Ca2+ Release to Lumen from ADP-sensitive Phosphoenzyme E1PCa2 without Bound K+ of Sarcoplasmic Reticulum Ca2+-ATPase*

During Ca2+ transport by sarcoplasmic reticulum Ca2+-ATPase, the conformation change of ADP-sensitive phosphoenzyme (E1PCa2) to ADP-insensitive phosphoenzyme (E2PCa2) is followed by rapid Ca2+ release into the lumen. Here, we find that in the absence of K+, Ca2+ release occurs considerably faster than E1PCa2 to E2PCa2 conformation change. Therefore, the lumenal Ca2+ release pathway is open to some extent in the K+-free E1PCa2 structure. The Ca2+ affinity of this E1P is as high as that of the unphosphorylated ATPase (E1), indicating the Ca2+ binding sites are not disrupted. Thus, bound K+ stabilizes the E1PCa2 structure with occluded Ca2+, keeping the Ca2+ pathway to the lumen closed. We found previously (Yamasaki, K., Wang, G., Daiho, T., Danko, S., and Suzuki, H. (2008) J. Biol. Chem. 283, 29144–29155) that the K+ bound in E2P reduces the Ca2+ affinity essential for achieving the high physiological Ca2+ gradient and to fully open the lumenal Ca2+ gate for rapid Ca2+ release (E2PCa2 → E2P + 2Ca2+). These findings show that bound K+ is critical for stabilizing both E1PCa2 and E2P structures, thereby contributing to the structural changes that efficiently couple phosphoenzyme processing and Ca2+ handling.

Sarcoplasmic reticulum (SR) 2 Ca 2ϩ -ATPase (SERCA1a) catalyzes Ca 2ϩ transport coupled with ATP hydrolysis against an ϳ10,000-fold concentration gradient (1)(2)(3)(4)(5)(6)(7)(8)(9). The ATPase is first activated by the binding of two cytoplasmic Ca 2ϩ ions at the transport sites with a submicromolar high affinity (E2 to E1Ca 2 , see step 1 in Fig. 1) and then autophosphorylated at Asp 351 by ATP to form a phosphoenzyme intermediate (EP) (step 2). This EP is "ADP-sensitive" (E1P) because it is rapidly dephosphorylated by ADP in the reverse reaction. Upon E1P formation, the bound Ca 2ϩ ions are occluded in the transport sites (E1PCa 2 ). Subsequently, E1PCa 2 undergoes its isomeric transition to an ADP-insensitive form (E2P), i.e. loss of ADP sensitivity, which results in a large reduction of Ca 2ϩ affinity and opening of the lumenal release gate, i.e. Ca 2ϩ deocclusion and release (steps [3][4]. Ca 2ϩ release in step 4 is very rapid, so that an E2PCa 2 intermediate state does not accumulate and in fact had never been found until we recently established its existence (10 -13) and successfully trapped it for the first time (14). Finally, E2P is hydrolyzed back to the inactive E2 form (step 5).
In E1PCa 2 3 E2P ϩ 2Ca 2ϩ , the A domain rotates parallel to the membrane plane and the P domain inclines to the A domain, thereby associating with each other to produce a compactly organized and inclined headpiece (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27). This tight structure is stabilized by critical interaction networks between the A and P domains at three regions (10 -14) (see Fig. 9 for details). The rotation and inclination of the domains result in motions and rearrangements of the transmembrane helices thereby disrupting the Ca 2ϩ sites and opening the lumenal gate. In the P domain, there is a specific K ϩ binding site (28); K ϩ binding here is crucial for rapid hydrolysis of E2P (28 -30). Recently, we further found (13) that the K ϩ in E2P is critical for reducing the lumenal Ca 2ϩ affinity that is required to achieve the high physiological Ca 2ϩ gradient and for rapid Ca 2ϩ release (E2PCa 2 3 E2P ϩ 2Ca 2ϩ ). Thus, bound K ϩ contributes to stabilization of the compactly organized and inclined E2P structure with its disrupted Ca 2ϩ sites and fully opened lumenal gate, probably by cross-linking the P domain with the A domain/M3-linker (13).
Despite these findings on the Ca 2ϩ release process and E2P, a possible role for K ϩ in E1PCa 2 has not been explored. The K ϩ site is situated at the bottom of the P-domain near the cytoplasmic ends of the transmembrane helices. Therefore, the lack of K ϩ binding might have a serious effect on the stability of the helices and Ca 2ϩ handling in E1PCa 2 . The E2-E1Ca 2 transition is markedly retarded, and its equilibrium is affected by the absence of K ϩ (31)(32)(33).
In this study, we explore a possible role of K ϩ in E1PCa 2 especially in regard to Ca 2ϩ occlusion. Results reveal that K ϩfree E1PCa 2 has an open Ca 2ϩ pathway to the lumen. Thus, the Ca 2ϩ binding sites face the lumen, and Ca 2ϩ can be released. The absence of K ϩ does not reduce the high Ca 2ϩ affinity (Ca 2ϩ site coordination probably unchanged), and yet the cytoplasmic gate is closed, and the lumenal gate is open. These changes probably do not involve large motions of the cytoplasmic domains and transmembrane helices. Therefore, bound K ϩ likely stabilizes the Ca 2ϩ occluded structure of E1PCa 2 by simply keeping the lumenal Ca 2ϩ pathway closed.
The structural role of K ϩ in E1PCa 2 is discussed in detail using crystal structures of Ca 2ϩ -ATPase with bound K ϩ .

EXPERIMENTAL PROCEDURES
Preparation of SR Vesicles-SR vesicles were prepared from rabbit skeletal muscle as described (34). The phosphorylation site content in the vesicles determined according to Barrabin et al. (35) was 4.49 Ϯ 0.22 nmol/mg vesicle protein (n ϭ 5).
Determination of EP-SR vesicles were phosphorylated with [␥-32 P]ATP as described in the legends to Figs. 2-6. In the experiments performed in Fig. 2, aliquots of the reaction mixture were spotted on the HAWP membrane filter (Millipore) and washed continuously with a chasing solution for the periods indicated. At the end of chase, the reaction was terminated by washing with 0.1 M HCl. To determine the amount of E2P in the phosphorylation mixture, the membrane was washed with an ADP solution for 1 s and then with 0.1 M HCl. The membrane was dried, and the radioactivity was measured by digital autoradiography. In Figs. 3 and 6, total EP was measured by quenching the phosphorylation reaction (in a test tube) with 5% (v/v) ice-cold trichloroacetic acid containing P i , whereas for E2P determination, the reaction was chased with ADP for 1 s and quenched by addition of the trichloroacetic acid. The precipitated proteins were separated by 5% SDS-PAGE at pH 6.0 according to Weber and Osborn (36). The radioactivity associated with the separated Ca 2ϩ -ATPase was quantitated by digital autoradiography (37). Rapid kinetic measurements in Fig. 3 were performed with a handmade rapid mixing apparatus (38).
Determination of Bound Ca 2ϩ -In the experiments performed in Figs. 2 and 6, SR vesicles were incubated with 45 CaCl 2 as per the figure legends, and an aliquot of reaction mixture was spotted on the HAWP membrane filter (Millipore). Then, the membrane was perfused with a chasing solution for indicated time periods using a rapid filtration apparatus RFS-4 (Bio-Logic, Claix, France). To estimate nonspecific 45 Ca 2ϩ binding, the same experiments were done in the presence of 1 M thapsigargin. Specific 45 Ca 2ϩ binding was obtained after subtracting this nonspecific binding.
Ca 2ϩ Uptake into SR Vesicles in a Single Turnover of EP-In the experiments performed in Figs. 4 and 5, SR vesicles were incubated with 45 Ca 2ϩ , and a single turnover of EP was initiated by adding ATP and excess EGTA using the handmade rapid mixing apparatus. After chasing the reaction, the mixture was spotted on the membrane filter and washed for ϳ10 s by an EGTA solution, as described in the figure legends. The background level of 45 Ca 2ϩ was determined without ATP and subtracted. This background level was Ͻ3% of the maximum Ca 2ϩ uptake level.
Miscellaneous-All of the reactions were performed at 4°C in 7 mM MgCl 2 and 50 mM MOPS/Tris (pH 7.3). Protein concentrations were determined by the method of Lowry et al. (39) with bovine serum albumin as a standard. Free Ca 2ϩ concentrations were calculated by the Calcon program. Data were analyzed by nonlinear regression using the program Origin (Microcal Software, Inc., Northampton, MA). Three-dimensional models of the enzyme were produced by the program VMD (40).

Time Courses of EP Decay and Ca 2ϩ
Release-The Ca 2ϩ -ATPase in SR vesicles was phosphorylated with MgATP in the presence of 0.1 M K ϩ , 10 M Ca 2ϩ , and Ca 2ϩ -ionophore A23187 (Fig. 2, A and B). The reaction reaches steady state within a few seconds, and almost all of the Ca 2ϩ -ATPase is in the ADP-sensitive form of EP (E1P) because of the rate-limiting E1P to E2P transition followed by rapid E2P hydrolysis in the presence of K ϩ (29,30).
When the reaction was chased with excess EGTA in 0.1 M K ϩ , the amount of EP decreases in a single exponential time course, and the EP during the decay is almost all ADP sensitive ( Fig. 2A). The bound Ca 2ϩ decreases concomitantly with E1P decay, i.e. E1PCa 2 to E2P transition. The result agrees with the established mechanism that the two Ca 2ϩ ions are occluded in E1PCa 2 , and Ca 2ϩ release into the lumen occurs very rapidly after the rate-limiting E1PCa 2 to E2PCa 2 transition, E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ (11)(12)(13)(14). Thus, the EP transition and Ca 2ϩ release are tightly coupled in the presence of K ϩ . Surprisingly, when E1PCa 2 formed as above in 0.1 M K ϩ , and A23187 was chased with excess EGTA in the absence of K ϩ , the Ca 2ϩ release is considerably (ϳ3ϫ) faster than the E1P decay via its transition to E2P (Fig. 2B). The result shows that in the absence of K ϩ , there is an E1P species without bound Ca 2ϩ and that the Ca 2ϩ ions are released from E1PCa 2 . We found essentially the same results in the presence of choline chloride in place of LiCl without K ϩ (data not shown). 45 Ca 2ϩ Uptake in Single Turnover of E1PCa 2 -We then examined whether this rapid Ca 2ϩ release from E1PCa 2 in the absence of K ϩ upon the EGTA chase occurs to the lumenal side or cytoplasmic side of the membrane. For this purpose, we performed a 45 Ca 2ϩ uptake assay in a single turnover of E1PCa 2 in the absence of ionophore, i.e. with sealed SR vesicles. In Fig. 3, for the single turnover of E1PCa 2 , the Ca 2ϩ -ATPase in the vesicles in 10 M Ca 2ϩ was phosphorylated by a simultaneous addition of [␥-32 P]ATP and excess EGTA in either the presence or the absence of 0.1 M K ϩ . Approximately half of the ATPase is phosphorylated rapidly to form E1PCa 2 both in the presence and absence of K ϩ , and then EP decays slowly, in contrast to the full phosphorylation achieved without the removal of Ca 2ϩ . In sealed vesicles (without A23187), EP decays at the same rate in the presence or absence of K ϩ . In the presence of K ϩ , nearly all EP is E1P (ADP- Ca 2؉ Release from E1PCa 2 of Ca 2؉ -ATPase DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 sensitive), whereas in the absence of K ϩ , E2P increases slowly to ϳ20% at ϳ2 s of the maximum amount of EP formed immediately after the ATP addition.
Then, in Fig. 4 (closed circles), the 45 Ca 2ϩ uptake assay during a single turnover of E1PCa 2 was performed by membrane filtration with an EGTA chase, i.e. with extensive EGTA washing of the filter for ϳ10 s under otherwise the same conditions as in the single turnover of E1PCa 2 in Fig. 3. During the ϳ10 s of EGTA washing, nearly all EP was dephosphorylated ( Fig. 3) as we intended; therefore, all of the bound 45 Ca 2ϩ in EP was released even at the first time point (0.1 s after the start when nearly all EP is E1PCa 2 ) either to the cytoplasmic side or lumenal side. If released to the cytoplasmic side, the 45 Ca 2ϩ will be lost from the filter by the EGTA wash, and levels will be reduced significantly from the ideal stoichiometry of two Ca 2ϩ ions transported in a single turnover of E1PCa 2 . However, the results (closed circles) clearly show a maximum uptake of ϳ1.7 Ca 2ϩ per EP in 0.1 M K ϩ and an even higher uptake of 1.8ϳ1.9 without K ϩ , very close to the ideal stoichiometry. Therefore, during a single turnover, the bound 45 Ca 2ϩ ions in E1PCa 2 formed in the absence of K ϩ are not released to the cytoplasmic side but to the lumen. It is concluded that in E1PCa 2 without K ϩ , the cytoplasmic gate is closed, but a Ca 2ϩ pathway to the lumen exists. Thus, the Ca 2ϩ binding sites face the lumen.
ADP Chase during Single Turnover of 45 Ca 2ϩ Uptake-In Fig. 4 (open circles), we assessed at each time point during the single turnover of E1PCa 2 the amount of 45 Ca 2ϩ remaining on the filter with the vesicles. For this purpose, we chased the reaction with ADP and excess EGTA at each time point, i.e. dephosphorylating to E1Ca 2 very rapidly in the reverse reaction and removing 45 Ca 2ϩ released to the cytoplasmic side. Both in the presence and absence of K ϩ , at 0.1 s (first time point) immediately after the ATP/EGTA addition, nearly maximum EP is already formed (all E1PCa 2 , Fig. 3), and all of the bound 45 Ca 2ϩ is removed by the ADP chase. Then, in the presence of 0.1 M K ϩ (A), the ADP-insensitive fraction and the amount of 45 Ca 2ϩ released into the lumen increased exponentially due to the forward E1PCa 2 decay via its transition to E2P with Ca 2ϩ release as expected from the established transport mechanism. In fact, the time course agreed with that of EP decay via the rate-limiting E1PCa 2 to E2P transition (Fig. 3, closed triangles).
On the other hand, in the absence of K ϩ (B), 45 Ca 2ϩ , and the ADP-insensitive fraction increase very rapidly (within the initial ϳ0.5 s) and suddenly slow, showing a clear biphasic time course. The second slow phase occurs at nearly the same  To determine the amount of E2P (squares), the phosphorylated sample was mixed with an equal volume of a solution containing 2 mM ADP and 5 mM EGTA, and then the reaction was terminated by trichloroacetic acid at 1 s after the ADP addition.
rate as the EP decay via the E1P to E2P transition (Fig. 3, open triangles) and the single exponential 45 Ca 2ϩ uptake in the presence of K ϩ (A), i.e. the normal transport process E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ . The initial rapid phase occurs at a significantly faster rate and to a higher extent than in E2P formation (Fig. 3) and therefore cannot be accounted for simply by formation of E2P. Actually, the initial phase is even faster than the Ca 2ϩ release from K ϩ -free E1PCa 2 revealed upon excess EGTA addition (without ADP) in A23187 in Fig. 2. The results suggest that, in K ϩ -free E1PCa 2 , different types of Ca 2ϩ sites are produced in the initial rapid phase; the Ca 2ϩ ions are not released to the cytoplasmic side even upon ADP-induced reverse dephosphorylation.
Behavior of 45 Ca 2ϩ at Site I in E1PCa 2 -We examined whether the above observed biphasic kinetics revealed by the ADP chase is related to the heterogeneity of the Ca 2ϩ sites I and II in E1Ca 2 . In E1Ca 2 , Ca 2ϩ bound at site II is rapidly exchanged with the cytoplasmic Ca 2ϩ , and the Ca 2ϩ bound at the deeper site I can be released to the cytoplasm only when site II is vacant (15,(41)(42)(43)(44). Therefore, we first labeled site I with 45 Ca 2ϩ by exchanging the site II-bound 45 Ca 2ϩ with nonradioactive Ca 2ϩ (supplemental Fig. S1). In Fig. 5B, we clearly observed a biphasic 45 Ca 2ϩ increase in the ADP-insensitive fraction in the absence of K ϩ as in Fig. 4B. The only difference is that, as expected, the total amount of 45 Ca 2ϩ uptake (0.8 -1.0 Ca 2ϩ per EP) is half of that in Fig. 4 in which both sites I and II are labeled by 45 Ca 2ϩ . The results show that the heterogeneity of the two Ca 2ϩ sites I and II in E1Ca 2 is not related to the biphasic 45 Ca 2ϩ increase revealed by the ADP chase in Fig. 4B. Furthermore, we observed a nonsequential release of two Ca 2ϩ ions from E1PCa 2 to the lumenal side upon removal of free Ca 2ϩ in the presence of A23187 without ADP, which therefore is not related to the biphasic 45 Ca 2ϩ increase in Fig. 4B (supplemental Fig. S2) (58,59).
These results show that there are two different types of E1PCa 2 , i.e. the normal Ca 2ϩ -occluded E1PCa 2 and another E1PCa 2 species that possesses lumen-facing Ca 2ϩ binding sites (opened lumenal pathway) and a closed cytoplasmic gate. The results further indicate that in the absence of K ϩ , the E1PCa 2 species with the lumen-facing Ca 2ϩ binding sites is rapidly produced from normal E1PCa 2 , and this process is revealed by the ADP chase as the initial rapid phase in Fig. 4B (see more in "Discussion" and a schematic model in Fig. 7).
Affinity of E1P for Lumenal Ca 2ϩ in Absence of K ϩ -In Fig.  6, we assessed the Ca 2ϩ affinity of the transport sites exposed to the lumen in K ϩ -free E1PCa 2 by determining the Ca 2ϩ binding to E1P in steady state in the presence of A23187. In Fig. 6A, the total amount of EP increased with increasing Ca 2ϩ concentration and reached its maximum level at ϳ0.5 M Ca 2ϩ due to high affinity Ca 2ϩ binding at the transport sites (E2 to E1Ca 2 transition). The total amount of EP at saturating Ca 2ϩ was half of the maximum Ca 2ϩ binding in E1Ca 2 (B); therefore, all Ca 2ϩ -ATPases are phosphorylated at saturating Ca 2ϩ . As replotted in Fig. 6C, ϳ60% of the maximum total amount of EP was E1P in steady state at saturating Ca 2ϩ under these conditions.
In Fig. 6B, the amount of bound Ca 2ϩ in steady state in the presence of A23187 was determined without washing the filter so as not to alter the equilibrium. As replotted in Fig. 6C with % values relative to the maximum Ca 2ϩ binding in E1Ca 2 , the bound Ca 2ϩ under the phosphorylating condition without K ϩ increases concomitantly with an increase in E1P, and their relative values are nearly the same. Note that if the affinity of the lumen-facing Ca 2ϩ sites of E1P without K ϩ is significantly lower than that of the high Ca 2ϩ affinity in E1 for the phosphorylation, the Ca 2ϩ binding curve would be shifted In the absence of the Ca 2ϩ ionophore, SR vesicles (SRV; 20 g/ml) were first incubated with 10 M 45 CaCl 2 for ϳ10 min, and then Ca 2ϩ uptake in a single turnover of EP was initiated by mixing with an equal volume of a solution containing 20 M ATP and 2 mM EGTA, as described in Fig. 3. After the indicated periods, the reaction was chased with an equal volume of a solution containing 2 mM EGTA without (closed circles) or with (open circles) 2 mM ADP. The mixture was immediately spotted on the membrane and washed for ϳ10 s with 1 ml of a 2 mM EGTA solution. The amount of 45 Ca 2ϩ on the membrane, i.e. transported into the vesicles and/or remained bound to the ATPase and not released to cytoplasmic side, was normalized to the maximum total amount of EP formed immediately after the addition ATP and EGTA (Fig. 3). In A, the time course obtained with the ADP chase was best described by a single exponential Ca 2ϩ uptake (solid line) with a rate constant of 0.49 s Ϫ1 and maximum Ca 2ϩ /EP value of 1.26. In B, it was best described by a double exponential (broken line) with a rate constant and maximum Ca 2ϩ /EP value of 5.1 s Ϫ1 and 0.66 for the fast phase and 0.24 s Ϫ1 and 0.98 for the slow phase (but it was not described by a single exponential increase shown by solid line with the rate constant of 1.54 s Ϫ1 and maximum value of 1.31). Note also that without the ADP addition, almost of all the bound Ca 2ϩ ions are transported into the vesicles during the ϳ10-s EGTA wash because the single turnover of EP is nearly completed in this period (see Fig. 3). DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 significantly to higher Ca 2ϩ concentrations, and the relative value of the bound Ca 2ϩ would become significantly smaller than that of E1P in the 0.1-10 M range. However, this is obviously not the case. We conclude that the affinity of the lumen-facing Ca 2ϩ sites of K ϩ -free E1P is as high as the cytoplasmic Ca 2ϩ affinity in E1.

DISCUSSION
Ca 2ϩ Release from E1PCa 2 in Absence of K ϩ -Our studies show that in the absence of K ϩ , Ca 2ϩ is released from E1PCa 2 to the lumenal side. This Ca 2ϩ release obviously precedes the conversion of the ADP-sensitive EP (E1P) to ADP-insensi-  ) in B in the absence of K ϩ are replotted after normalization to the maximum total amount of EP and to the maximum 45 Ca 2ϩ binding under the nonphosphorylating condition (E1) in the absence of K ϩ , respectively, and shown as % values. Solid lines show the least squares fit to the Hill equation, and the maximum values were 58% for E1P and 52% for bound 45 Ca 2ϩ , respectively. tive one (E2P); thus, there is a K ϩ -free E1P species without bound Ca 2ϩ (Fig. 2B). Evidently, a Ca 2ϩ pathway from the transport sites to the lumen is open at least to some extent in this species. K ϩ , probably bound to its specific site in the ATPase (28), therefore plays a critical role in E1PCa 2 to stabilize the transport sites in an occluded state. Notable also is our finding that the Ca 2ϩ affinity of the sites facing the lumen in K ϩ -free E1PCa 2 is as high as the cytoplasmic Ca 2ϩ affinity in the unphosphorylated E1 state (Fig. 6). Thus, the Ca 2ϩ binding sites are not disrupted in this K ϩfree E1PCa 2 structure, suggesting that the opening of the lumenal Ca 2ϩ pathway does not involve large structural changes such as those that occur during the EP conformation change. The observation also means that such a Ca 2ϩ -ATPase species cannot be involved in producing a Ca 2ϩ gradient across the membrane and therefore is unlikely to contribute significantly to active Ca 2ϩ transport. This is because, without a reduction in Ca 2ϩ affinity, lumenal Ca 2ϩ would rebind at low concentrations and inhibit the pump.
Our kinetic analysis of the lumenal Ca 2ϩ -induced reverse conversion E2P ϩ 2Ca 2ϩ 7 E2PCa 2 7 E1PCa 2 in wild type Ca 2ϩ -ATPase (13) has revealed that the K ϩ in E2P is critical for lowering the lumenal Ca 2ϩ affinity and for fully opening the lumenal gate, thereby accomplishing the high physiological Ca 2ϩ gradient and rapid Ca 2ϩ release E2PCa 2 3 E2P ϩ 2Ca 2ϩ . K ϩ stabilizes the E2P structure with disrupted Ca 2ϩ sites and a fully open lumenal gate. In the absence of K ϩ , the lumenal Ca 2ϩ affinity of E2P is ϳ2000 times lower than in E1P (K 0.5 values, 0.4 mM (13) and 0.15 M (Fig. 6), respectively). Therefore, the large structural change associated with the EP conformation change is obviously required, even in the absence of K ϩ , for disrupting the Ca 2ϩ sites. K ϩ binding in E2PCa 2 /E2P further reduces the Ca 2ϩ affinity to a level (K 0.5 value, 1.5 mM (13)) appropriate for producing the high physiological Ca 2ϩ gradient across the membrane.
Thus, bound K ϩ stabilizes both the Ca 2ϩ occluded structure of E1PCa 2 and the Ca 2ϩ -released structure of E2P. Thereby, K ϩ critically contributes to the successive structural changes and ensures strict and efficient coupling for EP processing and Ca 2ϩ handling in E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ , key events for Ca 2ϩ transport. Also notable is the fact that the K ϩ bound in the P domain is crucial for producing a catalytic site structure in E2P appropriate for its accelerated hydrolysis (28 -30).
Biphasic Ca 2ϩ Release in ADP Chase of Single Turnover of E1PCa 2 without K ϩ -In Fig. 7, we provide a schematic model to show the roles of K ϩ in the Ca 2ϩ transport and to account for the biphasic Ca 2ϩ release from K ϩ -free E1PCa 2 following an ADP chase during a single turnover (Fig. 4B, open circles). The fast initial phase may be accounted for by the rapid formation of sE1PCa 2 , with lumen-facing, high affinity Ca 2ϩ binding sites, in rapid equilibrium with normal E1PCa 2 . The bound 45 Ca 2ϩ ions cannot be released to the cytoplasmic side even upon ADP-induced reverse dephosphorylation (sE1Ca 2 ) but only to the lumenal side (yellow arrow). Because sE1P has high affinity, Ca 2ϩ rebinding occurs at low lumenal FIGURE 7. Schematic model for roles of K ؉ in EP processing and Ca 2؉ handling in Ca 2؉ transport. sE1PCa 2 is an E1PCa 2 species formed without K ϩ possessing a closed cytoplasmic gate and lumen-facing Ca 2ϩ binding sites (an opened lumenal pathway) with high Ca 2ϩ affinity (Fig. 6). sE1PCa 2 is in rapid equilibrium with the normal E1PCa 2 . Here, s denotes silent because this species is apparently absent in the presence of K ϩ and also because the bound Ca 2ϩ ions are not released to the cytoplasmic side even upon ADP-induced reverse dephosphorylation (to sE1Ca 2 ) in contrast to the normal E1PCa 2 reverse dephosphorylation. Actual active Ca 2ϩ transport is achieved by a large reduction of the Ca 2ϩ affinity during the normal sequence E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ (blue arrows). The schematic is based on crystal structural models for the ADP-sensitive and -insensitive EP states and E1Ca 2 , with the positions of the cytoplasmic N, P, and A domains, and membrane (orange layer) being approximate. The Ca 2ϩ sites in the transmembrane domain are depicted as occluded (closed cytoplasmic and lumenal gates) in normal E1PCa 2 , as lumen-facing and high Ca 2ϩ affinity with the closed cytoplasmic gate in sE1PCa 2 and sE1Ca 2 , and as lumenally opened with reduced Ca 2ϩ affinity in E2P and E2PCa 2 (immediately before the Ca 2ϩ release). DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 concentrations 3 and inhibits flux through this pathway. The slow second phase (Fig. 4B) most probably reflects the E1PCa 2 to E2P transition as in the single exponential Ca 2ϩ uptake in 0.1 M K ϩ (Fig. 4A and Fig. 7, blue arrows). The formation of sE1PCa 2 in rapid equilibrium with occluded E1PCa 2 necessarily lowers the steady state level of the latter species and hence Ca 2ϩ transport through the normal route. Thus, although progression to sE1PCa 2 is relatively fast, this pathway cannot contribute to gradient formation and ultimately slows normal transport. It is concluded that K ϩ ensures the normal structural process for Ca 2ϩ transport (blue arrows) by stabilizing the Ca 2ϩ -occluded structure of E1PCa 2 and disallowing opening of a lumenal Ca 2ϩ pathway (this study), and by stabilizing the E2P structure with disrupted Ca 2ϩ sites (greatly reduced affinity) and a fully opened lumenal gate (13).

Ca 2؉ Release from E1PCa 2 of Ca 2؉ -ATPase
Structural Role of Bound K ϩ in E1PCa 2 -The crystal structures provide a likely structural role of bound K ϩ in E1PCa 2 . In structures analogous to K ϩ -bound E1PCa 2 (E1PCa 2 ⅐AMPPN (22) and E1Ca 2 ⅐AlF 4 Ϫ ⅐ADP as well as E1Ca 2 ⅐AMPPCP (17)), K ϩ is specifically bound at the bottom part of the P domain and coordinated by the backbone carbonyl oxygens of Leu 711 , Lys 712 , and Ala 714 on P␣6 (sixth Pdomain ␣-helix) (near the catalytic Mg 2ϩ site Asp 703 /Asp 707 on P␣5 of this region) and by the Glu 732 side chain oxygen on P␣7 (Fig. 8). The importance of Glu 732 in the K ϩ -induced acceleration of E2P hydrolysis was shown through mutations (28).
The K ϩ ion and these ligands are distant from and not in direct contact with the transport sites from which Ca 2ϩ release occurs. On the other hand, adjacent to the K ϩ binding site on P␣6/P␣7 is P␣1, which is directly linked with the cytoplasmic end of M4 within the P domain. P␣6, P␣7, and P␣1 constitute the bottom part of one-half of the P domain and move together as a body during the transport cycle (7,18). Furthermore, P␣1 forms a hydrogen-bonding network with L6 -7 (a cytoplasmic short loop linking M6 and M7) and top parts of M3/M5. This interaction network is critical for proper arrangement of the transmembrane helices (48 -50). In fact, disruption of this network by mutations causes a marked retardation of the E2-E1 transition (48,49).
Because the bound K ϩ is deeply embedded and ligated within this part of the P domain (Fig. 8a), its absence would allow more flexibility of the structural components, such as 3 Note that the intravesicular volume of SR vesicles has been estimated to be in the range of 2-10 l/mg protein (45,46), and therefore, the release of Ca 2ϩ bound in EP (ϳ8 nmol/mg protein) into the lumen in a single turnover might increase the lumenal Ca 2ϩ to ϳ0.8 -4 mM. Although a fair amount of lumenal free Ca 2ϩ may be removed by low affinity Ca 2ϩ buffers such as calsequestrin (47), even a small rise in the lumenal Ca 2ϩ level might result in rebinding of lumenal Ca 2ϩ to sE1P because of its high affinity revealed in Fig. 6 (yellow arrow in Fig. 7). segmental fluctuations or wobbling, which in turn would impinge on the cytoplasmic regions of the transmembrane helices and probably destabilize the interaction network P␣1/L6 -7/M3/M5. The absence of K ϩ in fact markedly retards the E2 to E1 transition (31,32), and, as noted above, disruption of the P␣1/M3/M5/L6 -7 interaction network markedly retards the E1-E2 transition and also the E1P to E2P conformation change (48 -50). Opening of the lumenal pathway and Ca 2ϩ release from E1PCa 2 may be caused by such structural perturbations in the absence of bound K ϩ . As shown in the view from the lumen of the helices M4/ M5/M6/M8 ligating Ca 2ϩ in Fig. 8c, the space surrounded by these helices seems to be the only possible Ca 2ϩ exit pathway. M3 is in close contact at the lumenal end with the lumenal part of M4 (M4L), and they are connected by a short lumenal loop (L3-4). During the EP conformation change and subsequent Ca 2ϩ release (E1PCa 2 3 E2P ϩ 2Ca 2ϩ ), M3 and M4L incline together and move outward, thereby opening the putative Ca 2ϩ release pathway (lumenal gate) (19). The M3/M4L motion is produced by the large rotation and inclination of the A and P domains and by the consequent significant motions and rearrangements of the helices M1ϳ6, in which M1/M2 as a rigid body pushes M4L to open the Ca 2ϩ release gate ( Fig. 9) (19). The large motions concomitantly disrupt the Ca 2ϩ binding sites and reduce the Ca 2ϩ affinity (19). In K ϩ -free E1PCa 2 (ADP sensitive), these domain motions have not yet taken place, and the Ca 2ϩ sites are not disrupted and maintain a high affinity. Here, these motions are likely much less prominent and opening of the release pathway is simply the result of fluctuations and wobbling of the relevant helices, in particular M3/M4L.
The unique Ca 2ϩ coordination and particular make up of the M3 and M4 helices lend themselves to creating a release pathway while maintaining a high affinity. The Ca 2ϩ sites with properly positioned ligands are located at an unwound portion of the M4 helix creating intrinsic flexibility (Fig. 8).
On the other hand, M3 is a continuous helix from the cytoplasmic to the lumenal end and is located at the periphery of the transmembrane domain and is not closely associated with other helices including M1/M2 (except for M4L at the lumenal end). Thus, in the crystal structures analogous to E1PCa 2 , M3 seems not to have much steric restriction against possible outward movement, a shift that would open the Ca 2ϩ pathway. Therefore, if the cytoplasmic region of M3 is not fixed as occurs in the absence of bound K ϩ , its lumenal part and the associated M4L may become more mobile. Wobbling here could allow the Ca 2ϩ pathway to fluctuate between a closed and open state. The Ca 2ϩ sites are not necessarily disrupted because of the flexibility of the unwound structure of M4 and because the large motions of the A-P domains do not occur. (These are the motions that disrupt the Ca 2ϩ sites by inclining the cytoplasmic region of M4/M5.) Also, M3 is not involved directly in the Ca 2ϩ ligation.
Interestingly, at the lumenal end of M4L (Fig. 8c), there are bulky and hydrophobic residues (Tyr 294 /Tyr 295 /Lys 297 ), which may form hydrogen bonds, e.g. Tyr 294 /Tyr 295 with Glu 785 on L5-6. Lys 297 seems to seal the Ca 2ϩ channel (51). Tyr 295 is important for Ca 2ϩ transport activity and stabilizing E2 rela-tive to E1 (52). These residues may possibly function as the lumenal plug, and M3/M4L wobbling may destabilize their interactions helping to open the Ca 2ϩ pathway in K ϩ -free E1PCa 2 .
Importantly, in the crystal structures of analogues of E1PCa 2 , the cytoplasmic Ca 2ϩ gate is closed by the Ca 2ϩ ligand Glu 309 because Leu 65 on M1 locks the Glu 309 side chain configuration by van der Waals contact (8,9,18,53). Our observation shows that this cytoplasmic gate is closed in E1PCa 2 even without bound K ϩ , and therefore, the Glu 309 -gating with Leu 65 has not been affected.
Movement of K ϩ Binding Site during E1PCa 2 3 E2P ϩ 2Ca 2ϩ -K ϩ -bound crystal structures E1PCa 2 ⅐AMPPN and E2⅐AlF 4 Ϫ may be used as a model for the overall change in E1PCa 2 3 E2P ϩ 2Ca 2ϩ (Fig. 9). Hence, the P domain inclines to the A domain that also rotates and inclines (curved arrows), thus producing the A-P domain association in the most compactly organized and inclined headpiece structure, the Ca 2ϩ -released E2P. With this change, the cytoplasmic Ϫ are indicated by curved arrows. Note that the K ϩ site with bound K ϩ on the P domain moves down to the Gln 244 region on the A/M3-linker (blue arrow), thus likely cross-linking the P domain with the A/M3-linker. There are three critical interaction networks to realize and stabilize the compactly organized E2P structure. They are Tyr 122 HC forming a hydrophobic interaction cluster (violet Van der Waals spheres), the Val 200 loop (red loop), and TGES 184 (blue loop) (10 -13). Crystal structures of E2⅐BeF 3 Ϫ (21, 22), which are analogs of the E2P ground state (25), are not used here because they were formed without K ϩ (although the above noted changes are also seen with the E2⅐BeF 3 Ϫ crystals).
region of M4/M5 in the P domain inclines and disrupts the Ca 2ϩ sites (19). M2 inclines with the A domain motion and consequently M1, which forms a rigid V-shaped body with M2, pushes against the lumenal part of M4, and opens the lumenal gate (19). In these structural changes, the K ϩ site with bound K ϩ on the P domain moves down to the Gln 244 region on the A/M3linker (blue arrow) and brings in the Gln 244 side chain (or neighboring residues) as an additional coordination ligand. Thus bound K ϩ likely cross-links the bottom part of the P domain and the A/M3-linker. This cross-link must contribute to the stabilization of the compactly organized and inclined E2P structure with disrupted Ca 2ϩ sites and fully opened lumenal gate (13).
The A/M1Ј-linker of correct length has a critical function in inclining and compacting the E2P structure (14,27). The structure is stabilized by three critical interaction networks; at the Tyr 122 HC (hydrophobic interaction cluster involving the A and P domains and M2), at the Val 200 loop (ionic and hydrogen bonding interactions with the P domain residues), and at the TGES 184 loop (hydrogen bonding interactions with the P domain residue in the catalytic site) (10 -14). The TGES 184 loop of the rotated A domain protrudes into the catalytic site and blocks attack of ADP on the Asp 351 phosphate (causing the loss of ADP sensitivity). The Tyr 122 HC is produced upon A-P domain inclination induced by tension on the A/M1Јlinker (14,27) and is critical for reducing the Ca 2ϩ affinity and opening the lumenal gate, i.e. to deocclude/release Ca 2ϩ , E2PCa 2 3 E2P (11-13). All of the interaction networks are essential for these changes and are also necessary for the formation of the catalytic site with hydrolytic activity (10 -13). Importantly, the Val 200 loop and Tyr 122 HC are situated at the top and bottom of the A-P domain interface, respectively, and bound K ϩ is lower down and close to the membrane domain. Thus, these interaction networks including the K ϩ site are situated at positions most appropriate for stabilizing the compactly organized and inclined (thus strained) structure of E2P.
Ca 2ϩ Release into Cytoplasm and Uncoupling-It was previously observed with SR Ca 2ϩ -ATPase (54 -57) that Ca 2ϩ in E1PCa 2 can be released to cytoplasm upon direct hydrolysis to E1Ca 2 (not via its transition to E2P) under specific conditions such as with a raised lumenal Ca 2ϩ level. This causes ATP hydrolysis without Ca 2ϩ transport resulting in uncoupling. de Meis and co-workers (54 -56) further suggested that such uncoupled ATP hydrolysis functions as a heat-producing entity. This finding obviously differs from ours in that in K ϩfree E1PCa 2 , the Ca 2ϩ release pathway into the lumen is open, and the phosphoenzyme is not directly hydrolyzed.
In summary, we have found that E1PCa 2 without bound K ϩ has a perturbed structure with at least a partially open lumenal Ca 2ϩ release pathway and still with the Ca 2ϩ sites maintaining a high affinity. Thus, in the natural E1PCa 2 structure, bound K ϩ stabilizes the Ca 2ϩ in an occluded form by not allowing the pathway to open. Bound K ϩ also stabilizes E2P following disruption of the Ca 2ϩ sites and full opening of the lumenal gate (13). Thus, bound K ϩ has a crucial role in EP processing and Ca 2ϩ occlusion and release to the lumen in the sequence E1PCa 2 3 E2PCa 2 3 E2P ϩ 2Ca 2ϩ .