Introduction

High glucose levels stimulate the release of ATP into the extracellular space from several tissues and cell types such as endothelial cells, blood vessels, mesangial cells, macrophages and beta cells [1–5]. The net pharmacological effects of extracellular ATP are difficult to predict since ATP by itself can stimulate P2Y2, P2Y4 (in rodents), P2Y11 and P2X1–P2X7 receptors present on the cell surface [6]. Furthermore, ATP is rapidly degraded by ectonucleotidases to ADP [7], which can act on P2Y1, P2Y12 and P2Y13 receptors. ADP is then further degraded by ecto-5′-nucleotidase to adenosine, which in turn can activate four different adenosine receptors [8].

A large number of studies indicate that extracellular ATP and ADP have a key role in regulating insulin secretion by purinoreceptors [9–13]. Insulin-secretory granules contain 3.5 mmol/l ATP and 5 mmol/l ADP [14], and with a novel biosensor it has been demonstrated that glucose stimulation releases ATP from a single pancreatic beta cell to a local extracellular ATP concentration exceeding 25 µmol/l [4].

ADP has been shown to increase insulin release via P2Y receptors in vitro, in vivo and in human beta cells [9–11]. Beta cells are activated by ADP and can also propagate this signal via intermittent release of ATP [15]. However, ADP has also been shown to decrease insulin release [12, 13]. These discrepancies are probably in part due to involvement of several different purinoreceptors [13, 15–22]. Therefore, we wanted to examine whether endogenous release of purines may regulate insulin secretion and to determine the P2Y receptors involved. To do so we quantified the mRNA levels for all the purinergic P2Y receptors, and based on these findings we did pharmacological studies on insulin and glucagon secretion in isolated mouse islets. We found an unanticipated role for ADP acting on P2Y13 receptors as an autocrine inhibitor of insulin release both in vitro and in vivo.

Methods

Ethics

The study conforms to the Guide for the Care and Use of Laboratory Animals, US National Institute of Health (NIH Publication No. 85-23, revised 1996) and was approved by the Ethics Committee of Lund University, Sweden.

Animals

Female mice of the NMRI strain (B&K Universal, Sollentuna, Sweden), weighing 25–30 g were used for in vitro and in vivo studies. They were fed a standard pellet diet (B&K Universal) and tap water ad libitum.

Isolation of islets and beta cells

Collagenase (CLS 4) was obtained from Worthington Biochemicals, Freehold, NJ, USA. Bovine serum albumin was from ICN Biochemicals, High Wycombe, UK. IBMX (isobutylmethylxanthine), an inhibitor of phosphodiesterase, and all other drugs and chemicals were from Sigma Chemicals, St Louis, MO, USA or Merck AG, Darmstadt, Germany. Radioimmunoassay kits for insulin and glucagon determination were obtained from Diagnostika (Falkenberg, Sweden) and from Eurodiagnostica (Malmö, Sweden).

Pancreatic islets were isolated by retrograde injection of a collagenase solution via the bile–pancreatic duct. Islets were then collected under a stereomicroscope at room temperature. Purified islets that were to be used for total RNA extraction were collected directly in 1,000 μl TRIzol (Invitrogen, Paisley, UK) and stored at −80°C.

Beta cells were purified using a method previously used for enterochromaffin-like-cell purification [23] with some modifications. Freshly isolated mouse pancreatic islets (5,000 islets) were dispersed using trypsin digestion (0.9 mg/ml, Boehringer Mannheim, Mannheim, Germany) and omission of calcium in the presence of EDTA. After dispersion, the beta cell preparations were further purified by repeated counterflow elutriation using first a standard chamber and then a Sanderson chamber (Beckman, Palo Alto, CA, USA). The enriched cells from the standard chamber were collected at 25 ml/min at 380–560 g. They were purified further in a Sanderson chamber and collected at 18 ml/min and 800 g. This cell preparation consisted of about 80% beta cells. The beta cell preparation was then subjected to density gradient centrifugation. A stock solution of 60% iodixanol (Optiprep, Nycomed Pharma, Oslo, Norway), containing 1.2 mmol/l MgCl2, 15 mmol/l HEPES at pH 7.4 and 10 mg/ml BSA, was diluted to 15% and 10.8%, respectively, with medium consisting of (mmol/l): 140 NaCl, 1.2 MgSO4, 1 CaCl2, 15 HEPES at pH 7.4, 11 glucose, 0.5 dl-dithiothreitol and 10 mg/ml BSA. In a 15 ml centrifuge tube, 5 ml of the 15% iodixanol solution was overlaid with 5 ml 10.8% iodixanol. Enriched beta cells (1 × 106 cells/ml) were layered on the two layers of iodixanol and centrifuged (Beckman, Spinchron R centrifuge). A slow acceleration period (400 rpm) was followed by 5 min centrifugation during which the speed reached 1,000 rpm. Deceleration lasted 5 min. The cells in the light density fraction above the 10.8% cushion were collected. The yield of cells in this fraction was 1 ± 0.15 × 106 cells (mean ± SEM, n  =10).

The purity of each beta cell preparation was assessed by RIA measurement of insulin, glucagon, somatostatin and pancreatic polypeptide per mg protein [24]. At least 10 samples (tubes) of the cells were examined for each hormone. The final cell preparation consisted of around 95 ± 8% beta cells. The beta cells were then collected in 1,000 μl TRIzol (Invitrogen) and stored at −80°C.

RNA purification, cDNA synthesis and quantitative real-time PCR

Mouse islet and beta cell total RNA was isolated using a modified TRIzol method as described elsewhere [25]: after phase separation, the upper aqueous phase was incubated at −20°C overnight with 10 μg glycogen (Invitrogen) and centrifuged for 15 min at 12,000 g, 4°C. After addition of ethanol, the pellet was centrifuged for 10 min at 7,500 g, 4°C. Each individual RNA pellet was dissolved in RNase-free water and converted into cDNA using TaqMan reverse transcription (Applied Biosystems, Carlsbad, CA, USA) according to the manufacturer’s instructions.

All quantitative real-time PCR primers were designed using the VectorNTI software (Invitrogen, Informax, UK), using RefSeq mRNA sequences as templates. All PCR primers were designed to generate products of 199–201 base pairs (see Electronic supplementary material).

Light Cycler quantitative real-time PCR (Roche, Basel, Switzerland) was used to determine P2Y receptor expression relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) according to the manufacturer’s instructions as described elsewhere [25].

In vitro quantification of ATP release

ATP release was measured using an adaptation of the luciferase method previously described by Taylor et al. [26]. Briefly, MIN6c4 cells were grown in DMEM, 25 mmol/l glucose containing 15% FBS (Invitrogen). A total of 35,000 cells were plated in 35 mm Petri dishes and were allowed to grow for 24 h. The medium was then replaced with 1 ml serum-free DMEM containing 5.6 mmol/l glucose and the cells were cultivated for another 24 h. Subsequently, 1 ml 2× ATP SL reagent (BioThema, Handen, Sweden), in serum-free DMEM, 5.6 mmol/l glucose, was added and the Petri dish was incubated for 30 min at room temperature. After the baseline luciferase activity in the Petri dish had been recorded for 1 min, in a GloMax 20/20 Luminometer (Promega Corporation, Madison, WI, USA), 400 µl of the medium was removed. The test substances were added to the removed medium and returned to the Petri dish, where the luminescence was recorded for another 10 min.

Insulin and glucagon secretion from islets in vitro

The freshly isolated islets were pre-incubated for 30 min at 37°C in KRB, pH 7.4, supplemented with 10 mmol/l N-2 hydroxyethylpiperazine-N′-2-ethanesulfonic acid, 0.1% BSA and 4 mmol/l glucose before addition of the agents to be tested. Each incubation vial contained 12 islets in 1.0 ml buffer solution and was gassed with 95% O2–5% CO2 to obtain constant pH and oxygenation. After pre-incubation, the medium was changed and supplemented with various receptor agonists and antagonists. The islets were then incubated for 1 h at 37°C in a metabolic shaker (30 cycles per min). Aliquots of the medium were then removed for the assay of insulin and glucagon.

Measurement of cAMP levels

MIN6c4 cells, at 4 ×105 cells/well, were cultured for 3 days. After being washed with OptiMEM and treated with 25 µmol/l Rolipram, the cells were further incubated with the test compounds for 20 min. Incubations were terminated by addition of 0.1 mol/l hydrochloric acid. Cell lysates were centrifuged at 1,000 rpm for 10 min at 4°C. cAMP levels in cell extracts were measured by ELISA using the Cyclic AMP Enzyme Immunoassay kit (Cayman Chemical, Ann Arbor, MI, USA) according to manufacturer’s instructions.

In vivo measurements of insulin and glucagon secretion

Either MRS2211 (100 mg/kg) in physiological saline or saline only was administered intraperitoneally (i.p.) 10 min prior to i.p. glucose injection of 1.2 g/kg with or without MRS2211 (50 mg/kg) i.p., and blood samples were collected at 0, 3, 10 and 40 min post glucose administration as previously described [27].

Calculations and statistics

Calculations and statistics were performed using the GraphPad Prism 4.0 software (GraphPad Software, Inc., San Diego, CA, USA). Values are presented as mean ± SEM. Statistical significance was accepted when p < 0.05 (Mann–Whitney’s test).

Results

Effects of glucose on insulin and glucagon secretion in the presence of apyrase

Apyrase rapidly eliminates ATP and ADP by degradation to AMP [28], which in turn may then be degraded to adenosine by CD73 [7]. Addition of 1 U/ml apyrase (Sigma) to pancreatic islets in the presence of 1 mmol/l glucose stimulated an increase in insulin release from 0.27 ± 0.04 to 0.59 ± 0.06 ng islet−1 h−1 (p < 0.001). Raising the glucose level to 20 mmol/l resulted, as expected, in increased insulin release to 2.41 ± 0.18 ng islet−1 h−1. In the presence 1 U/ml of apyrase, this increase was attenuated to 1.31 ± 0.19 ng islet−1 h−1 (p < 0.001) (Fig. 1a).

Fig. 1
figure 1

Effect of the ATP and ADP-degrading enzyme apyrase (1 U/ml) on insulin and glucagon secretion in mouse islets at different glucose concentrations. Data are expressed as mean ± SEM, n = 10, ***p < 0.001 between control and apyrase for each glucose concentration. Black columns, apyrase; white columns, control

Full size image

Addition of 1 U/ml apyrase to pancreatic islets in the presence of 1 mmol/l glucose stimulated an increase in glucagon release from 33.9 ± 2.1 to 49.2 ± 2.5 pg islet−1 h−1 (p < 0.001). Raising the glucose level to 20 mmol/l resulted in slightly decreased glucagon release to 18.4 ± 1.3 pg islet−1 h−1 (p < 0.001). In the presence of 1 U/ml apyrase this decrease was counteracted and the glucagon level was increased to 40.1 ± 3.2 pg islet−1 h−1 (p < 0.001) (Fig. 1b).

ATP release from MIN6c4 cells

Release of ATP into the extracellular space was significantly elevated (4.8 times,p < 0.05) in the presence of glucose (16.6 mmol/l). The ATP release was unaffected by equimolar concentrations of mannitol but was inhibited by the calcium-channel blocker nifedipine (Fig. 2) [29].

Fig. 2
figure 2

ATP release from MIN6c4 cells into the extracellular space. The addition of 16.6 mmol/l glucose, but not mannitol, resulted in a significant increase (p < 0.05) in ATP release that was completely abolished by the addition of the Ca2+-channel blocker nifedipine (20 µmol/l). Black squares, glucose; black circles, mannitol; grey triangles, nifedipine; grey circles, PBS (control). SI, signal intensity

Full size image

P2Y receptor expression in mouse islets and beta cells

Real-time PCR revealed that, among the purine-sensitive P2Y receptors, the highest mRNA levels were found for the ADP receptors P2Y1 and P2Y13 in mouse pancreatic islets (6.7 ± 0.6 × 10−3 and 1.3 ± 0.6 × 10−3 respectively) and mouse beta cells (4.0 ± 3.0 × 10−3 and 1.8 ± 0.3 × 10−3 respectively) (ratio of GAPDH mRNA expression, Fig. 3). No expression of the ADP receptor P2Y12 could be detected.

Fig. 3
figure 3

Expression of mouse ADP receptors P2Y1, P2Y12 and P2Y13 in pancreatic islets and beta cells from the mouse (n = 3). Only trace amounts of P2Y12 were detected. Black columns, whole islets; white columns, isolated beta cells

Full size image

Effects of ADP on insulin and glucagon secretion

At 10 μmol/l, the stable ADP analogue 2-MeSADP, which activates both P2Y1 and P2Y13 receptors [30], did not have any effect per se on insulin secretion from mouse pancreatic islets at 8.3 mmol/l glucose (Fig. 4). However, after pre-treatment with the selective P2Y1 receptor antagonist MRS2179 [31], 10 μmol/l of 2-MeSADP resulted in a marked decrease in insulin secretion (p < 0.001), indicating an insulin inhibitory effect via the P2Y13 receptor. In contrast, pre-treatment with the selective P2Y13 receptor antagonist MRS2211 [32] resulted in a marked increase in insulin secretion (p < 0.001), indicating an insulin-stimulating effect via the P2Y1 receptor.

Fig. 4
figure 4

a Effect of the stable ADP analogue 2-MeSADP on insulin secretion in mouse islets at 8.3 mmol/l glucose. 2-MeSADP had a limited effect per se. In the presence of the P2Y1 antagonist MRS2179, 2-MeSADP reduced insulin secretion. In the presence of the P2Y13 antagonist MRS2211, insulin secretion was increased. Data are expressed as mean ± SEM, n = 9, ***p < 0.001. Black columns, 2-MeSADP; white columns, control. b Effect of 2-MeSADP on cAMP production in MIN6c4 cells. MIN6c4 cells were treated with 2-MeSADP (10 nmol/l) in the absence or presence of MRS2211 (10 μmol/l). Cells were treated with the compounds for 20 min, in the presence of 25 μmol/l Rolipram, and were extracted with 200 μl of 0.1 mol/l hydrochloric acid and analysed for cAMP. Data are expressed as mean ± SEM, n=5, ***p < 0.001

Full size image

To test whether endogenously released ADP regulates insulin release via P2Y13 receptors, we added the P2Y13 receptor antagonist MRS2211 (10 μmol/l). Interestingly, MRS2211 per se increased insulin secretion at 8.3 (p < 0.001) and 20 mmol/l glucose (p < 0.01, Fig. 5a).

Fig. 5
figure 5

Effect of the P2Y13 antagonist 10 μmol/l MRS2211 on (a) insulin and (b) glucagon secretion in mouse islets at different glucose concentrations. Data are expressed as mean ± SEM, n = 9, *p = 0.05, **p < 0.01, ***p < 0.001. Black columns, MRS2211; white columns, control

Full size image

A similar increase in glucagon secretion was seen with 10 μmol/l MRS2211 per se both at 8.3 (p < 0.001) and 20 mmol/l glucose (p < 0.01, Fig. 5b).

Effect of P2Y13 on MIN6c4 cell cAMP level

We studied the mechanism underlying the stimulatory effect of the P2Y13 agonist on insulin secretion. We examined the effect of 2-MeSADP on cAMP production. We found that 10 nmol/l of 2-MeSADP inhibited the cAMP level by 45% (p < 0.0001, Fig. 4b), while addition of the P2Y13 antagonist, MRS2211, reversed this effect of 2-MeSADP on cAMP levels (p < 0.0001).

In vivo measurements of insulin and glucagon secretion

Administration of the P2Y13 antagonist MRS2211 in vivo resulted in a significant increase of insulin and a diminished suppression of glucagon secretion following an i.p. injection of glucose (Fig. 6a, c). MRS2211 in vivo also resulted in a significant decrease of glucose following the i.p. injection of glucose (Fig. 6b).

Fig. 6
figure 6

Effect of the P2Y13 antagonist MRS2211 or saline on (a) insulin, (b) glucose and (c) glucagon secretion in vivo. MRS2211 in vivo resulted in a significant increase in insulin and decrease of glucose following the i.p. injection of glucose (a, b). Data are expressed as mean ± SEM, n  =10, *p < 0.05, **p < 0.01. Black squares, MRS2211; black triangles, saline

Full size image

Discussion

In this study we found evidence for glucose-stimulated endogenous release of insulin-regulating purines. In MIN6 cells, we observed that ATP release is induced by glucose, and that it can be blocked by the Ca2+ channel blocker nifedipine. Using real-time PCR, we identified high mRNA levels of the ADP receptors P2Y1 and P2Y13, but very low levels of the ADP receptor P2Y12, in mouse islets and beta cells. ADP was found to have a dual effect on beta cells, stimulating insulin release via P2Y1 receptors and inhibiting insulin release via P2Y13 receptors. We discovered that glucose stimulated ATP release and that the ATP metabolite ADP activated P2Y13 receptors, exerting an autocrine inhibitory effect on insulin secretion. The P2Y13 antagonist MRS2211 was shown to increase glucose-stimulated insulin release in vitro. In vivo administration of the P2Y13 antagonist MRS2211 resulted in increased release of insulin and lowered glucose levels, suggesting that P2Y13 antagonism might be of therapeutic value for the treatment of type 2 diabetes.

Purinergic nucleotides (ATP and ADP) have been found to serve as autocrine regulators of insulin release [15, 16, 18, 20, 22]. We confirmed that nucleotides are important regulators of insulin and hormone release by the addition of the enzyme apyrase, which removes ATP and ADP by degradation [30, 33]. At low glucose concentrations, the addition of apyrase led to increased insulin secretion, possibly due to reduced ADP-mediated P2Y13 activation. However, apyrase together with ecto-5′-nucleotidases also generates adenosine, which inhibits insulin release via A1 receptors [34]. At high glucose levels, we found inhibition of insulin secretion by apyrase, and this could possibly be explained by enzyme-mediated generation of adenosine. On the other hand, apyrase had a stimulating effect on glucagon release at both low and high glucose concentrations. This effect could be explained by removal of extracellular ATP, which has previously been demonstrated to inhibit glucagon release by interfering with the oscillatory Ca2+ signalling of the alpha cell [35].

It has been suggested that ATP and insulin are co-secreted from pancreatic beta cells [36]. Using MIN6 cells, we found that glucose significantly increases the release of ATP into the extracellular space, and that this release can be blocked by the Ca2+ channel blocker nifedipine, leading to the inhibition of Ca2+-mediated insulin exocytosis [37].

P2Y receptor research is hampered by the shortage of specific receptor antagonists. mRNA quantification is therefore a useful complement to investigate the importance of the many subtypes. In mouse beta cells and also in whole islets we found a very clear and consistent pattern, with high expression levels of P2Y1 and P2Y13 receptors, but extremely low or undetectable levels of P2Y2, P2Y4 [38] or P2Y12 receptors (P2Y11 is absent from the mouse genome). Therefore, we conclude that no pharmacologically relevant P2Y receptors for ATP are present in mouse beta cells or islets. As a consequence, extracellularly released ATP can only stimulate islet P2Y receptors after conversion into ADP and can then act on P2Y1 or P2Y13 receptors. ATP could still stimulate ionotropic P2X receptors on beta cells, but activation of P2X receptors has been shown to mediate only small and transient increases in insulin secretion at low glucose concentrations [39].

For the pharmacological experiments we used the stable ADP analogue 2-MeSADP to limit conversion of ADP into adenosine, which is known to inhibit insulin secretion [34]. 2-MeSADP alone did not have any measurable effect on insulin secretion at 8.3 mmol/l glucose. However, when the P2Y1 receptor was blocked by the specific P2Y1 antagonist MRS2179, a pronounced decrease in insulin secretion was observed, revealing an inhibitory effect of ADP via the P2Y13 receptor. In contrast, when the P2Y13 receptor was blocked by MRS2211 during 2-MeSADP stimulation, a marked increase in insulin secretion was seen. This indicates that ADP exerts opposing regulatory effects on insulin secretion via different receptor subtypes. ADP stimulates insulin secretion from beta cells via the Gαq-coupled P2Y1 receptor and inhibits insulin release via the Gαi-coupled P2Y13 receptor [30]. This finding may explain the sometimes contradictory results from previous studies [9–13] where ADP has been shown to both increase [9–11] and decrease [12, 13] insulin release via P2Y receptors. ADP acting on P2Y1 receptors has been reported to increase pulsatility of insulin release, but not Ca2+ rhythmicity in intact pancreas [16]. At the same time, selective P2Y1 receptor agonists have been shown to stimulate insulin release, as ADP stimulates Ca2+ transients via P2Y1 receptors [40]. The stimulatory effect of ADP via P2Y1 receptors was confirmed in our study. It is possible that the inhibitory effect of ADP via P2Y13 receptors dominated the experimental settings in which ADP decreased insulin release [12, 13].

However, the effects also depend on the presence of other cell types in the model. In studies using intact pancreas, P2Y1 receptors on endothelial cells could indirectly regulate insulin secretion via effects on blood flow or even via release of ATP from endothelial cells. It becomes even more complex in whole animal studies. The P2Y1 receptor knockout mouse exhibits both increased plasma insulin and plasma glucose [41]. The animals display increased body weight and a tendency towards glucose intolerance. To explain these findings it is speculated that P2Y1 receptors in the hypothalamus, liver and fibroblasts may be involved. In contrast to the P2Y1 receptor, the P2Y13 receptor has a more limited expression, mainly in haematopoietic cells [42], and now also in pancreatic alpha and beta cells.

In our in vivo experiments, insulin increased acutely in response to glucose, as expected, and this early response was unaffected by the P2Y13 antagonist MRS2211. However, the second phase (after 10 min), during which insulin is normally attenuated, was markedly affected by MRS2211 and insulin remained on the same level as in the acute phase. A possible explanation is that ATP is co-released with insulin during degranulation. After degradation to ADP, the ADP molecule binds to the P2Y13 receptor on beta cells and turns off insulin release. In the presence of a P2Y13 antagonist this negative feedback system is disrupted and insulin levels remain high.

For glucagon the in vivo experiments demonstrated a diminished suppression of glucagon secretion following an injection of glucose at both 3 and 10 min. The effects are probably mediated by a direct effect of MRS2211 on alpha cells, because secondary effects via increased insulin release would have decreased glucagon levels, instead of increasing them as we observed. Furthermore, these results are supported by the previously described influence of apyrase on isolated islets, and taken together they indicate an inhibitory effect either of ATP, or of its degradation products, on glucagon secretion. Although it is presently not known whether pancreatic alpha cells express P2Y13 receptors, it has been demonstrated that Gαi activation inhibits exocytosis of glucagon [43]. The well-described glucagon release-inhibiting peptide somatostatin, for example, is mediated by Gαi-coupled receptors [44]. Consequently, it is likely that inhibition of P2Y13 receptors on alpha cells would result in increased glucagon release.

The mechanism of action for the Gαi-coupled P2Y13 receptor is to decrease intracellular levels of the second messenger cAMP, and ADP has been shown to increase cAMP in pancreatic beta cells [45]. We found that an ADP analogue reduced cAMP levels and blocking the P2Y13 receptor with MRS2211 restored cAMP levels in beta cells. cAMP has been shown to improve insulin release and to protect against apoptosis and cell death [24]. Interestingly, the adenosine A1 receptor is also Gαi-coupled and also decreases cAMP levels. Caffeine, which blocks the adenosine A1 receptor, thus protects against the development of diabetes [46]. Based on their similar effects on cAMP, it is to be expected that a P2Y13 antagonist would have a similar protective effect against diabetes.

In conclusion, we have found a new role for extracellular ADP, whereby ADP acts on the P2Y13 receptor in an autocrine inhibitory pathway regulating insulin secretion in the mouse. Furthermore, we have shown that by means of a P2Y13 antagonist (MRS2211) it is possible to improve insulin secretion both in vitro and in vivo.