Critical evaluation of cardiac Ca2+-ATPase phosphorylation on serine 38 using a phosphorylation site-specific antibody.

The phosphorylation of the cardiac muscle isoform of the sarcoplasmic reticulum (SR) Ca(2+)-ATPase (SERCA2a) on serine 38 has been described as a regulatory event capable of very significant enhancement of enzyme activity (Hawkins, C., Xu, A., and Narayanan, N. (1994) J. Biol. Chem. 269, 31198-31206). Independent confirmation of these observations has not been forthcoming. This study has utilized a polyclonal antibody specific for the phosphorylated serine 38 epitope on the Ca(2+)-ATPase to evaluate the phosphorylation of SERCA2a in isolated sarcoplasmic reticulum vesicles and isolated rat ventricular myocytes. A quantitative Western blot approach failed to detect serine 38-phosphorylated Ca(2+)-ATPase in either kinase-treated sarcoplasmic reticulum vesicles or suitably stimulated cardiac myocytes. Calibration standards confirmed that the detection sensitivity of assays was adequate to detect Ser-38 phosphorylation if it occurred on at least 1% of Ca(2+)-ATPase molecules in SR vesicle experiments or on at least 0.1% of Ca(2+)-ATPase molecules in cardiac myocytes. The failure to detect a phosphorylated form of the Ca(2+)-ATPase in either preparation (isolated myocyte, purified sarcoplasmic reticulum vesicles) suggests that Ser-38 phosphorylation of the Ca(2+)-ATPase is not a significant regulatory feature of cardiac Ca(2+) homeostasis.

Regulation of Ca 2ϩ sequestration by cardiac sarcoplasmic reticulum has been identified as a key control point in cardiac muscle contraction (1,2). Stimulation of the rate of Ca 2ϩ uptake occurs during exercise, or following ␤-adrenergic stimulation (3) and is associated with an enhanced force of contraction and an increased rate of relaxation. This accounts for much of the positive inotropic and positive lusitropic effects of these interventions. In contrast, Ca 2ϩ sequestration by cardiac SR 1 is abnormally slow in the muscle of individuals with heart failure (4 -6). In animal studies, normalization of the rate of Ca 2ϩ sequestration has been shown to prevent progression of heart failure (7) and thus the molecular mechanism of control of Ca 2ϩ transport into the sarcoplasmic reticulum has become a focus for research into purposeful therapies to combat human heart failure (7,8).
Ca 2ϩ transport into cardiac SR is an enzymatic process performed by the (Ca 2ϩ -Mg 2ϩ )-ATPase (9) (or SERCA2a, Ref. 10), a member of the P-type ATPase family (11). The transport process involves the movement of two Ca 2ϩ from the cytoplasm into the lumen of the SR following the hydrolysis of a single molecule of ATP (12), although significantly lower coupling efficiencies have been measured empirically (13). The reaction cycle is relatively slow and thus a large number of SERCA2 molecules are expressed in cardiac muscle (up to 45% of total SR protein content, 14) to achieve the rates of Ca 2ϩ sequestration required by the kinetics of contraction and relaxation.
Regulation of Ca 2ϩ transport into the SR occurs on an acute time scale (seconds to minutes) through the transient modification of the proteins involved. The most important of these is phospholamban, which is a naturally occurring inhibitor of the Ca 2ϩ -ATPase. In its dephosphorylated state, phospholamban is able to interact with SERCA and reduce Ca 2ϩ -pump activity through a reduction in the apparent Ca 2ϩ affinity of the pump (15)(16)(17)(18). Phosphorylation of phospholamban on one of a number of sites (Ser-16, cAMP-dependent protein kinase; Thr-17, CaMKII; Ref. 19) abrogates the inhibitory influence of phospholamban (20), to reveal enhanced Ca 2ϩ transport activity at all physiological Ca 2ϩ concentrations. The kinetic basis of inhibition and subsequent activation of Ca 2ϩ transport is complex; involving contributions from the acceleration of particular reaction steps in the catalytic cycle of SERCA (21) and an increase in the coupling efficiency between ATP hydrolysis and Ca 2ϩ movement (13).
Other members of the P-type ATPase family are regulated by direct phosphorylation (e.g. plasma membrane Ca 2ϩ -ATPase, 22; Na ϩ ,K ϩ -ATPase, Ref. 23). A comparable form of regulation has been described for SERCA2 by a number of groups (24 -29), namely the phosphorylation of Ser-38 by calmodulin-dependent kinase II. Xu et al. (24) observed SERCA phosphorylation by CaMKII following the immunoprecipitaton of SERCA from a solubilized, kinase-treated SR preparation. In addition, Xu et al. (24) demonstrated the phosphorylation of purified SERCA2a by CaMKII, which was accompanied by a 1-fold stimulation of Ca 2ϩ -ATPase activity at all [Ca 2ϩ ]. Subsequent studies identified Ser-38 as the principal site of phosphorylation (25,30), and still further studies noted substantial enhancement of Ca 2ϩ -ATPase activity (1-1.5-fold increase in activity) despite the low stoichiometry of SERCA phosphorylation (12-13% of total SERCA; Ref. 26). Simple extrapolation of these latter data sets might suggest that phosphorylation of SERCA has the capacity to stimulate Ca 2ϩ -pump activity ϳ10fold, making this form of regulation more significant than that achieved by phospholamban.
The control of Ca 2ϩ -pump function by direct phosphorylation is not observed by all investigators. Odermatt et al. (30) de-tected no direct stimulation of Ca 2ϩ -pump function following CaMKII treatment, although phospholamban-mediated control was clearly evident. Furthermore this study noted a sizeable inhibition of basal Ca 2ϩ -pump activity upon brief exposure to Ca 2ϩ -free conditions (30). These Ca 2ϩ -free conditions were similar to the "control" conditions of Xu et al. (24), making it unclear whether Xu et al. (24) had described stimulation of pump activity in response to CaMKII phosphorylation or artifactual inhibition of pump activity under (Ca 2ϩ -free) control conditions. In separate studies Ready et al. (14) explored the phosphorylation of SERCA using highly purified cardiac SR and highly purified SERCA2a. CaMKII was unable to phosphorylate SERCA in either of these preparations, however the phosphorylation of a junctional SR protein of 100 kDa was noted. This phosphoprotein co-migrated with SERCA in onedimensional electrophoretic separations but did not share its distribution throughout the SR network. As such the identity of the phosphoprotein was suggested to be a protein unrelated to SERCA.
A number of physiological and pathophysiological observations might be explained in terms of altered SERCA activity upon phosphorylation of Ser-38. The lack of independent confirmation of the incidence or importance of Ser-38 phosphorylation of SERCA has prevented confident implication of this event as a molecular explanation of particular physiological observations (6,31). In the present study, we have produced and characterized an antibody specific for the Ser-38 phosphorylated form of SERCA2a in an attempt to evaluate whether Ser-38 phosphorylation occurs. This antibody is able to detect Ser-38 phosphorylation at levels of 0.03 pmol and above; however, it failed to detect SERCA in any experiment performed (included those involving kinase-treated SR containing up to 60 pmol of SERCA2a, and those involving myocytes exposed to CaMKII-activating interventions). As such we conclude that Ser-38 phosphorylation of SERCA does not occur to a significant extent in the situations examined.
Preparation of Phosphospecific Antibodies-Ca 2ϩ -ATPase peptide phosphorylated on Ser-38 ( 31 KLKERWGS(PO 4 )NEL 41 ) was prepared by the CaMKII phosphorylation of peptide 31 KLKERWGSNEL 41 . The phosphopeptide was purified to homogeneity by reverse phase high performance liquid chromatography. Peptide was conjugated to keyhole limpet hemocyanin (KLH) using carbodiimide cross linkage (33) and dialyzed extensively against buffer (50 mM Tris-HCl, pH 7.2, 150 mM NaCl). Adult New Zealand White rabbits were immunized with ϳ150 g KLH and attached peptide at 6-week intervals and immune serum collected 11 days after immunizations. Serum was prepared and stored at Ϫ70°C. A polyclonal antiserum is described herein: SERCA PS-38 raised to the phosphorylated peptide.
Production of Scaffolding Protein Titin I27 Domains (denoted (I27) 5 )-A gene encoding a concatamer of mutant forms of the I27 domain of titin was provided by Prof. Radford (University of Leeds, Ref. 34). The construct differs from the one described in Brockwell et al. (34) in that the two C-terminal cysteine residues have been deleted ( Fig.  2A). It is referred to as (I27) 5 throughout this study. (I27) 5 was expressed and purified as described in Brockwell et al. (34).
Calibration Standard: Peptide Conjugation to Scaffolding Protein-Purified phosphorylated Ser-38 peptide ( 31 KLKERWGS(PO 4 )NEL 41 ) (0.1 mol) was mixed with an excess of sulfosuccinimidyl 4-(N-maleimidomethyl)-cyclohexane-1-carboxylate (sulfo-SMCC) cross-linker (5 mol) in buffer containing 0.1 M sodium phosphate, 0.15 M NaCl, pH 7.2. After 1 h of incubation at room temperature, the maleimide-activated peptide was purified by gel filtration chromatography using a Superdex Peptide HR 10/30 column (Amersham Biosciences). The chromatography was performing using 0.1 M sodium phosphate, 0.15 M NaCl, pH 7.2, and a flow rate of 0.25 ml/min. Fractions of interest were pooled and urea was added to the fractions to make a final urea concentration of 9 M. (I27) 5 concatamer (0.1 mol) was added to the mixture and incubated for 2 h at room temperature. The conjugate was dialyzed extensively against water. Final product was stored at Ϫ20°C. The same procedure was followed to conjugate SERCA2a peptide (YLE-PAILE) to (I27) 5 concatamer. Protein concentration was determined by a BCA assay (35).
Phosphorylation of SR Proteins by CaMKII-Rabbit and canine cardiac sarcoplasmic reticulum vesicles were prepared as described previously (36). Phosphorylation reactions were conducted at 37°C in 100 l of buffer (adapted for Hawkins et al.,Ref. 26), containing 1 mg/ml SR vesicles and 50 mM Hepes-NaOH (pH 9.0), 5 mM MgSO 4 , 2 M calyculin A, and 1 M PKI. For phosphorylation by CaMKII (3 g/ml or 6 g/ml), the buffer contained 0.1 mM CaCl 2 and 37.5 g/ml of calmodulin. For the control samples, Ca 2ϩ (1 mM EGTA present), calmodulin and CaMKII were omitted from the assay medium. The phosphorylation reactions were initiated by the addition of ATP to a final concentration of 2 mM. In some experiments [␥-32 P]ATP (specific activity of 400 -550 cpm/pmol) was used. After 2 min of incubation, the reactions were stopped by adding 25 l of concentrated Laemmli sample buffer (5ϫstrength) resulting in a final solution composition of 62.5 mM Tris-HCl (pH 6.8), 4% glycerol, 2% SDS, 5% 2-mercaptoethanol, and 0.1% bromphenol blue (37). Samples were subjected to SDS-PAGE/Western blotting as described previously (38).
Isolated Cardiac Myocytes-Isolated rat ventricular myocytes were prepared from male Wistar rats according to the method of Frampton et al. (39). After isolation, cells were resuspended in 1 ml of a HEPESbased solution of the following composition (mM): NaCl 130; KCl 5.4; CaCl 2 1; MgCl 2 1.4; NaH 2 PO 4 0.4; HEPES 5; glucose 10; creatine 10, and taurine 20 at pH 7.4. Cells were counted using a counting chamber (Weber Scientific International Ltd.). Aliquots (100 l) of the myocyte suspension were placed in wells of a perspex container and stimulated via platinum electrodes at 0.5 Hz with a pulse width of 5 ms at room temperature. CaMKII activation in myocytes was provoked by exposure of myocytes to 1 M isoproterenol; to 2.5 mM extracellular calcium; to 1 M of the phosphatase inhibitor calyculin A; or to increased stimulated frequency (2.5 Hz). After 5 min, concentrated Laemmli sample buffer (5-strength), containing 50 mM NaF and 5 mM sodium pyrophosphate to prevent dephosphorylation of SR proteins during sample preparation (40), was added to each cell suspension. Samples were subjected to SDS-PAGE.
Phosphorimaging and Data Analysis-The amount of 32 P incorporation into the cardiac SR proteins was determined by phosphorimaging. Phosphorimage plates were exposed to PVDF membranes for 48 or 72 h, and densitrometric analysis of the data was conducted using Fuji BAS-1000 scanner (AIDA software for analysis). Phosphoproteins were also excised from PVDF membranes and quantified by scintillation counting.
ELISA-Competitive ELISA experiments were performed by standard protocols (42) using o-phenylene diamine as the substrate for peroxidase-conjugated second antibodies (Jackson Immunochemicals). ELISA plates were coated with phosphorylated Ser-38 peptide at 1 g/well, and phosphorylated, and de-phosphorylated Ser-38 peptides were used as competing agents as described in the relevant figure.
Statistics-All values are expressed as mean Ϯ S.D. for n experiments, and statistical significance was evaluated using the Student's t test for unpaired data.

RESULTS
The phosphorylation of SERCA2 on Ser-38 has been described as a regulatory feature capable of very significant activation of Ca 2ϩ -ATPase activity (24,26), although its occurrence and implication have been disputed (14,30). This site, although unique to SERCA2, is contained within a segment of the protein which is highly conserved between SERCA1 and SERCA2 (Fig. 1A), particularly from residue 39 onwards (43). As such, the two proteins are likely to display comparable structures in this region. This segment of SERCA1, for which two high resolution structures exist (11,44), is a surface ex-posed, highly mobile segment of the protein (Fig. 1B, highlighted in black). This segment remains solvent exposed in both conformational extremes of the enzyme (E1, E2; Fig. 1B). Assuming that the structure of this segment of SERCA2 is comparable, Ser-38 would be surface exposed and mobile. This would make it accessible to kinases and phosphatases in the major conformational states of the enzyme, and would also make it accessible to an antibody specific for this site. We have produced a phosphorylation site-specific antibody to this feature (highlighted in black in the equivalent structure of SERCA1, Fig. 1) in an effort to define the incidence and role of Ser-38 phosphorylation in cardiac muscle. A polyclonal antibody was produced to the sequence 31 KLKERWGS(PO 4 )NEL 41 , phosphorylated at Ser-38 as described under "Experimental Procedures." Fig. 2 shows that this polyclonal antiserum, SERCA PS-38, was specific for the phosphorylated peptide, because the phosphopeptide was a potent inhibitor of antibody binding to antigen (IC 50 18 nM), whereas the equivalent dephosphorylated peptide was unable to interfere with antibody: antigen recognition. Shading denotes immunogen sequence. B, x-ray crystal structure of SERCA1 in E1 conformation (44) and E2 conformation (11). Segment analogous to the (SERCA2) phosphoantibody epitope is highlighted in black, residue 38 side chain is displayed.

SERCA PS-38 Recognition of a Calibration Standard-Hav-
ing confirmed that polyclonal antibody SERCA PS-38 was specific for the phosphorylated Ser-38 epitope (Fig. 2), we examined the phosphorylation status of this residue in SERCA following exposure of cardiac SR vesicles to CaMKII. SERCA was not detected in these Western blot experiments (see Fig. 5 for an example). It was important to establish the basis of this negative result, to ensure that it was providing information about the incidence of Ser-38 phosphorylation, rather than recording a technical failing of the antibody or experiment. To this end, we constructed a calibration standard comprising the phosphopeptide epitope attached to an irrelevant scaffolding protein. The scaffolding protein was chosen as it contained a single site for peptide attachment (Fig. 3A) thus providing a uniform structure for the presentation of epitope peptide, ideal for accurate quantitation.
The scaffolding protein employed was a concatamer of five copies of a domain from titin (I27), mutated to remove all but one cysteine residue from the concatamer sequence (Fig. 3A, adapted from Brockwell et al., Ref. 34). Purified phosphoepitope peptide was conjugated to the (I27) 5 concatamer via the only cysteine residue in the protein (C47 in I27 domain 3, presented schematically in Fig. 3A), and the stoichiometry of covalent attachment of the peptide was evaluated by mass spectrometry. A low stoichiometry of peptide attachment to the concatamer (final product mass 54,124 Da, labeled calibration-38; Fig. 3B) was observed on this occasion, which comprises 5.4% of the total preparation. Nevertheless, this low level of conjugation to the concatamer proved sufficient for immunodetection (Fig. 3C). Fig. 3C shows that antibody SERCA PS-38 recognized the concatamer product decorated with the relevant phosphopeptide (calibration-38), but did not recognize the same concatamer (I27) 5 decorated with an irrelevant peptide (calibration-␣CLEP) even when 60 pmol of concatamer was presented. The calibration standard migrates as a single molecular species of ϳ60 kDa in SDS-PAGE. Furthermore, the phosphorylated epitope was detected by antibody SERCA PS-38 with high sensitivity, down to a limit of 0.1 pmol of epitope peptide using standard (SuperSignal West Pico, Pierce) ECL substrate (Fig. 3C).
The calibration standard (calibration-38) contained some minor contaminants. A contaminant of 51,230 Da, seen on the mass spectrum, does not appear to accept peptide (not detected in Western blot experiments, Fig. 3C). This material was included in the calculation of percentage product (calibration-38) as it made an appreciable contribution to total protein. A second contaminant of the (I27) 5 preparation is covalently labeled by peptide. It underlies the immunostaining of a complex of high M r (ϳ250 kDa, Fig. 3C). This product was undetectable in the mass spectrum and therefore is present in low amounts in the calibration-38 preparation. It does not make an appreciable contribution to total protein and was excluded from consideration in the quantification performed in this study.
Ser-38 Phosphorylation of SERCA Is Not Detected in Cardiac SR-The conditions for optimal phosphorylation of SERCA have been described precisely (26), and these conditions (pH 9.0, high ATP, high CaMKII) were deployed with rabbit and dog cardiac SR. Total SR proteins and a calibration standard series (calibration-38, 0.03-4.8 pmol) were separated by SDS-PAGE in 10% gels and immunoblotted with SERCA PS-38 following transfer to PVDF membrane (Fig. 4A). Antibody SERCA PS-38 detected the phosphorylated epitope in the calibration standard at loadings of 0.05 pmol and above; however, it failed to recognize SERCA protein, or any other protein, in the samples from dog (Fig. 4A). Immunorecognition of a protein of ϳ100 kDa, consistent in size to SERCA, was observed with rabbit SR samples (Fig. 4A). This interaction, however, was nonspecific, as it was not affected by inclusion of competing phosphoepitope peptide in the blot (Fig. 4B), and was observed without exposing the blot to primary antiserum (SERCA PS-38, Fig. 4C). Thus the immunoreactivity must be explained by recognition of an unidentified rabbit protein of ϳ100 kDa by the goat-anti-rabbit-peroxidase reagent.
Rabbit immunoglobulin G can be detected using Staphylococcus aureus protein A based reagents as an alternative to anti-rabbit IgG preparations. To overcome the nonspecific recognition of proteins in rabbit cardiac SR by our goat anti-rabbit peroxidase reagent, rabbit cardiac SR was treated with CaMKII and analyzed with antibody SERCA PS-38 using protein A-peroxidase detection techniques. With this approach, the phosphoepitope was detected in the calibration standard series at loadings in excess of 0.1 pmol, but SERCA was not detected (Fig. 5B) despite loading 10 g of CaMKII treated SR, which would contain 10 pmol of SERCA2a (26). Careful examination of this experiment shows the recognition of a protein of ϳ70 kDa in control and CaMKII-treated rabbit SR. This band was not eliminated by inclusion of the phosphorylated epitope peptide, and was not seen in numerous additional experiments. It represents a minor, nonspecific band detected by the high sensitivity methods being deployed to observe trace amounts of SERCA2 phosphorylation.
Thr-17 phosphorylation of phospholamban in the samples presented in Fig. 4A is shown in Fig. 4D. These data show enhanced Thr-17 phosphorylation of phospholamban following exposure to CaMKII (Fig. 4D, lanes 2 and 3), which confirms that the kinase was active against SR targets despite the fact that SERCA phosphorylation on Ser-38 was not observed. Taken together, the data of Figs. 4 and 5B demonstrate that CaMKII was active (Fig. 4D), that the immunodetection of the Ser-38 phosphoepitope was performed with high sensitivity (Figs. 4A and 5B), but that SERCA was not detected by the phosphospecific antibody (Figs. 4A and 5B). This suggests that there is no basal (pre-existing) Ser-38 phosphorylated SERCA in SR vesicles, and that CaMKII is unable to phosphorylated

Ser-38 Phosphorylation of SERCA2
this residue. It is possible that SERCA phosphorylation occurred at levels below that detected by the antibody (0.1 pmol in this experiment). However, SR loadings of 10 g contain at least 10 pmol SERCA (assuming SERCA is 10% of total SR protein, Ref. 26). Thus phosphorylation of Ͻ0.1 pmol represents phosphorylation of Ͻ1% of SERCA molecules.
Cardiac SR Phosphoprotein of 100 kDa Is Not SERCA-CaMKII phosphorylation of rabbit cardiac SR using radioactive ATP, employing the conditions of Hawkins et al. (26) results in the production of a phosphoprotein of ϳ100kDa (Fig. 5A, lane  2). The use of high concentrations of ATP appeared important, as this phosphoprotein was not observed when 0.2 mM ATP was used (data not shown). To evaluate whether SERCA was the phosphoprotein of ϳ100 kDa, this membrane was blotted with antibody SERCA PS-38 (Fig. 5B). Immunodetection sensitivity in these experiments was shown to be high (0.1 pmol of calibration-38, Fig. 5B); however, SERCA was not detected. The radioactivity associated with the ϳ100 kDa protein was quantified at the end of the experiment, at 0.16 pmol, while 10 pmol SERCA would be expected at that location. The amount of phosphoprotein was within the sensitivity range of antibody SERCA PS-38, but was not detected by this antibody. Further experiments were performed with both rabbit and rat cardiac SR. The loading of protein was increased to 60 g (60 pmol of SERCA) but antibody SERCA PS-38 did not detect SERCA (data not shown). This suggests that Ser-38 phosphorylation is not responsible for the incorporation of 32 P into a protein of ϳ100 kDa in rabbit cardiac SR.

Ser-38 Phosphorylation of SERCA2
SERCA Remains Dephosphorylated in Isolated Rat Ventricular Myocytes-In a further attempt to evaluate SERCA phosphorylation on Ser-38, we analyzed the pattern of (selected) protein phosphorylation in isolated cardiac myocytes exposed to stimulants able to activate CaMKII. Others have shown that an increase in electrical stimulation frequency (45), exposure to ␤-adrenergic agonists (46) and exposure to potent Ser/Thr phosphatase inhibitors (47) results in the activation of CaMKII in isolated rat ventricular myocytes. Rat myocytes treated for 5 min with each of the above intervention regimes (stimulated electrically at 0.5 Hz, unless otherwise stated) displayed enhanced CaMKII phosphorylation of phospholamban compared with control myocytes (Fig. 6B, lanes 2-5, see lane 1). The phosphorylation of SERCA on Ser-38 was examined in these myocyte samples using the most sensitive detection reagents available (SuperSignal West Femto; Pierce). This permitted the detection of 0.03 pmol of epitope in the calibration series using this antibody, but did not result in the immunodetection of SERCA with any of these treatments.
Thus we conclude that SERCA phosphorylation, if occurring at all, results in the generation of less than 0.03 pmol of Ser-38 phosphoprotein in the cells studied (10,000 viable myocytes). In previous studies, Ser-16 phosphorylation of phospholamban in myocytes following similar interventions was quantified at 8.5 pmol/1,000 cells (46), indicating the presence of at least 85 pmol of phospholamban in the 10,000 cells of the present study. As phospholamban and SERCA are expressed in similar amounts in cardiac muscle (2 phospholamban per SERCA, Ref. 16), we might expect 42 pmol of SERCA in the experiments performed. Our failure to detect Ser-38 phosphoprotein with the antibody described herein suggests that less than 0.1% of SERCA is phosphorylated in rat cardiac myocytes treated with CaMKII stimulants.

DISCUSSION
In the present study we have described a polyclonal antibody wholly specific for a phosphorylated Ser-38 epitope on SERCA2. The antibody was able to detect the phosphorylated epitope in a calibration standard with high sensitivity (0.03-0.1 pmol). However, it failed to recognize SERCA2 in cardiac SR samples from a variety of animal species, despite the presentation of large amounts of SERCA (10 -60 pmol) and the presence of a phosphoprotein of 100 kDa. This indicates that either SERCA is not phosphorylated on Ser-38, or that only a minor fraction of SERCA molecules (i.e. less than ϳ1%) are phosphorylated on Ser-38. CaMKII activation in isolated car- diac myocytes was achieved using four independent stimuli resulting in phospholamban phosphorylation on Thr-17. None of these resulted in detectable Ser-38 phosphorylation of SERCA using this antibody, despite immunodetection of 0.03 pmol of the calibration standard in the same experiment. These data provide no support for SERCA phosphorylation in myocytes, however the detection sensitivity limit of the approach employed necessitates conclusion that less than 0.1% of SERCA molecules become phosphorylated on Ser-38 in response to CaMKII activating stimuli in intact cardiac myo-cytes. This study does not provide evidence that Ser-38 phosphorylation of SERCA2a is a significant event in cardiac myocytes or cardiac SR preparations.
Immunodetection of the Ser-38 Phosphoepitope-As we have shown previously with other targets (38,41), polyclonal antisera specific for a phosphorylated epitope can be produced by immunizing rabbits with a short phosphopeptide sequence, purified to homogeneity and attached to an immunogenic carrier protein. The serum from one of the two animals immunized in this study contained high titer antibodies, wholly specific for the phosphorylated Ser-38 epitope of SERCA, as antigen recognition was inhibited by the phosphopeptide immunogen, but was unaffected by the equivalent dephosphorylated peptide sequence. This antiserum, however, was unable to recognize SERCA in any experiment performed.
The interpretation of negative data requires careful consideration, to ensure that technical shortcomings do not underlie the observations made. Such technical shortcomings would include: 1) deployment of an experimental format unsuitable for the antibody in question; 2) presentation of insufficient antigen (phospho-SERCA2); or 3) presentation of an epitope inaccessible to the antibody. Each of these can be discounted using experimental evidence or logical argument. The data derived from experiments using the peptide calibration standard (calibration-38) address the first two potential shortcomings directly. It is clear that Western blotting is a suitable format for the immunodetection of an epitope by this antibody, as detection of 0.1 pmol of epitope was seen in all experiments. Furthermore these experiments confirm that sufficient SERCA was presented in each experiment, as 10 -60 pmol of SERCA was analyzed in the various experiments performed.
Although we do not have experimental data which address the accessibility of the Ser-38 epitope directly, two separate lines of evidence suggest that it is accessible. Firstly, the threedimensional structure of SERCA1 shows that the equivalent segment of this isoform is solvent accessible in the enzyme (11). Furthermore this region undergoes significant movement as the enzyme undergoes conformational change in its reaction cycle (44). This infers that the segment is solvent accessible and highly mobile, both of which would assist antibody access to the site. In this study we are dealing with SERCA2 not SERCA1. The structure of SERCA2 is not available, however the primary sequence of the two enzymes is well conserved in this region particularly from residue 39 onwards (43). Thus it is likely that they share a similar structure in this region, and that the antibody epitope on SERCA2 is in a solvent-exposed, highly mobile segment of the protein.
A highly mobile, solvent accessible structure is a prerequisite for a kinase substrate. Generally, a protein segment subject to phosphorylation interacts with the active site of the kinase in an extended conformation (48) and thus the phosphorylation motif is located in a flexible loop, or at the N or C terminus of a protein (48). Therefore if we assume Ser-38 is a phosphorylation site, this segment of SERCA2 must be surface exposed and mobile. This would make the segment accessible to antibody binding, as has been described for many hundreds of other phosphoproteins.
SERCA2 Is Not Phosphorylated on Ser-38 by CaMKII-In this study we have examined the phosphorylation of SERCA on Ser-38 in isolated SR vesicles and isolated cardiac myocytes. In each case interventions that activate CaMKII have been deployed, using the optimal conditions described in the literature (26,(45)(46)(47). In experiments using cardiac SR vesicles and those using isolated rat myocytes, Thr-17 phosphorylation of phospholamban has demonstrated high levels of CaMKII activity focused on SR targets. Furthermore large amounts of sample FIG. 5. A cardiac phosphoprotein of 100 kDa is not recognized by SERCA2 antibody. Rabbit cardiac SR vesicles (1 mg/ml) were phosphorylated by exogenous CaMKII (6 g/ml) as described under "Experimental Procedures." Phosphorylation reactions were initiated by the addition of 2 mM [␥-32 P]ATP (specific activity 400 cpm/pmol). After 2 min at 37°C the reactions were terminated by the addition of Laemmli sample buffer (37)  have been analyzed in an attempt to identify the SERCA phosphoprotein (10 -60 pmol). Nevertheless, Ser-38 phosphorylation of SERCA has not been observed with the specific antibody.
In the SR experiments, a phosphoprotein of 100kDa was observed, when phosphorylation reactions were performed at high ATP concentration. This protein was not recognized by the SERCA PS-38 antibody. As such, we do not believe that this phosphoprotein is SERCA. Reddy et al. (14) has previously reported an SR phosphoprotein of 100 kDa, which was not SERCA. This phosphoprotein was restricted to junctional SR, whereas SERCA is distributed throughout junctional and longitudinal SR. The identity of the 100 kDa phosphoprotein was not described (14). SERCA phosphorylation has been observed by others (24 -30). These studies employ a variety of approaches (in particular immunoprecipitation and heterologous expression of SERCA mutants) to establish the identity of the phosphoprotein critically (24 -26, 30). In other cases the description of SERCA phosphorylation is made on the basis of coincident migration of SERCA and the phosphoprotein in a one-dimensional SDS-PAGE gel (27)(28)(29). This criterion is not sufficiently critical to enable identification of the phosphoprotein, and does not establish whether SERCA is a phosphoprotein.
Where the identification of SERCA as a phosphoprotein has been made using more rigorous criteria, the stoichiometry of SERCA phosphorylation reported is low (0.12 mol/mol, Ref. 26). It has been suggested that this might be explained in part due FIG. 6. SERCA phosphorylation in rat cardiac myocytes is not detected by antibody SERCA PS-38. Isolated rat ventricular myocytes were paced electrically at 0.5 Hz for 5 min unless otherwise stated. Cells (A, 10,000; B, 2,500; rod shaped cell count) were treated without further intervention (control lane 1) or with 1 M isoproterenol (lane 2); exposed to increased stimulation frequency (2.5 Hz; lane 3); exposed to 2.5 mM extracellular calcium (lane 4); or exposed to 1 M calyculin A (lane 5). A, calibration-38 (concentration range 0.04 -4 pmol) and rat myocytes were submitted to SDS-PAGE (10% polyacrylamide) and transferred to PVDF membrane. Blot was probed with antibody SERCA PS-38 (1: 5000) and visualized with goat anti-rabbit IgG peroxidase together with an enhanced chemiluminescent substrate (Su-perSignal West Femto Maximum Sensitivity Substrate, Pierce). B, in a parallel experiment, myocytes were separated by 15% SDS-PAGE and transferred to PVDF membrane. Threonine 17-phosphorylated phospholamban was detected using antibody PT-17 (1:5000). to significant existing phosphorylation (as is the case for RYR2, Ref. 38). Phosphorylation to these levels would result in the presentation of measurable amounts of Ser-38 phosphoprotein (1.2 pmol plus any pre-existing phosphoprotein in experiments employing 10 g of SR, Figs. 4 and 5). Such amounts would be readily detected in the present experiments, where detection sensitivity of 0.1 pmol (and lower) was observed. Our failure to observe SERCA recognition by the antibody described here, challenges previous conclusions that phosphorylation of Ser-38 of SERCA is a potent regulatory step of Ca 2ϩ transport (24,26).