α-Kinase Anchoring Protein αKAP Interacts with SERCA2A to Spatially Position Ca2+/Calmodulin-dependent Protein Kinase II and Modulate Phospholamban Phosphorylation*

The sarco-endoplasmic reticulum calcium ATPase 2a (SERCA2a) is critical for sequestering cytosolic calcium into the sarco-endoplasmic reticulum (SR) and regulating cardiac muscle relaxation. Protein-protein interactions indicated that it exists in complex with Ca2+/calmodulin-dependent protein kinase II (CaMKII) and its anchoring protein αKAP. Confocal imaging of isolated cardiomyocytes revealed the colocalization of CAMKII and αKAP with SERCA2a at the SR. Deletion analysis indicated that SERCA2a and CaMKII bind to different regions in the association domain of αKAP but not with each other. Although deletion of the putative N-terminal hydrophobic amino acid stretch in αKAP prevented its membrane targeting, it did not influence binding to SERCA2a or CaMKII. Both CaMKIIδC and the novel CaMKIIβ4 isoforms were found to exist in complex with αKAP and SERCA2a at the SR and were able to phosphorylate Thr-17 on phospholamban (PLN), an accessory subunit and known regulator of SERCA2a activity. Interestingly, the presence of αKAP was also found to significantly modulate the Ca2+/calmodulin-dependent phosphorylation of Thr-17 on PLN. These data demonstrate that αKAP exhibits a novel interaction with SERCA2a and may serve to spatially position CaMKII isoforms at the SR and to uniquely modulate the phosphorylation of PLN.

The sarco-endoplasmic reticulum calcium ATPase 2a (SERCA2a) is critical for sequestering cytosolic calcium into the sarco-endoplasmic reticulum (SR) and regulating cardiac muscle relaxation. Protein-protein interactions indicated that it exists in complex with Ca 2؉ /calmodulin-dependent protein kinase II (CaMKII) and its anchoring protein ␣KAP. Confocal imaging of isolated cardiomyocytes revealed the colocalization of CAMKII and ␣KAP with SERCA2a at the SR. Deletion analysis indicated that SERCA2a and CaMKII bind to different regions in the association domain of ␣KAP but not with each other. Although deletion of the putative N-terminal hydrophobic amino acid stretch in ␣KAP prevented its membrane targeting, it did not influence binding to SERCA2a or CaMKII. Both CaMKII␦ C and the novel CaMKII␤ 4 isoforms were found to exist in complex with ␣KAP and SERCA2a at the SR and were able to phosphorylate Thr-17 on phospholamban (PLN), an accessory subunit and known regulator of SERCA2a activity. Interestingly, the presence of ␣KAP was also found to significantly modulate the Ca 2؉ /calmodulin-dependent phosphorylation of Thr-17 on PLN. These data demonstrate that ␣KAP exhibits a novel interaction with SERCA2a and may serve to spatially position CaMKII isoforms at the SR and to uniquely modulate the phosphorylation of PLN.
The phosphorylation/dephosphorylation cycle is critical for controlling a diverse series of signaling processes in cell biology (1,2). Specificity of the phosphorylation/dephosphorylation event is in part achieved by selective employment of a protein kinase/phosphatase cascade and subcellular targeting (1,2). Both spatial and temporal specificity of signaling events is achieved by the compartmentalization of the signaling complexes through adaptor or anchoring proteins (1,2). Recent studies have highlighted novel aspects of integrating spatially and temporally the cAMP signaling cascades via a diverse family of protein kinase A anchoring proteins (AKAPs) 2 (3). The AKAPs are responsible for positioning the signaling complex via protein-protein interactions for effective and time-sensitive compartmentalization of the cAMP signal (4).
Although the intracellular targeting of protein kinase A to the effectors is being unraveled, little is known about the targeting of CaMKII activity, which is ubiquitously expressed and serves important roles in calcium signaling to guide synaptic transmission (2,5,6), gene transcription (7), cell growth (8), and excitation-contraction coupling (9 -11). Although four different isoforms of CaMKII (␣, ␤, ␦, and ␥) are expressed in a tissuespecific manner, cardiac tissue is shown to have predominance of CaMKII␦ C (cytosolic) and CaMKII␦ B (nuclear) isoforms, which serve roles in excitation-contraction coupling and cell growth, respectively (7,12). Studies have also revealed a significant level of a muscle-specific CaMKII ␤ isoform (CaMKII␤ 4 ) in skeletal and cardiac muscle (11,(13)(14)(15). In addition, the CAMK2A gene that encodes CaMKII␣ kinase in brain expresses an alternatively spliced non-kinase polypeptide designated ␣KAP in cardiac and skeletal muscle (14 -16). The ␣KAP has a unique amino acid stretch at the N terminus, which encodes a putative transmembrane domain, followed by the association domain of CaMKII␣. The association domain in the CaMKII gene family is a common feature important for oligomerization (15)(16)(17).
␣KAP is believed to be targeted to the SR membrane in skeletal muscle via the N-terminal hydrophobic sequence and has been proposed to recruit the muscle-specific CaMKII␤ 4 through dimerization with the association domain and regulate calcium transport (15). Data also suggest that ␣KAP along with the novel CaMKII␤ 4 are enriched in cardiac SR membranes implying a common regulatory role for these molecules in these two muscle types (13)(14)(15). Further, studies suggest a significant level of a muscle-specific CaMKII ␤ isoform (CaMKII␤ 4 ) in cardiac and skeletal muscle (14 -16). In addition, the CAMK2A gene that encodes CaMKII␣ kinase in the brain expresses an alternatively spliced non-kinase polypeptide designated ␣KAP in cardiac and skeletal muscle (14 -16). The ␣KAP has a unique amino acid stretch at the N terminus, which encodes a putative transmembrane domain, followed by the association domain of CaMKII␣. The association domain in the CaMKII gene family is a common feature important for oligomerization (15)(16)(17).
␣KAP is believed to be targeted to the SR membrane in skeletal muscle via the N-terminal hydrophobic sequence and has been proposed to recruit the muscle-specific CaMKII␤ 4 through dimerization with the association domain and regulate SR function (15). Data also suggest that ␣KAP along with the novel CaMKII␤ 4 are enriched in cardiac SR membranes, implying a common regulatory role for these molecules in these two muscle types (13)(14)(15). Further, studies suggest that CaMKII␤ 4 can recruit the glycolytic machinery to the SR membrane in cardiac and skeletal muscle and potentially serve to spatially modulate the supply of ATP for the calcium transport process (13,14). In view of the emerging concept of spatial and temporal control of signal transduction through kinase-anchoring proteins, we investigated further the role of ␣KAP at the SR membrane and found that it directly interacts with the calcium ATPase and serves to recruit CaMKII isoforms and modulate the phosphorylation of PLN at Thr-17, which is known to critically regulate calcium uptake and muscle relaxation (18). We propose a model in which ␣KAP would serve to modulate PLN phosphorylation and integrate the spatial and temporal control on calcium transport through direct binding to the calcium ATPase on the one hand and recruitment of CaMKII activity on the other.

EXPERIMENTAL PROCEDURES
Expression Constructs of CaMKII␦C, CaMKII␤C, ␣KAP, SERCA2a, and Phospholamban-The cloning of CaMKII␤ 4 and ␣KAP from myocardium has been described previously (14). These constructs were subcloned in frame either into pcDNA3-six-Myc, pcDNA3-GFP, or pGEX-2TK. SERCA2a, CaMKII␦ C , and PLN were cloned by reverse transcription-PCR. In brief, total RNA was extracted from a mouse heart using the Tripure isolation kit (Roche Applied Science). First strand cDNA was obtained using oligo(dT) primer and reverse transcriptase (Invitrogen). Specific primers were designed for SERCA2a, CaMKII␦ C , and phospholamban. PCR was performed using cDNA as a template and Platinum PCR Super-Mix (Invitrogen). The PCR products were cloned into either pcDNA3-six-Myc, pcDNA3-GFP, or pGEX-2TK and sequenced by an automated ABI sequencer using M13 and T7 sequencing primers, and sequences were analyzed with Seqaid II (University of Kansas) and BLAST.
Cell Culture and Transfection-HeLa cells were maintained at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Transfection was performed with FuGENE-HD (Roche Applied Science) according to the manufacturer's instructions. Primary culture of mouse cardiomyocytes was carried out following the isolation procedure described (19). Hearts from neonatal mice were washed with suspension minimum essential medium (Invitrogen), sliced into small pieces with scissors, and transferred into a cell dispenser containing 0.1% trypsin at 37°C. Enzymatic treatment was allowed for 30 min, and the yielded cells were spun down and collected in Hanks minimum essential medium (Invitrogen) supplemented with 5% fetal bovine serum. Enzymatic digestion was repeated three more times, and isolated cells were cultured on dishes. After 50 min of plating, the nonadherent cells, mostly consisting of cardiomyocytes, were transferred onto new dishes, and this procedure was repeated three more times to enrich for cardiomyocytes.
GST Fusion Protein Expression-Escherichia coli BL21 containing recombinant proteins were shaken in a special medium (Peptone 20, yeast extract 10, and NaCl, 7 g/liter) at 37°C. When the A 595 reached 0.5, the temperature was cooled to 28°C, and cells were induced by 0.1 mM 1-thio-␤-D-galactopyranoside and maintained at 28°C for 4 h. Then cells were pelleted by centrifugation and lysed by sonication in phosphatebuffered saline buffer containing 1% Nonidet P-40. Debris was removed by centrifugation, and supernatant was saved. The fusion proteins were isolated on glutathione-Sepharose beads 4B (GE Healthcare) by incubating the lysate for 60 min and then washed four times with the same buffer.
GST Pull-down and Calmodulin Binding-Mouse hearts were homogenized with TBS buffer (50 mM Tris, pH 7.5, 150 mM NaCl, and 1 mM EDTA). The homogenate was centrifuged at 7500 ϫ g for 20 min, and supernatant was transferred to new tubes and centrifuged at 100,000 ϫ g for 1 h at 4°C and considered as the crude SR fraction. The latter was solubilized with 0.5% Nonidet P-40 and 0.8% CHAPS for 2 h at 4°C and centrifuged for 20 min at 10,000 ϫ g, and the supernatant was retained as solubilized SR. For GST pull-down assays, GST alone and GST-␣KAP fusion proteins (10 g) were incubated with solubilized SR fraction (500 g) in 1 ml of TBS buffer for 2 h at 4°C. Beads were washed four times in TBS buffer, 2ϫ gel loading buffer was added, and bound proteins were resolved on SDS-PAGE. For isolation of CaM-binding proteins, detergentsolubilized SR fraction (500 g) was applied to a CaM-Sepharose column in the presence of 2 mM CaCl 2 and then washed with four volumes of buffer before elution with 5 mM EGTA in the same buffer but without CaCl 2 (20). Protein samples were resolved on SDS-PAGE and transferred onto polyvinylidene difluoride membrane and immunoblotted with different antibodies or stained with Coomassie Blue or silver stain to visualize protein bands that were further sequenced by liquid chromatography-MS/MS analysis (LTQ) at the Proteomics Resource Centre at the University of Ottawa. Protein samples were also analyzed by MALDI-TOF techniques at the proteomics facility at Queens University.
Immunofluorescence-Cells transfected with different cDNA expression constructs were fixed with 4% paraformaldehyde for 10 min and then washed four times with phosphate-buffered saline. These were incubated with anti-Myc antibody (Sigma) for 2 h, washed four times, and then incubated with secondary antibody (Alexa Fluor 594; Invitrogen) for 40 min, washed four times, and mounted in Vectashield mounting medium. The cells were observed under a Zeiss LSM 500 confocal laser-scanning system attached to a Zeiss Axiovert 200 M inverted microscope (Carl Zeiss). Cardiomyocytes were fixed in a similar manner as above; immunostained with anti-CaMKII␤, anti-␣KAP, and anti-SERCA2a antibodies; and analyzed by confocal microscopy.
Immunoprecipitation (IP) Assays-IP assay was carried out using c-Myc-agarose (catalog number A7470; Sigma) according to the manufacturer's protocol. Briefly, cells transfected in a 10-cm culture dish were scraped and lysed in TBS buffer. The resulting lysate was centrifuged at 12,000 ϫ g at 4°C for 10 min.
The supernatant was collected, and protein was measured by a Bradford kit (Bio-Rad). Myc-agarose and IgG-agarose beads (control) were washed four times with TBS buffer, and 250 g of cell lysate was incubated with 50 l of washed Myc-agarose beads. The mixture was centrifuged at 4°C for 60 min, and beads were collected and washed (four times) with 1 ml of cold TBS buffer for 10 min each time. The precipitated proteins were resolved on SDS-PAGE (10% gel) and immunoblotted with different antibodies as described above.
Deletion Mutations-Deletion constructs of ␣KAP were constructed with PCR using selected primers in which a stop codon (TGA) was introduced at the desired locations of the ␣KAP sequence. The PCR products generated were inserted into pcDNA3-six-Myc or pcDNA3-GFP and sequenced as described above to confirm identity.
Phosphorylation Assays-Phosphorylation of SERCA2a was carried out by co-transfecting SERCA2a-GFP and ␣KAP-Myc with either CaMKII␦ C -Myc or CaMKII␤ 4 -Myc. IP reactions were performed with Myc-agarose as described above, and precipitated proteins were incubated in assay buffer (20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 1 mM sodium orthovandate, 1 mM dithiothreitol, either 1 mM CaCl 2 plus 100 nM calmodulin or 5 mM EGTA, and [␥-32 P]ATP (0.4 ϫ 10 Ϫ5 cpm)) for 2 min at room temperature; the reaction was terminated with 50 l of SDS gel loading buffer; proteins were resolved on SDS-PAGE and gel-dried; and phosphoproteins were detected on x-ray films.
Phosphorylation of PLN was studied in the presence or absence of purified ␣KAP-GST (10 g). In brief, purified GST-PLN (20 g) was incubated with either GST-CaMKII␦ C or GST-CMKII␤ 4 (10 g). The control reactions include GST alone (20 g) instead of GST-PLN. The mixture of proteins was incubated in the phosphorylation assay buffer (as above except that hot ATP was replaced with cold ATP) for 2 min at 30°C and then terminated with 50 l of SDS gel loading buffer. Proteins were resolved on SDS-PAGE, transferred onto polyvinylidene difluoride membrane, and immunoblotted with anti-CaMKII polyclonal antibody or phospho-PLN antibody (Badrilla, UK) or anti-PLN antibody (Affinity Bioreagent).

RESULTS
␣KAP Exists in a Complex with SERCA2a-Previous data have shown that ␣KAP is a component of the SR membrane in skeletal and cardiac muscle (13)(14)(15). In order to define interacting components of ␣KAP, GST pull-down assays were performed with a detergent-solubilized SR fraction from cardiac muscle. The potential interacting proteins were revealed with Coomassie Blue staining. Characterization of the interacting proteins with respect to molecular mass revealed that ␣KAP specifically binds polypeptides of ϳ110, 86, 83, 74, 71, 52, 34, and 23 kDa (Fig. 1A, lane 2). The presence of these polypeptides was not detected in the pull down with GST alone (Fig. 1A, lane  4). Further, recombinant GST-␣KAP (Fig. 1A, lane 1) or GST alone (Fig. 1A, lane 3) in TBS buffer also lacked the presence of these polypeptides. The recombinant GST-␣KAP appeared as a 47-kDa form (mature form) and a 32-kDa truncated product (lanes 1 and 2), and GST alone appeared as 25 kDa on SDS gels (lanes 3 and 4).
In order to identify the specifically bound polypeptides, gel bands were excised and sequenced and analyzed by liquid chromatography-MS/MS. This revealed that ␣KAP exists in complex with a diverse array of proteins (Table 1), including aralar, aconitase 2, acyl-coenzyme A dehydrogenase, hydroxyacyl-Coenzyme A dehydrogenase, SERCA2a, CaMKII, and glyceraldehyde-3-phosphate dehydrogenase. Since CaMKII, ␣KAP, and SERCA2a are enriched in the SR, we further investigated their potential interactions. The GST pull-down experiments noted above were repeated, and the polypeptides present were probed in immunoblots with an anti-SERCA2a monoclonal antibody, which clearly identified a 110-kDa polypeptide in the GST-

The SERCA2a, ␣KAP, CaMKII, and CaM complex in SR-
The CaMKII activity of the SR can be isolated on calmodulin affinity columns (20). We reasoned that if SERCA2a, ␣KAP, and CaMKII exist in a complex, these proteins would co-purify together on CaM-Sepharose. CaM-Sepharose 4B was incubated with detergent-solublized SR and CaM-binding proteins eluted with EGTA, resolved in SDS-PAGE, and silver-stained (20). CaM-Sepharose specifically binds to a multitude of polypeptides of different molecular mass (Fig. 1C, lane 2), and analysis with either MALDI-TOF or liquid chromatography-MS/MS revealed that the ϳ450 kDa band was the ryanodine receptor, the 110 kDa band was SERCA2a, the 86 kDa band was 6-phosphofructose kinase, and the 23 kDa band was ␣KAP (Table 2). Further, immunoblotting of the EGTA-eluted proteins with a polyclonal anti-CaMKII antibody (Fig. 1D, bottom, lane 2) recognized ϳ70-, 60-, 52-, 54-, and 23-kDa polypeptides, which we have previously shown to represent the CaMKII␤ 4 , CaMKII␤, CaMKII␦, and ␣KAP, respectively (13,14,20). The anti-SERCA2a antibody recognized a 110-kDa polypeptide (Fig. 1D, top, lane 2), further confirming the specific retention of the calcium ATPase on the CaM affinity column. Since both SERCA2a and ␣KAP lack calmodulin binding motifs (14,16,21), their retention on the CaM affinity column could only be due to an association with other CaM-binding proteins, such as CaMKII or the ryanodine receptor (20,22). The Sepharose 4B exhibited some nonspecific binding of polypeptides (Fig. 1C, lane 1), but these appeared mostly distinct in terms of molecular mass from those retained specifically on the CaM-Sepharose 4B column and did not exhibit any cross-reactivity with anti-CaMKII or anti-SERCA2a antibodies (Fig. 1D, lane 1).
␣KAP Is a Common Binding Partner of SERCA2a and CaMKII␤ 4 -We further studied the interaction of ␣KAP with SERCA2a and CaMKII␤ 4 by transfection of expression con-structs in mammalian cells. In a series of experiments, cDNA of mouse SERCA2a fused to GFP was co-expressed with six-Myctagged ␣KAP in HeLa cells, and IP was carried out on cell lysates with anti-Myc. Western blot analysis with anti-SERCA2a monoclonal antibody revealed that SERCA2a-GFP was detected in the IP reactions of cell lysates containing ␣KAP-Myc ( Fig. 2A, middle, lane 1) and not in the reactions of cell lysates containing control pcDNA3-six-Myc vector ( Fig.  2A, middle, lane 2). The top panel in the same figure shows staining with anti-SERCA2a antibody and confirms the relative level of expression of SERCA2a protein in test and control cell lysates ( Fig. 2A, top, lanes 1 and 2). Similarly, the bottom panel in the figure shows staining with anti-Myc, which identified the presence of ␣KAP-Myc in cell lysates ( Fig. 2A, bottom, lane 1). The control experiment consisting of pcDNA3-Myc alone ( Fig.  2A, bottom, lane 2) does not show any protein band, because of the very small size of the six-Myc tag.
The samples from the IP reactions noted above were also immunoblotted with anti-GFP monoclonal antibody (Fig. 2B), which identified the expression of SERCA2a protein and con-   firmed the data obtained with the anti-SERCA2a antibody. In these sets of experiments, SERCA2a-GFP was only detected in the IP reactions with ␣KAP-Myc (Fig. 2B, middle, lane 1) and not in control pcDNA3-Myc vector (Fig. 2B, middle, lane 2). The top panel in the figure shows the expression of SERCA2a-GFP in test (Fig. 2B, top, lane 1) and control (Fig. 2B, top, lane 2) cell lysates, as identified with anti-GFP staining. In the bottom panel, anti-CaMKII␣ antibodies confirm the expression of ␣KAP in the cell lysates (Fig. 2B, bottom, lane 1). Notably, SERCA2a-GFP appeared as a doublet in this experiment. The presence of two protein bands for SERCA2a from expression constructs have been previously noted as well (23,24). We then examined the interactions between ␣KAP and CaMKII by co-transfection studies of HeLa cells with ␣KAP-Myc and CaMKII␤ 4 -GFP expression constructs. Cell lysates from mock (Fig. 2C, lane 1) and CaMKII␤ 4 -GFP plus ␣KAP-Myc-transfected (Fig. 2C, lane 2) cells were examined for the expression of these proteins. The IP reactions were carried out with anti-Myc antibodies and detected with anti-CaMKII␤ antibody. This showed the presence of CaMKII␤ 4 in the immunoprecipitates of ␣KAP-transfected cell lysates only (Fig. 2C, top, lane 4) and not in mock-transfected cell lysates (Fig. 2C,  top, lane 3). The IP reactions were also stained with anti-GFP antibodies, which confirmed the interaction between ␣KAP and CaMKII␤ 4 (Fig. 2C, second panel, lane 4). The third and fourth panel in same figure show staining with anti-␣KAP and anti-Myc antibodies, respectively, which identified an ␣KAP band in test lysates (Fig. 2C, lane 2) and in the IP reactions of test lysates (Fig. 2C, lane 4) but not in control lanes (Fig. 2C,  lanes 1 and 3).
We further visualized any co-localization of ␣KAP with SERCA2a and CaMKII␤ 4 by immunocytochemical staining of expressed proteins in HeLa cells as well as their endogenous distribution in primary cultures of cardiomyocytes. Confocal microscopy of HeLa cells transiently transfected with ␣KAP-Myc (red) and SERCA2a-GFP (green) shows that ␣KAP and SERCA2a appear on intracellular structures with a perinuclear staining, which is a characteristic feature of the endoplasmic reticulum membrane (Fig. 2D). Additionally, some ␣KAP staining also appeared on other intracellular structures. The SERCA2a staining appears to completely overlap with ␣KAP, although ␣KAP was also noted in additional locations. Immunocytochemical staining of cells transfected with ␣KAP-Myc and CaMKII␤ 4 -GFP showed a similar pattern of co-localization (Fig. 2E). Both ␣KAP-Myc (red) and CaMKII␤ 4 -GFP (green) staining were mostly concentrated in reticular structures around the nucleus, although the staining was also noted to spread through out the cell.
The subcellular localization of endogenous ␣KAP, SERCA2a, and CaMKII␤ 4 was also examined in mouse cardiomyocytes with confocal imaging (Fig. 3). The cardiomyocytes appeared as binucleated cells with a typical rectangular shape, and immunostaining with anti-␣KAP (red) and anti-SERCA2a (green) showed that these proteins are distributed in a punctuate fashion on reticular structures that most probably represent subcellular membranes in these cells. Although in a merged image, a significant overlap of the two signals (yellow) indicates colocalization, some regions with distinct distribution for ␣KAP (red) and SERCA2a were also evident (Fig. 3A). Immunostaining with anti-SERCA2a (red) and anti-CaMKII␤ 4 (green) shows the presence of the two proteins on reticular structures throughout the cell and concentrated in the perinuclear region (Fig. 3B). A merge of the images indicates some co-localization of SERCA2a and CaMKII␤ 4 in cardiomyocytes, but there are clear regions of distinct distribution for the two molecules as well.
Distinct Regions in ␣KAP Interact with SERCA2a and CaMKII␤ 4 -To assess any direct interactions between ␣KAP and SERCA2a or CaMKII␤ 4 , we made a series of deletion mutants of ␣KAP in pcDNA3-six-Myc (Fig. 4, A and B, top). Full-length ␣KAP, which encodes 200 amino acids, or its different deletion constructs were co-expressed with SERCA2a-GFP (Fig. 4A) or CaMKII␤ 4 -GFP (Fig. 4B). IP assays were performed on cell lysates with anti-Myc, and the precipitated proteins were analyzed by Western blots using anti-GFP. The ␣KAP-Myc could immunoprecipitate either SERCA2a (Fig. 4A) or CaMKII (Fig. 4B), and this implies a direct association of the two proteins with ␣KAP. The minimal sequence of ␣KAP that is required for the interaction with SERCA2a resides in amino acids 1-156 (Fig. 4A, bottom), since binding with SERCA2a is restored with additional c-terminal sequences in ␣KAP. The middle panel in Fig. 4A shows the expression level of SERCA2a in the different transfected cell lysates.
The interaction of ␣KAP with CaMKII␤ 4 is mediated by N-terminal sequences of ␣KAP as well (Fig. 4B). The mutant with amino acids 1-149 showed full interaction with CaMKII␤ 4 , but deleting additional amino acids substantially reduced the interaction between the two proteins, and the mutant with fewer than amino acids 1-83 did not interact with CaMKII␤ 4 (Fig. 4B, bottom). The middle panel in Fig. 4B shows expression of CaMKII␤ 4 in different deletion mutant cell lysates.

␣KAP Interacts with CaMKII␦ C and CaMKII␤ 4 -Since
CaMKII␦ is the most studied CaMKII isoform in cardiac tissues, we sought to determine if ␣KAP can bind CaMKII␦ as well. We co-expressed full-length ␣KAP-six-Myc and either CaMKII␤ 4 -GFP or CaMKII␦ C -GFP in HeLa cells. An IP reaction was performed with anti-Myc, and the analysis of the immunoprecipitated protein with anti-GFP showed that ␣KAP can interact with both CaMKII␤ 4 (Fig. 5A, top, lane 2) and CaMKII␦ C (Fig. 5A, top, lane 5). The control reaction involving mock-transfected cells did not show any polypeptide in those size ranges (Fig. 5A, top, lane 6). Lysates were also loaded on the gel to check the level of expressed proteins (5A, top) in cell lysates for CaMKII␤ 4

(lane 1), CaMKII␦ C (lane 4), and mocktransfected cells (lane 3).
The association domains of CaMKII␣ and ␣KAP are identical and share a high degree of homology with other CaMKII isoforms. Since ␣KAP interacts with SERCA2a through a part of its association domain, we examined if other CaMKII isoforms can also directly interact with SERCA2a. HeLa cells were transfected with SERCA2a-GFP and co-transfected with either ␣KAP-Myc or CaMKII-Myc isoforms (␦ C or ␤ 4 ). IP assays on cell lysates were performed with anti-Myc antibody and subjected to SDS-PAGE and analyzed on Western blots with anti-GFP. Data revealed that only ␣KAP can immunoprecipitate SERCA2a (Fig. 5B, middle, lane 2), whereas CaMKII␦ C (Fig. 5B,  middle, lane 3), CaMKII␤ 4 (Fig. 5B, middle, lane 4), or control reactions with mock-transfected cell lysates (Fig. 5B,  Transmembrane Domain in ␣KAP Is Critical for Membrane Targeting but Not SERCA2a Binding-␣KAP is thought to consist of an N-terminal transmembrane (TM) domain, followed by a nuclear localization signal (NLS) and the association domain (AD) of CaMKII␣ (Fig. 6A) (16). We examined the possible roles of the putative transmembrane domain to target ␣KAP to intracellular membranes and in the interaction with SERCA2a. A deletion mutant of ␣KAP that lacked the N-terminal 23 amino acids that are predicted to constitute the TM domain of ␣KAP (Fig. 6A) tagged with Myc (␣KAP ⌬TM -Myc) and wild type ␣KAP-Myc were transfected into HeLa cells along with SERCA2a-GFP. IP assays were performed on the cell lysates with anti-Myc, and SDS-PAGE, followed by Western blotting with anti-GFP, revealed that the ␣KAP ⌬TM has no significant effect on its SERCA2a binding ability (Fig. 6B, lane 4) compared with the ability of wild type ␣KAP to bind SERCA2a (lane 2). Fig. 6B also shows the expression of SERCA2a-GFP and wild type ␣KAP-Myc (lane 1) and SERCA2a-GFP and ␣KAP ⌬TM -Myc (lane 3) in cell lysates.
In a series of experiments to study targeting, ␣KAP was fused to GFP, and cells transfected with ␣KAP-GFP and ␣KAP ⌬TM -GFP were visualized by confocal microscopy. Cells transfected with ␣KAP-GFP showed reticular membranous type association, as evident from the punctate appearance without any nuclear localization (Fig. 6C, top panels). On the other hand, the deletion mutant ␣KAP ⌬TM -GFP exhibited a smooth non-reticular appearance throughout the cell, including localization in the nuclei that were identified with 4Ј,6-diamidino-2-phenylindole (blue) staining (Fig. 6C, bottom panels). CaMKII␤ 4 Phosphorylates Phospholamban-It is apparent from the findings described above that CaMKII␤ 4 is a novel   1 and 2) or CaMKII␤ 4 -GFP (lanes 4 and 5) or mock control (lanes 3 and 6). cardiac isoform of CAMKII that is targeted to the SR and SERCA2a through ␣KAP. We sought to determine the ability of CaMKII␤ 4 to phosphorylate phospholamban, which is a known regulator of SERCA2a activity and muscle relaxation (18,(25)(26)(27)(28). The ability of CaMKII (␦ C or ␤ 4 ) to phosphorylate SERCA2a or its subunit phospholamban was examined in HeLa cells that were co-transfected with SERCA2a-GFP, ␣KAP-Myc, and CaMKII-Myc isoforms (␦ C or ␤ 4 ). IP assays were performed with anti-Myc, and the precipitated proteins were subjected to calcium/CaM-dependent phosphorylation, as described under "Experimental Procedures." The phosphoproteins were separated in SDS-PAGE and visualized by autoradiography, which showed the presence of autophosphorylated polypeptides of ϳ64 kDa (CaMKII␦ C -Myc) and ϳ84 kDa (CaMKII␤ 4 -Myc) when phosphorylation assays were conducted in the presence of Ca 2ϩ /CaM (Fig. 7A, lanes 2 and 4) but not when phosphorylation was carried out in 5 mM EGTA in the absence of Ca 2ϩ /CaM (Fig. 7A, lanes 1 and 3). It is notable that no phosphorylation of the SERCA2a-GFP (ϳ140-kDa polypeptide), which was co-expressed in these cells, was detected due to CaMKII␦ C or CaMKII␤ 4 activity under these conditions. The CaMKII␦ C is known to phosphorylate Thr-17 on phospholamban and stimulate SERCA2a (18). We examined whether CaMKII␤ 4 can also serve to phosphorylate phospholamban by co-expressing PLN together with CaMKII␦ C or CaMKII␤ 4 as GST fusion proteins and assaying for the Ca 2ϩ / CaM-dependent phosphorylation of PLN as described above. Western blot analysis of phosphorylated proteins with anti-PLN phospho-Thr-17 revealed that both CaMKII␦ C and CaMKII␤ 4 are able to stimulate phosphorylation of PLN in a Ca 2ϩ /CaM-dependent manner (Fig. 7B, middle, lanes 11 and 12, respectively) compared with the EGTA control containing CaMKII␦ C (Fig. 7B, middle, lane 5) and CaMKII␤ 4 (Fig. 7B,  middle, lane 6). Phosphorylation of PLN by CaMKII␦ C in the  presence of EGTA (lane 5) may be due to the high level of expression of this fusion protein. A lower molecular size band was also evident and most likely represents a nonspecific reactivity of anti-PLN-phospho-Thr-17, since it was present in the GST control (lanes 4 and 10) as well. The immunoblot was stripped and stained with a PLN antibody to identify total PLN in the reactions (Fig. 7B, bottom). The top panel shows the immunoblot with anti-CaMKII to define the expression level of CaMKII␦ C and CaMKII␤ 4 . Buffer controls for any potential background or nonspecific immunostaining with various antibodies are also shown (lanes 1, 2, 7, and 8).
␣KAP Modulates Phospholamban Phosphorylation-In order to assess whether ␣KAP can modulate PLN phosphorylation, PLN, CaMKII␦ C , and CaMKII␤ 4 were expressed as GST fusion proteins and subjected to phosphorylation in the presence and absence of purified ␣KAP. Western blot analysis of phosphoproteins with anti-PLN phospho-Thr-17 was used to determine any effects on phospholamban phosphorylation. Fig.  8 shows purified CaMKII␦ C , PLN, and ␣KAP incubated in a phosphorylation assay with EGTA (lane 1) or Ca 2ϩ /CaM (lanes 2-4). The top panel (Fig. 8A) shows the expression of CaMKII␦ C (lanes 1-4), as assessed by immunoblotting with anti-CaMKII␦ C , and ␣KAP expression (lanes 1-3), as assessed by immunoblotting with anti-␣KAP. The middle panel (Fig.  8A) shows immunoblotting with anti-PLN phospho-Thr-17, and the bottom panel (Fig. 8A) shows the immunoblot with anti-PLN for loading control. Ca 2ϩ /CaM clearly activated CaMKII␦ C to stimulate phosphorylation of PLN (lane 4) compared with that in the presence of EGTA (lane 1). However, Ca 2ϩ /CaM-dependent PLN phosphorylation (lane 4) was markedly inhibited by the presence of ␣KAP (lane 2). The bottom panel (Fig. 8B) shows the expression of CaMKII␤ 4 ( lanes  1-4), as assessed by immunoblotting with anti-CaMKII␤, and ␣KAP expression (lanes 1-3), as assessed by immunoblotting with anti-␣KAP. The middle panel (Fig. 8B) shows immunoblotting with anti-PLN phospho-Thr-17, and the bottom panel

DISCUSSION
The data here show that ␣KAP can exist in a complex with CaMKII isoforms, SERCA2a, and CaM at the cardiac SR. ␣KAP directly binds SERCA2a as well as recruits CaMKII␦ C and CaMKII␤ 4 isoforms, which are prominent in cardiac tissue. Furthermore, ␣KAP can modulate PLN phosphorylation at Thr-17, which is known to regulate SERCA2a activity and SR function (18,(25)(26)(27)(28). SERCA2a binds directly to regions in the association domain of ␣KAP that are distinct from those that associate with CaMKII. Although the removal of the putative transmembrane domain of ␣KAP disengaged it from the SR membrane, as reported (15,30), it did not effect its interaction with SERCA2a or CAMKII, since the association domain of ␣KAP containing the binding sites is extramembranous. Further more, the CaMKII isoforms did not exhibit direct interactions with SERCA2a, indicating a central role for the ␣KAP interactions in positioning these enzymes at the SR membrane.
The CaMKII␦ C and CaMKII␤ 4 were both effective in phosphorylating PLN on Thr-17, and the presence of ␣KAP was found to down-regulate this phosphorylation event. The phosphorylation of PLN on Thr-17 due to CaMKII␦ C or at Ser-16 by protein kinase A has been positively correlated with an increase in SERCA2a activity, calcium uptake, and the rate of cardiac muscle relaxation (25)(26)(27)(28). The ability of CaMKII␤ 4 isoform to exist in a complex with ␣KAP and SERCA2a and phosphorylate Thr-17 on PLN suggests that this kinase would serve a physiological role in cardiac muscle function as well. In this regard, recent studies indicate that mice that lack CaMKII␦ C in the phospho-Thr-17 phosphorylation due to Ca 2ϩ /CaM-dependent CaMKII activity was quantified by densitometry and is shown as percentage of phosphorylation of that in the absence (Ϫ␣KAP) of any ␣KAP (control). Values presented are mean Ϯ S.E. of three independent experiments. The asterisks denote statistical significance with respect to control; **, p Ͻ 0.01. myocardium exhibit normal physiological function and pathological response to pressure overload (29). In view of our data on the expression and targeting of CaMKII␤ 4 activity to cardiac SR, we suggest that this CaMKII isoform could substitute for CaMKII␦ C and potentially contribute to normal cardiac biology and the acute response to stress in the CaMKII␦ C null mice (29). Further, CaMKII␤ 4 can bind glycolytic enzymes and regulate glyceraldehyde-3-phosphate dehydrogenase at the SR and has been implicated in the control of local ATP production to support calcium uptake (13,14). Thus, the data here on CaMKII␤ 4 and its associations at the SR membrane suggest an important and underappreciated role for this CaMKII isoform in cardiac function. It is noted that the CaMKII␦ C isoform has been studied in detail, and its involvement in excitation-contraction coupling, cardiac growth, and dysfunction has been clearly implicated (28). Although our data show that ␣KAP can target both CaMKII␦ C and CaMKII␤ 4 isoforms to the SR to modulate PLN phosphorylation, it is apparent that neither of these bind directly to SERCA2a; nor was SERCA2a a substrate of these protein kinases. Thus, the direct association of ␣KAP with SERCA2a described here may be critical for positioning CaMKII activity for the modulation of calcium uptake via phosphorylation of PLN.
Collectively, these findings position ␣KAP as a membrane protein that interacts with SERCA2a on the one hand and CaMKII on the other. In this sense, ␣KAP acts as a scaffold and an adaptor to promote the spatial positioning of these proteins to facilitate the modulation of SERCA2a function through PLN phosphorylation by CaMKII activity at the SR. Based upon these observations, we propose a model of a CaMKII-␣KAP-SERCA2a-PLN complex at the SR membrane, which will convey calcium and CaM sensitivity to the calcium transport mechanism in a localized and temporal manner (Fig. 9). Thus, when the cytosolic Ca 2ϩ concentration rises due to calcium release from the SR, it would bind CaM and activate anchored CaMKII activity to phosphorylate PLN and relieve the inhibition on SERCA2a and stimulate calcium sequestration into the SR and muscle relaxation (25)(26)(27)(28). Additionally, ␣KAP can itself modulate the level of PLN phosphorylation on Thr-17 and further fine tune SR function (Fig. 9). Although ␣KAP has been shown to assemble with and support ␣-CaMKII activity (15), how it can modulate CaMKII␦ C and CaMKII␤ 4 and the level of PLN phosphorylation on Thr-17 at the SR membrane remains to be defined.
It is notable that AKAP is critical for the spatial and temporal control of cAMP signaling, and a large repertoire of AKAPs have been described that function as scaffolds for targeting cAMP signaling enzymes to distinct subcellular sites (3,4). Cardiac and skeletal muscle tissue have been reported to express many distinct AKAPs, some of which are critical in the anchoring of protein kinase A to the effectors, such as the ryanodine receptor (31), L-type calcium channel (32), or PLN (33) and regulate contraction and relaxation (34,35). We noted that ␣KAP can associate with membrane proteins of the mitochondria and peroxisomes in myocardium, and confocal imaging demonstrated that ␣KAP exhibits distinct subcellular locations within the cardiomyocyte. ␣KAP may therefore be a part of different intracellular mem-brane complexes that tether unique CaMKII isoforms to effectors to influence a variety of cellular events and support the multifunctional nature of CaMKII activity (2). Given the diverse emerging role of the AKAPs in cAMP signaling cascades (36), it is conceivable that ␣KAP may provide an analogous scaffold/adaptor for the spatial and temporal control of Ca 2ϩ signaling at distinct subcellular locations. In this regard, the calcium release channel is also a substrate for CaMKII (29), and the potential role of ␣KAP in its targeting and regulation is also under investigation. FIGURE 9. Modeling the assembly of ␣KAP with SERCA2a, PLN and CaMKII at the cardiac SR. ␣KAP is targeted to the SR membrane via its N-terminal transmembrane domain and directly interacts with SERCA2a as well as CaMKII␦ C or -␤ 4 isoforms. In addition, ␣KAP can modulate the Ca 2ϩ /CaM dependent phosphorylation of PLN at threonine 17, which is known to regulate calcium uptake into the SR and muscle relaxation in response to calcium signals. Thus, ␣KAP serves a dual role of targeting and modulating the calcium transport process.