Pancreatic islets express a Ca2+-independent phospholipase A2 enzyme that contains a repeated structural motif homologous to the integral membrane protein binding domain of ankyrin.

Pancreatic islets express a Ca2+-independent phospholipase A2 (CaI-PLA2) activity that is sensitive to inhibition by a haloenol lactone suicide substrate that also attenuates glucose-induced hydrolysis of arachidonic acid from islet phospholipids and insulin secretion. A cDNA has been cloned from a rat islet cDNA library that encodes a protein with a deduced amino acid sequence of 751 residues that is homologous to a CaI-PLA2 enzyme recently cloned from Chinese hamster ovary cells. Transient transfection of both COS-7 cells and Chinese hamster ovary cells with the cloned islet CaI-PLA2 cDNA resulted in an increase in cellular CaI-PLA2 activity, and this activity was susceptible to inhibition by haloenol lactone suicide substrate. The domain of the islet CaI-PLA2 from amino acid residues 150-414 is composed of eight stretches of a repeating sequence motif of approximately 33-amino acid residues in length that is highly homologous to domains of ankyrin that bind both tubulin and integral membrane proteins, including several proteins that regulate ionic fluxes across membranes. These findings complement previous pharmacologic observations that suggest that CaI-PLA2 may participate in regulating transmembrane ion flux in glucose-stimulated β-cells.

Glucose-induced insulin secretion from pancreatic islet ␤-cells requires that glucose be transported into the ␤-cell and metabolized (1). Signals derived from glucose metabolism result in inactivation of plasma membrane ATP-sensitive K ϩ channels (K ATP ), 1 membrane depolarization, activation of voltage-operated Ca 2ϩ channels, influx of Ca 2ϩ , and a rise in cytosolic [Ca 2ϩ ], which triggers insulin exocytosis (2). Stimulation of islets with glucose also induces hydrolysis of arachidonic acid from islet membrane phospholipids (3), and the resultant ac-cumulation of nonesterified arachidonic acid (4) may facilitate Ca 2ϩ entry into ␤-cells (5) and amplify depolarization-induced insulin secretion (6).
Hydrolysis of arachidonic acid from membrane phospholipids in glucose-stimulated islets appears to be mediated in part by a phospholipase A 2 (PLA 2 ) enzyme that is catalytically active in the absence of Ca 2ϩ and that is inactivated by a haloenol lactone suicide substrate (HELSS) (7)(8)(9)(10). Treatment of islets with HELSS results in attenuation of the glucose-induced rise in ␤-cell cytosolic [Ca 2ϩ ] and in inhibition of insulin secretion (8 -10). The structure of islet Ca 2ϩ -independent phospholipase A 2 (CaI-PLA 2 ) is not known, but a CaI-PLA 2 enzyme has recently been cloned from CHO cells (11) and its sequence determined (GenBank accession number 115470). We report here the cloning, expression, and sequence analysis of a homologous enzyme from a rat pancreatic islet cDNA library (12).
Isolation of Pancreatic Islets, Preparation of Islet ␤-Cells and Non-␤cells, and Culture of COS-7, CHO, and HIT-T15 Cells-Islets were isolated aseptically from male Harlan Sprague Dawley rats, as described (13), by collagenase digestion of excised, minced pancreas, density gradient isolation, and manual selection under microscopic visualization (10). Isolated islets were transferred to Falcon Petri dishes containing 2.5 ml of cCMRL-1066, placed under an atmosphere of 95% air, 5% CO 2 , and cultured overnight at 37°C. Purified populations of islet ␤-cells and non-␤-cells were prepared by fluorescence-activated cell sorting of dispersed cells obtained by treating islets with the enzyme dispase, as described previously (5,7,10,14,15). COS-7 and CHO cells were cultured in Dulbecco's modified Eagle's medium (MEM) containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate. HIT-T15 cells were obtained from ATCC (Bethesda, MD) and were cultured in Ham's F12K medium (Life Technologies, Inc.) containing 7 mM D-glucose, 10% dialyzed horse serum, and 2.5% fetal bovine serum.
cDNA Cloning and Sequencing-A pair of PCR primers (sense, 5Ј-TATGCGTGGTGTGTACTTCC-3Ј and antisense, 5Ј-TGGAGCT-CAGGGCGACAGCA-3Ј) were designed based on the cDNA sequence of the CHO cell CaI-PLA 2 (11) and used in RT-PCR reactions with RNA from HIT-T15 insulinoma cells. These reactions yielded a 820-bp product. A 32 P-labeled form of this product was prepared by randomly primed labeling and used to screen an islet cDNA library prepared as described elsewhere (12) in the Lambda ZAP Express system (Stratagene). Clones that hybridized with the probe were isolated and further plaque-purified. The pBK-CMV plasmid containing the CaI-PLA 2 cDNA sequence as insert was excised in vivo according to the manufacturer's protocol. Insert sizes were determined after digestion with the restriction enzymes EcoRI and XhoI. The cloned cDNA in pBK-CMV was sequenced from the double strand by the method of Sanger et al. (17) using a Sequenase version 2.0 DNA sequencing kit (United States Biochemical Corp.) with T3 and T7 primers and specific synthetic oligonucleotide primers.
Expression and Purification of CaI-PLA 2 -After removal of the 5Јuntranslated region, the full-length rat islet CaI-PLA 2 cDNA was subcloned in-frame into the EcoRI and XhoI sites of pET-28c (Novagen) using standard techniques. The fidelity of the construct was verified by restriction analysis and sequencing. The pET28-CaI-PLA 2 construct then was transformed into the bacterial expression host Escherichia coli strain BL21(DE3) (Novagen). This system yields a fusion protein with polyhistidine and T7 epitope tag sequences joined to the sequence encoded by the insert. Cells transformed with pET28c containing no insert served as negative controls. Expression was induced by incubation of the cells with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h and assessed by 10% SDS-PAGE analysis of total protein followed by Coomassie Blue staining. The expressed protein was affinity-purified by nickel-chelate chromatography as described in the manufacturer's protocol (Novagen). The protein was eluted with imidazole and analyzed by 10% SDS-PAGE.
Immunoblotting Analyses-For Western blotting analyses, proteins were transferred to a nylon membrane, which was subsequently blocked (1 h, room temperature) with Tris-buffered saline-Tween (TBS-T, which contains 20 mM Tris-HCl, 137 mM NaCl, pH 7.6, and 0.1% Tween 20) containing 1% BSA and 3% milk powder. Following three washes with TBS-T, the blot was then incubated (1 h, room temperature) with a monoclonal antibody (1:50 dilution in TBS-T with 1% BSA) raised in a mouse against a synthetic peptide representing a 10-amino acid residue sequence ( 221 TPLHLACQMG 230 ) located at the C terminus of islet CaI-PLA 2 . The nylon membrane was washed three times in TBS-T and incubated (1 h, room temperature) with a goat anti-mouse IgG conjugated to horseradish peroxidase (Boehringer Mannheim) at a 1:3000 dilution in TBS-T containing 1% BSA. Detection of the secondary antibody was performed by enhanced chemiluminescence.
Examination of the Abundance of CaI-PLA 2 mRNA by Competitive RT-PCR-Competitive RT-PCR (18,19) was used to determine the abundance of CaI-PLA 2 mRNA in various preparations. In this approach, a competitor DNA species is prepared that contains the same primer template sequences as the target cDNA but that contains an intervening sequence that differs from the target in size or in restriction sites so that PCR products from the target and competitor can be distinguished (18,19). Using the competitor as an internal control, amounts of target cDNA can be determined by allowing known amounts of the competitor to compete with the target for primer binding during amplification (18,19).
To prepare the competitor DNA, two composite primers were synthesized (sense, 5Ј-TGCAGACCAGTTAGTATGGCTGTGGGCAAGGTCA-TCC-3Ј and antisense, 5Ј-TGGACAAGCTTCTGGAACTCCTTGGAGG-CCATGTAG-3Ј). These primers contain the CaI-PLA 2 primer sequence (underlined) attached to sequences that hybridize to rat glyceralde-hyde-3-phosphate dehydrogenase cDNA (20). This pair of primers was then used in PCR reactions with rat glyceraldehyde-3-phosphate dehydrogenase cDNA as template. In these reactions, the CaI-PLA 2 primer sequences are incorporated into the PCR product during amplification, and the intervening sequence derives from glyceraldehyde-3-phosphate dehydrogenase. The resultant PCR product (380 bp in length) was analyzed by agarose gel electrophoresis, isolated with a QIAEX gel extraction kit (QIAGEN), and used as the competitor DNA species in subsequent PCR experiments. In these experiments, the primer pair (sense 5Ј-TGCAGACCAGTTAGTATGGC-3Ј and antisense 5Ј-TGGA-CAAGCTTCTGGAACTC-3Ј) was used. These primers hybridize to the CaI-PLA 2 cDNA sequence and to the competitor DNA sequence. The PCR product derived from the CaI-PLA 2 cDNA is 528 bp in length and that from the competitor DNA is 380 bp in length. The products were then analyzed by 1% agarose gel electrophoresis and visualized with ethidium bromide. Product band intensity was then determined with an IS-1000 Digital Imaging System.
With a fixed amount of target and competitor, varying the PCR cycle number from 19 to 31 was found to yield a constant relative intensity of the target and competitor PCR product bands. In subsequent experiments, 28 PCR cycles were used. To construct a standard curve, the competitor DNA solution was diluted in the range of 10 Ϫ2 to 10 Ϫ6 , and aliquots of each dilution were added to a reaction mixture containing a fixed amount of CaI-PLA 2 cDNA. After PCR amplification, products were analyzed by agarose gel electrophoresis, and product band intensity was determined as above. The ratio of product band intensities was found to correspond to the amount of input DNA over the range of tested concentrations. Similar results were obtained in experiments in which the amount of competitor DNA was fixed and the amount of target DNA was varied.

COS-7 and CHO Cell Transfection and CaI-PLA 2 Activity Assay-
The full-length islet CaI-PLA 2 cDNA was subcloned into pcDNA3.1 (Invitrogen) in the forward orientation. Petri dishes (100 mm diameter) containing either COS-7 cells or CHO cells at about 60 -80% confluence were transfected with 20 g of plasmid DNA by calcium phosphate precipitation. After overnight culture, fresh MEM was added to each plate, and the cells were then cultured for an additional 2 days. At 72 h after transfection, cells were harvested after washing with ice-cold phosphate-buffered saline. For CaI-PLA 2 enzyme activity assays, the cells were collected by centrifugation (Beckman table-top centrifuge, 3000 rpm, 2 min, room temperature). These cells were then resuspended and washed three times with ice-cold phosphate-buffered saline and once with homogenization buffer (250 mM sucrose, 40 mM Tris-HCl, pH 7.5). The cells were pelleted again, resuspended in homogenization buffer, and disrupted by sonication. The homogenates were centrifuged (170,000 ϫ g, 60 min, 4°C) to obtain a cytosolic supernatant. The protein content of the cytosolic fraction was measured by Bio-Rad assay. CaI-PLA 2 activity was measured in aliquots of cytosol (100 l, approximately 25 g of protein) added to assay buffer (200 mM Tris-HCl, pH 6.0, 10 mM EGTA, total assay volume 200 l). As described in the appropriate figure legends, effects of preincubating (2 min, room temperature) cytosolic fractions with various concentrations of HELSS on CaI-PLA 2 activity were also determined, as were effects of varying the assay solution pH and effects of including nucleotide phosphates such as ATP in the assay solution. Reactions were initiated by injection of an ethanolic solution (5 l) of L-a-1-palmitoyl-2-[ 14 C]linoleoyl-phosphatidylcholine substrate (specific activity 50 mCi/mmol, final concentration 5 M). Assay mixtures were then incubated (3 min, 37°C, with shaking), and reactions were terminated by addition of butanol (100 l) and vortexing (8 s). After centrifugation (2000 ϫ g, 2 min), reaction products in 25 l of the upper (butanol) layer were analyzed by Silica Gel G thin layer chromatography in the solvent system petroleum ether/ethyl ether/acetic acid (80/20/1). The region of the TLC plate containing free fatty acid (R f 0.58) was identified with iodine vapor and scraped into scintillation vials, and the 14 C-labeled content was then determined by liquid scintillation spectrometry. The amount of [ 14 C]linoleate released was then converted to a PLA 2 specific activity value (pmol/mg protein ϫ min) as described elsewhere (7).
Dot Matrix Analysis of the Islet CaI-PLA 2 Amino Acid Sequence and Comparison to Other Sequences-Dot matrix plots (21) were constructed with the UWGCG programs COMPARE and DOTPLOT using a window of 30 and stringency of 15. Segments of 30 amino acids from the horizontal axis were compared with segments from the vertical axis, and a dot was placed in the appropriate position whenever the total score of aligned sequences exceeded a value of 15. These plots were used to examine internal repeating motifs in the islet CaI-PLA 2 sequence and to compare the sequences of ankyrin and of related proteins to the CaI-PLA 2 sequence. activity similar to that in islets (22). A 32 P-labeled form of the resultant 820-bp RT-PCR product was generated by randomly primed labeling and used to screen an islet cDNA library. Screening about 900,000 plaque-forming units resulted in the identification of five positive clones that were isolated, plaquepurified, and excised in vivo. The size of the cDNA inserts in the plasmids from each of the five clones was then determined after release of the inserts by digestion with the restriction enzymes EcoRI and XhoI. Each clone contained an insert of about 3.3 kilobases.

Cloning the Rat
Sequencing the inserts from both 5Ј-and 3Ј-ends revealed that each contained an identical 5Ј-sequence and an identical 3Ј-sequence ending in a poly(A) tail. A putative initiation codon (ATG) was observed 475 bp downstream of the 5Ј-end. Nucleotide sequencing of the entire 3288-bp insert from one clone revealed a single long open reading frame (2256 bp), which was predicted to encode a protein with 751 amino acid residues. The insert also contained 474 bp of 5Ј-untranslated region and 558 bp of 3Ј-untranslated region including a poly(A) tail. A presumptive polyadenylation signal (AATAAA) was present 14 bp upstream of the poly(A) tail (Fig. 1).
Deduced Amino Acid Sequence of the Rat Islet CaI-PLA 2 -The encoded protein has a calculated molecular mass of 83,591 daltons and a predicted isoelectric point of 6.6. The amino acid sequence of islet CaI-PLA 2 deduced from the nucleotide sequence of the cDNA does not exhibit significant homology to the sequences of known Ca 2ϩ -dependent PLA 2 enzymes, including the 85-kDa cytosolic PLA 2 or secretory PLA 2 (23). The islet CaI-PLA 2 exhibits 90% identity of nucleotide sequence and 95% identity of amino acid sequence to the CaI-PLA 2 recently cloned from CHO cells (11). One absolutely conserved region is that between His 421 and Glu 551 , which flanks and includes a GXSXG lipase consensus sequence (11,24,25). This sequence occurs in the islet CaI-PLA 2 as G 462 TSTG 466 .
Examination of the islet CaI-PLA 2 amino acid sequence using dot matix analysis revealed ( Fig. 2A) that the domain from amino acid residues 150 -414 is composed of eight strings of a repetitive sequence motif of approximately 33 amino acid residues in length. The alignment of the repeats in the islet CaI-PLA 2 sequence is illustrated in Fig. 2B. A BLAST search of the National Biomedical Research Foundation protein data base revealed significant homology of the repetitive sequence of the islet CaI-PLA 2 to a number of ankyrin-related proteins (ARP), including the Caenorhabditis elegans ARP (26) and both murine (27) and human (28) red cell ankyrin. Dot matrix anal-ysis revealed that the repetitive sequence motif of the islet CaI-PLA 2 is highly homologous to that of an 89-kDa domain of murine red cell ankyrin (Fig. 3A) that contains binding sites for tubulin and for several integral membrane proteins (28 -30). Alignment of the repeating domain of islet CaI-PLA 2 with the repeating domain of ankyrin indicated that both repeating motifs have very similar consensus sequences (Fig. 3B). The CHO cell CaI-PLA 2 is also known to contain ankyrin-like repeats (11).
Bacterial Expression of the Protein Encoded by the Cloned Islet CaI-PLA 2 cDNA-The open reading frame of the putative islet CaI-PLA 2 cDNA was subcloned in-frame into the bacterial expression vector pET-28c (Novagen) to generate a fusion protein containing the islet CaI-PLA 2 sequence and both polyhistidine and T7 epitope tag sequences. Upon isopropyl-1-thio-␤-D-galactopyranoside induction of bacterial cultures transfected with vector containing the islet CaI-PLA 2 insert, strong expression of the expected 87-kDa fusion protein occurred, and this material was not observed in control cultures (Fig. 4). The expressed protein was recognized by a monoclonal antibody raised against a synthetic peptide contained within the islet CaI-PLA 2 sequence.
Expression of CaI-PLA 2 Activity in COS-7 and CHO Cells After Transient Transfection with the Cloned Islet CaI-PLA 2 cDNA-For eukaryotic expression, the islet CaI-PLA 2 cDNA was inserted into the vector pcDNA3.1 (Invitrogen), and this construct (pcDNA3.1-CaI-PLA 2 ) was used to transfect both COS-7 and CHO cells. At 72 h after transfection with this construct or with vector containing no insert, CaI-PLA 2 activity was measured in cytosolic fractions prepared from these cells. No measurable CaI-PLA 2 activity was observed in the cytosol of either untransfected COS-7 cells (not shown) or COS-7 cells transfected with vector containing no insert (Fig. 5A). In contrast, COS-7 cells transfected with the pcDNA3.1-CaI-PLA 2 construct exhibited substantial amounts of cytosolic CaI-PLA 2 activity (Fig. 5A). This activity was inhibited by pretreatment of cytosol with a suicide substrate (HELSS) (Fig. 5A) that has previously been demonstrated to inhibit CaI-PLA 2 activity in islet cytosol (7). Similar results were obtained with CHO cells. Both untransfected CHO cells (not shown) and CHO cells transfected with vector containing no insert exhibited measurable CaI-PLA 2 activity (Fig. 5B). This was expected because a CaI-PLA 2 enzyme has been cloned from CHO cells (11). Transient transfection of CHO cells with the pcDNA3.1-CaI-PLA 2 construct, however, resulted in a 5.5-fold increase in cytosolic CaI-PLA 2 activity, and this activity was inhibited by treatment of cytosol with HELSS (Fig. 5B).

Characterization of CaI-PLA 2 Activity in COS-7 Cells Transiently Transfected with the Islet CaI-PLA 2 cDNA Construct-
The concentration dependence of inhibition by HELSS of CaI-PLA 2 activity in COS-7 cells transfected with the islet CaI-PLA 2 cDNA construct was found to be very similar to that of the CaI-PLA 2 activity in native HIT insulinoma cell cytosol (Fig. 6). Like the CaI-PLA 2 activity in islet and HIT cell cytosol (7,22), the CaI-PLA 2 activity in cytosol from transiently transfected COS-7 cells was optimal near neutrality and was stimulated by including 1 mM ATP in the assay solution (Fig. 7). The effects of ATP to stimulate cytosolic CaI-PLA 2 activity in transiently transfected COS-7 cells, however, differed in some respects from those observed with islet (7) or HIT cell (22) CaI-PLA 2 activity. First, the ATP concentration dependence for stimulation of CaI-PLA 2 activity from transiently transfected COS-7 cells was bell-shaped, and the stimulatory effect decreased at ATP concentrations exceeding 1 mM (not shown). CaI-PLA 2 activity in cytosol from islets (7) or HIT cells (22) is stimulated by ATP concentrations as high as 10 mM. Second, there was less nucleotide phosphate selectivity with the CaI-PLA 2 activity from transfected COS-7 cells than with islet (7) or HIT cell (22) activity. Although ATP was more effective than AMP or ADP, ADP did exert some activating influence on CaI-PLA 2 activity from transfected COS-7 cells (not shown). ADP does not stimulate HIT cell CaI-PLA 2 activity and sup-presses the stimulatory effect of ATP (22). In addition, GTP and UTP were nearly as effective as ATP in stimulating CaI-PLA 2 activity from transfected COS-7 cells (not shown), whereas GTP has little effect on HIT cell CaI-PLA 2 activity (22). Finally, the non-hydrolyzable ATP analog AMP-PCP stimulates CaI-PLA 2 activity from transfected COS-7 cells less effectively than ATP (not shown), whereas the two compounds are equipotent stimulators of HIT cell CaI-PLA 2 activity (22).
Expression of CaI-PLA 2 mRNA in Rat Islet Cells-To determine whether islet CaI-PLA 2 mRNA is specifically expressed in ␤-cells within islets, a competitive RT-PCR (18,19) method was developed. A competitor DNA was prepared that shares with CaI-PLA 2 cDNA sequences recognized by the primers but that yields a smaller product than that derived from the CaI-PLA 2 cDNA (Fig. 8A). The ratio of signals from the target and competitor was found to correspond to the input DNA over a wide range of concentrations (Fig. 8B). Single cell suspensions were then prepared from isolated islets and analyzed by fluorescence-activated cell sorting (5,7,10,14,15) to yield two populations of cells. One population contains over 90% ␤-cells, and the second contains 10 -15% ␤-cells and about 70% ␣-cells (5,7,10,14,15). RNA was then prepared from the two populations of cells and used as template in RT-PCR reactions with CaI-PLA 2specific primers and the competitor DNA. At equivalent amounts of input RNA, a substantially higher target to competitor ratio was observed with RNA from the ␤-cell-enriched population than with that from the ␣-cell-enriched population ( Fig. 9). This suggests that the RNA species amplified in RT-PCR reactions with the ␤-cell-enriched population did not arise from contaminating non-␤-cells and that ␤-cells contain CaI-PLA 2 mRNA.
Expression of CaI-PLA 2 mRNA was also examined in other tissues by competitive RT-PCR. At equivalent amounts of input RNA, the highest target to competitor ratio (TCR) among tissues examined was observed in brain (TCR 1.3). Target signal was also observed in lung (TCR 0.6), spleen (TCR 0.4), kidney (TCR 0.4), liver (TCR 0.2), heart (TCR 0.1), and skeletal muscle (TCR 0.1).

DISCUSSION
The studies described here indicate that pancreatic islet ␤-cells express an 84-kDa CaI-PLA 2 enzyme that is highly homologous to an enzyme recently cloned from CHO cells (11). The occurrence of ankyrin-like repeats in the islet CaI-PLA 2 sequence raises interesting possibilities about the function of this protein in ␤-cells. The domain of ankyrin that contains similar repeats is responsible for the binding of ankyrin to a number of integral membrane proteins that govern ionic fluxes across cellular membranes (29), including the Na ϩ /K ϩ -ATPase of renal basolateral membranes (31-33), a renal amiloridesensitive Na ϩ channel (34), the red cell anion exchanger (35,36), a cerebellar inositol-trisphosphate receptor (37), and a voltage-dependent Na ϩ channel in myelinated neurons (29).
Regulation of ionic fluxes at the ␤-cell plasma membrane is critical in the control of insulin secretion, and both inactivation of K ATP channels and activation of voltage-operated Ca 2ϩ channels are required for the induction of insulin secretion by glucose (2). One product of the action of PLA 2 on islet phospholipids is nonesterified arachidonic acid. Arachidonic acid is the vastly predominant sn-2 fatty acid substituent in islet phospholipids and comprises 36% of the total fatty acyl mass in islet phospholipids (9, 10). Nonesterified arachidonic acid facilitates FIG. 3. Alignment of the islet CaI-PLA 2 amino acid sequence with repeating domains of ankyrin. A, dot matrix analysis of the amino acid sequence of CaI-PLA 2 versus that of mouse erythrocyte ankyrin. The dot matrix analysis was performed using high stringency parameters (Ͼ20), as described in Fig. 2.  (lanes 1, 2, and 5) or by Western blotting with anti-CaI-PLA 2 antibody (lanes 3 and 4). Lanes 1 and 3 represent negative controls in which the expression host cell E. coli strain BL21(DE3) was transformed with pET28c vector containing no insert. Lanes 2 and 4 represent transformation of the host cell with the pET28-CaI-PLA 2 construct. For Western blotting analyses (lane 3 and 4), proteins were transferred to a nylon membrane and probed with a monoclonal anti-CaI-PLA 2 antibody, as described under "Experimental Procedures." Lane 5 represents expressed protein, which contains a polyhistidine tag sequence, after affinity purification by nickel-chelate chromatography.
Ca 2ϩ entry into cells through both voltage-operated (38) and receptor-operated (39) Ca 2ϩ channels and induces a rise in ␤-cell cytosolic [Ca 2ϩ ] that is dependent on Ca 2ϩ entry from the extracellular space (5). Both exogenous PLA 2 and the products of its action suppress K ATP channel activity in excised and cell-attached patches of plasma membranes from HIT insulinoma cells (40).
Taken together, the observations summarized in the preceding two paragraphs raise the possibility that the ankyrin-like repeat domains of the islet CaI-PLA 2 might serve to attach the (last three lanes), which contains the islet CaI-PLA 2 cDNA as insert. All assays were performed in buffer containing 10 mM EGTA. In the 1st (VO) and 4th (VI) lanes, cytosol was not pretreated before CaI-PLA 2 activity was measured. In the 3rd (VOϩH) and 6th (VIϩH) lanes, cytosol was incubated with the CaI-PLA 2 inhibitor HELSS before CaI-PLA 2 activity was measured. In the 2nd (VOϩE) and 5th (VIϩE) lanes, cytosol was incubated with ethanolic vehicle that did not contain HELSS before CaI-PLA 2 activity was measured. Cytosol was prepared from native (non-transfected) HIT-T15 cells and from COS-7 cells that had been transfected with the pcDNA3.1-CaI-PLA 2 construct, as described under "Experimental Procedures." Cytosol was preincubated at room temperature for 2 min in assay buffer containing no added Ca 2ϩ , 5 mM EGTA, and either no HELSS (control) or varied concentrations of HELSS (10 Ϫ14 to 10 Ϫ6 M). After the preincubation, the temperature of the assay solution was increased to 37°C, and enzymatic reactions were initiated by addition of substrate. CaI-PLA 2 activity was measured as described as under "Experimental Procedures" and is plotted as a percent of control activity observed in cytosol that was not exposed to HELSS. enzyme to ␤-cell K ATP channels or to voltage-operated Ca 2ϩ channels or to anchor the enzyme in the membrane in close proximity to such channels. The ␣-subunits of voltage-operated Ca 2ϩ channels exhibit striking similarities in sequence to voltage-operated Na ϩ channels (41,42) that interact with ankyrin through its repeating domain (29). Ankyrin is distributed in close proximity to voltage-operated Ca 2ϩ channels in the triad junctional structures of skeletal myocytes, although direct interaction of ankyrin with these channels has not been demonstrated (43).
The repeat sequences in ankyrin are also responsible for its binding to the cytoskeletal component tubulin (29). Although the precise role of the cytoskeletal apparatus in insulin secretion is not clearly established, a number of agents that interfere with microtubule assembly or disassembly inhibit insulin secretion (44 -47), and morphological evidence suggests a direct association of microtubules with insulin secretory vesicles (48). Electron microscopic evidence demonstrates secretory granules associated with microtubules and directed toward the plasma membrane (48). An association of the islet PLA 2 with microtubular structures mediated by its ankyrin-like repeat domain might facilitate interaction of the enzyme with secretory granule membranes and/or plasma membranes. It is of interest in this regard that treatment of parotid acinar secretory granules with PLA 2 greatly increases their tendency to fuse with plasma membranes (49), which is the final event in exocytosis.
The sensitivity of the islet CaI-PLA 2 to inhibition by the HELSS suggests that this enzyme could be the target within ␤-cells that is responsible for the ability of HELSS to inhibit the glucose-induced rise in ␤-cell cytosolic [Ca 2ϩ ] and insulin se-cretion (7-10), although HELSS has also recently been demonstrated to inhibit an isozyme of phosphatidate phosphohydrolase (50). HELSS does not inhibit any known Ca 2ϩ -dependent PLA 2 of groups I through IV (51-54), but Ca 2ϩ -independent PLA 2 activities from several sources are sensitive to inhibition by HELSS. These include the 84-kDa CHO cell CaI-PLA 2 (11) and a PLA 2 from the macrophage-like P388D1 cell line (51,52,55,56), which has recently been demonstrated to represent the mouse homolog of the CHO cell CaI-PLA 2 (61). In addition, Ca 2ϩ -independent, HELSS-sensitive PLA 2 activities thought to reside in 40-kDa proteins have been described in myocardium (53), HIT cells (22), and renal proximal tubular cells (57).
It is not yet known whether these 40-kDa proteins are related to the 84-kDa CaI-PLA 2 enzymes, although generation of the former from the latter by proteolytic processing is a formal possibility. It is also possible that the 40-kDa proteins are separate gene products that share sequence homology with the 84-kDa CaI-PLA 2 enzymes, perhaps including the serine lipase consensus sequence that is likely to confer sensitivity to HELSS. The 40-kDa proteins from myocardium (58) and HIT cells (22,59) are thought to be associated with 85-kDa protein FIG. 7. Influence of pH and of ATP on CaI-PLA 2 activity expressed in transfected COS-7 cells. Cytosol was prepared from COS-7 cells that had been transfected with the pcDNA3.1-CaI-PLA 2 construct, and CaI-PLA 2 activity was measured as in Fig. 6. For pH values from 7 to 9, assay solution was buffered with 100 mM Tris-HCl. For pH values from 4 to 6, assay solution was buffered with 100 mM sodium acetate. Open symbols reflect assays performed without ATP, and closed symbols reflect assays performed with ATP (1 mM). CaI-PLA 2 activity is plotted as dpm of [ 14 C]linoleate released from L-␣-1palmitoyl-2-[ 14 C]linoleoyl-phosphatidylcholine substrate.
FIG. 8. Determination of the abundance of CaI-PLA 2 mRNA by competitive RT-PCR. A, agarose gel electrophoresis of products obtained from competitive PCR reactions. Competitive RT-PCR was performed as described under "Experimental Procedures," and products were analyzed by 1% agarose gel electrophoresis and visualized with ethidium bromide. Lane 1 represents the product obtained when only CaI-PLA 2 cDNA and no competitor was included in the reaction mixture. Lane 10 represents the product obtained when only competitor DNA and no CaI-PLA 2 cDNA was included in the reaction mixture. Lanes 2-8 represent products obtained when a fixed amount of competitor DNA and serial 10-fold dilutions of CaI-PLA 2 cDNA were included in the reaction mixture. B, relationship of target to competitor signal ratio to input DNA. The relative intensities of the product bands from the gel in A were determined with an IS-1000 Digital Imaging System and plotted as a function of input target DNA.
subunits that are homologous to phosphofructokinase and that confer sensitivity to activation by ATP and to modulation of this effect by ADP. Experiments with the CaI-PLA 2 enzymes from CHO cells (62) and P388D1 cells (51) have failed to demonstrate association with phosphofructokinase-like proteins, although both enzymes behave as 300 -350-kDa multimers on gel filtration chromatography, as do CaI-PLA 2 activities from myocardium (58) and HIT cells (59). Whether the 84-kDa CaI-PLA 2 enzymes may exist as hetero-oligomers consisting of distinct catalytic and regulatory proteins in some cells but not others is a question that will require further examination.
Although the 84-kDa islet CaI-PLA 2 expressed in transiently transfected COS-7 cells is sensitive to activation by ATP under the assay conditions employed in this study, the nature of this effect differs in several respects from that of CaI-PLA 2 activities in islet and HIT cell cytosol and is in general quite similar to effects observed with the P388D1 cell CaI-PLA 2 (51). The possibility that the differences in nucleotide phosphate sensitivity between the islet CaI-PLA 2 expressed in transfected COS-7 cells and that in native islet cytosol might reflect the influence of an interacting protein that is expressed in islets but that is not abundantly expressed in COS-7 cells is under study. It is of interest that the CHO cell CaI-PLA 2 , after expression in an Sf9 cell host, has recently been demonstrated to interact avidly and selectively with both ATP affinity matrices and with calmodulin affinity matrices (60), suggesting that the enzyme may bind ATP and interact with other cytosolic proteins. FIG. 9. Relative abundance of CaI-PLA 2 mRNA in pancreatic islet ␤-cells and in non-␤-cells cells as estimated by competitive RT-PCR. Single cell suspensions were prepared from isolated pancreatic islets with the enzyme dispase and subjected to fluorescence-activated cell sorting to yield a population of nearly homogeneous ␤-cells and a population containing about 70% ␤-cells, 10 -15% ␤-cells, and 15-20% other cells. The second population is designated "non-␤-cells" in the figure. RNA was then prepared from the two populations of cells and used as template in competitive RT-PCR reactions with CaI-PLA 2specific primers and the competitor DNA species described under "Experimental Procedures." Products were analyzed, and the relative intensity of the target to competitor bands was determined as in Fig. 8.