Selective Fe2+-catalyzed Oxidative Cleavage of Gastric H+,K+-ATPase

In the presence of ascorbate/H2O2, Fe2+ ions or the ATP-Fe2+ complex catalyze selective cleavage of the α subunit of gastric H+,K+-ATPase. The electrophoretic mobilities of the fragments and dependence of the cleavage patterns on E 1 andE 2 conformational states are essentially identical to those described previously for renal Na+,K+-ATPase. The cleavage pattern of H+,K+-ATPase by Fe2+ ions is consistent with the existence of two Fe2+ sites: site 1 within highly conserved sequences in the P and A domains, and site 2 at the cytoplasmic entrance to trans-membrane segments M3 and M1. The change in the pattern of cleavage catalyzed by Fe2+ or the ATP-Fe2+ complex induced by different ligands provides evidence for large conformational movements of the N, P, and A cytoplasmic domains of the enzyme. The results are consistent with the Ca2+-ATPase crystal structure (Protein Data Bank identification code 1EUL; Toyoshima, C., Nakasako, M., Nomura, H., and Ogawa, H. (2000) Nature 405, 647–655), anE 1Ca2+ conformation, and a theoretical model of Ca2+-ATPase in anE 2 conformation (Protein Data Bank identification code 1FQU). Thus, it can be presumed that the movements of N, P, and A cytoplasmic domains, associated with theE 1 ↔ E 2 transitions, are similar in all P-type ATPases. Fe2+-catalyzed cleavage patterns also reveal sequences involved in phosphate, Mg2+, and ATP binding, which have not yet been shown in crystal structures, as well as changes which occur in E 1 ↔E 2 transitions, and subconformations induced by H+,K+-ATPase-specific ligands such as SCH28080.

The gastric H ϩ ,K ϩ -ATPase generates gastric acid by pumping protons or hydronium ions out of the parietal cell of the stomach in exchange for potassium. H ϩ ,K ϩ -ATPase, like the Na ϩ ,K ϩ -ATPase, has 10 trans-membrane helices in the catalytic ␣ subunit like other P2-type ATPases and a single transmembrane helix in the accessory ␤ subunit (1)(2)(3)(4)(5)(6). The catalytic subunits have 65% sequence identity, and both are K ϩ countertransport pumps. In addition to their difference in H ϩ to Na ϩ selectivity, and pumping stoichiometry, the H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase also differ in their inhibitor sensitivity. Ouabain inhibits the Na ϩ ,K ϩ -ATPase but not the H ϩ ,K ϩ -ATPase and, conversely, the imidazo [1,2␣]pyridine, SCH28080, 1 inhibits the H ϩ ,K ϩ -ATPase but not the Na ϩ ,K ϩ -ATPase.
The work in this paper addresses the question as to the similarity or difference in structural organization of the cytoplasmic region of the two enzymes, associated with E 1 7 E 2 transitions. The experiments utilize Fe 2ϩ -catalyzed oxidative cleavage of gastric H ϩ ,K ϩ -ATPase, a technique that we have described extensively for cleavage of renal Na ϩ ,K ϩ -ATPase (see Refs. 26 -28; for a recent review, see Ref. 29; see also Refs. 30 and 31 for specific Cu 2ϩ -catalyzed cleavages). With Fe 2ϩ ions or ATP-Fe 2ϩ complexes bound to Na ϩ ,K ϩ -ATPase, addition of ascorbate with H 2 O 2 generates OH ⅐ radicals locally (or the reactive metal-peroxyl intermediates), and these selectively cleave the polypeptide chain (32)(33)(34)(35). Because more than one peptide bond can be cleaved from the same metal site, the technique has the important feature of providing information on proximity of cleaved segments in the native protein. In the case of cleavage by the ATP-Fe 2ϩ complex, the properties indicate that the Fe 2ϩ ion is ligated by residues normally occupied by Mg 2ϩ ions, and thus cleavages reveal contacts of Fe 2ϩ or Mg 2ϩ ions (28). Cleavages catalyzed by Fe 2ϩ ions and the ATP-Fe 2ϩ complex have provided evidence that both E 1 3 E 2 and E 1 P 3 E 2 P conformational changes are associated with large movements in the cytoplasmic domains and also with a change in ligation of Mg 2ϩ ions between E 1 P and E 2 P (26 -28). Several cleavage points of the Na ϩ ,K ϩ -ATPase lie within the most highly conserved sequences of P-type pumps, and the experiments indicate that they play a major role in binding phosphate or the ␥-phosphate of ATP ( 369 DKTGT and 607 MVTGD in the P domain) or Mg 2ϩ ions ( 708 TGDGVNDS in the P domain, 212 TGESE in the A domain, and near 440 VAGDA in the N domain), and mediating the domain interactions. This hypothesis is consistent with the recently published 2.6-Å crystal structure of Ca 2ϩ -ATPase (36) and also of proteins with a homologous fold of the phosphorylation domain such haloacid dehalogenase (HAD), phosphonatase, and phosphoserine phosphatase (37)(38)(39)(40)(41). In particular, the proposed domain movements are consistent with inferred structural changes of Ca 2ϩ -ATPase (36,42). Throughout this paper, the structure of Ca 2ϩ -ATPase in an E 1 Ca conformation (PDB identification code 1EUL) and a theoretical model of the protein in an E 2 conformation (PDB identification code 1FQU) serve to illustrate the concepts and clarify experimental design (Figs. 1 and 2). Sequences and residue numbering depicted in the figures are for Ca 2ϩ -ATPase (SERCA), whereas sequences or residue numbering related to cleavage fragments refer to H ϩ ,K ϩ -ATPase or Na ϩ ,K ϩ -ATPase, respectively. Fig. 1 presents an overview of the two models and highlights residues at or near cleavage sites or in domains of interest. Fig. 2 presents a more detailed view of residues in the active site with a bound Fe 2ϩ , based on the E 2 model (1FQU). The E 2 model was produced by moving the cytoplasmic N, P, and A domains and the transmembrane segments of the known structure, in the E 1 conformation, so as to make them fit the electron density envelope of the molecule in E 2 at 8-Å resolution (42). It can be assumed that the differences of the N, P, and A domains in E 1 and E 2 conformations are much better defined than are those of the trans-membrane segments. Domain A consists of the N terminus and the loop between M2 and M3, the P domain consists of the N-terminal region of the loop between M4 and M5 (residues 330 -359) and the C-terminal region (residues 605-737), and the N domain consists of the residues between 359 and 605. One clear difference is that the N, P, and A domains are well separated in E 1 , whereas in E 2 the three domains are docked closely together. The structural organization of the molecule, as well as features of cleavages catalyzed by Fe 2ϩ ions and a complex of Cu 2ϩ ions with bathophenanthroline (31), led us to propose the existence of two Fe 2ϩ sites (29). Site 1 consists of residues within highly conserved sequences in the P domain ( 351 DKTGT, 701 TGDGVNDS, 623 MITGD) and A domain ( 181 TGES), respectively (SERCA sequences). As seen in Fig. 1, the highlighted residues Asp 351 , Thr 353 , Asp 703 , Asp 707 , Thr 625 , and Glu 183 indicate the location of the conserved sequences. Site 1 exists in E 2 conformations but disappears in E 1 confor-mations. It is striking that Asp 351 , Thr 353 , Asp 703 , Asp 707 , Thr 625 in the P domain can be virtually superimposed in E 1 and E 2 conformations. The major difference is that, in E 2 , the A domain docks with the N and P domains, the 212 TGES sequence in the A domain approaching the active site residues in the P domain and creating a cage for Fe 2ϩ binding (see Fig. 2). Residues Glu 439 and Lys 515 in the N domain are also highlighted in Fig. 1 and will be discussed in relation to cleavage catalyzed by the ATP-Fe 2ϩ complex. Cleavages of Na ϩ ,K ϩ -ATPase indicate that site 2 for Fe 2ϩ is located near the entrance of trans-membrane segments M3 (within the sequence 283 HFIH) and M1 and is insensitive to E 1 7 E 2 conformations. A homology model of the membrane domain, in which sequences of Ca 2ϩ -ATPase were replaced by those of Na ϩ ,K ϩ -ATPase, shows close proximity of 283 HFIH (M3) and 81 EWVK (M1) (31). The corresponding sequences of Ca 2ϩ -ATPase 255 EFGE and 49 LWEL are also close together in E 1 (Fig. 1, left). In E 2 (Fig. 1, right), these sequences appear to separate but, as mentioned above, the theoretical model may not be reliable in the trans-membrane region.
The similarity of the reaction mechanism of all P-type cation pumps, and the evidence for the roles of the conserved cytoplasmic residues implies that similar structural changes underlie E 1 7 E 2 transitions in other P-type pumps. The experiments described here test this hypothesis and demonstrate a striking similarity of the structural changes, associated with E 1 7 E 2 conformational transitions, in H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase, as well as applicability of the Fe 2ϩ -catalyzed cleavage technique.
Enzyme Preparations-The H ϩ ,K ϩ -ATPase was derived from hog gastric mucosa by previously published methods, using differential and density gradient centrifugation (43). The vesicles obtained have been shown to be over 90% cytoplasmic side out. The ion impermeability of the vesicles was determined by the difference in K ϩ stimulation of ATPase activity in the presence of KCl and in the presence of KCl and nigericin. The specific activity in the presence of nigericin was Ϸ135 mol of ATP hydrolyzed mg Ϫ1 protein h Ϫ1 , and in the absence of nigericin Ϸ10 mol mg Ϫ1 h Ϫ1 . Thus greater than 90% of the K ϩstimulated ATPase activity was dependent on the addition of the ionophore, nigericin. P i released was measured by the method of Yoda and Hokin (44) and protein concentration by the Lowry method (45). Pig kidney Na ϩ ,K ϩ -ATPase was prepared and assayed as described in Ref. 46. The specific activity was 15-20 mol⅐mg Ϫ1 ⅐min Ϫ1 .
Cleavage Reactions-Hog gastric microsomal membranes enriched with the gastric H ϩ ,K ϩ -ATPase were washed and resuspended at 1 mg/ml in 0.1 M Tris⅐HCl, pH 6.8 (E 1 conformation) or 0.1 M RbCl, 10 mM Tris⅐HCl, pH 6.8 (E 2 conformation). Freshly prepared ascorbic acid/Tris, pH 7.2, and hydrogen peroxide, each at 10 mM, were added to suspensions of washed membranes (0.25-0.5 mg/ml) and also various concentrations of phosphate, ATP, vanadate, magnesium, AMP-PNP, and EDTA, as indicated in figure legends. The cleavage reaction was initiated by adding FeSO 4 at a final concentration of 2-10 M and incubating at 20°C for 5-10 min. To stop the reaction, EDTA was added to a final concentration of 20 mM, followed by 5-fold concentrated gel sample buffer (0.1 M Tris⅐HCl, pH 7.4, 10% SDS, 10 mM EDTA, 10% ␤-mercaptoethanol, 40% sucrose, and 0.025% bromphenol blue). For defining cleavages mediated by the ATP-Fe 2ϩ complex, the enzyme was preincubated in a solution composed of 100 M FeSO 4 with ATP at concentrations between 0.1 and 2 mM and 0.1 M Tris⅐HCl, pH 6.8. Pig kidney Na ϩ ,K ϩ -ATPase (0.3 mg/ml) was suspended in a medium containing 10 mM Tris⅐HCl, pH 6.8, and 0.1 M RbCl. Cleavage was initiated by adding 2 M FeSO 4 and 10 mM ascorbate/H 2 O 2 and incubated for 5 min at 20°C. The cleavage reaction was stopped by adding 20 -50 mM EDTA, and then the 5-fold concentrated gel sample buffer was added at a 1: SDS-Gel Electrophoresis and Western Blotting-Samples were applied to 1.5-mm-thick 10% (34:1 acrylamide/methylene bisacrylamide) slab gels, using the Tricine buffer method of Schä gger and von Jagow (47). The gel and running electrophoresis buffers contained 1 mM EDTA. The gel was run for 4 -6 h at 70 V constant voltage, and the gel was then transferred to polyvinylidene difluoride or nitrocellulose membranes, as described previously (48). Gels included a lane with prestained molecular mass standards (Bio-Rad, 200 -14.4 kDa).
Anti-Q1030-Y1034, referred to as anti-DQELYY, or antibody 12.18, recognizing residues 681 DMDPSEL687 (49), was used for detecting fragments of the H ϩ ,K ϩ -ATPase ␣-subunit, and anti-K1012-Y1016, referred to as anti-KETTY, was used to detect fragments of Na ϩ ,K ϩ -ATPase. Immunoblots were developed by enhanced chemiluminescence (ECLϩplus kit) using anti-rabbit IgG horseradish peroxidase conjugate and the protocol supplied with ECLϩplus reagents from Amersham Pharmacia Biotech or stained by the BCIP/NBT color developing method using anti-rabbit IgG alkaline phosphate conjugate and the protocol from Promega. Blots were scanned using UMAX, Astra 4000U, or Hewlett Packard HP ScanJet 3200C equipment.
Antibody Preparation-The anti-DQELYY, which recognizes the C terminus of the ␣ subunit of H ϩ ,K ϩ -ATPase, was raised in rabbits, after coupling the peptide CDQELYY to keyhole limpet hemocyanin (Biological Services, Weizmann Institute). The antibody was purified on an affinity column of CDQELYY, prepared by coupling the peptide to a Poros activated affinity chromatography column (epoxide or amino), using a BioCad perfusion chromatography apparatus. The antibody was eluted off the column in a solution of 0.2 M Tris⅐HCl, pH 2, neutralized immediately with Tris base, diluted 1:1 with glycerol, 0.02% sodium azide was added for preservation, and the solution stored at Ϫ18°C until use. Immunoblots of H ϩ ,K ϩ -ATPase were probed with anti-DQE-LYY used at a dilution of 1:1000.

RESULTS
For greater ease of understanding of the experimental design and results, the reader is referred to Figs. 1 and 2 and Table I.  The table contains the homologous sequences

Structure of Ca 2؉ -ATPase in an E 1 conformation (left) and a theoretical model in an E 2 conformation (right).
The two structures were drawn from the atomic co-ordinates of PDB identification codes 1EUL and 1FQU, respectively. The highlighted residues indicate the sequences of Ca 2ϩ -ATPase equivalent to those of Na ϩ ,K ϩ -ATPase or H ϩ ,K ϩ -ATPase at the N termini of Fe 2ϩ -catalyzed or ATP-Fe 2ϩ -catalyzed cleavage fragments (see Table I).

TABLE I
Location and masses of cleavage fragments of H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase, and sequence alignment with Ca 2ϩ -ATPase Fragments of H ϩ ,K ϩ -ATPase are identified by sequences at or near the cleavage points by analogy to fragments of Na ϩ ,K ϩ -ATPase. The N termini of fragments of Na ϩ ,K ϩ -ATPase marked with an asterisk ‫)ء(‬ are known exactly; N termini of other fragments of Na ϩ ,K ϩ -ATPase are known to within 4 -10 residues (26 -28, 31). As explained in the text, two minor fragments of Na ϩ ,K ϩ -ATPase, with N termini near EWVK (site 2) and near DKTGT (site 1), respectively, were usually not seen with H ϩ ,K ϩ -ATPase, and are included here for completeness. Calculations of masses assume that the fragments of the ␣ subunits have the same N termini (indicated by the first residue) and extend to the C terminus. Column 4 presents experimental values of the apparent mass Ϯ S.E. (n ϭ 4) of fragments of H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase. Sequences and numbering refer to pig H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase ␣1 subunits and rabbit sarcoplasmic reticulum Ca 2ϩ -ATPase (SERCA1). ATPase, Na ϩ ,K ϩ -ATPase, or Ca 2ϩ -ATPase at or near the cleavage points, the location within the molecule, and theoretical or observed masses of the fragments. The experiments described below demonstrate that the electrophoretic mobility of fragments of H ϩ ,K ϩ -ATPase is indistinguishable from that of fragments of Na ϩ ,K ϩ -ATPase. The N termini of the latter have been identified, either exactly or to within 4 -10 residues, and are referred to as near EWVK (site 2), ESE (site 1), near HFIH (site 2), near DKTGT (site 1), near VAGDA (ATP-Fe 2ϩ site), near MVTGD (site 1), and VNDS (site 1 and ATP-Fe 2ϩ site), respectively (Table I) (26 -28). Accordingly, the corresponding fragments of H ϩ ,K ϩ -ATPase have been labeled ESE, near HFVD, near VIGDA, near MVTGD, and VNDS, respectively. Fig. 3A presents the basic observations on cleavage of hog gastric H ϩ ,K ϩ -ATPase catalyzed by 2 M Fe 2ϩ with 10 mM ascorbate/H 2 O 2 and, for comparison, cleavage fragments of pig kidney Na ϩ ,K ϩ -ATPase. Fragments were detected with antibodies that recognize the C termini of the ␣ subunits (anti-DQELYY or anti-KETYY, respectively). Thus, they extend from their N-terminal cleavage points to the C termini. Cleavage was more efficient at pH 6.8 than at 7.2, and subsequent experiments were done at pH 6.8. In media containing 0.1 M Tris⅐HCl or 0.1 M RbCl, we observed either a single fragment, near HFVD (site 2), or four fragments 230 ESE (site 1), near 299 HFVD (site 2), near 624 MVTGD (site 1), or 728 VNDS (site 1), respectively. In media containing 0.1 M Tris⅐HCl or RbCl, the H ϩ ,K ϩ -ATPase, like Na ϩ ,K ϩ -ATPase, is stabilized in E 1 or E 2 (Rb) conformations, respectively. The conformation dependence of the pattern is the same as described for Na ϩ ,K ϩ -ATPase, respectively (26,27), and, as mentioned above, the electrophoretic mobility of the four fragments is indistinguishable from that of the fragments of Na ϩ ,K ϩ -ATPase (see Table  I). Measured masses of fragments of both H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase are given in Table I. Two minor fragments of Na ϩ ,K ϩ -ATPase reported previously (26,27) with N termini near 81 EWVK and near 369 DKTGT, were usually not seen in the current experiments, and neither did we observe the corresponding minor fragments of H ϩ ,K ϩ -ATPase (but see Fig.  5B). The absence of the minor fragments is attributable to the necessity to digest both proteins in milder conditions than used previously, on account of nonselective cleavage of H ϩ ,K ϩ -ATPase and loss of the epitope, which binds the anti-CDQE-LYY. However, the fragmentation patterns of H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase were identical when cleavage was compared in the same conditions. Fig. 3B shows a blot of the experiment in Fig. 3A probed with a different antibody, Ab12.18, which recognizes residues 681-687 in the large cytoplasmic loop of the H ϩ ,K ϩ -ATPase ␣ subunit. The fragment at 728 VNDS was not seen because its N terminus lies downstream of this epitope, but otherwise the pattern and conformational sensitivity was exactly as in Fig. 3A, confirming the selectivity of the cleavages. An additional indication for selectivity may be obtained by showing that cleavages at the same site show a similar Fe 2ϩ concentration dependence. Titration at increasing Fe 2ϩ (above 5 M) is complicated by the tendency to nonspecific cleavage, but the complementary approach of chelating free Fe 2ϩ with EDTA avoids this difficulty. In the experiment in Fig. 3C, a progressive and complete suppression of all cleavages at 50 -500 M EDTA was observed as the [Fe 2ϩ ] (initially 2 M) was reduced, presumably to a negligible concentration. Although this kind of experiment is not easily analyzed in a quantitative way, the fragments at ESE, near MVTGD, and VNDS (site 1) appeared to show a similar dependence on EDTA concentration, whereas the fragment near HFVD (site 2) appeared to be somewhat more sensitive to EDTA. The reaction was incubated for 10 min for the gastric H ϩ ,K ϩ -ATPase and for 5 min for the Na ϩ ,K ϩ -ATPase at 20°C and stopped by adding 20 mM EDTA. Protein (2 g) was applied to each lane of SDS-polyacrylamide gel electrophoresis. A, hog gastric H ϩ ,K ϩ -ATPase (0.3 mg/ml) was suspended in a medium containing 0.1 M Tris⅐HCl, pH 6.8 or 7.2, or in a medium containing 10 mM Tris⅐HCl, pH 6.8 or 7.2, and 0.1 M RbCl. Cleavage conditions for Na ϩ ,K ϩ -ATPase were as described under "Experimental Procedures." Immunoblot was developed using the ECLϩplus kit. B, cleavage conditions as in A (pH 6.8). Immunoblot was stained with BCIP/NBT reagents. C, cleavage condition as in A, with indicated concentrations of EDTA. Immunoblots were stained with BCIP/NBT color reagents. Lane C, gastric H ϩ ,K ϩ -ATPase incubated without Fe 2ϩ ion. K ϩ -ATPase was determined in media containing either 0.1 M Tris⅐HCl or RbCl and increasing concentrations of Mg 2ϩ ions. Whereas cleavage in 0.1 M Tris⅐HCl was unaffected, as the Mg 2ϩ ion concentration was raised, in the medium containing 0.1 M RbCl, the fragments at ESE, near MVTGD, and at VNDS (site 1) were suppressed in parallel, whereas the fragment near HFVD (site 2) remained unaffected. Thus, sufficiently high Mg 2ϩ concentrations (10 -20 mM) converted the cleavage pattern characteristic of the E 2 conformation to one similar to that of the E 1 conformation. Presumably, Mg 2ϩ ions did not affect the pattern in the medium containing Tris⅐HCl because the enzyme is in an E 1 conformation even without the Mg 2ϩ ions. The effect of Mg 2ϩ ions in Fig. 2 is like that described previously for Na ϩ ,K ϩ -ATPase (27) (see "Discussion" for possible interpretations). With respect to the experiments presented in Fig. 5, it should be noted that concentrations of Mg 2ϩ ions lower than 2 mM had no significant effect on cleavages.
Previously it was shown for Na ϩ ,K ϩ -ATPase that inorganic phosphate ions, P i , selectively suppressed the cleavages near DKTGT and near MVTGD (site 1), and the combinations P i / Mg 2ϩ /ouabain or vanadate/Mg 2ϩ also suppressed those at ESE and VNDS (site 1), whereas the cleavages near EWVK (site 2) and near HFIH (site 2) were unaffected (27). Thus, paradoxically, although P i /Mg 2ϩ /ouabain or vanadate/Mg 2ϩ stabilize E 2 conformations, all cleavages at site 1 were suppressed, and the cleavage pattern was converted to that of an E 1 conformation. The experiments in Fig. 5 examined whether similar effects could be reproduced with H ϩ ,K ϩ -ATPase. As seen in Fig. 5A, inorganic phosphate ions indeed selectively suppressed the fragment near MVTGD (site 1), and in this case they also amplified fragments ESE and VNDS (site 1), whereas the fragment near HFVD (site 2) was not affected. The presence of 5 M vanadate plus 0.5 mM Mg 2ϩ ions (Fig. 5B) suppressed fragments ESE, near MVTGD, and VNDS almost completely, leaving the fragment near HFVD and a minor fragment marked with a star, thus producing an E 1 pattern of cleavage. Fig. 5 (C  and D) presents the effects of P i , Mg 2ϩ , and SCH28080, which could also be predicted to stabilize E 2 conformations. In a medium containing 0.1 M RbCl (Fig. 3C), SCH28080 somewhat amplified the typical E 2 -like cleavages, ESE, near MVTGD, and VNDS, more so in the presence of 0.5 mM Mg 2ϩ . Strikingly, in the presence of P i /Mg 2ϩ /SCH28080, all the fragments except that near HFVD were suppressed. In 0.1 M Tris⅐HCl (Fig. 5D), the presence of SCH28080 with 0.5 mM Mg 2ϩ induced appearance of fragments ESE, near MVTGD, and VNDS, showing that SCH28080 stabilized the E 2 conformation of the unphosphorylated enzyme. Again, with P i /Mg 2ϩ /SCH28080 fragments ESE, near MVTGD, and VNDS were suppressed and only that near HFVD, typical of an E 1 conformation, remained. Thus, Fig. 5 demonstrates that the effects of the pump ligands observed previously with Na ϩ ,K ϩ -ATPase are reproduced also in the case of H ϩ ,K ϩ -ATPase except, of course, that it is necessary to use SCH28080 rather than ouabain with Mg 2ϩ /P i (see "Discussion" and Fig. 8 for an interpretation of these phenomena).
Another set of experiments used an ATP-Fe 2ϩ complex to selectively cleave H ϩ ,K ϩ -ATPase (Figs. 6 and 7). In the experiment of Fig. 6 (left) conducted in a medium containing 0.1 M Tris⅐HCl, the gastric G 1 membranes were incubated with 10 M Fe 2ϩ , 10 mM ascorbate/H 2 O 2 , and increasing concentrations of ATP. In the absence of ATP, a single fragment near HFVD was observed, as expected for cleavage of the E 1 conformation. As the ATP concentration was raised to 100 -200 M, the fragment near HFVD (site 2) was suppressed and two fragments appeared, a major one, 728 VNDS, and a less prominent one, near 456 VIGDA. The mobility of the fragments at VNDS and near VIGDA is indistinguishable from that of two fragments of Na ϩ ,K ϩ -ATPase produced in similar conditions, 712 VNDS in the P domain and near 440 VAGDA in the N domain (28) (see Fig. 1, left). An indication for specific binding of the ATP-Fe 2ϩ complex is the finding that very high concentrations of ATP (1000 M) suppressed the cleavages producing the fragments at VNDS and near VIGDA. This indicates that an excess of uncomplexed ATP is able to compete with and displace the ATP-Fe 2ϩ complex from the site and so prevent the cleavages (see Ref. 28). In addition, the presence of an excess of Mg 2ϩ ions also suppressed the fragments at VNDS and near VIGDA, as would be expected if Mg 2ϩ ions compete with Fe 2ϩ for complexation with the ATP (not shown; see also Ref. 28). Another indication for specific binding of an ATP-Fe 2ϩ complex is the observation that, in a medium containing 0.1 M RbCl (right), it was necessary to add a much higher concentration of ATP (500 M) to induce appearance of the fragments at VNDS and near VIGDA, compared with that in the medium containing 0.1 M Tris⅐HCl. For the Na ϩ ,K ϩ -ATPase, it is well known that E 2 conformations bind ATP with a low affinity and that, by accelerating the rate of the conformational transition E 2 (Rb) 3 E 1 , high concentrations of ATP stabilize E 1 conformations to which it binds with a high affinity (50,(21)(22)(23). Thus, a simple explanation of the result is that the high concentration of ATP-Fe 2ϩ complex is required to induce conversion of the enzyme from an E 2 (Rb) conformation in 0.1 M RbCl, to the E 1 conformation to which the ATP-Fe 2ϩ complex binds and cleaves to the fragments at VNDS and near VIGDA, exactly as in the medium containing 0.1 M Tris⅐HCl. Overall, Fig. 6 shows that the bound ATP-Fe 2ϩ com-plex mediates selective cleavage of H ϩ ,K ϩ -ATPase essentially as described for Na ϩ ,K ϩ -ATPase (28).
In the presence of Na ϩ ions, the ATP-Fe 2ϩ complex phosphorylates Na ϩ ,K ϩ -ATPase due to the ability of the Fe 2ϩ to substitute for Mg 2ϩ ions (51,52). In this condition the E 2 P phosphoenzyme conformation is formed, and tightly bound Fe 2ϩ mediates cleavage at the ESE cleavage site in addition to that at the VNDS site. In the case of H ϩ ,K ϩ -ATPase, Na ϩ ions are not required for phosphorylation. If Fe 2ϩ ions were also able to substitute for Mg 2ϩ ions, one might observe cleavage at the ESE site even in the medium containing 0.1 M Tris⅐HCl. In Fig.  6 (left), a very small amount of fragment at ESE can be seen in the lanes with 200 and 500 M ATP, but one cannot draw conclusions on the basis of such a minor band. However, as seen in Fig. 7A, when we raised the Fe 2ϩ concentration to 100 M in the presence of 500 M ATP, the fragment at ESE was clearly visible, in addition to the fragments at VNDS and near VIGDA. When, by contrast, cleavage was done in the presence of the nonhydrolyzable analogue AMP-PNP, the fragment at ESE was not observed, although those at VNDS and near VIGDA appeared as expected. The difference between ATP and the nonphosphorylating analogue AMP-PNP provides one indication that the fragment at ESE is formed as a result of cleavage of the phosphoenzyme, E 2 P. A second indication came from an experiment that compared cleavage in media containing either 300 mM Tris⅐HCl or NaCl (Fig. 7B). It is striking that, whereas fragments at VNDS and near VIGDA were generated in either medium, the fragment at ESE is prominent only in the Tris⅐HCl medium and is almost absent in the NaCl medium. With Na ϩ ,K ϩ -ATPase, by contrast, the fragment at ESE appears only in the presence of Na ϩ ions, and is absent in a Na-free medium (28). The opposite ion specificity for generating the fragment at ESE in H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase is consistent with prior knowledge that Na ϩ ions stabilize an E 1 conformation, apparently in competition with H ϩ ions, but inhibit H ϩ ,K ϩ -ATPase at neutral pH (17). Thus, a simple explanation of the difference between H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase is that Na ϩ ions compete with H ϩ on H ϩ ,K ϩ -ATPase but are unable to catalyze phosphorylation, and so the fragment at ESE is not produced. The fragments at VNDS and near VIGDA appear in either medium because they are generated by the bound ATP-Fe 2ϩ or AMP-PNP-Fe 2ϩ complex. The conclusion from Fig. 7 (A and B) is that, like Na ϩ ,K ϩ -ATPase, H ϩ ,K ϩ -ATPase is indeed phosphorylated by ATP-Fe 2ϩ and the E 2 P conformation is cleaved within the conserved sequence, TGES.

DISCUSSION
The results in this paper show that Fe 2ϩ -catalyzed cleavage of gastric H ϩ ,K ϩ -ATPase allows detection and characterization not only of E 1 and E 2 conformational states of the protein, but also subconformations stabilized by different pump ligands. Because, previously, the conformations have not been as well characterized for the gastric H ϩ ,K ϩ -ATPase as for the Na ϩ ,K ϩ -ATPase or Ca 2ϩ -ATPase, Fe 2ϩ -catalyzed cleavage has turned out to be a valuable tool, revealing general mechanistic principles and structural organization of P-type pumps, as well as effects of H ϩ ,K ϩ -ATPase-specific ligands such as SCH28080. The similarity of cleavage of H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase shows that similar structural changes accompany E 1 7 E 2 transitions in these two pumps. As discussed in detail below, the structural changes inferred from cleavages of H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase fit very well with the Ca 2ϩ -ATPase crystal structure in the E 1 conformation (1EUL) and the theoretical model in an E 2 conformation (1FQU).

Location of Cleavage Sites
Assignment of cleavage sites on H ϩ ,K ϩ -ATPase depends on an accurate comparison of fragments with those of Na ϩ ,K ϩ -ATPase. For this purpose it was necessary to prepare antibodies that recognize the C terminus of the ␣ subunit of H ϩ ,K ϩ -ATPase, anti-DQELYY, which is analogous to anti-KETYY used with Na ϩ ,K ϩ -ATPase. Thus the fragments of both H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase extend from their respective N termini to the C terminus of the ␣ subunits, and can be com-pared. Because the mobility of all fragments of H ϩ ,K ϩ -ATPase was indistinguishable from that of comparable fragments of Na ϩ ,K ϩ -ATPase, the obvious inference is that the two proteins are cleaved at equivalent positions. Nevertheless, because the two ␣ subunits have only 65% sequence identity, a similar electrophoretic mobility of fragments does not necessarily imply equivalent cleavage positions. Table I compares the theoretical masses of the fragments of the two ␣ subunits based on their known sequences, assuming that they are cleaved at the same positions. For this comparison it is irrelevant that the exact cleavage position of Na ϩ ,K ϩ -ATPase is known only for two fragments (ESE and VNDS), whereas the others are known only approximately (near HFIH, near VAGDA, and near MVTGD respectively). The predicted masses for the pairs of fragments are all within Ϸ0.5 kDa of each other, differences that are undetectable with the gel system. Thus, the calculation suggests that H ϩ ,K ϩ -ATPase and Na ϩ ,K ϩ -ATPase are indeed cleaved at equivalent sites. That conclusion is much strengthened by the findings that ligands that stabilize the same conformations of both pumps have identical effects on appearance or disappearance of fragments.
Structural Changes Accompanying E 1 7 E 2 Conformations: Ligand Sites-In this section we discuss the inferences based on cleavages in terms of the two structures of the Ca 2ϩ -ATPase: 1EUL, which represents an E 1 ⅐Ca-bound form, and 1FQU, the theoretical model of an E 2 conformation with the bound inhibitors, decavanadate and thapsigargin. The comparison shows that the cleavage data are fully compatible with the domain movements seen in the 1EUL and 1FQU structures, and also predict features of ligand sites that have not been detected in the crystal structures.
Structural Changes Predicted by Cleavages in Fe 2ϩ Site 1-The parallel appearance of the fragments at ESE, near MVTGD, and at VNDS upon transition from the E 1 to E 2 (Rb) conformations (Fig. 3A), or their parallel disappearance when the Fe 2ϩ was chelated by EDTA (Fig. 3C), or when Mg 2ϩ ions were added to enzyme in an E 2 (Rb) conformation (Fig. 4), provide evidence that they are the products of cleavage in the same site. Based on these and similar findings with Na ϩ ,K ϩ -ATPase, the A and P domains could be predicted to interact in the E 2 conformation, and residues at cleavage sites in the A and P domains should come into proximity with each other and with the bound Fe 2ϩ (26,27). Fig. 1 illustrates these movements of the domains and sequences at the level of the whole molecule, whereas Fig. 2 depicts the site in detail, based on the E 2 structure (1FQU). Close proximity, between 4 and 6.5 Å, is indeed found between Gly 182 in the A domain and the residues Asp 703 , Gly 626 , and Thr 353 in the P domain (SERCA numbering). Because the exact position of the Fe 2ϩ is not known, it has been drawn to be about equidistant from Gly 182 , Thr 353 , Asp 703 , and Gly 626 . Cleavages at ESE and VNDS are the most prominent, and the Fe 2ϩ may in fact be closer to these residues. In any event the model is consistent with the prediction based on the cleavages. It was noted previously that, in both of the exactly defined cleavage sites of Na, K-ATPase, glycine residues precede the N-terminal residues, Glu 214 and Val 712 , respectively (26). Interestingly, Fig. 2 shows that glycines (highlighted in white) are found in all of the proximal sequences, either within the site (Gly 182 , Gly 626 ) or next to residues within the site (Gly 704 , Gly 352 ). The lack of a side chain on glycine residues may create the necessary space for an OH ⅐ radical to approach the polypeptide backbone. Thus, other cleavage sites may also contain glycine residues.
The experiment in Fig. 4, showing that a high concentration of Mg 2ϩ converted an E 2 -like to an E 1 -like cleavage pattern, can be interpreted to mean either that Mg 2ϩ ions directly compete with Fe 2ϩ ions in site 1, or that Mg 2ϩ ions convert the E 2 (Rb) conformation to an E 1 ⅐Mg conformation in which the Fe 2ϩ site 1 is disrupted. On the basis of this experiment alone, it cannot be decided whether the Fe 2ϩ in site 1 occupies a Mg 2ϩ binding site. However, the other evidence, discussed below, clearly indicates that the cleavage sites at TGES and TG-DGVNDS include residues that bind Mg 2ϩ ions in the presence of ATP or phosphate. Thus, Fe 2ϩ site 1, in the absence of ATP or phosphate, can be assumed to contain elements of the normal Mg 2ϩ site.
Cleavages Catalyzed by Fe 2ϩ in Site 2-Proximity of transmembrane segments M3 and M1, predicted from the cleavages of H ϩ ,K ϩ -ATPase or Na ϩ ,K ϩ -ATPase catalyzed by Fe 2ϩ in site 2 is shown in Fig. 1 (left). A homology model, in which transmembrane sequences of Ca 2ϩ -ATPase were replaced by those of Na ϩ ,K ϩ -ATPase, also showed the predicted proximity of the sequence HFIH (M3) and EWVK (M1) (31). The cleavage near M3 (HFVD) was seen clearly for H ϩ ,K ϩ -ATPase, but that near M1 was seen only in some experiments (Fig. 5B), and it is also a less prominent fragment with Na ϩ ,K ϩ -ATPase. Cleavages at site 2 are not affected by E 1 and E 2 conformations (Figs. 3), and neither do Mg 2ϩ ions compete with the Fe 2ϩ ions (Fig. 4). The similar cleavages of both Na ϩ ,K ϩ -ATPase and H ϩ ,K ϩ -ATPase and fit of the prediction with the structure of Ca 2ϩ -ATPase provide direct evidence for a similar organization of M3 and M1 in all three pumps.
Effects of Phosphate, Vanadate, Mg 2ϩ Ions, and SCH28080 on Cleavages in Fe 2ϩ Site 1-The paradoxical effects of P i , P i /Mg 2ϩ /SCH28080, or vanadate/Mg 2ϩ presented in Fig. 5, and similar effects noted previously with Na ϩ ,K ϩ -ATPase, can be explained by assuming that the P i , vanadate, or Mg 2ϩ ions directly interfere with Fe 2ϩ binding in site 1 (see Ref. 27). It was proposed that noncovalently bound phosphate interacts with Na ϩ ,K ϩ -ATPase at Asp 369 and in the 607 MVTGD sequence, displacing Fe 2ϩ from these contacts, but leaving the Fe 2ϩ in contact with the 708 TGDGVNDS and 212 TGES sequences (27). The result would be selective suppression of the cleavages near Asp 369 and near 607 MVTGD for Na ϩ ,K ϩ -ATPase or 624 MVTGD for H ϩ ,K ϩ -ATPase (Fig. 5A). The simultaneous amplification of cleavages at the ESE and VNDS sites (Fig. 5A) may indicate that the bound phosphate interacts with Fe 2ϩ in site 1 and induces tighter binding of the Fe 2ϩ , which is not saturating in these conditions. The combination of phosphate, Mg 2ϩ ions, and ouabain for Na ϩ ,K ϩ -ATPase or phosphate, Mg 2ϩ , and SCH28080 with H ϩ ,K ϩ -ATPase (see Ref. 53) should stabilize the phosphoenzyme E 2 -P⅐Mg. We suggested that the bound Mg 2ϩ ions interact with residues in the TG-DGVNDS and TGES sequences and bound phosphate and so prevents binding of Fe 2ϩ to site 1 (27). Similarly, the combination of vanadate and Mg 2ϩ ions should induce a stable complex, E 2 ⅐vanadate⅐Mg, in which Fe 2ϩ binding to site 1 would be hindered. Interference with Fe 2ϩ binding in only site 1 should leave Fe 2ϩ binding to site 2 intact and thus produce the E 1 -like cleavage pattern, despite the E 2 -stabilized conformations.
The hypothesis that phosphate interacts near MVTGD, that Mg 2ϩ ions interact within the TGDGVNDS and TGES sequences, and that Mg 2ϩ ions interact with bound phosphate or vanadate in E 2 -P or E 2 -vanadate conformations cannot be tested by reference to either the Ca 2ϩ -ATPase crystal structure or the E 2 theoretical model, because neither include Mg 2ϩ ions or phosphate. Recently, however, the atomic structures of two proteins of the HAD superfamily, phosphoserine phosphatase (PDB identification codes 1F5S and 1J97) (39,40) and phosphonoacetaldehyde hydrolase (PDB identification code 1FEZ) (41) have been determined in the presence of Mg 2ϩ ions and the product PO 4 3Ϫ (39) or the phosphate analogues, BF 3 Ϫ (40) and WO 4 2Ϫ (41). In the HAD superfamily, which includes the P-type ATPases, critical active site residues are conserved. Fig. 8 shows a similar spatial organization of five crucial residues Asp 351 , Thr 625 , Lys 684 , Asp 703 , and Asp 707 of Ca 2ϩ -ATPase (1EUL) and the corresponding residues Asp 11 , Ser 99 , Lys 144 , Asp 167 , and Asp 171 of phosphoserine phosphatase (1F5S) and, in the latter case, phosphate and hydrated Mg 2ϩ ion. The phosphate is in close contact with the active site Asp 11 and also with Ser 99 and Lys 144 with which it interacts by H-bonds, and the Mg 2ϩ ion interacts closely with Asp 11 and Asp 167 and the bound phosphate. A similar arrangement of WO 4 2Ϫ and Mg 2ϩ ions is seen in phosphonatase (41). Thus, the structure of these proteins with a related active site structure provides strong support for our hypothesis that phosphate interacts within the MVTGD sequence, Mg 2ϩ ions interact within the TGDGVNDS sequence, and bound phosphate and Mg 2ϩ interact with each other. Mutagenesis experiments on Na ϩ ,K ϩ -ATPase have also provided evidence that residues Asp 710 and Asn 713 in the TG-DGVNDS sequence are important for Mg 2ϩ binding in the E 1 conformation (54). The experiments were interpreted to exclude such a role for TGDGVNDS in E 2 , but relevant residues in E 2 were not identified. The prediction that Mg 2ϩ ions bind within the TGES sequence in the E 2 conformation cannot be tested by reference to the phosphoserine phosphatase and phosphonatase structures because these proteins lack an A domain and do not undergo the equivalent of E 1 7 E 2 transitions (see the next section).
Cleavage Mediated by the ATP-Fe 2ϩ Complex-Cleavages reveal points of contact of the Fe 2ϩ (or Mg 2ϩ ) ions when complexed with ATP, and we can now presume that these are the same for Na ϩ ,K ϩ -ATPase and H ϩ ,K ϩ -ATPase. High affinity binding of the ATP-Fe 2ϩ complex in E 1 conformations produces a major cleavage at the valine residue within the conserved TGDGVNDS sequence in the P domain, and less prominent cleavages in the N domain, very near or within the sequence 440 VAGDA (Na ϩ ,K ϩ -ATPase) or 456 VIGDA (H ϩ ,K ϩ -ATPase) (28, and see Fig. 6). The A domain is not in contact with bound ATP-Fe 2ϩ and is not cleaved. In the Ca 2ϩ -ATPase crystal structure, 1EUL, ATP is not present and the N domain containing the nucleotide binding pocket is far from the P domain containing the phosphorylated Asp 351 . When ATP-Mg 2ϩ is bound, the N and P domains must interact so as to allow the ␥-phosphate of ATP to approach the active site aspartate. Cleavage of Na ϩ ,K ϩ -ATPase and H ϩ ,K ϩ -ATPase mediated by the ATP-Fe 2ϩ complex now provides direct evidence for an N to P interaction. The prediction is that Mg 2ϩ bound to ATP is in close proximity to both TGDGVNDS in the P domain and 456 VIG-DASES (H ϩ ,K ϩ -ATPase) and 440 VAGDASES (Na ϩ ,K ϩ -ATPase) (or 437 VGEATET, the corresponding sequence of Ca 2ϩ -ATPase, in the N domain; see Table I). As seen in Fig. 1, Glu 439 of Ca 2ϩ -ATPase lies at the outer end of an ␣ helix that lines the ATP binding pocket, at the bottom of which lies Lys 515 , which can labeled with fluorescein 5-isothiocyanate. Evidently, this position of Glu 439 is such that, by tilting the N domain toward the P domain, it could well come into proximity with Asp 703 .
Fe 2ϩ ions are able to substitute for Mg 2ϩ ions in catalyzing phosphorylation and Na ϩ ,K ϩ -ATPase activity (51,52). As we showed recently (28), in the presence of Na ϩ ions and the ATP-Fe 2ϩ complex, the E 2 -P conformation and tightly bound Fe 2ϩ ions can be isolated kinetically, and in this state a major cleavage occurs near TGES in the A domain, whereas that at VNDS in the P domain is less prominent. The experiment in Fig. 7 showing cleavage of H ϩ ,K ϩ -ATPase at TGES implies that Fe 2ϩ can also substitute for Mg 2ϩ in catalyzing phosphorylation of the H ϩ ,K ϩ -ATPase, the Fe 2ϩ bound in E 2 P then catalyzing cleavage at TGES. Analysis of cleavages of Na ϩ ,K ϩ -ATPase or H ϩ ,K ϩ -ATPase by the ATP-Fe 2ϩ complex leads to the following concepts. First, in either E 1 or E 1 P conformations, the N domain docks onto the P domain, whereas the A domain is displaced to one side. The residues important for binding Mg 2ϩ (Fe 2ϩ ) ions lie within TGDGVNDS (P domain) and near VAGDA (N domain). Second, in the E 1 P 7 E 2 P conformational transition, the N to P interaction is disrupted by formation of A to P and A to N interactions. In E 2 P tightly bound Mg 2ϩ (Fe 2ϩ ) ions make contact both with TGES (A domain) and TG-DGVNDS (P domain) and the bound phosphate. We have argued that this change in Mg 2ϩ ligation is crucial in the change of reactivity of the phosphoenzymes from ADP sensitivity in E 1 P to water sensitivity in E 2 P (28).
The direct evidence for interaction of Fe 2ϩ or Mg 2ϩ ions with residues within TGES, TGDGVNDS, and the bound phosphate, based on cleavages of E 2 -P with tightly bound Fe 2ϩ , complements and greatly strengthens the same hypothesis based on inhibition of Fe 2ϩ -catalyzed cleavages in site 1 by P i /Mg 2ϩ / ouabain, P i /Mg 2ϩ /SCH28080, or vanadate/Mg 2ϩ , described above (27). Although the 1FQU theoretical model of the Ca 2ϩ -ATPase does not contain either Mg 2ϩ or phosphate ions, the close proximity of TGES to both TGDGVNDS and Asp 351 (Fig.  2) shows that the experimental evidence is quite compatible with the structure.

Conclusions
The major conclusion of this work is that the structural organization and changes in the cytoplasmic domains, associated with the E 1 7 E 2 transitions, are essentially the same for Na ϩ ,K ϩ -ATPase, H ϩ ,K ϩ -ATPase, and Ca 2ϩ -ATPase. Thus, it is highly likely that these features are common to all other P-type pumps. Presumably, the cytoplasmic domain interactions accompanying E 1 7 E 2 transitions are then relayed via the extended stalk helices of M4 and M5, to the trans-membrane segments in which the transported cations are occluded. Changes in tilt, or turn or position of the trans-membrane helices, driven by changes in the N, P, and A domains, may allow the cations to dissociate at the opposite surface from that at which they were occluded. A necessary corollary of the similarity of the organization and structural changes of the cytoplasmic domains, in the different pumps, is that their differences, namely cation specificity and inhibitor sensitivity, lie primarily in the trans-membrane segments and extracellular loops.