Group VIA Phospholipase A2 Forms a Signaling Complex with the Calcium/Calmodulin-dependent Protein Kinase IIβ Expressed in Pancreatic Islet β-Cells*

Insulin-secreting pancreatic islet β-cells express a Group VIA Ca2+-independent phospholipase A2 (iPLA2β) that contains a calmodulin binding site and protein interaction domains. We identified Ca2+/calmodulindependent protein kinase IIβ (CaMKIIβ) as a potential iPLA2β-interacting protein by yeast two-hybrid screening of a cDNA library using iPLA2β cDNA as bait. Cloning CaMKIIβ cDNA from a rat islet library revealed that one dominant CaMKIIβ isoform mRNA is expressed by adult islets and is not observed in brain or neonatal islets and that there is high conservation of the isoform expressed by rat and human β-cells. Binary two-hybrid assays using DNA encoding this isoform as bait and iPLA2β DNA as prey confirmed interaction of the enzymes, as did assays with CaMKIIβ as prey and iPLA2β bait. His-tagged CaMKIIβ immobilized on metal affinity matrices bound iPLA2β, and this did not require exogenous calmodulin and was not prevented by a calmodulin antagonist or the Ca2+ chelator EGTA. Activities of both enzymes increased upon their association, and iPLA2β reaction products reduced CaMKIIβ activity. Both the iPLA2β inhibitor bromoenol lactone and the CaMKIIβ inhibitor KN93 reduced arachidonate release from INS-1 insulinoma cells, and both inhibit insulin secretion. CaMKIIβ and iPLA2β can be coimmunoprecipitated from INS-1 cells, and forskolin, which amplifies glucose-induced insulin secretion, increases the abundance of the immunoprecipitatable complex. These findings suggest that iPLA2β and CaMKIIβ form a signaling complex in β-cells, consistent with reports that both enzymes participate in insulin secretion and that their expression is coinduced upon differentiation of pancreatic progenitor to endocrine progenitor cells.


Insulin-secreting pancreatic islet ␤-cells express a
Group VIA Ca 2؉ -independent phospholipase A 2 (iPLA 2 ␤) that contains a calmodulin binding site and protein interaction domains. We identified Ca 2؉ /calmodulindependent protein kinase II␤ (CaMKII␤) as a potential iPLA 2 ␤-interacting protein by yeast two-hybrid screening of a cDNA library using iPLA 2 ␤ cDNA as bait. Cloning CaMKII␤ cDNA from a rat islet library revealed that one dominant CaMKII␤ isoform mRNA is expressed by adult islets and is not observed in brain or neonatal islets and that there is high conservation of the isoform expressed by rat and human ␤-cells. Binary two-hybrid assays using DNA encoding this isoform as bait and iPLA 2 ␤ DNA as prey confirmed interaction of the enzymes, as did assays with CaMKII␤ as prey and iPLA 2 ␤ bait. His-tagged CaMKII␤ immobilized on metal affinity matrices bound iPLA 2 ␤, and this did not require exogenous calmodulin and was not prevented by a calmodulin antagonist or the Ca 2؉ chelator EGTA. Activities of both enzymes increased upon their association, and iPLA 2 ␤ reaction products reduced CaMKII␤ activity. Both the iPLA 2 ␤ inhibitor bromoenol lactone and the CaMKII␤ inhibitor KN93 reduced arachidonate release from INS-1 insulinoma cells, and both inhibit insulin secretion. CaMKII␤ and iPLA 2 ␤ can be coimmunoprecipitated from INS-1 cells, and forskolin, which amplifies glucose-induced insulin secretion, increases the abundance of the immunoprecipitatable complex. These findings suggest that iPLA 2 ␤ and CaMKII␤ form a signaling complex in ␤-cells, consistent with reports that both enzymes participate in insulin secretion and that their expression is coinduced upon differentiation of pancreatic progenitor to endocrine progenitor cells.
Phospholipases A 2 (PLA 2 ) 1 are a diverse group of enzymes that catalyze hydrolysis of sn-2 fatty acid substituents from glycerophospholipid substrates to yield a free fatty acid and a 2-lysophospholipid (1). The Group VIA PLA 2 designated iPLA 2 ␤ has a molecular mass of 84 -88 kDa and does not require Ca 2ϩ for catalysis (2). Various splice variants of iPLA 2 ␤ are expressed at high levels in testis (3), brain (4), and pancreatic islet ␤-cells (5), among other tissues.
The amino acid sequence of iPLA 2 ␤ contains an ankyrin repeat domain with eight strings of a repetitive motif of about 33 amino acid residues each (34). Ankyrin repeats link integral membrane proteins to the cytoskeleton and mediate protein-protein interactions in signaling (34 -38). Ankyrin binds to inositol trisphosphate receptors (37), for example, which are located on Ca 2ϩ -containing vesicles that release intracellular Ca 2ϩ when ␤-cells are stimulated with glucose (26 -31). Ankyrin G also associates with skeletal muscle postsynaptic membranes and sarcoplasmic reticulum (38), and CaMKII participates in regulating local [Ca 2ϩ ] gradients in subcellular zones involved in Ca 2ϩ signaling. CaMKII is an important Ca 2ϩ signaling effector and serves as a gauge that temporally integrates [Ca 2ϩ ] signal intensities (39), and calmodulin participates in several Ca 2ϩ -dependent processes in insulin secretion by ␤-cells (40,41). Calmodulin and iPLA 2 ␤ interact functionally (2,8,23,24,33), and the iPLA 2 ␤ domain from residues 650 -722 contains a calmodulin binding site (2).
During cell signaling, iPLA 2 translocates to membranes (22,25,32) where it interacts with regulatory proteins to effect cellular activation. To identify proteins that interact with iPLA 2 ␤ to understand better its role in signaling, we performed yeast two-hybrid screening and have found that iPLA 2 ␤ interacts with the specific CaMKII␤ isoform expressed in pancreatic islet ␤-cells. This interaction is demonstrated by multiple independent techniques, and the interaction affects both iPLA 2 ␤ and CaMKII␤ activities, thereby defining a signaling complex.
Screening of a Rat Brain cDNA Library in the Yeast Two-hybrid System-A rat brain cDNA library cloned into pACT2 vector (containing the LEU2 gene for selection) to produce fusion between proteinencoding DNA sequences and the DNA activation domain of GAL4 was used as prey. To produce the bait construct, full-length iPLA 2 ␤ cDNA cloned from rat pancreatic islets was ligated into the SalI-EcoRI sites of the two-hybrid BD pAS2-1 vector, which contained a TRP1 gene for selection, resulting in in-frame fusion of iPLA 2 ␤ with the DNA binding domain of the yeast GAL4 protein. The fidelity of constructs was confirmed by automated sequencing. The yeast strain AH109 was used for screening assays, and this strain contains HIS3 and lacZ reporter genes. Expression of each of these genes is regulated by a distinct GAL4-responsive promoter under control of a GAL4-responsive upstream activation site. Lack of autonomous activation by the iPLA 2 ␤/ DNA binding domain fusion product was demonstrated by plating cells transformed with bait alone on media lacking histidine. In these assays, both bait and prey plasmids were transformed simultaneously into AH109 yeast cells, which were plated on medium lacking leucine, tryptophan, and histidine and allowed to grow at 30°C for 4 days. Putative positive colonies were lifted onto filter paper and incubated with the chromogenic substrate X-gal. Interactions were confirmed when the blue ␤-galactosidase reaction product was evident after 4 h of incubation at room temperature. Plasmids were then recovered from yeast and transformed into DH5␣ bacterial cells using ampicillin for selection. Isolated plasmids were sequenced, and BLAST searches were performed against GenBank (NIH genetic sequence data base) to identify putative iPLA 2 ␤-interacting proteins.
Molecular Cloning of CaMKII␤ cDNA from a Rat Islet Library-Total RNA was isolated from adult rat islets as described previously (5). First strand cDNA was transcribed with avian myeloblastosis virus reverse transcriptase. PCR was performed using a pair of gene-specific primers designed from regions of cDNA sequence that are conserved in the mouse and rat brain CaMKII␤ cDNA sequences (sense, 5Ј-ATCGCCACCGC-CATGGCCACC-3Ј; antisense, 5Ј-CAGGCGCAGCTCTCACTGCAG-3Ј). A PCR band of 1,650 bp was gel purified, ligated into pGEM-T vector, and transformed into DH5␣ cells for amplification. DNA was purified and sequenced using T3 and T7 primers and gene-specific primers. Binary Yeast Two-hybrid Assays-The iPLA 2 ␤ cDNA was cloned from an adult rat islet library (5). Full-length iPLA 2 ␤ cDNA was ligated into BD vector pAS2-1 or AD vector pACT2 and used as bait or prey. Full-length CaMKII␤ cDNA was cloned into the AD vector pACT2 or BD vector pAS2-1 and used as prey or bait. Both bait and prey plasmids were transformed simultaneously into AH109 yeast cells, which were plated on restriction medium. After incubation (30°C, 4 days), colonies were lifted onto filter paper and screened as described above. Colonies that produced the blue ␤-galactosidase reaction product were considered positive for the interaction between iPLA 2 ␤ and CaMKII␤.
Cloning and Expression of His-tagged CaMKII␤, His-tagged iPLA 2 ␤, and FLAG-tagged Proteins in Sf9 Cells-Recombinant proteins were expressed in Spodoptera frugiperda (Sf9) cells using the Bac-to-Bac baculovirus expression system (Invitrogen) following the manufacturer's instructions, as described in detail elsewhere (2,23,33,42). cDNA containing the entire coding sequence of His-tagged CaMKII␤, Histagged iPLA 2 ␤, or FLAG-tagged iPLA 2 ␤ was cloned into the SalI-EcoRI site of the pFastBac-1 vector. The sequence of the insert was verified, and the plasmid was then transformed into DH10Bac cells. Recombinant bacmid DNA was isolated using an alkaline lysis protocol modified for high molecular weight plasmid purification. PCR analysis was performed with purified bacmid DNA and pUC/M13 forward and reverse primers to characterize the inserts in the recombinant bacmid DNA. The recombinant baculovirus was produced by transfecting the recombinant bacmid DNA into Sf9 cells. The baculovirus was amplified and used to infect Sf9 cell cultures to express the recombinant proteins (2,23,33,42).
Immunoblotting Analyses-Proteins were analyzed by SDS-PAGE and transferred to a nylon membrane that was subsequently blocked with 5% nonfat dry milk for 1 h. The membrane was washed and incubated for 1 h with polyclonal antibody (1:200) to iPLA 2 ␤ or CaMKII. The membrane was then incubated with secondary antibody (1:30,000) coupled to horseradish peroxidase, and the antibody complex was visualized by ECL.
Interaction of CaMKII␤ with iPLA 2 ␤ and Protein Pull-down Assays-In some experiments, both iPLA 2 ␤ and His-tagged CaMKII␤ proteins were coexpressed in Sf9 cells. The Sf9 cell cytosol containing iPLA 2 ␤ and His-tagged CaMKII␤ proteins was mixed with TALON metal affinity resin in the presence or absence of added Ca 2ϩ /calmodulin, the calmodulin antagonist W13, or the Ca 2ϩ chelator EGTA and incubated (room temperature, with shaking, for 1 h). The mixture was washed with 10 bed volumes of wash buffer (50 mM Na 2 HPO 4 , 500 mM NaCl, pH 7.8) twice and transferred onto a gravity-flow column. The His-tagged CaMKII␤ was eluted with elution buffer (50 mM Na 2 HPO 4 , 300 mM NaCl, and 200 mM imidazole, pH 7.8) and collected in 0.5-ml fractions. Desorbed proteins were visualized by immunoblotting analyses with antibodies to iPLA 2 ␤ or CaMKII␤.
In other experiments, iPLA 2 ␤ protein was first expressed in Sf9 cells and purified as described previously (23,33). Cytosol was prepared from Sf9 cells infected with baculovirus containing cDNA that encoded Histagged CaMKII␤ and mixed with 1 ml of TALON metal affinity resin, as described above. The resin was washed and mixed with purified iPLA 2 ␤ protein. The mixture was then incubated (30 min at room temperature with shaking), washed three times, and loaded onto a 5-ml gravity-flow column. Bound proteins were desorbed with elution buffer, collected in 0.5-ml fractions, and analyzed by immunoblotting with antibodies specific for iPLA 2 ␤ or CaMKII␤.
Immunoprecipitation of FLAG-tagged iPLA 2 ␤ and CaMKII␤ Expressed in Sf9 Cells-Sf9 cells expressing FLAG-tagged iPLA 2 ␤, CaMKII␤, or both were harvested by centrifugation, washed with phosphate-buffered saline, resuspended in cell lysis buffer supplemented with protease inhibitors, and homogenized by sonication. Cytosol was prepared by centrifugation (15,000 ϫ g, 20 min) and incubated with 100 l of anti-FLAG M2 affinity resin (2 h, 4°C, gentle rotation) in the presence or absence of 10 mM Ca 2ϩ chelator EGTA. Immunoprecipitated material was recovered by centrifugation and washed four times with wash buffer. Samples immunoprecipitated with anti-FLAG affinity resin were eluted with elution buffer (0.1 M glycine, pH 3.5). Aliquots (30 l) were analyzed by 10% SDS-PAGE, transferred onto a nylon membrane, and blotted with iPLA 2 ␤ or CaMKII␤ antibodies (1:200) followed by horseradish peroxidase-conjugated secondary antibodies (1:30,000).
The iPLA 2 ␤ activity assays were performed as described previously (5,22). Briefly, 100 l of sample was added to assay buffer containing 10 mM EGTA. Reactions were initiated by injecting 5 l of 1-palmitoyl-2-[1-14 C]palmitoyl-sn-glycerol-3-phosphorylcholine (specific activity 50 mCi/mmol, final concentration 5 M) in ethanol. The assay mixture was incubated (37°C, 5 min, with shaking), and the reaction was terminated by adding 100 l of butanol. A 25-l aliquot of the butanol layer was analyzed by Silica Gel G TLC as described previously (19). The amount of 14 C-labeled free fatty acid was determined by liquid scintillation spectrometry.
[ 3 H]Arachidonic Acid Release Measurements-INS-1 insulinoma cells (5 ϫ 10 5 cells/well) were prelabeled for 20 h with 0.5 Ci/ml [ 3 H]arachidonic acid and placed in serum-free medium for 1 h. The cells were washed three times with glucose-free RPMI 1640 medium to remove unincorporated radiolabel. Cells were treated with 20 M BEL or 8 M KN93 for 30 min before adding RPMI 1640 medium containing 0.5% bovine serum albumin and incubating for 1 h. The medium was then removed and replaced with fresh medium of identical composition, and the cells were incubated for 40 min. Supernatants and cells were separated by centrifugation (500 ϫ g, 5 min) and assayed for 3 H content by liquid scintillation spectrometry.
Coimmunoprecipitation of iPLA 2 ␤ and CaMKII␤ from INS-1 Cells-Immunoprecipitation was performed with a protein A-agarose slurry that had been washed twice with phosphate-buffered saline, mixed with a 10-l solution of antibody to CaMKII␤ or to iPLA 2 ␤, and incubated (room temperature, 40 min). The mixture was centrifuged, and the supernatant was discarded. The agarose-antibody complex in the precipitate was washed three times with phosphate-buffered saline, mixed with INS-1 cell cytosol, and incubated (overnight, 4°C, with shaking). Immunoprecipitates were collected by centrifugation, washed, boiled for 5 min in SDS-PAGE sample loading buffer, and analyzed by SDS-PAGE. Proteins were transferred to nylon membranes, and immunoblotting was performed with primary antibody to iPLA 2 ␤ or to CaMKII␤ and secondary antibody coupled to horseradish peroxidase, as described above.

RESULTS
Yeast Two-hybrid Screening Indicates That CaMKII␤ Is an iPLA 2 ␤-interacting Protein-To identify proteins that interact with iPLA 2 ␤, a yeast two-hybrid screen of a rat brain cDNA library was performed using iPLA 2 ␤ cDNA cloned from a rat islet cDNA library as bait. The commercially available rat brain cDNA library was used for screening because there are many biochemical similarities between brain and islets, including high iPLA 2 ␤ expression (19 -22, 34). Colonies were identified that activated transcription of both the HIS3 gene (permitting autotrophic selection) and the lacZ reporter gene (permitting X-gal analysis) in the presence of bait. Such colonies were purified by culture after serial dilution, and the sequences of their cDNA inserts were determined. One colony contained cDNA that encoded 241 residues of rat brain CaMKII␤ N-terminal amino acid sequence (residues 34 -274). Several other colonies also contained inserts with the CaMKII␤ sequence. This interaction was examined further because of its likely functional importance, which is suggested by the facts that calmodulin is an important ␤-cell Ca 2ϩ -binding protein (43) and that ␤-cells express high levels of CaMKII␤ (44,48), which regulates voltage-operated Ca 2ϩ channels involved in insulin secretion (7,(45)(46)(47). Insulinoma cell secretion is also potentiated by overexpressing CaMKII␤ (49) or iPLA 2 ␤ (22), and iPLA 2 ␤ binds calmodulin (2,34).
Cloning CaMKII␤ from a Rat Islet cDNA Library Reveals Tissue Specificity and a Developmental Profile of CaMKII␤ Isoform Expression-Pancreatic islets express distinct CaMKII isoforms, and adult rat islets express predominantly the CaMKII␤ isoform (48,49). To determine whether CaMKII isoform(s) expressed in rat islets, like those in rat brain, also interact with iPLA 2 ␤, we cloned CaMKII␤ cDNA from adult rat islets. Reverse transcription-PCR was performed using RNA isolated from rat islets as template and a pair of primers designed from regions of cDNA sequence which are conserved in rat and mouse CaMKII␤. The PCR product was cloned. Sequencing the insert revealed a putative initiation codon (ATG) at the 5Ј-end, a stop codon (TGA) at the 3Ј-end, and the entire coding sequence in an intervening single open reading frame. Fig. 1 illustrates the nucleotide and deduced amino acid sequences of the CaMKII␤ isoform cloned from adult rat islets (ACaMKII). Fig. 2A illustrates sequence alignments for CaMKII␤ from rat brain (BCaMKII), adult rat islets (ACaMKII), human ␤-cells (HCaMKII), and neonatal rat islets (NCaMKII).
The CaMKII␤ cDNA cloned from adult rat islet mRNA contained a complete coding sequence of 1,509 bp which encodes 503 amino acid residues (Fig. 1), and this CaMKII␤ isoform is distinct from previously (50, 51) described rat isoforms. Analysis of nucleotide sequences revealed that ACaMKII␤ differs from BCaMKII␤ (50) by the lack of sequence corresponding to the first (residues 316 -339) and third (residues 379 -393) variable domains ( Fig. 2A). ACaMKII␤ differs from NCaMKII␤ (51) by the absence of the sequence from residues 370 to 456 in the association domain (Fig. 2B). Sequence alignments revealed 99.4% amino acid sequence identity between ACaMKII␤ and the HCaMKII␤ isoform cloned from human insulinoma cells (48). The ACaMKII␤ and HCaMKII␤ sequences differ only in 3 amino acid residues in variable domain 2 (Fig. 2B).
To search for other subtypes of CaMKII␤ in adult rat islets, we performed a series of reverse transcription-PCR experiments using RNA from adult rat islets as template and primers designed from various regions of the ACaMKII␤ sequence, but we observed no other CaMKII␤ subtype in adult rat islets (not shown). Adult rat pancreatic islets and adult human ␤-cells thus express mRNA that encodes a CaMKII␤ isoform that differs from those in adult brain or in neonatal islets, and the latter two isoforms also differ from each other. There is thus both tissue specificity and a developmental profile of CaMKII␤ isoform expression, but there is little rat-to-human species heterogeneity in the CaMKII␤ isoform expressed in adult pancreatic islet ␤-cells.
Binary Yeast Two-hybrid Assays Confirm the Interaction between iPLA 2 ␤ and CaMKII␤-To confirm the interaction between iPLA 2 ␤ and ACaMKII␤ observed in yeast two-hybrid screening experiments, binary yeast two-hybrid assays were performed. We first used ACaMKII␤ as bait and iPLA 2 ␤ as prey. When bait or prey alone was transformed into yeast cells, no colonies grew in medium lacking leucine, tryptophan, and histidine, but when both bait and prey were transformed simultaneously into yeast cells, colonies formed and produced blue reaction products when treated with the chromogenic substrate X-gal (Fig. 3B, left column) that reflect interaction between ACaMKII␤ and iPLA 2 ␤. When the bait and prey DNA were switched (so that iPLA 2 ␤ was bait, and ACaMKII␤ was prey), similar results were obtained (Fig. 3B, right column). These results reflect a specific interaction between iPLA 2 ␤ and CaMKII␤.
To identify domains of the proteins essential for their interaction, we performed binary yeast two-hybrid assays using Nor C-terminal fragments of iPLA 2 ␤ as the bait or prey and N-or C-terminal fragments of CaMKII␤ as the prey or bait. Fig. 3B shows the schematic representation of wild-type iPLA 2 ␤ and CaMKII␤ proteins and of their N-and C-terminal fragments. When the N-terminal fragment of iPLA 2 ␤ (NiPLA 2 ␤) was used as the bait or prey and the N-terminal fragment of CaMKII␤ (NCaMKII␤) was used as the prey or bait, large colonies formed after incubation at 30°C for 4 days, and these colonies turned blue after incubation with the chromogenic substrate X-gal for 4 h at room temperature (Fig. 3C, lane 1). When an N-terminal fragment (NiPLA 2 ␤ or NCaMKII␤) was used as bait or prey and a C-terminal fragment (CiPLA 2 ␤ or CCaMKII␤) as the prey or bait, only small colonies formed after incubation at 30°C for 4 days (Fig. 3C, lanes 2 and 3). These colonies were lifted onto filter paper and incubated until they grew large enough to perform the X-gal assay. As illustrated in Fig. 3C  (lanes 2 and 3), these colonies failed to turn blue after incubation with the chromogenic substrate X-gal, indicating that the interactions between the C-terminal domains of iPLA 2 ␤ and CaMKII␤ are weak and nonspecific. No colonies formed when the C-terminal fragment CiPLA 2 ␤ was used as bait or prey and CCaMKII␤ as prey or bait (Fig. 3C, lane 4). These results demonstrate that the N-terminal domains of iPLA 2 ␤ and CaMKII␤ interact, but the C-terminal domains do not, in agreement with the initial library screening result that the N-terminal domain of CaMKII␤ (residues 34 -271) participates in the interaction with iPLA 2 ␤. In control experiments, expression of N-or C-terminal fragments of either protein as bait or prey alone resulted in no colonies, as expected (Fig. 3C,  lanes 5-8).
CaMKII␤ Can Be Expressed from Its DNA at High Levels in a Baculovirus-Sf9 Cell System and Retains Activity after Purification-Sf9 cells have been used to express iPLA 2 ␤ (2, 23, 33), and we found that His-tagged ACaMKII␤ can also be expressed at high levels in Sf9 cells infected with baculovirus containing its cDNA. Cytosol from Sf9 cells infected with baculovirus containing DNA encoding His-tagged ACaMKII␤ was loaded onto TALON metal affinity columns, which were then washed to remove nonadsorbed proteins. Interaction of His-tagged ACaMKII␤ with metal ions on the column resin was then disrupted with imidazole-containing buffers, and this caused desorption of His-tagged ACaMKII␤ protein, which was collected in 0.5-ml fractions of column eluant. Proteins in eluant fractions were analyzed by SDS-PAGE and visualized by immunoblotting using a CaMKII antibody to demonstrate expression and purification of His-tagged ACaMKII␤ (Fig. 4A). Purified His-tagged ACaMKII␤ retained catalytic activity reflected by phosphorylation of the synthetic substrate autocamtide-3 in the presence of added Ca 2ϩ /CaM. In the absence of added Ca 2ϩ /CaM little activity was detected (Fig. 4B). The intensity of the immunochemical signal for CaMKII␤ in the eluant fractions (Fig. 4A) correlated well with CaMKII␤ activity in these fractions (Fig. 4B).
ACaMKII␤ and iPLA 2 ␤ Interact with Each Other When Coexpressed in Sf9 Cells-To characterize further the interaction between the two proteins, His-tagged ACaMKII␤ and fulllength, untagged iPLA 2 ␤ (hereafter designated "native" iPLA 2 ␤) were coexpressed in Sf9 cells to determine whether His-tagged ACaMKII␤ could pull down native iPLA 2 ␤ from cell cytosol. Sf9 cells were coinfected with baculovirus that contained DNA encoding His-tagged ACaMKII␤ and with baculovirus that contained DNA encoding native iPLA 2 ␤. Cytosol was loaded onto TALON metal affinity columns, which were then washed as described above. Imidazole-containing buffer was used to desorb His-tagged CaMKII␤ and any proteins associ-ated with it. Aliquots of eluant fractions were analyzed by SDS-PAGE and immunoblotting with antibodies specific for CaMKII␤ or iPLA 2 ␤. His-tagged ACaMKII␤ (Fig. 5A, lower panel) and native iPLA 2 ␤ (Fig. 5A, upper panel) proteins eluted in the same fractions, as detected by immunoblotting. Activity assays for iPLA 2 ␤ (Fig. 5B) and ACaMKII␤ (Fig. 5C) indicate that both proteins retain activity after elution. The intensity of the immunochemical signals (Fig. 5A) correlated well with the activities of iPLA 2 ␤ (Fig. 5B) and CaMKII␤ (Fig. 5C) in the eluant fractions. Similar results were obtained using purified proteins from Sf9 cells (Fig. 6A). These findings support the conclusions from yeast two-hybrid assays that these two proteins interact with each other.
The Stoichiometry of the Interaction between iPLA 2 ␤ and CaMKII␤-To characterize further the interaction of iPLA 2 ␤ with ACaMKII␤, His-tagged ACaMKII␤ was adsorbed onto TALON metal affinity resin, and purified iPLA 2 ␤ was incubated with the resin. The resin was then washed and loaded into a gravity-flow column, and the interaction between the His tag and the immobilized metal ions was disrupted by elution with imidazole-containing buffer. Proteins in eluant fractions

FIG. 2. Alignment of deduced amino acid sequences of CaMKII␤ isoforms cloned from rat brain (B), adult rat islets (A), adult human ␤-cells (H), and neonatal rat islets (N).
A compares aligned sequences of CaMKII␤ isoforms from various sources. The variable region is shaded. B is a graphical alignment of the sequences. The region of difference in amino acid sequence between ACaMKII␤ and HCaMKII␤ is denoted by an asterisk (*).

FIG. 3. iPLA 2 ␤ interacts with CaMKII␤ in yeast cells.
A illustrates binary yeast two-hybrid assays performed using full-length iPLA 2 ␤ (or CaMKII␤) as bait and full-length CaMKII␤ (or iPLA 2 ␤) as prey. B contains schematic structures of wild-type iPLA 2 ␤ and CaMKII␤ and of constructs that correspond to N-or C-terminal fragments of each protein. An iPLA 2 ␤ N-terminal fragment that contains the ankyrin repeat domain and a C-terminal fragment that contains the catalytic site are shown, as are a CaMKII␤ N-terminal fragment that contains the catalytic domain and a C-terminal fragment that contains the association domain. C illustrates binary yeast two-hybrid assays involving coexpression of N-or C-terminal fragments of iPLA 2 ␤ and of CaMKII␤ as bait/prey pairs together (lanes 1-4) or, in control experiments, expression of an N-or C-terminal fragment of one of the proteins alone (lanes 5-8). The blue colonies reflect specific interactions between two proteins that constitute bait-prey partners in the binary yeast two-hybrid assay. The arrow identifies such blue colonies formed by the ␤-galactosidase reaction product after incubation with the chromogenic substrate X-gal.
were analyzed by SDS-PAGE and immunoblotting. Fig. 6A illustrates that His-tagged ACaMKII␤ (lower panel) and iPLA 2 ␤ (upper panel) eluted from the column in the same fractions, which provides additional evidence that these two proteins interact with each other. To determine the molar ratio of the two enzymes in the complex, the dose-response studies illustrated in Fig. 6B were performed. The amount of iPLA 2 ␤ enzyme pulled down by His-tagged ACaMKII␤ increases as the molar ratio increases up to 1:1 but does not increase further at a ratio of 2:1. This suggests that the two enzymes form a complex with 1:1 stoichiometry.
The Calmodulin Antagonist W13 Does Not Prevent the Interaction of CaMKII␤ with iPLA 2 ␤-Because both iPLA 2 ␤ and CaMKII␤ have calmodulin binding domains, calmodulin might mediate the interaction between these two proteins by forming a ternary complex. To evaluate this possibility, the interaction between iPLA 2 ␤ and CaMKII␤ was examined in the presence and absence of added calmodulin. FLAG-tagged iPLA 2 ␤ was expressed in Sf9 cells and purified with a FLAG M kit (Sigma). FLAG-tagged iPLA 2 ␤ was then mixed with TALON metal affinity resin that had previously been loaded with His-tagged ACaMKII␤ in the presence or absence of calmodulin and then washed. When calmodulin was not added, the calmodulin antagonist W13 was added to block binding of any contaminating calmodulin to the target proteins. Adsorbed proteins were eluted from the metal affinity resin with imidazole-containing buffer, and proteins in eluant fractions were analyzed by SDS-PAGE and immunoblotting with iPLA 2 ␤-specific antibody. Fig.  7A illustrates that added calmodulin is not required for the interaction between iPLA 2 ␤ and CaMKII␤ and that this interaction is not prevented by the calmodulin antagonist W13. These results are consistent with the findings that the CaM binding site(s) of iPLA 2 ␤ reside in its C-terminal domain (2) and that the interaction of iPLA 2 ␤ and CaMKII␤ occurs between their N-terminal domains (Fig. 3C).
The Ca 2ϩ Chelator EGTA Does Not Prevent the Interaction between iPLA 2 ␤ and CaMKII␤-The ability of iPLA 2 ␤ to bind calmodulin causes iPLA 2 ␤ preparations purified from cytosol to contain calmodulin, as detected by immunoblotting with calmodulin antibody (data not shown). Previous studies demonstrate that iPLA 2 ␤ dissociates from calmodulin-agarose in the presence of EGTA (23,39). To determine the role of calmodulin in the interaction between iPLA 2 ␤ and CaMKII␤, we performed an immunoprecipitation study of the interaction of FLAG-tagged iPLA 2 ␤ with CaMKII␤ in the presence and absence of EGTA. Fig. 7B illustrates that in the presence of 10 mM EGTA, FLAG-tagged iPLA 2 ␤ can still pull down CaMKII␤ from cytosol. The immunoblotting results in Fig. 7B illustrate that the amount of CaMKII␤ pulled down by FLAG-tagged iPLA 2 ␤ is unaffected by EGTA and suggest that calmodulin is not directly involved in the interaction between iPLA 2 ␤ and

FIG. 4. Expression of His-tagged CaMKII␤ in Sf9 cells and its adsorption to and desorption from metal affinity columns.
In A, cytosol from Sf9 cells that had been infected with baculovirus containing DNA that encodes His-tagged CaMKII␤ was incubated with TALON metal affinity resin, as described under "Experimental Procedures." The resin was then loaded into a gravity-flow column and washed with buffer, and His-tagged CaMKII␤ was eluted with imidazole-containing buffer and collected in 0.5-ml fractions. Proteins in aliquots of the load (L), wash (W), and elution fractions were analyzed by SDS-PAGE, and immunoblotting was then performed with CaMKII antibody. In B, the protein content of each fraction was measured, and CaMKII activity was determined in the presence (ϩ) or absence (Ϫ) of added Ca 2ϩ /CaM. When Ca 2ϩ /CaM was not added, 1 mM EGTA was added. For each assay, an aliquot of each eluant fraction was mixed with assay buffer, peptide substrate (autocamtide-3), and [␥-32 P]ATP, as described under "Experimental Procedures." Displayed values represent the means, and error bars denote S.E. (n ϭ 6).

FIG. 5. iPLA 2 ␤ interacts with CaMKII␤ when the two proteins are coexpressed in Sf9 insect cells.
In A, Sf9 cells were infected simultaneously with baculovirus containing full-length HisCaMKII␤ and iPLA 2 ␤ DNAs, cultured, and then homogenized, as described under "Experimental Procedures." Cytosol prepared from homogenates was loaded onto a TALON metal affinity column and washed with buffer. HisCaMKII␤ was eluted with imidazole-containing buffer and collected in 0.5-ml fractions. The proteins in aliquots of load (L), wash (W), and elution fractions were analyzed by SDS-PAGE, and immunoblotting was performed with antibodies to iPLA 2 ␤ (upper panel) or HisCaMKII␤ (lower panel). In B, an aliquot of load, wash, or elution fractions was added to assay buffer containing 10 mM EGTA, 1 mM ATP, and 1-palmitoyl-2-[ 14 C]linoleoyl-sn-glycero-3-phosphocholine substrate. Reactions to measure iPLA 2 ␤ activity were performed and terminated as described under "Experimental Procedures," and released [ 14 C]linoleic acid was isolated by TLC and measured by liquid scintillation spectrometry. Displayed values represent the means, and error bars denote S.E. (n ϭ 6). In C, an aliquot of load, wash, or elution fractions was mixed with assay buffer containing 0.1 mM ATP, 0.75 mM CaCl 2 , 20 g/ml calmodulin, 20 M autocamtide-3, and 2 Ci of [␥-32 P]ATP and incubated at 30°C for 3 min to determine CaMKII activity. An aliquot of the reaction mixture was applied to phosphocellulose paper, which was then washed. CaMKII activity was calculated from the amount of phosphorylated autocamtide-3, as determined by liquid scintillation spectrometric measurement of 32 P content. Displayed values represent the means, and error bars denote S.E. (n ϭ 6).
CaMKII␤. In control experiments, the N-terminal FLAGtagged alkaline phosphatase fusion protein was found not to pull down CaMKII␤ from cytosol, as expected.
The Activities of Both iPLA 2 ␤ and CaMKII␤ Increase When the Proteins Associate with Each Other-Because results from yeast two-hybrid assays and protein pull-down experiments indicate that the ACaMKII␤ and iPLA 2 ␤ proteins interact with each other, we next determined whether this interaction affects the catalytic activity of either enzyme. PLA 2 activity assays involved measuring radiolabeled free fatty acid release from phospholipid substrates and were performed in buffer supplemented with 10 mM EGTA and 10 mM ATP with no added Ca 2ϩ . Under these conditions, adding purified, recombinant, Histagged ACaMKII␤ to purified, recombinant, His-tagged iPLA 2 ␤ resulted in a statistically significant increase in PLA 2 activity (Fig. 8A). Results from dose-response studies under conditions where [iPLA 2 ␤] was constant and [CaMKII␤] was varied indicate that the maximal iPLA 2 ␤ activity is achieved at a 1:1 molar ratio of the two enzymes (Fig. 8B), which is consistent with the finding in Fig. 6B that iPLA 2 ␤ and CaMKII␤ form a complex with 1:1 stoichiometry.
CaMKII activity assays involved measurement of [ 32 PO 4 ] incorporation from [␥-32 P]ATP into a model peptide substrate. Fig. 9 illustrates that adding purified, recombinant, His-tagged iPLA 2 ␤ to purified, recombinant, His-tagged CaMKII␤ resulted in a statistically significant increase in CaMKII activity in the presence of added Ca 2ϩ /CaM. Without added Ca 2ϩ or CaM, CaMKII activity was low, and it was little affected by adding iPLA 2 ␤.
Arachidonic Acid and 2-Lysophosphatidylcholine Inhibit CaMKII␤ Activity-The above results suggest that iPLA 2 ␤ and CaMKII␤ form a complex and that this affects activities of both enzymes. To examine further the functional relationship be-tween the two enzymes, we measured effects of the iPLA 2 ␤ reaction products arachidonic acid and 2-lysophosphatidylcholine on CaMKII␤ activity. Fig. 10 illustrates that both arachidonic acid and 2-lysophosphatidylcholine inhibit CaMKII␤ activity in a concentration-dependent manner.
Arachidonic Acid Release from INS-1 Insulinoma Cells Is Suppressed by Inhibitors of CaMKII␤ and iPLA 2 ␤-To determine whether evidence for a signaling complex between iPLA 2 ␤ and CaMKII␤ could be observed in intact ␤-cells, we examined the effects of the CaMKII inhibitor KN93 and the iPLA 2 ␤ inhibitor BEL on [ 3 H]arachidonic acid release from prelabeled INS-1 insulinoma cells. Both KN93 and BEL are known to suppress insulin secretion from ␤-cells (9, 10, 19 -22). Fig. 11 illustrates that both the CaMKII inhibitor and the iPLA 2 ␤ inhibitor suppress [ 3 H]arachidonic acid release from INS-1 cells, which is consistent with an interaction of CaMKII␤ and iPLA 2 ␤ in ␤-cells to form a signaling complex.
CaMKII␤ and iPLA 2 ␤ Form a Complex in Insulin-secreting ␤ Cells-To confirm the formation of an iPLA 2 ␤⅐CaMKII␤ complex in ␤-cells, we determined whether the two enzymes can be coimmunoprecipitated from INS-1 insulinoma cells. Fig. 12A illustrates that both enzymes can be coimmunoprecipitated from parental INS-1 cells and from a stably transfected INS-1 cell line that overexpresses iPLA 2 ␤ (22) using antibodies against CaMKII (left panel). Similar results were obtained in coimmunoprecipitation experiments using antibodies against iPLA 2 ␤ (right panel). This demonstrates the existence of an iPLA 2 ␤⅐CaMKII␤ complex in intact ␤-cells. Fig. 12B illustrates that forskolin, which is an adenylyl cyclase activator that amplifies insulin secretion (22), increases the intensity of the FIG. 6. The stoichiometry of the interaction between iPLA 2 ␤ and CaMKII␤. In A, purified, recombinant, His-tagged CaMKII␤ from Sf9 cells was mixed with TALON metal affinity resin, and the resin was then washed. Bound CaMKII␤ was measured with a Coomassie protein assay kit. iPLA 2 ␤ protein expressed in Sf9 cells was purified as described previously (33), and 850 g (10 nmol) of the protein was mixed with metal affinity resin to which 570 g (10 nmol) of His-tagged CaMKII␤ had been adsorbed. The mixture was incubated at room temperature for 30 min with shaking, and the resin was washed and loaded onto a gravity-flow column. Bound proteins were eluted with imidazole-containing buffer and collected in 0.5-ml fractions. Proteins in aliquots of the load (L), wash (W), and elution fractions were analyzed by 10% SDS-PAGE, and immunoblotting was then performed with antibodies specific for iPLA 2 ␤ (upper panel) or CaMKII␤ (lower panel). In B, 200 l of metal affinity resin slurry to which 150 g (2.64 nmol) of His-tagged CaMKII␤ had been adsorbed was mixed with FLAG-tagged iPLA 2 ␤ in amounts that varied from 0 to 5.28 nmol. The mixture was incubated at 4°C overnight with shaking, and the resin was then washed to remove noncomplexed proteins. Proteins were eluted from the metal affinity resin and analyzed by 10% SDS-PAGE. Immunoblotting was then performed with primary antibodies specific for iPLA 2 ␤ (upper panel) or CaMKII␤ (lower panel). The resin was then washed, and adsorbed proteins were eluted with imidazole-containing buffer. Aliquots of wash and elution fractions were analyzed by SDS-PAGE and immunoblotting with antibody specific for iPLA 2 ␤ or CaMKII. In B, cytosol was prepared from baculovirus-infected Sf9 cells that expressed FLAG-iPLA 2 ␤, CaMKII␤ without a FLAG tag, or the control fusion protein N-terminal FLAG-tagged alkaline phosphatase (Flag-BAP). Binary mixtures of cytosols were prepared and incubated with anti-FLAG M2 affinity resin for 2 h at 4°C in the presence or absence of 10 mM EGTA. Immunoprecipitated material was recovered by centrifugation and washed four times with wash buffer. Samples immunoprecipitated with anti-FLAG affinity resin were eluted with buffer containing FLAG peptide. Proteins in the eluant were analyzed by 10% SDS-PAGE and transferred onto a nylon membrane, and immunoblotting was performed with iPLA 2 ␤ or CaMKII antibodies. immunochemical signal for iPLA 2 ␤ that coimmunoprecipitates with CaMKII␤ in INS-1 cells. This suggests that forskolin promotes formation of the iPLA 2 ␤⅐CAMKII␤ complex, and forskolin is also known to induce subcellular redistribution of iPLA 2 ␤ in INS-1 cells (22). DISCUSSION Major PLA 2 activities in pancreatic islet ␤-cells and insulinoma cells are Ca 2ϩ -independent, and much evidence indicates that iPLA 2 ␤ participates in signaling events involved in glucose-induced insulin secretion (19 -22, 34, 52). The iPLA 2 ␤ enzyme is also the predominant PLA 2 activity in hippocampus, where it catalyzes arachidonic acid release that is required for long term potentiation (4), which is an electrophysiologic analog of learning. CaMKII is also involved in both insulin secretion (6, 7, 9 -11, 45-49, 53) and long term potentiation (54 -56). The physiological functions of iPLA 2 ␤ and CaMKII thus appear to be linked in some cells, such as ␤-cells and neurons. Another isoform of CaMKII (CaMKII␣) interacts with Group IVA PLA 2 (cPLA 2 ) in vascular smooth muscle cells (57), and our findings indicate that CaMKII␤ interacts similarly with iPLA 2 ␤ to form a complex. Because ␤-cells express both CaMKII␤ and iPLA 2 ␤, a complex of these enzymes could affect ␤-cell function.
We first observed the complex between iPLA 2 ␤ and CaMKII␤ by using iPLA 2 ␤ as bait in yeast two-hybrid screening of a rat brain cDNA library. Formation of a complex between the two enzymes was confirmed in binary yeast twohybrid assays in which iPLA 2 ␤ was bait and CaMKII␤ was prey and in the converse assay configuration in which CaMKII␤ was bait and iPLA 2 ␤ was prey. Pull-down assays with recombinant, His-tagged proteins adsorbed to metal affinity matrices also provided direct evidence for the physical association of CaMKII␤ and iPLA 2 ␤. These findings clearly demonstrate that iPLA 2 ␤ and CaMKII␤ interact with each other. We have demonstrated here that an immunoprecipitatable complex of these two enzymes exists in insulinoma cells and that the amount of the complex increases upon stimulation of intact ␤-cells with forskolin, which is an adenylyl cyclase activator that amplifies insulin secretion and induces subcellular FIG. 9. Influence of iPLA 2 ␤ on CaMKII␤ activity. CaMKII␤ activity was measured in the presence (ϩ) or absence (Ϫ) of iPLA 2 ␤, Ca 2ϩ , and CaM. For each assay, His-tagged CaMKII␤ was mixed with assay buffer containing ATP, autocamtide-3 substrate, and [␥-32 P]ATP. CaMKII activity was calculated from the amount of phosphorylated autocamtide-2 as in Fig. 4. Values are represented as the mean Ϯ S.E. (n ϭ 5). Statistical significance is denoted by an asterisk, which indicates a p value Ͻ 0.01 compared with the group to which no iPLA 2 ␤ was added (ϪiPLA 2 ␤).
FIG. 10. Inhibition of CaMKII␤ activity by arachidonic acid (AA) and lysophosphatidylcholine (LPC). CaMKII␤ activity was measured in the presence or absence of arachidonic acid or lysophosphatidylcholine as in Fig. 4. For each assay, two separate measurements were performed simultaneously, one in the presence and the other in the absence of added Ca 2ϩ /CaM. Activity values were calculated from the difference between these two measurements and are represented as the mean Ϯ S.E. (n ϭ 3). Statistical significance is denoted by an asterisk (*), which indicates a p value Ͻ 0.05 compared with control. redistribution of iPLA 2 ␤ in ␤-cells (22).
We have demonstrated previously that depletion of internal Ca 2ϩ stores causes activation of iPLA 2 ␤ in ␤-cells (23) and in vascular smooth muscle cells (24). It has been demonstrated recently that iPLA 2 ␤ participates in SOC entry from the extracellular space (25,32), and this process is required for insulin secretion (26 -31). Lysophospholipid products of iPLA 2 ␤ activate SOC channels that mediate capacitative Ca 2ϩ influx (25,32), and CaMKII also affects Ca 2ϩ fluxes by potentiating SOC channel activity (58) and regulating T-type voltage-operated calcium channels (59). Our findings indicate that iPLA 2 ␤ interacts with the specific isoform of CaMKII␤ that is expressed in ␤-cells and that this interaction affects activities of both iPLA 2 ␤ and CaMKII␤. This suggests that CaMKII␤ and iPLA 2 ␤ form a signaling complex, and this complex represents a potential means to regulate SOC entry.
Alignment of the deduced amino acid sequences of HCaMKII␤ (48) and ACaMKII␤, which have been cloned from adult human ␤-cells and adult rat islets, respectively, reveals more than 99% sequence conservation, and this indicates that there is little species-to-species variation in pancreatic islet ␤-cell expression of CaMKII␤ isoforms. The expression pattern of CaMKII isoforms does change with development in islets, as reflected by the difference in isoforms expressed in neonatal and adult islets, and there is also tissue-to-tissue heterogeneity in CaMKII isoform expression, as reflected by the different isoforms expressed by islets and brain. The high degree of CaMKII␤ sequence conservation between rat and human islets and the fact that islets express only a single, predominant CaMKII␤ isoform is consistent with the possibility that the islet isoform has a special function in ␤-cells and that iPLA 2 ␤ and other proteins that interact with this enzyme modulate that function. It is thus of interest that expression of both iPLA 2 ␤ and of CaMKII␤ has recently been found to occur at the same stage of differentiation of pancreatic progenitor cells to endocrine progenitor cells during development (60).