Phospholamban Binds with Differential Affinity to Calcium Pump Conformers*

To investigate the mechanism of regulation of sarco-endoplasmic reticulum Ca2+-ATPase (SERCA) by phospholamban (PLB), we expressed Cerulean-SERCA and yellow fluorescent protein (YFP)-PLB in adult rabbit ventricular myocytes using adenovirus vectors. SERCA and PLB were localized in the sarcoplasmic reticulum and were mobile over multiple sarcomeres on a timescale of tens of seconds. We also observed robust fluorescence resonance energy transfer (FRET) from Cerulean-SERCA to YFP-PLB. Electrical pacing of cardiac myocytes elicited cytoplasmic Ca2+ elevations, but these increases in Ca2+ produced only modest changes in SERCA-PLB FRET. The data suggest that the regulatory complex is not disrupted by elevations of cytosolic calcium during cardiac contraction (systole). This conclusion was also supported by parallel experiments in heterologous cells, which showed that FRET was reduced but not abolished by calcium. Thapsigargin also elicited a small decrease in PLB-SERCA binding affinity. We propose that PLB is not displaced from SERCA by high calcium during systole, and relief of functional inhibition does not require dissociation of the regulatory complex. The observed modest reduction in the affinity of the PLB-SERCA complex with Ca2+ or thapsigargin suggests that the binding interface is altered by SERCA conformational changes. The results are consistent with multiple modes of PLB binding or alternative binding sites.

The sarco-endoplasmic reticulum Ca 2ϩ -ATPase (SERCA) 3 is a P-type ion pump that maintains the Ca 2ϩ gradient across the endoplasmic reticulum. In cardiac muscle cells, calcium (Ca 2ϩ ) sequestration by SERCA is critical for muscle relaxation during the cardiac cycle, and disordered Ca 2ϩ handling is associated with cardiac dysfunction (1)(2)(3). SERCA activity is regulated by phospholamban (PLB), a helical transmembrane peptide that reduces the apparent affinity of the pump for Ca 2ϩ . Inhibition of SERCA by PLB is partially relieved through phosphorylation of PLB by protein kinase A and Ca 2ϩ /calmodulindependent protein kinase (4), which alters the structure of the PLB-SERCA regulatory complex (5) and increases PLB oligo-merization into non-inhibitory pentamers (5,6). It is widely recognized that PLB inhibition of SERCA is also relieved by elevated Ca 2ϩ , but the mechanism of this functional effect is unclear. One possibility is that PLB binds selectively to the Ca 2ϩ -free "E2" conformation of SERCA and cannot bind to the Ca 2ϩ -bound "E1" conformation ( Fig. 1A, Dissociation Model). This model is supported by the observation that elevated Ca 2ϩ abolishes chemical cross-linking of the PLB transmembrane domain to reactive residues on SERCA (7)(8)(9)(10)(11) and reduces coimmunoprecipitation (12). Cross-linking was also prevented by the SERCA inhibitor thapsigargin (Tg) (7, 9 -11). Another recent study showed that oligomerization of a PLB-SERCA fusion construct was increased by micromolar Ca 2ϩ (13), consistent with the idea that Ca 2ϩ causes displacement of PLB from the inhibitory cleft, permitting PLB self-association into pentamers. Overall, these data suggest that only certain conformational substates of SERCA interact with PLB. The dissociation model predicts that PLB unbinds from SERCA during the period of systole (cardiac contraction) when Ca 2ϩ is high, and the regulatory complex reforms during diastole (cardiac relaxation) when the cytoplasmic Ca 2ϩ concentration is low.
An alternative theory was generated from in vitro measurements of fluorescence resonance energy transfer (FRET) between SERCA and PLB. Mueller et al. (14) showed that FRET was decreased, but not abolished, by Ca 2ϩ . This result suggested that relief of inhibition was accomplished by a conformational change of the regulatory complex, rather than unbinding of PLB from the pump. This study estimated a dissociation constant significantly lower than the expected in vivo concentrations of PLB and SERCA. Li et al. (15) also provided evidence from fluorescence spectroscopy that suggests that the binding of PLB and Ca 2ϩ is not mutually exclusive. The investigators suggested that the regulator PLB may be considered a subunit of the pump and remains bound to SERCA throughout the catalytic cycle (Fig. 1B, Subunit Model).
To test these alternative mechanisms, we expressed SERCA and PLB fused to fluorescent proteins to quantify the regulatory interaction by FRET in live cells. We have previously used this approach to measure the relative affinity of PLB and SERCA variants in AAV-293 cells (5, 16 -18). In the present study, we also expressed fluorescently labeled SERCA and PLB in adult ventricular myocytes to determine whether the regulatory complex is dynamically dissociated during the cardiac cycle.
Cer (17). PLB interacts similarly with SERCA1a (skeletal muscle isoform) and SERCA2a (cardiac isoform) (20,21), but with a slightly higher affinity for SERCA1a (17). Adenoviral vectors of canine Cer-SERCA2a and canine YFP-PLB were produced using the AdEasy system (Stratagene, La Jolla, CA). Cardiac ventricular myocytes were isolated from adult New Zealand White rabbits (22). All protocols were approved by the Loyola University Institutional Animal Care and Use Committee. Myocytes were transferred to culture vessels and washed with fresh PC-1 medium (Lonza, Basel, Switzerland). YFP-PLB and Cer-SERCA adenoviruses were added at a multiplicity of infection of 100. Cells were paced for 48 h in culture using a C-Pace EP pacer (IonOptix, Milton, MA) set to 10 volts with a frequency of 0.1 Hz and 5-ms pulse duration.
AAV-293 cells were cultured and transiently transfected using the MBS mammalian transfection kit (Stratagene) as described previously (19). Cells were co-transfected with plasmids encoding YFP-PLB and Cer-SERCA with a molar ratio of 20:1 and co-transfected with Cer-PLB and YFP-PLB with a molar ratio of 5:1. Transfected cells were trypsinized and replated onto poly-D-lysine-coated glass bottom dishes and allowed to attach for 2-3 h prior to imaging.
Live Cell Ca 2ϩ Uptake Activity Assay-To evaluate Ca 2ϩ transport activity by fluorescently labeled SERCA, we performed a live cell Ca 2ϩ uptake assay. Heterogeneously transfected populations of AAV-293 cells were labeled with the cellpermeant Ca 2ϩ indicator dye X-rhod-1 (AM) (Invitrogen). Transfected and untransfected cells were distinguished on the basis of the intensity of YFP fluorescence emission. Release of Ca 2ϩ from intracellular stores was accomplished by stimulating purinergic receptors with extracellular application of 2 mM ATP. Accumulation of Ca 2ϩ in the cytosol was quantified as an increase in X-rhod-1 fluorescence and was the net result of Ca 2ϩ release counterbalanced by Ca 2ϩ extrusion and uptake processes (including SERCA activity). Exogenous SERCA activity was detected as a decrease in ATP-stimulated cytosolic Ca 2ϩ accumulation relative to untransfected cells in the same microscopic field. 4 min after application of extracellular ATP, cells were treated with 10 M Tg to determine the size of the Ca 2ϩ store remaining in the ER.
Confocal Imaging-Cardiac myocytes were used 48 h after adenoviral infection. To visualize the sarcolemma and t-tubule system, cells were stained with FM 4-64 (Invitrogen) according to the manufacturer's instructions. Confocal imaging was performed using an inverted Leica TCS SP5 confocal microscope with a 63ϫ water immersion objective and an 8000-Hz resonant scanner. For localization experiments, Cer, YFP, and FM 4-64 were sequentially excited using an argon laser with wavelengths of 458, 514, and 543 nm, respectively. As an index of FRET, the YFP/Cer ratio was obtained with 458 nm excitation and simultaneous detection of YFP (525-560 nm) and Cer (470 -515 nm). Cells were incubated with 5 M X-rhod-1 AM Ca 2ϩ indicator dye (Invitrogen) and imaged with 543 nm HeNe laser excitation and emission at 570 -650 nm. Cells were perfused with PC-1 medium. Experiments were conducted in linescan (x-t) mode to study individual Ca 2ϩ transients or in imaging mode (x-y-t) to study alternating intervals of pacing and rest. For the latter, mean cytosolic Ca 2ϩ and YFP/Cer ratio were quantified by averaging the signal from the final 20 s of a 60-s interval of 1-Hz pacing, and these measurements were normalized to the pre-paced (rest) value. Comparisons of pacing intervals with rest were made using a one-sample Student's t test, where p Ͻ 0.05 was considered significant.
E-FRET Quantification and High Throughput Cell Scoring-Wide-field fluorescence microscopy was performed as described previously (18). For each sample, automated acquisition of a field of 48 images was performed using a motorized stage (Prior, Rockland, MA) controlled by the MetaMorph software. Focus was maintained by an optical feedback system (Perfect Focus System, Nikon). Images were obtained with a 40ϫ 0.75 N.A. objective with 100-ms exposure for each channel: Cer, YFP, and FRET (Cer excitation/YFP emission). Fluorescence intensity was automatically quantified with a multiwavelength cell scoring application in MetaMorph. Cell selection criteria included having a diameter of 35-75 pixels and having an average intensity of 100 counts above background. The data were transferred automatically to a spreadsheet for analysis. FRET efficiency was calculated according to and F Cer are the matching fluorescence intensities from FRET, YFP, and Cer images, respectively, and G represents FRET intensity corrected for the bleed-through of the channels. The factors a and d are bleed-through constants calculated as a ϭ F FRET /F YFP for a control sample transfected with only YFP-PLB and d ϭ F FRET /F Cer for a control sample with only Cer-SERCA. For our experimental setup, a and d were 0.082 and 0.82, respectively. To estimate the contribution of FRET between non-interacting proteins (nonspecific FRET), we have previously used non-fluorescent PLB to compete for binding. This abolished specific FRET, leaving 4% residual nonspecific FRET (18). In another study, we used a fluorescently labeled PLB that could not participate in FRET as a competitor. We determined a maximal residual nonspecific FRET of 5% (16) under similar conditions to those used here. Another group has also shown that FRET from SERCA to phospholamban can be reduced by competition with unlabeled phospholamban (24).
To independently quantify parameters related to structure and binding affinity, we measured FRET from a heterogeneous FIGURE 1. Alternative models of PLB regulation of SERCA. A, the dissociation model (7,9,10). PLB dissociates from SERCA when the pump assumes the E1 (Ca 2ϩ -bound) conformation during systole. B, the subunit model (14). PLB remains bound throughout the catalytic cycle as a subunit of the pump.
population of cells, as described previously (5,16,17,18,25,26). Observed FRET was compared cell-by-cell to YFP fluorescence, which was taken as an index of protein concentration (5, 16 -18, 26). The concentration dependence of FRET was initially fit by a hyperbolic function of the form FRET ϭ (FRET max ϫ [protein])/(K D 2ϩ[protein]) with all parameters independently fit. Because the mean FRET max at low and high concentrations of Ca 2ϩ or Tg was not significantly different when fit independently, this parameter was shared for subsequent global fit analysis. In this way, a single best FRET max value was obtained, with independent K D values for each condition. The dependence of K D 2 on the concentration of Ca 2ϩ or Tg was analyzed by fitting a Hill function of the form where n is the Hill coefficient and [x] is the concentration of Ca 2ϩ or Tg. Dilutions of Tg in PBS were added to the culture chambers and incubated at room temperature for ϳ30 min. prior to data collection. For Ca 2ϩ experiments, cells were permeabilized with 10 g/ml saponin for 1 min and then washed in a bath solution of composition 100 mM KCl, 5 mM NaCl, 2 mM MgCl 2 , 20 mM imidazole, 5 mM EGTA, and 2 mM ATP (pH 7.0), with varying free Ca 2ϩ . Ca 2ϩ concentration was calculated with Maxchelator (28) and validated with a Ca 2ϩ -sensitive electrode (Thermo Scientific).

Quantification of FRET from SERCA to PLB in Cardiac
Myocytes-Confocal imaging of Cer-SERCA expressed in adult rabbit ventricular myocytes showed a striated pattern of fluorescence as well as longitudinal streaks ( Fig. 2A), suggesting localization in the junctional and transverse sarcoplasmic reticulum, respectively. Counterstaining with the membrane dye FM 4-64 demonstrated that the t-tubule system was largely intact after 2 days of culture with pacing and showed that the registration of the observed striation pattern of PLB and SERCA was at the Z-lines of the sarcomeres. Fig. 2B is an overlay image showing a high degree of colocalization of YFP-PLB and Cer-SERCA, with an additional fraction of PLB visible at the sarcolemmal membrane (Fig. 2B, inset, arrow). Strong fluorescence was also visible in the perinuclear regions (Fig. 2B). We assessed PLB and SERCA diffusion by measuring fluorescence recovery after photobleaching. At the target region (Fig.  2C, arrow), the fluorescence of both proteins recovered over the course of several minutes, consistent with long range diffusion over many microns and across multiple sarcomeres. The data suggest that neither PLB nor SERCA is immobilized in the sarcoplasmic reticulum membrane. FRET between Cer-SERCA and YFP-PLB was quantified in these myocytes with the E-FRET method (17,29). FRET increased with protein concentration toward a maximum level (Fig. 2D), consistent with saturable binding of PLB to SERCA (18). This maximum (FRETmax ) was similar to the value previously measured in heterologous cells (5, 16 -18).
To determine whether the regulatory complex was dynamically responsive to changes in Ca 2ϩ , myocytes expressing fluorescent PLB and SERCA were stimulated with electrical pacing to elicit Ca 2ϩ transients. Fig. 3A shows the average of 21 consecutive Ca 2ϩ transients obtained by line-scan mode measurements of the indicator X-rhod-1 (black trace) normalized to diastolic Ca 2ϩ . We obtained simultaneous measurements of YFP and Cer fluorescence with 458 nm excitation. We did not detect a systole-to-diastole change in the normalized YFP/Cer ratio (Fig. 3A, red trace), suggesting that FRET from Cer-SERCA to YFP-PLB is not abolished during systole. The ␤-ad- renergic agonist isoproterenol (5 M) increased the amplitude and decreased the duration of the Ca 2ϩ transient (Fig. 3B), but FRET was still unaffected. To test whether FRET would be abolished by a larger prolonged elevation of cytosolic Ca 2ϩ , we alternated intervals of rest (no stimulation) with intervals of rapid (1-Hz) pacing (Fig. 3C). We observed a modest reduction in the normalized YFP/Cer ratio with rapid pacing apparent in the absence (Fig. 3C) or presence (Fig. 3D) of 5 M isoproterenol. Fig. 3E shows the mean cytosolic Ca 2ϩ elevation for control (n ϭ 6 cells) and after the addition of isoproterenol (n ϭ 4 cells) normalized to resting Ca 2ϩ , and Fig. 3F shows the corresponding mean YFP/Cer ratio. During pacing intervals, the mean cytosolic Ca 2ϩ was significantly elevated when compared with rest, both before (p Ͻ 0.01) and after (p Ͻ 0.01) isoproterenol. FRET was also significantly decreased when compared with rest, both before (p Ͻ 0.05) and after (p Ͻ 0.01) isoproterenol. Overall, the data are consistent with a small decrease in FRET during intervals of rapid pacing.
Quantification of FRET from SERCA to PLB in Heterologous Cells-We also quantified FRET from a large population of AAV-293 cells expressing a wide range of protein concentration (5, 16 -18). For better control of Ca 2ϩ over a wider range of concentrations, the cells were transiently transfected with Cer-SERCA and YFP-PLB and permeabilized with saponin in Ca 2ϩ / EGTA-buffered solutions. The data were similar to those obtained from ventricular myocytes (Fig. 2C) and were pooled for clarity in Fig. 4. The dependence of FRET on protein concentration at low Ca 2ϩ concentrations (100 pM) was well described by a hyperbolic function of the form FRET ϭ (FRETmax ϫ [protein])/(K D 2ϩ[protein]) (Fig. 4A, black curve). We did not observe a significant difference in FRET max after adding saturating concentrations of Ca 2ϩ (1 mM) (Fig. 4A, red curve), suggesting that the average structures of the Ca 2ϩ -free (E2) and Ca 2ϩ -bound (E1) regulatory complexes are similar. The hyperbolic fit also provided an estimate of the apparent dissociation constant (K D 2), which is the protein concentration that yields half-maximal FRET. K D 2 was modestly increased by saturating . We detected no change in PLB-SERCA FRET in response to beat-to-beat Ca 2ϩ elevations, before or after isoproterenol. C, prolonged elevations of Ca 2ϩ (black) were achieved with intervals of rapid (1-Hz) pacing after periods of rest (no stimulation). A modest decrease in FRET (red) was observed during rapid pacing. The blue lines represent the Ca 2ϩ and FRET traces smoothed by 5-s adjacent averaging. D, as in C, after 5 M isoproterenol, a small decrease in FRET (red) was observed during rapid pacing. E, averaging multiple experiments showed that rapid pacing increased mean [Ca 2ϩ ], with a larger increase observed after the addition of 5 M isoproterenol (iso). F, averaging multiple experiments showed that rapid pacing decreased the mean YFP/Cer ratio, and this decrease was larger after the addition of 5 M isoproterenol. The first, second, and third pacing intervals are marked a, b, and c in C-F. Ca 2ϩ , consistent with a small decrease in the binding affinity of SERCA for PLB. Fig. 4B demonstrates the concentration dependence of the increase in K D 2 from 100 pM to 1 mM Ca 2ϩ . K D 2 increased 41% with Ca 2ϩ , with an EC 50 of 410 nM and a Hill coefficient of 1.1. Data in Fig. 4B are mean Ϯ S.E. for five independent experiments, using an average of 5300 cells per experiment.
We also investigated the effect of SERCA inhibitor Tg on SERCA-PLB FRET because this ligand is reported to abolish cross-linking of the regulatory complex (10,30,31). When compared with Ca 2ϩ , the effect of Tg on K D 2 was similar in magnitude and direction. We observed no change in FRET max (Fig. 4C). Fig. 4D shows that K D 2 increased 78% with Tg, with an apparent EC 50 of 350 nM and a Hill coefficient of 0.7. Data in Fig.  4D are mean Ϯ S.E. for four independent experiments, using an average of 3900 cells per experiment. To verify that the effect of Ca 2ϩ and Tg on PLB-SERCA FRET was not an indirect result of altered PLB oligomerization, we measured the dependence of FRET from Cer-PLB to YFP-PLB. As observed previously (5, 16 -18), FRET increased with protein concentration toward a maximum (Fig. 4E, black curve). There was no effect on PLB-PLB binding with either Ca 2ϩ (Fig. 4E) or Tg (Fig. 4F). Data in Fig. 4, E and F, were pooled from a set of 470 cells. The data suggest that the observed change in apparent K D 2 (PLB-SERCA) was not an indirect effect of a change in K D 1 (PLB-PLB) (5,6,16,18,26,32) or nonspecific effects on the fluorescent proteins or the membrane structure.
Measurement of Transport and Regulatory Activity of Fluorescent Protein Fusion Constructs-To determine whether fluorescently labeled SERCA and PLB were functional, the constructs were expressed in live AAV-293 cells, and the cells were loaded with Ca 2ϩ indicator X-rhod-1 (Fig. 5A). Inositol 1,4,5trisphosphate-mediated Ca 2ϩ release was elicited by activation of purinergic receptors with application of 100 M ATP in the bath solution (Fig. 5B). In untransfected cells (Fig. 5A, UT), accumulation of Ca 2ϩ in the cytosol was detected as a large increase in X-rhod-1 fluorescence (Fig. 5B, black points). This fluorescence transient was significantly suppressed in cells expressing Cer-SERCA (Fig. 5B, red points). These results suggest that exogenous SERCA expression significantly increases Ca 2ϩ uptake and prevents cytosolic Ca 2ϩ accumulation. Exogenous Cer-SERCA also increased Tg-releasable ER Ca 2ϩ content (Fig. 5B) when compared with untransfected controls. Coexpression of YFP-PLB partially restored the ATP-stimulated Ca 2ϩ transient (Fig. 5B, blue points) and reduced the Tg-releasable ER Ca 2ϩ content, suggesting that inhibition of Cer-SERCA by PLB was intact. Cells expressing YFP-PLB alone (Fig. 5B, green points) showed a smaller ATP-induced Ca 2ϩ transient and a smaller ER Ca 2ϩ load when compared with untransfected cells, indicating that exogenous YFP-PLB regulates endogenous SERCA. Overall, the data suggest that the Cer and YFP fusion tags were benign with respect to catalytic and regulatory function.

DISCUSSION
Contrasting models have emerged to describe the mechanism of SERCA regulation by PLB. A key point of divergence of these models is whether the PLB dissociates from SERCA during systole (Fig. 1A) or remains bound as a subunit of the pump (Fig. 1B). The dynamics of this interaction are of interest because the proteins play a central role in cardiac Ca 2ϩ handling, and they are regarded as high value targets for prospective treatments of cardiac disease. The goal of this study was to directly test these prevailing models and quantify the PLB-SERCA interaction in the context of cardiac myocyte contraction and relaxation.
The PLB-SERCA Regulatory Complex in Cardiac Myocytes-Confocal microscopy showed the expected localization of PLB and SERCA in perinuclear membranes and in the junctional/ longitudinal sarcoplasmic reticulum of cardiac myocytes ( Fig.  2A). Interestingly, we also observed some PLB at the sarcolemma (Fig. 2B, inset, arrow). This is consistent with our previous observations of PLB in the plasma membrane of heterologous cells (19). It is not clear whether PLB has a physiological role at the sarcolemma or whether this localization is an artifact of overexpression. Others have proposed a physiological (33) or pathological (34) function for a sarcolemmal fraction of PLB. The high colocalization of SERCA with PLB was matched by significant FRET (Fig. 2D), indicating an intact regulatory complex in the membranes of live myocytes. We have previously shown that competition with unlabeled proteins reduces the observed FRET and FRET max (16,18), but here we observed that FRET max was similar to that observed in heterologous cells lacking endogenous PLB. This suggests that in the cells with the highest levels of expression, the endogenous PLB and SERCA are largely replaced by the exogenous proteins, consistent with the overexpression studies of other laboratories (35). Despite such robust steady-state FRET, we did not detect any change in the YFP/Cer ratio with single transients in paced cells, suggesting that FRET did not change with Ca 2ϩ under these conditions. The data are not consistent with the model that the PLB-SERCA regulatory complex is fully disrupted on a beat-to-beat basis by the elevated Ca 2ϩ of systole. However, we did detect a small (2%) reduction in FRET during intervals of prolonged rapid pacing relative to periods of rest (Fig. 3, C and D), a protocol that produces a relatively higher and more prolonged elevation of cytosolic Ca 2ϩ .
A complication for interpreting the significance of such relative changes is that reduced FRET may be due to a decrease in the number of interacting proteins or to a conformational change of the complex that increases the distance between the donor and acceptor fluorophores. To distinguish between these possibilities, we measured the dependence of FRET on the concentration of expressed protein. This assay also provides direct control over the Ca 2ϩ concentration. We found that the PLB-SERCA binding curve was modestly right-shifted by saturating Ca 2ϩ (Fig. 4A). The direction of this shift is consistent with decreased binding of PLB to E1 (Ca 2ϩ -bound) SERCA, but the magnitude of the change is much less than would have been predicted from the observation of others that PLB-SERCA chemical cross-linking is completely abolished by Ca 2ϩ (7)(8)(9)(10)(11). The 41% change in the apparent affinity of PLB for SERCA would support, at most, a 3% change in the PLB-SERCA binding equilibrium, depending on the actual in vivo concentrations of these binding partners. The Hill coefficient was 1.1 Ϯ 0.7 for the Ca 2ϩ dependence of binding affinity (Fig. 4B). This apparent lack of positive cooperativity is compatible with reports that showed that the SERCA conformational change occurs after binding of a single Ca 2ϩ to the first (high affinity) binding site (36 -39). Satoh et al. (40) also used FRET to detect a non-cooperative SERCA structure change under steady state conditions, although they also observed highly cooperative structure changes in response to Ca 2ϩ oscillations in cells.
The hypothesis that PLB inhibition of SERCA can be relieved without dissociation of the regulatory complex is in harmony with studies that have shown that PLB remains bound to SERCA after being phosphorylated (41,42). We have previously compared the effect of phosphomimetic mutations of PLB and phospholemman (PLM) on the respective regulatory complexes with SERCA or the Na ϩ /K ϩ -ATPase (NKA) (5,26). Both of these regulatory complexes persist after phosphomimetic mutations, although with reduced affinity and a small change in the quaternary structure. Importantly, this apparent stability of the PLB-SERCA binding equilibrium does not mean that the interaction is irreversible or static for particular molecules. Indeed, we have previously observed rapid exchange of PLB from its binding site on SERCA (19). High affinity binding is still compatible with rapid exchange of binding partners because the steady-state complex is the net result of binding and unbinding. The balance of these forward and reverse rates determines the binding affinity according to the relationship K D ϭ k off /k on . Our present data suggest that the balance of these rates is only modestly altered by SERCA conformational changes.
Regulatory Complex Structure-It is noteworthy that the pump ligands Tg and Ca 2ϩ did not significantly alter FRET max . This parameter is a measure of the intrinsic FRET of the regulatory complex and is sensitive to conformational changes that alter the distance between the chromophores at the centers of the fluorescent protein tags. With this assay, we have previously quantified small (4 Å) changes in probe separation distance from structure changes induced by phosphomimetic mutations (5,26). However, the present data suggest similar structures for the E1/E2 regulatory complexes. On the basis of some of the SERCA x-ray crystal structure solutions (Ca 2ϩ -bound (43) and Ca 2ϩ -free (44)), one might predict a distance change of tens of angstroms and a very large change in FRET max . Our present data are more consistent with the spectroscopy data of others that indicate a much smaller average structure difference between E1 and E2 (45,46).
Alternative Models of SERCA Regulation-Because the PLB-SERCA interaction persists during systole in paced myocytes and in heterologous cells subjected to saturating Ca 2ϩ , we conclude that relief of inhibition does not require unbinding of PLB from SERCA (Fig. 1B). This observation has implications for the search for small molecule modulators of SERCA activity. Evaluation of candidates should not focus exclusively on disruption of the complex. The present data suggest that PLB can bind to the E1 and E2-Tg enzymatic substates, although with slightly reduced affinity when compared with E2 (Fig. 6A). This does not mean that the E1 and E2 regulatory complexes are identical; rather, the observed modest change in apparent binding affinity suggests that the binding interface between PLB and SERCA has been altered. The changed binding interface may also cause the marked decrease in PLB-SERCA chemical crosslinking with Ca 2ϩ and Tg (11). One possibility is that the entire PLB transmembrane helix is displaced from the binding site in the SERCA M2/M4/M6 groove during the E2/E1 conformational change and that it then occupies an alternative binding site. In this regard, Glaves et al. (47) have postulated a secondary PLB binding site on M3 on the opposite side of SERCA. FIGURE 6. A model of PLB binding to SERCA conformers. A, the present study suggests that PLB can bind to the E1 and E2-Tg conformations of SERCA, but with reduced affinity when compared with the E2 substate. B, a comparison of computational models of the PLB-SERCA and PLM-NKA regulatory complexes (21,27), showing putative interactions between the regulatory peptide (red) with a groove formed by helices M2/M4/M6 (blue) or with helix M9 (green). Alternative binding sites may account for the differential binding affinities observed in the present study for Ca 2ϩ -free and Ca 2ϩ -bound SERCA.
However, translocation to the putative M3 site would be expected to result in a large increase in FRET max , which was not observed in the present study (Fig. 4A). An alternative hypothesis may be taken from structural studies of an analogous ionmotive ATPase, NKA, and its conjugate regulator PLM. Some structure solutions have shown that PLM interacts with the outside of the NKA M9 helix (23,48,49) instead of the M2/M4/M6 groove. Fig. 6B compares models of the regulatory complexes of SERCA and NKA (gray) (21,27) with the M2/M4/M6 groove highlighted in blue and M9 highlighted in green. The regulatory partners PLB and PLM are shown in red. The binding site on M9 is less than 20 Å from the canonical groove, so PLB could conceivably translocate between these sites without displacing the PLB cytoplasmic domain (altering FRET max ). Targeting additional pairs of cross-linkable residues may reveal whether PLB interacts with the outside M9 site on the Ca 2ϩ -bound (E1) SERCA. Future studies may also benefit from alternative cross-linking chemistries that do not require close apposition of compatible residues. In summary, we propose that PLB binds to adjacent alternative sites on SERCA, depending on the conformation of the pump. This model can accommodate the previous cross-linking (7-11, 30, 31) and spectroscopy (14,15) data, as well as the present FRET measurements obtained in cardiac myocytes.