Critical Role of Glu40-Ser48 Loop Linking Actuator Domain and First Transmembrane Helix of Ca2+-ATPase in Ca2+ Deocclusion and Release from ADP-insensitive Phosphoenzyme*

The functional importance of the length of the A/M1 linker (Glu40-Ser48) connecting the actuator domain and the first transmembrane helix of sarcoplasmic reticulum Ca2+-ATPase was explored by its elongation with glycine insertion at Pro42/Ala43 and Gly46/Lys47. Two or more glycine insertions at each site completely abolished ATPase activity. The isomerization of phosphoenzyme (EP) intermediate from the ADP-sensitive form (E1P) to the ADP-insensitive form (E2P) was markedly accelerated, but the decay of EP was completely blocked in these mutants. The E2P accumulated was therefore demonstrated to be E2PCa2 possessing two occluded Ca2+ ions at the transport sites, and the Ca2+ deocclusion and release into lumen were blocked in the mutants. By contrast, the hydrolysis of the Ca2+-free form of E2P produced from Pi without Ca2+ was as rapid in the mutants as in the wild type. Analysis of resistance against trypsin and proteinase K revealed that the structure of E2PCa2 accumulated is an intermediate state between E1PCa2 and the Ca2+-released E2P state. Namely in E2PCa2, the actuator domain is already largely rotated from its position in E1PCa2 and associated with the phosphorylation domain as in the Ca2+-released E2P state; however, in E2PCa2, the hydrophobic interactions among these domains and Leu119/Tyr122 on the top of second transmembrane helix are not yet formed properly. This is consistent with our previous finding that these interactions at Tyr122 are critical for formation of the Ca2+-released E2P structure. Results showed that the EP isomerization/Ca2+-release process consists of the following two steps: E1PCa2 → E2PCa2 → E2P + 2Ca2+; and the intermediate state E2PCa2 was identified for the first time. Results further indicated that the A/M1 linker with its appropriately short length, probably because of the strain imposed in E2PCa2, is critical for the correct positioning and interactions of the actuator and phosphorylation domains to cause structural changes for the Ca2+ deocclusion and release.

Sarcoplasmic reticulum Ca 2ϩ -ATPase (SERCA1a) 2 is a representative member of P-type ion transporting ATPases and catalyzes Ca 2ϩ transport coupled with ATP hydrolysis (Fig. 1) (Refs. 1, 2 and for recent reviews see Refs. [3][4][5][6][7]. In the catalytic cycle, the enzyme is activated by binding of two Ca 2ϩ ions to the transport sites (E2 to E1Ca 2 , steps 1 and 2) and then autophosphorylated at Asp 351 with MgATP to form the ADP-sensitive phosphoenzyme (E1P, step 3), which can react with ADP to regenerate ATP in the reverse reaction. Upon formation of this EP, the bound Ca 2ϩ ions are occluded in the transport sites (E1PCa 2 ). The subsequent isomeric transition to the ADP-insensitive form (E2P), i.e. the loss of the ADP sensitivity at the catalytic site, results in rearrangements of the Ca 2ϩ -binding sites to deocclude Ca 2ϩ , reduce the affinity, open the luminal gate, and thus release Ca 2ϩ into the lumen (steps 4 and 5). As an intermediate state in the EP isomerization/Ca 2ϩ -release process, E2PCa 2 has been postulated (e.g. see Ref. 8), although this state has never been identified. Finally, the E2P hydrolysis takes place and returns the enzyme into an unphosphorylated and Ca 2ϩ -unbound form (E2, steps 6 and 7). The transport cycle is totally reversible, e.g. E2P can be formed from P i in the presence of Mg 2ϩ and the absence of Ca 2ϩ by reversal of its hydrolysis, and the subsequent addition of high concentrations of Ca 2ϩ to E2P reverse the Ca 2ϩ -releasing step and the E1P to E2P isomerization.
The enzyme has three cytoplasmic domains as follows: the nucleotide binding (N), phosphorylation (P), and actuator (A) domains, and 10 transmembrane helices M1-M10 (Fig. 2). During the EP isomerization/Ca 2ϩ -release E1PCa 2 3 E2P ϩ 2Ca 2ϩ , the A domain largely rotates (by ϳ110°) parallel to the membrane and associates with the P domain (see Refs. 9 -17) (see E1⅐AlF x ⅐ADP (the E1PCa 2 ⅐ADP analog) 3 E2⅐MgF 4 2Ϫ (the E2⅐P i analog) in Fig. 2). The interactions of the A domain with the P domain in the E2P state occur at three regions (Fig. 2, semitransparent purple, blue, and orange on E2⅐Mg F 4 2Ϫ ): i.e. at the T 181 GES loop with the residues of the P domain around Asp 351 ; at the Val 200 loop (Asp 196 -Asp 203 ) with the polar residues of the P domain (Arg 678 /Glu 680 /Arg 656 /Asp 660 ); and at the Tyr 122 -hydrophobic cluster formed by seven hydrophobic residues gathered from the A domain (Ile 179 /Leu 180 /Ile 232 ), the P domain (Val 705 /Val 726 ), and the top part of M2 (the A/M2 linker region, Leu 119 /Tyr 122 ). The formation of the A-P domain interaction at the T 181 GES loop has been predicted to be critical for the loss of ADP sensitivity at the catalytic site, i.e. the E1P to E2P isomerization, by the structural and mutation studies (18 -20). The mutations at the latter two interaction regions were shown not to inhibit the E1P to E2P isomerization but to markedly retard the subsequent EP decay (19,21,22). Its kinetics were consistent with the view that there is a Ca 2ϩreleasing step from E2PCa 2 (E2PCa 2 3 E2P ϩ 2Ca 2ϩ ) before the E2P hydrolysis and that this Ca 2ϩ -releasing step is blocked and became the kinetic limit for the EP decay by the disruption of the A-P domain interactions at each of the latter two regions (19,21,22). It is therefore very interesting to know how the motions and interactions of the A and P domains progress during the postulated successive steps E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ as the key structural events in the energy coupling between the cytoplasmic and transmembrane domains. In this respect, it is also critical to clarify and distinguish the structural roles of the three linkers connecting the A domain with M1Ј/M1, M2, and M3 (A/M1, A/M2, and A/M3 linkers). Tyr 122 /Leu 119 involved in the aforementioned Tyr 122 -hydrophobic cluster is at the A/M2 linker region. The A/M3 linker, because of its strain, has been predicted to be important for the large rotation of the A domain in the EP isomerization (12,13).  Regarding the A/M1 linker, we recently found (23) that its shortening by deletions of any single residues within this linker (Glu 40 -Ser 48 ) blocks the E1P to E2P isomerization and the hydrolysis of the Ca 2ϩ -free form of E2P, whereas substitutions of any residues in this linker do not inhibit the function. Our results indicated that the A/M1 linker with its correct length critically contributes to the EP isomerization/Ca 2ϩ release and to the E2P hydrolysis, and we pointed out the possible importance of this linker in the proper positioning of the A and P domains for their motions and association during these processes. Therefore, in this study, we further explored the structural roles of this linker and the structural events occurring in the processes by elongating this linker with insertion of glycines (see Fig. 2).
Results demonstrated that the elongation of the linker markedly accelerates the E1PCa 2 to E2PCa 2 isomerization, strongly stabilizes E2PCa 2 that possesses two occluded Ca 2ϩ ions at the transport sites, and blocks the Ca 2ϩ deocclusion and release from E2PCa 2 . Thus, for the first time, the intermediate state E2PCa 2 was identified and trapped in this study. We were then able to characterize the structure of this state. Results revealed that the correct length of the A/M1 linker is critical for structural events in each of successive steps in E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ and E2P ϩ H 2 O 3 E2 ϩ P i , and they further suggested how the motions and interactions of the properly positioned A and P domains progress with the critical contribution of the linker to accomplish the successive structural events in these steps. Our study also revealed the importance of M1Ј directly connected with the A/M1 linker likely for forming the base of this linker.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression-The QuikChange TM site-directed mutagenesis method (Stratagene, La Jolla, CA) was utilized for the insertions and substitutions of residues in the rabbit SERCA1a cDNA. The ApaI-KpnI restriction fragments with the desired mutation were excised from the plasmid and ligated back into the corresponding region in the full-length SERCA1a cDNA in the pMT2 expression vector (24). The pMT2 DNA was transfected into COS-1 cells by the liposome-mediated transfection method. Microsomes were prepared from the cells as described previously (25). The "control microsomes" were prepared from COS-1 cells transfected with the pMT2 vector containing no SERCA1a cDNA. The amount of expressed SERCA1a was quantified by a sandwich enzyme-linked immunosorbent assay as described previously (26). Expression levels of the wild-type SERCA1a and mutants examined in this study were 2-3% of total microsomal proteins, except a mutant 4950525354S for M1Ј with serine substitutions of Leu 49 /Trp 50 / Leu 52 /Val 53 /Ile 54 , which showed markedly reduced expression (only ϳ15-20% of the wild type).
Ca 2ϩ -ATPase Activity-The rate of ATP hydrolysis was determined at 25°C in a mixture containing 1 g of microsomal protein, 0.1 mM [␥-32 P]ATP, 1 M A23187, 0.1 M KCl, 7 mM MgCl 2 , 0.55 mM CaCl 2 , 0.5 mM EGTA, and 50 mM MOPS/Tris (pH 7.0). The Ca 2ϩ -ATPase activity was obtained by subtracting the Ca 2ϩ -independent ATPase activity, which was determined in the presence of 5 mM EGTA without added CaCl 2 , otherwise as above. The ATPase activity/mg of expressed SERCA1a protein was cal-culated from the amount of expressed SERCA1a and the Ca 2ϩ -ATPase activity of expressed SERCA1a, which was obtained by subtracting the Ca 2ϩ -ATPase activity of the control microsomes from that of the microsomes expressing SERCA1a. This background level with the control microsomes was as low as 3% of the activity of microsomes expressing the wild-type SECRA1a.
Formation and Hydrolysis of EP-Phosphorylation of SERCA1a in microsomes with [␥-32 P]ATP or 32 P i and dephosphorylation of 32 P-labeled SERCA1a were performed under conditions described in the figure legends. The reactions were quenched with ice-cold trichloroacetic acid containing P i . Rapid kinetics measurements of phosphorylation and dephosphorylation were performed with a handmade rapid mixing apparatus (27), otherwise as above. The precipitated proteins were separated by 5% SDS-PAGE at pH 6.0 according to Weber and Osborn (28). The radioactivity associated with the separated Ca 2ϩ -ATPase was quantitated by digital autoradiography as described (29). The amount of EP formed with the expressed SERCA1a was obtained by subtracting the background radioactivity with the control microsomes. This background was less than 5% of the radioactivity of EP formed with the expressed wild-type SERCA1a. The amount of EP/mg of SERCA1a protein was calculated from the amount of EP thus obtained and the amount of expressed SERCA1a.
Ca 2ϩ Occlusion in EP-As described in Fig. 8 legend, the expressed mutant SERCA1a in microsomes was phosphorylated with ATP and 45 CaCl 2 , and then the mixture was diluted by a "washing solution" containing excess EGTA and immediately filtered through a 0.45-m nitrocellulose membrane filter (Millipore). The filter was washed four times with 2 ml of the washing solution, and 45 Ca 2ϩ remaining on the filter was quantitated. The amount of Ca 2ϩ specifically bound to the transport sites of EP in the expressed SERCA1a was obtained by subtracting the amount of nonspecific Ca 2ϩ binding, which was determined by including 1 M thapsigargin (TG) in the phosphorylation mixture, otherwise as above. This background subtraction is ensured by the fact that TG inhibits the Ca 2ϩ binding at the transport sites and the EP formation (30). The background level thus determined was ϳ60% of the total amount of 45 Ca 2ϩ remaining on the filter when the maximum amount of EP was present (i.e. at the zero time of EP decay in Fig. 8). It should be noted that the specifically bound Ca 2ϩ in EP thus determined represents the occluded one because it is not released even after the extensive washing by EGTA. The Ca 2ϩ occluded/mg of expressed SERCA1a protein was calculated from the amount of expressed SERCA1a and the amount of occluded Ca 2ϩ . The Ca 2ϩ occlusion resulted from Ca 2ϩ binding to E2P in the reverse reaction of the Ca 2ϩ -release process was also determined. In this case, E2P was first formed from P i in the absence of Ca 2ϩ , and 45 Ca 2ϩ was then added to E2P otherwise as described in Fig. 10 legend, and the amount of occluded 45 Ca 2ϩ was determined as above.
Limited Proteolysis of Major Intermediates and Western Blot Analysis-Major intermediates and its stable analogs of the Ca 2ϩ -ATPase were produced and subjected to the structural analysis by limited proteolysis with trypsin and proteinase K (PrtK) as described in Fig. 12 legend. The digests were separated by 10.5 or 7.5% SDS-PAGE, according to Laemmli (31), and blotted onto a polyvinylidene fluoride membrane and then incubated with IIH11 monoclonal antibody to the rabbit SERCA1a (Affinity Bioreagents), which recognizes an epitope between Ala 199 -Arg 505 . After incubation with secondary antibody (goat anti-mouse IgG-horseradish peroxidase-conjugated), the bound proteins were probed using an enhanced chemiluminescence-linked detection system (ECL Plus, GE Healthcare).
Miscellaneous-Protein concentrations were determined by the method of Lowry et al. (32) with bovine serum albumin as a standard. Data were analyzed by nonlinear regression using the program Origin (Microcal Software, Inc., Northampton, MA). Three-dimensional models of the enzyme were reproduced by the program VMD (33).

RESULTS
Ca 2ϩ -ATPase Activity-The specific Ca 2ϩ -ATPase activity of the expressed SERCA1a mutants was determined at saturating 50 M Ca 2ϩ and 25°C and compared with that of the wild type (Fig. 3). Insertion of one glycine between Gly 46 and Lys 47 (1Gi-46/47) and between Pro 42 and Ala 43 (1Gi-42/43) within the A/M1 linker slowed the ATPase activity by ϳ50%. Insertion of two or more glycines at each of these sites abolished the activity almost completely (2Gi-46/47 to 6Gi-46/47 and 4Gi-42/43). Thus, the elongation of the linker by the glycine insertion at the two different positions within the A/M1 linker exhibited the same inhibitory effects on the ATPase, indicating the importance of the correct length of this loop in the function. 3 We then examined the possible effects of the 4-amino acidinsertion in the C-and N-terminal regions of the A/M1 linker: between Val 53 and Ile 54 on the helix M1Ј and between Thr 22 and Gly 23 in the immediate vicinity of the Thr 25 -Tyr 36 helix (see Fig. 2b). In the mutants 4Ai-53/54 and 4Ai-22/23, alanines were inserted (for M1Ј, we intended to minimize possible disruption of the helical structure). These mutants exhibited the high ATPase activity (Fig. 3) and the Ca 2ϩ transport coupled with the ATP hydrolysis (data not shown). Thus the insertions at the adjacent regions of the A/M1 linker did not inhibit the activity. We inserted amino acids also in the immediate N-terminal region of the A/M1 linker, for example at His 32 -Leu 33 and at Gly 37 -His 38 ; however, the protein expression levels of these mutants were extremely low (less than ϳ10% of the wild type); therefore, their functional analysis was not possible.
We also investigated the possible importance of amphipathic property of the helix M1Ј (Trp 50 -Glu 58 in E1⅐AlF x ⅐ADP or Leu 49 -Gln 56 in E2⅐MgF 4 2Ϫ ), which is directly connected with the A/M1 linker and formed by kinking of M1. M1Ј lies on the membrane surface, having hydrophobic residues aligned on the membrane side (Leu 49 /Trp 50 /Val 53 /Ile 54 ) and the polar residues (Glu 51 /Glu 55 /Gln 56 /Glu 58 ) on the cytoplasmic side ( Fig.  2b). Therefore, the hydrophobic interactions of M1Ј with the membrane core and/or its hydrophilic interactions at the membrane surface may possibly be important for function (13). Previously the mutations of the single residues on the M1Ј region were found to have almost no or only a slight effect on the activity (23,34). In this study, we therefore introduced the extensive nonconservative substitutions as follows: the serine substitution of all Leu 49 /Trp 50 /Leu 52 / Val 53 /Ile 54 (4950525354S) and the alanine substitution of all Glu 51 /Glu 55 /Gln 56 /Glu 58 /Asp 59 (5155565859A). The mutant 4950525354S exhibited the markedly reduced ATPase activity (17% that of the wild type), whereas the mutant 5155565859A exhibited a fairly high activity (70% that of the wild type). The results indicate that the hydrophobic interaction of M1Ј with the membrane core may be important.
EP Formation from ATP and the E2-E1Ca 2 Transition-The amount of EP formed from ATP at saturating 50 M Ca 2ϩ was determined at steady state and 0°C with 10 M ATP under the   conditions otherwise the same as those for the ATPase assay (open bars in Fig. 4). All the mutants formed EP with the amounts comparable with that of the wild type (3.31 Ϯ 0.14 nmol/mg of the SERCA1a protein (n ϭ 4)) except that the mutant 4950525354S exhibited a somewhat reduced amount. The Ca 2ϩ affinity of the mutants in the E2 to E1Ca 2 transition was estimated by the Ca 2ϩ dependence of the EP formation from ATP and was found to be nearly the same as that of the wild type (see the Ca 2ϩ affinity and the Hill coefficient in Table  1). The mutant 4Ai-53/54 among those examined showed a slightly reduced affinity. We further found that the first-order rate constants of the E2 to E1Ca 2 transition (steps 1 and 2) in the examined mutants were nearly the same as in the wild type (Table 1). In this experiment, the enzyme was first preincubated in the absence of Ca 2ϩ at pH 6, where the equilibrium between E1 and E2 is most shifted to E2 (35), and then the phosphorylation was initiated by the simultaneous addition of saturating Ca 2ϩ and ATP. When ATP was added to the enzyme preincubated with Ca 2ϩ , otherwise as above, the EP formation was much faster in the mutants as well as in the wild type, and therefore the rates obtained above actually reflect the rate-limiting E2 to E1Ca 2 transition. Loss of ADP Sensitivity in EP-The E1P to E2P isomerization, i.e. the loss of the ADP sensitivity at the catalytic site, was analyzed at 0°C under the conditions otherwise the same as those for the ATPase assay. In Fig. 4 (closed bars), the amount of ADP-insensitive EP (E2P) accumulated in the steady state was determined 60 s after the addition of ATP in the presence of K ϩ , which strongly accelerates the hydrolysis of E2P and thus suppresses its accumulation in the wild type (36). In the mutants 1Gi-46/47 and 1Gi-42/43, the amount of the accumulated ADP-insensitive EP was very low as in the wild type, and it was markedly increased in the mutants with two or more glycine insertions in the A/M1 linker, 2Gi-46/47 to 6Gi-46/47 and 4Gi-

TABLE 1 Kinetic parameters determined for partial reaction steps
The affinity of the transport sites for Ca 2ϩ (K 0.5 ) and the Hill coefficient (n) in the E2 to E1Ca 2 transition was determined at 0°C by the EP formation from ATP in the presence of various concentrations of Ca 2ϩ under the conditions otherwise as described in the legend to Fig. 4 and by the least squares fit to the Hill equation. The rates of the E2 to E1Ca 2 transition in steps 1 and 2 were determined by the E1PCa 2 formation from the E2 state upon the simultaneous addition of saturating 100 M Ca 2ϩ and ATP at pH 6.0 under the conditions otherwise as in Fig. 4. In this EP formation, the E2 to E1Ca 2 transition is rate-limiting. The rates for the other steps were obtained at 0°C in the experiments in Fig. 5, A and C (E1Ca 2 to E1PCa 2 in step 3), Fig. 5, B and D (loss of ADP sensitivity (i.e. accumulation of ADP-insensitive EP from ADP-sensitive EP in step 4)), Fig. 6 (decay of EP formed from ATP in the presence of Ca 2ϩ (EP ATP )), and Fig. 7 (hydrolysis of E2P formed from P i in the absence of Ca 2ϩ (E2P Pi ) in steps 6 and 7). In parentheses, the values obtained with the wild type are normalized to 100%. For E2P to E2PCa 2 /E1PCa 2 , the accessibility of luminal Ca 2ϩ to the transport sites of E2P was assessed at 0°C in Fig. 10 by determining the affinity for Ca 2ϩ (K 0.5 ) and the Hill coefficient (n) in the reverse reaction, i.e. upon the addition of Ca 2ϩ to E2P and the consequent formation of E2PCa 2 or E1PCa 2 (see "Results" with Fig. 10 for details of the formation of these stable EP species in the reverse reaction). E1Ca   a The rate of the E1PCa 2 formation from E1Ca 2 in the presence of K ϩ was very similar to that in the absence of K ϩ (see Fig. 5, A and C) and therefore not shown for simplicity. b Not determined because the accumulation of ADP-insensitive EP was low. c The rate of the EP isomerization (loss of the ADP-sensitivity) must be faster because almost all of EP formed had become already ADP-insensitive during the EP formation (see Fig. 5). d Not determined because the EP formation from E1Ca 2 was slow (see Fig. 5A). e The apparent slow rate is probably due to the slowed EP formation from E1Ca 2 and ATP. The rate of the EP isomerization (loss of the ADP sensitivity) must be faster because a very large fraction of EP formed had become already ADP-insensitive during the EP formation (see Fig. 5).

E2 to
42/43 (compare with the total amount of EP shown in Fig. 4, open bar). In the mutants with the three or more glycine insertion, nearly all of EP was ADP-insensitive. The mutant 4950525354S for M1Ј also accumulated the ADP-insensitive EP almost exclusively. On the other hand, in the mutants 5155565859A for M1Ј and 4Ai-22/23 and 4Ai-53/54 for the adjacent regions of the A/M1 linker, the accumulation of ADPinsensitive EP was very low as in the wild type. The observed markedly enhanced accumulation of ADP-insensitive EP in the mutants suggests that the E1P to E2P isomerization was accelerated and/or that the rate of one of the reverse transitions, e.g. from E2PCa 2 to E1PCa 2 , was dramatically reduced. It is also possible that the decay of ADP-insensitive EP was blocked. In Fig. 5, A and B, the time course of accumulation of ADP-insensitive EP from E1Ca 2 and ATP was determined in the absence of K ϩ , in which the wild type accumulates a fair amount of ADP-insensitive EP and thus the kinetics can be compared. The phosphorylation from E1Ca 2 , i.e. the formation of ADP-sensitive EP (E1P) from E1Ca 2 , occurred very rapidly, and thus the total amount of EP reached its maximum level within a few seconds except for the slowed mutant 4950525354S (Fig. 5A). The accumulation of ADP-insensitive EP apparently proceeded with first-order kinetics. The rate and extent of the accumulation in the mutants 1Gi-46/47, 2Gi-46/47 (Fig. 5B), and 1Gi-42/43 (data not shown) in the absence of K ϩ were comparable with those of the wild type (see Table 1 for the rates). The accumulation of ADP-insensitive EP became extremely rapid in the mutants with three or more glycine insertions, 3Gi-46/47 and 4Gi-46/ 47. Actually, during the time course of EP formation in these mutants, nearly all of EP formed was already ADP-insensitive. Essentially the same results were obtained with the mutant 4Gi-42/43 as with these mutants (data not shown, but see Table 1). Thus in the mutants with the elongated A/M1 linker with the three or more glycine insertions, the EP isomerization was markedly accelerated, and the ADP-insensitive EP (E2P) accumulated exclusively. In these mutants, the rates of EP formation and the EP isomerization in the presence of K ϩ (Fig. 5, C and D) were almost the same as those in the absence of K ϩ (Fig. 5, A and B). In the mutant 2Gi-46/47, the fair amount of ADP-insensitive EP accumulated rapidly even in the presence of K ϩ (see Figs. 4 and 5D).
It should be noted that the mutant 4950525354S exhibited the almost exclusive accumulation of ADP-insensitive EP both in the absence (Fig. 5B) and presence ( Fig. 4) of K ϩ at steady state. In this regard, the removal of the hydrophobic property of M1Ј by the serine substitutions (removal of the likely hydrophobic interaction with the membrane core) caused the same consequence as that of the elongation of the A/M1 linker. Besides, the observed slow EP formation from E1Ca 2 and ATP in this mutant 4950525354S (Fig. 5A) suggests the importance of the hydrophobic property of this region in rapid structural changes for ATP binding and phosphorylation. The mutants 5155565859A on M1Ј and 4Ai-22/23 and 4Ai-53/54 adjacent to the A/M1 linker exhibited the extent and rate of the accumulation of ADP-insensitive EP almost the same as those of the wild type (see Table 1 for the rates, and Figs. 4 and 5B for the extent).
Decay of EP Formed from ATP and Ca 2ϩ -The decay of EP formed from ATP and Ca 2ϩ was determined at 0°C in the presence of K ϩ and is shown with the representative mutants in Fig. 6. The decay time courses were fitted well with a single exponential (Fig. 6A), and the rates were summarized in Table  1. The fraction of the ADP-insensitive EP (E2P) remaining in the decay course was also determined (Fig. 6B). In the wild type, the EP remaining was exclusively the ADP-sensitive EP (E1P), and this is consistent with the well known rate-limiting E1P to E2P transition in the ATPase cycle (36,37). The EP decay was slightly slowed in the single glycine-insertion mutants 1Gi-46/47 (Fig. 6A) and 1Gi-42/43 (see Table 1), being consistent with the slight reduction in the ATPase activity in these mutants (cf. Fig. 3). The EP decay was almost completely blocked in the mutants with two or more glycine insertions in the A/M1 linker, 2Gi-46/47 to 4Gi-46/47 (Fig. 6A) and 4Gi-42/43 (see Table 1). This is consistent with the complete loss of the ATPase activity in these mutants. EP present at the start and course of the decay reaction was exclusively the ADP-insensitive one (E2P) in the mutants 3Gi-46/47 and 4Gi-46/47, and ϳ50% in the mutant 2Gi-46/47 (Fig. 6B, as also shown in Figs. 4 and 5D). In the mutants with the elongated A/M1 linker with the two or more glycine insertions, these results show that the E1P-E2P isomerization was strongly shifted toward the ADPinsensitive EP (E2P) and that the decay of the ADP-insensitive EP was blocked. In the mutant 4950525354S on M1Ј, the EP decay was markedly slowed (Fig. 6A), and the EP remaining was almost exclusively the ADP-insensitive EP during the decay reaction (data not shown, but see Fig. 4). Thus in this mutant, the decay of the ADP-insensitive EP was markedly retarded. The mutants 5155565859A (Fig. 6A), 4Ai-22/23, and 4Ai-53/54 (Table 1) showed the rapid EP decay as the wild type, being consistent with their high ATPase activities.
Hydrolysis of E2P Formed from P i without Ca 2ϩ -The observed block of decay of the ADP-insensitive EP (E2P) formed from ATP and Ca 2ϩ in the mutants with the elongated A/M1 linker and 4950525354S might possibly be due to the block of hydrolysis of the Ca 2ϩ -free form of E2P. Therefore, in Fig. 7, the E2P hydrolysis was directly examined by first phosphorylating the enzyme with 32 P i in the absence of Ca 2ϩ and K ϩ and the presence of 35% (v/v) Me 2 SO, which extremely favors the E2P formation in the reverse reaction (38), and then by diluting the phosphorylated sample at 0°C with a large volume of solution containing nonradioactive P i and K ϩ without Ca 2ϩ . The conditions for the hydrolysis were thus otherwise made the same as those for the decay of EP formed from ATP with Ca 2ϩ in Fig. 6. Hydrolysis of 32 P-labeled E2P proceeded with firstorder kinetics as shown with the representative mutants, and the rates obtained were summarized in Table 1. To our surprise, in all the mutants with the elongated A/M1 linker 1Gi-46/47 to 6Gi-46/47, 1Gi-42/43, and 4Gi-42/43, the hydrolysis of E2P without bound Ca 2ϩ occurred as rapidly as in the wild type. The mutants 4950525354S, 5155565859A, 4Ai-22/23, and 4Ai-53/54 also exhibited the rapid E2P hydrolysis as the wild type.
Ca 2ϩ Occlusion in Stable E2P-The observed block of the decay of ADP-insensitive EP (E2P) formed from ATP with Ca 2ϩ (Fig. 6) and the rapid hydrolysis of E2P formed from P i without Ca 2ϩ (Fig. 7) in the mutants indicate that there may be a kinetic limit for the decay of E2P formed with Ca 2ϩ before the hydrolysis of E2P without bound Ca 2ϩ . This limiting step is possibly the Ca 2ϩ -releasing step from E2PCa 2 ; E2PCa 2 3 The total amounts of EP obtained at zero time (i.e. immediately before the EGTA addition) are normalized to 100%. Solid lines show the least squares fit to a single exponential, and the decay rates thus obtained are given in Table 1. For determination of ADP-insensitive EP (B), an equal volume (100 l) of a mixture containing 4 mM ADP, 1 M A23187, 0.1 M KCl, 7 mM MgCl 2 , 10 mM EGTA, and 50 mM MOPS/Tris (pH 7.0) was added to the above EGTA-added phosphorylation mixture at the indicated time. At 1 s after the ADP addition, the reaction was quenched by acid. The amount of the ADP-insensitive EP at each of the indicated times thus determined is shown as percentage of the total amount of EP determined at zero time. . At different times after the dilution, the E2P hydrolysis was quenched by acid. The amounts of E2P formed with 32 P i at zero time are normalized to 100%. Solid lines show the least squares fit to a single exponential, and the rates thus obtained are given in Table 1.
E2P ϩ 2Ca 2ϩ and E2PCa 2 may be stabilized and accumulated in the mutants. In Fig. 8, this possibility was directly examined by the determination of 45 Ca 2ϩ occlusion in E2P accumulated from ATP and 45 Ca 2ϩ . With the representative mutants 4Gi-46/47 (Fig. 8, A and C) and 2Gi-46/47 (Fig. 8, B and C), EP was first formed from ATP and 45 Ca 2ϩ at steady state in the absence or presence of the Ca 2ϩ ionophore A23187, and then the EP decay was initiated by the addition of excess EGTA. The amount of occluded 45 Ca 2ϩ was determined at the indicated time by membrane filtration with an extensive washing with a solution containing EGTA and A23187. The total amount of EP and the fraction of ADP-insensitive EP (E2P) was determined by the use of [␥-32 P]ATP and nonradioactive Ca 2ϩ .
The extremely slow EP decay was observed at 25°C with the mutant 4Gi-46/47 (Fig. 8A). At the zero time of the decay reaction, the amount of occluded Ca 2ϩ and that of EP were ϳ9 and 4 ϳ 4.5 nmol/mg of expressed SERCA1a protein, respectively, both in the presence and absence of A23187. These values gave the stoichiometry "two occluded Ca 2ϩ ions per one EP." The amount of 45 Ca 2ϩ remaining on the membrane filter decreased concomitantly with the EP decay, and therefore the stoichiometry was always found to be "two" during the EP decay course (Fig. 8C). This stoichiometry is in complete agreement with the presence of two Ca 2ϩ -binding (transport) sites in the ATPase molecule. Importantly, EP remaining during the decay time course was exclusively the ADP-insensitive one (E2P) in this mutant 4Gi-46/47 (data not shown, but see Fig. 6B). Thus, the results clearly demonstrated that EP accumulated in the mutant is the ADP-insensitive EP that possesses two occluded Ca 2ϩ ions, i.e. E2PCa 2 , and the Ca 2ϩ deocclusion from E2PCa 2 is extremely slowed by the elongation of the A/M1 linker, and therefore E2PCa 2 is accumulated exclusively. The decay of E2PCa 2 became faster by A23187 (although only slightly), being consistent with the mechanism that the Ca 2ϩ release occurs into lumen from E2PCa 2 . It should be noted that this type of experiment was not possible with the wild type because its EP decay is extremely rapid and completed during the EGTA washing (actually within 1 s).
In Fig. 8, B and C, the experiments were performed with the mutant 2Gi-46/47 in the presence of A23187 at 10°C. At this temperature, the EP decay was most conveniently followed in this mutant. The phosphorylated samples were filtered at the indicated time with and without addition of ADP immediately before the filtration. Approximately 50% of the total amount of  45 Ca 2ϩ specifically bound to the expressed SERCA1a, i.e. the occluded Ca 2ϩ in SERCA1a, was determined after this extensive washing by subtracting the amount of nonspecific Ca 2ϩ binding, which was determined by including 1 M TG in the above phosphorylation mixture, otherwise as described under "Experimental Procedures." The values are shown as "nmol per mg of SERCA1a applied on the filter" (‚, OE). By using [␥-32 P]ATP and nonradioactive CaCl 2 , the total amount of EP present in the above EGTAadded phosphorylation mixture was determined at the indicated time by acid quenching (छ, ࡗ). EP present was found to be exclusively the ADPinsensitive one (E2P) as also demonstrated with this mutant in Figs. 4 -6. The amounts of EP were also determined without the acid quenching but with the membrane filtration method as the above determination of the occluded 45 Ca 2ϩ . The amounts of EP thus determined by the two methods actually agreed well with each other, and therefore the acid-quenching method was EP was ADP-insensitive EP (E2P) throughout the decay reaction (compare open and closed squares in Fig. 8B, also see Fig.  6B). The total amount of EP, the amount of ADP-insensitive EP, and the amounts of occluded Ca 2ϩ determined without and with the ADP addition decreased very slowly and concomitantly (Fig. 8B). Thus, as plotted in Fig. 8C, the stoichiometry of the occluded Ca 2ϩ was always found to be two in the total amount of EP (E1P plus E2P) and in the amount of ADP-insensitive EP (E2P) throughout the decay reaction. The results show that two Ca 2ϩ ions are occluded in both the ADP-sensitive EP, i.e. E1PCa 2 , and the ADP-insensitive EP, i.e. E2PCa 2 , in the mutant 2Gi-46/47. The results are consistent with the view that the E1PCa 2 -E2PCa 2 equilibrium was largely shifted to E2PCa 2 , and the Ca 2ϩ deocclusion from E2PCa 2 was blocked in this mutant. It should be mentioned for the mutant 4950525354S that, as indicated by the kinetic analyses in Figs. 4 -7, E2PCa 2 is probably accumulated in this mutant from ATP and Ca 2ϩ as in the mutant 4Gi-46/47. The Ca 2ϩ -binding experiments were not possible, however, with this mutant because of its very low protein expression level (only ϳ15-20% that of 4Gi-46/47 and 2Gi-46/47 or wild type).
Formation of Stable E2PCa 2 from E2P and Ca 2ϩ in Reverse Reaction-As demonstrated with SR Ca 2ϩ -ATPase (39), the E1P to E2P transition and the Ca 2ϩ release into lumen can be reversed by the low affinity Ca 2ϩ binding from the luminal side to the transport sites of the Ca 2ϩ -free form of E2P. In Fig. 9, we examined with the representative mutants whether the stable E2PCa 2 can be produced in the reverse reaction. In the experiments, E2P was first formed from 32 P i without Ca 2ϩ in the presence of A23187 and 35% (v/v) Me 2 SO that strongly favors E2P (38), and subsequently the phosphorylation mixture was diluted 10-fold with a solution containing CaCl 2 to give a very high (saturating) Ca 2ϩ concentration of 20 mM. After 1 min of incubation with Ca 2ϩ , the mixture was further diluted 10-fold with a solution containing excess EGTA and 0.1 M KCl, and the decay of EP was followed at 0°C. Thus, the final conditions for this EP decay in Fig. 9 were made to be essentially the same as those for the decay of EP formed from ATP and Ca 2ϩ in the forward reaction in Fig. 6. We observed actually the same time courses in Fig. 9 as those in Fig. 6 for each of the representative mutants as well as for the wild type. Most importantly, in the mutants with the two or more glycine insertion in the A/M1 linker (2Gi-46/47, 3Gi-46/47, and 4Gi-46/47), the decay of EP formed from E2P with the subsequently added Ca 2ϩ was nearly completely blocked. Furthermore, EP remaining during the decay course was exclusively the ADP-insensitive one in the mutants 3Gi-46/47 and 4Gi-46/47, and ϳ50% in the mutant 2Gi-46/47 (Fig. 9B). The results show that the stable E2P, probably E2PCa 2 , is formed in these mutants in the reverse reaction from E2P and Ca 2ϩ as well as in the forward reaction with ATP and Ca 2ϩ . In the wild type, E2P formed from P i became exclusively the ADP-sensitive one (E1PCa 2 ) upon the Ca 2ϩ addition, being consistent with the previous observation (39).
With the representative mutant 4Gi-46/47, we then actually determined the 45 Ca 2ϩ occlusion in the stable E2P formed from E2P with the subsequently added 3 mM 45 Ca 2ϩ (rather than 20 mM because of the experimental limitation), otherwise as above. 4 We found that the amounts of stable E2P formed and the 4 The details of the experimental conditions are as follows. E2P was first formed from nonradioactive P i in 35% (v/v) Me 2 SO in the absence and presence of 1 M TG, otherwise as in Fig. 9. Subsequently, the mixture was diluted 10-fold at 0°C with a solution containing 3 mM 45  ) and for reducing nonspecifically bound Ca 2ϩ . The mixture was then subjected to the membrane filtration with the extensive washing by the above EGTA-containing washing solution, otherwise as described in Fig. 8. . At different times after this Ca 2ϩ -removal, the EP decay was quenched by acid, and the amounts of EP remaining were determined (A). The amounts of EP obtained at zero time (i.e. immediately before the addition of the EGTA solution) are normalized to 100%. It should be noted that the amount of E2P formed from P i (at zero time) was comparable with that of EP formed from ATP and Ca 2ϩ shown in Fig. 4 in the wild type and in each of the mutants (e.g. in the wild type, E2P formed from P i was 3.45 Ϯ 0.22 nmol/mg SERCA1a protein (n ϭ 4) and EP formed from ATP was 3.31 Ϯ 0.14 nmol/mg of the SERCA1a protein (n ϭ 4)). B, for determination of the ADP-insensitive EP, an equal volume (250 l) of a mixture containing 4 mM ADP, 1 M A23187, 0.1 M KCl, 7 mM MgCl 2 , 10 mM EGTA, and 50 mM MOPS/Tris (pH 7.0) was added to the above EGTA-diluted phosphorylation mixture at the indicated time. At 1 s after this addition, the reaction was quenched by acid. The amount of the ADP-insensitive EP determined at each of the indicated time is shown as percentage of the total amount of EP determined at zero time.

JOURNAL OF BIOLOGICAL CHEMISTRY 34437
occluded Ca 2ϩ were 1.65 Ϯ 0.13 and 3.33 Ϯ 0.25 nmol/mg of expressed SERCA1a protein (n ϭ 4), respectively, and thus the stoichiometry of the occluded Ca 2ϩ in E2P was 2.02. The results demonstrated that two Ca 2ϩ ions are occluded in the stable E2PCa 2 , which is formed in the reverse reaction as in the forward reaction. The results also indicated that the reverse transition from E2PCa 2 to E1PCa 2 was dramatically retarded (blocked) in the mutants 3Gi-46/47 and 4Gi-46/47. Accessibility of Luminal Ca 2ϩ to Transport Sites of E2P Formed from P i -We then measured the Ca 2ϩ concentration dependence of the Ca 2ϩ -induced formation of the stable EP from E2P and Ca 2ϩ in the reverse reaction (Fig. 10). In the experiments, E2P was first formed from P i without Ca 2ϩ in the presence of A23187, and subsequently the phosphorylation mixture was largely diluted with a solution containing CaCl 2 to give the free Ca 2ϩ concentrations indicated in Fig. 10. Immediately after this Ca 2ϩ addition, the EP decay was followed without removing Ca 2ϩ . As typically shown in Fig. 10, A and B, for the wild type and the mutant 2Gi-46/47, the decay of EP proceeded with two phases as shown previously in this type of experiment with the Ca 2ϩ -ATPase (40). The first and rapid phase corresponds to the hydrolysis of E2P without bound Ca 2ϩ . In the wild type, the second and very slow phase corresponds to the forward decay of E1PCa 2 that is formed from E2P and Ca 2ϩ , as well documented previously (39,40). In fact, nearly all the EP remaining in the decay course was the ADPsensitive one (E1P) in the wild type (data not shown).
In the mutants 2Gi-46/47 (Fig. 10B) and 3Gi-46/47 and 4Gi-46/47 (data not shown), the EP decay in the second phase was extremely slow, and actually almost no decay occurred during the period of observation. The decay of EP in these mutants was extremely slow even after removal of Ca 2ϩ as shown in Fig. 9. The EP remaining in this slow phase was almost exclusively the ADP-insensitive one (E2P) in the mutants 3Gi-46/47 and 4Gi-46/47 even at the highest Ca 2ϩ concentration 20 mM and ϳ50% in the mutant 2Gi-46/47 at 20 mM Ca 2ϩ (see Fig. 9).
The content of EP in the second and slow phase was obtained by extrapolating to the zero time and plotted versus the Ca 2ϩ concentrations (Fig. 10C). The content increased with the increase in the Ca 2ϩ concentration and was nearly saturated at ϳ10 mM Ca 2ϩ . K 0.5 was estimated to be 1.4 mM in the wild type and similarly 0.9 -1.3 mM in the mutants in Fig. 10 and other mutants as well (see Table 1). These values are actually consistent with the high Ca 2ϩ concentrations required for the Ca 2ϩinduced reverse reaction from E2P determined previously with SR Ca 2ϩ -ATPase as the access of luminal Ca 2ϩ to the transport sites of E2P (41)(42)(43)(44). The results indicate that in these mutants with the elongated A/M1 linker, the luminal Ca 2ϩ can access and bind to the transport sites of E2P as in the wild type, and the mutants produce the stable ADP-insensitive EP, i.e. E2PCa 2 .
In this context, it should be noted that the Ca 2ϩ -dependent increase in the stable EP in Fig. 10C reflects mostly the relative values between the Ca 2ϩ -dependent increasing rate of the formation of the stable EP versus the rate of the hydrolysis of the Ca 2ϩ -unbound E2P. Namely, the curve reflects mostly the relative rates between E2P ϩ 2Ca 2ϩ 3 E2PCa 2 (or further to E1PCa 2 ) versus E2P ϩ H 2 O 3 E2 ϩ P i , rather than the relative rates between the reverse and forward reactions in E2P ϩ 2Ca 2ϩ 7 E2PCa 2 (or E1PCa 2 ) (i.e. the Ca 2ϩ affinity). This is because the curve in Fig. 10C is the plot of the amount of EP stably remaining after the hydrolysis of the Ca 2ϩ -unbound E2P, and because the amount of the remaining EP is dependent on the rate of its formation relative to the rate of the E2P hydrolysis. Furthermore, the decay of the remaining EP in the second phase was extremely slow in the wild type as well as in the mutants, and thus virtually negligible as compared with its were phosphorylated with 32 P i in 2.5 l of a mixture as described in Fig. 9. The mixture was then cooled and diluted 100-fold at 0°C with 247.5 l of a solution containing various concentrations of CaCl 2 in 1 M A23187, 101 mM KCl, 1 mM EGTA, 7 mM MgCl 2 , and 50 mM MOPS/Tris (pH 7.0) to give the final Ca 2ϩ concentrations as indicated with different symbols. The amount of EP remaining at the indicated time after this Ca 2ϩ addition was determined and shown as percentage of the amount of EP at zero time, which was determined immediately before the Ca 2ϩ addition. The EP decay occurred in two phases. The first and rapid phase completed within a few seconds corresponding to the hydrolysis of E2P without bound Ca 2ϩ (see Fig. 7). Essentially the same results were observed in the mutants 3Gi-46/47 and 4Gi-46/47 (data not shown) as in 2Gi-46/47. C, content of EP in the slow and second phase at each Ca 2ϩ concentration was obtained by extrapolating to the zero time and plotted versus the Ca 2ϩ concentration.  Table 1. rapid formation and the rapid hydrolysis of the Ca 2ϩ -unbound E2P both in the wild type and mutants. 5 Importantly, the E2P hydrolysis rate was found to be essentially the same between the wild type and the mutants (Fig. 7). Therefore, the very similar Ca 2ϩ dependence curves of the wild type and mutants in Fig.  10C indicate that the rates of the formation of the stable EP upon the luminal Ca 2ϩ binding to E2P are very similar between them. Thus we concluded that E2P formed from P i without Ca 2ϩ in the mutants possesses the luminally opened Ca 2ϩ release pathway as in the wild type.
Decomposition of Stable E2PCa 2 by Thapsigargin-Thapsigargin (TG) is well known to bind very tightly to a specific site on the transmembrane helices of the Ca 2ϩ -free E2 form of SERCA and to fix their orientation to produce a very stable complex E2(TG) (10,30,45,46). In addition, it was previously suggested by the P i 7 HOH oxygen exchange (47) that TG stimulates the dephosphorylation (the acylphosphate hydrolysis reaction step) of the Ca 2ϩ -unbound form of E2P (especially in the presence of Me 2 SO). In Fig. 11, we examined a possible effect of TG on E2PCa 2 exclusively accumulated with the representative mutant 4Gi-46/47. E2PCa 2 was first formed in the forward reaction from ATP and Ca 2ϩ as in Fig. 6A, and also in the reverse reaction from P i with the subsequently added 20 mM Ca 2ϩ as in Fig. 9A. Then excess EGTA was added with or without TG, and the dephosphorylation was followed at 25°C. The results demonstrated that TG accelerates the dephosphorylation of E2PCa 2 strongly and equally in both cases, and therefore the E2PCa 2 accumulated is sensitive to TG. It is very likely that TG binds to E2PCa 2 , and its binding accelerates the Ca 2ϩ deocclusion and release from E2PCa 2 . This is because E2PCa 2 is the exclusively accumulated one, and there is no E2P and no E2 (that are known to bind TG), and because the decay of E2PCa 2 in the presence of TG took place with the rate 0.01 s Ϫ1 (Fig. 11), and this is still far slower than the hydrolysis of the Ca 2ϩ -released form of E2P. Actually the E2P hydrolysis of the mutants as well as of the wild type was completed within 1 s at 25°C under the conditions in Fig. 11 in the absence and presence of TG (data not shown). Consistently, the rate of the E2P hydrolysis was reported previously with SR Ca 2ϩ -ATPase to be 60 -120 s Ϫ1 in the absence or presence of TG (42,47) at this temperature (25°C) under very similar conditions. Thus the effect of TG on the decomposition of E2PCa 2 is shown here for the first time.
Structure of E2PCa 2 Revealed by Proteolytic Analysis as an Intermediate State between E1PCa 2 and E2P-In the transport cycle, the cytoplasmic three domains, N, P, and A largely move and change their organization states (7, 9 -17). These changes are definitely monitored as the changes in the resistance of the specific cleavage sites against trypsin and PrtK (15)(16)(17). As one of most notable examples, the tryptic T2 site Arg 198 on the outermost Val 200 loop (Asp 196 -Asp 203 ) of the A domain is rapidly cleaved in E1PCa 2 , by contrast, it is completely resistant in E2P (15)(16)(17). This is because the A domain largely rotates parallel to the membrane plane by ϳ110°, and the Val 200 loop, including Arg 198 , associates with the P domain by forming an ionic interaction network and thus blocks sterically against the tryptic attack, as seen in E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ (or E2⅐AlF 4 Ϫ ) (see Refs. 11-14) (see Figs. 2 and 13). In Fig. 12A, upper panel, the trypsin proteolysis was performed with the wild type for the major intermediates, and their structural analogs were stabilized by the appropriate ligands according to previous findings (15)(16)(17). The ATPase chain and its fragments were immunodetected with a monoclonal antibody that recognizes Ala 199 -Arg 505 (the tryptic fragment "A1") of SERCA1a. In all the structural states, the T1 site (Arg 505 ) on the outermost loop of the N domain was very rapidly cleaved to produce the fragment "A" (Met 1 to Arg 505 , as immunodetected) and the fragment "B" (Ala 506 to the C terminus Gly 994 , not immuno-monitored). In the structural states E2, E1Ca 2 , 5 Regarding the wild type, the observed biphasic behavior in the EP decay after the addition of the high concentrations of Ca 2ϩ in Fig. 10A may not be accounted for if we assume a rapid equilibrium in the reaction E1PCa 2 7 E2PCa 2 7 E2P ϩ 2Ca 2ϩ under these conditions, i.e. in the presence of high Ca 2ϩ concentrations (because if in the rapid equilibrium, a slowed (but still) single exponential decay after the addition of Ca 2ϩ may be expected).
In this regard, it may be of interest to note that in the previous analysis with SR Ca 2ϩ -ATPase by Nakamura (41) and Inesi and co-workers (42) for the decay of E1PCa 2 formed from ATP, a biphasic E1PCa 2 decay with a markedly slowed decay in second phase at the high luminal Ca 2ϩ concentrations was well documented, and actually the E1PCa 2 fraction in the second phase was increased with the increase in the Ca 2ϩ concentrations at submillimolar to approximate millimolar. To account for the biphasic decay, Inesi and co-workers (42) proposed the branched reaction mechanism from E1PCa 2 to E2PCa 2 , in which E1PCa 2 rapidly decays in the presence of bound ADP but very slowly in its absence and in the presence of the high concentrations of Ca 2ϩ . As a reason for the retardation of the E1PCa 2 decay at the high Ca 2ϩ concentrations in Fig. 10A, it is also possible that Mg 2ϩ bound at the catalytic metal site in E1PCa 2 was substituted by Ca 2ϩ , and hence the EP isomerization (thus the E1PCa 2 decay) was markedly slowed because of this bound Ca 2ϩ , as demonstrated previously by using the substrate CaATP with SR Ca 2ϩ -ATPase (54,55). The marked stabilization of the structure of the E1PCa 2 state by Ca 2ϩ bound at the catalytic site was actually demonstrated by the recent structural studies on its analogs with SR Ca 2ϩ -ATPase (56, 57).   (48,49). The positions of the Ca 2ϩ -ATPase chain and its fragments and those of the molecular mass markers are indicated on the left and right margins, respectively. Note also that the antibody immunodecorated trypsin (A) and PrtK (B), in addition to the Ca 2ϩ -ATPase fragments. E1⅐AlF x ⅐ADP (E1PCa 2 ⅐ADP analog), and E1PCa 2 accumulated exclusively with the wild type from ATP at a very high concentration (5 mM) of Ca 2ϩ , and the T2 site was rapidly cleaved to produce the fragment A1 as immunodetected and the fragment "A2" (Met 1 -Arg 198 , not immuno-monitored). By contrast, in the structural states of E2P without bound Ca 2ϩ stabilized by orthovanadate (E2V i ) and by Al 3ϩ /F Ϫ (E2⅐AlF 4 Ϫ , the transition state analog of the E2P hydrolysis (17)), the T2 site was completely resistant. The complete resistance was also found with E2⅐MgF 4 2Ϫ and E2⅐BeF x , the E2⅐P i analog, and the E2P ground state analog, respectively (17) (data not shown). These results with the wild type agree with the previous demonstration with SR Ca 2ϩ -ATPase (15)(16)(17).
In the lower panel of Fig. 12A for the representative mutant 4Gi-46/47, the distinct finding was obtained with EP accumu-lated from ATP and Ca 2ϩ , which was exclusively E2PCa 2 as demonstrated in Figs. 6 and 8. In E2PCa 2 , the T2 site was completely resistant to tryptic attack as clearly shown by the exclusive accumulation of the fragment A without any further cleavage (i.e. without any formation of A1). This is in sharp contrast to its rapid cleavage in E1PCa 2 accumulated from ATP with the wild type and in the E1PCa 2 structural state of the mutant and wild type stabilized as E1⅐AlF x ⅐ADP. In the Ca 2ϩ -unbound E2P structural analogs (E2V i , E2⅐AlF 4 Ϫ , E2⅐MgF 4 2Ϫ , and E2⅐BeF x ), the T2 site in the mutant was completely resistant as in the wild type. The results show that in E2PCa 2 accumulated with the mutant, the A domain is already largely rotated from its position in E1PCa 2 and associated with the P domain at the T2 site region (Val 200 loop) as in the Ca 2ϩ -unbound E2P state. This agrees with the fact that E2PCa 2 accumulated is ADP-insensi- tive, because the loss of the ADP sensitivity is brought about by the association of the largely rotated A domain with the P domain (13)(14)(15)(16)(17)21) (Fig. 2 and Fig. 13).
In Fig. 12B, the same set of experiments was performed with PrtK. As demonstrated previously with SR Ca 2ϩ -ATPase (48,49), in the E2 state, PrtK cleaved at Leu 119 on the top part of M2 (A/M2 linker region, see Figs. 2 and 13) producing the fragment "p95," and more slowly at Thr 242 on the A/M3 linker and Ala 746 on M5 producing the fragments "p81/83." The Leu 119 site became resistant in E1Ca 2 , and all the sites were nearly completely resistant in E1PCa 2 accumulated from ATP with the wild type and in the E1PCa 2 structural state (E1⅐AlF x ⅐ADP) produced with the wild type and with the mutant. The ATPase chain was also completely resistant in the Ca 2ϩ -unbound E2P structural analogs E2V i , E2⅐AlF 4 Ϫ (Fig. 12B), E2⅐MgF 4 2Ϫ , and E2⅐BeF x (data not shown). These observations in the wild type and the mutant are in complete agreement with the previous demonstration with SR Ca 2ϩ -ATPase (15)(16)(17). Distinct finding was obtained with E2PCa 2 accumulated with the mutant. Namely, a fairly rapid cleavage at Leu 119 occurred in E2PCa 2 to produce p95, and was in sharp contrast to its resistance in the E1PCa 2 state and the Ca 2ϩ -unbound E2P state. The results show that the structure at Leu 119 in E2PCa 2 differs distinctly from those in E1PCa 2 and in the Ca 2ϩ -free form of E2P.
In this regard, it is essential to note the fact that the structural bases rendering the complete resistance at Leu 119 are totally different between E1PCa 2 and the Ca 2ϩ -free form of E2P, as clearly seen in the structures E1⅐AlF x ⅐ADP and E2⅐MgF 4 2Ϫ (see Fig. 13). The resistance in E1PCa 2 (E1⅐AlF x ⅐ADP) is most likely because of the steric blocking against PrtK brought about by van der Waals contacts of the top part of M2, including Leu 119 with the top part of M4 (Asn 330 /Ile 332 ). On the other hand, in the Ca 2ϩ -free form of E2P (E2⅐MgF 4 2Ϫ and E2⅐AlF 4 Ϫ ), the resistance is most likely because of the steric blocking brought about by van der Waals contacts of Leu 119 with the hydrophobic residues in the interaction network "Tyr 122 -hydrophobic cluster" formed from the A and P domains and the top part of M2 (A/M2 linker region). This cluster actually consists of Ile 179 / Leu 180 /Ile 232 at the bottom part of the A domain, Val 705 /Val 726 of the P domain, and Leu 119 /Tyr 122 at the top part of M2. Then a structural basis for the observed rapid cleavage at Leu 119 in E2PCa 2 , the intermediate state between E1PCa 2 and E2P, can be deduced from the structural change E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ (or E2⅐AlF 4 Ϫ ), which is the presently available model for the overall change in E1PCa 2 3 E2P ϩ 2Ca 2ϩ (Fig. 13)

DISCUSSION
In this study, we explored the functional importance of the length of the A/M1 linker (Glu 40 -Ser 48 loop) and found that its elongation markedly accelerates the loss of the ADP sensitivity in EP (the E1P to E2P isomerization) but blocks almost completely the decay of the ADP-insensitive EP accumulated. This EP was demonstrated to be E2PCa 2 having two occluded Ca 2ϩ ions at the transport sites, and the Ca 2ϩ deocclusion and release from E2PCa 2 were blocked. On the other hand, the hydrolysis of the Ca 2ϩ -free form of E2P produced from the E2 state with P i in the mutants was as rapid as in the wild type. The stable E2PCa 2 of the mutants was also produced in the reverse reaction upon addition of high concentrations of Ca 2ϩ to E2P formed from P i without Ca 2ϩ , and the reverse isomerization from E2PCa 2 to E1PCa 2 was blocked in the mutants.
The results indicate that the EP isomerization/Ca 2ϩ -release process described as a single step, E1PCa 2 3 E2P ϩ 2Ca 2ϩ , consists of or can be dissected into the two successive steps E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ , i.e. the loss of ADP sensitivity at the catalytic site (the EP isomerization) and the subsequent Ca 2ϩ release into lumen. This mechanism actually agrees with the one previously postulated on SR Ca 2ϩ -ATPase, although the intermediate state E2PCa 2 has never been identified and its mere presence has been questioned (e.g. Ref. 50). 6 In 6 In this regard, we make the following two additional discussions in relation to previously proposed mechanisms. First, the standard view on the reaction mechanism with the sequential occurrence of E1P and E2P in the linear kinetic model (see Fig. 1) has been challenged, and instead an "out-ofphase coupling of the catalytic reactions" in the interacting ATPase molecules in their dimer was postulated (58,59). In this model, one ATPase molecule in the dimer precedes the catalytic steps one step ahead of the other, and therefore the presence of equimolar steady-state concentrations of E1PCa 2 and E2PCa 2 are expected to be observed. On the other hand, in this study with the mutants with the elongated A/M1 linker, we observed clearly the exclusive accumulation of E2PCa 2 (E2P possessing the stoichiometric amount of occluded Ca 2ϩ ) among all the EP present at the steady state with no E1PCa 2 . As shown in more detail in Figs. 4 -6 and 8, almost all of the EP accumulated at steady state with the wild type was E1PCa 2 , and the accumulation of the stable E2PCa 2 increased with the increase in the number of the inserted glycines in the A/M1 linker (i.e. 50% E2PCa 2 with 50% E1PCa 2 in 2Gi-46/47, and nearly 100% E2PCa 2 in the mutants with three or more glycine insertions). Thus, at least in our system in this study, the postulated out-of-phase coupling of the catalytic reactions is unlikely, or it might be possible that the postulated interactions between the ATPase molecules are disrupted in the mutants. As the second point of the additional discussions, the presence of a slow step in the reverse transition E2P to E1PCa 2 upon the luminal Ca 2ϩ binding to E2P has also been a subject of controversy, and the mere presence of E2PCa 2 has been questioned (50). Myung and Jencks (50) concluded on the basis of the observed very rapid reverse transition that there is no kinetic barrier in this reversal E2P ϩ 2Ca 2ϩ 3 E1PCa 2 , and that there is only one phosphorylated intermediate with bound Ca 2ϩ in the transport cycle (i.e. E1PCa 2 ). This reversal in the wild type upon the luminal Ca 2ϩ binding to E2P is certainly very rapid (as the kinetic results in Fig. 10A with the wild type indicated). On the other hand, in this study of the mutants with the elongated A/M1 linker, we also found the exclusive accumulation of E2PCa 2 in the reverse reaction from E2P and Ca 2ϩ as well as in the forward reaction with ATP and Ca 2ϩ . Thus, for the first time, we were able to trap the inter-this study, we could successfully trap and thus identify this E2PCa 2 state for the first time. E2PCa 2 accumulated was actually shown to possess the structural feature as the intermediate state between E1PCa 2 and the Ca 2ϩ -released form of E2P (Fig.  12), and the results revealed the critical importance of the length of the A/M1 linker in the Ca 2ϩ deocclusion and release from E2PCa 2 . Importantly, our previous study showed (23) that the substitutions of any residues in the A/M1 linker do not inhibit the reaction cycle, whereas the shortening of the linker by a deletion of any single residue in this linker almost completely blocks the E1PCa 2 to E2PCa 2 isomerization and the hydrolysis of the Ca 2ϩ -free form of E2P. Thus, the length of this linker is obviously critical for each of the successive three steps, E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ and E2P ϩ H 2 O 3 E2 ϩ P I ; and the shortening and the elongation of the linker both cause the severe defects but in distinct steps. 7 Based on the results and the crystal structures, we discuss below the structural roles of the A/M1 linker in each of the three steps and dissect the structural events occurring in these steps.
Loss of ADP Sensitivity; the E1PCa 2 to E2PCa 2 Isomerization-The loss of ADP sensitivity at the catalytic site (the EP isomerization) involves the large rotation of the A domain approximately parallel to the membrane plane and its association with the P domain (see E1⅐AlF x ⅐ADP 3 E2⅐Mg F 4 2Ϫ in Figs. 2 and 13). Therefore, the outermost T 181 GES loop of the A domain comes above and docks onto the phosphorylation site Asp 351 by forming an extensive hydrogen-bonding network with the residues of the P domain around Asp 351 , and thus the T 181 GES loop sterically blocks the access of the ADP ␤-phosphate to the Asp 351 -acylphosphate (13). In E2PCa 2 accumulated in the mutants with the elongated A/M1 linker, the rotation of the A domain and its docking with the P domain for the loss of the ADP sensitivity must already be achieved as it is ADP-insensitive. Such a structural state in E2PCa 2 was in fact clearly demonstrated by the observation that the tryptic T2 site Arg 198 on the Val 200 loop (Asp 196 -Asp 203 ), another outermost loop of the A domain juxtaposed to the T 181 GES loop, is completely resistant against tryptic attack as in the E2P state without bound Ca 2ϩ (Fig. 12).
The elongation of the A/M1 linker markedly accelerated the E1PCa 2 to E2PCa 2 isomerization to cause the exclusive accumulation of E2PCa 2 (Figs. 4 and 5), and in contrast, its shortening by a deletion of any single residue blocked this isomerization (23). Therefore, the length of the A/M1 linker needs to be sufficiently long for the E1PCa 2 to E2PCa 2 isomerization. The A/M1 linker is probably critical for the positioning of the A domain relative to the P domain, i.e. the height from the mem-brane plane for their docking. Actually in the structural analogs for the ADP-insensitive EP (E2⅐MgF 4 2Ϫ and E2⅐AlF 4 Ϫ ), the largely rotated A domain is positioned above about half of the P domain, including Asp 351 , i.e. about half of the P domain is located underneath the A domain (Figs. 2 and 13). Hence, the T 181 GES loop at the lower part of the A domain comes above Asp 351 and blocks the access of ADP bound on the N domain to the Asp 351 -acylphosphate (13).
For realizing the A-P domain association, the P domain also should move significantly from the original position in E1PCa 2 , i.e. incline to the underneath of the A domain. Such motion of the P domain, as well as the large rotation of the A domain, is probably achieved to produce E2PCa 2 and at least to some extent to cause their docking and thus the loss of ADP sensitivity. 8 (These motions are still not enough for the subsequent Ca 2ϩ -deocclusion/release from E2PCa 2 (see below for the Ca 2ϩ release).) As the nature of the wild type, it is known that the EP isomerization is a very slow process thus having a kinetic barrier, and this slow process is followed by a very rapid Ca 2ϩ release in the scheme in Fig. 1 (actually the isomerization is rate-limiting in the whole Ca 2ϩ transport cycle). This means that the isomerization from E1PCa 2 to E2PCa 2 is accompanied by some structural distortion or strain. Our findings show that such structural restriction for the EP isomerization is markedly relieved by the elongation of the A/M1 linker (this study) but markedly enhanced by its shortening (23). The findings further suggest that structural distortion or strain imposed in E2PCa 2 upon the EP isomerization in the wild type benefits to the work for the subsequent Ca 2ϩ release from E2PCa 2 (see below). This is because virtually no Ca 2ϩ -deocclusion/release from E2PCa 2 occurred in the mutants with the elongated A/M1 linker. It is tempting to speculate that the largely rotated A domain may be pushed (or even moved) upward upon its positioning above and docking onto the P domain and that such mounting motion could be the structural restriction related to the length of the A/M1 linker.
As a summary of this section, the sufficiently long length of the A/M1 linker is probably critical for the proper motions and mediate state E2PCa 2 by introducing the mutations, and thereby we identified the structural element that is critical for the Ca 2ϩ release from E2PCa 2 and for the rapid reverse transition from E2PCa 2 to E1PCa 2 . It should be noted, however, that in the wild type all the structural changes in these processes may be occurring as successive events; therefore, the intermediate state E2PCa 2 may not be trapped or detected as a stable one. 7 The shortening of the A/M1 linker by the deletion of single residues (23) and its elongation by the one to four glycine insertions (this study) did not cause serious consequences on the E2 to E1Ca 2 transition and the E1PCa 2 formation from E1Ca 2 and ATP and therefore on the structural changes in these steps. 8 It is important to note that in all the crystal structures E1Ca 2 , E1⅐AlF x ⅐ADP, E2⅐AlF 4 Ϫ , E2⅐MgF 4 2Ϫ , and E2(TG), Arg 334 /Arg 324 at the bottom part of the P domain forms an ionic and hydrogen-bonding interaction network with the top part of M2 (the A/M2 linker region) where the several polar and negatively charged residues are aligned (Glu 123 /Glu 121 /Glu 117 /Asn 114 / Glu 113 /Asn 111 /Glu 109 /Gln 108 ). In the change E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ or E2⅐AlF 4 Ϫ , Arg 334 moves downward from the Glu 123 /Glu 121 region to Asn 114 / Asn 111 in the interaction network (see Fig. 8 in Ref. 22). The structures therefore suggest that the interactions in the network may function as to guide and pull the P domain to cause its inclination toward the bottom of the A domain when the A domain largely rotates and M2 tilts away. Most importantly, the mutations in this interaction network involving Arg 324 / Arg 334 were previously found to block the loss of the ADP sensitivity, i.e. the E1PCa 2 to E2PCa 2 isomerization (22). Therefore, the inclination of the P domain occurs to some extent during the E1PCa 2 to E2PCa 2 isomerization. The strain of the A/M1 linker imposed in E2PCa 2 likely causes further inclination of the P domain in E2PCa 2 3 E2P ϩ 2Ca 2ϩ as the final process. Therefore, the A-P domain interface is further adjusted to achieve the Ca 2ϩ deocclusion/release, and Val 705 /Val 726 of the P domain reaches Ile 179 / Leu 180 /Ile 232 of the A domain and Tyr 122 /Leu 119 on the top part of M2 producing the Tyr 122 -hydrophobic cluster, which is critical for stabilizing the Ca 2ϩ -released structure of E2P. positioning of the A and P domains and their docking to produce the E2PCa 2 structure in the E1PCa 2 to E2PCa 2 isomerization. The structural events are markedly accelerated by the elongation of the A/M1 linker, and the structural state of E2PCa 2 thus produced with the mutants is very stable; therefore, the Ca 2ϩ -deocclusion/release from E2PCa 2 and the reverse change from E2PCa 2 to E1PCa 2 are blocked.
Ca 2ϩ Release from E2PCa 2 after the E1PCa 2 to E2PCa 2 Isomerization-After the E1PCa 2 to E2PCa 2 isomerization, our results showed that further structural changes should take place with a critical contribution of the A/M1 linker to rearrange the transmembrane helices and thereby deocclude Ca 2ϩ ions at the transport sites and release into lumen (E2PCa 2 3 E2P ϩ 2Ca 2ϩ ). Because the elongation of the A/M1 linker blocked the Ca 2ϩ deocclusion and release, it is clear that this linker needs to be appropriately short for this event. The structural requirement of the length of the linker is hence in sharp contrast to that for the preceding E1PCa 2 to E2PCa 2 isomerization, in which the linker needs to be sufficiently long. Our present and previous results (23) clearly showed the following: 1) the shortening of the A/M1 linker blocks the E1PCa 2 to E2PCa 2 isomerization; 2) its elongation on the other hand markedly accelerates this isomerization; and 3) its elongation blocks the subsequent Ca 2ϩ deocclusion and release from E2PCa 2 . These results strongly suggest that the A/M1 linker with its native and correct length in the wild type will be strained upon the motion of the A domain and its positioning above half the P domain in the E1PCa 2 to E2PCa 2 isomerization, and the strain thus imposed in E2PCa 2 will be utilized to cause the subsequent structural changes for the Ca 2ϩ deocclusion and release, E2PCa 2 3 E2P ϩ 2Ca 2ϩ . Then the question is what structural changes are produced by critical contributions of the A/M1 linker with its correct length, i.e. its strain, and how the changes are transmitted to the transmembrane region to rearrange the helices to deocclude and release Ca 2ϩ by the contribution of the A/M1 linker.
The essential changes of the transmembrane helices for the Ca 2ϩ release related with the motions of the cytoplasmic domains are seen in E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ as the model for the overall structural change in the EP isomerization/Ca 2ϩrelease process E1PCa 2 3 E2P ϩ 2Ca 2ϩ . As described in detail by Toyoshima and co-workers (13,51), the inclination of the P domain (i.e. its moving and tilting to the A domain) causes the sideward shift of the cytoplasmic part of M4 (M4C), downward movement of M4, bending of M5C, and also rotation of M6; these changes would destroy the Ca 2ϩ -binding sites. Also critical is the tilting of M2, which forms a V-shaped rigid body structure with M1Ј/M1 by van der Waals contacts. This rigid body moves (M1Ј rotates and M2/M1 tilts) as the A domain largely rotates (see Fig. 13 and Fig. 4 in Ref. 13). Especially, the top part of M2 at its junction with the A domain largely moves outward, and thus M2 largely tilts and pushes against the luminal half of M4 via M1 to open the luminal gate (13).
Because the A/M1 linker connects directly to the A domain with M1Ј/M1, it would be easy at first glance to assume that this linker functions for transmitting the motion of the A domain directly to the M1Ј/M1/M2 rigid body to cause its motion, and such linkage would be impaired by the elongation of the A/M1 linker. In more detail in the change E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ , M1Ј rotates away from its original position as the A domain rotates, and the top part of M2 comes to the position where M1Ј was originally positioned, and the top part of M1 also moves sideward (see Fig. 4 in Ref. 13). If the A/M1 linker is elongated, the linker, for example, cannot pull M1Ј/M1, and the coordination between the motion of the A domain and that of M1Ј/ M1/M2 would be impaired or not enough for the Ca 2ϩ deocclusion and release.
Nevertheless, to further clarify the critical structural contributions of the A/M1 linker in the Ca 2ϩ -deocclusion/release process, it is necessary to take into account the actual structural difference revealed between E2PCa 2 and the Ca 2ϩ -released form of E2P. Namely, in E2PCa 2 , the Tyr 122 -hydrophobic cluster is not properly formed yet as demonstrated by the rapid PrtK cleavage at Leu 119 in sharp contrast to its complete resistance with the properly formed cluster in the Ca 2ϩ -released form of E2P (Fig. 12). Most importantly, this cluster is produced by the largely moved three structural elements and their appropriate positioning, i.e. by the largely rotated A domain, the largely tilted M2, and the largely inclined P domain (see E1⅐AlF x ⅐ADP 3 E2⅐MgF 4 2Ϫ in Fig. 13). Therefore, in the mutants with the elongated A/M1 linker, the motions of these elements are not yet achieved enough to form the cluster, and they are not stabilized at the appropriate positions because of the lack of the proper cluster. In this context, it should be noted that our present observations are in accord with the previous mutation studies on this cluster (19,22) that the formation of this cluster and resulting strong interactions of the three structural elements are critical for the Ca 2ϩ -released structure of E2P. Actually, in the mutants with the elongated A/M1 linker, the Ca 2ϩ -released form of E2P or its structural analogs were shown to possess the proper Tyr 122 -hydrophobic cluster (Fig. 12) and the luminally opened Ca 2ϩ -release pathway (Fig. 10).
All these findings directed us that for understanding the Ca 2ϩ -release process from E2PCa 2 , it is necessary to further consider and dissect the possible roles of the A/M1 linker in the motions and proper positioning of these three structural elements to form the cluster. This implies even a possible contribution of the A/M1 linker to the motion of the P domain from E2PCa 2 to E2P. Note that the rearrangements of M4/M5 are essential for the Ca 2ϩ deocclusion/release, and they must be linked with the motion of the P domain from the E2PCa 2 state because M4/M5 is directly connected with the P domain. As discussed above for the E1PCa 2 to E2PCa 2 isomerization, in E2PCa 2 half of the P domain is probably positioned already underneath the A domain and associated with this domain by forming the two interaction networks at the Val 200 loop and at the T 181 GES loop. For this A-P domain association in the EP isomerization, the P domain may be inclined to some extent (as also described in Footnote 8). It is possible that the A/M1 linker with its strain likely functions to cause a further motion of the P domain from E2PCa 2 via the associated A domain, thereby accomplishing the rearrangements of M4/M5 required for the Ca 2ϩ -deocclusion/release from E2PCa 2 . It is tempting to speculate that the A domain is pulled by the strain of the A/M1 linker and pushes down the P domain located underneath the A domain. Thereby the P domain inclines further from the well as the catalytic site with the normal hydrolytic activity (Fig.  7). Thus, the Tyr 122 -hydrophobic cluster formed with the critical contribution of the A/M1 linker functions both for the Ca 2ϩ -deocclusion/release at the transport sites and for the formation of the catalytic site with the hydrolytic activity. Most importantly, as a consequence of this structural mechanism, a possible acylphosphate hydrolysis in E2PCa 2 without releasing Ca 2ϩ will be avoided, and hence the reaction sequence for the Ca 2ϩ release and subsequent acylphosphate hydrolysis will be ensured, thus accomplishing the energy coupling in the Ca 2ϩ transport.
During the E2P hydrolysis to E2, the luminal Ca 2ϩ gate should be closed, and the ATPase prevents possible Ca 2ϩ leakage from the lumen. On the basis of the structural change E2⅐MgF 4 2Ϫ 3 E2(TG), it was predicted (13) that the luminal gate closure involves the release of the A domain from the P domain (the loss of the interactions at the T 181 GES loop and Val 200 loop regions) and a tilting of the A domain upon the dephosphorylation (see Fig. 6 in Ref. 13). An interesting and specific question for us here was whether the strain of the A/M1 linker possibly contributes to the tilting of the A domain during the E2P hydrolysis and thus to the closure of the luminal gate. But this seems unlikely. This is because the rate in the E2P hydrolysis and the Ca 2ϩ affinity and rate in the E2 to E1Ca 2 transition in the mutants with the elongated A/M1 linker are normal ( Fig. 7 and Table 1), and therefore the normal structural changes probably take place during the E2P hydrolysis to E2 in the mutants as in the wild type.
Finally, it is worth noting that the strain of the A/M3 linker has been predicted to be important for the large rotation of the A domain in the EP isomerization (13,53). The A/M1 linker, the A/M2 linker (top part of M2), including Tyr 122 /Leu 119 , and the A/M3 linker therefore play distinct structural roles for the successive structural events in the E1PCa 2 to E2PCa 2 isomerization, the Ca 2ϩ -deocclusion/release, and the E2P hydrolysis. In the transport cycle, the energy gained by the Ca 2ϩ and ATP bindings will be transformed into the structural states of the phosphorylated intermediates producing the strain of the linkers and the motions and strong interactions of the A and P domains, and the conformational energy thus produced is utilized for the structural changes to deocclude and release Ca 2ϩ into lumen. The usefulness for the Ca 2ϩ -ATPase having flexible links between cytoplasmic domains and the transmembrane domain is also discussed extensively in the very recent paper by Toyoshima and co-workers (51).
In summary, our studies revealed the structural requirements of the length of the A/M1 linker. The linker needs to be as follows: 1) sufficiently long for the E1PCa 2 to E2PCa 2 isomerization, 2) appropriately short for the subsequent Ca 2ϩ deocclusion and release, E2PCa 2 3 E2P ϩ 2Ca 2ϩ , and 3) then again sufficiently long for the E2P hydrolysis, E2P ϩ H 2 O 3 E2 ϩ P i . The native length of the A/M1 linker in the wild type is therefore precisely designed for inducing and coordinating the successive structural events in these steps and for the energy coupling. On the basis of the results and structures, we dissected the structural events for these processes and the functions of the structural elements involved in the processes.