ϫ5 concentrated sample buffer were added and samples were loaded onto gels. Cleavage fragments were visualized by immunoblotting using the anti-KETYY antibody.

Molecular Dynamics
The initial three-dimensional model of the wild type Ca 2ϩ -ATPase was truncated to comprise only the P and N domains, including bound ATP and Mg 2ϩ , as described in a recent publication (13). The D351N 2 A. Seluanov and E. Bibi, unpublished data. mutation was performed using the SwisPdbViewer. The Asn 351 side chain was manually adjusted to remove clashes and to point the side chain N-atom in the direction of the ATP. The simulations were performed with the GROMACS simulation package (37,38), using the GROMOS96 united atoms force field (39) and the SPCwater model (40). The truncated protein model was solvated with 14,764 molecules of water and a physiological ion concentration of Na ϩ and Cl Ϫ ions in a cubic box with a minimum distance of 1.0 nm between protein and box boundaries. The excess negative charge of the protein model was neutralized by an excess of sodium ions, so that the system was electroneutral. Periodic boundary conditions were applied. Long range electrostatic interactions were treated with PME approximation (41,42). Van der Waals and short range electrostatic interactions were cut off at 0.9 nm. The system was coupled to an external water bath at 300 K and an external pressure reservoir set to 1 bar. The solvated system was minimized in 500 steps of steepest decent minimization, followed by a short 10-ps simulation of the system with position restraints on the protein and the ATP with a step size of 0.002 ps. This was followed by a 900-ps simulation of the whole system without restraints, again with a stepsize of 0.002 ps. The bonds of the system were restrained with the SHAKE algorithm (43).

Transformation of P. pastoris and Screening for Expression of ␣ and ␤ Subunits-
The pig ␣1 and ␤1 cDNAs were cloned into the pHIL-D2 expression vector, each under control of the AOX1 promoter. The genes are incorporated into the endogenous AOX1 chromosomal locus by homologous recombination at the 5Ј and 3Ј ends. By placing the His4 selectable marker between the ␣1 and ␤1 genes the possibility of recombination leading to insertion of the ␤1 gene without the ␣1 gene was avoided. Transformation of the wild type P. pastoris strain GS115 with the pHIL-D2 (␣/␤) expression vector led to appearance only of degraded fragments of the ␣1 subunit. Therefore all experiments were done with a protease-deficient strain of P. pastoris, SMD1165, transformed with the pHIL-D2 (␣/␤) vector or the "empty" vector, pHIL-D2, as a control. Colonies were selected for histidine prototrophy, and then for the Mut s phenotype, which detects clones in which the AOX1 gene is disrupted by integration of the genes of interest, and as a result of which they grow slowly because of their dependence on the AOX2 gene for metabolism of methanol (22). Between 10 and 20% of the His ϩ prototrophs showed the Mut s phenotype. In the experiment of Fig. 2, 11 such Mut s clones were grown for 5 days, either as colonies on plates, or in 5 ml of BMM liquid medium, with daily addition of 0.5% methanol. Fig. 2A shows results of the dot blot assay for the ␣1 DNA content, and Fig.  2B shows an immunoblot to detect the ␣1 subunit in the membranes prepared from these 11 clones. It is striking that the different clones showed a wide variation in content of ␣1 DNA and ␣1 subunit protein and that, qualitatively, there was good agreement between the DNA content and level of ␣1 subunit expressed. The empty vector control, of course, showed no expression of the ␣1 subunit. It is known that transformation of P. pastoris can result in multicopy integration into the AOX1 locus (32). The result in Fig. 2 is consistent with the assumption of multicopy integration, and shows that screening Mut s clones reveals those expressing the highest level of DNA or protein, which can then be used for optimization of expression (e.g. clone 91). As seen in Fig. 2B the ␣1 subunit expressed in the yeast membranes runs slightly behind the ␣1 subunit of renal Na ϩ ,K ϩ -ATPase. This could indicate that post-translational clipping off of the first five residues MGKGV, which occurs in native renal cells, does not occur in the yeast cells. Fig. 2C shows that the ␤1 subunit is expressed in clone 91 as two quite sharp bands of apparent mass 47 and 44 kDa, respectively, by contrast with the heavily glycosylated ␤ subunit of the renal Na ϩ ,K ϩ -ATPase. The finding in Fig. 2C demonstrates that the ␤1 subunit is only lightly glycosylated in the P. pastoris. Light glycosylation of heter-ologously expressed proteins is another known feature of P. pastoris (44).
Optimization of Expression of Na ϩ ,K ϩ -ATPase-A series of experiments was carried out in flask cultures to determine the optimal stage of growth in the BMM prior to methanol induction, the time course after induction with methanol, the temperature in the methanol induction phase, and different growth media. Western blots, using the anti-KETYY antibody, ouabain binding on crude membranes, or ouabain binding on whole cells, were used to assess the level of expression of the protein. Initial experiments, in which P. pastoris clone 91 (see Fig. 2) was grown to different densities in the BMG medium, from the log to stationary phase, the glycerol was removed, and then methanol was added for 5 days, which showed optimal expression for log phase cultures (A 600 2-4). If glycerol was not removed prior to induction with methanol, expression was greatly suppressed. Thus, optimization of the time course of methanol induction utilized log phase cultures after removal or utilization of glycerol. Fig. 3 shows a time course of methanolinduced expression, for a 3-liter scaled-up culture in a Spinner flask at 25°C. Maximal expression of the ␣1 subunit, and oubain binding in crude membranes occurred after 6 days of methanol induction. Methanol induction in a basal salt (see Invitrogen Pichia Expression kit) medium did not lead to significant expression. Thus all experiments utilized the BMM growth medium and 6 days of induction. In experiments to examine the optimal temperature for the methanol induction phase, expression of the ␣1 subunit or ouabain binding in crude membranes was found to be similar at 30 and 25°C, somewhat lower at 20°C, while there was no expression at either 15 or 37°C. A more detailed comparison of growth at 25 and 30°C revealed an interesting phenomenon (Fig. 4). At 25°C the cells grew to about twice the density compared with that at 30°C and, although the ouabain binding to crude membranes was similar, ouabain binding to whole cells was about 3-fold higher for cells grown at 30°C (see "Discussion"). For most purposes, cells were grown in the Spinner flasks at 25°C to maximize the yield of crude membranes. Mechanical disruption of cells from 3 liters of culture, ϳ100 g wet weight, provided about 500 mg of crude membranes with specific ouabain binding capacities of 30 -50 pmol mg of protein Ϫ1 . Turnover Number of the Recombinant Na ϩ ,K ϩ -ATPase-Ouabain binding and Na ϩ ,K ϩ -ATPase activity of the expressed protein has been determined and the turnover number calculated. Scatchard plots of oubain binding to different crude membrane preparations gave dissociation constants of 17-21 nM and B max values in the range of 30 -50 pmol mg of protein Ϫ1 (see also Table I). The measured range of Na ϩ ,K ϩ -ATPase activities at 37°C was 0.17-0.23 mol mg of protein Ϫ1 min Ϫ1 .
Using the ouabain binding B max values, as a measure of the site concentration, the range of calculated turnover numbers at 37°C was 5666 -7600 min Ϫ1 . The higher values of turnover number are close to that of the renal Na ϩ ,K ϩ -ATPase (33) and indicate that a large percentage the recombinant protein is functional. Attempts to purify the recombinant enzyme by the SDS/ATP procedure of Jorgensen (33) were not successful, because of inactivation by the SDS. The ouabain binding capacity of the D369N mutant, which is discussed below, was 24 pmol mg of protein Ϫ1 and, of course, it has no ATPase activity.
Chymotryptic Cleavage of Recombinant Na ϩ ,K ϩ -ATPase-The immunoblot in Fig. 5 shows that well characterized fragments generated by specific chymotryptic cleavage of renal Na ϩ ,K ϩ -ATPase (36), in E 1 Na and E 2 (Rb) conformations, are observed also for the recombinant protein. The yield of fragments is somewhat lower for the recombinant protein, a feature seen also for Fe 2ϩ -catalyzed cleavage, as discussed below. The experiment indicates that native structural organization of the renal enzyme is maintained in the recombinant Na ϩ ,K ϩ -ATPase.
Fe 2ϩ -catalyzed Oxidative Cleavage of WT and D369N Mutant- Fig. 6 presents results of Fe 2ϩ -catalyzed oxidative cleavage at 0°C of the wild type Na ϩ ,K ϩ -ATPase, and the D369N mutant in Na ϩ -or Rb ϩ -containing media, respectively (see Refs. 9 and 10). Equal quantities of WT and D369N mutant protein were applied to the gel. The fragments of the recombinant proteins are identified in immunoblots by comparison with the fragments produced by cleavage of renal Na ϩ ,K ϩ -ATPase (two right lanes). Cleavage of renal Na ϩ ,K ϩ -ATPase in a Na ϩ -containing media (E 1 Na ϩ conformation) leads to two fragments: 80 EWVK and 283 HFIH (lane marked NKA, Na). For the recombinant WT and also the D369N protein in a Na ϩcontaining medium, the two fragments appear in parallel as the Fe 2ϩ concentrations was raised from 10 to 300 M (Fig. 6,  left). By contrast, a comparison of cleavages of WT and D369N mutant in a Rb ϩ -containing medium (E 2 (Rb) conformation) showed a distinct difference (Fig. 6, right). Cleavage of the renal Na ϩ ,K ϩ -ATPase in the E 2 (Rb) conformation produces three additional fragments, 214 ESE, 608 MVTGD, and 712 VNDS (lane marked NKA, Rb). Cleavage of the WT recombinant enzyme led to the same three fragments, albeit in reduced yield compared with the renal enzyme. However, the specific cleavages of the E 2 (Rb) conformation were largely suppressed in the D369N mutant. At 10 M Fe 2ϩ none of the three fragments, 214 ESE, 608 MVTGD, and 712 VNDS, were observed. At high Fe 2ϩ concentrations, 100 and 300 M, small amounts of the fragments appeared (Fig. 6, right, lanes 6 and 8). These experiments provide strong evidence for two Fe 2ϩ sites in the WT: site 1 ( 214 ESE, 606 MVTGD, and 712 VNDS) and site 2 ( 80 EWVK and 283 HFIH), and distortion of Fe 2ϩ binding site 1 by the D369N mutation (see Ref. 14 and "Discussion").
In optimization experiments, cleavage of the recombinant   protein at 20°C produced only small amounts of the specific fragments, whereas cleavage at 0°C produced the fragments in an acceptable yield. This difference from the renal Na ϩ ,K ϩ -ATPase reflects a lower stability of the recombinant enzyme. Cleavage of the recombinant protein was improved at pH 6.5 compared with the usual pH 7.2, and requires higher Fe 2ϩ concentrations than renal enzyme due, probably, to the large amounts of contaminant proteins that bind the Fe 2ϩ . Previously, we proposed that the Fe 2ϩ site 1 is a normal Mg 2ϩ site in the E 2 conformations (12). However, it was difficult to prove this by competition between Fe 2ϩ and Mg 2ϩ , because Mg 2ϩ ions independently stabilize the E 1 conformation (9, 12). Because Fe 2ϩ can substitute for Mg 2ϩ ions as a co-factor in phosphorylation from ATP (19), it might act in this way also for ouabain binding. This hypothesis was tested in Fig. 7 and Table I. As seen in Fig. 7, cleavage of WT in the presence of P i /Fe 2ϩ /ouabain (E 2 -P⅐Fe 2ϩ ⅐ouabain) or Fe 2ϩ /ouabain (E 2 -Fe 2ϩ ⅐ouabain), followed by the Fe 2ϩ chelator Desferal, produces two fragments, which run parallel to the 214 ESE and 712 VNDS fragments observed in the E 2 (Rb) conformation in site 1 (compare lane NKA). The cleavages are completely suppressed in the D369N mutant. This result indicates that in WT the Fe 2ϩ is tightly bound in conditions of ouabain binding with or without P i , as might be expected if Fe 2ϩ substitutes for Mg 2ϩ ions (19). Site 2 cleavages are not seen because of chelation of free Fe 2ϩ by the Desferal. Ouabain binding to membranes of the WT and D369N mutant in the presence of Mg 2ϩ or Fe 2ϩ ions is shown in Table I. At a saturating concentration of ouabain (2 M), and Mg 2ϩ ions, the maximal capacity found for binding to WT and the D369N mutant was 40.9 and 29.5 pmol mg of protein Ϫ1 , respectively, and it was fully suppressed by EDTA. Fe 2ϩ at 1 mM partially substitutes for Mg 2ϩ on either WT or the D369N mutant (16.1 and 14.2 pmol mg of protein Ϫ1 , respectively), and chelation of Fe 2ϩ by Desferal largely prevents ouabain binding. The 1 mM Fe 2ϩ used in this experiment may not be saturating. In another experiment, a significant increase in ouabain binding was observed between 0.5 and 1 mM Fe 2ϩ , but the presence of 5 and 10 mM Fe 2ϩ precipitated the membranes, thus preventing experimental access to the higher concentrations. The experiment confirms that Fe 2ϩ is able to bind at the Mg 2ϩ site of both WT and D369N proteins (see "Discussion"). D369N Mutant-Fig. 8 presents experiments to test the ATP-Fe 2ϩ complex as a substrate for phosphorylation with ATP-Fe 2ϩ -Na ϩ , on the WT and D369N mutant (see Refs. 11 and 13). The lanes marked Na show the same cleavages as in Fig. 6 (left). In the presence of ATP, the ATP-Fe 2ϩ complex is formed and, for the WT, three new fragments are observed ( 214 ESE, near 440 VAGDA, and 712 VNDS), whereas 283 HFIH and 80 EWVK fragments are suppressed (lane marked ATP⅐Na, WT). These features are exactly those described and analyzed previously for cleavage of the renal Na ϩ ,K ϩ -ATPase (11,13). Suppression of the near 283 HFIH and near 80 EWVK fragments is because of chelation of Fe 2ϩ in the ATP-Fe 2ϩ complex and reduction of the free Fe 2ϩ concentration to a negligible value. The near 440 VAGDA and 712 VNDS fragments are the product of cleavage mediated by bound ATP-Fe 2ϩ acting as an affinity cleavage reagent in the E 1 ⅐Na⅐ATP-Fe 2ϩ conformation (11,13). The 214 ESE fragment is the product of cleavage of the E 2 -P conformation, upon phosphorylation by ATP-Fe 2ϩ plus Na ϩ ions, with Fe 2ϩ substituting for the Mg 2ϩ ion (11). None of these three fragments were observed for the D369N mutant (lane marked ATP⅐Na, D369N). Cleavage of E 2 -P can also be observed, without the cleavages mediated by bound ATP-Fe 2ϩ , by first generating the phosphoenzyme, then arresting phosphorylation with the iron chelator Desferal, and shortly afterward adding the ascorbate/H 2 O 2 (11). When E 2 -P is isolated kinetically in this way, a major cleavage is seen at the 214 ESE position (lane marked ATP⅐Na Desferal, WT). In the D369N mutant no cleavage is seen (lane marked ATP⅐Na Desferal, D369N). Lack of cleavage of the D369N mutant, characteristic of the E 2 -P form, is expected because the mutant cannot undergo phosphorylation. On the other hand, the absence of cleavages mediated by bound ATP-Fe 2ϩ was not expected because the D369N mutant is known to bind ATP, and indeed with a much higher affinity than the WT (26).

ATP-Fe 2ϩ and AMP-PNP-Fe 2ϩ -catalyzed cleavage of WT and
Figs. 9-11 examine in greater detail specific cleavages of the WT and D369N mutant mediated by the non-hydrolyzable analogue AMP-PNP-Fe 2ϩ acting as an affinity cleavage reagent. When bound to the renal Na ϩ ,K ϩ -ATPase, the AMP-PNP-Fe 2ϩ complex mediates the same cleavages as ATP-Fe 2ϩ , except for that dependent on phosphorylation (11,13). In Fig. 9 cleavage was examined in a Na ϩ -containing medium, at 10 or 100 M observation could indicate that AMP-PNP-Fe 2ϩ binds less well or is a less effective cleavage reagent on the D369N mutant compared with WT or, alternatively, that uncomplexed AMP-PNP is a much better competitor of the AMP-PNP-Fe 2ϩ complex in the D369N mutant compared with WT, and displaces AMP-PNP-Fe 2ϩ from the enzyme. Both effects could also apply. To distinguish these possibilities, the AMP-PNP concentration was varied systematically from 1 to 5000 M with a fixed Fe 2ϩ of 100 M, namely with AMP-PNP concentrations ranging from far below to far above the Fe 2ϩ concentration (Fig. 10). For both WT and the D369N mutant, as the AMP-PNP concentration was raised the site 2 cleavages (near 80 EWVK and near 283 HFIH) were progressively suppressed because of chelation of Fe 2ϩ , the near 440 VAGDA and 712 VNDS fragments were observed at the intermediate concentrations, and all cleavages were suppressed at the highest AMP-PNP concentrations (1000 -5000 M). Suppression at the high concentrations is because of competitive displacement of AMP-PNP-Fe 2ϩ by uncomplexed AMP-PNP (see Refs. 11 and 13). Fig. 11 represents graphs of the relative amounts of the prominent 712 VNDS fragment for WT and the D369N mutant based on scans. Clearly, the amount of 712 VNDS is significantly lower in the D369N mutant at all concentrations of AMP-PNP. Below 100 M AMP-PNP, the AMP-PNP-Fe 2ϩ complex is the major species present and at 1-5 M AMP-PNP the concentration of uncomplexed AMP-PNP must be very low. An approximate K 0.5 for the rising phase is 2-3 M AMP-PNP-Fe 2ϩ for both WT and the D369N mutant. By contrast the K 0.5 for the falling phase is much lower for the D369N mutant compared with WT (Ϸ300 M compared with Ϸ3000 M, respectively). The experiment indicates both that AMP-PNP-Fe 2ϩ mediates cleavages less effectively and also that free AMP-PNP is a better competitor in the D369N mutation.

P. pastoris as an Expression System for Na ϩ ,K ϩ -ATPase
The most important feature of the P. pastoris system is that rather large amounts of the membranes containing the fully functional recombinant Na ϩ ,K ϩ -ATPase, and no endogenous Na ϩ ,K ϩ -ATPase activity, can be readily obtained. In addition, the relatively high specific activity observed routinely, Ϸ50 pmol of ouabain binding/mg of protein, and the high yield of membranes, Ϸ300 -500 mg/3-liter culture, represent a significant advantage for the cleavage experiments described here and other biochemical work. This site density was achieved by utilizing the protease-deficient strain SMD1165, by screening transformed clones for multicopy integrants of the ␣1 subunit gene and Western blots of the expressed protein, and by opti-FIG. 6. Fe 2؉ -catalyzed oxidative cleavage of WT and D369N mutant in E 1 Na or E 2 (Rb) conformations. Membranes were suspended in 20 mM MES buffer, pH 6.5, and 130 mM NaCl or RbCl was added. Fe 2ϩ concentration was varied as indicated. C, control uncleaved membranes. NKA, pig kidney Na ϩ ,K ϩ -ATPase cleaved with Fe 2ϩ /ascorbate/H 2 O 2 in RbCl or NaCl containing media. mizing the kinetics of expression. The Na ϩ ,K ϩ -ATPase is expressed at the cell surface, as detected by ouabain binding on whole cells, but the findings in Fig. 4 suggest that a significant amount of the expressed protein is also located in the internal membranes. The 3-fold higher ouabain binding at the cell sur-face of cells grown at 30°C compared with 25°C, with little difference in specific activity of crude membrane preparations, implies that trafficking from internal membranes to the cell membrane might be optimal at 30°C. For isolation of membranes and characterization of the expressed protein, the yeast were normally grown at 25°C because the yield of cells was more than double that at 30°C. In view of the fact that methanol sustains only a slow growth of these cells at 25°C, the even slower growth at 30°C could reflect the extra energetic burden of trafficking the pumps or interference with trafficking of other membrane proteins. For analysis of oubain binding or cation fluxes on whole cells it would be advantageous to grow them at 30°C because of the higher surface density of pumps. Growth of the cells at 25°C in the Spinner flasks, reaching A 600 of Ϸ30, is lower than the very high cell densities (A 600 of 150 -200) reported for fermentor cultures (24,45). The protease-deficient strain of P. pastoris, SMD1165, may have a lower growth potential than wild type strains, but it is also likely that growth in a fermentor with full control of the oxygen supply and methanol feeding would lead to improved yields of cells. Although P. pastoris has not been utilized for expression of many integral membrane proteins, the full potential of this system has been realized for expression and purification of the multidrug resistance protein, MDR3 (23,24). Recently, a functional His-tagged Na ϩ ,K ϩ -ATPase has been expressed in the P. pastoris cells. 3 Thus, scaling up and purification of the recombinant Na ϩ ,K ϩ -ATPase should be feasible.

Properties of the Expressed Na ϩ ,K ϩ -ATPase
By several criteria the expressed protein is functional and structurally intact. The turnover number (about 7500 min Ϫ1 ), calculated from ATPase activities and ouabain binding data, is similar to that of the renal Na ϩ ,K ϩ -ATPase (33). The chymotryptic cleavage data (Fig. 5) show that the recombinant protein can adopt the E 1 Na or E 2 (K) conformations and expose chymotrypsin-sensitive bonds for selective proteolytic cleavage, as described for the renal enzyme (36). Similarly, oxidative cleavages mediated by bound Fe 2ϩ or the ATP-Fe 2ϩ complex (Figs. 6 -10), indicate that the E 1 7 E 2 conformational states, ATP-Fe 2ϩ binding, and phosphorylation from ATP-Fe 2ϩ are similar to those for the renal Na ϩ ,K ϩ -ATPase (9 -14). On the other hand, there are indications that the recombinant protein is less stable than renal Na ϩ ,K ϩ -ATPase, including inactivation by SDS, which precludes purification by the standard SDS/ATP procedure (33), and the necessity to carry out Fe 2ϩcleavage experiments at 0°C rather than at 20°C.

Fe 2ϩ -catalyzed Oxidative Cleavage
Two Fe 2ϩ Sites: A Role for Asp 369 in E 1 7 E 2 Conformational Changes-In the E 2 (Rb) conformation the 214 ESE, near 608 MVTGD, and 712 VNDS fragments cleavages are greatly suppressed by the D369N mutation compared with WT, whereas two other cleavages, the near 283 HFIH and near 80 EWVK fragments, found in either E 1 Na or E 2 (Rb) conformations are unaffected (Fig. 6). This finding provides clear evidence for two Fe 2ϩ sites: site 1 within the active site ( 214 ESE, near 608 MVTGD, and 712 VNDS fragments) and site 2 at the membrane-water interface of transmembrane segments M3 and M1 (near 283 HFIH and near 80 EWVK fragments), as inferred also previously on the basis of less direct evidence (46). The similar cleavages in the presence of either Fe 2ϩ /Rb or Fe 2ϩ /ouabain for WT (Figs. 6 and 7) and their suppression in the D369N mutant in both conditions, as well as Fe 2ϩ -dependent ouabain binding (Table I) binding site in the E 2 conformation, in the absence of the substrates ATP or P i .
Suppression of site 1 cleavages by the D369N mutation (Figs. 6 and 7) could occur if the mutation acted like Na ϩ ions to stabilize an E 1 conformation. However, prior evidence that the D369N mutation stabilizes an E 2 conformation (26) shows that the effect has another explanation. The low amounts of the 214 ESE, near 608 MVTGD, and 712 VNDS fragments observed at elevated Fe 2ϩ concentrations in Fig. 6 (right) may indicate that Fe 2ϩ ions do not bind tightly in the E 2 (Rb) conformation or that the efficiency of cleavage of the D369N mutant is low. Similarly, lack of cleavage of the D369N mutant in the E 2 ⅐ouabain conformation of Fig. 7 indicates either that Fe 2ϩ is bound but cannot catalyze oxidative cleavage, or that bound Fe 2ϩ dissociates rapidly from the protein and is chelated by Desferal. Either explanation implies that D369N mutation significantly distorts the Fe 2ϩ (Mg 2ϩ ) site 1 in the E 2 conformation.
The previous observation that D369N A mutants stabilize E 2 conformations by comparison with the WT enzyme shows that Asp 369 plays a role in E 1 7 E 2 conformational transitions (26). The present findings allow us to propose such a role for Asp 369 in the energetics of the A and P domain interactions, based on the Ca 2ϩ -ATPase structure. Fe 2ϩ -catalyzed cleavage of the WT or renal Na ϩ ,K ϩ -ATPase indicated that the E 1 Na 3 E 2 (Rb) conformation change brings the 212 TGESE sequence (A domain) into close proximity with the 369 DKTGT, 607 MVTGD, and 708 TGDGVNDS sequences (P domain) (9,14). The current experiments show that the D369N mutation alters the interaction of Fe 2ϩ (Mg 2ϩ ) ions with the 212 TGESE (A domain) and 708 TGDGVNDS (P domain) sequences in the E 2 conformation, and imply that the relationship between these sequences is altered in the E 2 conformations. Obviously, in the WT enzyme electrostatic repulsion between the charged residues of the TGES sequence (Glu 214 ) coming into proximity with those in 369 DKTGT and 708 TGDGVNDS sequences (Asp 369 and Asp 710 ) could hinder the A to P domain interaction, making the E 2 conformation less stabilized than the E 1 conformation in which the A and P domains are separate. In the D369N mutant, removal of the charge could decrease electrostatic repulsions, making the A to P interaction and thus the E 2 conformation energetically more favored. It was proposed previously that neutralization of the charge on the D369N A mutants mimics, at least in part, the bound phosphate and Mg 2ϩ ions in the WT enzyme (26). This is an interesting concept because decrease of electrostatic repulsion of the sort just discussed could also trigger the A to P domain movement characteristic of the normal E 1 -P 3 E 2 -P transition.
ATP-Fe 2ϩ (Mg 2ϩ ) Binding-The data in Figs. 9 -11 show that cleavage of the D369N mutant with AMP-PNP-Fe 2ϩ produces the same fragments 712 VNDS (P domain) and near 440 VAGDA (N domain) fragments as for WT, but only at elevated AMP-PNP-Fe 2ϩ concentrations, and a new cleavage near 367 CSDK appears. The effects result from a reduced efficiency of cleavage catalyzed by AMP-PNP-Fe 2ϩ and also a raised affinity for AMP-PNP in competing for AMP-PNP-Fe 2ϩ . The higher AMP-PNP binding affinity is consistent with the known effect of this mutation to raise ATP affinity (26). The findings have two specific implications. First, both of the cleavages at 712 VNDS and near 440 VAGDA fail to appear at low AMP-PNP-Fe 2ϩ and both are restored at elevated AMP-PNP-Fe 2ϩ . Therefore the mutation provides strong evidence that in the WT the Fe 2ϩ in ATP-Fe 2ϩ binds simultaneously in the P (D710) and N domains (within 440 12. Structural comparison between wild type and D351N mutant based on a 900 ps molecular dynamic simulation. Upper panel, the N domain is depicted in blue, the P domain in red, ATP in yellow, and the Mg 2ϩ ion in green. Whereas the overall -fold remains the same (RMSD between WT and D351N is 2.7 Å), the ATP conformation changes from bent in the wild type (left) to extended in the mutant (right). Lower panel, detail of the change in the ATP coordination. Whereas the WT form of Thr 353 is shielded from the Mg 2ϩ ion (green ball) by ATP (left), it becomes accessible in the mutant (right). In addition, Asp 351 points toward Lys 684 in the wild type, but toward the ␥-phosphate of the ATP in the mutant, where Lys 684 stabilizes the ␤-phosphate of ATP. In both cases, Mg 2ϩ remains coordinated with Glu 439 and Asp 703 . (26) or D351N, -A, and -T mutants of Ca 2ϩ -ATPase (30) indicate that neutralization of the charge of the active site aspartate removes repulsion with the ␥-phosphate of ATP. Thus, one could hypothesize that the ␥-phosphate plus Mg 2ϩ approaches Asn 351 or other nearby residues more closely than in the WT, and interacts more strongly.
To test the hypothesis we have calculated possible structural differences between the WT model and D351N by a molecular dynamics simulation, using the published model with bound ATP-Mg 2ϩ as the starting point (13). In the context of the simulation (Fig. 12), it can indeed be observed that, in the model of the mutated form, not only has the distance between the Mg 2ϩ ion and residues in the 367 CSDKTGT sequence shortened significantly (to both Asp 351 and Thr 353 ), but also Thr 353 appears to be more accessible to the ion. With Fe 2ϩ replacing Mg 2ϩ this would be compatible with the possibility of the new cleavage in the mutant at Asp 351 or Thr 353 . Mutagenesis of Ca 2ϩ -ATPase suggested that the side chain of Thr 353 interacts with ␤or ␥-phosphates or bridging oxygen of ATP, and the carbonyl group with the Mg 2ϩ ion (48). In both the calculated models of the WT and the Asn 351 mutant structures, the Mg 2ϩ ions are depicted as being closely bound to ␥and ␤-phosphates of ATP, to Asp 703 (P domain) and to Glu 439 (N domain) (see also Ref. 13).
In the framework of the model structures, differences in the coordination of ATP in the mutated and WT forms are readily detected. Whereas in the WT model, the ␥-phosphate of ATP is electrostatically repulsed by the negative charge of Asp 351 , in the model structure of the mutant, Asn 351 stabilizes ATP by hydrogen bonding to the ␥-phosphate. Furthermore, based on our simulations, stabilization occurs via Lys 684 , which in the model of the mutant form is coordinated with the ␤-phosphate of the ATP, whereas in the model of the WT, Lys 684 coordinates Asp 351 , stabilizing it in a conformation, which keeps it away from the ␥-phosphate of ATP. Thus, within the framework of the presented calculations, the observed increase in binding affinity upon D351N mutation can be explained by three contributions: first, loss of electrostatic repulsion between Asp 351 and ATP; second, gain of hydrogen bonds between Asn 351 and the ␥-phosphate; and third, further electrostatic stabilization of the ATP by Lys 684 .
It was speculated earlier that mutations of the Asp 351 of Ca 2ϩ -ATPase to either A or N or T induce a conformation analogous to the phosphorylation transition state (30). In our WT model, the distance between the active site, Asp 351 , and the ␥-phosphate is on the order of 8 Å, whereas in the model of the mutant the distance is significantly shortened, on the order of 3.5 Å. Thus the models of the wild type and the mutant support the notion that the latter might resemble a transition state analogue. In this context, one can also speculate that the loss of electrostatic attraction between Asp 351 and Lys 684 , which will have to occur to allow Asp 351 to approach the ␥-phosphate of the ATP, might be compensated by binding of Lys 684 to the ␤-phosphate of ATP, thus lowering the activation energy of phosphorylation. The model of the mutant structure (with Asp 351 instead of Asn 351 ) could resemble an E 1 ⅐ATP-Mg 2ϩ state with 2Ca 2ϩ ions bound and ready to phosphorylate Asp 351 , whereas the model of the WT structure would be analogous to an E 1 ⅐ATP-Mg 2ϩ form in the absence of Ca 2ϩ ions, and unable to phosphorylate Asp 351 . The equivalent conformations of Na ϩ ,K ϩ -ATPase would be E 1 (3Na)⅐ATP-Mg 2ϩ with 3Na ϩ ions bound and poised for phosphorylation (mutant structure), or E 1 ⅐ATP-Mg 2ϩ with ATP bound tightly but still unable to phosphorylate (WT structure), respectively.
The work described here shows how the Na ϩ ,K ϩ -ATPase expressed in P. pastoris can be utilized to study events in the active site by Fe 2ϩ -catalyzed oxidative cleavage. The recombi-nant Na ϩ ,K ϩ -ATPase in P. pastoris appears to provide a promising tool for studying the energy transduction mechanism by Fe 2ϩ -catalyzed oxidative cleavages, or other techniques, particularly when combined with molecular modeling.