Structural Basis for α1 Versus α2 Isoform-distinct Behavior of the Na,K-ATPase

We showed earlier that the kinetic behavior of the α2 isoform of the Na,K-ATPase differs from the ubiquitous α1 isoform primarily by a shift in the steady-stateE 1/E 2 equilibrium of α2 in favor of E 1 form(s). The aim of the present study was to identify regions of the α chain that confer the α1/α2 distinct behavior using a mutagenesis and chimera approach. Criteria to assess shifts in conformational equilibrium included (i) K+ sensitivity of Na-ATPase measured at micromolar ATP, under which condition E 2(K+) →E 1 + K+ becomes rate-limiting, (ii) changes in K′ATP for low affinity ATP binding, (iii) vanadate sensitivity of Na,K-ATPase activity, and (iv) the rate of the partial reaction E 1P →E 2P. We first confirmed that interactions between the cytoplasmic domains of α2 that modulate conformational shifts are fundamentally similar to those of α1, suggesting that the predilection of α2 for E 1 state(s) is due to differences in primary structure of the two isoforms. Kinetic behavior of the α1/α2 chimeras indicates that the difference inE 1/E 2 poise of the two isoforms cannot be accounted for by their notably distinct N termini, but rather by the front segment extending from the cytoplasmic N terminus to the C-terminal end of the extracellular loop between transmembranes 3 and 4, with a lesser contribution of the α1/α2 divergent portion within the M4-M5 loop near the ATP binding domain. In addition, we show that the E 1 shift of α2 results primarily from differences in the conformational transition of the dephosphoenzyme, (E 2(K+) →E 1 + K+), rather than phosphoenzyme (E 1P → E 2P).

We showed earlier that the kinetic behavior of the ␣2 isoform of the Na,K-ATPase differs from the ubiquitous ␣1 isoform primarily by a shift in the steady-state E 1 /E 2 equilibrium of ␣2 in favor of E 1 form(s). The aim of the present study was to identify regions of the ␣ chain that confer the ␣1/␣2 distinct behavior using a mutagenesis and chimera approach. Criteria to assess shifts in conformational equilibrium included (i) K ؉ sensitivity of Na-ATPase measured at micromolar ATP, under which condition E 2 (K ؉ ) 3 E 1 ؉ K ؉ becomes rate-limiting, (ii) changes in K ATP for low affinity ATP binding, (iii) vanadate sensitivity of Na,K-ATPase activity, and (iv) the rate of the partial reaction E 1 P 3 E 2 P. We first confirmed that interactions between the cytoplasmic domains of ␣2 that modulate conformational shifts are fundamentally similar to those of ␣1, suggesting that the predilection of ␣2 for E 1 state(s) is due to differences in primary structure of the two isoforms. Kinetic behavior of the ␣1/␣2 chimeras indicates that the difference in E 1 /E 2 poise of the two isoforms cannot be accounted for by their notably distinct N termini, but rather by the front segment extending from the cytoplasmic N terminus to the C-terminal end of the extracellular loop between transmembranes 3 and 4, with a lesser contribution of the ␣1/␣2 divergent portion within the M4-M5 loop near the ATP binding domain. In addition, we show that the E 1 shift of ␣2 results primarily from differences in the conformational transition of the dephosphoenzyme, (E 2 (K ؉ ) 3 E 1 ؉ K ؉ ), rather than phosphoenzyme (E 1 P 3 E 2 P).
The Na,K-ATPase or sodium pump is an integral membrane protein complex found in the plasma membrane of virtually all animal cells. It catalyzes the exchange of three intracellular Na ϩ ions for two extracellular K ϩ ions using the energy of hydrolysis of one molecule of ATP. Consequently, the sodium pump plays an essential role in the maintenance of the electrochemical alkali cation gradients, providing the driving force for the transport of various nutrients into the cell. The Na,K-ATPase is a member of the family of P-type ATPases, which during the course of their catalytic cycle undergo phosphoryl-ation and dephosphorylation of a conserved aspartate residue located in the large catalytic loop between transmembrane segments 4 and 5 of the catalytic ␣ subunit. During the catalytic cycle both dephospho-and phosphoenzymes undergo conformational transitions commonly referred to as E 1 7 E 2 and E 1 P 7 E 2 P, respectively. In addition to the large catalytic ␣ subunit, the Na,K-ATPase comprises a smaller, highly glycosylated ␤ subunit that is important for the proper folding of ␣ and its insertion into the plasma membrane. At present, four isoforms of ␣ and three isoforms of ␤ have been described, and these are distributed in a tissue-and developmentally dependent manner.
The ␣2 isoform is located primarily in skeletal muscle and in brain, predominantly in glial cells. Our earlier studies indicated that it differs from the ubiquitous ␣1 subunit primarily in the steady-state E 1 /E 2 equilibrium. Thus, compared with ␣1, the E 1 /E 2 poise of ␣2 is shifted toward E 1 . This shift is reminiscent of the changes in ␣1 effected by deleting 32 residues from its N terminus (mutant ␣1M32) which corresponds to the E 1 -shifted trypsinized kidney enzyme first described by Jorgensen (1). Except for a modest (Ϸ1.5-fold) increase in KЈ Na (12), ␣2 resembles ␣1 with respect to apparent affinity for extracellular K ϩ when ouabain-sensitive K ϩ influx is assayed under physiological conditions of ATP concentration. However, marked differences between ␣2 and ␣1 become apparent when, at micromolar ATP, the E 2 (K ϩ ) 33 E 1 conversion is ratelimiting. Under these conditions, ␣2* 1 resembles closely the ␣1M32 mutant. We showed previously (2, 3) that, compared with ␣1, both ␣2* and ␣1M32 have faster rates of K ϩ deocclusion as seen in K ϩ stimulation rather than ␣1-like inhibition of Na-ATPase activity at 1 M ATP, with a concomitantly lower KЈ ATP for low affinity ATP binding and decreased (50%) catalytic turnover.
One region of marked primary sequence diversity among the otherwise homologous ␣1, ␣2*, and ␣3* isoforms is the N terminus. In previous studies we compared the kinetics of ␣1/␣2 chimeras in which the first 32 residues were interchanged. The results showed that although removal of 32 residues from the N terminus of ␣1 yields an enzyme with ␣2*-like kinetics as mentioned above, substitution of the first 32 residues of ␣1 with those of ␣2* is without effect. Similarly, substituting the analogous N-terminal sequence of ␣1 into that of ␣2* does not dramatically alter the kinetics of ␣2. It was therefore suggested that either removal of the N terminus, as in the case of ␣1M32, or alteration of its primary or secondary structure, as in the case of ␣2*, likely results in a weakening of intramolecular interactions between the N terminus and some other regions of the ␣ protein (2).
The experiments described in the present study were designed to extend the chimera approach to additional domains of isoform diversity to pinpoint regions that confer the ␣1/␣2 distinct behavior. As a prelude to this analysis, we first show that cytoplasmic interactions that underlie the E 1 /E 2 conformational transitions seen with ␣1 (3,4) are notably similar in ␣2*. The subsequent analysis of ␣1/␣2 chimeras shows that the N-terminal segment encompassing the cytoplasmic N terminus and the first (M2-M3) cytoplasmic loop of ␣2*, the so-called "Actuator" or A domain (5) and, to a lesser extent, the most divergent portion of the nucleotide binding or N domain within the large M4-M5 loop, are responsible for the E 1 shift of ␣2*.

EXPERIMENTAL PROCEDURES
Mutagenesis, Transfection, and Cell Culture-All of the mutant and chimeric ␣'s used in this study were derived from the ouabain-insensitive rat ␣1 and ␣2* cDNAs previously described by Jewell and Lingrel (6). The E231K mutation was introduced into a rat ␣2* cDNA that had been excised from pRcCMV with HindIII (81, 3282) using PCR amplification with synthetic nucleotides and the TaqPlus polymerase reaction kit (Stratagene). The reaction mixtures were eluted from QIAquick spin columns (Qiagen) and digested with DpnI, EcoRI, and SalI. The EcoRI (780) -SalI (1202) ␣2* fragment was gel-purified and ligated into a modified pIBI␣2* shuttle vector in place of the wild type EcoRI-SalI fragment. Clones containing the E231K mutation (G800 3 A) were identified by the presence of adenosine at nucleotide 800, and the complete sequence of the substituted fragment was verified before excising the full-length ␣2E231K cDNA from the shuttle vector and ligating it into pRcCMV (Invitrogen). Orientation of the ␣2E231K cDNA in pRcCMV was determined by restriction enzyme analysis.
The ␣2M30 mutant was constructed by introducing a 30-residue deletion into the 5Ј HindIII (66) -EcoRI (780) restriction fragment cassette of the rat ␣2* cDNA as described previously (2). The mutant cassette was then ligated into the rat ␣2* cDNA in place of the wild type HindIII-EcoRI cassette. The full-length mutant cDNA was released from the shuttle vector by digestion with HindIII (66, 3284) and ligated into the expression plasmid pCDNA3.1 (Invitrogen), and orientation of the cDNA was determined by restriction analysis.
Chimeras ␣2-(1-309)/␣1 and ␣1-(1-311)/␣2 were prepared by exchanging 5Ј HindIII-BssHII fragments of the two cDNAs. The HindIII site of each cDNA is in the 5Ј-untranslated region, and the BssHII site splits the codons for Ala-347 in ␣1 and the corresponding Ala-345 in ␣2. It should be noted that the two isoforms have identical sequences in the region between residues 311 and 346 of ␣1. The full-length cDNAs for the chimeric proteins were then excised from the shuttle vectors and cloned into pRcCMV as above.
Membrane Preparation-NaI-treated microsomal membranes were prepared from the mutant cells as described earlier (6,9). Protein content was determined with a detergent-modified Lowry assay (10).
Enzyme Assays-Na,K-ATPase activity was measured as the release of 32 P i from [␥-32 P]ATP as previously described (11). Briefly and unless indicated otherwise, the membranes were preincubated for 10 min at 37°C with all reactants added except [␥-32 P]ATP. The reaction was initiated by the addition of [␥-32 P]ATP. Final concentrations for Na,K-ATPase activity measurements were 100 mM NaCl, 10 mM KCl, 3 mM MgSO 4 , 20 mM histidine (pH 7.4), 5 mM EGTA (pH 7.4), and 5 M ouabain (Sigma). 5 mM ouabain was used to determine base-line hydrolysis activity. As in earlier studies and unless indicated otherwise, assays of Na,K-ATPase activity were carried out using 1 mM ATP to maintain close to saturating ATP concentration and maximize sensitivity of assays of the relatively low activity cultured cells (cf. Refs. 3, 9, and 12). Na-ATPase activity was measured at 1 M ATP as described previously (2), with varying amounts of added KCl and choline chloride to maintain constant chloride (40 mM) concentration. Base-line activity was determined with 40 mM KCl replacing NaCl. For studies of vanadate sensitivity, inorganic orthovanadate (Fisher) solutions were prepared before the experiment and added with the [␥-32 P]ATP solution to initiate the reaction. Na,K-ATPase activities obtained at various vanadate concentrations and expressed as the percentage of that obtained in the absence of vanadate were analyzed by fitting the data to a one-compartment model using a nonlinear least-square analysis of a general logistic function, as described elsewhere (13). Curve fitting was carried out using the Kaleidagraph computer program (Synergy). Each experiment was carried out at least three times, with one or more chimeras analyzed concurrently with the ␣2* and ␣1 isoforms.
Rate of E 1 P 3 E 1 P-After formation of E 1 P in the presence of high chloride concentration (14), the rate of E 1 P 3 E 2 P was determined by measuring the rate of disappearance of total phosphoenzyme after rapid dilution of the salt (to allow "normal" relaxation of E 1 P 3 E 2 P) plus the addition of KCl to catalyze rapid hydrolysis of E 2 P (14, 15). Accordingly, the enzyme was first phosphorylated in medium containing 600 mM NaCl to stabilize E 1 P, 1 mM MgCl 2 , 1 mM EGTA, and 20 mM Tris-HCl (pH 7.4) with 1 M [␥-32 P]ATP for 30 s at 0°C to obtain maximal phosphoenzyme. Dephosphorylation was then initiated by 6-fold dilution with a "chase" medium containing final concentrations of 20 mM KCl, 10 M unlabeled ATP, 1 mM EGTA, and 20 mM Tris-HCl (pH 7.4), which simultaneously lowered the NaCl concentration to 100 mM. Samples were taken for measurement of [ 32 P]E for periods up to 30 s. Background phosphoenzyme levels were obtained by allowing the chase to continue for 60 s. The data were fitted to a first-order decay model using the Kaleidagraph nonlinear fitting program (Synergy). At least two different membrane preparations obtained from at least two different clones were assayed. The data presented are representative of at least three independent experiments. Each value shown is the mean Ϯ S.D. of triplicate determinations.

RESULTS
Cytoplasmic Interactions of ␣2*-Our earlier studies showed that the steady-state E 1 /E 2 conformational equilibrium of ␣2*, compared with that of the ubiquitous ␣1 isoform, is poised toward E 1 . In fact, the kinetic behavior of ␣2* is generally similar to that of mutants of ␣1 in which the poise is shifted toward E 1 . These mutations include a 32-residue deletion of the cytoplasmic N terminus (mutant ␣1M32) and a Glu-233 3 Lys replacement (mutant ␣1E233K) in the first cytoplasmic loop. Furthermore, the combination of these two mutations of the so-called Actuator domain (mutant ␣1M32E233K) in ␣1 results in a remarkably synergistic shift in poise toward E 1 state(s). The findings were interpreted to indicate that interactions between these regions of the Actuator domain (5) and the catalytic loop are critical for conformational coupling of the Na,K-ATPase (4).
Experiments carried out in the present study to address the question of whether cytoplasmic interactions of ␣2* are analogous to those of ␣1 are summarized in Fig. 1 and Table I. Expression of ␣2M30 and ␣2E231K in HeLa cells yielded functional enzymes capable of supporting cell growth in 1 M ouabain. Cells transfected with the double mutant ␣2M30E231K, analogous to ␣1M32E233K (4), failed to grow even in elevated K ϩ (cf. Ref. 16). Therefore, a comparable "mutation" was prepared by enzymatically cleaving the N terminus of ␣2E231K in the E 1 (Na ϩ ) conformation with trypsin (cf. Ref. 1). The single mutants and the cleaved ␣2E231K (␣2E231K-Tryp) were then assessed for shifts in the E 1 /E 2 equilibrium using the following criteria: (i) the effect of K ϩ on Na-ATPase activity as a measure of E 2 (K ϩ ) 3 E 1 (see Ref. 2), (ii) KЈ ATP for low affinity binding to E 2 (K ϩ ), KЈ ATP(L) , and (iii) sensitivity of Na,K-ATPase activity to inhibition by inorganic orthovanadate.
Thus, at micromolar ATP concentrations sufficient to saturate only the high affinity phosphorylation site the response of Na-ATPase to K ϩ is a sensitive means to characterize mutantspecific differences in the K ϩ -deocclusion pathway of the reaction cycle [E 2 (K ϩ ) 33 E 1 ϩ K ϩ ] that becomes rate-limiting under these conditions (2,17). As shown previously and summarized in Table I (upper panel), a low concentration of K ϩ (1 mM KCl) inhibits the Na-ATPase activity of ␣1 but stimulates that of ␣1M32 and ␣1E233K with a further and notably synergistic stimulation of the double mutant ␣1M32E233K. The  pattern for ␣2* cytoplasmic mutants is remarkably similar (see Fig. 1 and the lower panel of Table I). With respect to the ␣2 mutants, K ϩ stimulates the Na-ATPase activity of ␣2M30 and ␣2E231K, ϳ300 and 900%, respectively, and that of ␣2E231K-Tryp, more than 2000%. In fact, the apparent stimulation may be an underestimation of the true stimulation since trypsinolysis of ␣2E231K undoubtedly results in a heterogeneous pool of enzyme species that includes untrypsinized enzyme having inherently lower sensitivity to K ϩ activation. It should be mentioned that trypsinization of ␣2 in the presence of Na ϩ increased K ϩ activation such that ␣2-Tryp resembled ␣2M30 (data not shown; cf. tryptic cleavage of ␣1 described in Ref. 18), which is not surprising since the region encompassing the tryptic cleavage site (T2; see Ref. 1) is conserved between ␣1 and ␣2. These results suggest that ␣2M30, ␣2E231K, and ␣2E231K-Tryp all shift the E 1 /E 2 equilibrium of ␣2* progressively further toward E 1 .
As shown in Table I, ␣2M30, ␣2E231K, and ␣2E231K-Tryp also exhibit increased affinities for low affinity ATP binding relative to ␣2*, with a synergistic effect of the N-terminal deletion and the E231K mutation in the ␤-strand region of the M2-M3 loop. It should be noted that, as with the K ϩ stimulation of Na-ATPase activity shown above, the KЈ ATP(L) of ␣2E231K-Tryp is an overestimation of the actual value of the double mutant due to the presence of untrypsinized enzyme.
To gain further insight into the effects of these mutations on conformational equilibrium, we investigated the sensitivity of the Na,K-ATPase activity of the mutants to inhibition by vanadate. Inorganic orthovanadate is a transition state analog of inorganic phosphate that binds to P-type ATPases in the E 2 conformation during steady-state catalysis. Consequently, sensitivity of an enzyme to inhibition by vanadate is a measure of the proportion of enzyme in the E 2 state (19). As shown by the representative experiment in Fig. 2 and summarized in Table I, FIG. 3. Chimeras of ␣1 and ␣2. A, schematic representation of ␣1/␣2 chimeras. ␣1/␣2 chimeras were constructed as described under "Experimental Procedures" whereby the N terminus (␣2 nt /␣1, residues 1-65) and a portion of the catalytic M4/M5 loop of ␣1 (␣2 L /␣1, residues 429 -565) were substituted with the analogous residues of ␣2* either individually or together (␣2 (ntϩL) /␣1). Two additional chimeras were constructed in which the region encompassing the entire Actuator domain (residues 1-311 in ␣1 and the analogous region of ␣2*) was interchanged (␣1-(1-311)/␣2 and ␣2-(1-309)/ ␣1). B, sequence alignment of ␣1 and ␣2. The sequences for the regions described in A of the rat ␣1 and ␣2* were aligned using ClustalW, namely (i) the N-terminal segment and (ii) the L region of the N domain. Transmembrane segments are shaded.
␣2M30 and ␣2E231K are both less sensitive to vanadate inhibition than ␣2*, suggesting that these mutations further shift the E 1 /E 2 equilibrium of ␣2* toward E 1 . We were unable to determine accurately the IC 50 for vanadate inhibition of ␣2E231K-Tryp because the residual Na,K-ATPase activity was too low relative to the background (Ϸ15 nmol/mg⅐min).
Chimeras of the ␣1 and ␣2* Isoforms-The above analysis of cytoplasmic mutations of the ␣2* isoform of the Na,K-ATPase support the notion that interactions between the cytoplasmic domains of ␣2* that modulate conformational shifts are fundamentally similar to those of ␣1. The analysis also suggests that it is the ␣2-specific regions of the A domain and/or the catalytic domain with which A interacts that underlie the predilection of this isoform for the E 1 state(s). Accordingly, an ␣1/␣2 chimera approach was used to address this issue.
From a comparison of the primary structure of the cytoplasmic regions of ␣1 and ␣2* (Fig. 3B), it is important to note that the ␤ strand region in the M2-M3 loop of ␣1 encompassing Glu-233 is highly homologous to that of ␣2*. In contrast, the two isoforms differ significantly in the primary sequence of their N termini and in the sequence encompassing the ATP binding site in the M4-M5 cytoplasmic loop. Therefore, chimeras were constructed in which the entire N terminus (residues 1-65) and the divergent portion of the nucleotide binding (N) domain in the large M4-M5 loop (residues 429 -565) of ␣1 were substituted with the analogous residues of ␣2*. It should be noted that our original intent was to investigate 5 individual divergent regions contained within residues 429 -565 (see "Experimental Procedures," cassettes I-V) on the E 1 /E 2 equilibrium of ␣1. Because the individual domains showed no effect on the K ϩ inhibition of Na-ATPase of ␣1 relevant to the K ϩ deocclusion pathway, 2 we constructed an ␣1/␣2 chimera where all 5 cassettes of ␣2* were substituted into ␣1. Two additional chi-meras were constructed. In these chimeras, the first 311 residues of ␣1, which encompass the entire A domain, are interchanged between ␣1 and ␣2* because this region comprises the portion of the cytoplasmic domain that undergoes large rotational displacements during E 1 /E 2 transitions (5,20). A schematic representation and the designation of the chimeras are illustrated in Fig. 3A.
To assess the effect of these different regions on the E 1 /E 2 poise of ␣1, we first investigated the effect of K ϩ on Na-ATPase activity measured at micromolar ATP for each of the chimeras as described above. As shown in Fig. 4, the Na-ATPase activities of wild type ␣1 and chimeras ␣2 nt /␣1, 3 ␣2 L /␣1, and ␣2 (ntϩL) /␣1 are all inhibited by K ϩ . In contrast, wild type ␣2* and ␣2-(1-309)/␣1 are stimulated by K ϩ at least up to 1 mM, consistent with a faster E 2 (K ϩ ) 3 E 1 transition and, hence, with a preponderance of the E 1 conformation. It is noteworthy that although the ␣2 nt /␣1 construct encompasses the most divergent portion of the primary sequence, it is only upon replacement of the homologous N-terminal segment (residues 1-311) of ␣1 with that of ␣2*, chimera ␣2-(1-309)/␣1, that the shift toward E 1 is observed. Fig. 5 shows the determination of KЈ ATP(L) using the Lineweaver-Burk transformation of the simple Michaelis-Menten analysis of Na,K-ATPase activity as a function of ATP concentration, with values presented in the inset. Only the ␣2-(1-309)/␣1 chimera, like wild type ␣2*, showed an increased affinity for ATP. Thus, the K ϩ stimulation of ␣2-(1-309)/␣1 described in Fig. 4 correlates with its increased apparent affinity for ATP, consistent with a shift in the E 1 /E 2 poise in favor of E 1 . Chimeras ␣2 L /␣1 and ␣2 (ntϩL) /␣1 show no difference in KЈ ATP(L) relative to ␣1.
To further examine the contribution of the various domains to the poise of E 1 /E 2 , we used vanadate as a conformational probe for E 2 forms as described above. In one set of experiments, chimeras ␣1-(1-311)/␣2 and ␣2-(1-309)/␣1 were analyzed concurrently with ␣1 and ␣2* (Fig. 6A), and in another, the ␣2 L /␣1 chimera was compared with ␣1 and ␣2* (Fig. 6B). As shown previously (21), the Na,K-ATPase activity of ␣2* is ϳ25-fold less sensitive to vanadate than ␣1 (Fig. 6A). Interestingly, the ␣1-(1-311)/␣2 enzyme is 3.7-fold less sensitive to vanadate than ␣1 and resembles ␣1 with respect to K ϩ inhibition of Na-ATPase and KЈ ATP(L) , whereas ␣2-(1-309)/␣1 is 8.5-fold less vanadate-sensitive than ␣1 and resembles ␣2* with respect to K ϩ activation of Na-ATPase and KЈ ATP(L) . A modest shift in conformational poise is also indicated by the behavior of the ␣2 L /␣1 chimera. This replacement of the divergent region of the N domain of ␣1 by that of ␣2 reduces the vanadate sensitivity of ␣1 by a factor of 2.3. In other experiments (not shown) a similar Ϸ2-fold shift was effected by including the loop insertion of ␣2* with the ␣2* N terminus, i.e. a 2-fold higher IC 50 for vanadate seen for ␣2 nt /␣1 compared with ␣1, which was further doubled, resulting in a 4-fold higher IC 50 compared with ␣1 for ␣2 (ntϩL) /␣1. Taken together these findings imply a major contribution of the N-terminal segment encompassing residues 1-309 and a smaller contribution of the L region of ␣2* in effecting shifts away from E 2 .
It has been observed that the ␣2* isoform and the 32-residue deletion mutant of ␣1, ␣1M32, are both enzymes with their E 1 /E 2 equilibrium shifted toward E 1 forms. Both enzymes show similar K ϩ sensitivities of Na-ATPase activity at low ATP, KЈ ATP(L) values, catalytic turnovers, and K ϩ deocclusion rates. One major difference between the two enzymes is their sensitivity to inhibition by vanadate. Thus, ␣2* has a 25-fold lower IC 50 than ␣1M32. Because vanadate sensitivity reflects the steady-state levels of E 2 , we investigated differences not only in the E 2 (K ϩ ) 3 E 1 transition rate but also the E 1 P 3 E 2 P transition rate. As shown previously (22) and represented in Fig. 7, ␣1M32 has a 5-fold slower conversion of E 1 P to E 2 P than ␣1, consistent with a preference for the E 1 P form. This does not hold true for ␣2*; its E 1 P 3 E 2 P transition rate is only slightly slower than that of ␣1, providing an explanation for its higher sensitivity to vanadate compared with ␣1M32. The implication of vanadate sensitivity as a measure of the E 1 /E 2 poise during steady-state catalysis is discussed further below.

DISCUSSION
Our earlier studies showed a notable similarity between the ␣2* isoform of the Na,K-ATPase and the "E 1 -shifted" mutants of ␣1, particularly the deletion mutant ␣1M32. Nevertheless, the behavior of ␣1/␣2 chimeras in which the first 32 residues of the N termini were interchanged showed that the kinetic difference between ␣1 and ␣2* could not be explained by their distinct N-terminal 1-32 residue per se. More likely, those findings indicate that the difference is due to the interaction of the N terminus with another, isoform-distinct region(s) of the enzyme.
As a preliminary step toward defining the structural basis for ␣1/␣2 differences, we have used a mutagenesis approach to obtain evidence for cytoplasmic interactions between the N terminus, the M2-M3 loop, and large M4-M5 loop of ␣2*, as noted previously for ␣1 (3,4). As shown in Table I (lower panel), a kinetic analysis of cytoplasmic mutants of ␣2*, namely ␣2M30 and ␣2E231K, show that both mutations effect shifts in the E 1 /E 2 poise of ␣2* analogous to those seen for ␣1, although even further toward E 1. HeLa cells transfected with the double mutant, ␣2M30E231K, analogous to ␣1M32E233K, failed to grow. Noting that the catalytic turnover of ␣1M32E233K is Յ500 min Ϫ1 (4), it is likely that the catalytic turnover of ␣2M30E231K is much too low to support HeLa cell growth in 1 M ouabain. Consequently, the N terminus of ␣2E231K was enzymatically cleaved by trypsinolysis of membranes isolated from the ␣2E231K-transfected cells as originally described by Jorgensen (1). Like ␣1M32E233K, ␣2E231K-Tryp showed a strong, synergistic effect on E 1 7 E 2 , shifting it even further in favor of E 1 forms. Therefore, we conclude that the interaction of the cytoplasmic domains of ␣2* that modulate conformational shifts are fundamentally similar to those of ␣1.
As already mentioned, the primary structure of the ␤-strand region in the M2-M3 loop encompassing Glu-231 of ␣2* is highly homologous to that of ␣1 containing Glu-233. In contrast, the primary structures of the two isoforms differ significantly in their N termini (domain nt, 56% identity) 4 and a region within the N domain of the M4-M5 loop and referred to here as the L region (61% identity). It is noteworthy that for the remainder, the sequence identity is very high, namely 86% between the end of the nucleotide and the beginning of the L region (residues 429 -565; see Fig. 3B) and 92% from the Cterminal end of the L region to the C terminus of the protein. Therefore, ␣1/␣2 chimeras were constructed in which domains of divergent primary sequence were interchanged. Thus, the N terminus (nucleotide residues 1-65) and L region (residues 429 -565) of ␣1 were substituted with the analogous regions of ␣2*. This was done either individually (␣2 nt /␣1 and ␣2 L /␣1, respectively) or in combination (␣2 (ntϩL) /␣1) (see Fig. 3A for a schematic representation). In addition to the above, chimeras encompassing the entire Actuator domain, i.e. ␣1-(1-311)/␣2 and ␣2-(1-309)/␣1), were analyzed. As shown in recent structural studies, this domain undergoes large rotational motions (estimated at 110°in the sarcoplasmic reticulum Ca-ATPase) in the course of the conformational transitions (20). The present results show that although a switch of the entire N terminus is without effect, inclusion of the entire isoform-divergent N-terminal segment up to residue 309 of ␣2* is capable of conferring ␣2*-like kinetics to ␣1. This is apparent from the K ϩ activation of Na-ATPase activity at low ATP when E 2 (K ϩ ) 3 E 1 is rate-limiting as well as a decrease in KЈ ATP(L) . However, this N-terminal segment of ␣2* only partially decreases the vanadate sensitivity of ␣1 (Ϸ8.5-fold compared with 20 -25-fold for ␣2* relative to ␣1). In addition, however, the L region of ␣2* confers a 2.3-fold decrease in vanadate sensitivity to ␣1, such that synergistic effects of the two domains, residues 1-309 in the N-terminal segment and residues 427-562 in the N domain, can account for the ␣2* versus ␣1 differences.
Although it is attractive to hypothesize that the E 1 shift of ␣2-(1-309)/␣1 is due to the A domain, one cannot exclude the possible contribution of amino acid replacements in other regions, namely M1, M2, and M3 and the extracellular M1-M2 and M3-M4 loops. Of these regions, it is noteworthy that Coppi et al. (31) showed that an ␣2-(1-129)/␣1 chimera is not E 1shifted because, like ␣1, its Na-ATPase activity is inhibited by K ϩ at low ATP concentration (31). This result provides a basis for eliminating M1 and the M1-M2 loop as candidates for effecting the E 1 shift. For the rest, there are four replacements, namely Ser-132 3 A in M2, His-288 3 Gln in M3, and Glu-309 3 Gly and Thr-311 3 Ser in the M3-M4 loop. Experiments are currently under way to determine whether and/or which ones of these replacements are important or whether it is the A domain per se that confers the ␣2-like E 1 shift seen with the ␣2-(1-309)/␣1 mutant.
It is instructive to consider likely ␣1/␣2 differences in the interactions of the A domain with the N and P domains in the E 1 and E 2 conformations as well as interactions within regions of the A domain. With the Na,K-ATPase, metal-catalyzed oxidative cleavage studies of domain interactions reveal that in the E 1 conformation, the N domain docks onto the phosphorylation (P) domain, and A moves apart; in E 2 , A docks onto P, and N is displaced (32), consistent with putative domain interactions deduced from crystal structures of sarco(endo)plasmic reticulum calcium ATPase in E 1 and E 2 states (5,20,33). Concerted effects of ␣1 versus ␣2 distinct residues within the N-terminal segment encompassing domain A and the L region within domain N can be explained by the recent model for regulation of the aforementioned domain interactions regulated by the N terminus (22). Thus, assuming that the secondary structure of the ␣2* N terminus, like that of ␣1, has the propensity to form three short helices (H1, H2, and H3), intramolecular interactions between helices 1 and 2 of the N 4 As determined by BLAST, NCBI.  (22). The present kinetic analysis involves evaluation of the overall reaction cycle under different conditions of rate limitation or ligand perturbations as well as certain partial reactions relevant to conformation transitions. The usefulness of this approach is indicated by the comparison of ␣2* and ␣1M32 as summarized in Table II. Compared with ␣1, ␣1M32 but not ␣2*, exhibits a substantial slowing of the E 1 P 33 E 2 P conversion, providing an explanation for its much larger decrease in sensitivity to vanadate compared with ␣2*, i.e. IC 50 values of ␣1M32 and ␣2 are decreased ϳ500and 20-fold, respectively, compared with ␣1. Accordingly, the E 1 shift in E 1 /E 2 poise of ␣2* is due to a shift in dephosphoenzyme [E 2 (K ϩ ) 3 E 1 ] but not phosphoenzyme [E 1 P 3 E 2 P]. This result indicates the "nonequivalence" of the two conformational transitions and also highlights a limitation of the sole use of vanadate as a probe of conformation since the steady-state proportion of enzyme in the E 2 state reflects the rates of transition between both dephospho-and phosphoenzyme states.
This report deals only with the structure/function analysis of isoform-specific kinetic behavior. Isoform-specific interactions of the pump with other proteins such as cytosolic second messengers may be important for adapting sodium pump function to cell-specific requirements as suggested by Blanco et al. (23). Thus, targets for protein kinase C phosphorylation present in the N terminus of ␣1 are absent in ␣2. The significance of this difference is underscored by the observation that in transfected opossum kidney cells (24) protein kinase C␤-mediated phosphorylation of ␣1 at these residues promotes its translocation to the plasma membrane. Similarly, in ␣1 but not ␣2, the presence of Tyr-5 within a consensus sequence for phosphorylation by tyrosine kinases of the Src family has been implicated in the insulin stimulation of pump activity in the proximal convoluted tubules of the rat kidney (25). Furthermore, although insulin acts via tyrosine kinases in rat proximal convoluted tubules to stimulate pump activity by increasing the apparent Na ϩ affinity (26) in skeletal muscle, it promotes translocation of ␣2 to the plasma membrane (27) via the action of protein kinase C (28). There is also direct evidence for a role of the isoform-specific primary sequence within the M4-M5 loop domain, in regulation of the pump by second messengers. In agreement with our findings, Pierre et al. (29) failed to detect an effect of the introduction of an ␣2* distinct region of the large catalytic loop (residues 489 -499, contained within the L region) on the E 1 /E 2 conformational equilibrium of ␣1. However, a role for this domain in the isoform-specific response of transfected opossum kidney cells to protein kinase C activation was observed in the differential response of the isoforms to hormones via the action of second messengers. Taken together, the foregoing studies suggest that the primary sequence diversity may in part be relevant to isoform-specific pump regulation, separate from a role in isoform-distinct pump kinetics.
In conclusion, we have shown that the E 1 shift in the conformational equilibrium of ␣2 can be largely accounted for by the N-terminal third of the ␣ subunit that comprises mainly the A domain, with a small contribution of the isoform-specific sequence of the N domain within the M4-M5 loop. In addition, despite the kinetic similarities of the ␣2 isoform with cytoplasmic mutants of ␣1, such as ␣1M32, the E 1 shift of ␣2 results primarily from differences in the dephosphoenzyme conformational transition, i.e. E 2 (K ϩ ) 3 E 1 , with its E 1 P 3 E 2 P transition rate similar to that of ␣1.