Superinhibitory Phospholamban Mutants Compete with Ca2+ for Binding to SERCA2a by Stabilizing a Unique Nucleotide-dependent Conformational State*

Three cross-linkable phospholamban (PLB) mutants of increasing inhibitory strength (N30C-PLB < N27A,N30C,L37A-PLB (PLB3) < N27A,N30C,L37A,V49G-PLB (PLB4)) were used to determine whether PLB decreases the Ca2+ affinity of SERCA2a by competing for Ca2+ binding. The functional effects of N30C-PLB, PLB3, and PLB4 on Ca2+-ATPase activity and E1∼P formation were correlated with their binding interactions with SERCA2a measured by chemical cross-linking. Successively higher Ca2+ concentrations were required to both activate the enzyme co-expressed with N30C-PLB, PLB3, and PLB4 and to dissociate N30C-PLB, PLB3, and PLB4 from SERCA2a, suggesting competition between PLB and Ca2+ for binding to SERCA2a. This was confirmed with the Ca2+ pump mutant, D351A, which is catalytically inactive but retains strong Ca2+ binding. Increasingly higher Ca2+ concentrations were also required to dissociate N30C-PLB, PLB3, and PLB4 from D351A, demonstrating directly that PLB antagonizes Ca2+ binding. Finally, the specific conformation of E2 (Ca2+-free state of SERCA2a) that binds PLB was investigated using the Ca2+-pump inhibitors thapsigargin and vanadate. Cross-linking assays conducted in the absence of Ca2+ showed that PLB bound preferentially to E2 with bound nucleotide, forming a remarkably stable complex that is highly resistant to both thapsigargin and vanadate. In the presence of ATP, N30C-PLB had an affinity for SERCA2a approaching that of vanadate (micromolar), whereas PLB3 and PLB4 had much higher affinities, severalfold greater than even thapsigargin (nanomolar or higher). We conclude that PLB decreases Ca2+ binding to SERCA2a by stabilizing a unique E2·ATP state that is unable to bind thapsigargin or vanadate.

cannot be used to assess the Ca 2ϩ affinity of SERCA2a expressed alone, the direct effect of PLB on Ca 2ϩ affinity has yet to be determined. To overcome this limitation, here we com-pared the effects of a series of cross-linkable PLB mutants of increasing inhibitory strength on Ca 2ϩ binding to the Ca 2ϩ pump. If PLB competes with Ca 2ϩ for binding to the Ca 2ϩ -ATPase, then as PLB becomes a stronger inhibitor of enzyme activity, higher concentrations of Ca 2ϩ should be required to dissociate it from the Ca 2ϩ pump.
Earlier mutagenesis studies showed that the inhibitory function of PLB was enhanced by point mutations that either increased PLB monomer formation by destabilizing the PLB pentamer (e.g. L37A (3,4)), or otherwise enhanced PLB monomer binding interactions with the Ca 2ϩ pump (e.g. N27A (15) and V49G (16,17)). As these gain-of-function PLB mutants increased the K Ca of enzyme activation more than wild-type PLB, they were termed "supershifters" (3). In the present study, supershifting mutations were combined with the cross-linking mutation, N30C, to create two new PLB mutants, PLB3 (N27A,N30C,L37A-PLB) and PLB4 (N27A,N30C,L37A,V49G-PLB) (Fig. 2). PLB3 and PLB4 are predicted to be strongly inhibitory compared with N30C-PLB (which has a normal inhibitory strength (7)), while remaining cross-linkable to the Ca 2ϩ pump, thus allowing their physical interactions with SERCA2a to be measured simultaneously with their functional effects on enzyme activity. These cross-linkable PLB mutants of increasing inhibitory potency were used to demonstrate directly that PLB decreases the actual Ca 2ϩ binding affinity of the enzyme.
Our results indicate that PLB supershifters act by stabilizing a single conformation of SERCA2a, the nucleotide-bound E2 state, forming a ternary complex (E2⅐ATP⅐PLB) that is highly resistant to traditional Ca 2ϩ pump inhibitors like TG and vanadate, which also act at E2 (18,19). In addition, using these superinhibitory PLB mutants we gained new insights on what effect, if any, PLB has on the V max of enzyme activity (20 -23), and whether the catalytically inactive SERCA mutant, D351A (14), has substantially enhanced Ca 2ϩ binding affinity, as was reported earlier (24), but not subsequently confirmed (25).

EXPERIMENTAL PROCEDURES
Materials-The cross-linking agent KMUS was purchased from Pierce. [␥-32 P]ATP was obtained from PerkinElmer Life Sciences, and thapsigargin and sodium orthovanadate were purchased from Sigma.
Mutagenesis and Baculovirus Production-Mutation of canine SERCA2a and PLB cDNAs was conducted as described previously (4). For consistency with previous cross-linking studies, N30C-PLB was made on the Cys-less PLB background, in which Cys residues 36, 41, and 46 were mutated to Ala (7,10). N30C-PLB has been previously well characterized, and is fully functional with an inhibitory potency similar to wild-type PLB (7,10). In control experiments, FIGURE 1. Reaction cycle of SERCA2a. E1 and E2 represent the high and low Ca 2ϩ affinity conformations of SERCA2a, respectively. After sequential binding of two Ca 2ϩ ions to E1, the enzyme is phosphorylated with the ␥-phosphate of ATP at Asp 351 , forming the high energy intermediate, E1ϳP. Ca 2ϩ translocation across the SR membrane occurs during the E1 to E2 transition. TG inhibits Ca 2ϩ -ATPase activity by forming a dead-end complex with the enzyme in E2 (E2⅐TG) (18). E2⅐TG has a greatly reduced affinity for ATP relative to TG-free E2 (33,34). PLB cross-linking studies indicate that PLB binds preferentially to E2 with bound ATP (E2⅐ATP⅐PLB). PLB does not bind to E2⅐TG or E2⅐cyclopiazonic acid (7,10), E2-P (17), or to the Ca 2ϩ pump with Ca 2ϩ binding site 1 (12) or both sites (4,17) occupied. FIGURE 2. Complete amino acid sequences of the cross-linkable PLB mutants, N30C-PLB, PLB3, and PLB4. I and II designate cytoplasmic and transmembrane domains of PLB, respectively. Domain IA contains Ser 16 and Thr 17 , the residues phosphorylated in response to ␤-adrenergic stimulation. The point mutations N27A in domain IB (15) and L37A (3,4) and V49G (16,17) in domain II are all supershifting mutations. The N30C mutation in domain IB allows PLB to be cross-linked to Lys 328 of SERCA2a with KMUS (10). For consistency with previous publications, N30C-PLB was made on the Cys-less PLB background in which Cys residues 36, 41, and 46 were mutated to Ala (7, 10 -12, 17).
identical results were obtained when N30C-PLB was made on the wild-type PLB background with Cys residues 36, 41, and 46 unaltered (data not shown). cDNAs encoding PLB3 and PLB4 were generated on the wild-type PLB cDNA background inserted in the transfection vector pVL1393, using the QuikChange TM XL-Gold system (Stratagene). D351A was made similarly using canine cardiac SERCA2a cDNA as the template (10). All mutated cDNAs were confirmed by DNA sequencing of the plasmid vectors. Baculoviruses encoding mutated proteins were generated as described previously with BaculoGold TM (Pharmengen) linearized baculovirus DNA (10).
Protein Expression and Characterization-Sf21 insect cells were co-infected with baculoviruses encoding PLB and SERCA2a as described previously (4). Viral titers were adjusted to give an expression level of PLB to SERCA2a of ϳ4:1, as used in previous publications (7, 10 -12, 17). Cells were harvested 60 h after co-infection, washed with phosphate-buffered saline, and homogenized with a Polytron for 90 s at 15,000 ϫ g. Crude microsomal pellets were then collected by centrifuging at 48,000 ϫ g for 20 min. Microsomes were re-suspended at a protein concentration of 6 -10 mg/ml in 0.25 M sucrose, 10 mM MOPS (pH 7.0) and stored frozen in small aliquots at Ϫ40°C. Protein concentrations were determined by the Lowry method. PLB and SERCA2a contents in the membrane samples were determined by quantitative Western blotting with monoclonal antibodies 2D12 and 2A7-A1, respectively (7). Only membranes expressing PLB and SERCA2a at a molar ratio of ϳ4:1 were used for further analyses. As shown in Fig. 3, all PLB mutants were predominantly monomeric on SDS-PAGE. The low pentamer stability of N30C-PLB made on the Cys-less PLB background was reported previously (7).
Ca 2ϩ -ATPase Assay-Ca 2ϩ -ATPase activities were measured at 37°C in buffer containing 50 mM MOPS (pH 7.0), 100 mM KCl, 3 mM MgCl 2 , 3.0 mM ATP, 5 mM NaN 3 , 3 g/ml of the Ca 2ϩ ionophore, A23187, and 1 mM EGTA. Ionized Ca 2ϩ concentrations were set by varying the CaCl 2 concentration from 0 to 1.2 mM. Assays were conducted in the presence and absence of the anti-PLB monoclonal antibody, 2D12, which reverses PLB inhibition of SERCA2a (11,26). Ca 2ϩ -dependent ATPase activities were determined in a reaction volume of 1 ml containing 50 -100 g of membrane protein during a 30 -60-min incubation. P i release from ATP was measured colorimetrically (7). Maximal Ca 2ϩ -ATPase activities ranged between 15 and 25 mol of P i /mg of protein/h for all samples, which is ϳ25-40% of the maximal Ca 2ϩ -ATPase activity typically reported for dog cardiac SR vesicles (27). In some Ca 2ϩ -ATPase assays, small aliquots were taken from the assay tubes during the incubations, to simultaneously measure PLB cross-linking to SERCA2a (see below).
PLB Cross-linking to SERCA2a-In most experiments, crosslinking of N30C of PLB to Lys 328 of SERCA2a with KMUS was conducted identically as previously described (10). Cross-linking reactions were conducted with 11 g of membrane protein in 12 l of buffer. The final concentrations of PLB and SERCA2a in the cross-linking tubes were 1.2 and 0.3 M, respectively. Standard cross-linking buffer contained 50 mM MOPS (pH 7.0), 3.0 mM MgCl 2 , 100 mM KCl, 3 mM ATP, and 1 mM EGTA with zero to 1.2 mM added CaCl 2 . In some experiments, ATP concentrations were varied, or different nucleotides were used, as indicated in the figure legends. In the experiments with SERCA2a inhibitors TG and vanadate, TG was added from a 59 mM stock solution in ethanol, and sodium orthovanadate was added from a 15 mM stock solution in H 2 O. Cross-linking reactions were started by adding 0.75 l of 1.6 mM KMUS dissolved in Me 2 SO (final KMUS concentrations 0.1 mM), and the incubations were conducted for 2 min at room temperature. Reactions were stopped by adding 7.5 l of gel loading buffer containing 15% SDS and 100 mM dithiothreitol. The samples were subjected to SDS-PAGE, and Western blotting was performed with the anti-PLB antibody, 2D12, using 125 I-protein A for PLB visualization. In the experiment of Fig. 6, blots were probed directly with 125 I-2D12, and protein A was omitted (11). Radioactive signals (representing SERCA2a with bound PLB) were quantified using a Bio-Rad Personal Fx Phosphoimager. For economy of space, only the region of the autoradiographs containing PLB cross-linked to SERCA2a is displayed, except for the experiment shown in Fig. 7, A and B, in which the entire autoradiograph is shown.
For the data depicted in Figs. 4, 6, and 7, to assess Ca 2ϩ effects on PLB cross-linking to SERCA2a, experiments were conducted at 37°C in the same buffer used for measurement of Ca 2ϩ -ATPase activity, as described above. 15 min after initiation of Ca 2ϩ -ATPase reactions with ATP, 80-l aliquots containing 8 g of membrane protein were taken from Ca 2ϩ -ATPase assay tubes and cross-linked with 1 mM KMUS for 15 s, giving the maximal cross-linking obtainable at each Ca 2ϩ concentration tested. Reactions were stopped with gel loading buffer, and samples were then processed as described above. Cross-linking of PLB to D351A to assess Ca 2ϩ affinity was determined under identical conditions.
It should be pointed out that the heterobifunctional crosslinking agent KMUS reacts irreversibly with Lys 328 of SERCA2a and N30C of PLB whether the two proteins are bound or not. If the two proteins are bound when the cross-linker is added, then PLB is irreversibly cross-linked to SERCA2a by a single KMUS molecule. If the proteins are not bound when the cross-linker is added, then N30C of PLB can react with one KMUS molecule, and Lys 328 of SERCA2a can react with a second KMUS molecule, thus blocking additional cross-linking of the two proteins as new PLB⅐SERCA2a complexes are formed. Therefore, the amount of PLB⅐SERCA2a complex detected by chemical crosslinking in this study is essentially a "snapshot" of the amount of PLB⅐SERCA2a complex present at the time at which the crosslinker is added.
Phosphorylation of E1⅐Ca 2 by [␥-32 P]ATP-Phosphorylation of SERCA2a using [␥-32 P]ATP was conducted by incubating 11 g of membrane protein in 12 l of buffer containing 50 mM MOPS (pH 7.0), 3.0 mM MgCl 2 , 100 mM KCl, 1 mM EGTA, and 0 -1.2 mM CaCl 2 . Phosphorylation was initiated by adding a final concentration of 200 M [␥-32 P]ATP and conducted for 5 s at room temperature. Reactions were terminated with 7.5 l of acidic gel loading buffer (pH 2.4) containing 3% lithium dodecyl sulfate, and lithium dodecyl sulfate-PAGE was conducted under acidic conditions as recently described (17). After electrophoresis, proteins were transferred to nitrocellulose and the radioactive acylphosphoprotein bands were visualized by autoradiography and quantified with the Fx Phosphoimager.

RESULTS
Ca 2ϩ Activation of Ca 2ϩ -ATPase Activity and Ca 2ϩ Inhibition of PLB Cross-linking-In the present study, SERCA2a was expressed alone in Sf21 insect cells, or co-expressed with the three cross-linkable PLB mutants, N30C-PLB, PLB3, and PLB4, which were designed to be of increasing inhibitory strength. Prior to functional analyses, protein expression levels were quantified by Western blotting, and Ca 2ϩ -ATPase activities were then corrected for small variability in SERCA2a expression between preparations (Ϯ20%). Fig. 3 demonstrates that similar levels of SERCA2a and PLB were co-expressed in the different membrane preparations.
SERCA2a expressed alone exhibited typical ATP hydrolysis, with half-maximal activation of Ca 2ϩ -ATPase activity occurring at 0.16 M Ca 2ϩ (K Ca ϭ 0.16 M), and maximal enzyme activity reached at the saturating Ca 2ϩ concentration of 1-2 M ( Fig. 4A and Table 1). At Ca 2ϩ concentrations greater than 2 M, substantial back inhibition of the enzyme by Ca 2ϩ was observed (28,29). Co-expression of SERCA2a with N30C-PLB  . Ca 2؉ activation of Ca 2؉ -ATPase activity and Ca 2؉ inhibition of cross-linking. SERCA2a was expressed alone or co-expressed with N30C-PLB, PLB3, or PLB4 in Sf21 cells and SERCA2a and PLB expression levels were determined by Western blotting. Panel A depicts Ca 2ϩ -ATPase activities of membrane fractions measured as described under "Experimental Procedures." Enzyme activities were normalized to expression levels of SERCA2a expressed alone. The gray line intersecting the ordinate indicates the 50% V max value determined for SERCA2a expressed alone. Panel B shows cross-linking of the PLB mutants to SERCA2 determined under identical conditions as the Ca 2ϩ -ATPase assay. Aliquots were taken from the Ca 2ϩ -ATPase assay and crosslinked for 15 s with 1 mM KMUS at 37°C. Samples were then subjected to SDS-PAGE and immunoblotting with the anti-PLB antibody, 2D12. Protein bands in the upper panel show SERCA2a cross-linked with the PLB monomer. PLB cross-linking is quantified in the graph below. The graph in panel C was derived from the data in panels A and B. The percent of maximal PLB crosslinking to SERCA2a (determined in the absence of Ca 2ϩ ) was calculated at each Ca 2ϩ concentration for each PLB mutant, and then plotted against the percent inhibition of Ca 2ϩ -ATPase activity by PLB obtained at the same Ca 2ϩ concentration. The percent inhibition of Ca 2ϩ -ATPase activity by PLB at each Ca 2ϩ concentration was calculated by dividing the Ca 2ϩ -ATPase activity of membranes expressing SERCA2a plus PLB by the Ca 2ϩ -ATPase activity of membranes expressing SERCA2a alone, and multiplying by 100.
increased the K Ca value for enzyme activation ϳ2-fold, from 0.16 to 0.33 M, with little or no effect on the V max of the enzyme measured at 1-2 M Ca 2ϩ ( Fig. 4A and Table 1). The effect of N30C-PLB on enzyme activity observed here is identical to the effect of wild-type PLB reported previously (4). In contrast to N30C-PLB, PLB3 and PLB4 had large effects on both the K Ca of enzyme activation and on V max . The K Ca values were increased 3.3-(0.53 M Ca 2ϩ ) and 4.4-fold (0.70 M Ca 2ϩ ) by PLB3 and PLB4, respectively, when calculated based on the highest Ca 2ϩ -ATPase activity achieved by SERCA2a co-expressed with these two mutants, which occurred at 2 M Ca 2ϩ . It should be noted that at 2 M Ca 2ϩ concentration, the V max of the enzyme was inhibited by ϳ30% by both PLB3 and PLB4, relative to the same amount of SERCA2a expressed alone or with N30C-PLB (Fig. 4A). Nevertheless, complete reversal of Ca 2ϩ -ATPase inhibition by PLB3 and PLB4 did occur at much higher Ca 2ϩ concentrations (in the range of 100 -200 M), Ca 2ϩ concentrations at which significant back inhibition of the enzyme occurred. Thus, in Ca 2ϩ -ATPase assays, SERCA2a coexpressed with PLB3 or PLB4 can never achieve its maximal turnover rate, even though very high Ca 2ϩ concentrations do completely reverse Ca 2ϩ -ATPase inhibition by the supershifters. To correct for back inhibition of the enzyme by Ca 2ϩ , we also calculated the K Ca values for SERCA2a co-expressed with N30C-PLB, PLB3, and PLB4 using the V max value for SERCA2a expressed alone (Fig. 4A, gray lines intersecting the abscissa) (*K Ca values). When calculated by this method, the *K Ca values for SERCA2a co-expressed with N30C-PLB, PLB3, and PLB4 were 0.35, 0.88, and 1.49 M, respectively (Table 1, parentheses). These corrected *K Ca values indicate that 2.2-, 5.5-, and 9.3-fold higher Ca 2ϩ concentrations are required to half-maximally activate SERCA2a co-expressed with N30C-PLB, PLB3, and PLB4, respectively.
Next, the relative binding affinities of the PLB mutants for SERCA2a were estimated by measuring Ca 2ϩ inhibition of PLB cross-linking to the enzyme. All three PLB mutants crosslinked only to the cardiac Ca 2ϩ pump expressed in insect cell membranes, and all cross-linking results reported here depict the PLB monomer bound to SERCA2a (7, 10) (see "Experimental Procedures"). In the absence of Ca 2ϩ , strong cross-linking of all three PLB mutants to Lys 328 of SERCA2a was observed, and cross-linking was completely eliminated by increasing Ca 2ϩ concentration (Fig. 4B). The K i values for Ca 2ϩ inhibition of N30C-PLB, PLB3, and PLB4 cross-linking to SERCA2a were 0.35, 0.88, and 1.8 M Ca 2ϩ , respectively (Table 1). These K i values for Ca 2ϩ inhibition of cross-linking agree closely with the *K Ca values determined for half-maximal activation of Ca 2ϩ -ATPase activity. This demonstrates a strong correlation between PLB binding to E2, and decreased Ca 2ϩ affinity of SERCA2a determined by the Ca 2ϩ -ATPase assay. This conclusion is strengthened by plotting the percent maximal PLB cross-linking to SERCA2a (determined at each Ca 2ϩ concentration covering the Ca 2ϩ concentration range from 0.12 to 200 M), against the percent inhibition of Ca 2ϩ -ATPase activity determined at the same Ca 2ϩ concentrations (Fig. 4C). For all three PLB mutants, regardless of inhibitory strength, there was strong correlation (r 2 ϭ 0.97) between the extent of PLB crosslinking to SERCA2a and degree of enzyme inhibition. These data strongly suggest that SERCA2a with bound PLB is catalytically inactive, and point to competitive binding of PLB and Ca 2ϩ as the mechanism of enzyme inhibition.
Consistent with the Ca 2ϩ -ATPase results, similar shifts in K Ca by the different PLB mutants were observed when Ca 2ϩ stimulation of phosphoenzyme formation from ATP was monitored (Fig. 5). The K Ca for SERCA2a expressed alone was 0.11 M, whereas when co-expressed with N30C-PLB, PLB3, and PLB4, the K Ca values were 0.36, 1.16, and 2.05 M (Table 1), respectively. These K Ca values are nearly identical to the K i values determined for Ca 2ϩ inhibition of PLB cross-linking (gray lines). These results are particularly significant due to the fact that back inhibition of the Ca 2ϩ pump is not a factor when phosphoenzyme formation from [␥-32 P]ATP is monitored, therefore, no correction for loss of enzyme turnover at high Ca 2ϩ concentrations is required when K Ca values are estimated by this method.
Effect of 2D12 on Ca 2ϩ -ATPase Activity and PLB Crosslinking-The anti-PLB monoclonal antibody, 2D12, recognizes residues 7-13 of PLB and reverses PLB inhibition of the Ca 2ϩ pump by physically disrupting PLB binding to SERCA2a (11). 2D12 reverses enzyme inhibition by wild-type PLB (4, 26) and N30C-PLB virtually completely (11), but only partially reverses the effects of several supershifting PLB mutants on Ca 2ϩ -ATPase activity (28). This suggests that the PLB supershifters may bind more tightly to the Ca 2ϩ pump than wild-type PLB or

values (M) for Ca 2؉ -ATPase activation and phosphorylation of E1⅐Ca 2 with ͓␥-32 P͔ATP, and K i values (M) for Ca 2؉ inhibition of PLB cross-linking
SERCA2a was expressed alone or co-expressed with N30C-PLB (N30C), PLB3, or PLB4 in insect cell microsomes. Ca 2ϩ -ATPase activities and cross-linking were determined under identical conditions in the presence and absence of the anti-PLB antibody, 2D12, as described in the text. ND, not determinable. N30C-PLB, but this has not been demonstrated directly. Therefore, to confirm tighter binding of the PLB supershifters, and to show that the Ca 2ϩ affinity of the enzyme is restored commensurate with dissociation of PLB from SERCA2a, we measured the effect of 2D12 on N30C-PLB, PLB3, and PLB4 cross-linking to SERCA2a simultaneously with Ca 2ϩ -ATPase activity. Ca 2ϩ -ATPase activity of SERCA2a co-expressed with N30C-PLB, PLB3, and PLB4 was measured in the presence of 2D12 (2.2 M), a concentration sufficient to completely saturate PLB (Fig. 6, A-C). As shown previously (11), addition of 2D12 restored the Ca 2ϩ affinity of SERCA2a co-expressed with N30C-PLB nearly completely, decreasing the K Ca from 0.33 to 0.19 M Ca 2ϩ , compared with 0.16 M Ca 2ϩ for SERCA2a expressed alone (Table 1). On the other hand, 2D12 only partially restored the Ca 2ϩ affinity of the enzyme co-expressed with the superinhibitory mutants PLB3 and PLB4, shifting the K Ca values from 0.53 to 0.26 M for PLB3, and from 0.70 to 0.36 M for PLB4. Also, whereas 2D12 had little or no effect on the V max of the enzyme co-expressed with N30C-PLB, 2D12 increased the V max of the enzyme co-expressed with PLB3 and PLB4 significantly. In the presence of the antibody at 1-2 M Ca 2ϩ , Ca 2ϩ pumps co-expressed with PLB3 and PLB4 achieved 80 -95% of the maximal activity of Ca 2ϩ pumps expressed alone (gray lines).

Protein expressed K Ca values K i values, cross-linking Ca 2؉ -ATPase activity
The effects of 2D12 on PLB cross-linking during the same assay are shown in Fig. 6, D-F. Consistent with previous results, 2D12 inhibited cross-linking of N30C-PLB to the Ca 2ϩ pump nearly completely in the absence of Ca 2ϩ and at all Ca 2ϩ concentrations tested (7,11). This explains why Ca 2ϩ pump inhibition by N30C-PLB is 30% or less at each Ca 2ϩ concentration tested when Ca 2ϩ -ATPase activity was measured in the presence of 2D12. On the other hand, PLB3 and PLB4 cross-linking to SERCA2a in the absence of Ca 2ϩ was substantially reduced by addition of 2D12, but not eliminated altogether (25 and 53% maximal cross-linking persisted, respectively). Even in the presence of the antibody, Ca 2ϩ concentrations of 1 M or higher were required to completely dissociate PLB3 and PLB4 from the Ca 2ϩ pump (Fig. 6, E and F). These cross-linking results agree well with the results of the Ca 2ϩ -ATPase assays, which showed that the enzyme was significantly inhibited by PLB3 and PLB4 even in the presence of 2D12. In experiments not shown, the binding affinity of PLB for 2D12 was determined to be 0.1 M. Therefore, we conclude that the binding affinities of PLB3 and PLB4 for SERCA2a must be very high, at least within the range at which PLB binds 2D12.
Effect of Ca 2ϩ on PLB Cross-linking to D351A-To test directly for competition between PLB and Ca 2ϩ for binding to SERCA2a, we took advantage of the Ca 2ϩ pump mutant, D351A. During catalysis, Asp 351 is phosphorylated by ATP to form the high-energy acylphosphoprotein intermediate, E1ϳP⅐Ca 2 (Fig. 1). Replacement of aspartic acid at this position renders the enzyme catalytically inactive (14,30). Although inactive at the site of ATP hydrolysis, D351A retains the ability to bind Ca 2ϩ and maintains the thermodynamic equilibrium between E1 and E2 (14,24,25). Therefore, if PLB acts by stabilizing E2 and shifting the E1⅐Ca 2 7 E2⅐PLB equilibrium away from E1, then this effect should be fully reproducible with D351A. The advantage of using D351A for these experiments is that enzyme turnover is prevented; hence the system is at equilibrium with respect to Ca 2ϩ binding (Fig. 1). Consistent with previous results with SERCA1a, we first confirmed that the D351A mutant made from SERCA2a exhibited no Ca 2ϩ -ATPase activity, and was not phosphorylatable by [␥-32 P]ATP to form E1ϳP (30), nor by P i to form E2-P (14) (data not shown).
Next, the affinity of D351A for Ca 2ϩ was compared with that of wild-type SERCA2a by measuring Ca 2ϩ inhibition of N30C-PLB cross-linking. In the absence of Ca 2ϩ , D351A and wildtype SERCA2a bound comparable amounts of N30C-PLB (Fig.  7, A and B). However, a strikingly lower Ca 2ϩ concentration was sufficient to disrupt N30C-PLB cross-linking to D351A (K i ϭ 18 nM) compared with wild-type SERCA2a (K i ϭ 280 nM) (Fig. 7C). In fact, the Ca 2ϩ affinity of D351A determined by this method (18 nM) is ϳ9-fold higher than the Ca 2ϩ affinity of wild-type SERCA2a estimated by the Ca 2ϩ -ATPase assay (0.16 M in Table 1). Assuming that N30C-PLB decreases the Ca 2ϩ affinity of D351A by ϳ2-fold (as it does for wild-type SERCA2a), the Ca 2ϩ affinity of D351A expressed alone is likely even higher than this, in the range of 10 nM. This remarkably high Ca 2ϩ affinity for D351A was first reported by MacIntosh et al. (24), but subsequently not confirmed (25) (see "Discussion"). Fig. 7A also points out the highly specific nature of the PLB to the SERCA2a cross-linking reaction, with PLB cross-linking exclusively to the Ca 2ϩ pump protein expressed in Sf21 membranes. Ca 2ϩ inhibition of N30C-PLB, PLB3, and PLB4 cross-linking to D351A was then measured (Fig. 7D). As predicted from results with wild-type SERCA2a, progressively higher concentrations of Ca 2ϩ were also required to dissociate N30C-PLB (K i ϭ 18 Ϯ 3 nM), PLB3 (K i ϭ 131 Ϯ 25 nM), and PLB4 (K i ϭ 234 Ϯ 23 nM) from D351A (means Ϯ S.E. from 4 determinations). Fig. 7D demonstrates unambiguously that the supershifting PLB mutants inhibit Ca 2ϩ binding to D351A.
Effects of TG and Nucleotides on PLB Cross-linking-To confirm the relative binding affinities of the PLB mutants for SERCA2a, and to gain additional insights on the specific conformation of the Ca 2ϩ pump that binds PLB, we determined the effects of TG and nucleotides on PLB cross-linking to SERCA2a, measured in the absence of Ca 2ϩ . It was shown pre-viously that N30C-PLB binds preferentially to the E2 state of SERCA2a stabilized by bound nucleotide (7,10), and that TG antagonizes formation of this state.
When measured in the absence of ATP, TG potently inhibited the cross-linking of all three PLB mutants to SERCA2a (Fig.  8, A and B). The K i values for TG inhibition of N30C-PLB, PLB3, and PLB4 cross-linking to SERCA2a were low and similar (0.07, 0.07, and 0.10 M, respectively) ( Table 2), and within the range of Ca 2ϩ pumps present within the reaction tubes (ϳ0.3 M). Thus, in the absence of nucleotide, under conditions favoring formation of E2 (absence of Ca 2ϩ ), TG binds virtually stoichiometrically to the Ca 2ϩ pump (18,31), whether co-expressed with N30C-PLB, PLB3, or PLB4. However, addition of 3 mM ATP to the assay tubes dramatically increased the concen- tration of TG required to inhibit PLB3 and PLB4 cross-linking to SERCA2a. The K i values (Fig. 8, A and C) increased from 0.07 to 4.9 M for PLB3, and from 0.10 to 7.8 M for PLB4, whereas for N30C-PLB, addition of ATP only increased the K i value from 0.07 to 0.23 M ( Table 2). That is a remarkable 70-(PLB3) and 78-fold (PLB4) decrease in TG binding affinity induced by ATP when supershifting PLB mutants are present. It should be pointed out that the concentration of PLB present in the reaction tubes was ϳ1.0 M, which is considerably lower than the concentration of TG required to significantly inhibit cross-linking of PLB3 and PLB4 to the Ca 2ϩ -ATPase in the presence of ATP (Fig. 8C). Thus the affinity of the two PLB supershifters for E2⅐ATP must be even greater than the affinity of TG for E2⅐ATP, which is within the nanomolar range or lower (18,31). The same results with PLB3 or PLB4 were obtained whether membranes were preincubated with TG for 5 or 60 min prior to initiation of the cross-linking reactions with KMUS, indicating that the supershifters prevent formation of a dead-end complex by TG (18) under these conditions. Like ATP, ADP also dramatically increased the K i value for TG inhibition of PLB cross-linking to the Ca 2ϩ -ATPase, whereas AMP had no significant effect (Fig. 9A demonstrated with PLB4). These results confirm previous findings that both ATP and ADP, but not AMP, stabilize the E2 state that favors PLB binding (7). We then measured the binding affinity of SERCA2a for ATP determined at different concentrations of TG (Fig. 9B). Successively higher concentrations of ATP were required to stimulate PLB4 cross-linking to the Ca 2ϩ -ATPase when the concentration of TG was increased. In the absence of TG, the affinity of SERCA2a for ATP was 9 M; in the presence of 6.4 M TG, the affinity of the enzyme for ATP was decreased 10-fold, to ϳ100 M (Table 3). These K ATP values for SERCA2a measured in the absence FIGURE 7. Ca 2؉ effect on PLB cross-linking to D351A. Ca 2ϩ inhibition of N30C-PLB cross-linking to wild-type SERCA2a (A) and D351A (B) was determined under Ca 2ϩ -ATPase assay conditions, as described under "Experimental Procedures." PLB/SER designates the PLB monomer cross-linked to the Ca 2ϩ pump at 110 kDa, and free PLB monomers (PLB 1 ) and dimers (PLB 2 ) are visible below at 6 and 12 kDa, respectively. The full autoradiographs are shown, demonstrating the highly specific cross-linking reaction; PLB cross-linked exclusively to expressed wild-type SERCA2a or D351A in Sf21 membranes. C, graph of Ca 2ϩ inhibition of N30C-PLB crosslinking to wild-type SERCA2a and D351A. D, Ca 2ϩ inhibition of N30C-PLB, PLB3, and PLB4 cross-linking to D351A. K i values for Ca 2ϩ inhibition of cross-linking to D351A are listed under "Results." of Ca 2ϩ agree well with those in previous reports, and confirm for SERCA2a that TG significantly reduces the affinity of the enzyme for ATP at the modulatory nucleotide-binding site (32)(33)(34)(35). Collectively, these results demonstrate that PLB binds to a single conformation of SERCA2a, E2 with bound nucleotide, and this state is distinct from the E2 conformation binding TG (see "Discussion").
Similar nucleotide effects on PLB binding to D351A were noted. The K i values for TG inhibition of cross-linking of all three PLB mutants to D351A were low and similar when assessed in the absence of ATP, but dramatically increased for PLB3 and PLB4 when ATP was included ( Table 2). TG also substantially decreased the ATP binding affinity of Asp 351 (Fig.   9C and Table 3). Interestingly, the affinity of D351A for ATP was only about 2-fold greater than the affinity of wild-type SERCA2a for ATP (Table 3), which is substantially lower than the ATP binding affinity of D351A made from SERCA1a (24, 25) (see "Discussion").
Vanadate Effects on PLB Crosslinking-According to results above with TG, the binding affinities of N30C-PLB and the supershifters for the E2 state of SERCA2a are much higher than previously predicted (36). Therefore, to confirm these surprising results, we used a second lower affinity Ca 2ϩ pump inhibitor, vanadate, to estimate the binding affinities of the PLB mutants for SERCA2a. Vanadate inhibits the Ca 2ϩ -ATPase with micromolar affinity, and like TG is proposed to bind preferentially to the nucleotide-free, E2 conformation of the Ca 2ϩ pump (19). Fig. 10 shows that in the absence of Ca 2ϩ and ATP, vanadate inhibited cross-linking of all three PLB mutants to SERCA2a. However, significantly higher concentrations of vanadate were required to inhibit PLB3 (K i ϭ 46 M) and PLB4 (K i ϭ 380 M) cross-linking to SERCA2a, relative to N30C-PLB (K i ϭ 1.6 M, Fig. 10B). Moreover, maximal cross-linking of PLB3 and PLB4 to SERCA2a could only be inhibited by 80 and 60% at 1 mM vanadate, the highest concentration tested. When 36 M ATP was included in the buffer, the K i for vanadate inhibition of N30C-PLB cross-linking to SERCA2a was increased 125-fold, from 1.6 (no nucleotide) to 200 M vanadate (36 M ATP), and cross-linking of PLB3 and PLB4 to SERCA2a became nearly completely vanadate resistant (Fig. 10C). At 3 mM ATP, vanadate failed to inhibit cross-linking of any PLB mutant to the Ca 2ϩ -ATPase (data not shown). Thus, results with vanadate also show that PLB binds with surprisingly high affinity to the E2 state of SERCA2a when nucleotide is present.

DISCUSSION
In the present study, chemical cross-linking was used to monitor protein-protein interactions between the cardiac Ca 2ϩ pump and PLB mutants of increasing inhibitory strength to investigate the physical basis of enzyme inhibition by PLB. Using these cross-linkable PLB mutants, new insights were gained on the effect of PLB on the Ca 2ϩ affinity and V max of the enzyme, the binding affinity of PLB for the Ca 2ϩ pump, and the specific conformation of SERCA2a required for PLB binding.
Effect of PLB on Ca 2ϩ Binding Affinity and V max -The hallmark of PLB regulation of SERCA2a is its ability to decrease the apparent Ca 2ϩ affinity of the Ca 2ϩ -ATPase, while having little or no effect on the V max of the enzyme measured at saturating  Ca 2ϩ concentration (1, 2). However, whether PLB increases the K Ca of Ca 2ϩ -ATPase activation by decreasing the actual Ca 2ϩ binding affinity of the enzyme (7,12,22,37), or by affecting one or more catalytic steps in the reaction cycle (23, 26) has remained unclear. Here we addressed this question directly, using cross-linkable PLB mutants of increasing inhibitory potency (PLB4 Ͼ PLB3 Ͼ N30C-PLB). We showed that successively higher Ca 2ϩ concentrations were required to both activate the enzyme co-expressed with N30C-PLB, PLB3, and PLB4 and to dissociate N30C-PLB, PLB3, and PLB4 from the Ca 2ϩ pump. Moreover, there was a direct correlation between the degree of PLB binding to SERCA2a and the extent of PLB inhibition of Ca 2ϩ -ATPase activity at all Ca 2ϩ concentrations tested with all three PLB mutants (Fig. 4C). These results strongly suggest that PLB competes with Ca 2ϩ for binding to the Ca 2ϩ -ATPase, and that SERCA2a with PLB bound is catalytically inactive. Competition between PLB and Ca 2ϩ for binding to SERCA2a was confirmed using the Ca 2ϩ pump mutant D351A, which retains Ca 2ϩ binding, but cannot hydrolyze ATP. Progressively higher Ca 2ϩ concentrations were also required to dissociate the increasingly potent PLB mutants from D351A. Thus at each Ca 2ϩ concentration tested, progressively more E2⅐PLB was formed by the increasingly inhibitory PLB mutants, meaning that less E1 was available for Ca 2ϩ binding (Fig. 7D). Therefore, by stabilizing the enzyme in a Ca 2ϩ -free state, PLB decreases Ca 2ϩ binding to the pump and alters the kinetics of enzyme activation by Ca 2ϩ .
We also confirmed that PLB molecules of normal inhibitory strength (N30C-PLB) do not significantly affect the V max of the Ca 2ϩ -ATPase (1). This is contrary to conclusions of several recent studies in which PLB was reported to either decrease (22) or increase (20,21,23) the V max of the Ca 2ϩ -ATPase. Using our viral constructs and the 2D12 antibody, Waggoner et al. (22) recently noted a modest reduction (ϳ20%) in the V max of SERCA2a co-expressed with wild-type PLB compared with SERCA2a expressed alone. This is in disagreement with an earlier study in which no effect on V max was noted (37). We believe that the modest reduction in V max observed by Waggoner et al. (22) is more apparent than real. Fig. 4A points out that when Ca 2ϩ -ATPase activities are carefully corrected for Ca 2ϩ pump expression levels, there is little or no inhibition of the enzyme at saturating Ca 2ϩ concentrations when SERCA2a is co-expressed with PLB mutants of normal inhibitory potency (N30C-PLB). Moreover, this relief of Ca 2ϩ -ATPase inhibition at saturating Ca 2ϩ concentration is entirely FIGURE 9. Nucleotide effect on PLB4 cross-linking to wild-type SERCA2a and D351A. A, effect of 3 mM AMP, ADP, ATP, and no added nucleotide (Con) on PLB4 cross-linking to wild-type SERCA2a. TG concentrations were varied as indicated. B, ATP stimulation of PLB4 cross-linking to wild-type SERCA2a, determined at different TG concentrations. ATP concentrations were varied as indicated. C, ATP stimulation of PLB4 cross-linking to D351A, determined at different TG concentrations. consistent with the complete dissociation of N30C-PLB from SERCA2a observed at 1-2 M Ca 2ϩ by chemical cross-linking (Fig. 4B). The studies in which wild-type PLB and some other PLB mutants were reported to actually increase the V max of the Ca 2ϩ -ATPase were all conducted with the purified rabbit skeletal muscle enzyme co-reconstituted with purified PLB from detergent solution (20,21,23). In this case, it is possible that enzyme protection by PLB during the reconstitution process may have artifactually affected the results, as was recently suggested (23). Regardless, in multiple studies using cellular expression systems, no increase in V max by PLB has been noted (1,2). The results shown here illustrate how PLB is perfectly poised to regulate cardiac contractile kinetics in intact myocardium (Fig. 4). Ca 2ϩ concentrations within the cardiomyocyte range from nanomolar to 1-2 M (38), and the affinity of PLB for SERCA2a allows it to associate and dissociate from the enzyme over the same Ca 2ϩ concentration range at which contractile force develops. At low cytosolic Ca 2ϩ concentrations at which myofilament contractile force is low, the affinity of PLB for the Ca 2ϩ pump is high and enzyme inhibition by PLB is substantial, but still reversible by phosphorylation by protein kinase A (1,2) or by the 2D12 antibody (39). At high Ca 2ϩ concentrations yielding peak contractile force, PLB is completely dissociated from the Ca 2ϩ pump and the enzyme is maximally active. However, for the supershifting PLB mutants the situation is different. By virtue of their very high binding affinities for SERCA2a, the supershifters remain significantly bound to the Ca 2ϩ pump and continue to inhibit the enzyme at Ca 2ϩ concentrations that are normally saturating. At Ca 2ϩ concentrations high enough to dissociate these potent PLB molecules from the Ca 2ϩ pump (10 -200 M Ca 2ϩ ) (Fig. 4), significant back inhibition of the enzyme occurs. Thus, maximal Ca 2ϩ -ATPase activity can never be realized when SERCA2a is co-expressed with potent PLB supershifters, even after phosphorylation of PLB by protein kinase A or after addition of the 2D12 antibody. This may explain why transgenic mice overexpressing the most potent PLB supershifters develop heart failure and premature death (16).
D351A-Using PLB as a reporter molecule, we were able to estimate the Ca 2ϩ and nucleotide binding affinities of D351A relative to wildtype SERCA2a, and make comparisons with previous determinations made for the skeletal muscle enzyme (SERCA1a). In an earlier study, MacIntosh et al. (24) used 8-N 3 -TNP-ATP photolabeling to measure the Ca 2ϩ and ATP binding affinities of D351A (rabbit skeletal isoform) expressed in COS membranes. The authors found that relative to wild-type Ca 2ϩ -ATPase, D351A had an extraordinarily high affinity for both Ca 2ϩ (Ͼ10-fold increase) and ATP (20 -100-fold increase) (24). They postulated that Ala substitution at Asp 351 significantly increases the ATP affinity of the Ca 2ϩ -ATPase by relieving electrostatic repulsion between the ␥-phosphate of ATP and Asp 351 of the wild-type enzyme. Moreover, they proposed that mutationally induced conformational changes at the site of ATP binding within the cytoplasmic head group were transmitted to the Ca 2ϩ binding sites located at the membrane, substantially increasing the Ca 2ϩ affinity of the enzyme. The very high ATP affinity of D351A, but not the high Ca 2ϩ affinity, was confirmed in a subsequent study by Marchand et al. (25), also with SERCA1a. In this later report, ATP and Ca 2ϩ binding affinities were determined for the purified enzyme in detergent solution.
Here, using PLB cross-linking to estimate Ca 2ϩ affinity, we also noted an extremely high Ca 2ϩ affinity for D351A, this time using the cardiac muscle isoform (SERCA2a). Our results indicate that D351A has a Ca 2ϩ affinity at least 10 times higher than wild-type SERCA2a (Fig. 7A). This result is consistent with the earlier findings of MacIntosh et al. (24), but inconsistent with the results of Marchand et al. (25). It is well known that nonionic detergents like C 12 E 8 and dodecylmaltoside substantially decrease the Ca 2ϩ binding affinity of SERCA pumps (40,41), which may explain the failure of Marchand et al. (25) to detect an increase in Ca 2ϩ affinity for D351A.
Regarding ATP affinity, we determined a K d value of 9 M for the wild-type enzyme, which is well within the range reported by other investigators for ATP binding at the low-affinity modulatory binding site of E2 measured in the absence of Ca 2ϩ   [32][33][34][35]. For D351A, we noted a modest 2.3-fold increase in ATP affinity relative to wild-type SERCA2a (K d ϭ 4.0 M), in contrast to the two studies above that reported a much higher nucleotide binding affinity for D351A measured under similar conditions (24,25). However, our results appear to be consistent with the recently determined crystal structure of the E2(TG)⅐AMPPCP complex, representing E2 with ATP bound at the modulatory site (34). According to this structure, ATP fits more loosely into the modulatory site (ATP binding site in E2) relative to the catalytic site. When ATP is bound to E2, the ␥-phosphate is 9 Å away from the phosphorylation site, making the electrostatic repulsion between the ␥-phosphate and the negatively charged Asp 351 much less pronounced than what occurs when ATP is bound to E1 (34). Thus ATP affinity at the modulatory site may be less affected by the D351A mutation because the ␥-phosphate of ATP does not interact closely with Asp 351 when ATP is bound here. Nevertheless, our results with D351A demonstrate that there is long-range communication between the catalytic site and the Ca 2ϩ binding sites, and removal of the negative charge at Asp 351 strikingly enhances the Ca 2ϩ binding affinity at the two Ca 2ϩ binding sites in the membrane.
E2⅐ATP Conformation-Early studies showed that in the absence of Ca 2ϩ , the intrinsic tryptophan fluorescence of SERCA was substantially increased by the nucleotides, ATP and ADP, but was unaffected by AMP (32). This nucleotideinduced increase in fluorescence intensity was completely inhibited by TG, which was subsequently shown to reduce the affinity of the Ca 2ϩ -ATPase for ATP through uncompetitive inhibition (33)(34)(35). Similarly, PLB cross-linking to SERCA2a occurs in the absence of Ca 2ϩ , is enhanced by ATP and ADP, but inhibited by TG (7,11). Based upon these similarities, it was suggested that the physiological state detected by changes in fluorescence induced by nucleotide binding to E2 (32-35) is the unique E2⅐ATP state that binds PLB (7).
In a recent study by Jensen et al. (34), it was suggested that TG stabilizes the fully protonated H n E2 state of the Ca 2ϩ pump, and that ATP binding at the modulatory site stimulates deprotonation of E2, initiating the transition to E1. Here, using ATP stimulation of PLB cross-linking to measure ATP binding at different TG concentrations, we confirmed the ATP affinities of E2 and E2⅐TG reported previously (32)(33)(34)(35). Moreover, we showed that ATP dramatically increases the resistance of the E2⅐PLB complex to TG, shifting the K i values for TG inhibition of cross-linking by 100 -200-fold for the supershifters PLB3 and PLB4 (Table 2). Thus, the PLB supershifters and ATP interact synergistically at E2, stabilizing an E2⅐ATP⅐PLB ternary complex that is remarkably resistant to TG. These results suggest that the E2⅐ATP state detected by ATP-induced changes in Trp fluorescence (32)(33)(34)(35) and by chemical cross-linking of PLB (7,11), may be the deprotonated E2⅐ATP state with ATP bound at the modulatory site. Moreover, TG may inhibit formation of this specific conformation, not by blocking ATP binding, but by hindering ATP-stimulated deprotonation of the enzyme. Consistent with this interpretation, PLB does not bind to the P i (17) or vanadate (Fig. 10) bound forms of the enzyme, both of which interact with the protonated H n E2 state like TG (42). Also as observed with TG, ATP strongly enhances PLB cross-linking to SERCA2a in the presence of P i (17) and vanadate (Fig. 10), being able to compete for P i (43) or vanadate binding (19,44) to the Ca 2ϩ -ATPase.
Given the reputation of TG as an extremely potent, irreversible inhibitor of the Ca 2ϩ -ATPase (18,31), we were surprised to discover that TG did not disrupt the ternary complex between the PLB supershifters and E2⅐ATP, even when membranes were preincubated for up to 1 h with greater than stoichiometric concentrations of TG. Moreover, under these conditions favoring E2⅐ATP, PLB3 and PLB4 bind even more tightly to SERCA2a than does TG, the highest affinity SERCA inhibitor identified to date (18,31). Crystallographic studies have revealed that TG binds to E2 in a cavity formed between transmembrane helices M3, M5, and M7, near the cytoplasmic membrane surface (45). This is on the opposite face from the PLB binding site, which is predicted to extend along the groove formed between transmembrane helices M2, M4, and M9, based on cross-linking results (7,17,36). Our results suggest that binding of PLB at its site must drastically distort the TG binding pocket. Nonetheless, under enzyme turnover conditions, TG is the more powerful SERCA inhibitor. In the presence of Ca 2ϩ , the catalytic activity is completely inhibited by TG through formation of a dead-end complex (18), whereas Ca 2ϩ -ATPase inhibition by the PLB supershifters remains reversible, albeit at very high Ca 2ϩ concentrations (Fig. 4). It should be pointed out that TG binding to E1⅐Ca 2ϩ as well as to E2 has been noted in many studies (18, 34, 46 -48), and that the ability of TG to bind to different conformational states of SERCA may contribute to its apparently irreversible effect on Ca 2ϩ -ATPase activity.
The overall conclusion of this work is that PLB inhibits Ca 2ϩ binding to SERCA2a by stabilizing the enzyme in a Ca 2ϩ -free E2 state. Clearly, PLB binding has long-range conformational effects on both the cytoplasmic domains and the transmembrane domain, and these effects may be even more profound with the PLB supershifters. ATP binding to E2 accelerates the E2 to E1⅐Ca 2 transition by stimulating H ϩ /Ca 2ϩ cation exchange (34), while at the same time inducing structural changes that promote PLB binding. So, is the conformation of SERCA2a that binds to PLB really deprotonated E2⅐ATP, or Ca 2ϩ -free E1 (35), or perhaps something in between (47)? Until the crystal structure of PLB-bound SERCA2a is determined we have no way of knowing. It was recently suggested that TG "rigidifies" the transmembrane domain of the Ca 2ϩ pump, making it unresponsive to conformational changes occurring within the cytosolic domain (35). It is this ability of TG to fix the transmembrane helices that has enabled the Ca 2ϩ -free, TGbound enzyme to be crystallized, providing valuable structural information about the Ca 2ϩ -ATPase in different E2 states (45). However, all of the E2 structures determined to date have been in the presence of irreversible inhibitors like TG or cyclopiazonic acid (49), and it is unclear how closely these inhibitorbound structures resemble other, perhaps more physiological states of the enzyme (35). It is therefore our long-term goal to crystallize the Ca 2ϩ pump complexed with PLB to provide a structure of E2 stabilized by a reversible inhibitor that is physiologically active in the heart. Here, we have shown that in the absence of Ca 2ϩ , the binding affinities of the supershifters are severalfold higher than even TG, making the goal of crystallizing the Ca 2ϩ -free enzyme stabilized by PLB3 or PLB4 seem plausible.