cAMP-regulated guanine nucleotide exchange factor II (Epac2) mediates Ca²⁺-induced Ca²⁺ release in INS-1 pancreatic β-cells

Abstract

1. The signal transduction pathway responsible for cAMP-dependent Ca²⁺-induced Ca²⁺ release (CICR) from endoplasmic reticulum Ca²⁺ stores was assessed in the insulin-secreting cell line INS-1.

2. CICR was triggered by the GLP-1 receptor agonist exendin-4, an effect mimicked by caffeine, Sp-cAMPS or forskolin. CICR required influx of Ca²⁺ through L-type voltage-dependent Ca²⁺ channels, and was blocked by treatment with nimodipine, thapsigargin, or ryanodine, but not by the IP₃ receptor antagonist xestospongin C.

3. Treatment with the cAMP antagonist 8-Br-Rp-cAMPS blocked CICR in response to exendin-4, whereas the PKA inhibitor H-89 was ineffective when tested at a concentration demonstrated to inhibit PKA-dependent gene expression.

4. RT-PCR of INS-1 cells demonstrated expression of mRNA coding for the type-II isoform of cAMP-regulated guanine nucleotide exchange factor (cAMP-GEF-II, Epac2).

5. CICR in response to forskolin was blocked by transient transfection and expression of a dominant negative mutant isoform of cAMP-GEF-II in which inactivating mutations were introduced into the exchange factor's two cAMP-binding domains.

6. It is concluded that CICR in INS-1 cells results from GLP-1 receptor-mediated sensitization of the intracellular Ca²⁺ release mechanism, a signal transduction pathway independent of PKA, but which requires cAMP-GEF-II.

The mobilization of Ca²⁺ from endoplasmic reticulum (ER) Ca²⁺ stores is now recognized as playing a highly significant role in the regulation of multiple cellular functions. Gating of inositol trisphosphate receptor (IP₃-R) or ryanodine receptor (RYR) intracellular Ca²⁺ release channels produces an increase of [Ca²⁺]_i that serves as a stimulus for induction of gene expression, mitochondrial metabolism, excitation-contraction coupling, and hormone or neurotransmitter secretion (Berridge, 1998). Ca²⁺ is mobilized from IP₃-R- or RYR-regulated Ca²⁺ stores by a process referred to as Ca²⁺-induced Ca²⁺ release (CICR; Bootman & Berridge, 1995; Berridge, 1996, 1997). CICR results from the stimulatory action of Ca²⁺ at Ca²⁺ release channels, thereby initiating a regenerative Ca²⁺ wave propagating as a transient, and in some cases oscillatory increase of [Ca²⁺]_i. Although the mobilization of Ca²⁺ from IP₃-R-regulated Ca²⁺ stores is recognized as being of general importance for both excitable and non-excitable cells, the functional significance of CICR originating from RYR-regulated Ca²⁺ stores is only now becoming more fully understood.

The physiological importance of RYR-regulated CICR was first established in studies of cardiac myocyte excitation-contraction coupling where influx of Ca²⁺ through L-type voltage-dependent Ca²⁺ channels (VDCCs) was demonstrated to stimulate the opening of RYR Ca²⁺ release channels (Fabiato, 1983; Nabauer et al. 1989; Niggli, 1999). More recently, it has been suggested that endocrine cells and neurons also utilize RYR-regulated CICR as a means of mobilizing Ca²⁺ from Ca²⁺ stores (Islam et al. 1992, 1998; Leech et al. 1994; Gromada et al. 1995; Holz et al. 1999; Usachev & Thayer, 1999). In some cell types, this can lead to a stimulation of exocytosis (Cheek et al. 1993; von Ruden & Neher, 1993; Tse et al. 1993, 1997; Guo et al. 1996; Narita et al. 1998, 2000).

To explore what role RYR-dependent CICR might play as a stimulus for insulin secretion, we initiated studies utilizing the insulinotropic hormone glucagon-like peptide-1 (GLP-1; Drucker, 1998; Kieffer & Habener, 1999). GLP-1 promotes pancreatic β-cell depolarization by inhibiting ATP-sensitive K⁺ (K-ATP) channels (Holz & Habener, 1992; Holz et al. 1993), and it also facilitates insulin secretion via a more direct effect at the exocytotic secretory apparatus (Gromada et al. 1998b,c). An unusual aspect of GLP-1 signal transduction is its ability to mobilize β-cell Ca²⁺ stores in a cAMP-dependent manner (Gromada et al. 1995; Holz et al. 1999). The relevant Ca²⁺ release channels targeted by GLP-1 are most likely to correspond to RYR-2, the type-2 isoform of ryanodine receptor (Islam et al. 1998; Holz et al. 1999), and all available evidence indicates that GLP-1 sensitizes RYR-2 to stimulatory effects of cytosolic Ca²⁺, thereby increasing the likelihood that such channels will open in response to Ca²⁺ (Gromada et al. 1995; Holz et al. 1999).

Exactly which signal transduction pathway explains how GLP-1 triggers CICR has remained uncertain. GLP-1 stimulates cAMP synthesis (Thorens, 1992; Drucker, 1998), but has only a weak stimulatory effect on IP₃ production in β-cells (Zawalich et al. 1993; Zawalich & Zawalich, 1996). However, GLP-1 does stimulate IP₃ production in COS cells and Xenopus oocytes expressing recombinant GLP-1 receptors (Wheeler et al. 1993; Gromada et al. 1998a). GLP-1 also activates protein kinase A (PKA; Lester et al. 1997), Ca²⁺-calmodulin-dependent protein kinase (Susini et al. 2000), phosphatidylinositol 3-kinase (PI-3K; Buteau et al. 1999) and mitogen-activated protein kinases (MAPK; Frodin et al. 1995; Montrose-Rafizadeh et al. 1999).

The cAMP-dependent action of GLP-1 may not be limited to its effects mediated by PKA. A newly discovered family of cAMP-binding proteins has been characterized, molecules referred to as cAMP-regulated guanine nucleotide exchange factors (cAMP-GEFs, also known as Epac; Kawasaki et al. 1998; de Rooij et al. 1998; Leech et al. 2000). The cAMP-GEFs interact with the Ras-related small G protein Rap1 to initiate a cascade of protein kinase-mediated phosphorylation reactions catalysed by B-Raf and MAPK. Furthermore, the interaction of cAMP-GEFs with the Rab3 small G protein effector Rim is reported to stimulate exocytosis of insulin in β-cells (Ozaki et al. 2000). Therefore, there is reason to believe that activation of cAMP-GEFs might also explain, at least in part, how CICR in β-cells is triggered in response to agents (forskolin, 8-Br-cAMP, Sp-cAMPS) that produce an increase of [cAMP]_i analogous to that produced by GLP-1.

Here we examined what role cAMP-GEFs may play as mediators of CICR in the insulin-secreting cell line INS-1. The INS-1 cells are a radiation-induced rat insulinoma that respond to glucose and GLP-1 (Asfari et al. 1992; Skoglund et al. 2000), and were previously characterized as a model system in which to investigate RYR-mediated CICR (Gamberucci et al. 1999). Measurements of [Ca²⁺]_i were obtained under conditions in which wild-type or mutant isoforms of cAMP-GEFs were introduced by transient transfection. We report that RYR-regulated CICR is initiated, at least in part, by a PKA-independent signalling mechanism that most probably involves cAMP-GEF-II. Some of these finding have been reported in preliminary form (Kang & Holz, 2000, 2001).

METHODS

Cell culture

INS-1 cells (passages 70–90) were obtained from Dr Maryam Asfari (Asfari et al. 1992). Cell cultures were maintained in a humidified incubator (95 % air-5 % CO₂) at 37 °C in RPMI 1640 culture medium containing 10 mm Hepes, 11.1 mm glucose, 10 % fetal bovine serum (FBS), 100 U ml⁻¹ penicillin G, 100 μg ml⁻¹ streptomycin, 2.0 mml-glutamine, 1.0 mm sodium pyruvate and 50 μm 2-mercaptoethanol. INS-1 cells were passaged by trypsinization and subcultured once a week. Cell suspensions were plated on glass coverslips (25CIR-1; Fisher Scientific, Chicago, IL, USA) coated with 1 mg ml⁻¹ concanavalin A (type V; Sigma Chemical Co.), which facilitates adherence of cells to glass surfaces. All reagents for cell culture were obtained from Life Technologies (Rockville, MD, USA), and all experiments illustrated were performed in triplicate after a 16–36 h equilibration in culture medium in order to assure reproducibility.

Transfection of INS-1 cells for overexpression of wild-type and mutant cAMP-GEF-II

Wild-type cAMP-GEF-II (GenBank accession number AB021132) expression vector pSRα-cAMP-GEFII and mutant cAMP-GEF-II expression vector pSRα-cAMP-GEFII (G114E/G422D) were provided by Dr Susumu Seino (Chiba University, Japan; Ozaki et al. 2000). The mutant cAMP-GEF-II acts in a dominant negative manner because it contains inactivating G114E and G422D amino acid substitutions in the A and B cAMP-binding domains of the exchange factor, respectively. INS-1 cell cultures grown to 80 % confluence in 35 mm tissue culture dishes were subjected to transient transfection using 1 μg cAMP-GEF-II expression vector reconstituted in Lipofectamine Plus reagent (Life Technologies). At 24 h post-transfection, the cells were lifted by trypsinization, plated onto glass coverslips, and maintained in normal INS-1 cell culture medium. Experiments were then performed 48 h post-transfection. Transfection efficiency was estimated to be 10–20 %, as determined by use of an enhanced yellow fluorescent protein (EYFP) expression vector.

RIP2-EYFP for fluorescence-based selection of transfected cells in culture

In order to positively identify transfected INS-1 cells, the transfection cocktail containing 1 μg of cAMP-GEF-II expression vector was supplemented with 100 ng of a plasmid in which expression of EYFP was placed under the control of the rat insulin gene II promoter (RIP2). Expression of EYFP was achieved by replacement of a CMV promoter within the pEYFP-N1 expression vector (Clonetech, Palo Alto, CA, USA) for a −700 bp BamHI fragment of RIP2. EYFP fluorescence was monitored in living cells 2 days post-transfection using 513 nm excitation and 527 nm emission filters (Fig. 1A). Once an EYFP-positive cell was identified as having been transfected, the filter set was manually switched to a fura-2 filter set allowing ratiometric determinations of [Ca²⁺]_i (Fig. 1A; see below). The spectral properties of EYFP do not interfere with fura-2 based measurements of [Ca²⁺]_i (Fig. 1B; Kang & Holz, 2000), and less than 1 % crossover exists between fura-2 and EYFP when using filter sets selective for each probe (G. G. Holz, unpublished observations). Control experiments demonstrated that overexpression of EYFP in INS-1 cells does not interfere with KCl- or glucose-dependent Ca²⁺ signalling, nor does it interfere with glucose-dependent insulin gene promoter activity (G. G. Holz, unpublished observations).

Figure 1. Fluorescence microscopy allows co-detection of EYFP and fura-2 in single living INS-1 cells.

A, optical components of the imaging system. RIP2-EYFP was introduced into INS-1 cells by transient transfection, and cells were loaded with fura-2 by incubation in fura-2 AM-containing SES. For detection of EYFP, a xenon arc lamp delivered excitation light (Ex.) filtered at 513 nm and reflected by a 515 nm long pass (LP) dichroic mirror (DM). EYFP epifluorescence passed through a 527 nm emission filter (Em.) for detection by an intensified CCD camera interfaced with a digital video imaging system. For detection of fura-2, a rotating chopper mirror delivered excitation light of 340 or 380 nm to a second filter set in which was housed a 400 nm long pass dichroic mirror and a 510 nm emission filter. B, excitation spectra for EYFP (bold line) and fura-2 at two different concentrations of free Ca²⁺ (100 or 1350 nm). Little or no excitation of EYFP occurs in response to the 340 or 380 nm excitation light used for fura-2, and conversely, little or no excitation of fura-2 occurs in response to light delivered at the excitation maximum (513 nm) for EYFP.

Measurement of [Ca²⁺]_i

The fura-2 loading solution consisted of standard extracellular saline (SES) containing (mm): 138 NaCl, 5.6 KCl, 2.6 CaCl₂, 1.2 MgCl₂, 10 Hepes and 5.6 d-glucose. The pH was adjusted to 7.35 with NaOH and the osmolarity was adjusted to 295 mosmol l⁻¹ using H₂O. The SES was supplemented with 1 μm fura-2 acetoxymethyl ester (fura-2 AM; Molecular Probes Inc., Eugene, OR, USA), 2 % FBS and 0.02 % Pluronic F-127 (w/v; Molecular Probes Inc.). Cells were exposed to the fura-2 AM-containing solution for 20–30 min at 22 °C. The loading solution was then removed, and cells were equilibrated in fresh SES for 10 min at 22 °C. Under these conditions, we previously demonstrated that 9 % of fura-2 is compartmentalized, the remaining 91 % being restricted to the cytosol (Leech et al. 1994). Experiments were performed at 32 °C using an Eclipse TE300 inverted microscope (Nikon, Melville, NY, USA) outfitted with a temperature-controlled stage (Medical Systems Corp., Greenvale, NY, USA), a superfusion system, and a × 100 Nikon UVF oil immersion objective (NA = 1.3). Dual excitation wavelength microspectrofluorimetry was performed ratiometrically at 0.5 s intervals using a digital video imaging system outfitted with an intensified CCD camera (IonOptix Corp., Milton, MA, USA). A rotating chopper mirror delivered excitation light at 340 or 380 nm, and emitted light was measured at 510 nm (Fig. 1A). The average of five video frames of imaging data was used to calculate numerator and denominator values for determination of 340 nm/380 nm ratio values. The [Ca²⁺]_i was calculated according to established methods (Grynkiewicz et al. 1985) as:

(1)

where R is the experimentally derived fluorescence ratio, and R_min and R_max are ratios measured using calibration solutions containing no Ca²⁺ or saturating Ca²⁺, respectively. K_d is the dissociation constant (224 nm) describing the interaction of fura-2 and Ca²⁺, and β is the fluorescence ratio of free/bound fura-2 measured with 380 nm excitation. Calibration of the raw fluorescence values was performed using fura-2 pentapotassium salt dissolved in calibration buffers from Molecular Probes Inc. (Calcium Calibration Kit 1 with Mg²⁺). Values of R_min and R_max were 0.20 and 7.70 for the experiments described.

Sources of reagents and preparation of test solutions

GLP-1-(7–37), GLP-1-(7–36)-amide, exendin-4, exendin-(9–39), caffeine, thapsigargin, nimodipine, forskolin, H-89 and ryanodine were from Sigma. 8-Br-Rp-cAMPS and Sp-cAMPS were from BioLog Life Sciences Institute (Bremen, Germany). Xestospongin C was from Calbiochem (La Jolla, CA, USA). Test solutions were dissolved in SES prior to each experiment. For studies examining effects of peptides, the SES also contained 0.05 % human serum albumin (HSA, fraction V; Sigma) to protect against their binding to glass or plastic surfaces. No effect of HSA itself was observed. Test solutions were applied to individual cells from ‘puffer’ micropipettes using a PicoSpritzer II pressure ejection system (General Valve Corp., NJ, USA; Holz et al. 1993).

pCRE-Luc reporter gene assay

The pCRE-Luc plasmid (Stratagene, La Jolla, CA, USA) introduced into transformed JM109 cells was prepared using the Wizard Midi Prep procedure (Promega Inc., Madison, WI, USA). INS-1 cells were harvested as a single cell suspension, and Lipofectamine Plus containing 1 μg pCRE-Luc was added according to the manufacturer's protocol (Life Technologies). Cells were then pipetted onto 96-well culture plates (Costar, Fisher Scientific). Two days post-transfection, cells were lysed and assayed at 37 °C for luciferase activity using a Luciferase Assay Kit (Tropix, Bedford, MA, USA) in conjunction with a plate reading luminometer allowing automated injection of ATP and luciferin solutions (Model TR-717, Applied Biosystems, Foster City, CA, USA). Experiments were carried out in triplicate, and statistical analysis was performed using the ANOVA test combined with Fisher's PLSD test.

Analysis of cAMP-GEF mRNA expression by RT-PCR

Total RNA was prepared from INS-1 cells by the TRIzol method (Life Technologies), and cDNA synthesis was catalysed by AMV reverse transcriptase using oligo d(T) primers (Reverse Transcription System, Promega). A cDNA synthesis reaction to which no reverse transcriptase was added served as a negative control. The design of PCR primers was based on the published sequence for rat cAMP-GEF-II (Kawasaki et al. 1998; accession number U78517). Optimal PCR primers were selected using a computer-based analysis program (Lasergene 5.0, DNASTAR Inc., Madison, WI, USA). The cDNA was amplified using Taq DNA polymerase (TaKaRa Biologicals, Shiga, Japan). The thermal cycling parameters consisted of an initial denaturation step for 5 min at 95 °C, followed by 20 cycles consisting of: 30 s at 95 °C, 60 s at 57 °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 were extracted from agarose using Quiaex 2 (Quiagen, Inc., Valencia, CA, USA), and ligated into the pCR4TOPO vector (Invitrogen, San Diego, CA, USA) for transformation of OneShot cells (Invitrogen). Transformants were isolated by antibiotic resistance selection, and midiprep plasmid DNA was prepared (Wizard DNA Purification Kit, Promega) and sequenced for verification of PCR product identity.

RESULTS

Exendin-4 stimulates an increase of [Ca²⁺]_i in INS-1 cells

INS-1 cells loaded with fura-2 responded to the GLP-1 receptor agonist exendin-4 by exhibiting a transient increase of [Ca²⁺]_i (Fig. 2A). The response to exendin-4 exhibited an inflection on the rising phase of the Ca²⁺ spike, indicative of an initial increase of [Ca²⁺]_i that triggered a regenerative response after a short delay. This action of exendin-4 was quantified by measuring the response amplitude and duration at the 50 % and 10 % cut-off values, respectively (Fig. 2B). The mean response amplitude was 1100 ± 150 nm, and the mean response duration was 14 ± 3 s. The action of exendin-4 reported here is analogous to the previously reported stimulatory effects of GLP-1 on [Ca²⁺]_i in rat pancreatic β-cells (Holz et al. 1999). It is mediated by GLP-1 receptors because the action of 1 nm exendin-4 was reproduced by 1 nm of the GLP-1 receptor agonist GLP-1-(7–36)-amide, but blocked by 1 μm of the receptor antagonist exendin-(9–39) (Fig. 2C).

Figure 2. Elevation of [Ca²⁺]_i by GLP-1 receptor agonist exendin-4 in INS-1 cells.

A, local application of 10 nm exendin-4 (Ex-4) produced a transient increase of [Ca²⁺]_i. B, the response to exendin-4 was defined as a transient increase of [Ca²⁺]_i, the duration of which did not exceed 30 s when measured at the 10 % response amplitude cut-off, and the amplitude of which exceeded 200 nm when measured at the 50 % cut-off. C, population study demonstrating that CICR in response to 1 nm exendin-4 or 1 nm GLP-1-(7–36)-amide was blocked by prior exposure of INS-1 cells to 1 μm of the GLP-1 receptor antagonist exendin-(9–39). Experiments were performed using INS-1 cells equilibrated in saline containing 11.1 mm glucose, here and in subsequent figures. Arrows in A and B indicate a 30 s application of exendin-4 using a puffer pipette.

Pharmacological properties of CICR triggered by exendin-4

To validate that the fast transient increase of [Ca²⁺]_i observed in response to exendin-4 did in fact result from CICR, we performed pharmacological studies of Ca²⁺ signalling in INS-1 cells. Dose-response analysis (Fig. 3A) demonstrated that exendin-4 stimulated CICR over a concentration range of 0.1–10 nm of the peptide (EC₅₀ 0.8 nm). The action of exendin-4 to stimulate CICR was reproduced by caffeine (Fig. 3A, see also Fig. 4A), as well as by the cAMP-elevating agent forskolin (Fig. 3A and Fig. 4B), or by the cAMP agonist Sp-cAMPS (Fig. 3A). In agreement with the previously reported action of GLP-1 in stimulating CICR in rat β-cells (Holz et al. 1999), the actions of exendin-4 and forskolin in this assay were blocked by pretreatment of INS-1 cells with the L-type VDCC antagonist nimodipine (1 μm), or by pretreatment with the sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitor thapsigargin (1 μm; Fig. 3B and C). Of particular interest was the observation that there existed a synergistic interaction of exendin-4 and caffeine to trigger CICR (Fig. 3D). This interaction was eliminated by a 30 min pretreatment of INS-1 cells with a solution containing 5 mm caffeine and 1 μm ryanodine (Fig. 3D). The stimulation of CICR by caffeine was not simply a secondary consequence of this methylxanthine's known inhibitory effect on cAMP phosphodiesterase. The action of caffeine was not blocked by pretreatment with the cAMP antagonist 8-Br-Rp-cAMPS (Fig. 5A).

Figure 3. Pharmacological properties of Ca²⁺ signalling in INS-1 cells.

A, dose-response analysis illustrating the concentration range over which exendin-4, forskolin, Sp-cAMPS, and caffeine stimulated CICR. Such ‘all or nothing’ responses were quantified by determining the percentage of cells exhibiting a transient increase of [Ca²⁺]_i, as defined in Fig. 2B. B and C, CICR initiated by exendin-4 or forskolin was suppressed by nimodipine (1 μm) and by thapsigargin (1 μm). D, CICR in response to caffeine was potentiated by a low concentration of exendin-4 (500 pm). The synergistic interaction of caffeine and exendin-4 was blocked by pretreatment of cells for 30 min in a solution containing 5 mm caffeine and 1 μm ryanodine.

Figure 4. Pharmacological stimulation of CICR by caffeine and forskolin in INS-1 cells.

A, caffeine was applied twice at a concentration of 10 mm. B, forskolin was applied once at a concentration of 10 μm. Note that prior to administration of forskolin, this cell exhibited spontaneous oscillation of [Ca²⁺]_i. Arrows in A and B indicate a 10 s application of each test substance.

Figure 5. Analysis of the effects of 8-Br-Rp-cAMPS and xestospongin C.

A, CICR in response to a 10 s application of 10 mm caffeine (arrows) was not blocked by pretreatment with 200 μm of the cAMP antagonist 8-Br-Rp-cAMPS. B, CICR in response to a 10 s application of 10 μm forskolin (arrows) was not blocked by pretreatment with 20 μm of the IP₃ receptor antagonist xestospongin C.

A role for cAMP in support of CICR triggered by exendin-4

It has been suggested that cAMP-regulated, PKA-dependent phosphorylation sensitizes RYR-2 to stimulatory effects of cytosolic Ca²⁺, thereby facilitating CICR (Gromada et al. 1995; Holz et al. 1999). Consistent with this concept, CICR in response to the cAMP-elevating agent forskolin was unaffected by the membrane-permeant IP₃ receptor antagonist xestospongin C (Fig. 5B). As expected, pretreatment of INS-1 cells with the cAMP antagonist 8-Br-Rp-cAMPS (200 μm) inhibited the action of exendin-4 in this assay (Fig. 6A). To investigate whether exendin-4 exerted its stimulatory action via PKA, a well-characterized PKA antagonist (H-89) was tested. We found that the action of exendin-4 was not blocked by pretreatment with 1 μm H-89 (Fig. 6A), a concentration that allows selective inhibition of PKA (Hidaka & Kobayashi, 1992). A 10-fold higher concentration of H-89 reduced the action of exendin-4 (Fig. 6A), but also produced a generalized suppression of spontaneous Ca²⁺ oscillations (data not shown). This effect is most likely to be due to non-selective inhibition of a broad spectrum of protein kinases (Hidaka & Kobayashi, 1992; Bode et al. 1999).

Figure 6. Antagonist properties of 8-Br-Rp-cAMPS and H-89 assessed in INS-1 cells.

A, the cyclic AMP antagonist 8-Br-Rp-cAMPS (200 μm) inhibited CICR measured in response to a 30 s application of exendin-4 (10 nm). The PKA inhibitor H-89 did not block CICR in response to exendin-4 when tested at a concentration of 1 μm, but was an effective blocker when tested at 10 μm. B, CRE-dependent expression of luciferase in transfected INS-1 cells was stimulated by a 4 h exposure to forskolin (2 μm), and was partially inhibited by 8-Br-Rp-cAMPS (200 μm). Note that the action of forskolin was inhibited by a low (1 μm) as well as a high (10 μm) concentration of H-89. *P < 0.005, **P < 0.001, relative to Control.

The efficacy and specificity of 8-Br-Rp-cAMPS and H-89 were assessed in a luciferase-based reporter assay considered to be diagnostic for cAMP and PKA signalling (Skoglund et al. 2000). INS-1 cells were transfected with CRE-Luc, a plasmid in which expression of luciferase is directed by a promoter containing multimerized synthetic cAMP response elements (CREs). Cells were then assayed for luciferase activity following a 4 h exposure to 2 μm forskolin. Under these conditions, CRE-Luc activity was stimulated by forskolin (Fig. 6B), an effect inhibited by treatment with 8-Br-Rp-cAMPS (200 μm) or H-89 (1 μm). Therefore, 8-Br-Rp-cAMPS and H-89 are effective antagonists of cAMP and PKA signalling in INS-1 cells when used at low concentrations, and when tested within the appropriate experimental context. Taken together, this analysis indicates that CICR in response to exendin-4 results from activation of a cAMP signalling pathway that is relatively insensitive to H-89, and which might be, at least in part, independent of PKA. Consistent with this concept, RT-PCR analysis of INS-1 cells demonstrated expression of mRNA coding for rat cAMP-GEF-II (Table 1). Nucleic acid sequence analysis of these PCR products demonstrated them to be 100 % identical to cAMP-GEF-II (data not shown).

Table 1.

RT-PCR of INS-1 cell mRNA for cAMP-GEF-II (GenBank U78517)

Primer pair	Nucleic acid sequence	Sequence amplified	Predicted size (bp)	Detected in INS-1?
Forward no. 1	5′ CAGAACGGTGCGCTACTACAGGA 3′	716–738	408	Yes
Reverse no. 1	5′ TGCCAACCGTGAAGAAGATGACTA 3′	1100–1123	—	—
Forward no. 2	5′ GTGGGGACGTTTGAACTGATGAGC 3′	135–158	440	Yes
Reverse no. 2	5′ AGCCTGTACGCCTTGTGATTTCTG 3′	551–574	—	—
Forward no. 3	5′ GCCAGAACGGTGCGCTACTACA 3′	714–735	595	Yes
Reverse no. 3	5′ ACCCTCCCCAGAACCCAGACC 3′	1288–1308	—	—

RT-PCR was performed using 3 different PCR primer pairs complementary to the 3′ terminus of the rat cAMP-GEF-II coding sequence, and extending into the 3′-untranslated region.

Downregulation of CICR by expression of dominant negative cAMP-GEF-II

We next sought to more directly demonstrate a role for cAMP-GEF-II as a mediator of CICR by use of a molecular biological approach employing wild-type or dominant negative isoforms of the exchange factor. The dominant negative cAMP-GEF-II is a mutant isoform in which inactivating point mutations have been introduced into the two cAMP-binding domains (Ozaki et al. 2000). The cAMP-GEF-II cDNA was introduced by transient transfection in conjunction with use of RIP2-EYFP as a marker allowing positive identification of transfected INS-1 cells (Methods, Fig. 1). A population study was then conducted in order to assess the extent to which cAMP-GEF-II mediates CICR in response to forskolin. Cells not transfected exhibited the typical transient increase of [Ca²⁺]_i when challenged with 10 μm forskolin (Fig. 7A), and this response matched closely the pattern of CICR observed in cells transfected with wild-type cAMP-GEF-II (Fig. 7B). However, when cells were transfected with dominant negative cAMP-GEF-II, the action of forskolin was largely abrogated, and only a small, sustained increase of [Ca²⁺]_i was sometimes observed (Fig. 7C). These findings are summarized in the bar graph presented in Fig. 7D.

Figure 7. Inhibition of forskolin-stimulated CICR by dominant negative cAMP-GEF-II.

Single cell suspensions of INS-1 cells were treated with Lipofectamine Plus supplemented with RIP2-EYFP and wild-type or dominant negative cAMP-GEF-II expression vectors. Cells were then plated on glass coverslips for loading with fura-2 and measurement of [Ca²⁺]_i. A, untransfected cells not exhibiting EYFP fluorescence responded to application of forskolin by generating CICR. B, CICR in response to forskolin was also observed in transfected cells expressing EYFP and wild-type (WT) cAMP-GEF-II. C, CICR in response to forskolin was markedly inhibited in transfected cells expressing EYFP and mutant (MT) dominant negative cAMP-GEF-II. A residual response to forskolin was sometimes observed in these cells. It consisted of a small sustained increase of [Ca²⁺]_i. D, population study demonstrating antagonism of forskolin-induced CICR by dominant negative cAMP-GEF-II. *P ≤ 0.001, Student's t test. Arrows in A–C indicate application of 10 μm forskolin for 10 s.

DISCUSSION

A role for cAMP-GEF-II as a mediator of CICR

CICR in the INS-1 insulin-secreting cell line is demonstrated here to be triggered by a GLP-1 receptor agonist (exendin-4), by pharmacological agents that stimulate cAMP signalling (forskolin, Sp-cAMPS), and by a methylxanthine known to target Ca²⁺ release channels (caffeine). The source of Ca²⁺ mobilized by CICR resides in a thapsigargin- and ryanodine-sensitive compartment, consistent with the proposed role of RYR-2 as a downstream effector of GLP-1 signal transduction. A possible role for IP₃ receptors as mediators of this response appears unlikely given that CICR was not blocked by the IP₃ receptor antagonist xestospongin C. Instead, available evidence indicates that exendin-4 promotes CICR by sensitizing the RYR Ca²⁺ release mechanism to stimulatory effects of cytosolic Ca²⁺. Sensitization may result, at least in part, from activation of a novel cAMP signalling mechanism since CICR is shown here to be inhibited by the cAMP antagonist 8-Br-Rp-cAMPS, but is relatively insensitive to the PKA inhibitor H-89. INS-1 cells are also shown to express mRNA coding for cAMP-GEF-II, and overexpression of a dominant negative isoform of cAMP-GEF-II leads to a suppression of forskolin-induced CICR. Taken together, such findings provide evidence in support of an emerging model of Ca²⁺ signalling whereby cAMP-GEFs such as cAMP-GEF-II mediate the cAMP-dependent mobilization of Ca²⁺ from endoplasmic reticulum Ca²⁺ stores.

Ca²⁺ release channels responsible for CICR correspond to RYR-2

Multiple competing hypotheses exist concerning the origins of cAMP-dependent CICR in INS-1 cells, as well as in authentic pancreatic β-cells. It was proposed that the relevant Ca²⁺ release channels targeted by cAMP correspond to the IP₃-R (Liu et al. 1996), whereas evidence for a major role of RYR-2 was also provided (Islam et al. 1992, 1998). It should be noted that both types of ER Ca²⁺ release channels are sensitized by cAMP and PKA under appropriate experimental conditions (Burgess et al. 1991; Hajnoczky et al. 1993; Nakade et al. 1994; Hain et al. 1995; Ozawa, 1998; Marx et al. 2000). To complicate this picture further, it has been proposed that cAMP-elevating agents such as GLP-1 stimulate an increase of [Ca²⁺]_i by releasing Ca²⁺ stored in insulin-containing secretory granules (Nakagaki et al. 2000).

Although all of these possible mechanisms of Ca²⁺ signalling may play some role in GLP-1 signal transduction, the weight of current evidence favours a dominant role for RYR-2 in support of CICR. This conclusion is advanced on the basis of previous studies of insulin-secreting cells where it was demonstrated that CICR originating from a non-mitochondrial source of Ca²⁺ was blocked by ryanodine, but stimulated by caffeine, thimerosal, and 4-chloro-3-ethylphenol (Islam et al. 1992, 1998; Leech et al. 1994; Gromada et al. 1995; Willmott et al. 1995a; Chen et al. 1996; Gamberucci et al. 1999; Holz et al. 1999; Maechler et al. 1999). The action of caffeine is likely to be specific for ryanodine receptors, and is not secondary to this methylxanthine's inhibitory effect on phosphodiesterase, since the action of caffeine was not blocked by the cAMP antagonist 8-Br-Rp-cAMPS. It should also be noted that expression of RYR-2 was previously confirmed in INS-1 cells by immunoblot analysis (Gamberucci et al. 1999).

Despite such convincing evidence, the importance of RYR-2 as a mediator of CICR in β-cells has been debated, particularly from the standpoint of the purported stimulatory action of cyclic ADP-ribose (cADP-ribose) at the ryanodine receptor (Takasawa et al. 1998). Studies of permeabilized β-cells demonstrated no effect of ryanodine or agents (caffeine, cADP-ribose) known to interact with RYR-2 (Rutter et al. 1994; Willmott et al. 1995b; Webb et al. 1996; Tengholm et al. 1998, 2000). Such negative findings are possibly related to the labile nature of CICR when studied in permeabilized islet preparations. In contrast, studies of insulinoma cells demonstrated ryanodine to be an inhibitor of caffeine-induced CICR under conditions in which cells were pretreated with caffeine (Chen et al. 1996). This pharmacological manipulation recapitulates the ‘use-dependent’ antagonist action of ryanodine because it increases the affinity of RYR-2 for ryanodine by gating the channel from its closed to open, and subsequently blocked state (Bianchi, 1997). In this manner, the sensitivity of CICR to ryanodine is revealed. Using this strategy, we demonstrate that pretreatment with caffeine and ryanodine antagonizes CICR in INS-1 cells.

Conclusion

A role for cAMP in support of CICR might be expected if the filling state of ER Ca²⁺ stores is maintained by PKA-dependent sequestration of Ca²⁺ (Yaekura & Yada, 1998). It has also been proposed that sensitization of β-cell CICR results from PKA-mediated phosphorylation of RYR-2, as is known to occur in cardiac myocytes (Marx et al. 2000). We find that stimulators of cAMP signalling evoke CICR, whereas the cAMP antagonist 8-Br-Rp-cAMPS suppresses it. Furthermore, we demonstrate that a high concentration of H-89 produces a generalized suppression of Ca²⁺ signalling. However, when H-89 is tested at a lower concentration, one selective for PKA, we find it to be ineffective. Control experiments using cells transfected with CRE-Luc demonstrate further that the PKA-dependent action of forskolin at the cyclic AMP response element is blocked by a low concentration of H-89. It can be concluded, therefore, that high concentrations of H-89 suppress CICR via non-specific effects, rather than by selective inhibition of PKA. Such findings are reminiscent of the prior study of Bode and co-workers in which it was proposed that GLP-1 mobilizes ER Ca²⁺ stores in a PKA-independent manner (Bode et al. 1999). Taken together, our new findings lead us to propose that the cAMP-dependent stimulation of CICR might result from an alternative cAMP signalling mechanism, one involving cAMP-GEF-II (Fig. 8). Such a model is particularly noteworthy in light of the newly recognized role of cAMP-GEFs as mediators of β-cell stimulus-secretion coupling (Ozaki et al. 2000; Leech et al. 2000). Although it has yet to be ascertained exactly how cAMP-GEF-II exerts its action in this system, previous studies demonstrated that Rap1b, an immediate downstream effector of cAMP-GEF-II, does interact with SERCA (Lacabaratz-Porret et al. 1998). Alternatively, cAMP-GEF-II might interact with ion channels such as RYR-2, a possibility suggested by its reported ability to exhibit direct protein-protein interactions with the sulfonlyurea receptor (SUR1), a subunit of cell surface K-ATP channels (Ozaki et al. 2000).

Figure 8. Model illustrating the role of cAMP and cAMP-GEF-II as mediators of CICR.

Occupancy of GLP-1 receptors stimulates cAMP production, an effect mediated by G proteins and adenylyl cyclase (Ad. Cyclase). SERCA-mediated Ca²⁺ uptake maintains the filling state of ER Ca²⁺ stores. The action of cAMP is mediated by cAMP-GEF-II. Sensitization of the ER Ca²⁺ release mechanism by cAMP allows RYR-regulated CICR in response to entry of Ca²⁺ through L-type voltage-dependent Ca²⁺ channels (VDCCs).