Phospholamban regulates the Ca2+-ATPase through intramembrane interactions.

There is clear evidence for direct regulatory protein-protein interactions between phospholamban (PLN) and the Ca2+-ATPase of cardiac sarcoplasmic reticulum (SERCA2a) in cytoplasmic domains, but there is less clear evidence for regulatory interactions in the transmembrane domains of the two proteins. We have now coexpressed SERCA isoforms with the transmembrane sequence of PLN and with epitope-tagged transmembrane sequences of PLN to study intramembrane interactions in the absence of cytoplasmic interactions. Coexpression of the transmembrane sequence of phospholamban (Met-PLN28-52) with SERCA1a, SERCA2a, and SERCA3 inhibited Ca2+ transport by lowering apparent Ca2+ affinity. Addition of the hemagglutinin (HA) epitope to the transmembrane sequence of PLN (HA-PLN28-52) or deletion of PLN residues 21-29 (PLN1-20-PLN30-52) "supershifted" apparent Ca2+ affinity to values lower than those observed with native PLN without uncoupling Ca2+ transport from ATP hydrolysis. Inhibition by PLN1-20-PLN30-52 or by Flag-PLN28-52 was reversed by PLN antibody or by Flag antibody, demonstrating that inhibition by these constructs is reversible and that the inhibitory constructs are properly oriented in the membrane. These results suggest that PLN modulates the apparent Ca2+ affinity of SERCA2a through intramembrane interactions, which are disrupted at long range and in concert with disruption of the well characterized cytoplasmic interactions.

The hallmark of phospholamban (PLN) 1 inhibition of sarco-(endo)plasmic reticulum Ca 2ϩ -ATPases (SERCAs) is a decrease in apparent Ca 2ϩ affinity, which can be reversed by phosphorylation of PLN at Ser 16 or Thr 17 (1,2). This function makes PLN a key element in the inotropic response of the heart to ␤-adrenergic agonists. The 52 amino acids contained in each subunit of PLN homopentamers are predicted to be divided into three domains (3). PLN domain Ia (amino acids 1-20) is highly charged and largely helical (4), domain Ib (amino acids [21][22][23][24][25][26][27][28][29][30] is polar and unstructured, and domain II (amino acids 31-52) is neutral, very hydrophobic, and helical. Protein-protein interactions between the cytoplasmic domains of PLN and SERCA2a have been demonstrated before, but not after, phosphorylation of PLN and in the absence of elevated Ca 2ϩ (5). It is clear that there is a functional interaction site between the cytoplasmic domains of PLN and SERCA2a, since mutation of any of 13 of the 20 amino acids making up PLN domain Ia (6) or of any of residues KDDKPV 402 in the cytoplasmic domain of SERCA2a (7) diminishes the ability of PLN to inhibit SERCA2a.
In earlier studies, the addition of soluble synthetic PLN 1-31 suppressed V max without affecting Ca 2ϩ affinity of purified SERCA2a, while the in vitro reconstitution of purified SERCA2a with an unphysiological 100-fold molar excess of synthetic PLN 28 -47 lowered the apparent Ca 2ϩ affinity of SERCA2a (8). Attempts to reproduce these experiments revealed that the addition of excess PLN 26 -52 uncoupled Ca 2ϩ transport from ATP hydrolysis (9). Thus these experiments did not provide very strong evidence for a transmembrane interaction site between SERCA2a and PLN. Moreover, studies by Kirchberger et al. (10) showed that mild proteolysis, which removed most of the cytoplasmic sequence of PLN, activated SERCA2a by raising its apparent Ca 2ϩ affinity. The interpretation of the results of this study, however, is clouded by the fact that tryptic digestion of a membrane system could have pleiotropic effects.
In order to evaluate the functional effects of intramembrane interactions between PLN and SERCA2a, we added different epitope tags to the NH 2 terminus of PLN domain II and coexpressed the constructs with SERCA isoforms, thereby achieving in vivo reconstitution between the two proteins. We have found that PLN domain II inhibits SERCA2a by lowering its apparent affinity for Ca 2ϩ . We propose that interactions between transmembrane domains of PLN and SERCA2a are inhibitory but are modulated through long range coupling to the cytoplasmic interaction sites in the two molecules.
PLN construct cDNAs and SERCA cDNAs were cotransfected into * This work was supported by grants from The Human Frontier Science Program Organization and The Heart and Stroke Foundation of Ontario. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  HEK-293 cells using the calcium phosphate precipitation method (6). In a typical experiment, 8 g of PLN cDNA and 8 g of SERCA2a cDNA (1:1) were added to each dish. In some cases, however, 4 g of PLN 1-20 -PLN 30 -52 were mixed with 8 g of SERCA2a cDNA and 4 g of pMT2 DNA (1:2 ratio) or 12 g of Met-PLN 28 -52 were mixed with 4 g of SERCA3 cDNA (3:1 ratio). In other cases, 0.32 (1:10), 1.6 (1:2), 3.2 (1:1), 6.4 (2:1), 9.6 (3:1), or 12.8 (4:1) g of PLN cDNA were cotransfected with 3.2 g of SERCA2a cDNA, pMT2 DNA being added to make a total of 16 g of DNA per transfection. Reducing SERCA2a cDNA from 8 g to 3.2 g reduced V max of Ca 2ϩ transport, but not K Ca .
Microsome Preparation and Ca 2ϩ Uptake Assay-Microsomes were prepared and assayed for Ca 2ϩ transport activity and data were analyzed as described previously (6).
Assay of ATPase Activity-ATPase activity was measured under conditions identical to those used for measurement of Ca 2ϩ transport (6). The malachite green procedure for phosphate determination, developed as a phosphatase assay (14,15), was adapted to the microscale required to measure the inorganic phosphate liberated from ATP during Ca 2ϩ transport by transfected microsomes. The reaction was started by the addition of 80 l of microsomes (1 mg/ml) to 600 l of Ca 2ϩ transport reaction mixture. After 5, 10, 15, 20, and 25 min, the reaction was stopped by the transfer of 100 l of reaction mixture to 45 l of malachite green reagent mixture in a 96-well microplate. The malachite green reagent mixture was made by mixing 0.122% malachite green hydrochloride in 6.2 N H 2 SO 4 , 5.76% ammonium paramolybdate tetrahydrate, and 11% Tween 20 in a volume ratio of 100:66:2. Color development was quenched after 10 s by the addition of 45 l of 15.1% sodium citrate dihydrate. Inorganic phosphate liberated in the ATPase reaction was quantified by comparison of absorbance at 570 nm with standard curves generated with known amounts of Na 2 HPO 4 in the presence of 5 mM ATP. At 15 min, 150 l of the same microsomal sample was withdrawn and filtered for measurement of 45 Ca 2ϩ uptake. Background Ca 2ϩ uptake and ATP hydrolysis, defined as those values obtained from cells transfected only with pMT2 vector DNA, were subtracted from each data point. Under these conditions, ATP hydrolysis was found to be linear for at least 25 min.
Phosphorylation of Microsomes-Microsomes (0.8 mg/ml) were phosphorylated by 333 units/ml of cAMP-dependent protein kinase catalytic subunit (Sigma) in 0.15 M KCl, 0.25 M sucrose, 10 mM Tris-HCl, pH 7.5, 20 M CaCl 2 , 3 mM ␤-mercaptoethanol, and 25 M ATP for 5 min at room temperature. The cAMP-dependent protein kinase catalytic subunit was omitted in controls. The samples were then assayed for Ca 2ϩ transport assay as described previously (6).
Treatment of Microsomes with Anti-PLN or Anti-Flag Antibody-For detection of functional changes induced by anti-PLN antibody, microsomes transfected with SERCA2a and PLN or PLN 1-20 -PLN 30 -52 (1 mg/ml) were incubated with 0.25 mg/ml anti-PLN antibody 1D11, a generous gift from Dr. Robert Johnson, Merck Sharp and Dohme, in 10 mM Tris-HCl, pH 7.5, 0.225 M sucrose, 0.135 M KCl, 0.18 M CaCl 2 , 2.7 mM ␤-mercaptoethanol, and 0.015 M NaCl for at least 30 min on ice. Alternatively, microsomes (0.8 mg/ml) from cells cotransfected with SERCA2a and Flag-PLN 28 -52 were incubated for 1 h on ice with 0.8 mg/ml anti-Flag antibody M2 (IBI) in a buffer containing 70 mM KCl, 117 mM sucrose, 80 mM NaCl, 9.34 M CaCl 2 , 1.4 mM ␤-mercaptoethanol, 5.33 mM sodium phosphate, and 4.67 mM Tris-HCl, pH 7.5. Buffer was added to control samples to achieve the same salt composition in samples plus and minus antibody. Ca 2ϩ transport was assayed as described previously (6).
Immunoblotting was carried out as described previously (6), except that 0.2-micron polyvinylidene difluoride membranes (Bio-Rad) were used, and 10 g of of protein were dissolved in loading buffer at room temperature before separation on 12.5% polyacrylamide gels. Antibody binding was detected by chemiluminescence using an ECL Western blotting detection system (Amersham Corp.).

RESULTS
PLN, when coexpressed with SERCA2a, shifts the curve of Ca 2ϩ dependence of Ca 2ϩ uptake toward a lower apparent Ca 2ϩ affinity ( Fig. 1, A-D). Apparent K Ca (the Ca 2ϩ concentration that gives half-maximal Ca 2ϩ transport activity) was shifted from pCa 6.63 to pCa 6.31 when PLN and SERCA2a cDNAs were cotransfected in a 1:2 or 1:1 ratio ( Fig. 1; Table I).
Increasing the ratio of PLN cDNA to SERCA2a cDNA in the transfection system to 2:1, 3:1, or 4:1 did not shift apparent K Ca beyond pCa 6.21 under conditions in which an increase in cDNA in the transfection assay induced an increase in PLN synthesis as judged by Western blotting (Fig. 2A).
Met-PLN 28 -52 , when coexpressed with SERCA2a, also shifted K Ca significantly to pCa 6.46 (Table I). Unfortunately, we could not assess the level of expression of Met-PLN 28 -52 , since we had no antibody against domain II. We added epitope tags to PLN domain II in an attempt to measure synthesis of the PLN derivatives but found that they also resulted in very effective inhibitory constructs. Myc-PLN 30 -52 lowered apparent Ca 2ϩ affinity to pCa 6.29, Flag-PLN 28 -52 lowered apparent Ca 2ϩ affinity to pCa 6.22, and HA-PLN 28 -52 "supershifted" apparent Ca 2ϩ affinity to pCa 6.00.
Our most effective inhibitor of SERCA2a was PLN 1-20 -PLN 30 -52 . When the ratio of PLN 1-20 -PLN 30 -52 cDNA to SERCA2a cDNA in the cotransfection system was 1:2, the PLN construct supershifted apparent K Ca to pCa 5.76. When PLN 1-20 -PLN 30 -52 and SERCA2a cDNA were transfected in a weight ratio of 1:1, the K Ca measured in three of five experiments was supershifted to less than 5.61 pCa units, and in two of five experiments no Ca 2ϩ uptake was observed, even at 10 M free Ca 2ϩ ( Fig. 1B; Table I).
In all of these experiments SERCA2a was expressed at about the same level in the absence and in the presence of coexpressed PLN as judged by Western blot analysis (Fig. 2, A and  B). The expression level of PLN 1-20 -PLN 30 -52 , however, appeared to be lower than that of PLN as judged by Western blotting with the same antibody (Fig. 2B). Fig. 2B shows that PLN 1-20 -PLN 30 -52 is a mixture of oligomer and monomer much like intact PLN, and in Fig. 2C we show that Flag-PLN 28 -52 is also a mixture of oligomers and monomer. Since supershifts occurred without apparent overexpression of PLN 1-20 -PLN 30 -52 and since we could not achieve supershifting by adding more PLN cDNA to the transfection reaction, we conclude that the supershifts result from a qualitative change in the efficiency of the inhibitory peptides and are not related to overexpression.
The PLN constructs also suppressed Ca 2ϩ transport by SERCA1a (16) and SERCA3 (17), lowering the apparent affinity for Ca 2ϩ in each case (Fig. 1, C and D; Table I). Although SERCA3 has a lower apparent Ca 2ϩ affinity than either SERCA1a or SERCA2a (18 -20), NH 2 -terminal-truncated and epitope-tagged PLN constructs lowered its Ca 2ϩ affinity even further. The rank order for potency of suppression of Ca 2ϩ uptake by the various PLN constructs was the same for SERCA1a and SERCA2a, but the order of inhibition for Flag-PLN 28 -52 and HA-PLN 28 -52 was inverted for SERCA3.
To evaluate the possibility that the transmembrane domain of PLN might uncouple Ca 2ϩ uptake from Ca 2ϩ -ATPase (9) we measured Ca 2ϩ -dependent ATPase activity and Ca 2ϩ uptake simultaneously. As shown in Table II, the inhibition of Ca 2ϩ uptake by PLN constructs was mirrored by inhibition of Ca 2ϩdependent ATPase activity, even though values of Ca 2ϩ uptake and ATP hydrolysis varied from experiment to experiment depending on expression levels. Those samples which exhibited no Ca 2ϩ transport activity also had no ATPase activity. Thus it is unlikely that PLN constructs uncoupled Ca 2ϩ translocation from ATP hydrolysis.
Since PLN 1-20 -PLN 30 -52 could potentially be phosphorylated by cAMP-dependent protein kinase (PKA), we examined the question of whether phosphorylation of PLN domain Ia could exert a regulatory influence over domain II in this chimeric peptide. Unfortunately, PLN 1-20 -PLN 30 -52 was not phosphorylated by PKA under conditions in which PLN was phosphorylated, leading to an increase in Ca 2ϩ uptake at a low Ca 2ϩ concentration.
Monoclonal antibodies against PLN domain Ia can also reverse PLN inhibition of SERCA2 activity (21-23), and these observations led us to test whether antibodies against PLN or Flag epitopes might reverse the inhibition of SERCA2a by these epitope-tagged constructs. We observed that SERCA2a activities inhibited by either PLN 1-20 -PLN 30 -52 or Flag-PLN 28 -52 were enhanced by either the anti-PLN antibody 1D11 or the anti-Flag monoclonal antibody M2 at low Ca 2ϩ concentrations (Table III). These antibodies had little effect on the Ca 2ϩ uptake activity of microsomes transfected with SERCA2 alone.

DISCUSSION
We found that PLN constructs which lacked domain Ia, the site of interaction with cytoplasmic domains of SERCA2, were able to inhibit SERCA2a by lowering its apparent affinity for Ca 2ϩ . Met-PLN 28 -52 alone shifted K Ca significantly, Myc-PLN 30 -52 and Flag-PLN 28 -52 were as effective as PLN in lowering apparent Ca 2ϩ affinity, and HA-PLN 28 -52 supershifted apparent Ca 2ϩ affinity to values lower than that which could be achieved even with overexpression of native PLN. The inhibitory effects of the epitope-tagged PLN domain II constructs were not correlated with any specific characteristic of the epitope. Met-PLN 28 -52 was inhibitory without an epitope and the three added epitopes were not structurally related. Moreover, they were negatively charged (Ϫ3 for Myc and Flag; Ϫ2 for HA) instead of positively charged like PLN (ϩ3) or the domain Ia epitope PLN 1-20 (ϩ2). PLN 1-20 -PLN 30 -52 in which domain Ib (Pro 21 -Asn 29 ) was deleted was the strongest inhibitor of SERCA2a. Transfection with a PLN 1-20 -PLN 30 -52 cDNA to SERCA2a cDNA ratio of 1:1 led to complete inhibition of Ca 2ϩ transport even at 10 M Ca 2ϩ , and lowering of the ratio to 1:2 brought the K 0.5 for Ca 2ϩ activation into the measurable range. Western blotting with an anti-PLN antibody did not provide any indication that PLN 1-20 -PLN 30 -52 was overexpressed by comparison with PLN (Fig. 2B).
Our PLN constructs also suppressed Ca 2ϩ transport by SERCA1a and SERCA3, lowering their apparent affinity for Ca 2ϩ . The site of cytoplasmic interaction with PLN, KDD-KPV 402 in SERCA2, is conserved as KNDKPI 402 in SERCA1a but altered to QGEQLV 402 in SERCA3 (17), consistent with the fact that SERCA1a activity is inhibited by native PLN, while SERCA3 activity is scarcely affected by its coexpression with PLN (7,20). By contrast, the sequences of most of the transmembrane helices are highly conserved among the three SERCA isoforms (17). Even though SERCA3 had a lower apparent affinity for Ca 2ϩ than either SERCA1a or SERCA2a (18,19), epitope-tagged or NH 2 -terminal-truncated PLN constructs, but not PLN itself, lowered the apparent affinity of SERCA3 for Ca 2ϩ even further. The rank order for potency of suppression of Ca 2ϩ uptake by the various PLN constructs was similar for SERCA1a, SERCA2a, and SERCA3. These results are consistent with the view that it is a transmembrane interaction site and not the cytoplasmic interaction site that is responsible for lowering the apparent Ca 2ϩ affinity of all three SERCA molecules.
It has been suggested that the transmembrane domain of PLN might uncouple Ca 2ϩ uptake from Ca 2ϩ -ATPase (9). In our experiments, however, the inhibition of Ca 2ϩ uptake by PLN constructs was mirrored by their inhibition of Ca 2ϩ -dependent ATPase activity, and those samples which exhibited no Ca 2ϩ transport activity had no ATPase activity. Thus those of our PLN constructs which were tested did not uncouple Ca 2ϩ translocation from ATP hydrolysis in SERCA2.
We could not determine whether phosphorylation of PLN domain Ia could exert a direct regulatory influence over domain II in chimeric peptides, since PLN 1-20 -PLN 30 -52 was not phosphorylated by PKA. This lack of recognition of the PKA phosphorylation site in PLN 1-20 -PLN 30 -52 is consistent with a change in the exposure of the domain Ia sequence to cytoplasmic molecules in this construct. A monoclonal antibody against PLN domain Ia, however, activated PLN 1-20 -PLN 30 -52 -inhibited SERCA2a just as it does for PLN-inhibited SERCA2a (21)(22)(23). An antibody against the Flag epitope also reversed the inhibition of SERCA2a by Flag-PLN 28 -52 . Thus it is apparent that the inhibition of SERCA2a, induced by different PLN constructs, can be reversed by modulation of the cytoplasmic domain of the chimeric PLN constructs, just as modulation of the cytoplasmic domain of intact PLN by phosphorylation or antibody interaction can modulate PLN inhibition of SERCA2a.
Conclusions concerning the orientation of two of our chimeras can be drawn from these studies. The activating PLN and Flag antibodies can only be effective from the cytoplasmic side of the sealed Ca 2ϩ -impermeable microsomes used in this study. Accordingly, the epitopes in the chimeric molecules PLN 1-20 -PLN 30 -52 and Flag-PLN 28 -52 must be cytoplasmic and the transmembrane sequence must be in the same orientation in the chimeras and in native PLN.
From these results we propose a model of PLN-SERCA2a interaction in which PLN interacts with SERCA2a in at least two sites, one in the cytoplasmic sequences of PLN and SERCA2a and one within the transmembrane sequences of PLN and SERCA2a. We propose that the interaction between the transmembrane sequences of PLN and SERCA2a inhibits SERCA2a by altering its apparent Ca 2ϩ affinity. We also propose that interaction between PLN domain Ia and the cytoplasmic domain of SERCA2a is not by itself inhibitory (8), but it can modulate the inhibitory interactions in the transmembrane domains through long range coupling. If the cytoplasmic interaction is disrupted by PLN phosphorylation or binding of antibody, the inhibitory intramembrane interactions are also disrupted. If the inhibitory transmembrane interaction sites are FIG. 2. Immunoblotting of SERCA2a and PLN domain II constructs. In all panels, 10 g of microsomal protein were separated on 12.5% polyacrylamide gels and the gels were divided horizontally at a level corresponding to the mobility of a protein of M r 46,000. The upper segment was electroblotted onto 0.2-micron polyvinylidene difluoride membranes for 1.5 h, and the lower segment was blotted for 45 min. The upper segments in A and B were incubated with anti-SERCA2 monoclonal antibody IID8F6, and the lower segments in A and B were incubated with anti-PLN monoclonal antibody 1D11. Panel C was incubated with anti-Flag monoclonal antibody M2. HRP-conjugated antimouse IgG (Promega) was used as the secondary antibody in all panels. After addition of the substrates for chemiluminescence, the membranes were exposed to Kodak BioMax MR film for the following times: disrupted by elevated Ca 2ϩ concentrations, leading to Ca 2ϩ binding to the high affinity Ca 2ϩ binding and translocation sites in the transmembrane domain of SERCA molecules, then the regulatory cytoplasmic interaction sites are also disrupted.
In a striking analogy, long range interactions between the catalytic ATP hydrolytic site in the cytoplasmic headpiece domain of SERCA1 and the Ca 2ϩ binding and translocation sites in the transmembrane domain, mediated through a stalk sector, are an integral feature of Ca 2ϩ transport by SERCA molecules (24,25). It is now clear that PLN also has functional cytoplasmic (domain Ia) and transmembrane (domain II) domains, which are separated by a stalk sector (domain Ib) and that long range interactions occur between these functional domains. This long range coupling may occur entirely within the PLN molecule or it may be mediated by conduction through the SERCA2a molecule from its cytoplasmic to its transmembrane sites of interaction with PLN. In this case, a four-site regulatory circuit, possibly involving the catalytic site in the cytoplasmic domain of SERCA2a and the Ca 2ϩ binding and translocation sites in the transmembrane domain of SERCA2a, might best describe the interactions between SERCA2a and PLN.
We propose that PLN interacts only weakly with SERCA3 because the lack of compatible cytosolic interaction sites obviates the enmeshing of the transmembrane interaction sites. The transmembrane sequences of PLN constructs in which domain I is replaced or in which domain Ia is located nearer to the membrane surface (PLN 1-20 -PLN 30 -52 ) may interact directly with the transmembrane sequence of SERCA3, because steric hindrances created by the poor fit of the cytoplasmic interaction sites in the full length molecule are bypassed in the truncated molecule. a Ca 2ϩ /ATP ratio was not determined because both Ca 2ϩ uptake and Ca 2ϩ -ATPase activity were at background levels. b Ca 2ϩ /ATP ratio was not determined because we detected neither Ca 2ϩ uptake nor ATPase activity in two of five samples in which Ia-PLN 30 -52 and SERCA2a were cotransfected in a ratio of 1:1, and we detected very little activity in one sample, even at pCa ϭ 5.5.

TABLE III Effects of anti-Flag antibody (A) and anti-PLN antibody 1D11 (B) on the inhibition of SERCA2a by phospholamban domain II constructs
Microsomes were incubated with or without anti-Flag antibody M2 (A) or anti-PLN antibody 1D11 (B) and subjected to Ca 2ϩ transport assay as described under "Materials and Methods." Data are mean values Ϯ SD.