Neutralization of the Charge on Asp369 of Na+,K+-ATPase Triggers E1 ↔ E2 Conformational Changes*

This work investigates the role of charge of the phosphorylated aspartate, Asp369, of Na+,K+-ATPase on E1 ↔ E2 conformational changes. Wild type (porcine α1/His10-β1), D369N/D369A/D369E, and T212A mutants were expressed in Pichia pastoris, labeled with fluorescein 5′-isothiocyanate (FITC), and purified. Conformational changes of wild type and mutant proteins were analyzed using fluorescein fluorescence (Karlish, S. J. (1980) J. Bioenerg. Biomembr. 12, 111–136). One central finding is that the D369N/D369A mutants are strongly stabilized in E2 compared with wild type and D369E or T212A mutants. Stabilization of E2(Rb) is detected by a reduced K0.5Rb for the Rb+-induced E1 ↔ E2(2Rb) transition. The mechanism involves a greatly reduced rate of E2(2Rb) → E1Na with no effect on E1 → E2(2Rb). Lowering the pH from 7.5 to 5.5 strongly stabilizes wild type in E2 but affects the D369N mutant only weakly. Thus, this “Bohr” effect of pH on E1 ↔ E2 is due largely to protonation of Asp369. Two novel effects of phosphate and vanadate were observed with the D369N/D369A mutants as follows. (a) E1 → E2·P is induced by phosphate without Mg2+ ions by contrast with wild type, which requires Mg2+. (b) Both phosphate and vanadate induce rapid E1 → E2 transitions compared with slow rates for the wild type. With reference to crystal structures of Ca2+-ATPase and Na+,K+-ATPase, negatively charged Asp369 favors disengagement of the A domain from N and P domains (E1), whereas the neutral D369N/D369A mutants favor association of the A domain (TGES sequence) with P and N domains (E2). Changes in charge interactions of Asp369 may play an important role in triggering E1P(3Na) ↔ E2P and E2(2K) → E1Na transitions in native Na+,K+-ATPase.

The kinetic mechanism of P-type cation pumps is now well established. Active cation transport involves covalent phosphorylation by ATP and dephosphorylation of an aspartate residue coupled to cation movements mediated by E 1 7 E 2 2 conformational changes. The molecular mechanism is, of course, the central question of energy transduction. Molecular structures of the sarcoplasmic reticulum Ca 2ϩ -ATPase (SERCA1) in several conformations are available (1)(2)(3). Two crystal structures of the Na ؉ ,K ؉ -ATPase, consisting of ␣, ␤, and FYXD subunits (4), in an E 2 (2Rb)-MgF 4 2Ϫ conformation (equivalent to E 2 (2Rb)⅐P) have been published recently (5,6). The architecture of the ␣ subunit is very similar to the Ca 2ϩ -ATPase, consisting of head, stalk, and membrane sectors, with 10 trans-membrane segments and a cytoplasmic sector consisting of N (nucleotide binding), P (phosphorylating), and A (anchor or actuator) domains. Compared with the first published structure of the Na ϩ ,K ϩ -ATPase at 3.5 Å (5), the more recent structure at 2.4 Å (6) reveals greater detail of the cation binding domain, resolution of the ␤ subunit ectodomain, and an FXYD protein ectodomain.
In the case of Na ϩ ,K ϩ -ATPase, three Na ؉ ions and two K ؉ ions are transported in the E 1 ATP 3 E 1 P(3Na) 3E 2 P and the E 2 P 3 E 2 (2K) 3 E 1 ATP halves of the cycle, respectively, at the expense of one molecule of ATP. Binding of ATP with low affinity to E 2 (2K) is the first step in the catalytic cycle and is followed by accelerated conversion of E 2 (2K)ATP to E 1 ⅐ATP in which ATP is bound with high affinity (7)(8)(9). In the case of Ca 2ϩ -ATPase, two Ca 2ϩ ions and two protons are transported in the E 1 ATP 3 E 1 P 3 E 2 P E 2 P 3 E 2 3 E 1 ATP halves of the cycle, respectively.
The mechanism of energy transduction by P-type cation pumps is best addressed by reference to the Ca 2ϩ -ATPase, which has been crystallized in most of the relevant conformations, reviewed recently in detail (1)(2)(3). Crystal structures of Ca 2ϩ -ATPase show mainly rigid body movements of the N, P, and A domains, mediated by the flexible linkers between the N and P and between the A domain and trans-membrane segment (M1-M3), coupled mechanically to movements of M1-M6 trans-membrane segments that allow alternate access of two Ca 2ϩ ions and two protons to the occlusion sites within M4, M5, M6, and M8. M7-M10 segments appear to act as immovable anchors. In the E 1 ATP state, N and P domains are in close proximity cross-linked by the bound ATP, whereas the A domain is displaced to one side. Following phosphorylation to E 1 P, the conformational change to E 2 P involves a large rotation of the A domain bringing it into close proximity with the P domain, and also the N domain, whereas the N domain is displaced from the P domain. In particular, the conserved sequence TGES (A domain) comes close to the phosphorylated aspartate (P domain). Following hydrolysis of E 2 P and dissociation of phosphate, ATP bound to E 2 induces the change back to the E 1 state, with the A domain displaced, and the P and N domains come back into close proximity.
This study was prompted by the following paradox. In the case of Ca 2ϩ -ATPase, dissociation of ADP from E 1 P triggers the E 1 P 3 E 2 P transition. This arises because the ADP (in the E 1 P structure) and the TGES loop (in the E 2 P structure) occupy the same position relative to the P domain (3). In the case of Na ϩ ,K ϩ -ATPase, ADP, like ATP, is known to stabilize the E 1 conformation (9,10). Thus, it is clear that ADP must dissociate from E 1 P(3Na) to allow the transition to E 2 P. However, in the absence of ADP or ATP, an E 1 conformation is stable either with bound Na ؉ or without Na ؉ in a sufficiently high ionic strength buffer (9,(11)(12)(13). This is different from Ca 2ϩ -ATPase in which E 2 is the principal conformation in the absence of Ca 2ϩ ions at neutral pH (14 -16). In other words, for Na ϩ ,K ϩ -ATPase a spontaneous conformational change of the dephosphoenzyme E 1 3 E 2 or E 1 (3Na) 3 E 2 does not occur even when ADP is absent, but it occurs only in the context of the phosphoenzyme E 1 P(3Na) 3 E 2 P. Thus, a question arises whether dissociation of ADP from E 1 P(3Na) is sufficient or whether the covalently bound phosphate/Mg 2ϩ itself plays a role in triggering the E 1 P(3Na) 3 E 2 P conformational change.
One way to address this issue is to investigate a possible role of the phosphorylated aspartate, Asp 369 . D369N/D369A/ D369T/D369E mutants of Na ϩ ,K ϩ -ATPase have been expressed previously in different cells and shown to abolish ATP hydrolysis (17,18). Pedersen et al. (19) expressed wild type and D369N and D369A mutants in Saccharomyces cerevisiae and found that, although they are inactive, the D369N/D369A mutants have a very high affinity for ATP, compared with wild type, but have a similar ADP binding affinity. By analyzing antagonism of ATP binding by K ؉ ions, it was also suggested that the E 1 7 E 2 equilibrium is altered, favoring E 2 conformations. The D369N mutant was later expressed in Pichia pastoris (20) and studied by specific oxidative cleavages mediated by Fe 2ϩ , which substitutes for the catalytic Mg 2ϩ (21). The D369N mutation strongly suppressed specific Fe 2ϩ -catalyzed cleavages. Compared with the wild type Ca 2ϩ -ATPase, mutants of its corresponding phosphorylated aspartate, D351N/D351A/ D351T, also bind ATP with a very high affinity (22). A recent crystal structure of D351A with bound ATP is virtually identical to the wild type, showing that the high affinity is due to removal of electrostatic repulsion between the ␥-phosphate of ATP and the Asp 351 carboxylate, and not to a conformational change (23).
Fluorescence probes are very useful tools for studying E 1 7 E 2 conformational changes. They allow detailed characterization of specificities and affinities of ligands that stabilize E 1 or E 2 forms and especially measurement of rates of individual E 1 3 E 2 and E 2 3 E 1 transitions (8,11,13). Fluorescence probes used extensively to study native Na ϩ ,K ϩ -ATPase include intrinsic tryptophan fluorescence (11), noncovalently bound formycin nucleotides (8) and eosin (24), as well as covalently bound labels such as FITC 3 (13), iodoacetamidofluorescein (25,26), and the electrochromic shift dye RH421 (27,28). FITC covalently labels the ␣ subunit of renal Na ϩ ,K ϩ -ATPase, predominantly at Lys 501 within the ATP-binding site (the N domain) (13,29). The fluorescein chromophore is sensitive to the pH of its local environment. Because of the rather large fluorescence changes (25-35%) associated with E 1 7 E 2 conformational changes observed for the FITC-labeled renal Na ϩ ,K ϩ -ATPase, it was possible to carry out detailed equilibrium and stopped-flow fluorescence studies (13, 30 -33). FITC-labeled sarcoplasmic reticulum Ca 2ϩ -ATPase (16,34,35) and gastric H ؉ ,K ؉ -ATPase (36) have also been used extensively to monitor E 1 7 E 2 conformational changes.
The fluorescence methods described above have not been applied to recombinant Na ϩ ,K ϩ -ATPase. Recent publications from this laboratory describe expression of Na ϩ ,K ϩ -ATPase in P. pastoris and purification of the recombinant protein in a detergent-soluble functional state in quantities up to 1 mg (20,37,38). The high yield of purified protein now makes it possible to carry out detailed biochemical and biophysical work. Development of fluorescein-labeled recombinant Na ϩ ,K ϩ -ATPase described here has allowed us to examine directly effects of charge neutralization on Asp 369 on E 1 7 E 2 equilibria and to investigate its mechanism.

EXPERIMENTAL PROCEDURES
Materials-Escherichia coli XL-1 blue strain was used for propagation and preparation of various plasmid constructs. Yeast Lytic Enzyme from ICN Biomedicals, Inc. (catalogue number 152270), was used for transformation of yeast. P. pastoris protease-deficient strain SMD1165 (his4, prb1) was used for transformation. DDM (catalogue number D310) and C 12 E 8 (25% w/w, catalogue number O330) were purchased from Anatrace. Synthetic SOPS (sodium salt)) was obtained from Avanti Polar Lipids and stored as chloroform solutions. BD Talon TM metal affinity resin (catalogue number 635503) was obtained from Clontech. Cholesterol, ouabain (O3125), and FITC (catalogue number F7250) were from Sigma. [ 32 P]ATP and [ 3 H]ouabain were obtained from Amersham Biosciences. All other materials were of analytical grade.
Yeast Transformation and Selection; Screening for Expression; Yeast Growth and Induction of Protein Synthesis-10 g of linear DNA, obtained by digestion of the pHIL-D2 (wild type or mutant ␣ 1 /His 10 -␤ 1 ) with NotI, were used to transform spheroplasts of P. pastoris SMD1165, and His ϩ Mut s transformants were selected (20). Mut s clones were grown in 5 ml of BMG cultures, and expression was induced with methanol for 5 days. Small scale membrane preparations were made. Samples were run on SDS-PAGE, and relative expression in the different Mut s clones was determined by Western blots, using the antibody anti-KETYY antibody (at a dilution of 1:3000), which recognizes the C terminus of the ␣ subunit. Ouabain binding was also carried out to determine the highest expressing clones that were subsequently used routinely.
For large scale growth and expression of the recombinant Na ϩ ,K ϩ -ATPase (20,37), single colonies propagated in YPD medium were inoculated into 5 ml of BMG liquid medium, which was incubated for 48 h with vigorous shaking at 30°C. The culture was further diluted 100-fold to a final volume of 300 ml and grown in the BMG medium (glycerol 1%) for a further 24 h to an A 600 of 2-6. Cells were then poured into 3 liters of BMG medium (with 0.2% glycerol) in Bellco Spinner Flasks TM with air supplement (0.5 liter/min) and magnetic stirring at 250 rpm, and temperature was maintained at 30°C. After 24 h, cell growth had stopped (A 600 ϭ 2-4). Expression of the Na ϩ ,K ϩ -ATPase was induced by adding 0.5% methanol daily for 5 days. Routinely, the wild type was induced at 25°C and mutants at 20°C.
The maximal A 600 achieved under these conditions was ϳ10 -15. On the 6th day cells were collected and stored at Ϫ20°C.
Membrane Preparations and FITC Labeling of Membranes-Cells were broken with glass beads, and membranes were prepared as described previously (20). Membranes were stored at Ϫ80°C in a solution containing 10 mM MOPS/Tris, pH 7.4, and 25% glycerol, with a mixture of protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin, chemostatin, and leupeptin). Roughly 1 g of membrane protein was obtained per 3 liters of culture. For FITC labeling (see Ref. 13,31), membranes were suspended at 2 mg/ml in a buffer containing 50 mM NaCl, 1 mM EDTA, 20 mM Tris, pH 9.2, with protease inhibitors. Fluorescein 5Ј-isothiocyanate (usually 1-2 M FITC dissolved in DMSO) was added, and membranes were incubated for 1 h at 20°C in the dark. The suspension was then diluted 3-fold with an ice-cold solution of 100 mM MOPS, pH 6.45, mixed for 10 min, and centrifuged at 100,000 ϫ g for 80 min. The pelleted labeled membranes were resuspended at 1 mg/ml in 10 mM MOPS/Tris, pH 7.4, and 25% glycerol. Fluorescein bound to ␣ subunit, separated by SDS-PAGE, was detected with a Typhoon fluorescence imager with excitation supplied by a blue laser and emission at 520 nm set by a monochromator.
Protein was determined by scanning Coomassie-stained gels for the ␣ subunit content of the purified recombinant enzyme and compared with known amounts of pig kidney Na ϩ ,K ϩ -ATPase. Ouabain binding to yeast membranes was assayed using [ 3 H]ouabain essentially as described (19). Na ϩ ,K ϩ -ATPase activities and covalent phosphorylation were done as described (38,40,41).
Equilibrium and Stopped-flow Fluorescence Measurements-Equilibrium fluorescence changes were measured in a Varian fluorimeter at room temperature (20 -23°C). 10 -15 g of FITC-labeled purified recombinant enzyme were incubated for 30 min at room temperature and then added to a stirred cuvette containing 2 ml of the following solution: 150 mM choline chloride, 10 mM Hepes (Tris), pH 7.5 (and other pH values between 7 and 8), or 10 mM Mes (Tris) (pH range 5.5-6.5). The same buffer with no enzyme was used to bring the fluorescence level to zero before addition of the enzyme. Excitation was at 495 nm, and emission was at 520 nm, with both slits adjusted to 5 nm. After stabilization of the fluorescence, ligands such as RbCl, KCl or NaCl, P i (Tris), and vanadate (Tris) were added until no further fluorescence change occurred. For equilibrium titrations, graded concentrations of ligands such as RbCl were added, and the fluorescence change was recorded. For each addition, a correction was made for the dilution effects of the added volume.
Stopped-flow fluorescence measurements at 23°C were made using an Applied Photophysics stopped-flow fluorimeter. Excitation via a monochromator was at 495 nm, and emission was measured via a 515-nm cutoff filter. The two syringes were loaded with 1.2 ml of solution. Each push utilized 70 l per syringe. 3-4 pushes were used to fill the flow tubes, thus allowing up to 10 replicates for repetitive measurement in each condition. Dead time was ϳ1.5 ms. Purified enzyme was incubated for 30 min at room temperature before mixing with the relevant solution. Traces were smoothed with the use of an "oversample" function of the machine. 4 -9 such traces were then averaged to produce a single trace and analyzed as described below.
Solutions-Total ionic strength of all solutions in syringe 1 and 2 was maintained at 175 mM, consisting of 10 mM Hepes, pH 7.5, and choline chloride, RbCl, KCl, NaCl, or other ligands such P i (Tris) or vanadate (Tris) at a total ionic strength of 165 mM. Solutions not containing monovalent cations or other ligands consisted of 10 mM Hepes, pH 7.5, and 165 mM choline chloride. The final concentrations of all ligands not present in both syringes was half that given below. The design of the experiments follows that developed previously (8,11,13).  Derivation of Kinetic Parameters-All fits were done using the KaleidaGraph program (Synergy Software).
1) Rb ϩ titrations in equilibrium measurements were fitted to the following form of the Hill Equation 1, where n ϭ Hill coefficient, and K 0.5 ϭ concentration of Rb ϩ ions required for half-maximal fluorescent signal. For each curve ⌬F max was found by fitting the raw data, and the ratio of ⌬F/⌬F max was then recalculated. Values of ⌬F/⌬F max for replicate experiments were combined, and the best fit average parameters n H Ϯ S.E. and K 0.5 Ϯ S.E. were recalculated. This normalization enables comparison of the curves in different conditions or for wild type and mutant proteins.
2) Time courses in the stopped-flow fluorescence experiments were fitted either to single or double exponential function as shown in Equations 2 and 3, The "end point" is the constant fluorescence intensity after the reaction has reached equilibrium; k 1 and k 2 are rates of first and second exponentials. a 1 and a 2 are amplitudes of first and second exponentials.

RESULTS
Expression of Wild Type and Mutants of Na ϩ ,K ϩ -ATPase in P. pastoris-Expression of pig ␣ 1 ␤ 1 subunits and T212A, D369A, D369N, and D369E mutants was done as described under "Experimental Procedures" and as described previously (20). The wild type was expressed optimally at 25°C, and mutants were expressed at 20°C. The expression levels of the proteins in the yeast membranes were estimated by immunoblots of the ␣ subunit and also ouabain binding. The range of ouabain binding was as follows: wild type, 20 -40 pmol/mg protein; D369A, 20 -30 pmol/mg protein; D369N, 15-35 pmol/mg protein; D369E, 3-7 pmol/mg protein; and T212A, 20 -30 pmol/mg protein. Attempts to express E214A and E214Q mutants were not successful.
Labeling and Purification of Fluorescein-labeled Na ϩ ,K ϩ -ATPase; Optimization of Fluorescence Signals-We have taken two possible approaches to labeling the recombinant enzyme with FITC to detect fluorescence signals associated with E 1 7 E 2 conformational changes (see Fig. 3 for examples of the latter). As one approach, we purified the protein and then attempted to label it with FITC at pH 9.0. However, K ϩ -and Na ϩ -induced fluorescein fluorescence changes were not observed, probably because the high pH during labeling inactivates the protein. The protein is much more stable in the yeast membrane than in the detergent-soluble state (40) and should survive the incubation at pH 9.0. Therefore, we labeled the Na ϩ ,K ϩ -ATPase in the yeast membranes with FITC before purifying the fluorescein-labeled enzyme. This strategy turned out to be successful. The expressed Na ϩ ,K ϩ -ATPase constitutes no more than 0.5-1% of the yeast membrane proteins, and therefore FITC could label many proteins at pH 9.0. However, at low concentrations of FITC, one could expect to label Na ϩ ,K ϩ -ATPase more selectively at Lys 501 than at other lysines and minimize labeling of other proteins. Unrelated fluoresceinlabeled proteins should be removed during the purification. With these assumptions, extensive optimization of labeling, purification of labeled Na ϩ ,K ϩ -ATPase, and K ϩ -and Na ϩ -dependent fluorescence changes were undertaken. FITC labeling prevents ATP binding and inactivates covalent phosphorylation by ATP (13). As seen in Table 1, incubation of the yeast membranes with 10 M FITC for 1 h at pH 9.0 inactivated phosphorylation by 84.6%, whereas 50 M inactivated fully. Therefore, 10 M FITC was chosen initially, although subsequent optimization experiments showed that lower concentrations of FITC were preferable. A more sensitive criterion for selective labeling by FITC at Lys 501 is that it should be suppressed by ATP (13). Fig. 1 shows an experiment in which membranes expressing wild type and the mutant proteins were incubated with 1 M FITC at pH 9.0 for 1 h in the absence and presence of 1-1000 M ATP. The fluorescence of the bound fluorescein was detected at 520 nm. Indeed, the labeling of the ␣

TABLE 1 Inhibition by FITC of phosphorylation of recombinant Na ؉ ,K ؉ -ATPase
Wild-type membranes suspended in a medium containing 50 mM NaCl, 1 mM EDTA, 20 mM Tris, pH 9.2, were incubated with FITC at 0.10 or 50 M for 1 h at room temperature. The membranes were then centrifuged to remove the FITC and resuspended in the normal medium. Phosphorylation by ATP was then measured. subunit was progressively suppressed by increasing concentrations of ATP. This is indicative of selective labeling. For comparison with labeling of the ␣ subunit, labeling of an unrelated protein (Fig. 1, asterisk) was not affected by the presence of ATP. Several other unrelated bands were also labeled unselectively and are not shown. As a semi-quantitative measure of the effect of ATP concentrations, the fluorescence intensity associated with the ␣ subunit was scanned for each concentration of ATP, and the values relative to the control are recorded below each lane in Fig. 1. One obvious finding was that the concentration of ATP required to suppress labeling of the D369A mutant was much lower than that required to suppress labeling of the wild type, D369E, and T212A mutants (compare the values at 1 M). Because the ATP binding affinity of the D369A mutant is known to be greatly increased compared with that of the wild type (19), this observation provides a further indication for selective labeling of Lys 501 . It is also noticeable that 25-50% of the labeling in the different clones was not suppressed by ATP, even at 1000 M. This indicates that some of labeling of the ␣ subunit is unselective, even in optimal conditions. After labeling the protein with FITC, the next step was solubilization by ␤-DDM and the purification of the wild type and mutant Na ϩ ,K ϩ -ATPase as described under "Experimental Procedures." The proteins were eluted in a medium containing 150 mM imidazole, 0.1 mg/ml C 12 E 8 , 0.05 mg/ml SOPS, and 0.01 mg/ml cholesterol, as described, but the elution buffer was nominally devoid of both Na ϩ and K ϩ ions, because the fluorescence experiments involved measurement of responses after addition of Na ϩ and K ϩ . Fig. 2 presents an example of purification of wild type and two mutant proteins (D369N and T212A) after labeling membranes in the optimal conditions discussed below. The proteins are all about 80 -90% pure, and the ␣ subunit was clearly labeled as detected by the fluorescein fluorescence (the ␤ subunit does not bind FITC and is not fluorescent). D369A was purified to the same extent as the proteins shown in Fig. 2, whereas D369E was less pure due to the lower level of expression (data not shown). The final step was the detection and optimization of fluorescein fluorescence changes. Fig. 3 shows a standard equilibrium fluorescence experiment with excitation at 495 nm and emission at 520 nm. The detergentsoluble fluorescein-labeled Na ϩ ,K ϩ -ATPase was added to a medium containing 166 mM choline chloride, which stabilizes an E 1 conformation. Upon addition of sufficient Rb ϩ or K ϩ ions, the protein is converted to an E 2 (2Rb)/(2K) conformation, and the fluorescence drops. Upon subsequent addition of sufficient Na ϩ , the protein is re-converted to an E 1 Na conformation, and the fluorescence change reverts to the original level. Fig. 3 shows that this behavior is seen for the wild type and all the mutants, indicating that E 1 7 E 2 transitions can be detected in each case. One difference already evident in these initial experiments is that reversion of the signal in D369N is incomplete at 50 mM NaCl. The experiment in Fig. 3 shows the signals for typical optimized preparations and denotes the amplitude of the signal changes, for example 13.7% in the case of the wild type protein. However, in initial experiments a fluorescence change of only 7% was achieved with the wild type protein.
Optimization of the labeling conditions included variation of pH during labeling, FITC concentration, and time and temperature of the labeling reaction. Eventually, optimal signal amplitudes were obtained after labeling the membranes at room temperature with 1 M FITC for 1 h at pH 9.0 (the conditions in Fig. 1).  The effects of Rb ϩ (K ϩ ) and Na ϩ in the high ionic strength buffer are quite similar to those observed previously using the native membrane-bound renal Na ϩ ,K ϩ -ATPase except for a lower amplitude, typically 12-15% for recombinant wild type, compared with 25-35% for renal Na ϩ ,K ϩ -ATPase (13). Another similar property is that the emission maximum for the fluorescein-labeled recombinant Na ϩ ,K ϩ -ATPase was 520 nm and did not change upon addition of Rb ϩ or K ϩ ions (not shown). Thus, the fluorescence change represents a change in fluorescence intensity of the bound fluorescein and not a spectral shift, as also found previously for renal Na ϩ ,K ϩ -ATPase. These features provide strong evidence that the basic E 1 7 E 2 mechanism is preserved in the purified detergent-soluble recombinant protein and can be studied using fluorescein fluorescence changes.
Analysis of E 1 7 E 2 (2Rb) Conformational Transitions of Wild Type and D369N/D369A Mutants-This section compares equilibrium fluorescence titrations of the E 1 7 E 2 (2Rb) transitions and rates of the individual transitions E 2 (2Rb) 3 E 1 Na and E 1 3 E 2 (2Rb) measured in a stopped-flow fluorimeter. Rb ϩ ions are congeners of K ϩ ions and are used routinely in place of K ϩ ions because the rates of transitions are slower, but some experiments have also been done with K ϩ ions.
Equilibrium titrations of the dependence on Rb ϩ concentration of the fluorescence change accompanying the E 1 7 E 2 (2Rb) transition were analyzed by recording the response to graded additions of RbCl. Fig. 4 shows typical Rb ϩ titrations for wild type, D369A, D369N, D369E, and T212A mutants. It is obvious that the curves for wild type, D369E, and T212A mutants differ significantly from those for the D369N and D369A mutants. The titration curves were sigmoid and were fitted to a Hill function leading to a value of the K 0.5 Rb and Hill coefficient n, summarized in Table 2 (and also K ϩ titration data for the wild type). The clear finding is that the K 0.5 Rb for the D369A and D369N mutants is severalfold lower (6.22-fold for D369N and 3.3-fold for D369A) than for the wild type, D369E, and T212A, which are quite similar. The Hill coefficients, in the range 1.6 -2.3, are not significantly different between wild type and mutants. The Hill equation is used here empirically to obtain a value for K 0.5 Rb and n. Although, in principle, no mechanistic significance can be attributed to n H , a value close to 2 is, of course, suggestive of occlusion of two Rb ϩ ions, as is known to be the case (42).
The following simple kinetic Scheme 1 and the accompanying Equation 4 depict occlusion of two Rb ϩ ions coupled to the conformational change and illustrate an important point on the analysis of the mutants.
K R1 and K R2 are the dissociation constant for the two Rb ϩ ions and K C is a conformational equilibrium constant. In an equilibrium fluorescence titration between E 1 and E 2 (2Rb) the following relationship will hold (Equation 4), It can be shown that K 0.5 Rb is a function of all three constants K R1 , K R2 , and K C , and without independent information it is not possible to estimate the individual constants. Thus, the Hill equation is more useful as an empirical measure of the "apparent affinity" of the Rb ϩ ions. However, the point of showing the equation is that the difference in K 0.5 Rb between wild type and D369N and D369A mutations could reflect differences in either or both the intrinsic dissociation constants, K R1 and K R2 , and the conformational equilibrium constant, K C . Thus, it is necessary to obtain independent evidence to distinguish between the possibilities. Measurement of the rates of the individual transitions E 2 (2Rb) 3 E 1 Na and E 1 3 E 2 (2Rb) by stopped-flow fluorescence allows analysis of the mechanism of the change in K 0.5 Rb seen in equilibrium titrations. It is assumed that the cation binding and dissociation steps are rapid by comparison with the conformational change in either direction. Thus, the rate constants of the observed fluorescence changes are limited by the rates of the conformational changes. Fig. 5 shows an experiment that measures the rate of E 2 (2Rb) 3 E 1 Na. The enzyme Representative titrations of the E 1 7 E 2 (Rb) change for wild type (WT) and mutants. The data have been normalized, as described under the "Experimental Procedures," to permit comparison between the different proteins. Solid lines represent curves fitted to the data and provide the parameters in Table 2 for Rb ϩ . was pre-stabilized in the E 2 (2Rb) conformation and then mixed rapidly with Na ϩ ions, which stabilize the E 1 Na conformation detected by the progressive rise in fluorescence. The rate for D369N and D369A is much slower than for the wild type. The traces were fitted to single exponential functions and gave the values of the rate constants in Table 3. The data show that the rate of E 2 (2Rb) 3 E 1 Na for D369A is 13-fold and that for D369N is 20-fold slower than for the wild type. It was also a repeated finding that a larger effect on the conformational change was seen for D369N compared with D369A (see Tables  2 and 3). The rate of E 2 (2K) 3 E 1 Na was also measured for wild type and is faster than that for E 2 (2Rb) 3 E 1 Na (Table 3). Fig. 6 shows experiments to measure the rates of the opposite conformational transition E 1 3 E 2 (2Rb) at several Rb ϩ concentrations (20, 30, and 83 mM) for wild type and D369N or D369A mutants. In this type of experiment, the proteins are suspended in an Na ϩ -free high ionic strength medium and mixed with different concentrations of Rb ϩ ions in the stopped-flow fluo-rimeter. The conversion of the E 1 to the E 2 (2Rb) conformation is associated with a decrease in fluorescence. Evidently, the rates of this transition for wild type and mutants are quite similar. The rate constants obtained from fits to an exponential function are plotted against Rb ϩ concentrations in Fig. 7A, which shows clearly that there is little or no difference between the wild type and either mutant (exact values Ϯ S.E. are given in Table 3). According to the kinetic scheme above, for a Rb ϩ (K ϩ )-induced conformational transition the observed rate constant shows a saturating dependence on Rb ϩ (K ϩ ) concentration. The data in Fig. 7A show the rates of the transition increased as the Rb ϩ concentration was raised, in the case of all the wild type and mutant recombinant proteins, although saturation was not reached at the highest Rb ϩ concentration used (but see Fig. 7B for saturation by K ϩ ions). A technical limitation of this experiment was that the accessible range of Rb ϩ concentrations was limited by the total ionic strength, and the number of Rb ϩ concentrations that could be analyzed, particularly for the mutants, was limited due to the necessity to use large amounts of protein. However, because the rate of E 1 3 E 2 (2Rb) was unaffected at any of several Rb ϩ concentrations, it is unlikely that the D369N/D369A mutations affect either the maximal rate of E 1 3 E 2 (2Rb) or the intrinsic affinity for Rb ϩ ions. Thus, the simple conclusion is that the mechanism involves a large reduction in the rate of E 2 (2Rb) 3 E 1 Na with little or no effect on the rate of E 1 3 E 2 (2Rb) or Rb ϩ affinity. Fig. 7B provides an interesting mechanistic insight into the lack of effect of the mutations on E 1 3 E 2 (Rb). For the wild type protein, we compared the rates of E 1 3 E 2 (2K) and E 1 3 E 2 (2Rb) at increasing concentrations of K ϩ and Rb ϩ ions, respectively. The raw data for E 1 3 E 2 (2K) were fitted to exponential decay curves, and the rate constants are plotted as a function of K ϩ concentration. The data for the Rb ϩ are the same as in Fig. 7A. The figure shows that the rate for E 1 3 E 2 (2K) is about 3-fold higher than for E 1 3 E 2 (2Rb) and also shows saturation behavior in the accessible range of K ϩ concentrations. The data for E 1 3 E 2 (2K) have been fitted to a hyperbola with k E1 3 E2(K) 125 Ϯ 15 s Ϫ1 , and the K K was 39 Ϯ 10.5 mM. As we discuss below, the difference with K ϩ and Rb ϩ ions shows that the rate of E 1 3 E 2 (2ϫ) is limited by events in the cation-binding sites rather than in the active site.
Effects of pH on E 1 7 E 2 (2Rb) Conformational Transitions of Wild Type and D369N Mutant-It was reported many years ago that pH affects the conformational equilibrium E 1 7 E 2 (2Rb), by a "Bohr"-like effect (12). In particular it was observed that acidic pH favors E 2 (2Rb) and alkaline pH favors E 1 . The observations in Figs. 8 and 9 and Table 4 provide a mechanistic explanation of this Bohr-like effect. The Rb ϩ titrations in Fig. 8 (and the fitted parameters in Table 4) show that a reduction of the pH from 7.5 to 5.5 has a large effect on the K 0.5 Rb of the wild type, the value being reduced by 10-fold from 3.45 Ϯ 0.1 to 0.35 Ϯ 0.03 mM. By contrast, the K 0.5 Rb for the D369N mutant is reduced only from 0.91 Ϯ 0.02 at pH 7.5 to 0.37 Ϯ 0.01 mM at pH 5.5, a factor of 2.45-fold. An increase in apparent affinity of the wild type for Rb ϩ caused by raising the proton concentration by 100-fold must be due to an effect of the pH on the conformation of the protein. (Note for comparison that raising the pH even further to 8.0 actually causes a fall in  Table  3. Wild type, black; D369N, gray; D369A, light gray.

Rates of E 2 (2Rb) 3 E 1 Na and E 1 3 E 2 (2Rb) conformational changes
Fitted rate constants for the traces in Figs. 5 and 6 are given. Rate constants of E 2 (2Rb) 3 E 1 Na for wild type and Asp 369 represent averages from several independent experiments. The data for E 2 (2K) 3 E 1 Na are from a separate experiment and that for E 1 3 E 2 (2K) are derived from Fig. 7. K 0.5 Rb, which might be explained by a reduction of direct competition between Rb ϩ ions and protons between pH 7.5 and 8.0.) Consistent with the notion that lowering the pH below 7.5 stabilizes the E 2 (2Rb) form, the stopped-flow measurements in Fig. 9 show that the rate of E 2 (2Rb) 3 E 1 Na for the wild type is much slower at pH 6.0 compared with 7.5 (pH 7.5 as follows: k E2(Rb) 3 E1 ϭ 0.42 Ϯ 0.002; pH 6.0 and k E2(Rb) 3 E1 ϭ 0.073 Ϯ 0.001 s Ϫ1 ), quite similar to the effect of the D369A and D369N mutants at pH 7.5. In summary the effect of lowering pH from 7.5 to 5.5 on the conformational transition of the wild type is similar to the effect of neutralizing the charge on Asp 369 in the D369N mutation. Furthermore, the effect of pH is largely abolished in the D369N mutation itself.

E 1 7 E 2 Transitions of Wild Type and D369N/D369A Mutant Induced by Phosphate and Vanadate-As
shown previously, addition of inorganic phosphate/Mg ϩ or vanadate/ Mg ϩ to fluorescein-labeled renal Na ϩ ,K ϩ -ATPase in an E 1 conformation, without Na ϩ and K ϩ ions, stabilizes an E 2 conformation, detected by fluorescence quenching, and subsequent addition of Na ϩ ions reverses the fluorescence change (13,30,32). The equilibrium data in Fig. 10 and stopped-flow fluorescence data in Fig. 11 and Table 5 compare fluorescence responses to phosphate and vanadate of the wild type and mutant proteins and reveal two new phenomena. For the wild type, the fluorescence quench induced by phosphate or vanadate occurs only in the presence of Mg 2ϩ ions (Fig. 10), similar to the response seen for renal Na ϩ ,K ϩ -ATPase (13). The new finding is that for both the D369A and D369N mutants, the fluorescence quench response to phosphate is observed without added Mg 2ϩ ions, i.e. Mg 2ϩ ion is not required for the phosphate-dependent stabilization of E 2 (Fig. 10). Vanadate differs from phosphate in that the response of D369N and D369A is like the wild type protein and requires the presence of Mg 2ϩ ions (Fig. 10). Addition of Na ϩ ions at high concentrations (100 mM), after vanadate/Mg 2ϩ , partially reversed the fluorescence change of the wild type, as found previously for renal Na ϩ ,K ϩ -ATPase (30). Interestingly, the vanadate/Mg 2ϩ -induced signal change for D369A was also partially reversed by 100 mM NaCl but that for D369N was not reversed by NaCl, showing again that the FIGURE 7. Rate constants of E 1 3 E 2 (2؋) at different concentrations of the monovalent cation. A, rate constants of E 1 3 E 2 (2Rb) at 20, 30, and 83 mM RbCl for wild type (WT) and D369N/D369A mutants. The graph depicts the fitted rate constants from the data in Table 3. The values of the rate constants for D369N and D369A are virtually identical and cannot be seen as separate points. Note that the solution in syringe 1 consists of 20 -50 g of FITC-labeled enzymes, 165 mM choline chloride, plus 10 mM Hepes, pH 7.5, and the solution in syringe 2 consists of choline chloride ϩ RbCl to a total concentration of 165 mM plus 10 mM Hepes, pH 7.5. B, comparison of rate constants of E 1 3 E 2 (2ϫ) at different Rb ϩ or K ϩ concentrations for wild type. The data for K ϩ represents fitted rate constants from an experiments similar to that in Fig. 6 for a range of K ϩ concentrations from 10 to 83 mM. The solid line for K ϩ represents the hyperbolic fit with the parameters k E2(K) 3 E1 125 Ϯ 15 s Ϫ1 and the K K 39 Ϯ 10.5 mM. The data for Rb ϩ are taken from Table 3. D369N is more strongly poised toward E 2 . The concentrations of phosphate and vanadate required to induce these signal changes were in the several millimolar and tens of micromolar range, respectively, but attempts to determine accurate con-centration dependences were not satisfactory due to a slow time course or small signal size and difficulty in determining the end points after graded addition of phosphate or vanadate.
We have also looked at the rate of fluorescence quenching in the stopped-flow fluorimeter when the proteins were mixed with phosphate/Mg 2ϩ (wild type, D369N) or phosphate alone (D369A) (E 1 3 E 2 ⅐P) or vanadate/Mg 2ϩ (E 1 3 E 2 ⅐vanadate) at high concentrations that induce the full response (Fig.  11). These experiments show a second clear difference between wild type and mutants. As seen in Fig. 11, the rate of quenching by either phosphate/Mg 2ϩ or vanadate/Mg 2ϩ is very slow for the wild type protein, but a much faster component was observed for both D369N and D369A with either phosphate or vanadate. The rate constants of all the curves were fitted to double exponential functions, giving the values of k 1 and k 2 recorded in Table 5. In the case of the wild type, k 2 is very slow and may represent an artifact due to photo-bleaching of the bound fluorescein. For quenching induced by phosphate, k 1 for the D369N and D369A mutants is 25-34-fold faster than the k 1 of the wild type. It is noticeable that k for these mutants has about the same rate as the k 1 of the wild type. For quenching induced by vanadate/Mg 2ϩ , the k 1 was about 11-fold (D369A) or 141-fold (D369N) faster than the k 1 of the wild type, and again, the k 2 for the mutants has about the same rate as the k 1 of the wild type. Possible explanations of these observations are given below.

DISCUSSION
This study introduces the use of the purified fluoresceinlabeled recombinant Na ϩ ,K ϩ -ATPase and its mutants for detailed study of E 1 7 E 2 conformational changes. This represents a significant advance in functional analysis compared with experiments on Na ϩ ,K ϩ -ATPase expressed at rather low levels in mammalian cells or Xenopus oocytes or even after expression in yeast S. cerevisiae or P. pastoris membranes without purification, which permit studies such as ligand binding (ATP, ouabain), Rb ϩ occlusion, or proteolytic and Fe 2ϩ -catalyzed oxidative cleavages (20,43). The selective effects of ligands (K ϩ , Na ϩ , P i /Mg 2ϩ , vanadate/Mg 2ϩ , effects of pH) on the fluorescein fluorescence responses of the purified detergent-soluble recombinant Na ϩ ,K ϩ -ATPase are quite similar to those found previously for native membrane-bound renal Na ϩ ,K ϩ -ATPase, although the amplitudes of the responses are smaller (12-15% for wild type compared with 25-35% for renal Na ϩ ,K ϩ -ATPase). The specific effects of the ligands are diagnostic so that the basic enzymatic mechanism is preserved. In a complementary approach, fluorescence signals of the electrochromic shift dye RH421 have also been detected recently for  Table 4. At pH values lower than 7.0, a linear drift was subtracted from the fluorescence signal changes. FIGURE 9. Stopped-flow traces of E 2 (2Rb) 3 E 1 Na for wild type at pH 7. 5 and 6. Conditions were as under "Experimental Procedures," except that solutions at pH 6.0 contained 10 mM Mes. Each trace represents the average of 5-10 replicates. The rate was measured at pH 6.0 rather than at 5.5 to avoid a fluorescence drift that interferes with the measurement.

TABLE 4 Equilibrium titrations of the E 1 7 E 2 (2Rb) conformational change at different pH values
Parameters were obtained by fitting the normalized data in Fig. 8 to the Hill equation. the detergent-soluble purified recombinant Na ϩ ,K ϩ -ATPase, 4 similar to RH421 fluorescence changes for native renal Na ϩ ,K ϩ -ATPase (27). RH421 signals reflect changes in the local electrical field accompanying cation binding and dissociation and are also indicative of an intact catalytic mechanism. Tryptophan fluorescence changes of glycerol-stabilized purified detergent-soluble Ca 2ϩ -ATPase have also been described recently, as another example of the utility of the fluorescence approach (44).
Comparison of E 1 7 E 2 Conformational Transitions of Wild Type and D369N/D369A/D369E Mutants-Replacement of the negatively charged aspartate by asparagine and alanine in the D369N/D369A mutants reveals a clear-cut stabilization of the E 2 conformation (Figs. 4 -7 and Table 2). One important observation is that substitution of the aspartate with the negatively charged glutamate, D369E, led to behavior much like the wild type, as judged by the equilibrium fluorescence titrations (Fig. 4). Another mutant, T212A, which does not change the charge within the active site, behaved similarly. Thus, the observed changes in conformational transitions of the D369N/ D369A mutants can be attributed to neutralization of the negative charge of the aspartate. The present observations are consistent with the proposal (19) that D369N/D369A mutants stabilize E 2 , but the direct measurements show that the effect is much larger than inferred from effects of K ϩ ions on ATP binding. More importantly, the fluorescence measurements allow elucidation of the detailed mechanism. Stabilization of the E 2 conformation was detected by a lower K 0.5 Rb in equilibrium titrations, 3.3-fold for D369A and 6.2-fold for D369N ( Fig.  4 and Table 2). The origin of this change in apparent affinity for Rb ϩ ions is a large reduction in the rate of E 2 (2Rb) 3 E 1 Na, 13-fold for D369A and 20-fold for D369N, with little or no effect on the rate of E 1 3 E 2 (2Rb) or intrinsic Rb ϩ binding affinity (Figs. 5-7 and Table 3). Although the absolute values of the conformational equilibrium constant K C with Rb ϩ ions could not be calculated due to lack of saturation of E 1 3 E 2 (2Rb), one can calculate a lower limit value as k E1 3 E2(2Rb) 83 mM, k E2(2Rb) 3 E1Na . The values calculated from the rate constants in Table 3 are for wild type, 57.62 Ϯ 1.1; D369N, 1069 Ϯ 19.2; D369A, 948 Ϯ 30.2, i.e. 16.5-18.5-fold in favor of E 2 (2Rb) for the mutants. For the wild type, the value of K C with K ϩ ions could be calculated as k maxE1 3 E2(2K) :k E2(2K) 3 E1Na and equals K C ϭ 37.9 Ϯ 4.6 (see Table 3). 5 For the wild type, lowering the pH from 7.5 to 5.5 had a similar effect on the K 0.5 Rb in equilibrium titrations and rate of E 2 (2Rb) 3 E 1 Na as the D369N/D369A mutants ( Fig. 8 and Table 4). Furthermore, because replacing the protonatable carboxyl by the neutral carboxamide group in D369N abrogated about 75% of the effect of pH on K 0.5 Rb compared with the wild type, the inference is that the effect of reducing pH on the wild type is due largely to protonation of the carboxyl of Asp 369 and neutralization of its charge. Thus, in essence,   Table 5. the membrane Bohr effect described in Ref. 12 has the same origin as the neutralization of the charge in the D369N/ D369A mutations.
Structural Basis for Effect of D369N/D369A Mutations on E 1 7 E 2 (2Rb) Conformational Changes- Fig. 12 depicts residues in the active site of Na ϩ ,K ϩ -ATPase in the recent structures of the E 2 ⅐P(2Rb)-like conformation with bound MgF 4 2Ϫ , an analogue of bound phosphate, and Mg 2ϩ ion or depicted also without the bound MgF 4 2Ϫ and Mg 2ϩ (Fig. 12, A and B) (5,6). The conserved TGES sequence of the A domain (Fig. 12, red) comes close to the P domain (green), and in particular Glu 214 (A domain) is poised above the MgF 4 2Ϫ . As seen in the 2ZXE structure with bound MgF 4 2Ϫ and Mg 2ϩ removed (Fig. 12B, doubleheaded arrow), Glu 214 is the closest charged residue of either the A or N domain to the active site Asp 369 . Because an important characteristic of the E 2 conformation is the interaction of A-P and A-N domains, a simple hypothesis is that removal of electrostatic repulsion between the negative charges of Asp 369 and Glu 214 stabilizes the E 2 conformation.
Using Coulomb's law, we can calculate approximately the change in electrostatic repulsion between Asp 369 and Glu 214 , when the charge on Asp 369 is removed, and predict the effect on the conformational equilibrium. This follows a similar calculation, which determined the effects of neutralization of the charge on Asp 351 of Ca 2ϩ -ATPase on ATP affinity (23). The nearest approach distance between Asp 369 and Glu 214 (PDB code 2ZXE) is 6.2 Å, and in the homologous structure of Ca 2ϩ -ATPase (PDB code 1WPG), the distance between Asp 351 and Glu 183 is 6.5 Å. In the E 2 conformation of Ca 2ϩ -ATPase (PDB code 1IWO) without bound MgF 4 2Ϫ /Mg 2ϩ , the P and A domains are more separated, and the closest distance between Asp 351 and Glu 183 is 10.8 Å. For the calculation, we will assume that the distance between Asp 369 and Glu 214 in an E 2 (2Rb) conformation of Na ϩ ,K ϩ -ATPase without MgF 4 2Ϫ /Mg 2ϩ lies in the range 6.5-10 Å. The electrostatic potential between Asp 369 and Glu 214 is calculated from Coulomb's law as in Ref. 23. The range of values is 111 to 72 mV. 6 Assuming that the effect of the D369N/D369A mutations on K C is due only to removal of the electrostatic repulsion with Glu 214 , the predicted effect is calculated from the Nernst equation, 6 and the range of values is 85-to 18-fold at 6.5 or 10 Å distance, respectively. The experimental range of the 16.5-18.5-fold change in poise of K C toward E 2 , calculated from the kinetic data, falls in this predicted range with a distance between Asp 369 and Glu 214 close to 10 Å. Of course, these calculations depend on assumptions of the dielectric constant and the distance within the active site, but they certainly seem to be reasonable. Fig. 12B depicts another important point, namely that outside of the active site the A domain makes very few contacts with the P and N domains. However, two salt bridges, Arg 544 -Glu 216 (N-A) and Arg 685 -Glu 231 (P-A) can be clearly detected. The interactions of the residues in these salt bridges in different conformations provide a simple but convincing explanation of the well established effect of ATP to accelerate the rate of E 2 (2K) 3 E 1 Na in wild type Na ϩ ,K ϩ -ATPase and, as a consequence, the effects of D369N/D369A on the rate of E 2 (2K) 3 E 1 Na. Comparison of the Na ϩ ,K ϩ -ATPase structure in the E 2 (2Rb)⅐P conformation with the Ca 2ϩ -ATPase structure with a bound ATP analogue in an E 1 -AMPPCP conformation (Fig.  12C) (45) shows that the homologous arginines of Ca 2ϩ -ATPase, Arg 678 (Arg 685 ), and Arg 560 (Arg 544 ), directly bind the AMPPCP. Mutations of Arg 544 in the Na ϩ ,K ϩ -ATPase have also shown that this residue is crucial for high affinity binding of 6 The electrostatic potential ⌬ is calculated from the relationship ⌬ Ϸ e⅐k/ D⅐r, where e is the electronic unit charge (1.6.10 Ϫ19 C); k is the Coulomb constant (9⅐10 9 N⅐m 2 ⅐C Ϫ2 ); D is the dielectric constant in the site, assumed to have a value of 20; and r is the distance (in meters). The Nernst equation: lnK C⌬ ϭ 0 ϭ lnK C ϩ (ZF/RT)⌬Ϫ or logK C,⌬ ϭ 0 /K C ϭ ⌬/57 (in mV).  ATP and ADP (46), and Arg 678 in Ca 2ϩ -ATPase was shown to be cross-linked to the N domain by glutaraldehyde, and this is disturbed by ATP binding (47).Thus it is clear that in E 2 (2K) binding of ATP to Arg 544 and Arg 685 will compete with Glu 216 and Glu 231 of the A domain for these residues. In other words, ATP and the A domain compete for the interaction with the P and N domains, and attraction of ATP for Arg 544 and Arg 685 and repulsion of Glu 216 and Glu 231 will destabilize the interaction of the N and P domains with the A domain and stabilize the ATP-mediated interaction of the N to P domain. The result of this competition is the greatly accelerated rate of E 2 (2K) 3 E 1 Na by ATP bound to E 2 (2K) with a low affinity and stabilization of E 1 by ATP bound with high affinity. The explanation of this effect of ATP implies that the slow rate of E 2 (2Rb) 3 E 1 Na, without ATP, is limited, at least partially, by the high activation energy required to break these electrostatic interactions.
Because the total strength of N-A and P-A domain interactions represents the balance of attractive and repulsive forces, one can presume that removal of the Asp 369 -Glu 214 repulsion in the D369N/D369A mutants allows the A domain to approach the N and P domains more closely and increase their interaction energy via the Arg 544 -Glu 216 and Arg 685 -Glu 231 proximities. This will, in turn, increase further the activation energy required to break them and slow the rate of E 2 (2Rb) 3 E 1 Na compared with the wild type. Fig. 12C also shows that Arg 560 and Arg 678 interact with the oxygens on the ␤-phosphate and on the sugar moiety of the ATP analogue, respectively. Therefore, another implication is that ADP should be as effective as ATP in repulsing the A domain and accelerating the E 2 (2K) 3 E 1 Na transition. Acceleration of E 2 (2K) 3 E 1 Na by ADP, as for ATP, has indeed been inferred from effects of ADP on K ϩ -K ϩ exchange reactions (48), proteolytic digestion (10), and demonstrated directly using an iodoacetamidofluorescein-labeled Na ϩ ,K ϩ -ATPase (49). Finally, it is of interest that for Ca 2ϩ -ATPase, although the two arginines (Arg 560 and Arg 678 ) are conserved, the residue at position Glu 216 of Na ϩ ,K ϩ -ATPase is Val 185 in Ca 2ϩ -ATPase, and the residue at position Glu 231 is Asp 203 in Ca 2ϩ -ATPase. Thus, only one of the salt bridges can be formed in the E 2 P or E 2 conformations. Although it has been known for many years that ATP activates the E 2 3 E 1 transition of Ca 2ϩ -ATPase (50), ATP accelerates the rate of E 2 7 E 1 in Ca 2ϩ -ATPase much less (20-fold at pH 6.0 and less at higher pH (51)) than it accelerates the rate of E 2 (2K) 3 E 1 Na of Na ϩ ,K ϩ -ATPase (100 -150-fold) (11,49,52). The lack of effect of the D369N and D369A mutants on the rate of E 1 3 E 2 (2Rb) indicates that, in these conditions, the limiting step is not affected by neutralization of the charge on Asp 369 but is the same as in the wild type. The result in Fig. 7B shows that the rate of this transition depends on the nature of the bound cation and that with K ϩ being considerably faster than that with bound Rb ϩ . Thus, the experiment indicates that the limiting process is the movement of trans-membrane segments with occluded K ϩ or Rb ϩ ions. The rate of these movements should be the same in wild type and D369N/D369A mutants.
Effects of Phosphate and Vanadate-The finding (Fig. 10) that phosphate induced the E 1 3 E 2 ⅐P transition in the D369N/ D369A mutants even without added Mg 2ϩ ions, by contrast with the wild type, can be understood by reference to the structure in Fig. 12A. The Mg 2ϩ ion interacts with the MgF 4 2Ϫ , with the carboxyl groups of Asp 369 and Asp 710 , with the carbonyl of Thr 371 , and via a water molecule with Gly 213 . Binding of Mg 2ϩ ions to Asp 710 has also been inferred by mutation work (53 2Ϫ and reduce the charge repulsion with Glu 214 . The structure also clarifies two related previous findings. First, D369A binds ouabain weakly even without added Mg 2ϩ ions, although binding to wild type is absolutely dependent on Mg 2ϩ ions (19). Ouabain binding stabilizes the E 2 conformation, and again, one would infer that Mg 2ϩ is required in the wild type, but not in D369A, to allow Glu 214 (A domain) to approach Asp 369 (P domain). Second, Fe 2ϩ -dependent oxidative cleavages in the active site, characteristic of the E 2 conformation, are largely suppressed in the D369N mutant (20). Because Fe 2ϩ substitutes for Mg 2ϩ ions in catalyzing phosphate transfer, the result is consistent with suboptimal binding of Fe 2ϩ to the D369N mutant. It is interesting that the D369N/D369A mutants behave differently with vanadate, compared with phosphate, in that Mg 2ϩ ions are still required to stabilize the E 2 conformation. Vanadate is a phosphate transition state analogue, binds much more tightly than phosphate, and could be expected to show different interactions. Indeed, Asp 710 , which is a Mg 2ϩ -ligating residue, has been shown to be much more important for vanadate compared with phosphate binding in E 2 (53).
The scheme in Fig. 13 can explain simply the finding that, for the D369N/D369A mutants, the rate of conversion of E 1 3 E 2 ⅐P or E 1 3 E 2 ⅐vanadate showed a fast component, by contrast to the slow rate of conversion for the wild type ( Fig. 11 and Table  5). Both phosphate and vanadate are assumed to bind much more tightly to E 2 than to E 1 . By the left-hand path, E 2 ⅐P or E 2 ⅐vanadate are stabilized by preferential binding of phosphate or vanadate to E 2 and prevention of the back reaction, E 2 3 E 1 . The slow observed rate of fluorescence change is presumably limited by the E 1 3 E 2 transition without bound phosphate or vanadate. This is assumed to be the path for the wild type. By contrast, the right-hand path may apply to the D369N/D369A mutants. Although phosphate and vanadate bind weakly to E 1 , neutralization of the charge on the Asp 369 allows a fast confor- Asp 369 of Na ؉ ,K ؉ -ATPase and Conformational Changes NOVEMBER 6, 2009 • VOLUME 284 • NUMBER 45 JOURNAL OF BIOLOGICAL CHEMISTRY 31049 mational change from E 1 ⅐P 3 E 2 3 P or E 1 ⅐vanadate 3 E 2 ⅐vanadate, i.e. bound phosphate or vanadate induce the fast component observed in Fig. 11. Again, it is interesting that the effect with D369N is stronger than with D369A, at least with vanadate. The slow component may correspond to the fraction of molecules, which reach the E 2 ⅐P or E 2 ⅐vanadate conformation by the slow left-hand route E 1 3 E 2 3 E 2 ⅐P.
Role of Electrostatic Interactions of Asp 369 in the Normal Catalytic Cycle-Bound Mg 2ϩ ions are already known to play a crucial role in reducing negative charge repulsion, between the ␥-phosphate of bound ATP and the aspartate, enabling close proximity necessary for phosphorylation. D369N/D369A mutants of Na ϩ ,K ϩ -ATPase and D351A/D351N/D351T mutants of Ca 2ϩ -ATPase have a very high ATP affinity, which is the result of removal of electrostatic repulsion between the ␥-phosphate and the aspartate (19,22,23). In the wild type, binding of Mg 2ϩ ions to the Asp 369 (Na ϩ ,K ϩ -ATPase) or Asp 351 (Ca 2ϩ -ATPase) and the ␥-phosphate reduces this repulsion. Indeed, in the case of Ca 2ϩ -ATPase ATP binds only weakly in the absence of Mg 2ϩ ions (22).
We propose here an additional role of charge interactions of Asp 369 in the normal catalytic cycle, namely as triggers of E 1 P(3Na) 3 E 2 P and E 2 (2K) 3 E 1 Na conformational transitions. As discussed in the Introduction, dissociation of ADP from E 1 P(3Na) is necessary but may not itself be sufficient to trigger the spontaneous conformational change to E 2 P. The lack of requirement of Mg 2ϩ ions for the phosphate-induced conformational change (Fig. 10), and the fast phosphate-or vanadate-induced E 1 3 E 2 ⅐P or E 1 3 E 2 ⅐vanadate transitions, observed for the D369N/D369A mutants (Fig. 11), suggest that the behavior of the mutants resembles that of normal wild type E 1 P(3Na) with bound Mg 2ϩ ions. Thus E 1 P(3Na) 3 E 2 P may be triggered by neutralization of the charge on Asp 369 resulting from binding the divalent cation, Mg 2ϩ , together with the covalently bound phosphate (which is a mixture of -C-O-PO 3 2Ϫ and -C-O-POH 3 Ϫ at physiological pH). The interconversion between conformational states has been described as "fairly stochastic," i.e. one in which the energy barriers between states are comparable with the thermal energy (2,3). In the stochastic process of approach, binding, and disengagement of A to and from P and N domains, the lack of repulsion between Asp 369phosphate/Mg 2ϩ and Glu 214 , as well as attraction of the bound Mg 2ϩ ion to the A domain via Gly 213 and a water molecule, should favor approach of the A domain to the P and N domains. E 2 P, with Mg 2ϩ ions tightly bound, may then be stabilized via the Arg 685 -Glu 231 and Arg 544 -Glu 216 salt bridges between A and P and A and N domains, respectively. After dephosphorylation of E 2 P, phosphate/Mg 2ϩ dissociates, and the negative charge on Asp 369 should favor disengagement of the N-A and P-A domains in E 2 (2K) due to repulsion of the Glu 214 in the A domain. Dissociation of A from P and N domains in E 2 (2K) and transition to E 1 is then strongly accelerated by ATP, which breaks the Arg 685 -Gly 231 and Arg 544 -Glu 216 salt bridges, as discussed above.