Immunodetection of phosphorylation sites gives new insights into the mechanisms underlying phospholamban phosphorylation in the intact heart.

Phosphorylation site-specific antibodies, quantification of 32P incorporation into phospholamban, and simultaneous measurements of mechanical activity were used in Langendorff-perfused rat hearts to provide further insights into the underlying mechanisms of phospholamban phosphorylation. Immunological detection of phospholamban phosphorylation sites showed that the isoproterenol concentration-dependent increase in phospholamban phosphorylation was due to increases in phosphorylation of both Ser16 and Thr17 residues. When isoproterenol concentration was increased at extremely low Ca2+ supply to the myocardium, phosphorylation of Thr17 was virtually absent. Under these conditions, 32P incorporation into phospholamban, due to Ser16, decreased by 50%. Changes in Ca2+ supply to the myocardium either at constant β-adrenergic stimulation or in the presence of okadaic acid, a phosphatase inhibitor, exclusively modified Thr17 phosphorylation. Changes in phospholamban phosphorylation due to either Ser16 and/or Thr17 were paralleled by changes in myocardial relaxation. The results indicate that cAMP- (Ser16) and Ca2+-calmodulin (Thr17)-dependent pathways of phospholamban phosphorylation can occur independently of each other. However, in the absence of β-adrenergic stimulation, phosphorylation of Thr17 could only be detected after simultaneous activation of Ca2+-calmodulin-dependent protein kinase and inactivation of phosphatase. It is suggested that under physiological conditions, this requisite is only filled by cAMP-dependent mechanisms.

tion, is critically involved in the regulation of cardiac contraction and relaxation. Phosphorylation of phospholamban by either cAMP-dependent protein kinase (PKA) or Ca 2ϩ -calmodulindependent protein kinase (CaMKII) causes dissociation of phospholamban from the pump, thus increasing ATPase activity and the rate of Ca 2ϩ uptake by the SR (1,2). The increased rate of SR Ca 2ϩ uptake enhances the rate of the Ca 2ϩ transient decline and increases the Ca 2ϩ available for subsequent release, inducing increases in cardiac relaxation and contractility, respectively. In vitro experiments indicate that PKA and CaMKII phosphorylate phospholamban at two different sites, Ser 16 and Thr 17 , respectively (3). These phosphorylations are independent of each other, and when both are operating, they appear to have an additive action (4). In the intact heart, ␤-adrenergic stimulation phosphorylates phospholamban at both sites (5), which indicates that PKA-and CaMKII-dependent pathways are also working in the functioning heart. Whether these phosphorylation mechanisms are independent of each other and additive, as described in the isolated SR membranes, remains unknown. Different attempts to phosphorylate phospholamban by CaMKII in the intact heart have systematically failed unless cAMP levels within the cell increase (6 -11). This consistent finding strongly suggests an interaction between PKA and CaMKII pathways of phospholamban phosphorylation in the intact heart. The nature of this interaction as well as the cause for the difference between the in vivo and in vitro results have never been explored.
The availability of phosphorylation-site specific antibodies to phospholamban, which precisely discriminate between Ser 16 and Thr 17 phosphorylation sites (12), prompted us to reexamine the issue. Combination of this technique with the quantitative assessment of phospholamban phosphorylation by radiochemical labeling of ATP pools and simultaneous measurements of mechanical parameters allowed us to characterize the PKA and CaMKII-dependent mechanisms of phospholamban phosphorylation in the intact heart and their relative physiological roles on cardiac performance.

EXPERIMENTAL PROCEDURES
Heart Perfusions-Experiments were performed in isolated hearts from male Wistar rats (250 -350 g body wt) perfused according to the Langendorff technique as described previously (8). The composition of the physiological salt solution (PSS) was (in mM): 128.3 NaCl, 4.7 KCl, 1.35 CaCl 2 , 20.2 NaHCO 3 , 0.4 NaH 2 PO 4 , 1.1 MgCl 2 , 11.1 glucose, and 0.04 Na 2 EDTA. This solution was equilibrated with 95% O 2 , 5% CO 2 to give a pH of 7.4. The mechanical activity of the heart was assessed by either sewing an isometric strain gauge arch (Micro Measurements, type MA-06 -030LB-120) to the left ventricular wall or passing into the left ventricle a latex balloon connected to a pressure transducer (Namic, perceptor DT disposable transducer). The initial length of the gauge was set by stretching the segment attached by approximately 30%. The balloon was filled with aqueous solution to achieve a left ventricular end-diastolic pressure of 8 -14 mm Hg. No differences were observed between the mechanical data obtained by measuring either the isometric tension or the isovolumic pressure, and they were considered together for statistical analysis. Hearts were perfused with PSS for 10 -15 min for stabilization and then for the next 4 min with either PSS (control) or different interventions, as described under "Results." To quantify 32 P incorporation into phospholamban, hearts were perfused for 60 min with recirculation with PSS containing 10 Ci/ml 32 P i after the stabilization period and previously to the interventions assessed. At the end of the experimental period, the ventricles were freeze-clamped, pulverized, and stored at Ϫ70°C until biochemical assay.
Preparation of SR Membrane Vesicles-Membrane vesicles were prepared as described previously (8), except that the pulverized tissue from each heart was homogenized in 6 volumes of a medium containing (in mM): 30 KH 2 PO 4 (pH 7.0), 5 Na 2 EDTA, 25 NaF, 300 sucrose, 1 phenylmethylsulfonyl fluoride, and 1 benzamidine. Samples from 32 P-labeled perfused hearts were homogenized in the same medium except that the phosphate was replaced by 20 mM Tris-HCl (pH 7.0). Protein was measured by the method of Bradford (13) using bovine serum albumin as standard. The yield was 1-2 mg of membrane vesicles protein per g of cardiac tissue.
Electrophoresis and Western Blot Analysis-SDS-PAGE was performed using 10% acrylamide slab gels according to Porzio and Pearson (14) as described previously (8). Samples for electrophoresis were not boiled unless stated. For immunological detection of phospholamban phosphorylation sites, 10 g of membrane protein were electrophoresed per gel lane. Proteins were transferred to PVDF membranes (Immobilon-P, Millipore) and probed according to Drago and Colyer (12) with monoclonal antibody to phospholamban (1:5000) or polyclonal antibodies raised to a phospholamban peptide (residues 9 -19) phosphorylated at Ser 16 (1:10,000) or at Thr 17 (1:5000) (PhosphoProtein Research, UK). Immunoreactivity was visualized by peroxidase-conjugated antibodies using a peroxidase-based chemiluminescence detection kit (Boehringer Mannheim, Germany). The signal intensity of the bands on the film was quantified by optical densitometric analysis. To assess 32 P incorporation into phospholamban, 300 g of membrane protein were electrophoresed per gel lane. Gels were run in duplicate in order to use one of them for autoradiography and the other for liquid-scintillation counting. Quantitative results were expressed as pmol of 32 P incorporated into phospholamban per mg of SR protein based on the specific activity of 32 P in phosphocreatine (15).
Phosphorylation of Isolated SR Membranes-Phosphorylation of SR membrane vesicles was carried out at 30°C in 0.1 ml of reaction medium containing 50 g SR protein, 50 mM Tris-HCl (pH 7.0), 5 mM EGTA, 5 mM Mg-acetate, and 200 M ATP. PKA-dependent phosphorylation was catalyzed by 30 units/ml of the PKA-catalytic subunit. Alternatively, PKA-catalytic subunit and EGTA were omitted, and phosphorylation was carried out by the endogenous CaMKII present in the SR membranes in the presence of 0.5 mM Cl 2 Ca, 1 M calmodulin, and 1 M protein kinase A inhibitor peptide 5-24 amide. Reaction was stopped after 1 min (PKA) or 5 min (CaMKII) with SDS sample buffer.
Other Assays-Myocardial cAMP content was measured as described previously (8) by radioimmunoassay (16) using a commercially available kit (DuPont NEN) with acetylation of the samples. Results are expressed as pmol of cAMP per mg of wet weight. PP1 activity was assayed by measuring dephosphorylation of 32 P-labeled phosphorylase a (17). One activity unit was that amount of enzyme that catalyzes the release of 1 mol of 32 P/min.
Statistics-All data are expressed as the mean Ϯ S.E. of n preparations. Student's t test for paired or unpaired observations (mechanical and biochemical results, respectively) was used to test for statistical differences. p Ͻ 0.05 was considered statistically significant.

RESULTS
Immunodetection of Site-specific Phosphorylated Phospholamban-Specific antibodies to phospholamban, Ser 16 and Thr 17 -phosphorylated phospholamban peptides, were tested in SR membrane vesicles phosphorylated by the catalytic subunit of PKA or by the endogenous SR CaMKII (Fig. 1). The monoclonal antibody to phospholamban identified both the pentameric form of phospholamban (Fig. 1A, lanes a and c) and the monomeric form obtained by boiling the sample prior to electrophoresis (Fig. 1A, lanes b and d). Antibody to Ser 16 phosphopeptide only recognized phospholamban phosphorylated by PKA (Fig. 1B, lanes a and b) and ignored CaMKII-phosphorylated phospholamban (Fig. 1B, lanes c and d). Conversely, an-tibody to Thr 17 phosphopeptide exclusively recognized CaMKII-phosphorylated phospholamban (Fig. 1C, lanes c and d). Antibodies to Ser 16 phosphopeptide and to Thr 17 phosphopeptide did not recognize unphosphorylated phospholamban (data not shown).
Isoproterenol Concentration-dependent Induced Increase in Phospholamban Phosphorylation. Independence of the PKA Cascade- Fig. 2, panel A, shows an autoradiograph of SR membrane vesicles isolated from hearts perfused with 32 P in control conditions and in the presence of 3, 30, and 300 nM isoproterenol. Isoproterenol increased phosphorylation of phospholamban and reached a "plateau" at 30 nM. Panel B shows overall results of this experimental series.  Fig. 3B. The results indicate that the isoproterenol concentration-dependent increase in phospholamban phosphorylation (Fig. 2) is due to an isoproterenol concentration-dependent increase in the phosphorylation of both Ser 16 and Thr 17 residues. In all cases, a maximum was reached at 30 nM isoproterenol. Table I shows mean values of intracellular cAMP levels and of mechanical parameters obtained at the different isoproterenol concentrations.
In an attempt to diminish Ca 2ϩ supply to the myocardium and therefore block phospholamban phosphorylation by CaMKII, hearts were perfused with the same isoproterenol concentrations as above but in the presence of 0.07 mM [Ca] o plus 400 nM nifedipine. Myocardial contractility was abolished at any of the isoproterenol concentrations studied. Under these conditions, an isoproterenol concentration-dependent phosphorylation of Ser 16 was observed without detectable changes in phosphorylation of Thr 17 (Fig. 3A, lower panel). Mean results of three different experiments of this type are shown by the filled symbols in Fig. 3B. At each isoproterenol concentration, Ser 16 was phosphorylated to the same extent, independently of the degree of CaMKII-induced phosphorylation of Thr 17 . These findings indicate that PKA-dependent phospholamban phosphorylation is independent of the CaMKII pathway in the intact functioning heart.
Additivity of PKA-and CaMKII-dependent Phosphorylation of Phospholamban in the Intact Heart-To study the relative contribution of both phosphorylation sites to the total phospholamban phosphorylation after ␤-adrenoreceptor stimulation, hearts were perfused with 32 P, and phosphorylation of phospholamban was quantified at the "plateau" of the dose-response curve to isoproterenol (30 nM (Fig. 4A) and myocardial relaxation without changes in cAMP levels (Table I). Immunodetection of the site-specific phosphorylated phospholamban revealed that the decrease in phospholamban phosphorylation is exclusively due to a decrease in Thr 17 phosphorylation (Fig.  4B). Note that the isoproterenol-induced increase in Ser 16 phosphorylation was the same at any of the [Ca 2ϩ ] o assayed. Under conditions of extremely low Ca 2ϩ supply to the myocardium, in which Thr 17 phosphorylation was virtually absent, total phospholamban phosphorylation decreased by approximately 50%. These findings demonstrate the additivity of PKA and CaMKII pathways of phospholamban phosphorylation, in agreement with the in vitro results (4). Furthermore, both mechanisms contribute to the relaxant action of ␤-adrenergic stimulation (Table I).
CaMKII-dependent Phospholamban Phosphorylation. Role of Phosphatases-To address the participation of phosphatases in the degree of phospholamban phosphorylation, rat hearts were perfused with 32 P at two different [Ca] o in the absence and in the presence of the phosphatase inhibitor OA (0.1 M). This OA concentration decreased PP1 activity by 82.5 Ϯ 1.5%. Only in the presence of OA did the increase in [Ca] o enhance phospholamban phosphorylation (Fig. 5). This enhanced phosphorylation of phospholamban was associated with a significant increase in contractility and a decrease in half relaxation time (Table I). In 4 out of 11 experiments, the addition of OA at 3.85 mM [Ca] o produced a heart contracture. The cause for this contracture was not explored in the present experiments but might be due to the effects of OA on phosphatases other than those regulating phospholamban phosphorylation (18,19). The failure to detect phospholamban phosphorylation after increasing [Ca] o , in the absence of OA, was also observed when contractility was increased by poststimulation potentiation (PSP) (9) to levels similar to those evoked by maximal ␤-adrenergic stimulation (PSP of 88 Ϯ 9% versus isoproterenol of 72 Ϯ 7%). Phospholamban phosphorylation in pmol of 32 P/mg of SR protein was: PSP, 20.0 Ϯ 1.0 and control, 24.7 Ϯ 1.2. Immunological detection of the two phosphorylation sites of phospholamban (Fig. 6A) showed that the increase in [Ca] o did not increase phosphorylation of either Thr 17 or Ser 16 residues. The same increase in [Ca] o in the presence of 1 M OA increased phosphorylation of Thr 17 without affecting Ser 16 (Fig. 6A). This

FIG. 2. Isoproterenol concentrationdependent increase in 32 P incorporation into phospholamban.
Panel A, autoradiograph of SR membrane vesicles isolated from rat hearts perfused with 32 P and then without (lane a) or with different isoproterenol concentrations (Iso, lanes b to d). Samples (300 g) were subjected to SDS-PAGE and autoradiography. PHL H and PHL L designate the pentameric and monomeric forms of phospholamban, respectively. Panel B, isoproterenol dose-response curve for 32 P incorporation into phospholamban. Results are expressed as percentage of the maximal 32 P incorporation into phospholamban achieved in each experimental series (n ϭ 4). Maximal 32 P incorporation into phospholamban was 278.0 Ϯ 26.1 pmol/mg of SR protein. concentration of OA decreased PP1 activity by 97.9 Ϯ 1.0%. The overall results of this series are shown in Fig. 6B. These findings indicate that in the intact functioning heart CaMKII-dependent phospholamban phosphorylation can be detected in the absence of intracellular cAMP increases, provided phosphatases are inhibited.

DISCUSSION
Phosphorylation site-specific antibodies are a novel experimental tool to recognize a phosphorylated epitope of a protein.
The phosphorylation site-specific antibodies to phospholamban (12) used in the present experiments failed to detect the unphosphorylated protein and proved to be highly specific in the discrimination between the two sites of phosphorylation of phospholamban since no cross-reactivity with the other site of phosphorylation was observed (Fig. 1). The combination of this technique with the classical isotopic labeling technique of quantification of phospholamban phosphorylation, along with simultaneous measurements of mechanical parameters, allowed us a detailed characterization of the two signaling cascades of phospholamban phosphorylation, the relationships between them, and their physiological significance in the intact heart.
In agreement with the in vitro findings, the present results demonstrate that in the intact functioning heart, PKA-and CaMKII-dependent pathways of phospholamban phosphorylation may work independently of each other (Figs. 3 and 6) and that when both mechanisms are operating they have an additive action (Fig. 4). Previous works have consistently shown that CaMKII-dependent phospholamban phosphorylation and changes in myocardial relaxation can only occur in the intact heart when cAMP levels within the cell increase (7)(8)(9)(10)(11). This finding was interpreted as an interrelationship between PKA and CaMKII cascades, which would favor dual site phosphorylation of phospholamban after ␤-adrenoreceptor stimulation. This conclusion was in sharp contrast with the independence of both pathways described in the in vitro systems. The reason for the apparent discrepancy between the in vitro and the in vivo results as well as the nature of the interaction between the two phosphorylation cascades in the intact heart are not yet understood. Several mechanisms have been considered to explain this interaction, among which are the following. 1) Only the increases in intracellular Ca 2ϩ evoked by cAMP-dependent mechanisms are large enough to activate CaMKII pathway. 2) CaMKII cascade could be activated only by compartmentalized increases in Ca 2ϩ evoked by cAMP increases. 3) CaMKII could be activated by cAMP-dependent mechanisms unrelated to intracellular Ca 2ϩ increase. 4) Phosphatases that dephosphorylate phospholamban could be inhibited by a PKA-dependent mechanism. The fact that increases in contractility (reflecting cytosolic Ca 2ϩ ) similar to that evoked by isoproterenol, i.e. PSP, failed to phosphorylate phospholamban allows to rule out the first possibility. Similar conclusions were previously obtained (9). Furthermore, and relevant to the first three possibilities, we are presenting evidence showing that activation of CaMKII cascade and phosphorylation of Thr 17 residue could be detected even in the absence of high cAMP levels (Fig. 6). The present findings give support to the fourth hypothesis. Our results indicate that the nature of the interaction between PKA and CaMKII cascades of phospholamban phosphorylation lies in a basic mechanism underlying any phosphorylation process, i.e. the degree of phosphatase activity in the different experimental conditions. The major phosphatase that dephosphorylates phospholamban is a form of PP1 associated to the SR (20). Phosphorylation of Thr 17 residue in the intact heart could be

Effects of different interventions on mechanical parameters and intracellular cAMP levels in the intact heart
Rat hearts were perfused with the indicated interventions and then freeze-clamped and processed as described under "Experimental Procedures." Results are mean Ϯ SEM of n hearts (parentheses). PSS: physiological salt solution; Nife: nifedipine; ND: not determined. ⌬, indicate differences with respect to control. Control for mechanical data are values obtained immediately before the intervention. Control for biochemical data are values obtained in the PSS group. detected in two situations, at high intracellular cAMP levels and in the presence of OA. It has been proposed that SRassociated PP1 could be inhibited by PKA-dependent phosphorylation by two different but related mechanisms. First, as described for PP1 associated to glycogen particles (21) and to SR in skeletal muscle (22,23), PKA phosphorylation of the PP1-regulatory subunit would release the catalytic (C)-sub-unit, which would prevent PP1 from dephosphorylating phospholamban (20). Second, the thermostable protein inhibitor 1, when phosphorylated by PKA, becomes a potent inhibitor of PP1 C-subunit (24). Evidence for an isoproterenol-induced phosphorylation and increased activity of inhibitor 1 and for a reduction in SR-associated PP1 activity have been reported in the intact heart and isolated myocytes (25)(26)(27). PP1 is also inhibited by OA (18). We have found that increases in [Ca] o , in the presence of OA, phosphorylated phospholamban exclusively in Thr 17 . This phosphorylation was associated to a decrease in half relaxation time. Higher OA concentrations were described to produce an increase in phospholamban phosphorylation and an enhanced myocardial relaxation (28). A faster intracellular Ca 2ϩ decline induced by OA has also been observed in isolated myocytes (29). Unfortunately, none of these studies looked for sites of phosphorylation of phospholamban. In the present results, both situations in which we were able to detect Thr 17 phosphorylation, high cAMP and OA, have a common feature, and this is the inhibition of phosphatases.
In contrast to the in vivo results, phosphatase inhibition was not required to detect phosphorylation of either Ser 16 or Thr 17 residues of phospholamban in the in vitro systems (Fig. 1). This might be due to the already low phosphatase activity of the SR membrane preparation commonly used for the in vitro assays. It has been shown that PP1 activity decreased during the standard procedure of isolation of SR membrane vesicles (20) and that high salt treatment (used to wash out the membranes from myofibrillar proteins) also inhibits SR-associated PP1 activity (22).
For several years, different laboratories including our own, attempted to define the physiological role of the CaMKII pathway of phosphorylation of phospholamban (6 -11, 29). We are presenting evidence that indicate that both phosphorylation of phospholamban as well as the physiological expression of this phosphorylation, i.e. the increase in myocardial relaxation and the consequent increase in contractility, requires simultaneous stimulation of protein kinases and inhibition of phosphatases. Physiologically, this requisite is only filled by ␤-adrenergic stimulation. Although in the absence of high intracellular cAMP levels, the physiological meaning of CaMKII cascade appears as negligible, in its presence it is not. The present results indicate that under maximal ␤-adrenergic stimulation, activation of CaMKII cascade accounts for about 50% of phos- pholamban phosphorylation. This CaMKII-induced phospholamban phosphorylation is closely associated with an increase in the relaxant capacity of the intact ventricle.