Functional Co-expression of the Canine Cardiac Ca2+Pump and Phospholamban in Spodoptera frugiperda (Sf21) Cells Reveals New Insights on ATPase Regulation*

The utility of the baculovirus cell expression system for investigating Ca2+-ATPase and phospholamban regulatory interactions was examined. cDNA encoding the canine cardiac sarco(endo)plasmic Ca2+-ATPase pump (SERCA2a) was cloned for the first time and expressed in the presence and absence of phospholamban in Spodoptera frugiperda(Sf21) insect cells. The recombinant Ca2+ pump was produced in high yield, contributing 20% of the total membrane protein in Sf21 microsomes. At least 70% of the expressed pumps were active. Co-expression of wild-type, pentameric phospholamban with the Ca2+-ATPase decreased the apparent affinity of the ATPase for Ca2+, but had no effect on the maximum velocity of the enzyme, similar to phospholamban’s action in cardiac sarcoplasmic reticulum vesicles. To investigate the importance of the oligomeric structure of phospholamban in ATPase regulation, SERCA2a was co-expressed with a monomeric mutant of phospholamban, in which leucine residue 37 was changed to alanine. Surprisingly, monomeric phospholamban suppressed SERCA2a Ca2+ affinity more strongly than did wild-type phospholamban, demonstrating that the pentamer is not essential for Ca2+ pump inhibition and that the monomer is the more active species. To test if phospholamban functions as a Ca2+ channel, Sf21 microsomes expressing either SERCA2a or SERCA2a plus phospholamban were actively loaded with Ca2+ and then assayed for unidirectional45Ca2+ efflux. No evidence for a Ca2+ channel activity of phospholamban was obtained. We conclude that the phospholamban monomer is an important regulatory component inhibiting SERCA2a in cardiac sarcoplasmic reticulum membranes, and that the channel activity of phospholamban previously observed in planar bilayers is not involved in the mechanism of ATPase regulation.

The utility of the baculovirus cell expression system for investigating Ca 2؉ -ATPase and phospholamban regulatory interactions was examined. cDNA encoding the canine cardiac sarco(endo)plasmic Ca 2؉ -ATPase pump (SERCA2a) was cloned for the first time and expressed in the presence and absence of phospholamban in Spodoptera frugiperda (Sf21) insect cells. The recombinant Ca 2؉ pump was produced in high yield, contributing 20% of the total membrane protein in Sf21 microsomes. At least 70% of the expressed pumps were active. Coexpression of wild-type, pentameric phospholamban with the Ca 2؉ -ATPase decreased the apparent affinity of the ATPase for Ca 2؉ , but had no effect on the maximum velocity of the enzyme, similar to phospholamban's action in cardiac sarcoplasmic reticulum vesicles. To investigate the importance of the oligomeric structure of phospholamban in ATPase regulation, SERCA2a was coexpressed with a monomeric mutant of phospholamban, in which leucine residue 37 was changed to alanine. Surprisingly, monomeric phospholamban suppressed SERCA2a Ca 2؉ affinity more strongly than did wild-type phospholamban, demonstrating that the pentamer is not essential for Ca 2؉ pump inhibition and that the monomer is the more active species. To test if phospholamban functions as a Ca 2؉ channel, Sf21 microsomes expressing either SERCA2a or SERCA2a plus phospholamban were actively loaded with Ca 2؉ and then assayed for unidirectional 45 Ca 2؉ efflux. No evidence for a Ca 2؉ channel activity of phospholamban was obtained. We conclude that the phospholamban monomer is an important regulatory component inhibiting SERCA2a in cardiac sarcoplasmic reticulum membranes, and that the channel activity of phospholamban previously observed in planar bilayers is not involved in the mechanism of ATPase regulation.
Phospholamban is a pentameric transmembrane phosphoprotein regulator of the Ca 2ϩ -transport ATPase of cardiac sarcoplasmic reticulum (1,2). In the dephosphorylated state, phospholamban inhibits the Ca 2ϩ pump by decreasing the apparent affinity of the ATPase for Ca 2ϩ (3,4). Inhibition of the Ca 2ϩ pump is relieved by phosphorylation of phospholamban at serine 16 or threonine 17 or by the binding of a phospholamban monoclonal antibody to this cytoplasmic phosphorylation domain, resulting in a substantial increase in Ca 2ϩ transport into cardiac sarcoplasmic reticulum vesicles at low ionized Ca 2ϩ concentration (5)(6)(7). Purified phospholamban also forms Ca 2ϩ channels in lipid bilayers (8), but the functional role of this channel activity is ill defined (9). The physiological importance of phospholamban is demonstrated by recent work with cardiomyocytes (7) and phospholamban knockout mice (10), where it was shown that ablation of phospholamban regulatory function greatly augments the intracellular Ca 2ϩ transient and myocardial contractility, and at the same time attenuates the cardiac response to ␤-adrenergic agents such as isoproterenol.
To understand the molecular mechanism of phospholamban regulation, several mammalian cell expression systems have recently been developed in which phospholamban and the Ca 2ϩ -ATPase are co-expressed after transient transfection of cells with plasmid expression vectors (11)(12)(13). These studies have provided useful insights into the mechanism of phospholamban inhibition, including identification of some of the amino acid residues of phospholamban required for Ca 2ϩ pump regulation (14). However, the cell expression systems used to date have several drawbacks, including low transfection efficiencies, low expression levels, and low membrane yields, making detailed biochemical and kinetic characterizations with use of these systems difficult (11)(12)(13).
In the work described here, we have examined the utility of the baculovirus cell expression system for investigating phospholamban and Ca 2ϩ -ATPase regulatory interactions. An important strength of this system is that virtually all of the insect cells are infected using viral expression vectors, ensuring that very high levels of foreign protein expression are achieved (15). We recently reported on the use of this system for the expression and mass purification of canine cardiac phospholamban and several of its protein mutants from Sf21 1 cells (16,17). The purified, recombinant protein was successfully reconstituted into proteoliposomes to study its secondary structure (18) and oligomeric organization in the lipid bilayer (19). Successful co-reconstitution with Ca 2ϩ pumps purified from rabbit skeletal muscle (16) and canine myocardium (20) was also achieved. Here, we report on the further development of this system for functional co-expression of phospholamban with the canine cardiac Ca 2ϩ pump (SERCA2a). Microsomes isolated from infected Sf21 cells exhibit high levels of ATP hydrolysis and active Ca 2ϩ transport, and, furthermore, cardiac-like coupling between phospholamban and SERCA2a is retained. With the baculovirus system, we also demonstrate that a monomerforming mutant of phospholamban (17,19), unexpectedly, is a stronger inhibitor of SERCA2a activity than is the pentamer, suggesting that the monomer may be the key molecular species regulating the Ca 2ϩ pump in sarcoplasmic reticulum membranes. No evidence for a Ca 2ϩ channel activity of phospholamban was obtained.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP, 45 CaCl 2 , and 125 I-labeled protein A were purchased from DuPont NEN. Nucleic acid-synthesizing and -modifying enzymes were obtained from Promega. Sf21 cells were purchased from Invitrogen, and the BaculoGold TM system was obtained from Pharmingen.
Cloning Canine SERCA2a cDNA-A canine cardiac gt10 cDNA library (21) was screened in duplicate with 5Ј end-labeled oligonucleotide probes corresponding to base pairs 1-33 and 2935-2970 of rabbit SERCA2a cDNA (22). SERCA2a cDNA encoding the full-length canine cardiac Ca 2ϩ pump was excised from phage genomic DNA using EcoRI and subcloned into the EcoRI polylinker site of pBluescript. The SERCA2a cDNA clone contained approximately 250 base pairs of 5Јuntranslated sequence, a 2991-base pair open reading frame, and approximately 850 base pairs of 3Ј-untranslated sequence, which included a 39-base pair poly(A) tail. The entire protein coding region of the canine SERCA2a cDNA was sequenced in both directions by the dideoxy method (21).
Expression of SERCA2a and Phospholamban in Sf21 Cells-Wildtype phospholamban and L37A-PLB were expressed in Sf21 insect cells as recently described (16,17). To express canine SERCA2a in Sf21 cells, the XmaI insert encoding the Ca 2ϩ pump was excised from pBluescript and inserted into the XmaI site of the baculovirus transfer vector pVL1393 (15). This XmaI insert contained the entire protein coding region of the Ca 2ϩ pump, 90 base pairs of 5Ј-untranslated sequence, and the entire 3Ј-untranslated region including 11 base pairs excised from the pBluescript polylinker. Recombinant baculovirus containing the canine SERCA2a cDNA was obtained after co-transfection of Sf21 cells with pVL1393 and linearized baculovirus DNA using the BaculoGold TM system (17). The Ca 2ϩ pump and phospholamban were expressed in Sf21 insect cells grown in suspension (1.5 ϫ 10 6 cells/ml) at 27°C in Grace's insect cell medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Atlanta Biologicals), 0.1% Pluronic F-68 (Life Technologies, Inc.), 50 g/ml gentamicin, and 2.5 g/ml amphotericin B. Microsomes were isolated from insect cells harvested 48 h after infection with baculoviruses. For expression of the Ca 2ϩ pump alone, a multiplicity of infection of 10 (viruses per cell) was used. For co-expression of the Ca 2ϩ pump and phospholamban, a multiplicity of infection of 15 was used for SERCA2a, and of 5 for wild-type phospholamban and L37A-PLB.
Isolation of Microsomes from Sf21 Cells-Virus-infected Sf21 cells in 600 ml of suspension (9 ϫ 10 8 cells) were sedimented, washed twice with phosphate-buffered saline, and resuspended in 50 ml of medium containing 10 mM NaHCO 3 , 0.2 mM CaCl 2 , plus the following protease inhibitors: aprotinin (10 g/ml), leupeptin (2 g/ml), pepstatin A (1 g/ml), and Pefabloc (0.1 mM), which were included throughout the entire preparative procedure. Cells were disrupted by N 2 cavitation using a Parr Cell Disruption Bomb 4635 (Parr Instruments, Moline, IL), and cellular homogenates were diluted into an equal volume of ice-cold medium containing 500 mM sucrose, 300 mM KCl, 6 mM MgCl 2 , and 60 mM histidine (pH 7.4). Homogenates were centrifuged at 1000 ϫ g for 20 min. Supernatants were recovered, diluted with 0.25 volume of 3 M KCl, and centrifuged at 10,000 ϫ g for 20 min. Supernatants were again recovered and centrifuged at 100,000 ϫ g for 30 min. Pellets were washed in 250 mM sucrose, 600 mM KCl, 3 mM MgCl 2 , 30 mM histidine (pH 7.4) and sedimented as before. Final pellets (Sf21 microsomes) were resuspended at approximately 5 mg/ml in 250 mM sucrose, 30 mM histidine (pH 7.4) and stored in small aliquots at Ϫ40°C. The average yield per 600 ml of infection was 40 mg of microsomal protein.
Ca 2ϩ Uptake and Ca 2ϩ -ATPase Assays-Ca 2ϩ uptake was measured radiometrically using the Millipore filtration technique, and ATPase activity was determined by measuring release of P i from ATP colorimetrically (23). Assays were conducted at 37°C with 100 g of Sf21 microsomal protein in 1 ml of reaction medium consisting of 50 mM MOPS (pH 7.0), 100 mM KCl, 3 mM MgCl 2 , 2.7 mM ATP, 10 mM oxalate, 5 mM NaN 3 , with 2 mM EGTA and 0.20 to 1.8 mM CaCl 2 (containing tracer amounts of radioactive 45 Ca) to give the desired ionized Ca 2ϩ concentrations, determined as described previously (23). Ca 2ϩ uptake was terminated at selected times by vacuum filtering 10 g of membrane protein through glass-fiber filters, which were washed twice with 5 ml of 150 mM NaCl. 45 Ca 2ϩ accumulated inside membrane vesicles was monitored by liquid scintillation counting. ATPase activity was terminated at selected times by adding 50 l of reaction medium to 300 l of ice-cold 25 mM EDTA (pH 8.0). P i liberated was monitored by adding 2 ml of malachite green reagent and reading the absorbance at 660 nm (23). Basal ATPase activity, measured in the absence of added Ca 2ϩ with 2 mM EGTA, was subtracted from total ATPase activity to yield the Ca 2ϩ -ATPase activities reported. Prior to initiating assays, microsomes were preincubated for 20 min on ice with or without affinity-purified anti-phospholamban monoclonal antibody 2D12, at a membrane protein to antibody ratio of 1:1 (4,23). 45 Ca 2ϩ Efflux Assay-45 Ca 2ϩ efflux from Sf21 microsomes was measured after active Ca 2ϩ loading in the presence of 25 mM potassium phosphate, a Ca 2ϩ -precipitating agent that allows ready exchange of accumulated 45 Ca 2ϩ (24). Ca 2ϩ loading was performed at 37°C with 100 g of microsomal protein with or without the phospholamban monoclonal antibody in 10 ml of reaction medium containing 50 mM histidine (pH 7.0), 100 mM KCl, 3 mM MgCl 2 , 2.7 mM ATP, 5 mM NaN 3 , plus 25 mM potassium phosphate and 50 M added 45 Ca 2ϩ . After membrane vesicles had accumulated approximately 700 nmol of Ca 2ϩ /mg of protein, efflux of radioactive Ca 2ϩ from the vesicles was initiated by two methods: under conditions in which Ca 2ϩ uptake was stopped completely by addition of 2 mM EGTA, and under conditions in which Ca 2ϩ -ATPase remained active, by addition of 9 mM nonradioactive Ca 2ϩ and 10 mM EGTA (2 M ionized Ca 2ϩ concentration) to give a 180-fold dilution of the radioactive Ca 2ϩ in the medium. Ca 2ϩ load remaining within the microsomes was monitored at selected times by vacuum filtering 10 g of protein through glass-fiber filters, which were washed once with 5 ml of 150 mM NaCl. As a control, calcium efflux was also initiated by adding 3 g/ml of the Ca 2ϩ ionophore A23187.
SDS-PAGE and Immunoblotting-Prior to electrophoresis, microsomal proteins were solubilized at 37°C for 5 min in dissociation medium that contained 62.5 mM Tris (pH 6.8), 5% glycerol, 5% SDS, 5 mM dithiothreitol, and 0.0025% bromphenol blue. SDS-PAGE was conducted by the method of Laemmli (25) using 7.5-15% polyacrylamide (see Fig. 2) or by the method of Porzio and Pearson (26) using 8% polyacrylamide (see Fig. 6) (27). Gels were stained with Coomassie Blue, or proteins were transferred to nitrocellulose for immunoblotting. Nitrocellulose sheets were probed with anti-SERCA2a monoclonal antibody 2A7-A1 for detection of cardiac Ca 2ϩ pumps or with monoclonal antibody 2D12 for detection of phospholamban (28). Antibody binding proteins were visualized using 125 I-protein A followed by autoradiography. Labeling intensities were quantified with use of a GS-250 molecular imager (Bio-Rad).
Miscellaneous Methods-Procedure I canine cardiac microsomes enriched in sarcoplasmic reticulum were isolated as described previously (27). Purification of recombinant canine phospholamban from Sf21 cells was conducted as described elsewhere (16,17). The Ca 2ϩ -ATPase was purified from canine cardiac sarcoplasmic reticulum vesicles by 2A7-A1 monoclonal antibody affinity chromatography, using the methodology described by Reddy et al. (16). Protein concentrations were determined by the Lowry method using bovine serum albumin as the standard.

RESULTS
Cloning of Canine Cardiac SERCA2a cDNA-Most of the detailed biochemical work characterizing phospholamban and Ca 2ϩ -ATPase regulatory interactions has been conducted with sarcoplasmic reticulum preparations isolated from dog heart (1, 3). These well characterized preparations are enriched in the two proteins and are prepared in relatively high yield (27), which has greatly facilitated the biochemical studies. Ironically, however, the canine cardiac Ca 2ϩ pump has not been cloned or expressed to date. Therefore, to express canine SERCA2a in Sf21 cells, we first had to obtain the cDNA clone for this isoform. The SERCA2a cDNA clone was isolated and found to contain a 2991 base pair open reading frame encoding a polypeptide of 997 amino acid residues with a calculated molecular weight of 109,619. The deduced amino acid sequence of dog SERCA2a displays 98 -99% identity with mammalian SERCA2a Ca 2ϩ -ATPases cloned from human (29), rabbit (22), cat (30), rat (31), or pig (32) species, and 95% identity with avian SERCA2a Ca 2ϩ -ATPase cloned from chicken (33) (Fig. 1).
Expression and Functional Assay of SERCA2a in Sf21 Cells-Microsomes were isolated from Sf21 insect cells infected with the SERCA2a-encoding baculovirus. Separation of the microsomal membrane proteins by SDS-PAGE followed by Coomassie Blue staining revealed the abundant expression of an 110-kDa protein, which migrated with identical mobility as the Ca 2ϩ pump in canine cardiac sarcoplasmic reticulum vesicles ( Fig. 2A). The expressed protein was not visible in microsomes obtained from control (i.e. wild-type virus-infected) Sf21 cells. The identity of the expressed protein as the Ca 2ϩ pump was The canine SERCA2a amino acid sequence deduced from the cDNA sequence is reported in single-letter amino acid code. SERCA2a sequences from human (29), rabbit (22), cat (30), rat (31), pig (32), and chicken (33) species are shown for comparison. Identical residues are indicated by hyphens and the asterisk denotes the stop codon. Residue numbers are in the right margin. The DNA sequence has been submitted to GenBank TM (accession no. U94345).
confirmed by immunoblotting with a monoclonal antibody recognizing SERCA2a (28) (Fig. 2B), demonstrating that the recombinant Ca 2ϩ pump is expressed in Sf21 cell microsomes at levels approaching those of the native Ca 2ϩ -ATPase in cardiac sarcoplasmic reticulum vesicles. Quantitative immunoblotting using the purified canine cardiac Ca 2ϩ -ATPase as a standard revealed that the recombinant Ca 2ϩ pump accounted for 20 -25% of the total protein of Sf21 microsomes (data not shown), similar to Ca 2ϩ pump content reported in cardiac sarcoplasmic reticulum vesicles (34). Thus Sf21 cells infected with the SERCA2a-encoding baculovirus readily express the full-length Ca 2ϩ pump.
To confirm the functional integrity of recombinant SERCA2a, Ca 2ϩ transport assays were conducted with microsomal membranes isolated after N 2 cavitation of Sf21 cells. When assayed at the saturating Ca 2ϩ concentration of 1 M, the Ca 2ϩ transport rate was increased 10-fold in Sf21 microsomes expressing SERCA2a compared with control microsomes, confirming that the recombinant Ca 2ϩ pump was indeed functional (Fig. 3A). The Ca 2ϩ transport rate and the maximal level of Ca 2ϩ accumulation achieved was similar to that previously observed with use of canine cardiac microsomes (46). Ca 2ϩ activation of Ca 2ϩ transport was found to be halfmaximal at approximately 0.1 M ionized Ca 2ϩ (Fig. 3B), demonstrating that SERCA2a displays a high Ca 2ϩ affinity when expressed in the absence of phospholamban ( Fig. 3B; see Table  II). This apparent Ca 2ϩ affinity of SERCA2a is similar to that exhibited by the rabbit skeletal muscle Ca 2ϩ pump when expressed in Sf9 cells and assayed under similar conditions (35). As expected, a phospholamban monoclonal antibody (4, 36) had no effect on Ca 2ϩ transport when SERCA2a was expressed by itself (Fig. 3, A and B). Note that at the lower Ca 2ϩ concentrations tested (Յ0.1 M), microsomes containing SERCA2a accumulated at least 20 times more Ca 2ϩ than did control microsomes, due to the negligible activity of the endogenous Ca 2ϩ pump in Sf21 cells at low Ca 2ϩ concentration (Fig. 3B). Failure of the endogenous Ca 2ϩ pump in insect cells to react with SERCA1 (35)-or SERCA2-specific antibodies coupled with its low apparent affinity for Ca 2ϩ suggests that it may be the product of the SERCA3 gene (37), but this remains to be tested.
In control experiments we observed that the low background level of Ca 2ϩ uptake, as well as the Ca 2ϩ uptake attributable to SERCA2a expression, was completely inhibited by thapsigargin (data not shown).
Functional Coupling between Recombinant SERCA2a and Phospholamban Assayed by Ca 2ϩ Uptake-Previously we demonstrated that baculovirus-infected Sf21 cells express recombinant canine cardiac phospholamban with high efficiency (16,17). To test for functional coupling between canine cardiac SERCA2a and phospholamban we therefore co-infected Sf21 cells with viruses carrying the cDNAs encoding both proteins and isolated membranes for Ca 2ϩ transport assays. Fig. 2 shows that co-expression of phospholamban with SERCA2a in Sf21 cells produced similar levels of the two proteins in insect cell microsomes compared with the protein levels detected in canine cardiac sarcoplasmic reticulum vesicles. Phospholamban was visible in insect cell microsomes either by Coomassie Blue staining ( Fig. 2A) or by immunoblot analysis (Fig. 2B) and exhibited the characteristic conversion from the pentameric to monomeric forms (2) by boiling in SDS. Control Sf21 microsomes (WTV) contained no detectable phospholamban.
To test for functional coupling between SERCA2a and phospholamban in Sf21 microsomes, Ca 2ϩ transport assays were conducted at low (30 nM) and high (1 M) Ca 2ϩ concentrations in the presence and absence of anti-phospholamban monoclonal antibody 2D12, which blocks the inhibitory interaction between phospholamban and the cardiac Ca 2ϩ -ATPase (4,7,23,36). At low Ca 2ϩ concentration, Ca 2ϩ transport by Sf21 microsomes containing both proteins was stimulated 8-fold by addition of the phospholamban monoclonal antibody (Fig. 4A). However, at high Ca 2ϩ concentration, Ca 2ϩ transport by the same microsomes was unaffected by the antibody (Fig. 4B). Since the same monoclonal antibody increased Ca 2ϩ uptake by canine cardiac sarcoplasmic reticulum vesicles approximately 10-fold at a low Ca 2ϩ concentration, but had no effect at saturating Ca 2ϩ concentration (4, 23), we conclude that the recombinant SERCA2a Ca 2ϩ pump is tightly coupled to phospholamban in Sf21 microsomes in a similar fashion to that in cardiac sarcoplasmic reticulum vesicles.
Ca 2ϩ Affinities Monitored by ATP Hydrolysis-It has proven problematical to measure ATP hydrolysis by recombinantly expressed Ca 2ϩ pumps in previous studies, due to the low protein expression levels obtained and the interference by endogenous ATPase activities present in the microsomal preparations (38,39). The abundant expression of canine SERCA2a in Sf21 cells suggested that it would be possible to measure ATP hydrolysis by this recombinant Ca 2ϩ pump, which is a more direct method than assay of Ca 2ϩ transport for estimating apparent Ca 2ϩ affinities. Table I demonstrates that Ca 2ϩindependent (basal) ATPase activity was very low in Sf21 microsomes (ϳ200 nmol of P i /mg of protein/8 min). In membranes containing SERCA2a, the Ca 2ϩ -dependent ATPase activity was at least 20 times greater than the basal ATPase activity when measured at a saturating Ca 2ϩ concentration. Even at the lowest ionized Ca 2ϩ concentration tested (30 nM), ATP hydrolysis by SERCA2a was still four times greater than basal ATPase activity (Table I). ATP hydrolysis by the endogenous Ca 2ϩ pump was barely detectable (WTV) compared with that exhibited by SERCA2a. The maximal Ca 2ϩ -ATPase activity reported in Table I for SERCA2a expressed in Sf21 microsomes (33 mol of P i /mg of protein/h) is 55 times greater than the ATPase activity recently reported for SERCA2a expressed in HEK-293 membranes (0.6 mol of P i /mg of protein/h) (40).
Plots depicting Ca 2ϩ activation of ATP hydrolysis by SERCA2a, expressed in the presence and absence of phospholamban, are shown in Fig. 5. SERCA2a expressed alone had a high apparent Ca 2ϩ affinity (K Ca value ϭ 105 Ϯ 10 nM), which was unaffected by the phospholamban antibody ( Fig. 5A and Table II). Co-expression of phospholamban with SERCA2a decreased the Ca 2ϩ affinity by a factor of two, but this decrease in Ca 2ϩ affinity was removed by the phospholamban monoclonal antibody, shifting the Ca 2ϩ activation curve to the left (Fig. 5B and Table II). At the saturating Ca 2ϩ concentration of 2.4 M, the antibody had no effect on ATP hydrolysis (Fig. 5B). Thus phospholamban primarily decreases the Ca 2ϩ affinity of the pump, whether measured by assay of Ca 2ϩ transport or by ATP hydrolysis, but has no effect on the V max of the enzyme. It should be pointed out that the Ca 2ϩ -ATPase activities determined in these studies were about 2-3 times greater than the Ca 2ϩ transport rates measured, giving apparent coupling coefficients (Ca 2ϩ ions transported per ATP molecule hydrolyzed) of approximately 0.3-0.5. Similar low coupling coefficients are obtained with the use of canine cardiac sarcoplasmic reticulum vesicles, and are believed to be due to a significant proportion of leaky vesicles that hydrolyze ATP but are unable to retain accumulated Ca 2ϩ (4,23). We detected no differences in coupling coefficients between microsomes expressing SERCA2a alone, or microsomes expressing SERCA2a plus phospholamban.
Co-expression of a Monomeric Mutant of Phospholamban with SERCA2a-L37A-PLB is monomeric in SDS solution (17) or when reconstituted in phospholipid membranes (19). Since L37A-PLB is essentially depolymerized in lipid membranes, we co-expressed it with SERCA2a in Sf21 cells to test if the phospholamban monomer is sufficient for inhibition of the ATPase. Fig. 6 is an immunoblot showing that L37A-PLB was entirely monomeric on SDS-PAGE when co-expressed with SERCA2a in Sf21 cell microsomes. The expression levels of L37A-PLB and the Ca 2ϩ pump were similar to that achieved with coexpression of wild-type phospholamban and the pump (Fig. 6). Ca 2ϩ -transport assays conducted at 30 nM Ca 2ϩ revealed that L37A-PLB is a potent inhibitor of SERCA2a, and that this inhibition is reversed by addition of the phospholamban monoclonal antibody (Fig. 7A). ATPase activity was measured to assess Ca 2ϩ activation of SERCA2a when co-expressed with L37A-PLB (Fig. 7B). Like wild-type phospholamban, L37A-PLB shifted the Ca 2ϩ activation curve to the right, but, remarkably, the extent of the shift in the K Ca value was significantly greater with monomeric phospholamban (approximately 5-fold) than with wild-type phospholamban (approximately 2-fold) ( Table II). The reduction in Ca 2ϩ affinity by L37A-PLB was largely reversed by addition of the monoclonal antibody, and at the highest Ca 2ϩ concentration tested, only a marginal effect of the antibody was noted (Fig. 7B). (At ionized Ca 2ϩ concentrations greater than 10 M, no effect of the antibody was observed.) Similar Ca 2ϩ affinity shifts were observed when Ca 2ϩ uptake assays were conducted instead of measuring ATP hydrolysis (Table II). These results demonstrate that monomeric phospholamban is actually a stronger inhibitor of the Ca 2ϩ pump than is wild-type phospholamban. However, the basic mechanism of inhibition of SERCA2a by the two phospholambans is the same, in that the apparent K Ca value is increased with little effect on the V max of the enzyme measured at saturating Ca 2ϩ concentration. 45 Ca 2ϩ Efflux from Sf21 Microsomes Expressing Phospholamban-Purified phospholamban incorporated into planar  lipid bilayers forms Ca 2ϩ channels (8), and it has been proposed that Ca 2ϩ efflux through phospholamban is involved in its mechanism of ATPase regulation (9). This hypothesis has been difficult to test with cardiac sarcoplasmic reticulum vesicles, due to the presence of other channels in these membranes and the lack of adequate control membranes that contain SERCA2a but no phospholamban. Here we tested for a Ca 2ϩ efflux role for phospholamban by using Sf21 microsomes expressing SERCA2a alone, or microsomes expressing SERCA2a plus wild-type phospholamban. Microsomes were actively preloaded with 45 Ca 2ϩ in the presence of 25 mM phosphate, a Ca 2ϩprecipitating agent that readily exchanges Ca 2ϩ (24). 45 Ca 2ϩ efflux was then initiated under active transport conditions by diluting extravesicular 45 Ca 2ϩ with excess unlabeled Ca 2ϩ -EGTA buffer, or under conditions in which the Ca 2ϩ pump was inactivated by chelating extravesicular Ca 2ϩ completely with EGTA. When Ca 2ϩ efflux was initiated by the unlabeled 40 Ca 2ϩ chase (Fig. 8A), no difference in the rate of efflux was detected between SERCA2a-containing microsomes and microsomes containing SERCA2a plus phospholamban. Measurement of 45 Ca 2ϩ efflux when the Ca 2ϩ pump was not cycling, by chelation of all of the extravesicular Ca 2ϩ with EGTA, also showed that co-expression of phospholamban with SERCA2a did not increase the Ca 2ϩ efflux rate from microsomes (Fig. 8B). Addition of 3 g/ml of the Ca 2ϩ ionophore A23187 resulted in the rapid release of all of the accumulated Ca 2ϩ under both conditions (Fig. 8, A and B), demonstrating that the Ca 2ϩ inside the microsomes was readily releasable. Furthermore, the phospholamban monoclonal antibody had no significant effect on Ca 2ϩ efflux under either efflux condition (Fig. 8, A and B). Therefore, phospholamban does not exhibit a significant Ca 2ϩ channel activity in Sf21 microsomes when co-expressed with a functional SERCA2a Ca 2ϩ pump.
FIG. 6. Immunoblot showing co-expression of monomeric phospholamban and Ca 2؉ pump. 10 g of microsomes from Sf21 cells expressing monomeric phospholamban and the Ca 2ϩ pump (SERCA/L37A) were subjected to SDS-PAGE and immunoblotting, along with 10 g of microsomes expressing wild-type phospholamban and the Ca 2ϩ pump (SERCA/PLB). Blots were probed with antibodies as described in Fig. 2. Note that boiling in SDS (Ϯ Boil) was required to dissociate the wild-type phospholamban pentamer into monomers, whereas L37A-PLB was entirely monomeric without boiling in SDS.

DISCUSSION
To understand the molecular mechanism of phospholamban regulation of the Ca 2ϩ pump, an efficient expression system for these two proteins is desirable, which yields reasonable quantities of the recombinant proteins with functional interactions preserved. Here we have achieved this goal with the use of the baculovirus cell expression system. With this system, the SERCA2a ATPase contributes 20% of the total microsomal protein, a protein content comparable to that found in cardiac sarcoplasmic reticulum vesicles, and an expression level far greater than that previously reported (see below). Moreover, phospholamban co-expressed with SERCA2a is functional, producing regulatory effects identical to those observed with use of cardiac sarcoplasmic reticulum vesicles. Since preparation of microsomes from 600 ml of infected insect cells yields 40 mg of membrane protein (almost one-half the yield of sarcoplasmic reticulum vesicles from one dog heart), it is now possible to quickly and easily synthesize milligram quantities of the recombinant Ca 2ϩ pump using the baculovirus system.
Previous studies investigating recombinant SERCA Ca 2ϩ pumps resulted in levels of cellular protein expression much lower than those presently reported. For example, expression of SERCA1a in COS cells yielded 150 g of microsomal protein containing 2 g of Ca 2ϩ -ATPase (38). SERCA2a and SERCA3 Ca 2ϩ pumps have been expressed at comparable levels (37), where microsomal membranes contained approximately 50 -100 pmol of Ca 2ϩ pump per mg of total protein, or about 1-2% Ca 2ϩ -ATPase. SERCA1a has also been expressed in yeast cells (yielding 0.6% of total membrane protein in secretory vesicles) (39) and in baculovirus-infected Sf9 cells (yielding 1-3% of total membrane protein in microsomes) (35). A drawback to some of these systems, however, is that Ca 2ϩ -ATPase activities could not be measured, due to the low expression levels achieved and interference from basal ATPase activities present in the microsomal preparations (38,39). Additionally, in some of these studies a significant fraction of the recombinant Ca 2ϩ pumps were apparently inactive (12,41). As reported here, SERCA2a accounted for 20% or more of the total microsomal protein from infected Sf21 cells. Moreover, we estimate that 70 -80% of the expressed Ca 2ϩ pumps were active, relative to activity in cardiac sarcoplasmic reticulum membranes, by comparison of the ATPase hydrolysis rate to the amount of protein expressed determined by immunoblotting. Thus we have achieved expression levels of a highly active SERCA Ca 2ϩ pump 10 -20 times greater than those previously reported. In the study of Skerjanc et al. (35) the same baculovirus system was used to express the rabbit skeletal muscle Ca 2ϩ pump in Sf9 cells, but the level of the recombinant protein obtained was about 10-fold lower than that presently reported. The higher expression level observed presently could be due to the use of Sf21 cells instead of Sf9 cells or simply due to the fact that canine SERCA2a expresses more efficiently in insect cells than does rabbit SERCA1a; however, we have not investigated either of these possibilities. Even with the lower expression level in insect cells achieved by Skerjanc et al. (35), it was still possible to measure direct binding of 45 Ca 2ϩ to a recombinant Ca 2ϩ pump for the first time. Thus, we anticipate that Sf21 microsomes expressing canine SERCA2a will be a very useful system for future studies investigating detailed biochemical aspects of enzyme function. Previously we demonstrated that Sf21 cells express canine cardiac phospholamban very effectively, allowing purification of milligram quantities of the protein (16,17), and here we have shown that, when canine SERCA2a is co-expressed with phospholamban, the two proteins are functionally coupled. Both Ca 2ϩ -ATPase and Ca 2ϩ -uptake activities of Sf21 microsomes were stimulated manyfold at low ionized calcium concentration by the phospholamban monoclonal antibody, which produces the same effect as phosphorylation of phospholamban by protein kinases (7), but no effect of the antibody was noted at saturating Ca 2ϩ concentration. Thus, use of the baculovirus cell expression system provides additional strong evidence that the main regulatory effect of phospholamban is on the apparent Ca 2ϩ affinity of the Ca 2ϩ pump, but not on the V max of the enzyme. A similar regulatory effect of phospholamban on the Ca 2ϩ pump has been reported with use of canine cardiac sarcoplasmic reticulum vesicles (4,6,23,42), with use of the purified and reconstituted proteins (16,20), with use of the recombinant proteins expressed in HEK-293 membranes (43), and with use of phospholamban knockout mice (10). Although Kirchberger and co-workers (44) have recently challenged the idea that the main regulatory effect of phospholamban is on the K Ca value for Ca 2ϩ transport, most evidence concurs that dephosphorylated phospholamban strongly suppresses the Ca 2ϩ affinity of the enzyme, with an insignificant effect on V max . The argument by Kirchberger and co-workers (44) that the 45 Ca 2ϩ FIG. 8. Ca 2؉ efflux from Sf21 microsomes. Microsomes were isolated from Sf21 cells expressing SERCA2a (SERCA) or SERCA2a plus phospholamban (SERCA/PLB). Microsomes were actively loaded with 45 Ca 2ϩ in the presence of 25 mM phosphate, and Ca 2ϩ efflux was then initiated by cold Ca 2ϩ chase (panel A) or by chelation of all of the extravesicular Ca 2ϩ with EGTA (panel B), as described under "Experimental Procedures." Rapid Ca 2ϩ efflux was also induced by addition of a Ca 2ϩ ionophore (ϩA23187). ϩAb denotes incubations conducted in the presence of the phospholamban monoclonal antibody. uptake assay used by others (4,6,16,43) artifactually reports a Ca 2ϩ affinity change is negated by our results with ATPase activity measurements, where a similar K Ca shift by phospholamban is observed by a completely independent and more direct method.
Use of the Sf21 system allowed us to test if the phospholamban pentamer is essential for SERCA2a regulation. To examine this issue, SERCA2a was co-expressed with L37A-PLB, which is monomeric on SDS-PAGE (17) and, more importantly, also when reconstituted in lipid membranes (19). Surprisingly, we observed that L37A-PLB was a more effective suppressor of SERCA2a Ca 2ϩ affinity than wild-type phospholamban. Using an electron paramagnetic resonance technique, Cornea et al. (19) recently demonstrated that wild-type phospholamban exists in two physical states in the lipid bilayer; one state is composed of pentamers (80% of total phospholamban) and the other monomers (20% of total phospholamban), giving about an equimolar ratio of monomers to pentamers. These two states are in dynamic equilibrium, and phosphorylation of phospholamban by cAMP-dependent protein kinase shifts the equilibrium completely toward pentamers (19). Since we show here that the phospholamban monomer is a more effective inhibitor of ATPase activity than is the pentamer, the results suggest that phosphorylation of phospholamban in the sarcoplasmic reticulum membrane could relieve inhibition of the Ca 2ϩ pump by at least two mechanisms: one, by disrupting the physical interaction between phospholamban and the Ca 2ϩ pump (45) and, two, by promoting the more complete association of phospholamban into pentamers (19), which are relatively ineffective inhibitors compared with monomers. More sophisticated techniques will be required to test if the phospholamban pentamer by itself is capable of inhibiting the Ca 2ϩ pump, but the results presented presently in combination with those of Cornea et al. (19) clearly indicate that at least one monomeric mutant of phospholamban is a much stronger inhibitor of SERCA2a activity than is wild-type phospholamban. Our results are consistent with a recent preliminary report of Kimura et al. (47), who identified several additional monomer-forming mutations in the leucine zipper region of phospholamban (17), which produced similar potent inhibition of the Ca 2ϩ pump like L37A-PLB reported presently. Thus the increased effectiveness of L37A-PLB detected here is probably a direct consequence of its monomer status and not related to the change in the amino acid per se.
Although we reported earlier that purified phospholamban forms Ca 2ϩ channels in lipid bilayers (8), we found no evidence here for phospholamban acting as a Ca 2ϩ efflux channel when co-expressed with SERCA2a under conditions in which the two proteins are tightly coupled. A similar inability of phospholamban to form Ca 2ϩ channels was noted by Reddy et al. (16) in a study in which phospholamban was successfully co-reconstituted with the skeletal muscle Ca 2ϩ pump in phospholipid vesicles. Thus we believe that a Ca 2ϩ efflux role for phospholamban in its mechanism of regulation of SERCA2a is very unlikely and that recent models proposing such a role (9) should be viewed with some skepticism.
In conclusion, we have demonstrated that the baculovirus cell expression system is ideally suited for investigating phospholamban and SERCA2a regulatory interactions. High levels of protein expression are achieved, with preservation of functional coupling. With use of this system, it is now possible to carry out detailed biochemical and kinetic analyses while investigating both normal and mutated proteins.