cAMP-dependent Mobilization of Intracellular Ca2+ Stores by Activation of Ryanodine Receptors in Pancreatic β-Cells

Glucagon-like peptide-1 (GLP-1) is an intestinally derived insulinotropic hormone currently under investigation for use as a novel therapeutic agent in the treatment of type 2 diabetes mellitus. In vitro studies of pancreatic islets of Langerhans demonstrated that GLP-1 interacts with specific β-cell G protein-coupled receptors, thereby facilitating insulin exocytosis by raising intracellular levels of cAMP and Ca2+. Here we report that the stimulatory influence of GLP-1 on Ca2+ signaling results, in part, from cAMP-dependent mobilization of ryanodine-sensitive Ca2+ stores. Studies of human, rat, and mouse β-cells demonstrate that the binding of a fluorescent derivative of ryanodine (BODIPY FL-X ryanodine) to its receptors is specific, reversible, and of high affinity. Rat islets and BTC3 insulinoma cells are shown by reverse transcriptase polymerase chain reaction analyses to express mRNA corresponding to the type 2 isoform of ryanodine receptor-intracellular Ca2+ release channel (RYR2). Single-cell measurements of [Ca2+] i using primary cultures of rat and human β-cells indicate that GLP-1 facilitates Ca2+-induced Ca2+ release (CICR), whereby mobilization of Ca2+ stores is triggered by influx of Ca2+ through l-type Ca2+ channels. In these cells, GLP-1 is shown to interact with metabolism ofd-glucose to produce a fast transient increase of [Ca2+] i . This effect is reproduced by 8-Br-cAMP, but is blocked by a GLP-1 receptor antagonist (exendin-(9–39)), a cAMP antagonist ((Rp)-cAMPS), anl-type Ca2+ channel antagonist (nimodipine), an antagonist of the sarco(endo)plasmic reticulum Ca2+ ATPase (thapsigargin), or by ryanodine. Characterization of the CICR mechanism by voltage clamp analysis also demonstrates a stimulation of Ca2+ release by caffeine. These findings provide new support for a model of β-cell signal transduction whereby GLP-1 promotes CICR by sensitizing intracellular Ca2+ release channels to the stimulatory influence of cytosolic Ca2+.

Glucagon-like peptide-1 (GLP-1) 1 is a potent blood glucoselowering hormone that stimulates secretion of insulin from pancreatic ␤-cells, and which is under intensive clinical investigation for use in treatment of type 2 diabetes mellitus (1,2). The two naturally occurring and biologically active isopeptides of GLP-1 with therapeutic potential are GLP-1-(7-37) and GLP-1-(7-36)amide (3,4). They are derived by post-translational processing of proglucagon, and are synthesized and secreted by enteroendocrine L-cells of the distal intestine. The timing of the secretion of GLP-1 from these cells is such that circulating levels of the hormone rise coincident with the postprandial increase of blood glucose concentration. By binding to its receptors located on ␤-cells, GLP-1 synergizes with glucose to induce insulin secretion, thereby increasing the concentration of circulating insulin to a level above that attributable to glucose alone. This augmentation of glucose-stimulated insulin secretion by the gut hormone GLP-1 is termed the incretin effect and plays a critical role in the maintenance of systemic glucose homeostasis (5).
The GLP-1 receptor located on ␤-cells is a G protein-coupled receptor that mediates stimulatory effects of the hormone on adenylyl cyclase (6). GLP-1 raises intracellular levels of cyclic cAMP, activates protein kinase A (PKA), and exerts pleiotropic effects on ␤-cell signal transduction (7)(8)(9). PKA-mediated protein phosphorylation interacts with metabolism of D-glucose at a late step in stimulus-secretion coupling to facilitate exocytosis of insulin-containing secretory granules (10 -12). GLP-1 also produces cellular depolarization (13)(14)(15) and a large increase of [Ca 2ϩ ] i (14, 16 -20), effects reproduced by activators of cAMP signaling (21)(22)(23). The best evidence available indicates that the GLP-1-induced rise of [Ca 2ϩ ] i serves as an important trigger for exocytosis of insulin. The vital role Ca 2ϩ plays in this process is emphasized by studies demonstrating that the insulin secretagogic effect of GLP-1 is markedly attenuated by treatments that prevent the rise of [Ca 2ϩ ] i (16,24).
We have focused on early events in GLP-1 signal transduction in order to ascertain how the rise of [Ca 2ϩ ] i that serves as a trigger for secretion of insulin is achieved. In general, the alterations of [Ca 2ϩ ] i observed upon exposure of ␤-cells to GLP-1 can be described as having transient (seconds) or sustained (minutes) kinetics. The sustained rise of [Ca 2ϩ ] i results from influx of Ca 2ϩ that is a consequence of three distinct mechanisms. First, GLP-1 potentiates glucose-induced closure of ATP-sensitive K ϩ channels (K-ATP) (13), thereby generating cellular depolarization, activation of voltage-dependent Ca 2ϩ channels (VDCCs), and influx of Ca 2ϩ . Second, GLP-1 exerts a direct stimulatory influence on the entry of Ca 2ϩ through dihydropyridine-sensitive (L-type) VDCCs (9). Third, GLP-1 stimulates the opening of Ca 2ϩ -activated nonselective cation channels that are permeant to Ca 2ϩ as well as to Na ϩ (14,25). These three processes act in concert to augment oscillatory electrical activity, Ca 2ϩ influx, and pulsatile secretion of insulin characteristic of whole islets of Langerhans.
More poorly understood is the fast transient increase of [Ca 2ϩ ] i observed in response to GLP-1 (18,19). Here we report that this action of the hormone requires the presence of extracellular Ca 2ϩ , can occur independent of any change of membrane potential, and is reproduced by pharmacological activators of cAMP signaling. Typically, such transient effects of hormones on [Ca 2ϩ ] i are considered to be a consequence of the mobilization of Ca 2ϩ stores. GLP-1 might stimulate release of inositol trisphosphate (IP 3 )-sensitive Ca 2ϩ stores because it increases inositol phosphate production in COS-7 cells transfected with the recombinant GLP-1 receptor (26) or in Xenopus oocytes injected with GLP-1 receptor mRNA (27). However, GLP-1 has only a small effect on inositol phosphate production in freshly isolated islets (24,28) or HIT-T15 insulinoma cells (16). Therefore, an issue of central importance to our understanding of GLP-1 signal transduction concerns exactly how cAMP interacts with ␤-cell Ca 2ϩ stores to promote their mobilization.
It has been proposed that Ca 2ϩ is mobilized by GLP-1 during the process of Ca 2ϩ -induced Ca 2ϩ release (CICR), whereby influx of Ca 2ϩ across the plasma membrane triggers release of Ca 2ϩ from Ca 2ϩ stores regulated by ryanodine receptor Ca 2ϩ release channels (20). This hypothesis is controversial due to conflicting findings concerning what role ryanodine receptors play as regulators of ␤-cell Ca 2ϩ signaling. One objection raised concerns the uncertain effectiveness of cyclic ADP-ribose (cADPR) in this system. cADPR is a candidate second messenger that mobilizes ryanodine-sensitive Ca 2ϩ stores in some cell types (29). It is derived from ␤-NAD ϩ via an enzymatic process catalyzed by an ADP-ribosyl cyclase designated CD38 (30). Okamoto and co-workers reported that an IP 3 -insensitive Ca 2ϩ store was mobilized by cADPR in rat islet microsomes (31)(32)(33). However, cADPR failed to act in ob/ob mouse ␤-cells or RINm5F insulinoma cells (34 -38). Ca 2ϩ signaling in rat ␤-cells was also not influenced by an antagonist of cADPR (37,39), and levels of cADPR in rat islets were insensitive to alterations of glucose concentration (40), findings that contradict the Okamoto hypothesis. Therefore, the significance of ryanodine receptors as mediators of GLP-1 signal transduction remained unclear.
Here we report functional expression of ryanodine receptors in ␤-cells derived from human, rat, and C57BL/6J mouse islets. Assays of islet mRNA by the reverse transcriptase-polymerase chain reaction (RT-PCR) indicated a preferential expression of the type 2 isoform of ryanodine receptor (RYR2) in these tissues. The subcellular location of ryanodine receptors was visualized in ␤-cells by fluorescence microscopy using BODIPY FL-X ryanodine as a fluorophore. Ca 2ϩ imaging and patch clamp analyses of ␤-cells demonstrated that GLP-1 produces a fast transient increase of [Ca 2ϩ ] i as a consequence of CICR initiated by influx of Ca 2ϩ through L-type VDCCs. It is concluded that this process of CICR is mediated by a cAMP-signaling system that targets a ryanodine-sensitive source of intracellular Ca 2ϩ .

EXPERIMENTAL PROCEDURES
Preparation of Human, Rat, and Mouse Islet Cell Cultures-Human islets were obtained from Dr. C. Ricordi (Diabetes Research Institute, University of Miami School of Medicine, Miami, FL). Rat islets were isolated from pentobarbital-anesthetized male Sprague-Dawley rats (200 -250 g; Charles River Laboratories, Wilmington, MA) according Lacy and Kostianovsky (41) by digestion with collagenase (Roche Molecular Biochemicals). Mouse islets were isolated by digestion of pancreata obtained from male C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME). Islets were dispersed into single cell suspensions by digestion with trypsin-EDTA, and by trituration through a fire-polished Pasteur pipette. Cell suspensions were plated onto glass coverslips (25CIR-1; Fisher) coated with 1 mg/ml concanavalin A (type V; Sigma), which facilitates adherence of ␤-cells to glass surfaces. Cell cultures were maintained in a humidified incubator (95% air, 5% CO 2 ) at 37°C in RPMI 1640 culture media containing 11.1 mM glucose, 2 mM glutamine, 10% fetal bovine serum (FBS), 100 units/ml penicillin, and 100 g/ml streptomycin. RT-PCR Analyses of Ryanodine Receptor Isoforms-Total RNA was extracted from rat islets and insulinoma cell cultures using guanidinium thiocyanate. The RNA was fractionated by density gradient centrifugation on a CsCl cushion, and poly(A) mRNA was isolated by chromatographic separation on an oligo(dT) affinity column. First strand cDNA synthesis was initiated using oligo(dT) primers and avian myeloblastosis virus reverse transcriptase (Copy Kit; Invitrogen Corp., Carlsbad, CA). A cDNA synthesis reaction to which no reverse transcriptase was added served as a negative control. PCR primer pairs corresponding to RYR isoforms were identified by analysis of the published sequences (42,43) of RYR cDNAs using Oligo version 4.0 primer analysis software (National Biosciences Inc., Plymouth, MN). The design of the PCR primers is listed in Table I. PCR reactions were catalyzed using recombinant Taq DNA polymerase, PCR reaction buffer, and dNTPs supplied by TaKaRa Biologicals (Shiga, Japan). The thermal cycling parameters consisted of an initial denaturation step for 5 min at 95°C, followed by 36 cycles consisting of: 30 s at 95°C, 60 s at 55°C, and 90 s at 72°C. The final extension step was for 5 min at 72°C. PCR products were resolved by 1% agarose gel electrophoresis, and DNA was stained with ethidium bromide for fluorescence detection. PCR products corresponding to RYR were extracted from agarose gels by solubilization in NaI and by extraction with silica (GeneClean II; Bio 101 Inc., La Jolla, CA). The purified PCR products were ligated into the pCRII cloning vector by the T/A cloning method (Invitrogen Corp.). Ligation products were used to transform INV␣FЈ cells (Invitrogen Corp.), and transformants were isolated by antibiotic resistance selection on agar plates containing ampicillin. Individual clones were selected and amplified by inoculation of Luria broth. Cloned plasmid DNAs were extracted from bacterial cultures using the Rapid Pure Miniprep procedure (Bio 101 Inc.). Determination of plasmid DNA sequences containing RYR PCR products was accomplished by the dideoxynucleotide sequencing method of Sanger in combination with polyacrylamide gel electrophoresis (Sequenase version 2.0 DNA sequencing kit; U. S. Biochemical Corp.).
Fluorescence Microscopy, Immunocytochemistry, and FluorImager Analyses-The binding of BODIPY FL-X ryanodine (Molecular Probes Inc., Eugene, OR) to its receptor was evaluated by fluorescence microscopy using primary cultures of dispersed islet cells. An Eclipse E600 upright microscope (Nikon, Melville, NY) equipped with 20, 40, and 100ϫ Plan Fluor objectives was interfaced with a digital video imaging system comprising a Photometrics Ltd. Sensys cooled CCD camera (Tucson, AZ) and a MacIntosh G3 computer (Apple Computers, Cupertino, CA) running IP Lab Spectrum software (Scanalytics Inc., Fairfax, VA). A mercury lamp served as the excitation light source, and the emitted light was filtered using standard fluorescein and rhodamine filter sets. Individual islet cells were allowed to adhere to the numbered grid surfaces of CELLocate coverslips (55-m square size; Eppendorf, Hamburg, Germany). The cells were fixed in 4% paraformaldehyde and permeabilized using 0.1% Triton X-100. The fixed and permeabilized cells were incubated at room temperature in Hanks' balanced salt solution containing the specified concentration of BODIPY FL-X ryanodine. Immunocytochemistry for detection of insulin-like immunoreactivity was performed using 5% normal goat antiserum as a blocking agent. The primary antiserum was guinea pig anti-insulin (Dako, Carpinteria, CA) used at a dilution factor of 1:100. The secondary antiserum was rhodamine-conjugated goat anti-guinea pig (Jackson ImmunoResearch Laboratories, West Grove, PA). Quantitative analyses of islet cell suspensions were performed using a Molecular Dynamics FluorImager running ImageQuant software (Sunnyvale, CA). An argon laser served as the excitation light source, and fluorescence was imaged on 96-well tissue culture plates (Falcon 3072; ϳ20,000 islet cells/well) using a fluorescein 530-nm discriminating filter set. Estimates of RYR2 affinity were based on the use of permeabilized cells where access of BODIPY FL-X ryanodine to its binding site was unimpeded. Although the apparent K d for fluorophore binding was ϳ4 nM, higher concentrations of ryanodine were required to block CICR in living cells, where access of ryanodine to its binding site was impeded.
Measurement of Intracellular Ca 2ϩ Concentration-Cells were maintained on glass coverslips 24 -48 h prior to each experiment. The cell culture media was replaced with a standard extracellular salt solution (SES) containing (in mM): 138 NaCl, 5.6 KCl, 2.6 CaCl 2 , 1.2 mM MgCl 2 , 10 mM HEPES (pH 7.4), and 7.5 mM D-glucose. The SES was supplemented with 1 M fura-2 acetoxymethyl ester (fura-2-AM; Molecular Probes Inc.), 2% fetal bovine serum, and 0.02% Pluronic F-127 (w/v; Molecular Probes Inc.). Cells were exposed to the fura-2 AM-containing solution for 15 min at 22°C. The loading solution was then removed, and the cells were equilibrated in fresh SES for 10 min at 22°C. Experiments were performed at 32°C using a Zeiss IM35 inverted microscope (Thornwood, NJ) outfitted with a temperature-controlled stage, a superfusion system, and a 100ϫ Nikon UVF objective. Dual wavelength excitation microspectrofluorimetry was performed ratiometrically at 1-s intervals using a digital video imaging system (Ion-Optix Corp., Milton, MA). Calibration of the raw fluorescence values was performed as described previously (22) using fura-2 pentapotassium salt dissolved in calibration buffer solutions from Molecular Probes (Calcium Calibration Kit 1 with Magnesium). Insulin-containing ␤-cells were positively identified at the end of each experiment by fluorescence microscopy in combination with immunocytochemistry.
Patch Clamp Electrophysiological Techniques-The resting potential and holding current were measured under current clamp or voltage clamp using the tight seal, whole cell, perforated patch configuration (13,14). Pipettes were pulled from borosilicate glass (Kimax-51; Kimble Glass Inc., Vineland, NJ), fire-polished (final tip resistances of 2-3 megohms), and tip-dipped in a standard patch pipette solution containing (in mM): 95 Cs 2 SO 4 , 7 MgCl 2 , and 5 HEPES (pH 7.4). Pipettes were then back-filled with patch pipette solution supplemented with 240 g/ml nystatin. The pipette solutions did not contain fura-2 and were considered to be nominally Ca 2ϩ -free. Patch pipettes were connected to the head stage of an EPC-9 patch clamp amplifier (Heka Electroniks, Lambrecht, Germany) interfaced with a personal computer running Pulse version 8.0 acquisition software (Instrutech Corp., Mineola, NY). The series resistance (R s ) and cell capacitance (C m ) were monitored following seal formation, and experiments were conducted when R s declined to 12-25 megohms and C m increased to 10 -40 picofarads. In voltage clamp experiments, R s was compensated for by 60%. Electrical access to the cytosol was confirmed by noting approximately Ϫ60-mV resting potential, and by noting an increase of [Ca 2ϩ ] i in response to a voltage step from Ϫ70 to 0 mV. Experiments were rejected if such a rise of [Ca 2ϩ ] i was not observed. Measurements of membrane potential and current were digitized using an Instrutech VR-10A digital data recorder and stored on magnetic tape for subsequent analysis using pClamp version 6.0 software (Axon Instruments Corp., Foster City, CA).
Preparation of Test Solutions-GLP-1- , exendin-4, and exendin-(9 -39) were from Peninsula Laboratories (Belmont, CA). 8-Br-cAMP, nimodipine, ryanodine, caffeine, and thapsigargin were from Sigma. (Rp)-cAMPS was from BioLog Life Sciences Institute (Bremen, Germany). Test solutions were dissolved in SES prior to each experiment. For studies examining the effects of peptides, the SES also contained 0.05% human serum albumin (fraction V; Sigma) to protect against their absorption to glass or plastic surfaces. Solutions to which no Ca 2ϩ was added were prepared by substituting MgCl 2 for CaCl 2 . Solutions containing 44 mM KCl were prepared by substituting KCl for NaCl on an isosmotic basis. Test solutions were applied to individual cells from micropipettes using a PicoSpritzer II pneumatic pressure ejection system (General Valve Corp., NJ) as described previously (13).

BODIPY FL-X Ryanodine as a Fluorescent Probe for Detection of Ryanodine Receptors-
We tested for expression of high affinity ryanodine receptors in pancreatic ␤-cells by using the fluorophore BODIPY FL-X ryanodine. Cells derived from dispersed human, rat, or mouse islets were allowed to adhere to coverslips, on which was etched a numbered grid system. This allowed positive identification of individual cells by microscopic inspection. Cells were fixed with paraformaldehyde, permeabilized with Triton X-100, and bathed for 1 h at 22°C in saline containing BODIPY FL-X ryanodine. A microscope-based digital imaging system was used to acquire images of fluorescence attributable to binding of BODIPY FL-X ryanodine to its receptors.
Specific binding of BODIPY FL-X ryanodine to human islet cells was observed using a low concentration (1 nM) of the fluorophore (Fig. 1, panel A). A perinuclear pattern of fluorescence was observed (panel A), consistent with expression of ryanodine receptors in a subcellular compartment that is likely to include the endoplasmic reticulum. The affinity of BODIPY FL-X ryanodine for these receptor sites appeared to be cell type-specific because not all cells in a cluster of human islet cells were labeled (arrows, panel B). BODIPY FL-X, itself, was not effective in this assay (data not shown), thereby confirming that the unconjugated fluorophore has no binding affinity. These human islet cells labeled by BODIPY FL-X ryanodine included authentic ␤-cells, as demonstrated by their co-expression of insulin-like immunoreactivity (cf. panels C and D). Quantitative analyses of human islet cells using a FluorImager demonstrated that the binding of BODIPY FL-X ryanodine was saturable and reversible ( Fig. 2A). Pretreatment of human islet cells with 1 M non-fluorescent ryanodine blocked the subsequent binding of BODIPY FL-X ryanodine to its receptor (Fig.  2B). The IC 50 for displacement of binding was determined to be ϳ4 nM. These experiments conducted with human islet cells were reproduced successfully using islets derived from Sprague-Dawley rats and C57BL/6J mice (data not shown).
RT-PCR Analyses of Ryanodine Receptor Isoforms-RT-PCR analyses of the three known isoforms of RYR were performed using PCR primers that match the published partial sequences (42,43) of cDNAs corresponding to mouse ryanodine receptors (Table I). These primers target regions of RYR cDNA that code for the M1-M4 membrane-spanning and carboxyl terminus portions of the RYR protein (44). The samples of cDNA we tested were derived from poly(A) mRNA extracted from rat islets. Amplification of rat islet cDNA using an RYR2-specific primer set (F04/R04) demonstrated that a product of the expected size (423 bp) was resolved by gel electrophoresis (Fig. 3A). In contrast, PCR products were not obtained when we amplified aliquots of a rat islet cDNA synthesis reaction mixture to which no reverse transcriptase was added ( Fig. 3A; negative control). The cloned 423-bp PCR product derived from rat islets was sequenced in both directions and evaluated by a National Center for Biotechnology BLAST analysis using the GenBank data base. Determination of the nucleic acid sequence confirmed that it was homologous (Ͼ90% identity) to the RYR2 sequence derived by screening of a mouse cDNA library (42). This sequence corresponds to sequence 13084 -13506 (see Ref. 45) of rabbit cardiac RYR2.
It may be argued that our samples of rat islet cDNA are contaminated by cDNAs derived from non-endocrine cell types that express RYR2. Therefore, we sought to determine if RYR2 is detectable by RT-PCR of cDNA derived from insulin-secreting cell lines that serve as model systems for analysis of ␤-cell function. To this end, poly(A) mRNA was extracted from mouse BTC3 cells for use in synthesis of cDNA. As was the case for rat islet cDNA, the RYR2-specific F04/R04 primer set generated a 423-bp PCR product when using BTC3 cDNA as template (Fig.  3A). When cloned and characterized, this BTC3 PCR product was found to be identical in sequence to that reported for mouse RYR2. Three additional primer sets were chosen to validate the presence of RYR2 mRNA in rat islets or insulinoma cells. RYR2 F01/R01 and F02/R02 primer pairs (Table I) generated PCR products of the expected sizes (303 and 258 bp, respectively) when used for amplification of cDNA from rat islets (Fig. 3B) and BTC3 cells (Fig. 3C). Sequence analysis confirmed that these PCR products correspond to RYR2. The 303-and 258-bp PCR products are homologous to sequences 14757-15059 and 14802-15059 of rabbit cardiac RYR2 (45). We also observed that the RYR2 F03/R03 primer pair generated a product of the expected size (1179 bp) when amplifying cDNA from rat islets (data not shown). The RYR2 F03/R03 primers target a sequence in mouse and rabbit cardiac RYR2 that encompasses the entire carboxyl terminus starting at position 14805 of rabbit RYR2 and extending into the 3Ј-untranslated region (42,45).
We also sought to determine whether type 1 or type 3 isoforms of RYR are detectable by RT-PCR in rat islets or insulinoma cells. Therefore, RT-PCR was performed using RYR1-or RYR3-specific PCR primer sets (Table I). As an internal control, the same batches of cDNA that yielded RYR2 were used in each PCR reaction. Four primer sets corresponding to RYR1 were tested. Of these, two sets (RYR1-F01/R01; F02/R02) yielded faint PCR products of the appropriate sizes (298 and 385 bp, respectively) when amplifying cDNA from BTC3 cells but not rat islets (data not shown). The 298-bp PCR product was sequenced, whereby it was determined to be identical to that previously reported for mouse RYR1 (42,43). Also tested were four RYR3-specific primer sets. Of these, only one set (RYR3-F01/R01) generated a PCR product of the expected size (344 bp). This abundant PCR product was detected when amplifying cDNA derived from rat islets, BTC3 cells, and mouse brain (data not shown). Sequence analysis determined it not to correspond to any known sequence in GenBank. Instead, it corresponded most closely (93% identity) to mt19d02.rl, an expressed sequence tag (accession no. A178752) derived from a mouse cDNA library.
A Physiological Role for GLP-1 and Ryanodine Receptors in ␤-Cell Ca 2ϩ Signaling-GLP-1 is proposed to influence Ca 2ϩ signaling in ␤-cells by mobilizing ryanodine-sensitive Ca 2ϩ stores (9,20). To test this possibility, measurements of [Ca 2ϩ ] i were obtained from single rat ␤-cells maintained in primary cell culture and equilibrated in saline containing 7.5 mM Dglucose. Application of 10 nM GLP-1-(7-37) produced a significant increase of [Ca 2ϩ ] i in 34 of 50 cells. These responses were tabulated on the basis of kinetic parameters (Table II) Labeling of cells by the fluorophore was quantitated using an argon laser-based imaging system. The maximal intensity of fluorescence detected was assigned an arbitrary value of 1.0. Measurements were normalized to this value. A, cells were exposed to 1 nM BODIPY FL-X ryanodine for 20, 40, 60, or 80 min at 22°C. This led to a time-dependent increase of labeling that decreased following removal of the fluorophore from the extracellular solution (as indicated by Wash). B, high affinity binding of BODIPY FL-X ryanodine to its receptor was diminished by prior treatment of islet cells for 60 min at 22°C with the indicated concentration of non-fluorescent ryanodine. A concentration of 4 nM ryanodine reduced the binding of 1 nM BODIPY FL-X ryanodine by 50%. or a slowly developing and sustained increase lasting several minutes (Fig. 4B). Biphasic responses consisted of an initial transient increase of [Ca 2ϩ ] i followed by a sustained effect (Fig. 4C).
These actions of GLP-1 resemble those effects known to be produced by pharmacological activators of cAMP signaling. The membrane-permeant cAMP agonist 8-Br-cAMP produced a fast transient increase of [Ca 2ϩ ] i similar in time course and magnitude to that produced by GLP-1 itself (Fig. 4A). A similar effect was also obtained during exposure of ␤-cells to 10 M forskolin or 10 M of the cAMP agonist (Sp)-cAMPS (data not shown). As expected for a signal transduction mechanism mediated by cell surface receptors, the action of GLP-1 exhibited desensitization, whereas the effect of 8-Br-cAMP was readily repeatable (Fig. 4A). Voltage clamp analysis using the perforated patch technique demonstrated that a fast transient increase of [Ca 2ϩ ] i could also be measured when GLP-1 was applied to cells in which the membrane potential was maintained at Ϫ50 mV but not Ϫ80 mV (Fig. 4D). Under these conditions, the increase of [Ca 2ϩ ] i was not associated with a detectable change of membrane current ( Fig. 4D; trace labeled I h ). Therefore, the fast transient increase of [Ca 2ϩ ] i was unlikely to result from influx of Ca 2ϩ as a consequence of the activation of nonselective cation channels (14,25) or cyclic nucleotide-gated ion channels.
Detailed pharmacological studies were performed to characterize the signaling system by which GLP-1 stimulates a fast transient increase of [Ca 2ϩ ] i . To achieve this goal, the transient response was studied in isolation using cells for which no sustained increase of [Ca 2ϩ ] i was observed. The action of GLP-1 was characterized with respect to: 1) the requirement for a normoglycemic concentration of extracellular D-glucose, 2) the ligand selectivity of the GLP-1 receptor, 3) the action of (Rp)-cAMPS as a membrane-permeant antagonist of cAMP signaling, 4) the action of nimodipine as a dihydropyridine antagonist of L-type Ca 2ϩ channels, 5) the action of thapsigargin as a blocker of endoplasmic reticular Ca 2ϩ uptake mediated by the SERCA class of Ca 2ϩ ATPases, and 6) the action of ryanodine as an antagonist of CICR mediated by intracellular Ca 2ϩ release channels.
We observed that equilibration of ␤-cells in a normoglycemic concentration (7.5 mM) of glucose allowed for a fast transient The design of the PCR primers was based on the published partial sequences of mouse ryanodine receptors (42,43). Listed are the four sets of forward (F01-F04) and reverse primers (R01-R04) that target the type 1, 2, or 3 ryanodine receptor isoforms. The predicted sizes (in base pairs) of the PCR products are listed on the right.

RT-PCR analysis of RYR2 in rat islets and mouse Be-taTC3 insulinoma cells.
A, RT-PCR amplification of cDNA derived from rat islets or BetaTC3 cells was achieved using RYR2-specific primer pair 4, designated as RYR2 F04/R04 (Table I). A 423-bp PCR product was resolved by gel electrophoresis, as indicated by arrowheads in lanes designated as (ϩ) Rev. Trans. (for reverse transcriptase). No PCR product was detected when amplifying a cDNA synthesis reaction mixture to which no reverse transcriptase was added (lanes designated as (Ϫ) Rev. Trans.). B and C, RT-PCR amplification of cDNA derived from rat islets (B) or BetaTC3 insulinoma cells (C) was accomplished using RYR2-specific primer pairs 1 and 2, designated as RYR F01/R01 and RYR F02/R02 (Table I). The 303-and 258-bp PCR products were resolved by gel electrophoresis and are indicated by arrowheads in lanes designated as (ϩ) Rev. Trans.  Fig. 5A; Table II). In marked contrast, the actions of GLP-1 and 8-Br-cAMP were diminished by equilibration of cells in lower concentrations of glucose (Fig. 5A). Responses observed during exposure of cells to 2.0 or 0.8 mM glucose were measured not as a spike-like increase of [Ca 2ϩ ] i , but as a small sustained effect (data not shown). This diminishment in response size was accompanied by a decrease in the fraction of cells exhibiting any response at all (Fig. 5A). Evidently, these fast transient increases of [Ca 2ϩ ] i are all-or-none phenomena generated in a manner strictly dependent on exposure of ␤-cells to a normoglycemic concentration of glucose. This conclusion is consistent with the established role of glucose metabolism as a positive regulator of Ca 2ϩ uptake, thereby maintaining the filling state of intracellular Ca 2ϩ stores (46,47). It was also established that the stimulatory action of GLP-1 in this assay exhibited pharmacological specificity characteristic of the cloned GLP-1 receptor isolated from a rat islet cDNA library (3,6). GLP-1-(7-37) and GLP-1-(7-36)amide isopeptides were effective, whereas GLP-1-(8 -37) was ineffective (Fig. 5B). Exendin-4, a peptide isolated from Heloderma, and which resembles in structure GLP-1, also exhibited agonist activity (Fig. 5B). In contrast, no effect of GLP-1-(7-37) was observed after treatment of cells with exendin-(9 -39) (Fig. 5B), a peptide antagonist of the GLP-1 receptor. A cAMP signaling system that regulates influx of Ca 2ϩ through L-type Ca 2ϩ channels is likely to be an important determinant of this response. Equilibration of cells for 30 min at 37°C in saline containing the cAMP antagonist (Rp)-cAMPS blocked the fast transient increases of [Ca 2ϩ ] i measured in response to GLP-1 or 8-Br-cAMP (Fig. 5C). The actions of GLP-1 and 8-Br-cAMP were also blocked by brief exposure of cells to a Ca 2ϩ -free solution (Fig. 5C), or by treatment with the L-type Ca 2ϩ channel antagonist nimodipine (Fig. 5C). It was also observed that this signaling system targets an intracellular source of Ca 2ϩ that is mobilized as a consequence of Ca 2ϩ influx. No effect of GLP-1 was observed after treatment of cells with the SERCA Ca 2ϩ ATPase inhibitor thapsigargin (Fig. 5C).
The stimulatory actions of GLP-1 and 8-Br-cAMP that resulted in a fast transient increase of [Ca 2ϩ ] i were blocked by prior equilibration of islet cells for 5 min at 37°C in saline containing 44 mM KCl and 500 nM ryanodine (Fig. 5D). This concentration of KCl produces cellular depolarization and Ca 2ϩ influx, thereby promoting high affinity binding of ryanodine to Ca 2ϩ release channels that open as a consequence of CICR (44). No such antagonist action of 500 nM ryanodine was observed using islet cells not pretreated with 44 mM KCl (Fig. 5D). Pharmacological concentrations of ryanodine (20 -100 M) were also observed to be effective provided that islet cells were incubated for 30 min at 37°C in saline containing 5.6 mM KCl (data not shown). Our use of 44 mM KCl to "unmask" the potent antagonist action of ryanodine is reminiscent of a previous study of RINm5F cells in which ryanodine was reported to be an antagonist of Ca 2ϩ signaling only under conditions in which cells were pretreated with caffeine (48). Such observations may help explain conflicting reports documenting the effectiveness (48), limited effectiveness (49), or lack of effect (35, 50 -53) of ryanodine as a blocker of Ca 2ϩ signaling in ␤-cells.
Caffeine as a Sensitizer of Ca 2ϩ -induced Ca 2ϩ Release Mediated by Ryanodine Receptors-Caffeine is a methylxanthine that acts as a co-agonist with cytosolic Ca 2ϩ to facilitate CICR mediated by ryanodine receptors in cardiac myocytes (54). Whether ryanodine receptors also mediate stimulatory effects of caffeine on [Ca 2ϩ ] i in ␤-cells remains controversial. We examined what effect caffeine exerts on Ca 2ϩ signaling in HIT-T15 insulinoma cells, a hamster ␤-cell line that expresses GLP-1 receptors (16), L-type VDCCs (16,21), and the type 2 isoform of RYR (55). We found that application of caffeine  (Fig. 6A). This response was similar in magnitude and time course to that produced by GLP-1 or 8-Br-cAMP in rat ␤-cells (cf. Figs. 4A and 6A). As might be expected for a process of CICR, the action of caffeine was critically dependent on the initial [Ca 2ϩ ] i . There existed a positive correlation between the [Ca 2ϩ ] i measured prior to and during application of caffeine (Fig. 6B). This correlation appears to explain our finding that the action of caffeine was potentiated by prior exposure of HIT-T15 cells to a depolarizing concentration (44 mM) of KCl that produced a rise of [Ca 2ϩ ] i on its own (Fig. 6A).
It is important to note that the action of caffeine was unrelated to its previously reported inhibitory effect on K-ATP (51). Caffeine produced an increase of [Ca 2ϩ ] i in HIT-T15 cells voltage clamped at Ϫ70 mV (Fig. 6C). This action of caffeine was potentiated by repetitive cellular depolarization to 0 mV (Fig.  6C). Such repetitive stimuli generated influx of Ca 2ϩ through VDCCs, thereby explaining the gradual increase of [Ca 2ϩ ] i observed prior to application of caffeine ( Fig. 6C; arrow with dots). From these observations, it is evident that a close functional coupling existed between influx of Ca 2ϩ and the subsequent mobilization of caffeine-sensitive Ca 2ϩ stores in this cell type. Support for such a mechanism of CICR is also provided by our finding that the interaction of 44 mM KCl and caffeine to produce a fast transient increase of [Ca 2ϩ ] i was attenuated by prior exposure of HIT-T15 cells to ryanodine (Fig. 6D). DISCUSSION The precise mechanism by which cAMP-elevating agents such as GLP-1 regulate Ca 2ϩ signaling and insulin secretion in ␤-cells continues to be a topic of considerable controversy. Here we present findings in support of an important role for the type 2 isoform of ryanodine receptor as a mediator of intracellular Ca 2ϩ signaling activated by GLP-1-(7-37). RT-PCR analyses of rodent islet cells and insulinoma cell lines indicated a preferential expression of RYR2 in comparison to RYR1 or RYR3. These receptors were visualized by fluorescence microscopy in fixed, permeabilized human, mouse, and rat ␤-cells using the BODIPY-FL-X ryanodine fluorophore. Functional studies also demonstrated that application of ryanodine to ␤-cells blocked the fast transient increase of [Ca 2ϩ ] i observed in response to GLP-1-(7-37). Taken together, these findings appear to substantiate an important role for ryanodine receptors as mediators of GLP-1 signal transduction.
Evidence for Expression of Ryanodine Receptors in Pancreatic ␤-Cells-Islam and colleagues have used the RNase protection assay to detect RYR2 mRNA in ob/ob mouse islets and BTC3 cells (56). The riboprobes used in that study were derived by in vitro transcription of RYR cDNAs equivalent to partial sequences of RYRs 1-3 from mouse (42,43). These sequences were used in the present study to identify RYR-specific PCR primers for RT-PCR analyses of rat islets. Our findings complement those of Islam and co-workers, whereby we demon- strate expression of RYR2 mRNA corresponding to a region of the receptor that encompasses the transmembrane spanning domains as well as the carboxyl terminus. Takasawa and colleagues have also used RT-PCR to detect RYR2 mRNA in islets obtained from C57BL/6J mice but not ob/ob mice or RINm5F insulinoma cells (33). It was proposed that the ob/ob model of diabetes/obesity is characterized by a defect of Ca 2ϩ signaling such that levels of RYR2 protein are reduced (33). Should this prove to be the case, it may provide an explanation for the failure of previous studies to detect functional expression of ryanodine receptors in ob/ob mouse islets (35,37,(51)(52)(53) or RINm5F cells (Refs. 34, 36, and 50, but see Ref. 48).
Interpretation of findings obtained by RNase protection or RT-PCR analysis is complicated by uncertainties concerning whether or not there is functional expression of ryanodine receptor protein. Therefore, we tested for expression of ryanodine receptor binding sites by fluorescence microscopy using the BODIPY FL-X ryanodine fluorophore. Specific high affinity binding of BODIPY FL-X ryanodine was observed in human islet, rat islet, and C57BL/6J mouse islet cells. The binding was detected in permeabilized cells, was found to be saturable and reversible, and was blocked by pretreatment of cells with nonfluorescent ryanodine. Not all cells were labeled with BODIPY FL-X ryanodine, but those that did included ␤-cells that stained positively with anti-insulin antiserum. Therefore, BODIPY FL-X ryanodine might serve as a useful fluorescent probe for future studies examining the subcellular distribution of ryanodine receptors in organellar structures of the ␤-cell.
Caffeine as a Probe for Ca 2ϩ Signaling Mediated by Ryanodine Receptors-It is well established that IP 3 receptors play a highly significant role in the process by which hormones and neurotransmitters mobilize Ca 2ϩ stores in the ␤-cell (57-60). It is also known that it is the type 3 isoform of the IP 3 receptor that is expressed at high levels in this cell type (61). Much less well understood is the functional significance of IP 3 -insensitive Ca 2ϩ stores. An IP 3 -insensitive non-mitochondrial source of Ca 2ϩ was observed by Islam and co-workers to be mobilized in a synergistic fashion during treatment of RINm5F cells with thimerosal and caffeine (50). Although thimerosal and caffeine will potentiate CICR from ryanodine-sensitive Ca 2ϩ stores, the mechanism by which caffeine exerts its action in RINm5F cells was called into question. Studies of ob/ob mouse ␤-cells failed to demonstrate a caffeine-sensitive source of intracellular Ca 2ϩ (35,51,52). It was argued instead that caffeine raised [Ca 2ϩ ] i not by stimulatory effects on ryanodine receptors, but by inhibitory actions on K-ATP, whereby cellular depolarization produced influx of Ca 2ϩ via VDCCs (51).
The findings of the present study contradict several of these assertions and provide new evidence for an important role of ryanodine receptors in ␤-cell Ca 2ϩ signaling. We observed GLP-1 and caffeine to be effective under conditions of voltage clamp, producing a fast transient increase of [Ca 2ϩ ] i in a manner independent of K-ATP (Figs. 4D and 6C). As expected for a process of CICR, the actions of ryanodine and caffeine were shown to be facilitated by KCl-induced depolarization (Figs. 5D and 6A) or by stepwise changes of membrane potential (Fig.  6C). A positive correlation was also established linking an increase of basal [Ca 2ϩ ] i to an increased responsiveness of HIT-T15 cells to caffeine (Fig. 6B). Most importantly, the actions of GLP-1 and caffeine were found to be blocked by ryanodine (Figs. 5D and 6D). Such observations provide compelling evidence for a ryanodine-and caffeine-sensitive source of intracellular Ca 2ϩ that is mobilized as a consequence of CICR, a process activated in a direct manner by GLP-1. Notably, these findings differ from those of Islam and co-workers (56), who  4 -6). The trace for [Ca 2ϩ ] i was interrupted for ϳ300 s during which a light was turned on in order to monitor the superfusion system. observed no such direct stimulatory action of cAMP to release Ca 2ϩ stores in mouse ob/ob ␤-cells.
An Interaction of cAMP, L-type VDCCs, and Ryanodine Receptors to Regulate CICR-We report that GLP-1 stimulates a fast transient increase of [Ca 2ϩ ] i under conditions in which ␤-cells are equilibrated in a normoglycemic but not hypoglycemic concentration of glucose. This action of the peptide is mediated by receptors, the pharmacological properties of which match those of the GLP-1 receptor cloned from rat islets. Analyses of the signal transduction mechanism stimulated by GLP-1 demonstrates that its effects on [Ca 2ϩ ] i are mediated by cAMP. The action of GLP-1 is blocked by the cAMP antagonist (Rp)-cAMPS, and is mimicked by the agonist 8-Br-cAMP. Evidence is also presented implicating two distinct types of Ca 2ϩ channels as targets of the GLP-1 signaling system. Treatments that targeted L-type Ca 2ϩ channels (nimodipine) or Ca 2ϩ release channels (ryanodine) render GLP-1 ineffective. Finally, the source of Ca 2ϩ mobilized by GLP-1 is shown to be depleted by thapsigargin, thereby implicating the SERCA class of Ca 2ϩ ATPases in its regulation.
These findings are interpreted in the model illustrated in Fig. 7, which shares important features with previous models (20,56,69). GLP-1 stimulates production of cAMP and activation of PKA, thereby exerting a dual effect. The hormone augments influx of Ca 2ϩ and an increase of [Ca 2ϩ ] i by promoting the opening of L-type Ca 2ϩ channels (step 1). It also sensitizes RYR2 to stimulatory influences of cytosolic Ca 2ϩ , thereby facilitating CICR from a Ca 2ϩ store, the filling state of which is maintained by metabolism of glucose (step 2). The fast transient increase of [Ca 2ϩ ] i that results is measured in cells voltage clamped at Ϫ50 mV but not Ϫ80 mV becuse GLP-1 promotes partial activation of the L-type Ca 2ϩ current, thereby allowing the opening of Ca 2ϩ channels at a membrane potential (Ϫ50 mV) slightly less negative than the resting potential. This influx of Ca 2ϩ is responsible for CICR mediated by RYR2 sensitized to the stimulatory effects of cytosolic Ca 2ϩ . Such a model is consistent with studies demonstrating that PKA facilitates permeation by Ca 2ϩ of L-type Ca 2ϩ channels (62) and ryanodine receptors (63,64).
It should be emphasized that the model presented in Fig. 7 does not rule out a role for IP 3 receptors as mediators of GLP-1 signal transduction. GLP-1 has a small stimulatory effect on inositol phosphate production in rat islets (24,28), and cAMP enhances Ca 2ϩ release from IP 3 -sensitive stores in hepatocytes (65,66). Furthermore, permeation by Ca 2ϩ of the type 1 IP 3 receptor is facilitated by PKA (67), although such an effect has yet to be demonstrated for the type 3 IP 3 receptor expressed in ␤-cells. The model also does not take into account actions of GLP-1 to induce a sustained increase of [Ca 2ϩ ] i , an effect upon which the transient response may or may not be superimposed (cf. Fig. 4, B and C). We propose that such sustained actions of GLP-1 result from its inhibitory effect on ATP-sensitive K ϩ channels (13) and/or its stimulatory action on nonselective cation channels (14,25).
Conclusion-It is now established that influx of Ca 2ϩ through L-type Ca 2ϩ channels produces an increase of [Ca 2ϩ ] i that serves as a powerful trigger for secretion of insulin. Exocytosis occurs in localized patches of ␤-cell plasma membrane where secretory granules co-localize with L-type Ca 2ϩ channels (68). Influx of Ca 2ϩ through these channels triggers exocytosis by producing a fast transient increase of [Ca 2ϩ ] i that is spatially restricted (68). The efficacy of Ca 2ϩ as a trigger for exocytosis is limited by processes (Ca 2ϩ buffering, extrusion, re-uptake) that maintain [Ca 2ϩ ] i at a level subthreshold for secretion. By facilitating Ca 2ϩ influx and CICR, GLP-1 may allow for a localized and explosive increase of [Ca 2ϩ ] i that triggers secretion in a highly efficient manner. Therefore, the functional significance of CICR initiated by GLP-1 in the ␤-cell may be that it serves to amplify any increase of [Ca 2ϩ ] i arising from entry of Ca 2ϩ via the L-type Ca 2ϩ channels. In this manner, GLP-1 might stimulate two interrelated and closely coupled processes (Ca 2ϩ influx and Ca 2ϩ mobilization) that determine whether or not an increase of [Ca 2ϩ ] i is sufficient to reach threshold for initiation of exocytosis.