Effect of Orthophosphate, Nucleotide Analogues, ADP, and Phosphorylation on the Cytoplasmic Domains of Ca2+-ATPase from Scallop Sarcoplasmic Reticulum*

The effects of orthophosphate, nucleotide analogues, ADP, and covalent phosphorylation on the tryptic fragmentation patterns of the E1 and E2 forms of scallop Ca-ATPase were examined. Sites preferentially cleaved by trypsin in the E1 form of the Ca-ATPase were detected in the nucleotide (N) and phosphorylation (P) domains, as well as the actuator (A) domain. These sites were occluded in the E2 (Ca2+-free) form of the enzyme, consistent with mutual protection of the A, N, and P domains through their association into a clustered structure. Similar protection of cytoplasmic Ca2+-dependent tryptic cleavage sites was observed when the catalytic binding site for substrate on the E1 form of scallop Ca-ATPase was occupied by Pi, AMP-PNP, AMP-PCP, or ADP despite the presence of saturating levels of Ca2+. These results suggest that occupation of the catalytic site on E1 can induce condensation of the cytoplasmic domains to yield a unique structural intermediate that may be related to the form of the enzyme in which the active site is prepared for phosphoryl transfer. The effect of Pi on the E2 form of the scallop Ca-ATPase was also investigated, when it was found that formation of E2-P led to extreme resistance toward secondary cleavage by trypsin and stabilization of enzymatic activity for long periods of time.

The sarco(endo)plasmic reticulum ATPases (SERCAs) 1 transport Ca 2ϩ against an electrochemical gradient from the cytoplasm into the intracellular membranous compartment of the sarcoplasmic or endoplasmic reticulum and play a major role in Ca 2ϩ homeostasis in both muscle and non-muscle cells (1,2). In the course of transporting Ca 2ϩ , these enzymes pass through a number of relatively well defined intermediate biochemical states that can be isolated and studied. These include forms in which the ␤-carboxyl group of a specific aspartyl residue (Asp 351 in the case of rabbit SERCA1a) is present as an acylphosphate mixed anhydride. Much of the experimental data has been interpreted in terms of the enzyme being able to exist in two basic forms: the E 1 state, in which the enzyme possesses high affinity binding sites for Ca 2ϩ and can be phosphorylated by ATP but not by P i , and the E 2 state, which possesses low affinity Ca 2ϩ sites and can be phosphorylated by P i but not by ATP (3).
Comparison of the tryptic digestion patterns of rabbit SERCA1a in its Ca 2ϩ -bound and -free forms has provided some of the strongest direct evidence supporting such a model where the enzyme can exist in two fundamentally different conformational states (4 -6). Recent studies have found that the E 1 (Ca 2ϩ ) 2 form of rabbit SERCa1a shows clear structural differences to the enzyme in its vanadate-stabilized Ca 2ϩ -free (E 2 ) state (7)(8)(9)(10). In the E 1 form, the Actuator (A) domain is isolated from the nucleotide (N) and phosphorylation (P) domains, whereas in the E 2 form all three domains are clustered together.
The Ca-ATPase from the cross-striated part of the adductor muscle of the sea scallop has been the subject of a number of biochemical and structural studies (11,12). Provided the E 2 form of the scallop Ca-ATPase is stabilized against loss of activity, it adopts a dimeric type of quaternary organization (13) that is absent in the E 1 form of the enzyme (14), and in both the E 1 ϳP and E 2 -P states the enzyme is arranged in parallel instead of antiparallel helical strands in the tubular vesicles with only a single asymmetric subunit in each unit cell (p1 lattice) (15).
In the work reported here, tryptic digests of scallop SERCA detected differences in conformation between the E 1 and E 2 forms of the enzyme associated with the N and P subdomains as well as with the A domain. It was found that some features of the proteolytic stability normally associated with the E 2 state were also observed when ligands ranging from AMP-PNP to simple orthophosphate were bound to the catalytic site of the unphosphorylated E 1 form of the enzyme, despite saturation of the Ca 2ϩ -binding sites. Thus, occupation of the catalytic site on E 1 could induce structural changes that resembled in some respects those produced by emptying of the high affinity Ca 2ϩ sites; however, this modified structural form of the E 1 state of the scallop Ca-ATPase was unique and differed from E 2 . The effect of covalent phosphorylation of the scallop Ca-ATPase was also investigated, when the E 2 -P form was found to be exceptionally stable, both in terms of enzymatic activity and resistance to proteolysis. In contrast, the E 1 ϳP form was highly susceptible to trypsin with a digestion pattern resembling that of the E 1 (Ca 2ϩ ) 2 .

EXPERIMENTAL PROCEDURES
Deep-sea scallops (Placopecten magellanicus) were obtained from the Marine Biology Laboratory, Woods Hole, MA.
Preparation of Scallop FSR-SR vesicles were made from the crossstriated part of the adductor muscle as described previously (16,17). One minor modification to the procedure was the removal of glycogen granules from the final preparation. Glycogen and its associated proteins are very common contaminants of SR vesicle preparations from rabbit skeletal muscle (18), and the same is true of scallop muscle membranes. In the case of the scallop SR, it was found that a simple and effective method to eliminate much of the glycogen was to layer the purified membrane fraction suspended in 0.32 M sucrose, 0.1 M KCl, 1 mM CaCl 2 , 20 mM MOPSNa, pH 7.0, onto 1.5 M sucrose, 0.1 M KCl, 1 mM CaCl 2 , 20 mM MOPSNa, pH 7.0, and centrifuge the preparation for 3 h at 1.5 ϫ 10 5 ϫ g. The glycogen particles collected as a clear button-like pellet at the bottom of the centrifuge tube, whereas the membranes, now largely free of glycogen, banded at the 0.32-1.5 M sucrose interface. Preparation of DOC-extracted scallop FSR was done as described previously (16).
Tryptic Digests-DOC-extracted scallop FSR was suspended at 1 mg ml Ϫ1 in 20% v/v ethylene glycol (Pierce), 0.15 M KCl, 50 mM MOPSNa, pH 7.0. The E 1 pattern of tryptic fragments was obtained when the free Ca 2ϩ concentration was above ϳ10 M, but typically a total of 1 or 3 mM CaCl 2 was present for digests in the E 1 state. For digests of the E 2 form, 10 mM EGTANa replaced CaCl 2 . Digestion was at room temperature with TPCK-treated trypsin (12,000 units/mg, dissolved in 1 mM HCl; Sigma) added in a 1:30 w/w ratio to SR protein (giving 400 units of activity/mg SR protein). Digestions were usually terminated by addition of AEBSF (Calbiochem) to a final concentration of 20 mM, followed by transfer of the sample to ice. The simultaneous addition of HCl to 1 mM with the 20 mM AEBSF proved a very effective way of stopping the digestion. The samples were centrifuged at 16,000 ϫ g for 1/2 h at 4°C and the supernatant removed. The pellets containing the membranebound products were washed to remove trapped protease by resuspension in 0.32 M sucrose, 1 mM AEBSF, 1 mM CaCl 2 (E 1 digests) or 1 mM EGTA (E 2 digests), 25 mM MOPSNa, pH 7.0, followed by recentrifugation. After repeating the washing step, the trypsin-free samples were finally resuspended in 0.32 M sucrose, 1 mM CaCl 2 (E 1 digests), or 1 mM EGTA (E 2 digests), 25 mM MOPSNa, pH 7.0, before addition of an equal volume of 2ϫ Laemmli or Tricine sample buffer.
Electrophoresis-The discontinuous Tris-glycine and Tris-Tricine systems (19,20) were used for running SDS-polyacrylamide gels. Sodium thioglycholate (0.1 mM) was present in the sample denaturation medium and cathode buffer.
Electroblotting and N-terminal Sequencing-SDS gels were blotted onto Immobilon-PSQ polyvinylidene difluoride in a medium of 10% v/v MeOH, 10 mM CAPSNa, pH 11, at 4°C with a Bio-Rad Trans Blot apparatus. Blots were stained with 0.02% w/v Coomassie Brilliant Blue R.250 in 1 mM HCl, 50% v/v MeOH. Bands of interest were sent to the University of Florida for N-terminal sequencing.
Phosphorylation with P I -For digestions in the E 2 -P state, DOCextracted scallop FSR was phosphorylated with P i essentially as described previously (15,21) at room temperature for 15 min before addition of trypsin.
Detection of E 2 -P Using TNP-ADP Superfluorescence-This was carried out using an SLM 8000c spectrofluorimeter thermostated at 25°C. TNP-ADP (Molecular Probes) was added to 35 g/ml Ϫ1 DOC-extracted scallop FSR suspended in 20% v/v glycerol, 15 mM EGTA-Tris, 15 mM MgCl 2 , 50 mM Mes-Tris, pH 6.0, to a final concentration of 2 M. Phosphorylation was typically induced by addition of P i in the form of H 3 PO 4 -Tris to 8.5 mM. Steady-state fluorescence measurements using 4-nm slits were carried out with excitation at 403 nm, and emission followed at 532 nm.
Enzyme Assays-The dependence of the Ca 2ϩ -activated ATPase activity of the DOC-extracted scallop FSR on ATPMg 2Ϫ concentration was determined using a coupled enzyme assay as previously described (16), except that 5 mM Mg 2ϩ was present (as MgCl 2 ) in excess of the AT-PMg 2Ϫ concentration, according to Neet and Green (22). Assays were carried out in a Perkin-Elmer Lambda 40 spectrophotometer in a volume of 1 ml. The cell was thermostated at 25°C.
Protein Concentration-The bicinchoninic acid method (23) was used.

RESULTS
General Observations-As previously described, on SDS gels the undigested DOC-extracted scallop FSR fraction showed essentially a single band of ϳ110 kDa known to contain the polypeptides of the Ca-ATPase (24) and a Na-Ca exchanger (25), the latter probably of sarcolemmal origin. Based on phosphorylation levels of the scallop SR obtained with both ATP and P i (15,21) and on its specific Ca 2ϩ -activated ATPase activity (13,16), the Ca-ATPase must represent at least 90% of the 110-kDa material. None of the membrane-bound proteolytic fragments described in this report was derived from the Na-Ca exchanger, although soluble tryptic peptides originating in the exchanger have been identified (25). Traces of 2-3 peptides of 28 -32 kDa that arose from contamination of the SR by other elements of the sarcolemma (possibly gap junctions) were sometimes present (15,25,26). Gels of tryptic digests where the scallop Ca-ATPase was in the E 2 state (ϩ EGTA) showed two strong bands of apparent molecular mass ϳ56 and ϳ47 kDa, together with much weaker bands corresponding to peptides of ϳ88, ϳ37, and 22-25 kDa (Fig. 1A). Digestions in the E 1 state (10 M free Ca 2ϩ or above) gave a pattern in which both the 56-and 47-kDa bands were less intense than in E 2 digests (Fig. 1B), whereas the 88-, 37-, and 22-25-kDa bands were significantly stronger.
Attempts to sequence the 110-and 56-kDa tryptic fragments by the Edman method suggested that they both had blocked N termini. Autoradiography of SDS gels of membranes that had been first proteolyzed in the E 2 state and then phosphorylated with [␥-32 P]ATP showed labeling of the 56-kDa band with 32 P as well as the intact ATPase (not shown). Rabbit SERCA1a has a blocked (acetylated) N terminus, as of course does its A tryptic fragment, and the tryptic A fragment representing the N-terminal half of SERCA1a contains Asp 351 . Thus, the 56-kDa peptide formed in scallop tryptic digests contained Asp 350 and corresponded as expected to the scallop A tryptic peptide. 2 The polypeptide with the apparent size of 47 kDa formed in the same (E 2 ) digests was found by Edman N-terminal sequencing of its polyvinylidene difluoride blot to be produced by cleavage at Lys 504 -Val 505 (scallop sequence, Ref. 27) in the N domain and thus represented the scallop B tryptic fragment.
Under otherwise identical conditions, a more rapid overall digestion process occurred in the E 1 compared with the E 2 state. Comparison of time courses obtained when the scallop Ca-ATPase was in its E 1 and E 2 forms (Fig. 1, A and B) showed that in the E 1 state the A band was lost much more rapidly than in E 2 . Although the B fragment was more stable than the A fragment in digests carried out in the presence of Ca 2ϩ (when the enzyme was in its E 1 form), it was nevertheless broken down faster than in the presence of EGTA (when the enzyme was in its E 2 form). Therefore, Ca 2ϩ -dependent cleavages occurred in both the N-and C-terminal halves of the scallop Ca-ATPase.
N-terminal sequencing showed that the 88-kDa fragment, which was formed in larger amounts in the E 1 than the E 2 state, arose by cleavage at Arg 197 -Ala 198 (scallop sequence, Ref. 27) in the A domain. As with rabbit SERCA1a, this will be designated the T 2 cleavage site. Therefore, for a subpopulation of the scallop Ca-ATPase molecules in the E 1 state, the primary tryptic cleavage site was T 2 rather than T 1 . The 37-kDa fragment was relatively stable under E 1 conditions and was often observed to be a very prominent band on SDS gels of tryptic digests made with the E 1 form of the scallop Ca-ATPase (Fig.  1A). N-terminal sequencing showed that this peptide arose by cleavage at the Lys 582 -Phe 583 peptide bond (scallop sequence, Ref. 27) in the N domain, which will be designated as the T 3 site in the scallop enzyme. N-terminal sequencing of the 22 and 24-kDa peptides preferentially formed in the E 1 state showed that formation of both of these peptides involved cleavage of the Lys 727 -Ser 728 bond in the C-terminal component of the P domain. The Lys 727 -Ser 728 bond will be designated as the T 4 tryptic cleavage site in the scallop Ca-ATPase.
Effect of AMP-PNP, AMP-PCP, and ADP on the E 1 Form of Scallop SR Ca-ATPase-When AMP-PNP was present at concentrations at or above ϳ0.1 mM in tryptic digests of the scallop Ca-ATPase in its E 1 state (with free Ca 2ϩ Ͼ 10 M), substantial stabilization of both the A and B fragments was observed (Fig.  2, A and B), and there was a significant reduction in the amount of 37-kDa fragment that accumulated. Although AMP-PNP greatly stabilized the A fragment (which contains the T 2 site) and inhibited formation of the 37-kDa peptide by cleavage at T 3 in the B fragment, it had little effect on the accessibility of the T 4 site to trypsin. AMP-PNP did not qualitatively modify the pattern of tryptic peptides formed from the E 2 form of the Ca-ATPase, but both the A and B bands became somewhat stronger as the AMP-PNP concentration was raised to 5 mM (Fig. 2B). This may suggest some additional stabilization through the nucleotide analogue binding to a low affinity site.
Addition of AMP-PCP or ADP to tryptic digests of scallop Ca-ATPase gave very similar effects to AMP-PNP over the same (0.1-1 mM) concentration range. In summary, the overall effect of the nucleotide ligands was to make the pattern of proteolytic products formed from the E 1 scallop Ca-ATPase closer to that normally seen in the E 2 state, with strong A and B bands on SDS gels and a weak 37-kDa band.
Effect of Orthophosphate on the Tryptic Cleavage of E 1 -The effect of a range of concentrations of orthophosphate on the tryptic cleavage pattern produced in the presence of Ca 2ϩ (enzyme in the E 1 state) is shown in Fig. 3. As the concentration of P i was increased to 20 mM, there were significant reductions in the amounts of the 37-and 88-kDa fragments formed, i.e. the cleavages at the T 3 site in the N domain and the T 2 site in the A domain were inhibited. P i strongly stabilized the A fragment (Met 1 -Lys 504 ), but as with the nucleotide ligands, there was no significant protective effect on the T 4 site. Thus, the P i -bound polyphosphate (5 mM) in E 1 digests produced similar effects to those seen with P i .
Effect of Orthophosphate on the Tryptic Cleavage of E 2 -Thus, P i profoundly affected the tryptic cleavage of the E 1 form of the scallop Ca-ATPase. The effect of P i on the tryptic cleavage of E 2 was then examined. As described above, cleavage at the T 2 , T 3 , and T 4 sites was inhibited when the scallop Ca-ATPase was in its E 2 form under the usual conditions in the presence of 0.13 or 0.15 M K ϩ at pH 7. The presence of 20 mM P i did not in any way modify the products of the E 2 digest made under these standard conditions, as expected. When the E 2 form of the Ca-ATPase was digested under identical conditions but with no K ϩ present, it was rapidly degraded into small fragments (not shown), as anticipated from previous studies that showed that in the absence of K ϩ the enzyme adopts a loose and open conformation (21). Orthophosphate (20 mM) had no effect on the very extensive proteolysis of E 2 in the standard digestion medium lacking K ϩ at pH 7.
The E 2 -P Form of Scallop Muscle SERCA Is Very Stable, with a Tightly Folded Conformation-The studies described above on the effect of P i on tryptic cleavage of E 2 had been carried out under conditions that did not promote formation of E 2 -P. The effect of covalent phosphorylation of E 2 was then investigated. Previous studies had shown that membranous scallop Ca-ATPase could be phosphorylated with P i to yield a form of the enzyme corresponding to the well-studied E 2 -P form of rabbit SERCA1a (15). Enhancement of steady-state fluorescence (super fluorescence) of TNP-ADP at 532 nm associated with formation of E 2 -P (28, 29) was used to characterize the affinity of the binding site for P i on scallop Ca-ATPase involved in formation of E 2 -P (see "Experimental Procedures"). The intensity increase was half maximal at 3.4 mM added P i , consistent with the expected affinity of the enzyme for P i (30).
When the E 2 form of the scallop Ca-ATPase was suspended in 20% v/v Me 2 SO and 15 mM Mg 2ϩ in the absence of P i , some limited stabilization of the A and B tryptic fragments was observed (Fig. 4, lane 2). However, when 20 mM P i was present together with 20% v/v Me 2 SO and 15 mM Mg 2ϩ in the K ϩ -free medium at room temperature to induce formation of the E 2 -P state, both the A and B fragments became extremely resistant to secondary cleavage by trypsin (Fig. 4, compare lanes 2 and  3). Even after exposure to trypsin for 1/2 h at room temperature, the A and B peptides formed from the E 2 -P enzyme remained essentially intact. Cleavage was thus effectively restricted to the T 1 site with very little secondary proteolysis, whereas after 1/2 h a significant amount of small proteolytic debris had accumulated in the comparable digest of the E 2 form. Therefore, although the T 1 site remained accessible to trypsin after phosphorylation of the E 2 to the E 2 -P form of scallop SERCA, the rest of the cytoplasmic part of the molecule adopted a conformation that was exceptionally resistant to further attack by trypsin. Because K ϩ activates hydrolysis of E 2 -P (31), it was not possible to directly compare the stabilities of E 2 and E 2 -P in the presence of K ϩ . It was noted that the SDS complex of the residual undigested (intact) E 2 -P form of the Ca-ATPase polypeptide migrated more slowly than the SDS complex of the undigested E 2 form (Fig. 4), suggesting that the E 2 -P-SDS complex was more extended than the E 2 -SDS complex. It is known from studies of rabbit SERCA1a that there is binding site for ADP on E 2 -P (29). When 4 mM ADP was included in tryptic digests of the E 2 -P form of scallop Ca-ATPase (Fig. 4, lane 4), the A fragment became more susceptible to tryptic cleavage and traces of the 22-24 kDa doublet appeared in the digest, indicating that the T 4 site had become more exposed. Thus, binding of ADP perturbed the structure of E 2 -P.
In the course of these studies, it was found that provided the scallop Ca-ATPase was maintained in the K ϩ -free E 2 -P phosphorylation medium, it could be kept at room temperature for extended periods of time (Ͼ10 days) without loss of activity. Again, as judged by TNP-ADP superfluorescence, both membranous and C 12 E 8 -solubilized scallop Ca-ATPase phosphorylated with P i were very stable with little decay of the signal, provided the samples were kept in the dark between measurements to prevent photobleaching. The stability of scallop FSR in the E 2 -P state, formed in the absence of Ca 2ϩ , K ϩ , and Na ϩ , was in complete contrast to the extremely rapid loss of activity that occurs with the membranous unphosphorylated E 2 form of the scallop enzyme under comparable conditions (11,13). Because inactivation of the unphosphorylated Ca 2ϩ -free (E 2 ) scallop Ca-ATPase involves an irreversible loss of the Ca 2ϩ -binding sites, phosphorylation of the enzyme in the absence of Ca 2ϩ with P i may thus stabilize otherwise labile empty Ca 2ϩ -binding sites.
In the context of the above results, it was of interest to compare the effect of tryptic digestion of the Ca-ATPase phosphorylated from ATP (ϩ Ca 2ϩ ) to that described above where the enzyme had been phosphorylated from P i (ϩ EGTA). Because the enzyme was found to be inactivated by secondary tryptic cleavages within the A and B fragments, phosphorylation was carried out before proteolysis in the absence of Mg 2ϩ in the presence of K ϩ , i.e. under conditions where most (Ͼ90%) of the phosphorylated enzyme was in the ADP-sensitive form (E 1 ϳP) as previously described (15). The pattern of tryptic fragments produced from the E 1 ϳP preparation of scallop Ca-ATPase as visualized with Coomassie Blue was indistinguishable from that of the unphosphorylated E 1 form of the Ca-ATPase, with significant breakdown of the A and B fragments.
Dependence of Enzyme Activity on ATPMg 2 Concentration-Information about the number and type of nucleotide-binding sites on the scallop SERCA was important for interpretation of some of the above results. Many studies of rabbit SERCA1a have suggested the coexistence of catalytic-and regulatory nucleotide-binding sites on that enzyme (33,34), whereas there is evidence for only one type of site on the SERCA of the cold-resistant wood frog (35). Thus, the dependence of the Ca 2ϩ -activated ATPase activity of deoxycholate-extracted scallop muscle SERCA vesicles on ATP concentration was examined according to Eadie-Hofstee (36), where the activity (v o ) is plotted against the ratio of activity to ATPMg concentration as shown in Equation 1.
The graph displayed two limbs, one with a slope corresponding to an apparent Michaelis constant K H of 0.29 mM, which extrapolated to give a k cat of 368.5 min Ϫ1 , and the other to a K m of 4.6 M and a k cat of 92.4 min Ϫ1 , (assuming a molecular mass of 110 kDa and that the Ca-ATPase constituted 95% of the total protein in the deoxycholate-extracted vesicles used in the experiments.) The central portion of a Hill plot of the data had a slope of n H ϭ 0.6, while the double reciprocal (Lineweaver-Burke) plot was convex upwards (not shown). These results strongly resembled those obtained with rabbit SERCA1a (37,38), which have been interpreted in terms of a high affinity catalytic nucleotide-binding site and a low affinity regulatory site. DISCUSSION The affinities of the catalytic-and regulatory nucleotidebinding sites on SERCa1a for ATP (2-5 M and 0.3-1 mM, respectively, (38) are very similar to the K m and K H of scallop Ca-ATPase found here. There is much evidence for two separate binding sites for nucleotide on each Ca-ATPase polypeptide chain (8,39,40), with one of the sites functioning as an allosteric regulator site and the other as the active (catalytic) site. Thus, there are two potential nucleotide-binding sites on the scallop Ca-ATPase that have to be taken into consideration with regard to stabilization of the scallop A and B fragments in the E 1 form of the Ca-ATPase by P i , AMP-PNP, AMP-PCP, and ADP.
The affinity of the catalytic site for AMP-PNP on rabbit SERCA1a (K d ϳ 75-90 M (36, 41, 42) is in keeping with the concentration of that ligand that stabilized the A and B fragments in tryptic digests of the E 1 form of scallop Ca-ATPase. Electron paramagnetic resonance experiments using spin-labeled rabbit SERCA1a (43) suggest that AMP-PCP binds to two sites on rabbit SERCA1a with dissociation constants of 50 and 650 M. The lower value fits with stabilization of the A and B fragments of the scallop Ca-ATPase through AMP-PCP binding to the catalytic site. Although metal-free ADP binds only weakly to the phosphorylated E 1 ϳP(Ca 2ϩ ) 2 form of rabbit SERCA1a (K d values ϳ0.73 mM) (44), it binds more strongly to the unphosphorylated enzyme with reported K d values of 12-50 M (45). These are consistent with stabilization of the A and B fragments in digests of the E 1 form of scallop Ca-ATPase by ADP being bound to the catalytic site.
In the case of the rabbit enzyme, E 1 and E 2 have the same affinity for P i (5-10 mM) (30), whereas the K d of the catalytic site for P i on the E 2 form of scallop Ca-ATPase was found in the studies reported here to be 3.4 mM on the basis of TNP-ADP fluorescence measurements. These values are in the concentration range where stabilization by P i of the A and B fragments of the E 1 form of the scallop enzyme was manifested, so that binding of P i to E 1 is likely to be mediated through its occupation of some part of the catalytic site. Occupation of only that part of the active site that interacts with the very small orthophosphate ligand was sufficient to cause a very substantial reorganization of the three subdomains (A, N, and P) that comprise most of the cytoplasmic region of the scallop Ca-ATPase. Although the nucleotides were effective at significantly lower concentrations than P i , this probably primarily reflects their higher binding affinities. Thus, binding of the adenosine moiety was not essential for stabilization of the E 1 form of scallop Ca-ATPase. Binding of P i or the polyphosphate moiety of nucleotide to the Ca-ATPase stabilizes a conformation where the cytoplasmic domains of the enzyme lie in close proximity to one another so that tryptic cleavages are inhibited in the A and N domains.
There is already good evidence that binding of nucleotide can modify the conformation of rabbit skeletal muscle and scallop adductor muscle SERCA. In particular, the rate of a conformational change associated with the binding of Ca 2ϩ is increased by ATP, and binding of ATPMg changes the conformation of the enzyme to one that is activated for phosphorylation (46). In fact, ADP, AMP-PCP, and AMP-PNP accelerate Ca 2ϩ binding in a pH-dependent manner (47), and changes in the amide I and II regions of the IR spectrum show that binding of nucleotide modifies the conformation of rabbit SERCA1a (48). In the case of the scallop Ca-ATPase, addition of 5 mM ATPMg 2Ϫ to the E 2 form of the scallop Ca-ATPase leads to ϳ4 additional thiol groups becoming less reactive toward the thiol reagent 5,5Ј-dithiobis(2-nitrobenzoate) (DTNB), whereas addition of 6 mM ATP to the E 1 form in the absence of Mg 2ϩ causes ϳ5 additional thiol groups to become inaccessible to DTNB (21).
All of the ligands that stabilized the A and B fragments in Ca 2ϩ -saturated (E 1 ) scallop Ca-ATPase against further digestion by trypsin induced a form of the enzyme with some conformational features in common with the Ca 2ϩ -free E 2 state, despite Ca 2ϩ being bound. However, there is no evidence that, for example, AMP-PNP lowers the affinity of the enzyme for Ca 2ϩ , thereby producing an E 2 -like state. On the contrary, occupation of the nucleotide-binding site has been reported to increase, not decrease, affinity of the Ca-ATPase for Ca 2ϩ (48). The T 4 site on E 1 was not protected by any of the agents, so that the modified form of E 1 induced by binding of substrate analogues or P i was not identical to E 2 and possessed a unique structure. Hence, although binding of Ca 2ϩ can profoundly modify interactions among the cytoplasmic subdomains, those regions of the enzyme still retain some independence from the transmembrane portion of the enzyme in their response to the binding of ligands.
It has been pointed out that the binding of Ca 2ϩ alone is not sufficient to position the nucleotide substrate close enough to Asp 351 for transfer of the ␥-phosphoryl group (49). The results reported here suggest that the additional reorganization of the catalytic center necessary for this to happen may originate in conformational changes caused by occupation of the active site, an example of the induced fit phenomenon. The occlusion of potentially sensitive sites toward trypsin that occurs when substrate analogues are bound to the catalytic nucleotide-binding site in the E 1 form of the Ca-ATPase may be related to the closing of the residual distance between Asp 350 (scallop sequence) and the ␥-phosphoryl group of ATP.
Recently, it was found that treatment of rabbit SERCA1a with the phosphate transition state analogues F Ϫ and orthovanadate caused the enzyme to become extremely resistant to the action of several proteases (50,51). In the current study, formation of the E 2 -P covalent adduct by addition of P i to the E 2 form of the scallop SR membranes suspended in a K ϩ -free medium prevented an otherwise rapid degradation of the A and B fragments by trypsin. This behavior suggests an unusually condensed conformation for the scallop enzyme in its E 2 -P state; however, because the E 2 -P-SDS complex had a lower mobility than the E 2 -SDS complex, overall the E 2 -P form may not be folded into a more globular shape than E 2 . The unphosphorylated E 2 form of scallop Ca-ATPase loses activity very rapidly and is highly susceptible to proteolysis in the absence of K ϩ , yet formation of the aspartyl phosphate at the active site was able to lock the enzyme into a conformation that was both structurally and functionally exceedingly stable. Thus, the effect of the covalent modification was propagated throughout the cytoplasmic domains of the enzyme.
The resistance of the E 2 -P form of scallop Ca-ATPase toward trypsin and the great stability of its enzyme activity compared with E 2 , both under conditions where K ϩ are absent, are very likely to have a common origin in the unusual state of the active site in E 2 -P. The synergistic binding of P i and Mg 2ϩ to E 2 is endothermic, but a substantial entropy increase associated with the release of a large number of water molecules from the active site and the hydration shell of P i allows the process to be driven thermodynamically (27,52). The activity of water at the catalytic site in the E 2 -Mg 2ϩ -P i ternary complex is therefore very low; this very non-polar environment promotes spontaneous formation of the acyl phosphate from P i and the ␤-carboxyl side chain of Asp 351 and then protects it from hydrolysis (3). The very hydrophobic nature of the active site in E 2 -P may underlie its great stability as well as the reported differences in quaternary and tertiary structure between E 2 -P and E 2 (15,(53)(54)(55).
Stabilization by P i of E 1 in a form where Ca 2ϩ -sensitive tryptic cleavage sites on the A and N domains are occluded toward trypsin and the great resistance of E 2 -P toward trypsin are consistent with what is known about the binding site for P i . All three cytoplasmic domains may have to be intimately juxtaposed to form the site in combination with the hinge region. Thus, atomic models suggest that formation of the non-polar cavity in E 2 -P may involve movement of the A domain close to the catalytic site (39). Support for such a structure comes from studies using proteinase K digestion of SERCA1a (56) and chymotryptic digestion of the Na, K-ATPase (57). Further strong evidence that the A domain contributes to the P i -binding site derives from site-directed mutagenesis and chemical modification studies of Gly 233 and Arg 198 , both located in L 23 , (58, 59). The close proximity of Glu 485 -Asp 489 to Thr 170 -Leu 172 in the E 2 crystal structure of SERCA1a suggests that the N domain may also be involved in the binding of P i , and there is strong evidence that the C-terminal part of the P domain of SERCA1a is needed for P i to bind (60). The h1-h2 hinge connecting the N and P domains must participate in the active site in E 2 -P because it contains Asp 351 . There is also good evidence that the hinge is directly involved in the binding of P i , because the phosphate transition-state analogue orthovanadate binds close to Thr 353 in h1 (61), and the 601 DPPR motif in h2 provides ligands for the binding of P i (62). Formation of the P i -binding site thus appears to involve all the major structural elements of the cytoplasmic region of the Ca-ATPase. Orthophosphate or the phosphate moieties of nucleotides may therefore act to draw the cytoplasmic domains and the hinge together and so stabilize a compact structure resistant to trypsin.
In conclusion, large-scale structural changes can be initiated in the Ca-ATPase not only by the binding and release of Ca 2ϩ but also by interaction of phosphate groups with the active site. Another example of a major structural reorganization in a protein associated with P i binding/release is the very large conformational change that occurs when P i dissociates from the S-1 head of myosin after hydrolysis of ATP (32).