Membrane Perturbation of ADP-insensitive Phosphoenzyme of Ca2+-ATPase Modifies Gathering of Transmembrane Helix M2 with Cytoplasmic Domains and Luminal Gating

Ca2+ transport by sarcoplasmic reticulum Ca2+-ATPase involves ATP-dependent phosphorylation of a catalytic aspartic acid residue. The key process, luminal Ca2+ release occurs upon phosphoenzyme isomerization, abbreviated as E1PCa2 (reactive to ADP regenerating ATP and with two occluded Ca2+ at transport sites) → E2P (insensitive to ADP and after Ca2+ release). The isomerization involves gathering of cytoplasmic actuator and phosphorylation domains with second transmembrane helix (M2), and is epitomized by protection of a Leu119-proteinase K (prtK) cleavage site on M2. Ca2+ binding to the luminal transport sites of E2P, producing E2PCa2 before Ca2+-release exposes the prtK-site. Here we explore E2P structure to further elucidate luminal gating mechanism and effect of membrane perturbation. We find that ground state E2P becomes cleavable at Leu119 in a non-solubilizing concentration of detergent C12E8 at pH 7.4, indicating a shift towards a more E2PCa2-like state. Cleavage is accelerated by Mg2+ binding to luminal transport sites and blocked by their protonation at pH 6.0. Results indicate that possible disruption of phospholipid-protein interactions strongly favors an E2P species with looser head domain interactions at M2 and responsive to specific ligand binding at the transport sites, likely an early flexible intermediate in the development towards ground state E2P.

The cytoplasmic part of the second transmembrane helix, M2, plays a crucial role in coupling A-domain motion and tilting of the P domain during the rearrangements of transport sites 4,6-9 .
The E2P ground state, transition state (E2~P ‡ ), and product complex (E2·P i ) in the E2P hydrolysis process are mimicked by the stable structural analogs , respectively, as produced with the respective phosphate analogs for different configurational states 4 . Their crystal structures, without or with the potent inhibitor thapsigargin (TG), have been solved at atomic level 7,8,10,11 following purification of the protein using a non-ionic detergent octaethylene glycol monododecyl ether (C 12 E 8 ). Commensurate with the structural changes mentioned, the crystal structures are subtly different although the overall molecular structure of the compactly organized cytoplasmic A, P, and N domains with tightly bound BeF 3 − and occluded Mg 2+ at the catalytic site and the arrangement of transmembrane helices are similar. Namely, the E2·BeF 3 − crystal produced at pH 7.0 in 50 mM Mg 2+ has wide open transport sites (luminal gate open) with one bound Mg 2+ 11 and that at pH 5.7, where the transport sites are protonated and Mg 2+ is absent, the luminal access pathway is less open 8 (Fig. 2). The structures with bound TG at a cavity surrounded by M3, M5, and M7, namely E2·BeF 3 − (TG), and those of E2·AlF 4 − (TG) and E2·MgF 4 2− (TG), are different again, and the luminal gate is tightly closed. The closure is associated with formation of hydrophobic interaction network, the Tyr 122 -hydrophobic cluster (Y122-HC) by Leu 119 / Tyr 122 on the cytoplasmic part of M2 and five residues of the gathered A and P domains (Ile 179 /Leu 180 (A), Val 705 / Val 726 (P)) and A/M3-linker (Ile 232 on the loop connecting the A domain with M3). Significantly, in the E2·BeF 3 − crystals without TG, where the gate is open, the side chains of Leu 119 /Tyr 122 are close but pointing away from the other gathered five residues, indicative of weaker domain interactions here (Fig. 2).
Extensive mutation and kinetic studies have demonstrated [12][13][14][15] that all seven residues involved in Y122-HC including Leu 119 /Tyr 122 are crucial for opening the gate, reducing Ca 2+ affinity, and allowing rapid Ca 2+ -release (E2PCa 2 → E2P + 2Ca 2+ ), and for subsequent gate-closure and the formation of a catalytic site with hydrolytic ability. Investigation of the structural changes during these events has been aided by proteolytic digestion patterns, including a prtK site at Leu 119 4,16,17 . The site is exposed in the unphosphorylated E2 form but protected in E2·BeF 3 − , E2·AlF 4 − , and E2·MgF 4 2− as well as in the TG-bound forms of these analogs. Thus susceptibility to prtK attack or otherwise seems a good indicator of the state of the gathering of the head domains on M2. Significantly, E2PCa 2 an early E2P species, is uniquely susceptible to attack, an indication of a loose arrangement of head domains on M2 prior to progression to ground state E2P 4,18,19 .  Unexpectedly, we now find that low, non-solubilizing concentrations of C 12 E 8 render the prtK site at Leu 119 in E2P (E2·BeF 3 − ) susceptible to attack. It is as though the detergent has released constraints at the transmembrane helices to favor a state closer to that on Ca 2+ binding to the luminal sites, namely E2PCa 2 . The phenomenon uncovers a hitherto undescribed intermediate just prior to ground state E2P, stabilized by detergent that is uniquely susceptible to diverse ligand binding and cross-protein conformational changes. It shows that phospholipid-protein interactions directly participate the conformational changes associated with luminal gating events and expedite Ca 2+ release.

Results
PrtK-cleavage of Leu 119 -site in E2·BeF 3 − with C 12 E 8 at pH 7.4. In Fig. 3a, prtK-proteolysis of E2·BeF 3 − is performed at pH 7.4 in 0.1 M K + without and with a non-solubilizing low concentration of C 12 E 8 . In the absence of C 12 E 8 , E2·BeF 3 − is completely resistant to prtK both without and with A23187 as found previously 4 . In the presence of C 12 E 8 , a 95-kDa fragment (p95) is produced by specific prtK-cleavage at the Leu 119 -site without any other cleavages. Cleavage is accelerated by 30 mM Mg 2+ , but no cleavage occurs in the absence of C 12 E 8 even at 30 mM Mg 2+ . In Fig. 3d, the Mg 2+ concentration dependence of the specific prtK-cleavage rate at the Leu 119site is determined in C 12 E 8 and different monovalent cations (K + , Na + , and Li + ) at 0.1 M. The rate increases with increasing Mg 2+ concentration -binding to a low affinity site favors exposure. The cleavage is faster in Na + and K + as compared with that in Li + or in the absence of monovalent cation, thus K + or Na + binding at the K + site on the P domain 20,21 increases prtK attack at Leu 119 .
E2·BeF 3 − cleavage in C 12 E 8 is inhibited by thapsigargin (TG), which binds tightly to a cavity surrounded by M3, M5, and M7, fixing the arrangement of transmembrane helices with a tightly closed luminal gate 22 − with bound Mg 2+ is enlarged in the three top panels. In the three bottom panels, the view of transport sites from the luminal side as indicated by a large green arrow is shown. The A, P, and N domains and cytoplasmic part of M2 are yellow, cyan, pink, and purple, respectively. The Mg 2+ and water molecules at the Ca 2+ binding sites (transport sites) and Na + bound at the K + (Na + ) site on the P domain are green, red, and blue spheres, respectively. The seven residues involved in the formation of  without or with 4 μM TG ("TG"), as indicated. The "E2·TG" state of SR vesicles un-treated with the metal fluoride was subjected to the proteolysis as a control. In (d), the rate of prtK digestion of 110 kDa-ATPase chain in C 12 E 8 at pH 7.4 was determined at various concentrations of MgCl 2 in 0.1 M KCl, NaCl, or LiCl or in the absence of these salts, otherwise as in (a) and as described under "METHODS". The fragments indicated on the right of a panel are p95 produced by the prtK-cleavage at the Leu 119 -site on M2, p81/p83 produced by the prtK-cleavage at the Thr 242 -site on A/M3-linker (p83) and Ala 746 on M5 (p81) 16,39 , and the tryptic A1 fragment produced by cleavage at the Arg 198 -site on the A fragment (N-terminal half), which is formed very rapidly together with the B fragment (C-terminal half) by cleavage at Arg 505 -site 40 .
Scientific RepoRts | 7:41172 | DOI: 10.1038/srep41172 well as without Mg 2+ , the 110-kDa ATPase chain is very rapidly cleaved producing p95 and p81/p83 fragments by cleavages at Leu 119 and at Thr 242 (p83) and Ala 746 (p81), respectively, in agreement with previous findings 16  In Fig. 3a, the trypsin proteolysis was performed as described above with prtK. In the BeF 3 − -free state with bound TG as a control ("E2·TG") in which the A and P domains are not fixed, the Arg 198 -site is cleaved producing the A1 and A2 fragments (the A2 fragment is not seen because it is at the gel front) as found previously 6 . In E2·BeF 3 − , the A1 and A2 fragments are not produced regardless of the presence of C 12 E 8 and 30 mM Mg 2+ , thus the Arg 198 -site is completely resistant, consistent with association of the A and P domains by an ionic network as seen in the E2·BeF 3 − crystal structures 8,11 . E2·BeF 3 − in C 12 E 8 is completely resistant to prtK at pH 6.0. In Fig. 3e, prtK-proteolysis was performed at pH 6.0 otherwise as in Fig. 3a. At this pH the luminal transport sites are expected to be protonated. No cleavage of the 110 kDa-ATPase chain occurred even in C 12 E 8 and 30 mM Mg 2+ . The tryptic Arg 198 -site was also completely resistant at pH 6.0 as at pH 7.4 without and with C 12 E 8 and 30 mM Mg 2+ .
E2·AlF 4 − and E2·MgF 4 2− are completely resistant to prtK even in C 12 E 8 at pH 7.4 and 6.0. E2·AlF 4 − , the analog for the transition state E2~P ‡ is completely resistant to prtK at pH 7.4 and 6.0 even in the presence of C 12 E 8 both without and with 30 mM Mg 2+ (Fig. 3b,f). The Arg 198 -site is also protected from trypsin in all these conditions. E2·MgF 4 2− , the analog for the product complex (E2·P i ) is completely resistant to prtK and to trypsin in all these conditions as E2·AlF 4 − (Fig. 3c,f).
Hydrophobic nature of the nucleotide/catalytic site revealed by TNP-AMP superfluorescence. TNP-AMP binds to the ATP binding site with a very high affinity and develops an extremely high "superfluorescence" in the E2P ground state and its analog E2·BeF 3 − 4,25 . The TNP moiety binds at the adenine position in the N domain and the superfluorescence can be ascribed to a favorable TNP moiety Phe 487 interaction and site-occlusion that excludes non-specific water and increases hydrophobicity by the contribution of Arg 174 on the A domain at the A-N interface on the TNP binding pocket 26 . The superfluorescence is completely probably through TNP-Phe 487 mal-alignment and water influx here. In Fig. 4, the superfluorescence development in E2·BeF 3 − upon the TNP-AMP binding at saturating 4 μM was examined without and with C 12 E 8 at pH 7.4 and 6.0 and various concentrations of Mg 2+ in 0.1 M K + or Li + . There was almost no effect of C 12 E 8 on superfluorescence development. Specific K + binding on the P domain 20,21 also had virtually no effect (compare the data in K + with those in Li + ). Increasing Mg 2+ concentration up to 60 mM caused only slight decrease. The results show that the catalytic/ nucleotide site, starting from the E2P ground state, is not affected by C 12 E 8 , Mg 2+ , K + , and protonation of transport sites. luminal Ca 2+ at sub-mM to ~mM concentration is able to bind and cause reverse isomerization E2P + 2Ca 2+ → E2PCa 2 → E1PCa 2 , which contributes to the proper setting of luminal Ca 2+ concentration through "back-door inhibition". This reverse process as well as the forward EP isomerization is mimicked and characterized with the structural analogs E2·BeF 3 − (E2P), E2·BeF 3 − ·Ca 2 (E2PCa 2 , the transient intermediate state before the Ca 2+ -release), and E1Ca 2 ·BeF 3 − (E1PCa 2 ) 4,17-19 . In Figs 5 and 6, the effect of luminal Ca 2+ on E2·BeF 3 − was examined at pH 7.4 in C 12 E 8 or A23187, various concentrations of Mg 2+ , and 0.1 M K + or Li + . Here it should be noted that the E1Ca 2 ·BeF 3 − complex is not stable and rapidly decomposes to E1Ca 2 in the presence of a high concentration of Ca 2+ (due to Ca 2+ -substitution at the unoccluded catalytic Mg 2+ site in E1Ca 2 ·BeF 3 − 17 ), on the other hand, it is very rapidly isomerized to E2·BeF 3 − releasing Ca 2+ upon the removal or reduction of luminal free Ca 2+ concentration (to below ~100 μM) as the process mimics the isomeric transition E1PCa 2 → E2P + 2Ca 2+ 17 . Also, the E1Ca 2 ·BeF 3 − complex decomposes to E1Ca 2 upon ADP binding, mimicking the ADP-induced reverse dephosphorylation of E1PCa 2 , and upon TNP-AMP binding probably analogous to the ADP-induced process, in contrast to a stable E2·BeF 3 − with bound ADP or TNP-AMP 17 . In Fig. 5, taking these known properties into account, we first determined the overall time course of the Ca 2+ -induced E2·BeF 3 − reverse conversion and decomposition to E1Ca by adding an excess EGTA after various times of incubation with 0.5 mM Ca 2+ thereby converting the remaining E1Ca 2 ·BeF 3 − to the stable E2·BeF 3 − species, and in addition adding TNP-AMP to determine superfluorescence development to estimate the total amount of E2·BeF 3 − and E1Ca 2 ·BeF 3 − species remaining at the time of EGTA addition. In Fig. 6, prtK proteolysis was performed for a short period during the 0.5 mM Ca 2+ incubation and without the EGTA addition to identify the structural states of EP species under representative conditions in Fig. 5 (although the Ca 2+ -induced process proceeds).
First in Fig. 5 where TNP-AMP superfluorescence is examined, we found both with K + and without K + (with LiCl) that the Ca 2+ -induced reverse conversion/decomposition of E2·BeF 3 − is considerably slower in C 12 E 8 than in A23187, and increasing Mg 2+ to ~20 mM in C 12 E 8 causes a marked retardation or almost complete inhibition. The retardation by Mg 2+ in C 12 E 8 is much stronger and occurs at much lower Mg 2+ concentration than in A23187. In the absence of both C 12 E 8 and A23187, i.e. with an impermeable SR membrane, no conversion nor decomposition of E2·BeF 3 − occurs with Ca 2+ , therefore the Ca 2+ -induced decomposition is due to the Ca 2+ access from the luminal side as found previously 4,17 . Regarding the K + effect, the luminal Ca 2+ -induced conversion/decomposition of E2·BeF 3 − is considerably faster in K + than in its absence, therefore specific K + binding 20,21 accelerates the process.
Then in Fig. 6a, prtK-proteolysis was performed to identify the structural state stabilized in C 12 E 8 with, most typically, 30 mM Mg 2+ in the absence of K + during luminal Ca 2+ -induced E2·BeF 3 − reverse conversion and decomposition. Here, the sample was incubated first with 0.5 mM Ca 2+ for 10 s, and then with a high concentration of prtK for various times without removal of Ca 2+ . The proteolytic pattern was compared with those of BeF 3 − -free E1Ca 2 and of E1Ca 2 ·BeF 3 − that is formed and stabilized perfectly under the previously identified most appropriate conditions, i.e. at pH 7.0 with 0.7 mM Ca 2+ and 15 mM Mg 2+ in 0.1 M K + in the absence or presence of A23187 17 ; in these states, p81/p83 fragments are produced due to cleavage at Thr 242 (p83) and Ala 746 (p81) without production of the p95-fragment (Fig. 6b). In C 12 E 8 and Ca 2+ (Fig. 6a), E2·BeF 3 − both without and with 30 mM Mg 2+ is degraded slowly as compared with E1Ca 2 , producing the stable p95 fragment as seen with E2·BeF 3 − in C 12 E 8 without Ca 2+ (cf. Fig. 3) and a small amount of p81/p83 fragments, which degrade rapidly as the BeF 3 − -free E1Ca 2 state. Note also that the 110-kDa ATPase chain degradation is much slower and formation of the rapidly degrading p81/p83 fragments is much less in 30 mM Mg 2+ than without Mg 2+ . The results show that E2·BeF 3 − in C 12 E 8 and Ca 2+ is resistant to the luminal Ca 2+ -induced reverse conversion to E1Ca 2 ·BeF 3 − , which can be interpreted as very slow Ca 2+ binding to luminal transport sites and what slow conversion occurs is markedly retarded by 30 mM Mg 2+ . These results accord with those using superfluorescence as the indicator in Fig. 5.
In the presence of A23187, as seen in Fig. 6a, formation of the p81/p83 fragments from E2·BeF 3 − in Ca 2+ occurs without any p95 fragment, as with E1Ca 2 and E1Ca 2 ·BeF 3 − in A23187 (cf. Fig. 6b) indicating a fast conversion of E2·BeF 3 − to E1Ca 2 ·BeF 3 − without the detergent and with the ionophore. These results together with the retardation by Mg 2+ of loss of TNP-AMP superfluorescence (Fig. 5) indicate that E1Ca 2 ·BeF 3 − is formed from E2·BeF 3 − without detergent on luminal Ca 2+ binding and further decomposed to E1Ca 2 , and that Mg 2+ at a high    Fig. 6b, it can be seen that under conditions where E1Ca 2 ·BeF 3 − is perfectly stable in A23187 17 , the addition of C 12 E 8 in place of A23187 produces the same proteolytic pattern as developed with E2·BeF 3 − in C 12 E 8 and Ca 2+ . The results reveal that the E2·BeF 3 − state is produced and stabilized in C 12 E 8 even under conditions that perfectly stabilize E1Ca 2 ·BeF 3 − in the absence of C 12 E 8 . This was further verified by superfluorescence development and loss upon TNP-AMP addition in Fig. 6c, which was performed on the basis of previous findings 17 that E1Ca 2 ·BeF 3 − rapidly decomposes to the non-fluorescent E1Ca 2 state upon TNP-AMP binding whereas E2·BeF 3 − with bound TNP-AMP is stable, and also that the superfluorescence intensity is greater in E2·BeF 3 − than in E1Ca 2 ·BeF 3 − (by approximately 25%). In Fig. 6c, E1Ca 2 ·BeF 3 − was first formed under the conditions in Fig. 6b without A23187 and C 12 E 8 , and then A23187 or C 12 E 8 added. After 10 s, superfluorescence upon TNP-AMP addition was recorded. In A23187 or in its absence, superfluorescence development is followed by its rapid loss, which is due to E1Ca 2 ·BeF 3 − decomposition to E1Ca 2 on TNP-AMP binding 17 . In C 12 E 8 , greater superfluorescence develops and its loss is considerably slower than in A23187. The results show again that in C 12 E 8 , E2·BeF 3 − is formed even under conditions that perfectly stabilize E1Ca 2 ·BeF 3 − (although E2·BeF 3 − is slowly decomposed to the non-fluorescent E1Ca 2 state via E1Ca 2 ·BeF 3 − in high Ca 2+ and decomposition by TNP-AMP). Fig. 7, the effects of C 12 E 8 , K + , and Mg 2+ on the forward E2P hydrolysis rate were examined at pH 7.4 and 6.0. Here E2P was first formed in the reverse reaction of hydrolysis from the Ca 2+ -deprived E2 state and 32 P i in 7 mM Mg 2+ without or with C 12 E 8 (or with A23187) in 20% (v/v) Me 2 SO, conditions that favor E2P formation. Then hydrolysis was initiated by a 20-fold dilution in non-radioactive P i , various concentrations of Mg 2+ , and 0.1 M K + (Fig. 7a) or Li + (Fig. 7b) at the desired pH. In K + at pH 7.4, C 12 E 8 markedly retards hydrolysis as found previously at pH 7.5 27 , and increasing Mg 2+ concentration in C 12 E 8 hardly affects the rate (perhaps a slight increase), but the cation decreases the rate in the absence of C 12 E 8 . Because this decrease is observed both without and with A23187 (an ionophore for Ca 2+ and Mg 2+ ) and because Me 2 SO (used for the P i -induced E2P formation) does not permeabilize the SR membrane, the hydrolysis reaction rate itself is likely affected by Mg 2+ at the cytoplasmic side. At pH 6.0 in K + , hydrolysis is much slower than at pH 7.4, as is well known 28 , and C 12 E 8 and Mg 2+ have almost no effect on the slowed rate.

E2P hydrolysis. In
In the absence of K + (Fig. 7b), E2P hydrolysis at both pH 7.4 and 6.0 is much slower than in 0.1 M K + (by ~10-fold at the respective pH), in agreement with the well-known acceleration of hydrolysis by specific K + binding on the P domain 20,21 . In the absence of K + , hydrolysis in C 12 E 8 is only slightly slower than that without C 12 E 8 . Mg 2+ at ~10 mM somewhat increases the rate although the rate is still much slower than that in the presence of K + . In summary, induction of the detergent-stabilized state strongly inhibits hydrolysis at pH 7.4, but not following protonation of the transport sites at pH 6.0, and only in the presence of K + .

Discussion
Ca 2+ transport by Ca 2+ -ATPase includes phosphorylated intermediates where Ca 2+ is occluded at the transport sites and then released to the lumen, i.e. E1P[Ca 2 ] → E2P + Ca 2+ . During this process the A domain swings around and engages with the P domain and neck region of the protein at the cytoplasmic part of M2 (Fig. 1). The A-domain rotation inclines the P-domain by pulling an A/M1′ -link, pushing M4 down towards the lumen to release the Ca 2+ 8,18,19 . There is evidence that the gathering and interaction of A and P domains at the cytoplasmic part of M2 occurs progressively. Namely, changes, which are linked to deocclusion and opening of the luminal access channel with an affinity reduction, are followed by constrictions to limit access, protonation, and finally closure, and all these changes are synchronized with catalytic site preparations for hydrolysis 4,13,15,[17][18][19] . Part of the development is seen with the Leu 119 prtK cleavage site, being exposed in E2PCa 2 , hidden in E2P, E2~P ‡ and E2·P i , and exposed again in E2 4,18,19 . We found here that non-solubilizing concentrations of C 12 E 8 uncovers the Leu 119 prtK site of E2P, as depicted in its analog E2·BeF 3 − . This indicates that membrane perturbation drives the -free Ca 2+ -ATPase ("E1Ca 2 ") in A23187 and in C 12 E 8 was subjected to proteolysis otherwise as above. Note that the slow decomposition of E2·BeF 3 − in Ca 2+ in the absence of A23187 and C 12 E 8 (a) is probably due to slow Ca 2+ permeation into the SR vesicles lumen 17 . (c) E1Ca 2 ·BeF 3 − was produced by incubating SR vesicles for 30 min with 100 μM BeCl 2 and 2 mM KF in the absence of A23187 and C 12 E 8 otherwise as in (b), then C 12 E 8 or A23187 was added to give 0.02 mg/ml and 2.5 μM, respectively. At 10 s after this addition, TNP-AMP was added to give a saturating 4 μM, and the fluorescence monitored; trace b, without C 12 E 8 and A23187; traces c and d, in A23178 and in C 12 E 8 , respectively. Trace e, the fluorescence monitored with E2·BeF 3 − in the presence of 2 mM EGTA without adding Ca 2+ . Trace a, the nonsuperfluorescent E1Ca 2 level (BeF 3 − -free Ca 2+ -ATPase) in 4 μM TNP-AMP.
Scientific RepoRts | 7:41172 | DOI: 10.1038/srep41172 intermediate towards one more like that with bound Ca 2+ , and points to an earlier catalytic intermediate with a looser arrangement in the head region, as expected for early engagement of the rotated A domain. The responsiveness of E2P to membrane perturbation and the detergent-induced state to ligand binding (Ca 2+ , Mg 2+ , K + , H + , and TG) through changes in exposure of the Leu 119 prtK site at the cytoplasmic part of M2 points to flexible and rather unstable forms. These properties are due most probably not only to its unoccupied transport sites and associated circle of negative charges, but also to a loose meeting of domains and neck region with largely unsecured interactions at the cytoplasmic part of M2. The downward thrust of M4 (by a full turn of an α -helix 8 ), together with M3, is probably partly stabilized by surrounding phospholipids and insertion of non-ionic detergent between them could be disruptive. In the head region the interactions at Leu 119 involve the formation of Y122-HC, a hydrophobic interaction network of Tyr 122 /Leu 119 with the A and P domains and A/M3-linker involving seven residues (Fig. 2). As mentioned above, the interactions are likely progressive, loose at first as the A domain engages followed by incremental tightening in E2P to the fully stabilized state in E2~P ‡ and E2·P i . Indeed, in the E2·BeF 3 − crystal structures (formed in the presence of C 12 E 8 ) with the bound Mg 2+ or with protonation without the Mg 2+ , Leu 119 /Tyr 122 on M2 are close but not yet associated with the five other gathered residues involved in Y122-HC formation. The knitting of Leu 119 and Tyr 122 with the other residues is seen in the crystal structures of analogs of the next intermediates, E2~P ‡ and E2·P i . Accumulating interactions fit perfectly with the staggered changes at the luminal transport sites, from closed to open to closed again.
Stabilization of the early detergent-induced state is seen in the forward direction of catalysis coming from E1PCa 2 (E1Ca 2 ·BeF 3 − ) and in the backward direction with Ca 2+ binding to the luminal sites of E2P (E2·BeF 3 − ), using both TNP-AMP superfluorescence and the prtK sites as probes. Our results suggest that the E2·BeF 3 − structural state favored in C 12 E 8 and stabilized by Mg 2+ represents one between E1PCa 2 and Ca 2+ -released E2P, i.e. the transient E2P state immediately following Ca 2+ release denoted as E2P * with luminally open and vacant low affinity transport sites (E2P * Ca 2 → E2P * in Fig. 8  manifests itself in the competitive inhibition by Mg 2+ of luminal Ca 2+ -induced reverse isomerization E2P+ 2Ca 2+ → E1PCa 2 29 . Notably also, the dephosphorylated E1 state is able to accommodate one Mg 2+ at the transport sites and forms E1·Mg, which favors high affinity Ca 2+ -binding resulting in a rapid E2 → E1·Mg → E1Ca 2 transition 30,31 (Fig. 1). Thus it seems that Mg 2+ binds at the empty transport sites both in the unphosphorylated and phosphorylated states and modifies transport function.
The indicates that Mg 2+ accesses E2P * with a much higher affinity than E2P. Thus the transport sites appear more open and accessible to Mg 2+ on the luminal side in the Leu 119 -site cleavable E2P * state than in the prtK-resistant E2P ground state. In fact, in the E2·BeF 3 − crystal with bound Mg 2+ at the transport sites, the sites are actually more open to the lumen than in the structure without Mg 2+ (Fig. 2). Note also that in E2·AlF 4 − and E2·MgF 4 2− (E2~P ‡ and E2·P i ) and in E2·BeF 3 − with bound TG, Ca 2+ cannot bind as the luminal gate is tightly closed 4,7,8 , and the Leu 119 -site is completely resistant to prtK regardless of the presence of C 12 E 8 (Fig. 3). These findings suggest that the structural change reflected by prtK resistance at Leu 119 is associated with luminal gating, supporting the above conclusion that substantial luminal gate closure occurs in E2P * → E2P, which probably involves gathering of Leu 119 /Tyr 122 with the engaged A and P domains to accomplish the Y122-HC network. Then the passage is completely sealed in E2~P ‡ and E2·P i (E2·AlF 4 − and E2·MgF 4 2− ) 4 . Previous kinetic analysis of the luminal Ca 2+ -induced reverse isomerization E2P + 2Ca 2+ → E1PCa 2 indicated 14 that the luminal Ca 2+ access to the transport sites in E2P is rate-limiting. This is described in Fig. 8 with the equilibrium E2P * ↔ E2P, where the former state is more open and the latter relatively closed. This view agrees with our finding on the Ca 2+ release kinetics E1PCa 2 → E2PCa 2 → E2P+ 2Ca 2+ 15 that the E2P structure proceeds from a luminally open state for Ca 2+ release (corresponding to E2P * in Fig. 8) to a closed state (E2P) with the structural contribution of Leu 119 /Tyr 122 . The observation that Mg 2+ hardly alters the forward E2P hydrolysis rate in C 12 E 8 (Fig. 7a) can be accounted for by a rapid Mg 2+ binding/release relative to the hydrolysis reaction process, and implies that Mg 2+ binding favors the forward reaction.
At pH 6.0 in which the transport sites are protonated, the Leu 119 -site is completely resistant to prtK regardless of the presence of C 12 E 8 , and the E2P hydrolysis rate is not affected by C 12 E 8 . In Fig. 8, the protonated structural state with the prtK-resistance revealed in C 12 E 8 is denoted as E2P( * ) to be discriminated from the prtK-cleavable E2P * state without protonation. Protonation neutralizes charges at the Ca 2+ -binding sites and stabilizes the arrangement of transmembrane helices via a hydrogen bonding network 8 , which lowers Ca 2+ -accessibility (without completely closing the gate as seen in the E2·BeF 3 − crystal formed at pH 5.7 8 ). The protonated state proceeds promptly to subsequent hydrolysis with tight gate closure E2P+ H 2 O → E2~P ‡ (E2·BeF 3 − → E2·AlF 4 − ), as indicated previously by kinetic analysis of E2P hydrolysis 5 .
K + in the presence of C 12 E 8 accelerates both forward E2P hydrolysis and luminal Ca 2+ -induced reverse conversion of E2·BeF 3 − (Figs 5 and 7). These findings are in complete agreement with the known role of specific K + binding on the P domain in accelerating both forward hydrolysis 20,21 and luminal Ca 2+ -induced reverse conversion of E2P 14 . K + binding likely destabilizes both E2P and E2P * in Fig. 8, thus promoting rapid transport. Ca 2+ yet at a high affinity) were previously identified by the elongation of the A/M1′ -linker 18,19 and by substitutional mutation of Leu 119 and Tyr 122 15 , but they are transient intermediates and have never been trapped or identified in wild type 15,18,19 ; therefore they are shown in brackets. The E2P structural states found in this study at pH 7.4, the Leu 119 -cleavable state and the prtK-resistant state are denoted as E2P * and E2P, respectively. The prtK-resistant state found in C 12 E 8 at pH 6.0 (i.e. with protonation of the transport sites) is denoted as E2P( * ). Note that Y122-HC formation on gathering of Leu 119 /Tyr 122 on M2 with engaged A and P domains occurs progressively during E2P processing and couples with luminal gating (see more in "Discussion").
Scientific RepoRts | 7:41172 | DOI: 10.1038/srep41172 Finally, induction of the detergent-stabilized state, an early intermediate to ground state E2P, shows how phospholipids are intimately involved in the latter's stabilization. Membrane perturbation effects during the transport cycle may be under-appreciated as fundamental to the mechanism.

Methods
Preparation of SR vesicles and treatment with BeF x , AlF x , and MgF x . SR vesicles were prepared from rabbit skeletal muscle as described 32,33 , in which all the methods were carried out in accordance with institutional laws and regulations of the Asahikawa Medical University and the experimental protocols were approved by the Animal Experimentation Ethics Committee of the Asahikawa Medical University (license number 16006). The content of the phosphorylation site in the vesicles and the Ca 2+ -dependent ATPase activity were determined as described 32,33 . E2·BeF 3 − , E2·AlF 4 − , and E2·MgF 4 2− were produced by incubating the SR vesicles with the respective metal fluoride and by washing the unbound ligands as described previously 4 . Formation and hydrolysis of E2P. The SR vesicles were phosphorylated with 0.1 mM 32 P i at 25 °C for 10 min in 20% (v/v) Me 2 SO in the absence of Ca 2+ , after which the samples were cooled and diluted 20-fold by a solution containing 2.1 mM non-radioactive P i to initiate the hydrolysis of 32 P i -labeled E2P, otherwise as described in detail in the legend to Fig. 7. The reaction was quenched with ice-cold trichloroacetic acid containing P i . The precipitated proteins were separated by 5% SDS-PAGE at pH 6.0 according to Weber and Osborn 34 . The radioactivity associated with the separated Ca 2+ -ATPase was quantified by digital autoradiography as described 35 . Rapid kinetics measurement of hydrolysis was performed with a handmade rapid mixing apparatus and the rate of hydrolysis was determined with the least-squares fit to a single exponential, as described 35 . Proteolytic analysis. SR vesicles (0.45 mg/ml protein) were subjected to proteolysis at 25 °C by addition of trypsin (at 0.3 mg/ml, L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated) or proteinase K (prtK, at 0.1 mg/ml, Sigma) as described previously 6,16 , otherwise as indicated in the figure legends. The proteolysis was terminated by trichloroacetic acid, and the samples were subjected to Laemmli SDS-polyacrylamide gel electrophoresis 36 , and densitometric analyses of the gels stained with Coomassie Brilliant Blue R-250, as described 6,16 . The degradation rate of 110-kDa ATPase chain with prtK was determined by least-squares fit of a single exponential to the time course (0-150 min) as described previously 16 .
Fluorescence measurements. The TNP-AMP fluorescence of the Ca 2+ -ATPase (0.06 mg/ml protein, TNP-AMP from Molecular Probes ® Life Technologies) was measured on a RF-5300PC spectrofluorophotometer (Shimadzu, Kyoto, Japan) with excitation and emission wavelengths 408 and 540 nm (with band widths 5 and 10 nm), as described previously 4 . Miscellaneous. Protein concentrations were determined by the method of Lowry et al. 37 with bovine serum albumin as a standard. Three-dimensional models of the enzyme were reproduced by the program VMD 38 . The values presented are the mean ± s.d. (n = 3-4).