Conformational Fluctuations of the Ca2+-ATPase in the Native Membrane Environment

Digestion with proteinase K or trypsin yields complementary information on conformational transitions of the Ca2+-ATPase (SERCA) in the native membrane environment. Distinct digestion patterns are obtained with proteinase K, revealing interconversion of E1 and E2 or E1∼P and E2-P states. The pH dependence of digestion patterns shows that, in the presence of Mg2+, conversion of E2 to E1 pattern occurs (even when Ca2+ is absent) as H+ dissociates from acidic residues. Mutational analysis demonstrates that the Glu309 and Glu771 acidic residues (empty Ca2+-binding sites I and II) are required for stabilization of E2. Glu309 ionization is most important to yield E1. However, a further transition produced by Ca2+ binding to E1 (i.e. E1·2Ca2+) is still needed for catalytic activation. Following ATP utilization, H+/Ca2+ exchange is involved in the transition from the E1∼P·2Ca2+ to the E2-P pattern, whereby alkaline pH will limit this conformational transition. Complementary experiments on digestion with trypsin exhibit high temperature dependence, indicating that, in the E1 and E2 ground states, the ATPase conformation undergoes strong fluctuations related to internal protein dynamics. The fluctuations are tightly constrained by ATP binding and phosphoenzyme formation, and this constraint must be overcome by thermal activation and substrate-free energy to allow enzyme turnover. In fact, a substantial portion of ATP free energy is utilized for conformational work related to the E1∼P·2Ca2+ to E2-P transition, thereby disrupting high affinity binding and allowing luminal diffusion of Ca2+. The E2 state and luminal path closure follow removal of conformational constraint by phosphate.

The Ca 2ϩ -ATPase of sarcoendoplasmic reticulum membranes (SERCA) 2 includes multiple isoforms and splice variants with variable tissue distribution. In this study, we used the SERCA1a isoform of skeletal muscle, a well characterized enzyme (1,2) that utilizes the free energy of ATP for Ca 2ϩ transport against a concentration gradient. The functional unit is a protein monomer consisting of 994 amino acid residues. The sequence is folded into a cluster of 10 segments forming a transmembrane region, and three relatively large domains ("N", "P," and "A") protruding from the cytosolic surface of the membrane (3,4). The ATPase cycle begins with high affinity binding of two Ca 2ϩ derived from the cytosolic medium ("outside"), followed by ATP utilization to form a phosphorylated enzyme intermediate. Isomerization of the phosphoenzyme intermediate is then coupled to active transport of the bound Ca 2ϩ across the membrane ("inside"). Hydrolytic cleavage of the phosphoenzyme is the final step that allows enzyme turnover.
The cooperative character of Ca 2ϩ binding as well as the relatively large distance between the catalytic site in the headpiece and the Ca 2ϩ -binding sites of the ATPase within the transmembrane region imply that conformational rearrangements of the ATPase protein are involved in the mechanism of catalytic activation and energy transduction. Within the general context of cation transport, these rearrangements were envisioned as interconversions of the E1 and E2 conformations in the ground state of the enzyme and the E1-P to E2-P conformations of the phosphorylated intermediate. In fact, conformational changes were initially detected by spectroscopic experimentation (5). High resolution structures were then obtained by crystallographic studies and attributed to different catalytic intermediates (6). On the other hand, the occurrence of conformational transitions in the native membrane environment is revealed by changes in the patterns of proteolysis (7). We report here a series of experiments on limited proteolysis with proteinase K or trypsin, yielding complementary information on the conformational effects of pH, temperature, catalytic ligands, and the specific inhibitor thapsigargin (TG). The experimentation was extended to measurements of ATP hydrolysis and Ca 2ϩ transport, thereby providing an understanding of how functional events and conformational transitions are related in the native membrane environment, and how free energy is utilized for conformational work linked to active transport.
infected with adenovirus vectors carrying rabbit WT (3) or mutant cDNA (9). Ca 2ϩ uptake by SR vesicles was measured at 25 or 35°C in a reaction mixture containing 50 mM MOPS, pH 7, 80 mM KCl, 3 mM MgCl 2 , 50 g of microsomal protein/ml, and 50 M CaCl 2 , including 45 Ca-labeled isotopic tracer. ATP (3 mM) was added to start the reaction, and at various times, the protein contained in the 1-ml reaction mixture was loaded onto a 0.45-m Millipore filter by vacuum suction and washed with 15 ml of 2 mM LaCl 3 and 10 mM MOPS, pH 7.0. The filter was then processed for determination of radioactivity by scintillation counting. A zero time sample was taken before addition of ATP.
ATPase activity was measured at 25°C, in a reaction mixture containing 30 g of SR protein/ml, 50 mM MOPS, pH 7 (or 50 mM HEPES, pH 8), 50 mM KCl, 3 mM MgCl 2 , 1 g of A23187 ionophore, and 2 mM EGTA, in the presence or in the absence of 2 mM CaCl 2 . The reaction was started by addition of 2 mM ATP, and samples were taken at serial times for P i determination by colorimetry (10).
Enzyme phosphorylation with P i was estimated following equilibration (5 min at 25°C) of SR vesicles (0.12 mg/0.2 ml) with 2.0 mM 32 P i , in a medium containing 100 mM MES, pH 6.0 (or 100 mM HEPES, pH 8), and 3 mM EGTA, in the presence or the absence of 10 mM MgCl 2 and/or 0.1 mM free Ca 2ϩ . The reaction was quenched with 0.2 ml of 2.0 M perchloric acid. The quenched reaction mixture (0.3 ml) was filtered through 0.45-m Millipore filters, and the protein collected on the filters was washed three times with 5 ml of cold 0.125 M perchloric acid, once with 5 ml of cold water, and finally dissolved with dimethylformamide and processed for scintillation counting. Control experiments were performed by quenching samples with perchloric acid before addition of radioactive substrate.
Limited proteolytic digestion was performed in reaction mixtures containing 50 mM MOPS, pH 7.0 (or MES, pH 6.0, or HEPES, pH 8), 50 mM NaCl, 0.05 mg of SR microsomal protein/ml (or 0.4 mg of COS-1 cell microsomes containing recombinant SERCA/ml) and 0.05 mg of proteinase K or trypsin. CaCl 2 , MgCl 2 , EGTA, and AMPPCP were added as indicated in the figures. In some experiments, 2 mM KF and 0.1 mM AlCl 3 were added, and the reaction mixture was incubated for 30 min to obtain the fluoroaluminate complex, in the absence or in the presence of nucleotide. Following incubation at 25°C for various time intervals, the reaction was quenched with trichloroacetic acid (2.5%), and the protein was solubilized with a medium containing lithium dodecyl sulfate (1%), MOPS (0.312 M), pH 6.8, sucrose (3.75%), ␤-mercaptoethanol (1.25 mM), and bromphenol blue (0.025%). The samples were then subjected to electrophoretic analysis on 12% gels, and the protein bands were stained with Coomassie Blue R-250. Alternatively, Western blots were obtained using the monoclonal antibody MA3911 or MA3912 (Affinity Bioreagents), followed by goat anti-mouse IgG horseradish peroxidase-conjugated secondary antibodies and visualization with an enhanced chemiluminescence-linked detection system (Pierce). The MA3911 antibody reacts preferentially with the amino-terminal region of the ATPase, whereas the MA3912 reacts preferentially with the carboxyl-terminal region.

RESULTS
Electrophoresis of ATPase protein subjected to limited digestion with proteinase K (11) in the presence of Ca 2ϩ shows an 83-kDa band corresponding to the proteolytic fragment intervening between Glu 243 and the carboxyl terminus, and a 28-kDa band corresponding to the complementary fragment between Thr 242 and the amino terminus ( Fig. 1). Mapping of the fragments with respect to their proximity to the amino or carboxyl termini was confirmed by use of complementary monoclonal antibodies. A 29 -30-kDa band, due to cleavage at Ser 350 -Thr 357 and Ile 611 , is observed as well (12). A faint 54-kDa band corresponding to the segment between Glu 243 and Val 734 /Val 747 is also visible. The pattern of digestion appears to be the same at pH 6 -8, although proteolysis occurs more rapidly at alkaline pH. This pattern, obtained in the presence of Ca 2ϩ , is attributed to the E1 state (including various sequential conformations yielding high Ca 2ϩ affinity, cooperative binding of 2 Ca 2ϩ , and catalytic activation; see "Discussion").
If the digestion is performed at pH 6 in the absence of Ca 2ϩ , additional 95 and 14 kDa bands are noted (Fig. 1), corresponding to complementary fragments between Lys 120 and the carboxyl terminus and Leu 119 and the amino terminus, respectively. This pattern, obtained in the absence of Ca 2ϩ , is attributed to the E2 state (i.e. the ground state of enzyme in the absence of Ca 2ϩ , but it is also observed with the phosphorylated intermediate formed by reaction with P i and the dead end complex of E2 with TG (see below)).
We find that when the pH is raised (in the absence of Ca 2ϩ ) the E2 digestion pattern changes to that observed in the presence of Ca 2ϩ (i.e. E1 pattern), as the 95 kDa is much reduced at pH 7 and disappears at pH 8. This change in digestion pattern does not occur in the absence of Mg 2ϩ (Fig. 2, WT) indicating that, in the absence of Ca 2ϩ , alkaline pH and Mg 2ϩ have a com- The 1st lane on the left shows a sample with proteinase K denatured before addition of ATPase. The E1 patterns yields complementary 83-kDa (carboxyl terminus) and 28-kDa bands (amino terminus), whereas the E2 pattern yields additional 95-kDa (carboxyl terminus) and 14-kDa bands (amino terminus). Note that the E1 pattern is obtained even in the absence of Ca 2ϩ when the pH is raised. Digestion (30 min at 25°C) and electrophoresis are as explained under "Materials and Methods." The free Ca 2ϩ concentration, when present, was 50 M. Protein was stained with Coomassie Blue.
plementary effect in inducing a transition to the E1 digestion pattern.
We considered the possibility that control of the ATPase conformation yielding the E1 or E2 digestion patterns resides within the Ca 2ϩ -binding sites, even in the absence of Ca 2ϩ . We therefore performed proteolysis experiments using recombinant protein containing a single mutation of either the Glu 771 or Glu 309 in the Ca 2ϩ -binding site I and II, respectively (13). We found that, in the absence of Ca 2ϩ , digestion of E309Q yields the E1 pattern at any pH, regardless of whether Mg 2ϩ is present or not (Fig. 2). This indicates that in the WT protein ionization of the Glu 309 carboxyl group has an important role for transition to the E1 conformation. On the other hand, it is also shown in Fig. 2 that the E771Q mutant acquires in great part the E1 digestion pattern as the pH is raised from 6.0to 7.0, either in the presence or in the absence of Mg 2ϩ . This suggests that in the WT protein, the Glu 771 carboxyl group is involved in conformational stabilization of the E2 state. Lack of such stabilization then allows disruption of the E2 conformation when ionization of the Glu 309 (and/or other residues) carboxyl chain is produced by raising the pH. Overall, these experiments indicate that H ϩ binding to acidic residues at the Ca 2ϩ sites favors the E2 digestion pattern, whereas H ϩ dissociation favors the E1 digestion pattern even in the absence of Ca 2ϩ . The effect of Mg 2ϩ observed with WT protein is likely due to a requirement for additional perturbation when both Glu 309 and Glu 771 contribute to stabilization of E2 (see "Discussion").
Because the physiological role of shifting E2 to E1 is usually observed in the presence of Ca 2ϩ , resulting in catalytic activation, we checked whether acquisition of the E1 digestion pattern at pH 8 in the absence of Ca 2ϩ , but in the presence of Mg 2ϩ , would be accompanied by ATPase activation. We found no ATP utilization under these conditions (not shown), indicating that Ca 2ϩ (M) binding to E1 is still required to obtain catalytically active E1⅐2Ca 2ϩ . At any rate, this group of experiments suggests that in the WT protein and in the absence of Ca 2ϩ , E2 is stabilized by a cooperative mechanism, including the Glu 309 and Glu 771 side chains. H ϩ dissociation from their acidic chains favors a conformation yielding the E1 digestion pattern.
Effects of AMPPCP and ATP-It was previously reported that nucleotide binding protects the ATPase from digestion by proteinase K (7,14). The inactive analog AMPPCP must be used instead of ATP when Ca 2ϩ is present, to avoid rapid substrate consumption by the activated enzyme. We show here that in the presence of Ca 2ϩ the protective effect of AMPPCP is the same at pH 6 -8, whereas digestion retains the E1 pattern (Fig.  3). Half-maximal protection is obtained with a 0.1 mM concentration of AMPPCP, as determined by gel densitometry.
In the absence of Ca 2ϩ , ATP has a low protective effect at pH 6, whereas the digestion pattern remains that of E2. However, as the pH is raised to pH 8, as expected ( Fig. 1), the digestion pattern is transformed to that of E1. In this case, a more prominent protection by ATP (as well as AMPPCP) is observed (Fig.  3). This indicates that in the E1 state obtained in the presence of Mg 2ϩ and alkaline pH, the level of ATP protection is similar to that observed in the presence of Ca 2ϩ . Nevertheless, as stated above, addition of Ca 2ϩ is still required to obtain catalytic activation.
Phosphoenzyme Intermediate and the E1-P to E-2P Transition-Following ATP binding in the presence of Ca 2ϩ , formation of phosphorylated enzyme intermediate is the next  Fig. 1). Note that the WT protein acquires the E1 pattern at alkaline pH only in the presence of Mg 2ϩ . The E309Q mutant shows the E2 pattern at any pH, in the presence as well as in the absence of Mg 2ϩ . The E771Q mutant acquires the E1 pattern as the pH is raised in the presence as well as in the absence of Mg 2ϩ . Digestion was for 30 min at 25°C. The mutant protein is indicated by Western blotting. step in the ATP catalytic cycle. It was previously shown that formation of a stable fluoroaluminate transition state analog of E1ϳP⅐2Ca 2ϩ protects the ATPase from digestion with proteinase K (7,15). When we studied the proteinase K digestion pattern of the phosphoenzyme analog in the presence and in the absence of Ca 2ϩ (E-AlF 4 ⅐2Ca 2ϩ versus E-AlF 4 ), we found that E-AlF 4 ⅐2Ca 2ϩ was digested with the E1 pattern at acid as well as alkaline pH (Fig. 4). On the other hand, the E-AlF 4 (as well as the MgF 4 ) analog of E2-P exhibited the E2 digestion pattern at pH 6, while exhibiting the E1 digestion pattern at pH 8 (even though Ca 2ϩ was absent). This experiment indicates that the E1-P to E2-P transition requires H ϩ binding, and E2-P is stabilized by H ϩ binding in analogy to E2. For this reason, digestion with proteinase K exhibits the E1-P pattern at alkaline pH, even in the absence of Ca 2ϩ (Fig. 4). Therefore, H ϩ /Ca 2ϩ exchange is a determining factor in the E1ϳP⅐2Ca 2ϩ to E2-P transition, and Ca 2ϩ dissociation depends on this transition.
We also studied the effect of pH on the ability of the enzyme to acquire protection from proteinase K digestion following exposure to P i in the absence of Ca 2ϩ . In this case, the E2-P ground state of the catalytic cycle is obtained by reverse phosphorylation. It is shown in Fig. 5A that formation of E2-P by equilibration of the ATPase with P i (i.e. reversal of the ATPase cycle) produces strong protection, and the remaining digestion exhibits the E2 pattern. Protection by P i appears similar to that produced by TG (Fig. 5A). However, whereas experiments with TG yield the E2 pattern and a low rate of digestion under all conditions (including alkaline pH, absence of Mg 2ϩ , and presence of Ca 2ϩ ), experiments with P i yield protection and E2 pattern of digestion only under conditions permitting its covalent reaction with the enzyme (i.e. acid pH, presence of Mg 2ϩ , and absence of Ca 2ϩ ). It is shown in Fig. 5B that the same conditions required for ATPase protection from proteinase K (i.e. acid pH in the absence of Ca 2ϩ ) are also required for formation of phosphorylated intermediate by utilization of P i (16). The P i concentration yielding half-maximal activity was found to be ϳ0.5 mM, both for protection and phosphorylation. It is then apparent that detection of the E1 or E2 digestion patterns is a convenient expedient to monitor both the E1 to E2 and E1-P to E2-P through the ATPase cycle.

Temperature Dependence of Proteolytic Digestion and Effects of Catalytic
Substrates-It is noteworthy that protection by catalytic substrate (Fig. 3) does not involve a primary effect on the digestion pattern but rather a delay of the time course of digestion occurring with the same pattern (E1 or E2, respectively). This suggests that protection may be due to constraint of conformational fluctuations related to internal protein dynamics (17) and required for suitable interaction of the ATPase with the proteolytic enzyme. This is best demonstrated by comparing the ATPase digestion with trypsin at various temperatures.
Trypsin produces an initial cut on the surface of the ATPase N domain, acting on a site (Arg 505 ) freely accessible from the aqueous medium. Consequently, a 57-and 46-kDa complementary fragment are produced immediately, with no apparent temperature dependence under our conditions. On the other hand, further digestion is much slower, as the pertinent site (Arg 198 ) is enclosed within the ATPase structure. In fact, it is shown in Fig. 6 that further digestion, yielding the 35-and 23-kDa fragments from the 57-kDa fragment and even smaller fragments derived from the 46-kDa fragment, is temperature-   Fig. 1). The leftmost lanes show samples (controls) with proteinase K denatured before addition of ATPase. B, high levels of ATPase phosphorylation with P i (E2-P) are obtained only at acid pH and in the absence of Ca 2ϩ . Enzyme phosphorylation with P i was obtained by equilibration (5 min at 25°C) of SR vesicles (0.12 mg/0.2 ml) with 2.0 mM 32 P i , in a medium containing 100 mM MES, pH 6.5 (or 100 mM HEPES, pH 8), and 3 mM EGTA, in the presence or the absence of 10 mM MgCl 2 and 3.1 mM CaCl 2 . Quenching and phosphoenzyme determination are as described under "Materials and Methods." dependent and is much faster at 35 than at 25°C. Protection by ATP is also much more evident at 35°C (Fig. 7). It is interesting that the effect of ATP appears to be more pronounced on the 23-kDa than on the 35-kDa fragment, suggesting greater exposure of the latter in the ATP-bound conformation of the enzyme.
In the experiments shown in Figs. 6 and 7, the effect of temperature does not involve direct trypsin activation, because the first cut is produced very fast at all temperatures studied. It is clear that the effect of temperature is to promote conformational fluctuations and unfolding of the ATPase, whereby trypsin access to further proteolytic sites is facilitated. Protection by ATP binding is then produced by domain cross-linking and stabilization of conformational fluctuations, thereby limiting productive interaction of the ATPase with trypsin. We also found that the ADP⅐E1ϳP⅐2Ca 2ϩ fluoroaluminate analog exhibits very strong protection of the ATPase from digestion with trypsin. The protection is most evident at 35°C (Fig. 8, left  panel). If ADP is omitted, protection by the E1ϳP⅐2Ca 2ϩ analog is significantly weaker (Fig. 8, right panel). The experiments with trypsin digestion demonstrate that substrate (nucleotide) binding and the subsequent formation of phosphoenzyme intermediate produce strong stabilization of conformational fluctuations in the ATPase protein.
Functional Effects of Alkaline pH-Addition of ATP to SR vesicles is followed by rapid Ca 2ϩ transport, coupled to utilization of ATP. In the absence of oxalate trapping, maximal levels of Ca 2ϩ uptake are reached within 2 min, and these levels are lower at alkaline than at neutral or acid pH (Fig. 9). The levels of uptake are not limited by passive leak because if TG is added after reaching maximal level, at any pH, no reduction of these levels is observed for several minutes even though the ATPase activity is totally inhibited (not shown). In fact, the maximal levels of uptake are limited by the high Ca 2ϩ concentration formed in the lumen of the vesicles and the E2-P⅐2Ca 2ϩ dissociation constant. Assuming a luminal volume of 10 l per mg of SR protein (18), the luminal Ca 2ϩ concentrations reached at pH    6 -8, can be calculated to be 9.8, 8.1, and 3.7 mM, respectively. However, if we consider the significant quantity of Ca 2ϩ binding to internal sites such as calsequestrin, the free luminal Ca 2ϩ concentration at alkaline pH is likely to be within the 0.1 mM range. The lower levels of Ca 2ϩ uptake observed at alkaline pH indicate that lack of protons is a limiting factor for Ca 2ϩ /H ϩ exchange and net Ca 2ϩ dissociation from the phosphoenzyme into the lumen of the vesicles. This is consistent with Figs. 1 and  5 showing that the E1 and E1-P states (high Ca 2ϩ affinity) are favored by alkaline pH, whereas the E2 and E2-P states (low Ca 2ϩ affinity) are favored by acid pH.

DISCUSSION
Original diffraction studies (19,20) provided detailed representations of the ATPase crystal structure in the presence of high Ca 2ϩ or in the absence of Ca 2ϩ and presence of TG. The two resulting structures are quite different, including an open (E1-2Ca 2ϩ ) or closed (E2-TG) headpiece configuration, where the N, P, and A domains are separate or clustered, and their displacement is coupled to rearrangement of transmembrane segments to increase or decrease the binding affinity of the two Ca 2ϩ sites. Additional crystal structures were obtained in the presence of ligands (i.e. fluoroaluminate or MgF 4 ) producing stable transition state analogs of phosphoenzyme in the presence (i.e. analog of E1-P⅐2Ca 2ϩ ) or in the absence (i.e. analog of E2-P) of Ca 2ϩ (21)(22)(23). These structures provide very informative atomic models representing diverse conformations of mechanistic relevance, acquired by the ATPase protein under crystallization conditions, although ground states in the absence of analogs are expected to present some differences. On the other hand, with the experiments reported here, we studied the occurrence of conformational fluctuations and transitions in the native membrane environment, as they are revealed by the patterns and time course of ATPase digestion by proteinase K or trypsin. We characterized the dependence of these transitions on pH, temperature, catalytic ligands, and a specific inhibitor, and thereby their involvement in the mechanism of ATP utilization for Ca 2ϩ transport.
A specific feature of the experiments with proteinase K is the diversity of digestion patterns obtained in the presence or in the absence of Ca 2ϩ . This difference is related to the appearance of additional fragments (95 and 14 kDa), because of a more favorable exposure of the Leu 119 /Lys 120 site in the E2 configuration (Fig. 10). This site is located within the M2 helix that extends from the lumen of the SR vesicles to the A domain, and is inaccessible to proteinase K in the E1⅐2Ca 2ϩ state. On the other hand, in the E2⅐TG structure, the Leu 119 / Lys 120 is located within a short helical segment that is evidently accessible to proteinase K, because of unwinding of the M2 helix around Asn 111 and Ala 115 . This conformation change allows a longer path for M2 and rotation of the A domain to form the compact headpiece observed in the E2⅐TG crystal structure. The limited proteolysis observed in our experiments reveals then a functionally relevant transition from an E2 to an E1 (or "vice versa") pattern. It should be understood that in fact the E1 pattern is a common feature of sequential states related to H ϩ dissociation, cooperative binding of 2Ca 2ϩ , and catalytic activation (see below). Although only the final state (E1-2Ca 2ϩ ) has been characterized by crystallography (20), the occurrence of sequential conformational adjustments was inferred by experiments on Ca 2ϩ binding and catalytic activation (24).
We now find that, in the native membrane environment, transition between the two patterns of digestion is not strictly dependent on the presence of Ca 2ϩ . In fact, the E1 pattern can be obtained by raising the pH in the presence of Mg 2ϩ . This suggests that, at the physiological pH, the enzyme resides mostly in an E1 state even in the absence of Ca 2ϩ (25). However, Ca 2ϩ binding to E1 is still required to obtain catalytically active E1⅐2Ca 2ϩ . In fact, even though ATP can bind in the absence of Ca 2ϩ (Fig. 3), its terminal phosphate acquires a productive position (interacting with Asp 351 and Mg 2ϩ ) only in the presence of Ca 2ϩ (26 -29) (see also supplemental Fig. 1).
The results obtained with the E309Q and E771Q mutants (Fig. 2) indicate that, even in the absence of Ca 2ϩ , control of the E2 to E1 transition resides within the empty Ca 2ϩ sites (Fig. 2). Various residues that are likely to be involved in the stabilization of the empty Ca 2ϩ sites in the E2 configuration are shown in Fig. 11. Glu 771 is a critical residue as its mutation abolishes completely cooperative Ca 2ϩ binding at both sites (30). Glu 309 is also a critical residue, because its mutation interferes with Ca 2ϩ binding at the second site, as well as long range catalytic activation (31). It is likely that a specific network of hydrogen bonding, involving Glu 771 , Glu 309 , Asn 796 and other residues (32), is critical to the stability of the E2 state (Fig. 11). Measurements of charge movements upon Ca 2ϩ binding indicate that these residues release their protons within the 6 -7 pH range (33). Glu 309 appears most important because its (single) mutation to Gln (Fig. 2) yields the E1 digestion pattern at any pH. Its ionization is evidently important for transition to the E1 conformation. Glu 771 is also important because its (single) mutation to Gln allows transition to E1, even in the absence of Mg 2ϩ . In this case, however, a pH rise is required, most likely to induce H ϩ dissociation from Glu 309 and/or possibly other residues.
The Mg 2ϩ requirement for the E2 to E1 transition of the WT ATPase as the pH is raised (Fig. 2, top) indicates that, even following H ϩ dissociation from Glu 309 and Glu 771 , the E2 conformation may still be retained in virtue of residual weak interactions, which are then disrupted by Mg 2ϩ binding. In fact, under similar conditions, Mg 2ϩ binding occurs in the P domain, coordinated by Asp 703 and Asp 707 . Its presence is likely to interfere with stability of the A domain within the gathered headpiece, thereby allowing displacement of M2 and exposure of Leu 119 /Lys 120 site to proteinase K. On the other hand, in addition to its primary location at the catalytic site, Mg 2ϩ may also bind to the empty Ca 2ϩ sites. It was reported that, in the absence of Ca 2ϩ , interaction of Mg 2ϩ with SR-ATPase induces a fluorescence change because of pH-sensitive binding to the Ca 2ϩ sites and formation of an E⅐Mg dead-end complex (34). Furthermore, fluorescence measurements suggest that Mg 2ϩ , in the millimolar concentration range, is able to bind to the empty Ca 2ϩ sites, and this binding is competitively inhibited by Ca 2ϩ and H ϩ (35). At any rate, note that under all circumstances Ca 2ϩ binding is still required to yield the catalytically active E1⅐2Ca 2ϩ complex (Scheme 1).
It is of interest that the E1 and E2 patterns of digestion by proteinase K are also observed in the fluoroaluminate transition state analogs of E1ϳP⅐2Ca 2ϩ and E2-P (Fig. 4), as well as in the physiological E2-P state obtained by ATPase phosphorylation with P i in the absence of Ca 2ϩ (Fig. 5). This is helpful in demonstrating the effect of H ϩ and of H ϩ /Ca 2ϩ exchange in the E1ϳP⅐2Ca 2ϩ to E2-P conformational transition. It is clear that, following ATP utilization, alkaline pH favors the E1-P over the E2-P state (Scheme 1 and supplemental Fig. 1) not only from the functional (36) but also from the structural (Fig. 4) standpoint.
Another important aspect of our experiments is the evidence for prominent conformational fluctuations of the ATPase E1 and E2 ground states in the native membrane environment, as demonstrated (Fig. 6) by the thermal energy requirement for exposure of the Arg 198 site (Fig. 10) to trypsin. Consider that stabilization by multiple weak interactions depends on anisotropy, and thermally induced motion reduces electrostatic forces because of averaging effects by dipole rotations. In fact, spectroscopic studies have shown conformational heterogeneity of the ATPase protein (37), as well as prominent effects of ligands on the internal dynamics of the enzyme protein (17). Strong restraint of fluctuations is then observed upon acquisition of intermediate catalytic states. This is the case, for example, of the phosphoenzyme obtained with P i (i.e. E2-P) that displays a high resistance to digestion (Fig. 5), whereas the E2 digestion pattern is maintained. Stabilization upon formation of the ADP⅐E1ϳP⅐2Ca 2ϩ intermediate analog occurs as well FIGURE 11. Atomic model (stereo pair) of the empty Ca 2؉ -binding sites in the E2(TG ؉ DBHQ) structure (Protein Data Bank code 2agv). The explicit presence of hydrogen atoms was deducted from continuum electrostatic calculations (32). Our experiments demonstrate that in fact the E2 state is stabilized by the presence of these protons, and H ϩ dissociation favors transition to E1 even in the absence of Ca 2ϩ (and in the absence of TG and DBHQ stabilization). The broken lines show hydrogen bonds. Hydrogen bonds between main chain atoms are shown in green. SCHEME 1. Reaction scheme of the Ca 2؉ -ATPase catalytic and transport cycle based on the original Post-Albers model. The scheme includes exchange of 2H ϩ for 2Ca 2ϩ , as well as the dead-end complex formed by stabilization with the inhibitor TG. Our experiments demonstrate a pH dependence of the E2 to E1 transition, whereby H ϩ dissociation from acidic residues at the binding sites favors transition of 2H ϩ ⅐E2 to E1 even in the absence of Ca 2ϩ . Furthermore, the experiments demonstrate that the E1 and E2 ground states undergo strong conformational fluctuations. These fluctuations are tightly constrained by interaction with substrate, whereby thermal activation and substrate-free energy are required for the E1ϳP⅐2Ca 2ϩ to the E2-P⅐2Ca 2ϩ transition. Therefore, conformational work is a key for transduction of ATP free energy to active transport. The structures of some of the sequential states (or analogs thereof) were solved by crystallographic studies and are now available in the literature (for review see Ref. 6, and supplemental Fig. 1). (Fig. 4), whereas the E1 pattern of digestion with proteinase K is retained. Thermal activation and substrate-free energy are then required to overcome conformational stabilization and promote catalytic turnover. It is well known that phosphoenzyme can be obtained by utilization of ATP at low temperature (2-3°C), but thermal activation energy is required for progress to hydrolytic cleavage (38). On the other hand, regarding utilization of substrate-free energy, consider that a 3 orders of magnitude reduction of the binding affinity constant, as produced with the E1ϳP⅐2Ca 2ϩ to E2-P⅐2Ca 2ϩ transition, entails an input of 8 -9 kcal per cycle (Ϫ2RT ln(10 3 M Ϫ1 /10 6 M Ϫ1 )). It is then apparent that a large portion of the energy derived from ATP is actually utilized for conformational work to promote this transition. In fact, under conditions limiting the E1ϳP⅐2Ca 2ϩ to E2-P⅐2Ca 2ϩ transition (slippage of the pump), hydrolytic cleavage of E1ϳP⅐2Ca 2ϩ yields extra heat rather than work (39). It is noteworthy that the conformational work related to the E1ϳP⅐2Ca 2ϩ to E2-P⅐2Ca 2ϩ transition results in disruption of the high affinity Ca 2ϩ sites and opening of the exit path on the luminal side (see Scheme 1). Considering a 10 Ϫ3 M K d , luminal dissociation of Ca 2ϩ from E2-P.2Ca 2ϩ will proceed spontaneously as long as the luminal Ca 2ϩ concentration is below mM. Furthermore the K eq for hydrolytic cleavage of E2-P to E2⅐P i is nearly 1, and the K d value for P i dissociation from E2⅐P i is ϳ1 ϫ 10 Ϫ3 M. Therefore, these reactions require negligible energy input to proceed forward. Therefore, the return to the E2 ground state and closure of the luminal Ca 2ϩ path occur as soon as the long range constraint (Fig. 5) produced by the presence of phosphate at the catalytic site is relieved (Scheme 1). A full account of the K eq values for the ATPase sequential reactions, and the resulting free energy changes (including second order reactions), is given in Ref. 40.
Additional evidence for the functional relevance of dynamic fluctuations and conformational transitions, as revealed by patterns of proteolysis, is provided by the effects of the specific inhibitor TG. In fact, ATPase digestion by protein kinase K in the presence of Ca 2ϩ occurs with the E2 pattern if TG is present (i.e. just as if Ca 2ϩ were not present). It is clear that the mechanism of inhibition is based on interference with conformational transitions, whereby the enzyme is locked in an inactive dead-end complex (Scheme 1), which is in some respects different from the physiological E2 state (Fig. 5), but nevertheless prevents dynamic fluctuations and conformational transitions that are required for catalytic turnover.