Changes in Steady-state Conformational Equilibrium Resulting from Cytoplasmic Mutations of the Na,K-ATPase α-Subunit*

Mutations comprising either deletion of 32 amino acids from the NH2 terminus (α1M32) or a Glu233 → Lys substitution in the first M2-M3 cytoplasmic loop (E233K) of the α1-subunit of the Na,K-ATPase result in a shift in the steady-state E 1 ↔E 2 conformational equilibrium towardE 1 form(s). In the present study, the functional consequences of both NH2-terminal deletion and Glu233 substitution provide evidence for mutual interactions of these cytoplasmic regions. Following transfection and selection of HeLa cells expressing the ouabain-resistant α1M32E233K double mutant, growth was markedly reduced unless the K+concentration in the culture medium was increased to at least 10 mm. Marked changes effected by this double mutation included 1) a 15-fold reduction in catalytic turnover (V max/EPmax), 2) a 70-fold increase in apparent affinity for ATP, 3) a marked decrease in vanadate sensitivity, and 4) marked (≈10-fold) K+activation of the Na-ATPase activity measured at micromolar ATP under which condition the E 2(K) → →E 1 pathway is normally (α1) rate-limiting and K+ is inhibitory. The decrease in catalytic turnover was associated with a 5-fold decrease in V max and a compensatory ≈3-fold increase in expressed α1M32E233K protein. In contrast to the behavior of either α1M32 or E233K, α1M32E233K also showed alterations in apparent cation affinities.K′Na was decreased ≈2-fold andK′K was increased ≈2-fold. The importance of the charge at residue 233 is underscored by the consequences of single and double mutations comprising either a conservative change (E233D) or neutral substitution (E233Q). Thus, whereas mutation to a positively charged residue (E233K) causes a drastic change in enzymatic behavior, a conservative change causes only a minor change and the neutral substitution, an intermediate effect. Overall, the combined effects of the NH2-terminal deletion and the Glu233substitutions are synergistic rather than additive, consistent with an interaction between the NH2-terminal region, the first cytoplasmic loop, and possibly the large M4-M5 cytoplasmic loop bearing the nucleotide binding and phosphorylation sites.

The Na,K-ATPase couples the hydrolysis of one ATP molecule to the translocation of 3 Na ϩ and 2 K ϩ ions against their electrochemical gradients, thus maintaining the normally high K ϩ and low Na ϩ concentrations inside animal cells. This enzyme complex comprises a large subunit, ␣ (molecular mass, 112 kDa) and a small subunit, ␤ (molecular mass, 35 kDa). ␣ and ␤ have been cloned and sequenced from a variety of tissues (see Ref. 1). The functional unit may be a heterodimer (␣␤) 2 , although a monomeric ␣␤ unit can occlude both Na ϩ and (K ϩ )Rb ϩ , consistent with its being the minimal unit required for transport (2). ␣ is the catalytic subunit, which spans the membrane probably 10 times and includes the cytoplasmic catalytic domain and the extracellular cardiac glycoside binding site(s) (3). Although this enzyme complex has eluded efforts to obtain ordered three-dimensional crystals of sufficient quality to allow precise description of key structural features such as topology and cation ligating structure(s), protein chemical and molecular biology techniques are providing important information about structure/function relationships (for review, see Ref. 4).
The reaction mechanism is probably consecutive (Cleland's ping-pong mechanism), whereby Na ϩ is released before K ϩ binds (5,6). Most reaction schemes are based on the original Albers-Post mechanism involving phosphorylated and dephosphorylated forms of the enzyme, both of which undergo conformational transitions (E 1 P 3 E 2 P and E 2 3 E 1 ), which are coupled to ion translocation steps. The enzyme (and phosphoenzyme) include conformations with ion binding sites accessible to the cytosol, to the extracellular milieu and in a state in which the transported ion is occluded. ATP interacts not only with high affinity to catalyze phosphorylation of E 1 (ATP ϩ E 1 3 E 1 P), but also with low affinity in a manner which effects release of occluded K ϩ from the form E 2 (K), i.e. E 2 (K) ϩ ATP 3 Although considerable information has been obtained regarding transmembrane-located residues involved in cation binding and occlusion, the structural basis for conformational coupling of the scalar energy of ATP hydrolysis to the vectorial movement of Na ϩ and K ϩ remains a major unresolved issue. Previous site-specific alterations of cytoplasmic residues within several P-type ATPases, particularly those of the sarcoplasmic reticulum Ca-ATPase, have identified residues in both the M2-M3 and larger catalytic M4-M5 cytoplasmic loops, which support a model in which the E 1 P 3 E 2 P conformational transition is transmitted from the phosphorylation site to the cation binding pocket via the ␤-strand structure in M2-M3 and a "stalk" region connecting M4 to the catalytic loop (for review, see Ref. 7). It is pertinent that mutation of Leu 332 at the putative boundary of M4 and the stalk region connecting M4 to the M4-M5 loop affects conformation coupling as evidenced in the shift in equilibrium in favor of E 1 P (8).
Another cytoplasmic region involved in conformational cou-pling is the amino terminus as first reported in studies of specific proteolytic cleavage of the Na,K-ATPase by Jørgensen (9). In later studies, we showed that the consequence of deletion of residues 1-32 (mutant ␣1M32), as well as the nonconservative mutation Glu 233 3 Lys (E233K) in the M2-M3 cytoplasmic loop, is the displacement of the E 1 -E 2 conformational equilibrium in favor of E 1 . The pertinent kinetic changes included a decrease in catalytic turnover, a higher apparent affinity for ATP at the low-affinity binding site, and a faster rate of formation of E 1 from E 2 (K) (Refs. 10 and 11). A noteworthy feature of the E233K mutant is that Glu 233 is in the ␤-strand region of M2-M3 previously identified as having a role in the E 1 P 3 E 2 P conformational step(s). Based on the premise that conformation-dependent interactions may occur between the NH 2 terminus, the ␤-strand in the M2-M3 loop, and the phosphorylation site in the M4-M5 loop of the ␣1 enzyme, we have expressed double mutants comprising both the ␣1M32 deletion and substitution of Glu 233 . We show that the magnitudes of the kinetic changes resulting from the double mutation give strong support to the existence of mutual interactions between these regions.

EXPERIMENTAL PROCEDURES
Mutagenesis, Transfection, and Cell Culture-HeLa cell lines expressing pRc/CMV (Invitrogen) constructs of rat ␣1, ␣1M32, and E233K are the same as those described previously (11,12). The E233D and E233Q mutations were each prepared in the SalI(875)-BamHI(1780) cassette of rat ␣1 in M13mp18, using the site-specific, oligonucleotidedirected mutagenesis technique of Kunkel (13) as described previously (14). Mutant cassettes were completely sequenced to verify the presence of the mutation and the absence of any unplanned substitutions. The mutated SalI-BamHI restriction fragments containing each of the substitutions for E233 were then excised, gel-purified, and ligated into either wild type rat ␣1 or rat ␣1M32 cDNA in a modified pIBI30 shuttle vector, in place of the wild type SalI-BamHI cassette. After verifying the presence of mutant sequences and the sequences at the SalI and BamHI sites, the full-length mutant cDNAs were excised from the shuttle vector with HindIII and ligated into pRc/CMV (Invitrogen). Orientation of the ␣1 cDNAs were determined by restriction analysis.
Membrane Preparation-NaI-treated microsomal membranes were prepared from the mutant cells as earlier described (14,16), and the protein concentration was determined with a detergent-modified Lowry assay (17).
Growth Curves-The cells were grown in 24-well plates and at the times indicated in the text, the cells were trypsinized for 10 min at 37°C, after which the cell counts were performed in a hematocytometer.
Enzyme Assays-Assays of Na,K-ATPase activity were carried out as described earlier by measuring the release of 32 P i from [␥-32 P]ATP (18). Unless indicated otherwise, the membranes were preincubated for 10 min at 37°C with all the reactants except ATP. The reaction was initiated by adding 0.2 volumes of ATP to 0.8 volumes of preincubated membranes. Final concentrations were 1 mM ATP, 3 mM MgSO 4 , 20 mM Tris-HCl (pH 7.4), 5 mM EGTA (pH 7.4), and 10 M ouabain (Sigma). A final concentration of 10 mM ouabain was used to determine the baseline hydrolysis activity. The cation affinities were determined with 100 mM NaCl and varying KCl concentrations for KЈ K and 20 mM KCl and varying NaCl concentrations for KЈ Na . Appropriate amounts of choline chloride were added to keep the ionic strength constant. To optimize the proportion of specific [␥-32 P]ATP hydrolysis in the cation affinity assays, the concentration of ATP in the assay mix for the ␣1M32E233K mutant was lowered from 1 to 0.1 mM, which saturates the enzyme. K ϩ (Rb ϩ ) fluxes were performed as earlier described (12). Vanadate (orthovanadate from Fisher) was made up freshly before the experiment and was added at the concentrations indicated. The final Na ϩ and K ϩ concentrations were 100 and 10 mM, respectively. The Na-ATPase activities were measured with 1 mM MgSO 4 , 20 mM histidine (pH 7.4), 5 mM EGTA (pH 7.4), 20 mM NaCl, 1 M ATP, 10 M ouabain and KCl concentrations as indicated. For base-line activities, Na ϩ was omitted and 20 mM KCl included, with choline chloride added to maintain a constant (40 mM) chloride concentration.
Phosphoenzyme was determined as described earlier (10). Membranes were treated with ouabain (20 M ouabain, 0.4 mM MgSO 4 , and 10 mM Tris-glycylglycine (pH 7.4)) and preincubated for 10 min at 37°C. The suspension was then treated with oligomycin (0.2 mg/ml) as described elsewhere (10). Phosphorylation was carried out in a total volume of 200 l at 0°C for 10 s with 100 mM NaCl, 10 mM Trisglycylglycine (pH 7.4), 5 mM EGTA (pH 7.4), 1 mM MgSO 4 , and 1 M [␥-32 P]ATP (specific activity, 10,000 -20,000 cpm/pmol). For the baseline activity, 100 mM NaCl was replaced by 50 mM KCl and 50 mM choline chloride and oligomycin omitted (replaced with vehicle (ethanol)). The reaction was stopped by adding 800 l of ice-cold 5% trichloroacetic acid containing 5 mM ATP and 2.5 mM NaH 2 PO 4 , after which the precipitated protein was collected, and the radioactivity was assessed as described previously (18).
Values shown are the averages Ϯ S.D. Unless indicated otherwise, all experiments were carried out in triplicates, and a representative of at least three separate experiments is shown. Experiments were carried out with two different clones, except for ␣1M32E233D, in which case the results are from at least two different membrane preparations of the same clone.

RESULTS
As in earlier studies with HeLa cells expressing ouabainresistant rat ␣1 Na,K-ATPase, we used a culture medium containing 1 M ouabain to select stable HeLa cell lines expressing the mutant ␣1 enzymes. In most cases, the functional alterations caused by the mutations did not preclude this selection procedure and the stable expression of the mutants enabled us to characterize their functional properties. An exception was the double mutant ␣1M32E233K, which failed to yield any ouabain-resistant HeLa cells in normal Dulbecco's modified Eagle's medium following numerous transfections. On the assumption that the combined effects of the NH 2 -terminal 1-32 residue deletion and the E233K substitution might be additive or even synergistic, and that the resultant secondary decrease in catalytic turnover might prevent survival of the ␣1M32E233K-transfected cells, we attempted to promote growth by raising extracellular K ϩ as done previously by Arguello and Lingrel in their studies with mutant S775A defective in cation ligation (19).
When HeLa cells transfected with ␣1M32E233K were selected with 1 M ouabain in Dulbecco's modified Eagle's medium containing 20 mM KCl, a few ouabain-resistant colonies were obtained. In the experiment shown in Fig. 1, cells cultured in 20 mM K ϩ were either maintained in that medium or transferred into normal medium with 5.4 mM K ϩ . Growth rates were followed by counting the cells for periods up to 10 days. As shown, the doubling time for the wild type ␣1-transfected cells was 16.4 Ϯ 0.4 h, whereas that for the ␣1M32E233K mutant growing in 5.4 mM K ϩ was 61.5 Ϯ 7.5 h. Adjustment of the K ϩ concentration to 20 mM maintained the growth rate close to normal (23.5 Ϯ 0.5 h). Another experiment (not shown) indicated that 10 mM K ϩ was sufficient to restore the growth rate to that of the cells grown in 20 mM K ϩ . As described below and consistent with our original premise, elevated K ϩ concentration compensated for the markedly reduced pump turnover and the lower apparent affinity for K ϩ observed in kinetic analysis of ␣1and ␣1M32E233K-transfected cells as described below.
Ligand Interactions-At micromolar ATP concentrations sufficient to saturate the high-affinity phosphorylation site, the response of Na ϩ -dependent ATP hydrolysis to varying concentrations of K ϩ is a convenient and sensitive indication of mutant-specific differences in the E 2 (K) 3 3 E 1 pathway of the Na,K-ATPase reaction (10,11). This part of the reaction becomes rate-limiting at low ATP concentration, and as shown first by Post et al. (20), K ϩ inhibits Na-ATPase activity of the ␣1 enzyme. In contrast, K ϩ activates Na-ATPase of ␣1M32 and E233K mutants, consistent with a rapid deocclusion via E 2 (K) 3 E 1 K 3 E 1 ϩ K ϩ in these mutant enzymes (10,11).
In the experiments shown in Fig. 2, we compared the K ϩ activation profiles of the ␣1M32E233K mutant and the single mutants, ␣1M32 and E233K. The maximal stimulations by K ϩ are 1000, 135, and 200%, respectively, suggesting synergism and hence mutual interactions of the NH 2 -terminal and M2-M3 domains. Fig. 2 also shows the K ϩ response profiles of mutants in which substitution of Glu 233 is either conservative (aspartate), neutral (glutamine), or positive (lysine), both in the case of single mutants (compare E233D, E233Q, and E233K in Fig. 2A) and double mutants (compare ␣1M32E233D, ␣1M32E233Q, and ␣1M32E233K in Fig. 2B). As shown, the magnitude of K ϩ activation is dependent on the charge of the residue, being minimal with E233D and moderate with E233Q compared with E233K and similarly, minimal with ␣1M32E233D and moderate with ␣1M32E233Q compared with ␣1M32E233K.
The notion that the distinct K ϩ activations signal differences in the E 1 -E 2 conformational equilibrium is underscored by the alterations in ATP dependence of the Na,K-ATPase activity shown in Fig. 3. As in previous studies (10) a model describing a branched pathway of the E 2 (K) deocclusion limb of the Albers-Post mechanism is relevant. In one branch of the pathway, low-affinity ATP binding to the K ϩ -occluded enzyme, E 2 (K), is followed by rapid deocclusion (ATP⅐E 2 (K) 3 ATP⅐E 1 ⅐K 3 ATP⅐E 1 ϩ K ϩ ), and in the other branch, slow release of K ϩ from E 2 (K) via the sequence E 2 (K) 3 E 1 K 3 E 1 ϩ K ϩ is followed by high-affinity ATP binding to E 1 . A comparison of the results shown in Fig. 2 with those of Fig. 3 and Table I indicate that the magnitudes of K ϩ stimulation of the mutant enzymes at low ATP is directly related to their increased apparent affinities for ATP.
The experiments shown in Fig. 3 and Table I include previous results with ␣1, the ␣1M32 truncated mutant, and E233K. As reported previously, both the wild type ␣1 and truncated ␣1M32 enzymes have low (KЈ ATP(L) ) and high (KЈ ATP(H) ) apparent affinities for ATP (10). For ␣1 the values are 331 Ϯ 44 M notable is the high ATP affinity of the ␣1M32E233K mutant, the value obtained being virtually the same as KЈ ATP(H) of the wild type ␣1 enzyme. To date, a similar remarkable change in apparent affinity for ATP has not been observed with other functional mutants. These data and those for the single mutants of Glu 233 also provide insight into the nature of the interaction of the ␤-strand of M2-M3 encompassing the charged Glu 233 with another region, most likely the M4-M5 catalytic domain as discussed below.
Based on the Albers-Post model and discussed in detail by Eisner and Richards (21), a change in apparent affinity for ATP should alter the apparent affinity for K ϩ and vice versa. To determine whether the striking increase in ATP binding to ␣1M32E233K alters apparent cation affinities, KЈ K and KЈ Na were determined from the plots shown in Fig. 4, A and B. With the data fitted to a noncooperative model, the apparent affinity of ␣1M32E233K for K ϩ was 2.4-fold lower than that of ␣1. The KЈ K for the double mutant was 1.2 Ϯ 0.36 mM compared with 0.50 Ϯ 0.11 mM for ␣1. A similar difference in affinity for extracellular K ϩ activation of ( 86 Rb ϩ )K ϩ influx was observed in transport experiments performed with intact ␣1and ␣1M32E233K-transfected cells (not shown). A change in KЈ Na (2.2-fold) was also observed, but in the opposite direction, with values of 1.00 Ϯ 0.08 mM and 2.19 Ϯ 0.20 mM for ␣1M32E233K and ␣1, respectively. This increase in apparent affinity for Na ϩ and decrease in apparent affinity for K ϩ support the idea that the conformational equilibrium of ␣1M32E233K favors the high-affinity Na ϩ binding conformation, E 1 .
Vanadate Sensitivity-To gain further insight into the effect of the ␣1M32E233K mutation on the steady-state E 1 -E 2 conformational equilibrium, we used vanadate as a conformational probe. Vanadate has the ability to exist in a stable trigonal bipyramidal structure and may, as such, act as a transition state analog of inorganic orthophosphate (22). It competes with P i in binding to the E 2 conformation of P-type ATPases and is able to form a stable intermediate whereby the pump activity is inhibited (23). As shown in Fig. 5, the ␣1M32 and E233K mutants are at least 100-fold less sensitive to vanadate inhibition than ␣1. ␣1M32E233K is not inhibited by vanadate present up to at least 10 mM, at which concentration a paradoxical increase in activity was consistently observed. Although we have no explanation for this stimulation, it is plausible that it represents an effect due to a minor contaminant, possibly a form of vanadate other than orthovanadate, and that the stimulation is masked in other less resistant mutants by the inhibitory effects of orthovanadate.
Catalytic Turnover of Single and Double Mutations: Effect of Charge of Residue 233-We showed previously that the ␣1M32 and E233K mutants as well as the ␣2 isoform have reduced catalytic turnovers, which probably reflects the poise in the steady-state E 1 -E 2 equilibrium in favor of E 1. Catalytic turnover was determined as the ratio of V max under optimal conditions of ATP, Na ϩ , and K ϩ , to the maximal level of phosphoenzyme formed in the presence of Na ϩ , with oligomycin added to  trap the enzyme as Na 3 ⅐E 1 P (24,25). Compared with ␣1, the catalytic turnovers of ␣2, ␣1M32, and E233K were reduced approximately 50%.
In the present study, we extended the estimates of turnover to double mutants as well as single mutants in which Glu 233 was replaced with aspartate or glutamine as well as lysine. As shown in Fig. 6, the turnover effected by the double mutation, ␣1M32E233K, is 5% that of the wild type ␣1 enzyme. This reduction in catalytic site activity is far greater than predicted by the combined effects of the single mutations, which should be no greater than the product of the fractional reductions effected by the single mutations, namely 0.45 ϫ 0.55. The results also show that the extent of alteration (decrease) in turnover is dependent on the charge of residue 233.
Interestingly, comparison of phosphoenzyme formed in the absence and presence of oligomycin indicates an oligomycin-dependent increase (2-fold) in total phosphoenzyme of ␣1, which diminishes in the various mutants as a function of the change in charge at residue 233, deletion of the NH 2 terminus, and combinations of the two alterations (Fig. 6, inset). Thus, the FIG. 4. Effect of the mutation ␣1M32E233K on apparent cation affinities. ATP hydrolysis was assayed with varying cation concentrations as described under "Experimental Procedures." The data are presented as Na,K-ATPase activity as percent of V max . A, apparent affinity for K ϩ . When fitted to a noncooperative model, the V max (nmol/ (mg ϫ min)) values were estimated to be 124 and 45 for ␣1 and ␣1M32E233K, respectively. The KЈ K (mM) values were 0.50 Ϯ 0.11 for ␣1 and 1.22 Ϯ 0.36 for ␣1M32E233K. q, ␣1; E, ␣1M32E233K. B, apparent affinity for Na ϩ . When fitted to a noncooperative model, the V max (nmol/(mg ϫ min)) values were estimated to be 83 and 25 for ␣1 and ␣1M32E233K, respectively. The KЈ Na (mM) values were 2.19 Ϯ 0.20 for ␣1 and 1.00 Ϯ 0.08 for ␣1M32E233K. Legends are the same as in A. graded decrease in turnover and in oligomycin stimulation parallels the putative shifts in steady-state E 1 -E 2 conformational equilibrium toward E 1 form(s). DISCUSSION We showed previously that changes caused by mutation of glutamate 233 to lysine in the catalytic ␣1 subunit of Na,K-ATPase alters the equilibrium between major conformational states of the dephospho-and phosphoenzyme forms during steady-state catalysis in favor of E 1 forms. A generally similar conformational shift was observed when residues 1-32 were removed from the cytoplasmic amino terminus of ␣1 (9,26,12). Evidence for the conformational shift was derived from kinetic studies showing that these mutations cause (i) an increase in the E 2 (K) 7 E 1 ϩ K ϩ equilibrium in favor of E 1 , accountable for by an increase in the rate of formation of E 1 from E 2 (K), (ii) an increase in apparent affinity for ATP at the step E 2 (K) ϩ ATP 3 ATP⅐E 1 ϩ K ϩ when the overall reaction is measured at 37°C, (iii) less oligomycin-dependent increase in steady-state level of phosphoenzyme, which indicates a preponderance of E 1 P versus E 2 P form(s) in the mutants (cf. Fig. 6), and (iv) a decreased catalytic turnover. The decrease in sensitivity to vanadate is also diagnostic of a decrease in steady-state level of E 2 required for P i (vanadate) binding. Although mutation of amino acids (27) and tryptic cleavage (28,29) in this ␤-strand region of the Ca-ATPase of the sarcoplasmic reticulum (reviewed by Møller et al. (4)) and tryptic cleavage of the Na,K-ATPase (30) block the E 1 P to E 2 P conversion (for review, see Ref. 31), mutations in this region of yeast proton pumps, like the Glu 233 3 Lys mutation in Na,K-ATPase, have resulted in active enzyme which can cycle through the entire reaction, but with reduced catalytic turnover and decreased sensitivity to vanadate (32-34).
In the present study, the extent of the changes in kinetic behavior of the double mutant in which NH 2 -terminal deletion and Glu 233 3 Lys were combined is remarkable. The magnitudes of the changes in the various kinetic parameters of ␣1M32E233K are far greater than predicted from "additive" effects of the two mutations. This synergy was apparent in the K ϩ sensitivity profiles of Na-ATPase, the apparent affinities for ATP as well as the overall catalytic site turnover. As shown earlier, the response of Na-ATPase to K ϩ at micromolar ATP reflects either (i) the relative rate of the formation of E 1 from E 2 (K) via the high-affinity branch (E 2 (K) 3 E 1 ⅐K 3 E 1 ) versus the low-affinity branch (E 2 (K) ϩ ATP 3 ATP⅐E 1 ⅐K 3 ATP⅐E 1 ) of the reaction pathway and/or (ii) the relative ATP affinity for deocclusion via the low-affinity pathway.
Presumably, the extreme poise in conformational equilibrium of ␣1M32E233K in favor of E 1 precludes binding of ATP with low affinity during the catalytic cycle, to the extent that maximal activity is observed with ATP present at a concentration sufficient to saturate only the high-affinity site. This indicates that the enzyme can cycle through all the intermediates of the reaction cycle, without invoking low-affinity ATP binding, giving credence to the notion of a single physical site (35). Most likely, the E 1 P to E 2 P phase of the reaction cycle becomes rate-limiting to the extent that the overall catalytic turnover is reduced 15-fold. In contrast, turnover was decreased only Ϸ50% by each of the two mutations alone. In fact, the very slow turnover of ␣1M32E233K, though offset partly by a 3-4-fold increase in functional enzyme as measured by the maximal amount of phosphoenzyme, compromised the survival and growth of ␣1M32E233K-transfected cells in medium with regular K ϩ concentration. It is also clear that this mutant is remarkably insensitive to vanadate.
The concept that conformational coupling in P-type ATPases occurs via interaction of the ␤-strand in the M2-M3 loop with the catalytic domain, as envisioned previously by Green and Stokes (36), was based largely on studies of the Ca-ATPase. Involvement of the M2-M3 loop of Na,K-ATPase in structural rearrangements associated with ligand binding and phosphorylation has also been apparent in distinctive conformational changes revealed by proteolytic cleavage patterns (37). Whereas evidence for M2-M3/M4-M5 loop interactions was weak using a yeast two-hybrid system (38), conclusive evidence was obtained in recent studies of specific iron-catalyzed cleavages of the ␣-subunit (39). These interactions appear to involve the ␤-strand of M2-M3, and regions near the phosphorylation site and putative hinge region of the M4-M5 domain, but not the putative ATP binding site. Furthermore, the distinct cleavages in the presence of Na ϩ versus K ϩ indicate conformationdependent interactions whereby the two loops interact when the enzyme is in the E 2 form and come apart in the E 1 conformation. The behavior of the ␣1M32E233K mutant indicates that the NH 2 terminus of the Na,K-ATPase impacts interactions between the M2-M3 and M3-M4 loops. Although the structural basis for this effect remains unknown, it may be implied that the NH 2 terminus favors the interaction of the two cytoplasmic loops, because truncation shifts the equilibrium toward E 1 .
The kinetic alterations effected by the various mutations of residue 233 indicate that the shift is minimal with the conservative substitution of glutamate 233 by aspartate and moderate with substitution by glutamine compared with the nonconservative substitution by the positively charged lysine residue. This was evident with both the series of single mutants, E233D, E233Q, and E233K, as well as the double mutants, ␣1M32E233D, ␣1M32E233Q, and ␣1M32E233K. Although the sites of interaction between the NH 2 terminus and the M2-M3 and/or M4-M5 loops remain to be determined, it is likely that salt-bridge interactions are involved as suggested earlier (31, 40). Presumably such salt-bridge or ionic interactions occur close to the conformational-dependent specific iron-catalyzed cleavages. It may be relevant that an analysis of the behavior of a 1-27 deletion mutant of ␣1 1 indicates that the NH 2 -terminal region critical in modulating the E 1 -E 2 conformational equilibrium is the highly charged sequence 27 MDELKK 32 .