Glucagon-Like Peptide 1 Elevates Cytosolic Calcium in Pancreaticβ -Cells Independently of Protein Kinase A¹

Abstract

Glucagon-like peptide 1 (7–36)amide (GLP-1) is an insulinotropic intestinal peptide hormone with a potential role as antidiabetogenic therapeutic agent. It mediates a potentiation of glucose-induced insulin secretion, by activation of adenylate cyclase and subsequent elevation of cytosolic free calcium,[ Ca²⁺]_cyt. We investigated the role of protein kinase A (PKA) in GLP-1 signal transduction, using isolated mouse islets as well as the differentiated β-cell line INS-1. Two specific inhibitors of PKA, (Rp)-adenosine cyclic 3′,5′-phosporothioate (Rp-cAMPS, up to 3 mm) and KT5720 (up to 10 μm), did not inhibit the GLP-1-induced[ Ca²⁺]_cyt elevation. Another PKA inhibitor, H-89, reduced the [Ca²⁺]_cyt elevation only when applied at high concentrations (10–40 μm), higher than sufficient for PKA inhibition in many cell types. Furthermore, at these concentrations, H-89 also inhibited presumably PKA-independent processes such as glucose-induced [Ca²⁺]_cyt elevations and intracellular calcium storage. This suggests a PKA-independent action of H-89. Similarly to H-89, the potent but unselective protein kinase inhibitor staurosporine inhibited the GLP-1-induced [Ca²⁺]_cyt elevation only at high concentrations, at which it also inhibited glucose-induced [Ca²⁺]_cyt elevations. The same observations as with GLP-1 were made when adenylate cyclase was stimulated with forskolin, for selective examination of signal transduction downstream of receptor and G protein. Our results suggest that the GLP-1-induced[ Ca²⁺]_cyt elevation is mediated independently of PKA and thus belongs to the yet-little-characterized ensemble of effects that are mediated by binding of cAMP to other target proteins.

GLUCAGON-LIKE peptide 1 (7–36)amide (GLP-1) is an insulinotropic intestinal peptide hormone (1). Its signal transduction in pancreatic β-cells has received recent interest, prompted particularly by a potential role of GLP-1 as an antidiabetogenic therapeutic agent. GLP-1 mediates a potentiation of glucose-induced insulin secretion. This effect is derived from an elevation of cytosolic free calcium, [Ca²⁺]_cyt (2, 3), consecutively triggering insulin granule exocytosis. In addition, it is established that GLP-1 mediates a calcium-independent stimulation of the secretion machinery, resulting in enhanced recruitment of secretory granules to the plasma membrane (4, 5).

GLP-1 signal transduction still is not fully understood. It seems that GLP-1 mediates its effects mainly by stimulation of adenylate cyclase and subsequent elevation of cAMP. In contrast, the ensuing steps mostly are not known or are subject to controversial discussion. Effects of elevated cAMP are often mediated by protein kinase A (PKA). At least one important exception, however, is the modulation of ion channels. Several types of channels can be modulated by binding of cAMP, independently of phosphorylation by PKA (6).

Recently, it was proposed that also the GLP-1-induced[ Ca²⁺]_cyt elevation is associated with direct ion channel modulation by cAMP. In a study that was mainly performed with insulinoma cells, but included some measurements with native ratβ -cells, GLP-1 was found to activate an inward current with properties typical for nonselective cation channels (7). Earlier, in excised patches from a less frequently used insulinoma cell line (CRI-G1), a nonselective cation channel had been described with direct modulation by cAMP, independent of PKA (8, 9). It was therefore suggested (7) that GLP-1 elevates[ Ca²⁺]_cyt via cAMP-induced opening of a nonselective cation channel, leading to depolarization by sodium influx and subsequent opening of voltage-dependent calcium channels (VDCCs). Surprisingly, Rp-cAMPS, a PKA inhibitor that acts by blocking the cAMP binding site (10), mediated similar effects as GLP-1 (7). This was viewed as further support for the cAMP-regulated cation channel thesis. It was proposed that a less specific cAMP binding site on the channel (than in PKA) could have been activated by Rp-cAMPS, reflecting the findings in excised patches from CRI-G1 cells (9). However, no other, structurally distinct PKA inhibitors were examined, and the effect of Rp-cAMPS pretreatment was not tested. Thus, the participation of PKA was not assessed conclusively.

Other studies, however, have reached the opposite conclusion: mediation of the GLP-1-induced [Ca²⁺]_cyt rise by PKA activation. In a study with βTC3 insulinoma cells (11), the elevating action of GLP-1 on[ Ca²⁺]_cyt was completely blocked by a low concentration of Rp-cAMPS (10 μm), suggesting PKA involvement. In another study, preincubation with the PKA inhibitor H-89 (12) abolished the calcium elevations induced by the cAMP agonist dibutyryladenosine 3′,5′-cyclic monophosphate in rat β-cells (13). Although a high concentration of H-89 (40 μm) was required, partly inhibiting the glucose-induced calcium response, these results were taken as evidence that cAMP-mediated calcium elevations depend on PKA. Finally, an earlier study reported that Rp-cAMPS inhibited the facilitating effect of GLP-1 on glucose-induced depolarization in a rat β-cell preparation that displayed low glucose sensitivity (14).

These conflicting observations prompted us to readdress this issue, employing both a suitable β-cell model and adequate tools for PKA inhibition. For this purpose, we chose a combination of mouse islets on the one hand and the differentiated, glucose-sensitive β-cell line INS-1 (15) on the other. We considered regular islets as most useful because they allow us to assess β-cell behavior within a natural environment (16, 17). The risk of preparation-induced artefacts is much lower than in single, dispersed β-cells, where it is an obvious problem. On the other hand, β-cells in isolated islets may be influenced by the adjacent endocrine cell types in an uncontrolled manner. Therefore, we used INS-1 cells, in addition.

We employed a new, highly specific PKA inhibitor, the staurosporine derivative KT5720 (18), as well as the unselective, but very potent, kinase inhibitor staurosporine itself (19) to avoid interpretation problems caused by the obvious limitations of Rp-cAMPS and H-89. Both, if solely used, do not allow a sufficient clarification of the role of PKA for [Ca²⁺]_cyt under GLP-1 stimulation.

Materials and Methods

Solutions and chemicals

For islet isolation, HBSS buffer was used (sterile, from Life Technologies, Inc., Eggenstein, Germany). Calcium measurements in islets were performed with a solution containing 120 mm NaCl, 4.7 mm KCl, 2.5 mm CaCl₂, 1.2 mm MgCl₂, 5 mm NaHCO₃, and 20 mm HEPES buffer, adjusted to pH 7.4 (at 37 C) with NaOH. Measurements of[ Ca²⁺]_cyt in INS-1 cells were performed with a modified Krebs-Ringer buffer containing 134 mm NaCl, 4.7 mm KCl, 2 mm CaCl₂, 1.2 mm MgSO₄, 1.2 mm KH₂PO₄, 5 mm NaHCO₃, 10 mm glucose, and 25 mm HEPES buffer, adjusted to pH 7.4 (at 37 C) with NaOH.

The cell culture chemicals RPMI 1640, FCS, and solutions containing penicillin, streptomycin, EDTA, trypsin, or pyruvate, were obtained from Life Technologies, Inc. Fura-2 acetoxymethylester and fura-2 potassium salt were from Molecular Probes, Inc. (Eugene, OR). GLP-1 (7–36)amide was obtained from Bachem Biochemica GmbH (Heidelberg, Germany). Forskolin and H-89 were from BIOMOL Research Laboratories, Inc. (Hamburg, Germany). Staurosporine was purchased from Sigma Chemical Co. (Deisenhofen, Germany). KT5720 was from Calbiochem (Bad Soden, Germany). Rp-cAMPS was from Biolog (Bremen, Germany). Collagenase, from Clostridium histolyticum, 0.9 U/mg, was from Serva (Heidelberg, Germany).

For insulin determination in the perifusion experiments, a commercially available RIA was used, the SRI-13K Sensitive Rat Insulin RIA Kit, with rat insulin as standard and 100% cross-reactivity for mouse insulin, from Linco Research, Inc. (St. Charles, MO), purchased through Labodia, (Chanta-Merloz, Yens, Switzerland).

Measurements of cytosolic calcium,[ Ca²⁺]_cyt, in the rat β-cell line INS-1

INS-1 cells, passages 80–100 (kindly donated by C. B. Wollheim, University of Geneva, Switzerland) were grown in RPMI 1640 medium with 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mm sodium pyruvate, 10 mm HEPES, and 50μ m mercaptoethanol, as described (15). After 6–8 days, before confluency, the cells were detached with EDTA/trypsin, then maintained in spinner culture for 2–3 h, at 37 C, in the same medium, but with 25 mm HEPES and 5% FCS. The cells were then loaded with fura-2, by spinner culture in the presence of 2μ m fura-2 acetoxymethylester (fura-2/AM) for 30 min, at 37 C. After washing, the cells were kept in spinner culture at room temperature and were used subsequently for calcium measurements.

Measurements of [Ca²⁺]_cyt were performed with 10⁶ cells/ml in stirred, thermostated (37 C) cuvettes in a spectrofluorimeter (Perkin-Elmer GmbH, Langen, Germany, LS 50 B). Excitation and emission wavelengths were set to 340 and 505 nm, respectively. Calibration and compensation for extracellular fura-2 were done as described before (20).

Preparation of mouse islets for measurements of islet cytosolic calcium

Mouse islets were prepared with collagenase treatment of pancreata from DBA/2 mice (male, 6–8 weeks old, fed ad libitum, obtained from Charles River Deutschland GmbH, Sulzfeld, Germany). After preparation, the isolated islets were cultured for 3–6 days in RPMI 1640 (11 mm glucose) supplemented with 10% FCS, 100 μg streptomycin/ml, 100 U penicillin/ml, at 37 C, gassed with 95% O₂-5% CO₂. For calcium measurements, islets were then loaded with fura-2 in this medium, by 40 min incubation with 5 μm fura-2/AM, at 37 C.

Measurements of islet cytosolic calcium,[ Ca²⁺]_cyt, with digital imaging fluorescence microscopy

Single fura-2-loaded islets were transferred to a coverslip, which formed the bottom of an open superfusion chamber. The chamber was mounted onto the stage of an inverted microscope (Zeiss Axiovert 135 TV, Carl Zeiss, Oberkochen, Germany), held by an aperture in a thermostated metal block on the stage. The islets in the chamber were superfused continuously at 1 ml/min, using a peristaltic pump. The chamber vol was 700 μl. Solution changes were accomplished rapidly by means of a valve attached to an 8-chambered superfusion reservoir. The reservoir and the metal block on the microscope stage were thermostated to 37 C. Measurements of [Ca²⁺]_cyt were performed using a Zeiss/Attofluor RatioVision digital imaging system, with alternating excitation of the cells at 334 and 380 nm, monitoring of the resultant emission at 520 nm by an intensified CCD camera (512 × 512 pixels), and subsequent digitizing of the signal. For determination of [Ca²⁺]_cyt, the ratio (R) of the emissions at the two excitation wavelengths was formed, and[ Ca²⁺]_cyt was calculated according to the published equation: [Ca²⁺]_cyt = (R-R_min)/(R_max-R) × dissociation constant (K_d) × (S_f2/S_b2) (21). Calibration was done by measuring two external standards, containing calcium-saturated and calcium-free fura-2, respectively. This yielded R_max, R_min, S_f2, and S_b2. As K_d, 224 nm was used (21).

Measurement of insulin secretion from mouse islets

The same islet preparation, with islets 3–4 days after isolation (as for the calcium measurements) was also used for measurements of insulin secretion. For each experiment, 50 islets were kept in a perifusion chamber, as described previously (22), at 37 C, and perifused at a rate of 1 ml/min with Krebs-Ringer buffer supplemented with 10 mm HEPES buffer (adjusted to pH 7.4 with NaOH) and 1 mg BSA/ml. A valve allowed switching of perifusion buffers, for perifusion with different glucose concentrations and other added substances. All perifusion buffers were saturated with carbogen gas, 95% O₂-5% CO₂. The perifusate was collected over 1 or 3 min, as demonstrated by the spacing of data points (see Fig. 3C). The experiments started with 20 min perifusion with 3 mm ambient glucose, before sampling. Insulin was determined with the SRI-13 K RIA from Linco Research, Inc.

Statistical analysis

Results are usually presented as mean ± sd. Differences between results were analyzed with Student’s t test (unpaired). The results of islet calcium measurements were usually obtained with islets from at least three distinct preparations, except when only two experiments were performed. In the latter case, islets from two independent preparations were used.

Results

Effects of protein kinase inhibitors on cAMP-associated[ Ca²⁺]_cyt elevations in the β-cell line INS-1

Regarding the potential value of GLP-1 in diabetes therapy, one important property is the glucose-dependency of its insulinotropic action on β-cells. GLP-1 is typically effective at glucose concentrations that, by themselves, stimulate insulin secretion. GLP-1 signal transduction should therefore be assessed with cells that already are stimulated by glucose. First, we examined the GLP-1-induced[ Ca²⁺]_cyt elevation in the β-cell line INS-1. INS-1 is a clonal, differentiated β-cell line that can be used to avoid disadvantages of isolated primary β-cells, such as potential alterations by the isolation procedure. The experiments were performed at 10 mm glucose. In INS-1 cells, in which a graded glucose responsiveness is preserved, this glucose concentration mediates an intermediate stimulation of glucose-dependent processes, including glucose-induced [Ca²⁺]_cyt elevation (15).

In addition to GLP-1, the effect of the adenylate cyclase activator forskolin was examined. The use of forskolin enabled us to examine, selectively, GLP-1 signal transduction downstream of receptor and G protein. This seemed useful, in light of possible kinase inhibitor side actions on receptor activation and coupling to adenylate cyclase.

In INS-1 cells, GLP-1 (Fig. 1A) or forskolin (Fig. 1C) induced a biphasic [Ca²⁺]_cyt elevation. The effects of GLP-1 and forskolin were distinguished only by an indentation in the GLP-1 traces between initial peak and plateau, and by a lower plateau in the case of GLP-1. These differences may reflect desensitization of the GLP-1 receptor. A maximal effect on [Ca²⁺]_cyt required 10⁻⁸m GLP-1 (10⁻⁹, 5 × 10⁻⁹, 10⁻⁸, and 2 × 10⁻⁸m examined) or 1μ m forskolin (0.1, 0.5, 2, and 5 μm examined), respectively. These concentrations were thus employed throughout the study. The height of the initial calcium peak, after GLP-1 or forskolin addition, was used to assess the action of the inhibitors.

The two distinct inhibitors of PKA Rp-cAMPS and KT5720 had a similar effect on the [Ca²⁺]_cyt elevations induced by GLP-1 or forskolin. Neither high concentrations of Rp-cAMPS (up to 1 mm) nor KT5720 (up to 10μ m) inhibited [Ca²⁺]_cyt elevations that were induced by GLP-1 or forskolin (Fig. 1); 1 mm Rp-cAMPS mediated a very small calcium elevation during the first 2 min after its addition, peaking at 9 ± 4 nm (n = 5). The PKA inhibitor H-89 reduced the[ Ca²⁺]_cyt elevations at higher concentrations (10–40 μm, Fig. 1). However, 5 μm H-89, in many studies (23–27) mediating a strong antagonistic effect on PKA-dependent processes, did not produce a significant inhibition (Fig. 1).

In light of the strong inhibitory effect of higher H-89 concentrations, it was of interest to examine whether H-89 can have actions on cytosolic calcium possibly not mediated by PKA inhibition. For this purpose, we investigated the effect of H-89 on calcium release from intracellular stores. Calcium transport by intracellular stores inβ -cells, so far, does not seem to be influenced by the cAMP pathway and PKA, although this issue is still under investigation. To examine calcium stores, 1 μm thapsigargin was added to INS-1 cells under the same conditions as GLP-1 or forskolin in the experiments shown in Fig. 1. Thapsigargin inhibits the calcium pumps of intracellular stores, which empties the stores rapidly, via a leak conductance of the store membrane (28). This effect is used to assess the filling state of the stores. H-89 had a pronounced inhibitory effect on calcium store filling. Thapsigargin elevated[ Ca²⁺]_cyt in INS-1 cells by 249 ± 23 nm (mean ± sd) in the absence of H-89. This calcium rise was decreased to 110 ± 20 nm, 48 ± 5 nm, or 19 ± 4 nm in the presence of 10, 20, or 40 μm H-89, respectively (n = 4).

After H-89, we investigated the effect of staurosporine. Staurosporine is a very potent, albeit unselective, protein kinase inhibitor (19). Because of its high potency, it is especially suited to reveal a modulatory influence of protein kinases, including PKA. As H-89, staurosporine inhibited the [Ca²⁺]_cyt elevations induced by GLP-1 or forskolin only at higher concentrations (Fig. 2).

Effects of PKA inhibitors on GLP-1-induced[ Ca²⁺]_cyt elevations in mouse islets

To examine primary β-cells within their natural environment, we performed experiments with mouse islets.[ Ca²⁺]_cyt measurements with mouse islets reflect the [Ca²⁺]_cyt changes in β-cells (16, 17, 29) because this is the dominant islet cell type and all islet cells are coupled electrically via gap junctions (30). First, it was of interest to establish the glucose concentration at which the cAMP-mediated [Ca²⁺]_cyt elevation was maximal. We examined the effects of 8, 11, 15, and 20 mm glucose. The experiments were always started at 3 mm glucose. At this glucose concentration,[ Ca²⁺]_cyt remained at a low and stable value, as can be seen, for example, in the beginning of the traces in Fig. 3, A and B. Increasing the glucose concentration produced a calcium peak that was followed by oscillations at 8, 11, and 15 mm and by a plateau-like calcium elevation at 20 mm glucose. Additional superfusion of the islets with 10⁻⁸m GLP-1 induced a further calcium elevation, with attenuation or abolition of calcium oscillations (Fig. 3A). For a better resolution of the GLP-1 effect, superfusion with the peptide was kept short and limited to complete evolution of the initial peak in most experiments, to allow for return to the prestimulatory[ Ca²⁺]_cyt values during the experiment.

Examining glucose concentrations of 8, 11, 15, and 20 mm, the effect of GLP-1 was highest at 8 and 11 mm glucose. Because the glucose-induced calcium changes were less variable at 11 than at 8 mm glucose, all further experiments were conducted at 11 mm glucose. The effect of 1μ m forskolin (Fig. 3B) again, as in INS-1 cells, was similar to that of 10⁻⁸m GLP-1. Some variability between islets was encountered in the absolute calcium levels (e.g. see Fig. 3, A and B). However, the effects of GLP-1 or forskolin were remarkably constant in relation to other calcium elevations in the experiments, such as the glucose-induced effects.

To verify the relevance of our data, we examined cAMP-induced effects on insulin secretion in our islet preparation (Fig. 3C). Our measurements showed that the preparation responded to glucose or cAMP elevations in a typical way. Elevation of ambient glucose from 3–11 mm produced a short decrease followed by a pronounced peak-like increase in insulin secretion. The intermittent secretion decrease is reminiscent of similar calcium decreases in some of our measurements (e.g. see Fig. 6A) or in experiments by others (31). A glucose secretion peak, followed by return to (or nearly to) prestimulatory levels, has often been observed in secretion experiments with mouse islets (32, 33).

In the secretion measurements, cAMP elevation (by forskolin) had a somewhat different effect than in the calcium measurements. The secretion increase was larger and more protracted than the glucose-induced peak, in comparison with the calcium measurements. This most likely reflects the well-described calcium-independent part of cAMP-mediated stimulus secretion coupling, consisting mainly of enhanced translocation of secretory granules to the cell surface (4, 5). In summary, our secretion data reproduced the findings of others, with mouse islets (32–34), and thus confirm the relevance of our calcium measurements with islets. We did not perform any secretion experiments with INS-1 cells, because it is well established that this cell line is not well suited to examination of insulin secretion, in contrast to its higher similarity to primary β-cells with regard to calcium and membrane potential changes as well as metabolism (15).

None of the three PKA inhibitors under study abolished the GLP-1-induced [Ca²⁺]_cyt elevation in islets, even when used in concentrations in excess of those that have mediated a complete inhibition of PKA-dependent processes in other studies.

In the presence of 1 mm Rp-cAMPS (Fig. 4A), GLP-1 elevated[ Ca²⁺]_cyt in the same manner as in the absence of this inhibitor (Fig. 3A). Rp-cAMPS reduced amplitude and increased frequency of the glucose-induced calcium oscillations (Fig. 4A). The mechanism of this effect remains unclear. It could be attributed to inhibition of residual PKA activity generated, for example, by glucagon from islet α-cells (35). However, similar effects were observed also after stimulation with GLP-1 or forskolin. The oscillation amplitude was further reduced by increasing the Rp-cAMPS concentration to 3 mm (not shown). This, however, did not alter the GLP-1-induced calcium elevation. At 3 mm Rp-cAMPS, added to the superfusion medium 10 min before GLP-1, 10⁻⁸m GLP-1 raised[ Ca²⁺]_cyt to 291 nm (mean of two experiments). The effect of GLP-1, furthermore, was not affected by prolonged pretreatment with Rp-cAMPS. When 1 mm Rp-cAMPS was added to the superfusion medium 25 min before GLP-1, 10⁻⁸m GLP-1 raised [Ca²⁺]_cyt to 293 nm (mean of two experiments). Even the highest Rp-cAMPS concentration examined (3 mm) did not produce a[ Ca²⁺]_cyt elevation.

KT5720 (10 μm, added before GLP-1), attenuated the GLP-1-mediated[ Ca²⁺]_cyt elevation in islets slightly (Fig. 4B). KT5720 influenced the glucose-induced calcium oscillations in a fashion similar to that of Rp-cAMPS, lowering amplitude and increasing frequency. However, KT5720 also mediated a slight decrease of the mean[ Ca²⁺]_cyt during glucose- induced calcium oscillations (Fig. 4B), in contrast to Rp-cAMPS. The attenuation of the GLP-1-induced calcium elevation by KT5720 may partly be caused by the preceding decreased mean[ Ca²⁺]_cyt.

H-89, at high concentrations (20 μm, Fig. 5A; or 40 μm, Fig. 5B), produced a considerable inhibition of the GLP-1- induced [Ca²⁺]_cyt elevation in islets. This inhibition, however, was paralleled by a proportional inhibition of the underlying effects of glucose on[ Ca²⁺]_cyt (Fig. 5). This action comprised two separate effects. As with Rp-cAMPS and KT5720, H-89 decreased amplitude and increased frequency of the glucose-dependent[ Ca²⁺]_cyt oscillations. Still, it also mediated a very pronounced decrease of the mean cytosolic calcium level during glucose-dependent calcium oscillations.

Effects of the unselective protein kinase inhibitor staurosporine on cAMP-associated [Ca²⁺]_cyt elevations in mouse islets

Because staurosporine, in contrast to the specific PKA inhibitors Rp-cAMPS and KT5720, displayed an inhibitory action on GLP-1- or forskolin-induced[ Ca²⁺]_cyt elevations in INS-1 cells, a closer examination of its effects in mouse islets was of interest. Here, we also performed experiments with forskolin, for selective assessment of signal transduction downstream of receptor and G protein. We found that staurosporine is also a pronounced inhibitor of glucose-induced[ Ca²⁺]_cyt changes in islets, and not only of the forskolin-induced calcium elevation (Fig. 6). The effect of forskolin disappeared in parallel to the inhibition of glucose-dependent[ Ca²⁺]_cyt changes by staurosporine (Fig. 6, C and D). The same was found for the GLP-1-induced[ Ca²⁺]_cyt elevation in islets (not shown).

Discussion

We found that three structurally distinct inhibitors of PKA left the GLP-1- or forskolin-induced[ Ca²⁺]_cyt rise in pancreatic β-cells grossly untouched, even under conditions that have proven suitable for suppression of other PKA-dependent processes, in many cell types. This finding suggests that the GLP-1-induced, cAMP-mediated[ Ca²⁺]_cyt elevation in β-cells is independent of PKA and thus very likely belongs to the interesting, but still only little characterized, ensemble of effects that are triggered by binding of cAMP to target proteins other than PKA.

Rp-cAMPS (10, 36, 37) is selective for PKA, because it competes for the cAMP binding site of the enzyme. Rp-cAMPS has been used successfully to identify a number of PKA-dependent processes in diverse cell types. In most cases, 10⁻⁴m or lower concentrations of Rp-cAMPS, with a preincubation time not longer than 15 min, have been sufficient for complete inhibition of PKA-mediated effects (36, 37). In our study, a 10-fold higher concentration of the inhibitor did not inhibit GLP-1- or forskolin-induced[ Ca²⁺]_cyt elevations, either in INS-1 cells or in islets. This observation clearly argued against involvement of PKA. However, a potential disadvantage of Rp-cAMPS is its slow penetration into cells (36, 37). Keeping the presently-used high concentrations in mind (up to 3 mm in islets), this was not a likely cause for the absence of an inhibitory action. Still, we nevertheless examined the effects of the more lipophilic, and thus more permeant, inhibitors KT5720 (18) and H-89 (12).

In the majority of studies with KT5720, concentrations below 10μ m have been used for complete inhibition of PKA-dependent processes (38–40). In our study, 10 μm KT5720 mediated only a slight inhibition of the GLP-1-induced [Ca²⁺]_cyt elevation in islets. Lack of a relevant inhibitory action of KT5720 was furthermore confirmed by its insignificant effect in INS-1 cells.

Unlike the other two inhibitors, H-89 produced a considerable inhibition of GLP-1- or forskolin-induced[ Ca²⁺]_cyt elevations, albeit at rather high concentrations. At these concentrations, H-89 also inhibited glucose-dependent [Ca²⁺]_cyt elevations in islets, and calcium transport by intracellular stores in INS-1 cells. These processes probably do not depend on PKA to such a large extent, although (particularly) the involvement of PKA in calcium storage is still under investigation. The glucose-induced[ Ca²⁺]_cyt elevation involves closure of ATP-sensitive potassium channels and opening of L-type VDCCs (41). PKA has no central role, if any at all, in the nutrient-mediated closure of ATP-sensitive potassium channels (42). Furthermore, in contrast to other cell types, PKA has only a very small activating effect on L-type VDCCs in β-cells (5, 42). Calcium transport by intracellular stores in β-cells, so far, does not seem to be under the influence of PKA (43, 44). In conclusion, the results revealed pronounced unspecific inhibitory actions of higher H-89 concentrations against several calcium transport processes. This obviously demonstrates a limited usefulness of H-89 as PKA inhibitor in studies on[ Ca²⁺]_cyt regulation.

In contrast to our study, a number of cAMP-induced processes were reported to be inhibited by concentrations of H-89 below 1 or 10μ m (45, 23–27).

In the present study, H-89 at concentrations below 10 μm had only small inhibitory effects. Considering the high concentration of H-89 required for inhibition of the GLP-1- induced[ Ca²⁺]_cyt elevation, its effects on glucose-induced [Ca²⁺]_cyt elevations and calcium storage, and the lacking inhibitory action of the other inhibitors, it has to be assumed that the effect of H-89 on GLP-1 calcium signaling is of an unspecific nature, different from PKA inhibition.

The results we obtained with staurosporine, a potent (but unselective) protein kinase inhibitor, also did not suggest involvement of PKA. Again, high concentrations were required for inhibitory effects, and these concentrations also inhibited glucose-induced[ Ca²⁺]_cyt elevations. Remarkably, already 20 nm staurosporine abolished the GLP-1-induced augmentation of insulin secretion in perifused rat islets (46). This effect can be attributed to a PKA-regulated step in insulin secretion that is downstream of GLP-1-induced depolarization and[ Ca²⁺]_cyt elevation (4, 5). With regard to[ Ca²⁺]_cyt, we observed inhibitory effects only at staurosporine concentrations approximately 2 orders of magnitude higher. Staurosporine, like H-89, may thus have acted unspecificly. Both staurosporine (19) and H-89 (12) inhibit protein kinases by competing with ATP for its binding site. In β-cells, ATP binding sites exist with regulatory function for[ Ca²⁺]_cyt and insulin secretion; for example, on ATP-sensitive potassium channels (41). Higher concentrations of H-89 and staurosporine may mediate inhibitory actions by binding to these sites.

Other investigators have reached different conclusions about PKA involvement in the GLP-1-induced depolarization and[ Ca²⁺]_cyt elevation. In a study with ratβ -cells, the cAMP-mediated [Ca²⁺]_cyt elevation was abolished by pretreatment with 40 μm H-89 for 20 min (13). However, similar to our observations, this also led to a partial inhibition of glucose-mediated[ Ca²⁺]_cyt elevations. As we have shown, these results, obtained with a high concentration of H-89, do not allow us to reach conclusions on a mediation of the GLP-1-induced[ Ca²⁺]_cyt rise by PKA.

In studies with isolated rat (14) or mouse (5) β-cells, Rp-cAMPS inhibited the GLP-1-induced depolarization. This effect was attributed to a decreased inhibition of ATP- sensitive potassium channels by GLP-1. Because the results have been obtained with the selective PKA inhibitor Rp-cAMPS, the observations most likely reflect involvement of PKA. However, several points argue against central relevance of these observations. An earlier investigation, with mouse islets, has provided substantial evidence against an involvement of ATP-sensitive potassium channels in this process (34). Elevation of cAMP did not decrease β-cell potassium conductance. These results are now complemented by our data arguing against PKA involvement. Furthermore, the recent molecular characterization of the β-cell ATP-sensitive potassium channel has not yielded evidence for an inhibitory modulation by PKA (47, 48).

The inhibitory effects of Rp-cAMPS in isolated β-cells have been observed either at a low glucose concentration, 5 mm (5), or with cells that displayed a considerably reduced responsiveness to glucose (14). Interestingly, a very small forskolin-mediated decrease of rubidium efflux at 3 mm glucose, pointing to a cAMP-mediated decrease of β-cell potassium conductance, was furthermore found in mouse islets (34). It is possible that PKA has a slight inhibitory action on ATP-sensitive potassium channels that is manifest at substimulatory glucose concentrations. On the other hand, it cannot be ruled out that the observations reflect altered properties of β-cells, because of the isolation procedure. In any case, the present study provides results with general relevance for the mechanisms of GLP-1 action on[ Ca²⁺]_cyt, because glucose concentrations were used that enable maximal GLP-1 effects, and the examination of islets reflects β-cell activity in a more physiological environment.

Some indirect evidence against PKA involvement in the GLP-1-induced [Ca²⁺]_cyt rise has already been presented by others. In one study, the PKA inhibitor Rp-cAMPS surprisingly mimicked the depolarization and[ Ca²⁺]_cyt rise that was induced in βTC6 insulinoma cells by GLP-1, the cAMP analog 8-bromo-cAMP, or the PKA agonist Sp-cAMPS (7). It was concluded that the effects were mediated by a cAMP-binding site with lower specificity than that in PKA. For example, the results seemed compatible with direct activation of an ion channel by cAMP. No further use of Rp-cAMPS was made in that study, and no other PKA inhibitors were tested. The observation of prominent agonistic properties of Rp-cAMPS is not confirmed by our findings. We only found a comparatively small and protracted[ Ca²⁺]_cyt rise by Rp-cAMPS in INS-1 cells, and none at all in islets. It is conceivable that these differences are attributable to the rapid, focal application of substances by a micropipette in the mentioned study (7). This may have led to an extraordinarily rapid concentration increase of Rp-cAMPS within the cells, which may have unmasked a weak partial agonistic activity of the inhibitor (9). In summary, care should be taken in the interpretation of such results, with regard to PKA involvement.

In CRI-G1 insulinoma cells, a nonselective cation channel has been characterized in excised membrane patches that is activated by addition of cAMP, probably reflecting PKA-independent regulation by cAMP (8, 9). Our results encourage an intensified search for such channels inβ -cells.

In summary, PKA-independent [Ca²⁺]_cyt elevation seems to be one of two branches of GLP-1 signal transduction in β-cells, the other branch being the already-known PKA- dependent stimulation of the secretory machinery, resulting in increased recruitment of secretory granules to the plasma membrane. The divergent features of these two branches could be associated with similarly divergent functional roles, in the normal as well as in the pathological state.

Acknowledgments

We thank C. B. Wollheim, University of Geneva, for the gift of the INS-1 cell line.

References