Atrial Natriuretic Peptide Affects Stimulus-Secretion Coupling of Pancreatic β-Cells

ANP plays an important role in the regulation of blood volume and blood pressure (1). ANP also is involved in the regulation of food intake and lipid and glucose homeostasis. Cellular effects of ANP are mediated by a plasma membrane–associated receptor with guanylate cyclase activity (GC-A receptor). Thus, activation of the receptor results in an increased cyclic guanosine monophosphate (cGMP) concentration. GC-A receptors are expressed in murine β- and α-cells and in the insulin-secreting tumor cell line INS-1E (2). However, functional studies about effects of ANP on β-cells are controversial. In cultured mouse islets, ANP increases glucose-stimulated insulin secretion (GSIS), and the effect is suggested to be mediated by closure of K_ATP channels and an increase of the cytosolic Ca²⁺ concentration ([Ca²⁺]_c) (3). In the current study, β-cells of a global GC-A receptor knockout mouse (GC-A-KO) have been used to investigate the link between this receptor and islet function. The interpretation of the data is somewhat limited because a global GC-A-KO can alter systemic parameters, including blood pressure and lipid and glucose homeostasis (e.g., through effects on insulin resistance), which can retroact on the functional status of β-cells before islet isolation. A weak insulinotropic effect also has been observed in perfused rat pancreas (4), and another study reported marked hypoglycemia after intravenous ANP infusion in rats (5). On the contrary, acute incubation of isolated rat islets with ANP did not influence insulin secretion (6,7), and long-term culture with ANP even inhibited insulin production and GSIS (7). In healthy male volunteers, infusion of ANP slightly elevated plasma insulin and moderately increased blood glucose concentrations (8–10). Taken together, the mode of action on pancreatic islets remains to be elucidated.

In vitro data with isolated islets and/or β-cells are sparse, and the experiments with ANP infusion are difficult to interpret because one cannot discriminate between effects on β-cells and effects on peripheral organs or blood flow. Long-term effects were studied in mouse models lacking the GC-A receptor; however, the data are inconsistent. Global homozygous GC-A receptor deletion has led to enhanced fasted blood glucose concentration, whereas glucose tolerance and insulin sensitivity remained unchanged in GC-A-KO mice after 12 weeks of high-fat or standard diet compared with control mice (3). In contrast, mice with heterozygous receptor deletion that were not hypertensive developed impaired glucose tolerance after a high-fat diet (11). Humans with genetic variants predisposing to low plasma concentrations of ANP and brain natriuretic peptide (BNP) (mainly secreted by the heart) exhibit a high risk for the development of hypertension and heart hypertrophy (12,13). Epidemiological studies also found an association between low ANP (and BNP) plasma concentrations and obesity, insulin resistance, and the metabolic syndrome (14–16). According to the concept of Gruden et al. (17), the lack of the beneficial effects of the natriuretic peptides on adipose tissue, skeletal muscle, and β-cells promotes the development of type 2 diabetes. The current study takes advantage of a β-cell–specific GC-A-KO (βGC-A-KO) mouse to clarify the effects of ANP on stimulus-secretion coupling of β-cells.

C57BL/6N mice (wild type [WT]) were bred in the animal facility of the Department of Pharmacology at the University of Tübingen in Germany. GC-A-KO mice and their WT littermates (l-WT) were provided by Dr. M. Kuhn (Physiological Department, University of Würzburg, Würzburg, Germany). As previously described (18), βGC-A-KO mice were generated by crossing rat insulin II promoter (RIP)-Cre mice (RIP-Cre founders of the Herrera strain) (19) with floxed GC-A mice of a C57BL/6/Sv129 background (20). Deletion of GC-A protein in islet cells was verified by immunohistochemistry (18). The Guide for the Care and Use of Laboratory Animals and German laws were followed.

The details for cell and islet preparation have been previously described (21). Briefly, collagenase was injected into the ductus pancreaticus, and exocrine tissue was digested for ∼5 min. Islets were handpicked, and clusters/single cells were made by trypsin digestion of islets.

Standard whole-cell and cell-attached recordings were done with a bath solution that contained (in mmol/L) 140 NaCl, 5 KCl, 1.2 MgCl₂, 2.5 CaCl₂, or 10 CaCl₂ (for measurements of membrane potential), glucose as indicated, and 10 HEPES, pH adjusted to 7.4 with NaOH. The same bath solution was used for the determination of [Ca²⁺]_c and the mitochondrial membrane potential (ΔΨ). The pipette solution for standard whole-cell measurements of K_ATP currents contained (in mmol/L) 130 KCl, 4 MgCl₂, 2 CaCl₂, 10 EGTA, 20 HEPES, and Na₂ATP as indicated, pH adjusted to 7.15 with KOH. For cell-attached recordings, the pipette solution contained (in mmol/L) 130 KCl, 1.2 MgCl₂, 2 CaCl₂, 10 EGTA, and 20 HEPES, pH adjusted to 7.4 with KOH. Cell membrane potential recordings were performed with amphotericin B (250 μg/mL) in the pipette solution, which contained (in mmol/L): 10 KCl, 10 NaCl, 70K₂SO₄, 4 MgCl₂, 2 CaCl₂, 10 EGTA, and 20 HEPES, pH adjusted to 7.15 with KOH. Fura-2 acetoxymethylester was obtained from Molecular Probes (Eugene, OR). RPMI1640 medium was from PromoCell (Heidelberg, Germany), penicillin/streptomycin from Gibco BRL (Karlsruhe, Germany), atrial natriuretic factor (1-28) (mouse, rabbit, rat) trifluoroacetate salt from Bachem (Weil am Rhein, Germany), Rp-8-Br-PET-cGMPS from Biolog (Bremen, Germany), and PKI 14-22 amide, myristoylated (myr-PKI) from Tocris Bioscience (Wiesbaden, Germany). All other chemicals were purchased from Sigma (Deisenhofen, Germany) or Merck (Darmstadt, Germany) in the purest form available.

K_ATP currents and membrane potentials were recorded with an EPC-9 patch-clamp amplifier by using PATCHMASTER software (HEKA Elektronik, Lambrecht, Germany). Channel activity of the single channels was measured at the actual membrane potential in the cell-attached mode. Point-by-point analysis of the current traces reveals an open probability (P_o) owing to all active channels (N) in the patch and is thus presented as NP_o. Whole-cell K_ATP current was evoked by 300-ms voltage steps from −70 to −60 mV and −80 mV. Under these conditions, the current is completely blockable by K_ATP channel inhibitors (22).

Mitochondrial ΔΨ was measured as Rh123 fluorescence at 480-nm excitation wavelength as previously described (23).

Details have been previously described (21). In brief, cells were loaded with 5 μmol/L Fura-2 acetoxymethylester for 30 min at 37°C. Fluorescence was excited at a 340- and 380-nm wavelength, and fluorescence emission was filtered (LP515) and measured by a digital camera. [Ca²⁺]_c was calibrated in vitro by using Fura-2 pentapotassium salt (24).

Details for steady-state incubations have been previously described (21). For perifusions, 50 islets were placed in a bath chamber and perifused with 3 mmol/L glucose for 60 min before the beginning of the experiment. Samples for the determination of insulin were taken every 2 min.

Each series of experiments was performed with islets or islet cells of at least three independent preparations. Mean ± SEM is given for the indicated number of experiments. Statistical significance of differences was assessed by Student t test for paired values. Multiple comparisons were made by ANOVA followed by Student-Newman-Keuls test. P ≤ 0.05 was considered significant.

Insulin secretion of isolated mouse islets was measured in vitro to evaluate whether GC-A receptor stimulation by ANP results in changed insulin secretion. First, ANP was added in the second phase of insulin release after increasing the glucose concentration from 3 to 10 mmol/L (Fig. 1A). Under these conditions ANP increased insulin secretion induced by 10 mmol/L glucose from 7.6 ± 1.2 to 8.7 ± 1.3 pg insulin/(islet ⋅ min) (P ≤ 0.001) and insulin area under the curve (AUC) (Fig. 1B). For these experiments, islets from C57BL/6N mice (WT) were used. ANP effects at a threshold glucose concentration of 6 mmol/L were tested in steady-state incubation. In islets of l-WT, ANP significantly increased insulin secretion from 0.24 ± 0.06 to 0.33 ± 0.07 ng insulin/(islet ⋅ h) (P ≤ 0.05), whereas it was ineffective in islets from βGC-A-KO mice [0.24 ± 0.04 vs. 0.23 ± 0.06 ng insulin/(islet ⋅ h)] (Fig. 1C). The stimulating effect of ANP also was absent in islets from SUR1-KO mice that lacked functional K_ATP channels (Supplementary Fig. 1A).

[Ca²⁺]_c was measured in the presence of 6 mmol/L glucose in islet cell clusters of l-WT and βGC-A-KO mice. In 6 mmol/L glucose, [Ca²⁺]_c was at basal values in most cells (i.e., no oscillations occurred). These cells with basal Ca²⁺ concentration were selected for investigation of the effect of ANP (Fig. 1D). Glucose at a concentration of 15 mmol/L was added at the end of each experiment to test whether cells were metabolically intact and thus able to show a response to ANP. Maximal [Ca²⁺]_c (max[Ca²⁺]_c) before (basal) and after application of ANP was calculated for each of the 76 and 74 experiments performed with l-WT and βGC-A-KO cells, respectively. The mean of the max[Ca²⁺]_c is an indirect measure for the percentage of ANP-responsive cells because it mirrors the number of responsive cells (Fig. 1E). Figure 1D shows the typical response to ANP for cell clusters of each genotype. The summary of the data is shown in Fig. 1E. In l-WT β-cells, ANP increased max[Ca²⁺]_c from 68 ± 4 to 270 ± 33 nmol/L (P ≤ 0.001). Switching to the bath solution with ANP also augmented max[Ca²⁺]_c in βGC-A-KO β-cells from 68 ± 4 to 135 ± 23 nmol/L (P ≤ 0.01). This increase in the βGC-A-KO β-cells is significantly lower than that in the l-WT β-cells. Of note, the mean value for max[Ca²⁺]_c for βGC-A-KO cells in Fig. 1E is not 0, which may be a result of spontaneous Ca²⁺ transients occurring sporadically in single cells or small clusters after a change in bath solution from a stimulatory glucose concentration to a threshold concentration for induction of Ca²⁺ oscillations. To emphasize the significance of the data, we also calculated the percentage of cells for each of the three mouse preparations per genotype, which was 42 ± 12% for l-WT cells and 11 ± 5% for β-GC-A-KO cells. Because of the high variability between days and the small and limited number of mice, this data evaluation was not statistically significant (P = 0.08).

K_ATP channel activity was measured with β-cells from WT mice in the cell-attached mode in the presence of 0.5 mmol/L glucose to test whether ANP affects these channels. ANP at a concentration of 10 nmol/L reduced the NP_o from 100% under control conditions to 62 ± 10% (P ≤ 0.01) (Fig. 2A and B). Changes in K_ATP channel activity can be caused by altered mitochondrial metabolism (25). Therefore, the effects of ANP on the mitochondrial ΔΨ were measured, which can be taken to estimate mitochondrial ATP production (26). Neither in the presence of 15 mmol/L glucose nor in the presence of 4 mmol/L glucose did 10 nmol/L ANP alter ΔΨ (Supplementary Fig. 2). These data argue against an influence of ANP on ATP formation. To examine whether the GC-A receptor is involved in the inhibitory effect of ANP, β-cells from βGC-A-KO mice and l-WT mice were used. In l-WT β-cells, NP_o was reduced from 100% to 29 ± 11% (P ≤ 0.01) (Fig. 2C and D) upon addition of 10 nmol/L ANP. In contrast, ANP was without effect in βGC-A-KO β-cells (100% vs. 119 ± 15%) (Fig. 2E and F). In accordance with K_ATP single-channel current measurements, ANP increased the electrical activity of β-cells of WT mice. The fraction of plateau phase (FOPP), which is the percentage of time with spike activity, increased from 47 ± 5% to 65 ± 5% (P ≤ 0.05) (Fig. 3A and B). However, ANP did not change electrical activity in β-cells obtained from SUR1-KO mice (Fig. 3C and D). In these experiments, β-cells did not oscillate because the membrane potential is more depolarized in this genotype. Therefore, data were analyzed by determining the number of action potentials 2 min before change of the bath solution.

The global GC-A-KO leads to reduced expression of the K_ATP channel subunits SUR1 and Kir6.2 in β-cells (3). To elucidate a possible difference in K_ATP current density between the two genotypes, the maximal K_ATP current that developed after formation of the standard whole-cell configuration was measured without ATP in the patch pipette in l-WT (Fig. 4A) and βGC-A-KO β-cells (Fig. 4B). The data revealed no difference in the K_ATP current density (21 ± 2 vs. 23 ± 1 pA/pF in l-WT and βGC-A-KO, respectively) (Fig. 4C).

Because GCs synthesize the second messenger cGMP (27), we tested whether cGMP can mimic the effect of ANP on K_ATP channels. The membrane-permeable analog 8-Br-cGMP reduced NP_o of β-cells from l-WT mice in the cell-attached configuration from 100% to 72 ± 8% (P ≤ 0.05) (Fig. 5A and B). Preincubation of the cells with the PKG inhibitor Rp-8 completely blunted the effect of ANP on K_ATP channel activity of β-cells from WT mice measured in the cell-attached configuration (100% vs. 110 ± 23%) (Fig. 5C and D), indicating the dependence of the effect of ANP on K_ATP channels on this protein kinase. However, ANP still increased insulin secretion in the presence of Rp-8 and 6 mmol/L glucose [0.14 ± 0.03 to 0.19 ± 0.03 ng insulin/(islet ⋅ h); P ≤ 0.05] (Fig. 5E). Besides stimulating the PKG, cGMP can activate or inhibit various types of phosphodiesterases (PDEs) (28). With respect to insulin secretion, PDE3B is the most important PDE in β-cells (29), which is inhibited by cGMP (28). Because inhibition of PDE3B should inhibit the degradation of cAMP, cGMP and cAMP signaling pathways may converge on this PDE. Thus, ANP may affect the cAMP concentration and, consequently, insulin secretion by a PKA-dependent pathway. The specific PDE3B blocker cilostamide markedly increased insulin secretion in the presence of 10 mmol/L glucose, showing that this pathway is present in β-cells (Supplementary Fig. 3A). After inhibition of PDE3B by cilostamide, ANP no longer was able to increase insulin secretion (Supplementary Fig. 3B), pointing to a significant role of cAMP in the ANP effect on insulin secretion. Treatment of the cells with the PKA inhibitor myr-PKI did not suppress the inhibitory effect of ANP on K_ATP channel activity of β-cells from WT mice (85 ± 15% vs. 28 ± 12% with myr-PKI vs. with myr-PKI and ANP, respectively; P ≤ 0.01) (Fig. 5F and G). Insulin secretion was not significantly enhanced by ANP in the presence of myr-PKI, but a tendency was discernible (Fig. 5H). Figure 6 demonstrates the well-known effect that glucagon-like peptide 1 (GLP-1) potentiates GSIS. Of note, Fig. 6 also shows that the action of GLP-1 on secretion is augmented by 90-min preincubation of the cells with ANP [1.7 ± 0.2 vs. 2.3 ± 0.4 ng insulin/(islet ⋅ h) with GLP-1 vs. with GLP-1 and ANP, respectively; P ≤ 0.05]. Ten-minute preincubation with ANP was ineffective [3.2 ± 0.4 vs. 3.2 ± 0.6 ng insulin/(islet ⋅ h); n = 6]; after 20 min preincubation, a tendency to increase the action of GLP-1 appeared [1.7 ± 0.4 vs. 2.0 ± 0.5 ng insulin/(islet ⋅ h); n = 12]. The potentiating action of ANP on the GLP-1 effect was absent in islets from SUR1-KO mice, pointing to a significant role of K_ATP channels in this intensification (Supplementary Fig. 1B). The potentiating effect is most likely a result of inhibition of the PDE3B by cGMP (see above) and increased cAMP concentration. It can be mimicked by cilostamide. Insulin secretion amounted to 1.6 ± 0.2 ng insulin/(islet ⋅ h) with GLP-1 alone versus 3.8 ± 0.2 ng insulin/(islet ⋅ h) with GLP-1 and cilostamide (P ≤ 0.001) (Supplementary Fig. 3A).

An ANP-induced increase in insulin secretion was easier to detect at 6 mmol/L glucose, the threshold concentration for the induction of insulin secretion, than at 10 mmol/L glucose. Identical steady-state incubation experiments with 10 mmol/L glucose did not reveal a significant increase in insulin secretion (data not shown). However, with 10 mmol/L glucose, an effect of ANP could be detected in perifusion experiments. This kind of experiment allows discrimination between the effects of a drug on first and second phase of secretion. The results revealed an increase of GSIS by ANP in the second phase. Ropero et al. (3) showed an augmentation of insulin secretion by ANP in steady-state experiments in the presence of the threshold concentration of 7 mmol/L glucose but did not mention whether other glucose concentrations were tried. ANP seems to be less effective on insulin secretion than on other parameters of the stimulus-secretion coupling, which may be due to insulin secretion being measured with whole islets and, for example, membrane potential or [Ca²⁺]_c with dispersed cells. The capsule of connective tissue that surrounds the islets can restrain access of drugs to islet cells. Therefore, one should keep in mind that in vivo ANP reaches the cells through the capillaries, not across the capsule.

The particular effectiveness of ANP at the threshold glucose concentration also is assessed with [Ca²⁺]_c measurements. ANP considerably increased max[Ca²⁺]_c above the basal values obtained at 6 mmol/L. The special physiological role of ANP in β-cells possibly augments GSIS in a coordinated action, with incretins at blood glucose concentrations occurring at the beginning of a meal. In humans with metabolic disorders, low plasma concentrations of ANP (14–16) may contribute to the impairment of insulin secretion in addition to a reduced incretin effect.

The patch-clamp data clearly demonstrate that ANP-mediated inhibition of K_ATP single-channel activity is caused by activation of GC-A receptors on β-cells because the effect is abolished in βGC-A-KO cells. K_ATP channel current density was the same in both genotypes, which contrasts findings in β-cells with a global GC-A-KO, showing diminished K_ATP channel activity and reduced expression of both K_ATP channel subunits compared with WT cells (3). The findings may explain the higher rate of insulin secretion in islets of the global GC-A receptor KO compared with WT islets, an effect not observed in the β-cell–specific KO model (Fig. 1C). As expected, the inhibitory effect of ANP on K_ATP channel activity was accompanied by a depolarization of the plasma membrane and an increase in [Ca²⁺]_c and insulin secretion attributable to GC-A activation. ANP did not affect membrane potentials of cells from SUR1-KO mice, suggesting that ANP does not influence the activity of other ion channels than K_ATP channels. Stimulation of GC-A receptors should result in an increase in the cGMP concentration. Dankworth (18) showed that the ANP-induced increase in cGMP is much higher in islets of l-WT mice than in those of βGC-A-KO mice. The inhibitory effect of ANP on K_ATP channels was mimicked by a membrane-permeable cGMP analog and completely blocked by the PKG inhibitor Rp-8, strongly suggesting the involvement of the GC-A/cGMP/PKG signaling pathway in the action of ANP on β-cell K_ATP channels (Fig. 7). This pathway also is proposed for the rapid action of estrogen (30) and the effect caused by small amounts of nitric oxide (31) on β-cell function. Which PKG is involved in this pathway in β-cells is unknown, and Rp-8 is not isoform specific. Because two studies exclude the expression of PKGI in β-cells (32,33), PKGII is the most likely candidate. Pancreatic K_ATP channels comprise SUR1 (sulfonylurea receptor) and Kir6.2 (inwardly rectifying K⁺ channel) subunits. PKG is proposed to have dual effects in regulating SUR1/Kir6.2 channels: indirect activation of the channels by phosphorylation of cellular compounds not directly linked to K_ATP channels and inhibition by phosphorylation of channel proteins or tightly coupled proteins (34). In pancreatic β-cells, we, like Ropero et al. (35), found that the latter pathway seems to prevail. After preincubation with Rp-8, ANP still significantly augmented insulin secretion (Fig. 5E), which apparently contrasts with the patch-clamp data. Considering that cGMP inhibits the PDE3B (28), an increase in insulin secretion could be explained by increased cAMP concentration and activation of PKA (see below). The data even suggest that the cGMP/cAMP signaling pathway is more important for the final ANP effects on β-cell function than the cGMP/K_ATP pathway.

ANP can increase the cAMP concentration through GC-A receptor activation, cGMP formation, and inhibition of the PDE3B by reducing cAMP degradation. Inhibition of PKA did not influence the inhibitory effect of ANP on K_ATP channel activity, making a direct link between PKA and K_ATP channels unlikely (Fig. 5F and G). From these patch-clamp experiments, one would expect that ANP still activates insulin secretion when PKA is blocked. In our experiments, there was a tendency but no significant effect (Fig. 5H). However, PKA directly interferes with exocytosis (e.g., by increasing the Ca²⁺ sensitivity of the exocytotic machinery) (36). This effect may be alleviated by PKA inhibition. The experiments with cilostamide support the hypothesis that the increase of the cAMP concentration is an essential step in the action of ANP. In the presence of cilostamide, ANP did not enhance insulin secretion, most likely because PDE3B is fully inhibited and cAMP maximally increased under these conditions. The current data suggest that ANP augments cAMP through cGMP-mediated inhibition of the PDE3B. cAMP activates the PKA, which influences exocytosis, and exchange protein activated by cAMP (Epac), which interferes with K_ATP channels by rendering them more sensitive to ATP (37) (Fig. 7). The cAMP/Epac/K_ATP pathway seems to be indispensable for the action of ANP on β-cell function and to prevail in the cAMP/PKA/exocytosis pathway because ANP did not increase insulin secretion in islets from SUR1-KO mice (Supplementary Fig. 1A). The cGMP/cAMP pathway is mediated by the ANP-induced activation of the GC-A receptor because ANP-induced augmentation of insulin secretion is completely blunted in islets from βGC-A-KO mice (Fig. 1C).

The incretin hormone GLP-1 increases insulin secretion by increasing the cAMP concentration. This effect is potentiated by preincubation with ANP. A significant potentiating effect was seen after 90 min of preincubation with ANP at room temperature. Twenty minutes of preincubation led at least to a tendency to potentiate the action of the incretin hormone. We assume that penetration through the capsule is slow (see above), especially at room temperature. Although we cannot entirely rule out a genomic effect for this potentiation, a cytosolic interaction of the hormones seems much more likely. The ANP/cGMP/PDE3B/cAMP pathway may also be involved in the additive effect of ANP and GLP-1 in β-cells. This assumption is confirmed by the observation that the PDE3B inhibitor cilostamide potentiates the GLP-1 effect on GSIS. Our hypothesis is in accordance with earlier findings. Overexpression of PDE3B in insulin-secreting rat insulinoma cells leads to a decrease of cAMP concentration and GSIS. Furthermore, the ability of GLP-1 to potentiate insulin secretion is impaired (38). Thus, we assume that ANP inhibits PDE3B, which leads to reduced degradation of cAMP and finally increases the effectiveness of GLP-1 (Fig. 7). Because this potentiation requires K_ATP channels, it is most likely mediated by the cAMP/Epac/K_ATP pathway.

In conclusion, the data point to a dual action of ANP in pancreatic β-cells (Fig. 7): 1) ANP activates the GC-A/cGMP/PKG pathway wherein phosphorylation by PKG blocks K_ATP channels and 2) ANP inhibits PDE3B and thus increases cAMP concentration, which positively influences insulin secretion through the PKA and Epac pathway. The second pathway seems to be essential for ANP-mediated enhancement of insulin secretion.

Acknowledgments. The authors thank Isolde Breuning, Institute of Pharmacy, University of Tübingen, for excellent technical support. The authors thank Dr. Michaela Kuhn, Physiological Department, University of Würzburg for kindly providing the knockout mice and corresponding littermates.

Funding. This work was supported by Deutsche Forschungsgemeinschaft grant DR 225/9-1 (to G.D.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.U., J.K., and J.S. researched data. M.D. contributed to the discussion and study design and edited the manuscript. P.K.-D. evaluated data and edited the manuscript. G.D. designed the study, wrote and edited the manuscript, and contributed to the discussion. G.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 51st European Association for the Study of Diabetes Annual Meeting, Stockholm, Sweden, 14–18 September 2015.