Acyl-Ghrelin Influences Pancreatic β-Cell Function by Interference with K_ATP Channels

Ghrelin, released mainly by the X/A-like cells within the oxyntic glands of the stomach, plays a decisive role in the gut-brain axis regulating nutrient intake. The active form of the peptide hormone is acylated ghrelin (or acyl-ghrelin [AG]) (for review, see Müller et al. [1]), although the desacylated ghrelin (or unacylated ghrelin [UAG]) may have physiological functions, too, as we describe later. Plasma AG concentration rises before food intake and rapidly decreases postprandially. AG induces the secretion of appetite stimulators (e.g., agouti-related peptide (AgRP) and neuropeptide Y [NPY]), which are expressed in the feeding center of the hypothalamus (2). Thus, AG acts as a hunger-inducing hormone that regulates food intake and energy expenditure. Of note, in animals, including humans, plasma AG concentration strongly depends on nutrition status (3). Numerous other physiological functions of AG have been discovered (e.g., in the cardiovascular and gastrointestinal system). AG affects learning and memory and is important in the regulation of glucose metabolism (1,4). Ghrelin is also produced in the endocrine pancreas in ε-cells of the islets (5), thus a paracrine role of AG is conceivable. Pathophysiologically, altered AG concentrations, or rather changes in the UAG-to-AG ratio (6), may be relevant for obesity and type 2 diabetes mellitus.

The expression pattern of the AG receptor, the growth hormone secretagogue receptor 1a (GHSR1a), is still a matter of debate (i.e., it remains unclear by which receptors AG exerts its distinct effects in different tissues outside the hypothalamus). Dezaki et al. (7) reported that AG diminished glucose-stimulated insulin secretion (GSIS) in rodents via binding to GHSR1a in β-cells. Canonically, this receptor is coupled to the G_αq signaling pathway (for review, see Yin et al. [8]). Authors of some studies suggested that GHSR1a in β-cells display a unique coupling to the G_αi/cAMP pathway with subsequent modulation of ion-channel activities including inhibition of TRPM2 and opening of K_v2.1 channels (9,10). In contrast, Park et al. (11) concluded that the GHSR1a expressed in β-cells heterodimerizes with the somatostatin (SST) receptor (SSTR5) and establishes a G_αi- or G_αq-coupled signaling pathway, depending on the energy balance or, more precisely, the ratio of AG to SST. Recently, DiGruccio et al. (12) and Adriaenssens et al. (13) postulated that the GHSR1a is not expressed in β-cells but, rather, in δ-cells. According to their hypothesis, AG induces the secretion of SST via binding to the canonically G_αq-coupled GHSR1a. Subsequently, SST binds to SST receptors (SSTRs) in β-cells and inhibits GSIS. According to this model AG modulates β-cell function via a paracrine action mediated by SST. It is also demonstrated that AG decreases GSIS in healthy volunteers (14), pointing to a significant function of this hormone in metabolic regulation in humans as well, although the underlying mechanisms are even less clear than in non-human animals.

UAG is the degradation product of AG originating from desoctanoylation in the plasma (15). The possible UAG receptor is still a matter of debate. It became evident that UAG induces various effects in different tissues (e.g., a reduction in fat mass, increased vascular remodeling, and a decrease of cytokine-induced apoptosis of islet cells; for review, see Delhanty et al. [16]). Studies also revealed that the plasma UAG concentration is decreased in obese patients compared with normal-weight participants, whereas the plasma AG concentration remains more or less unchanged, shifting the UAG-to-AG ratio to lower values (6,17). Moreover, UAG can antagonize orexigenic properties of AG, as well as its diabetogenic effects and detrimental action on insulin sensitivity (16,18,19).

Taken together, the mode of action of AG remains contradictory and far from being completely understood. In this study, we addressed these conflicts and aimed to clarify how AG influences β-cell function, as well as possible antagonistic effects of UAG and the importance of GHSR1a in AG action.

C57Bl/6N wild-type (WT), SUR1-knockout (KO), and exchange protein activated directly by cAMP 2 (Epac2)-KO mice (global KOs) with C57Bl/6N background were bred in the animal facility of the Department of Pharmacology at the University of Tübingen. The principles of laboratory animal care, revised by the US National Institutes of Health in 1985, were followed (20), and the experiments were conducted according to German laws. Isolation and culture of islets were performed as previously described (21).

Recordings of cytoplasmic calcium concentration ([Ca²⁺]_c) and measurements of ATP-sensitive K⁺ (K_ATP) current and membrane potential (V_m) in the perforated-patch configuration were performed with a bath solution that contained (in mmol/L): 140 NaCl, 5 KCl, 1.2 MgCl₂, 2.5 CaCl₂, glucose as indicated, and 10 HEPES, at pH 7.4, adjusted with NaOH. Krebs-Ringer HEPES solution for insulin secretion was composed of (in mmol/L): 120 NaCl, 4.7 KCl, 1.1 MgCl₂, 2.5 CaCl₂, glucose as indicated, 10 HEPES, and 0.5% BSA, at pH 7.4, adjusted with NaOH. In experiments with 60 mmol/L KCl, the NaCl concentration was reduced appropriately. For recording of K_ATP current and V_m in the perforated-patch mode, pipette solution consisted of (in mmol/L): 10 KCl, 10 NaCl, 70 K₂SO₄, 4 MgCl₂, 2 CaCl₂, 10 EGTA, 20 HEPES, and amphotericin B (250 µg/mL), at pH 7.15, adjusted with KOH. For the patch-clamp recordings in the inside-out configuration, ATP-containing and ATP-free bath solution was composed of (in mmol/L): 130 KCl, 4.6 CaCl₂, 10 EDTA, and 20 HEPES, with pH adjusted to 7.2 with KOH. Pipette solution for the inside-out configuration contained (in mmol/L) 130 KCl, 1.2 MgCl₂, 2 CaCl₂, 10 EGTA, and 10 HEPES; pH was adjusted to 7.4 with KOH.

Islets and islet cell clusters were cultured in RPMI 1640 medium (11.1 mmol/L glucose) enriched with 10% FCS and 1% penicillin/streptomycin. MIN6 cells were cultured in DMEM (25 mmol/L glucose) containing 10% FCS and 1% penicillin/streptomycin. AG (human) was obtained from Biomol (Hamburg, Germany). UAG (human) and the ghrelin receptor inverse agonist K-(D1-NaI)-FwLL-NH₂ were purchased from Tocris Bioscience (Wiesbaden-Nordstadt, Germany); the ghrelin receptor antagonist [D-Lys³]-GHRP-6 and the SSTR2–5 antagonist H6056 were from Bachem (Bubendorf, Switzerland). Fura-2 acetoxymethylester (Fura-2-AM) was ordered from Biotrend (Köln, Germany). RPMI 1640 medium, FCS, penicillin/streptomycin, and trypsin were from Invitrogen (Karlsruhe, Germany). All other chemicals were purchased from Sigma-Aldrich (Taufkirchen, Germany) or Carl Roth (Karlsruhe, Germany) in the purest form available.

The [Ca²⁺]_c was measured in β-cell clusters at 37°C after incubation with 5 μmol/L Fura-2-AM for 35 min. Fura-2-AM was excited alternately at 340 nm or 380 nm and emission filtered (LP515). The [Ca²⁺]_c was expressed as the ratio of the emitted light at 340 nm to the emitted light at the 380 nm excitation wavelength. Mean [Ca²⁺]_c and the frequency of [Ca²⁺]_c oscillations were calculated for 15 min before solution change and at the end of the experiment, respectively, if not indicated otherwise.

Details for insulin secretion in steady-state incubations have been described previously (22). Isolated islets from WT mice were kept overnight in RPMI 1640 culture medium with 11.1 mmol/L glucose. MIN6 were seeded at a density of 2 × 10⁵ cells/well. The culture medium was supplemented with 0.65 mmol/L β-mercaptoethanol. Insulin secretion in steady-state incubation was performed 2 days after being plated. MIN6 cells were washed and incubated with Krebs-Ringer HEPES solution containing 0.5 mmol/L glucose for 30 min at room temperature. Afterward, the insulin secretion of MIN6 was made as described in the previous sentence. Levels of insulin secretion and content were determined by radioimmunoassay (Merck Millipore, Darmstadt, Germany).

K_ATP whole-cell currents, single channel activity, and V_m were recorded with an EPC-9 patch-clamp amplifier using Patchmaster software (HEKA, Lambrecht, Germany). For measurements of V_m, patch-clamp recordings were performed in the perforated-patch configuration in the current clamp mode at a holding current of 0 mA. For determination of V_m and spike frequency, average plateau potential and action potentials (APs) were evaluated 1 min before solution change. For determination of the K_ATP whole-cell current, 300-ms pulses were performed every 15 s from the holding potential at −70 to −60 and −80 mV. The amplitude of currents elicited by voltage steps from the holding potential to −60 mV was taken for evaluation. Data of the last three pulses in each interval were averaged. Single-channel activity was measured in the inside-out patch configuration at a holding potential of −50 mV. For determination of mean current density, the current during the last 30 s before solution change was evaluated.

Means ± SEM are given for the indicated number of experiments. For paired values, statistical differences were assessed by Student t test. ANOVA followed by Student-Newman-Keuls test was conducted for multiple comparisons. P ≤0.05 was considered statistically significant.

The data sets generated and analyzed during this study are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

To evaluate the effects of AG on stimulus-secretion coupling (SSC), [Ca²⁺]_c and V_m were determined. AG (10 nmol/L) diminished typical glucose-stimulated oscillations of [Ca²⁺]_c at 10 mmol/L glucose in WT β-cells (Fig. 1A). The mean [Ca²⁺] determined as the Fura-2-AM ratio (340 nm to 380 nm) was decreased from 0.63 ± 0.03 under control conditions to 0.58 ± 0.03 in the presence of AG (Fig. 1B, left). The mean [Ca²⁺] at the nadir between the Ca²⁺ peaks amounted to 0.4 ± 0.2 in this series of experiments. The frequency of the [Ca²⁺]_c oscillations/15 min was diminished from 2.7 ± 0.2 under control conditions to 1.7 ± 0.2 in the presence of AG (Fig. 1B, right). This was caused by the effect of AG on V_m (Fig. 1C). AG hyperpolarized V_m from −37.8 ± 2.1 mV in the presence of 10 mmol/L glucose to −53.0 ± 2.6 mV with AG (Fig. 1D, left) and lowered AP frequency from 40.0 ± 8.0 APs/min at 10 mmol/L glucose to 6.4 ± 11.4 APs/min with AG (Fig. 1D, right). Consequently, AG attenuated steady-state GSIS at a suprathreshold glucose concentration [10 mmol/L glucose: 3.9 ± 0.8 ng insulin/(islet∗h) vs. 10 mmol/L glucose with AG: 3.0 ± 0.7 ng insulin/(islet∗h)] (Fig. 1E). AG had no effect at a low or threshold glucose concentration (3 and 6 mmol/L glucose, respectively) and tended to diminish secretion at 8 mmol/L glucose, a concentration directly above the threshold. AG also reduced GSIS in the insulin-secreting cell line MIN6 (Fig. 1F). Overall, AG attenuated important parameters of SSC and thus exerted insulinostatic effects.

Contrary to the results of AG with WT β-cells, AG did not alter [Ca²⁺]_c in β cells isolated from SUR1-KO mice lacking functional K_ATP channels (23) (Fig. 2A and B). Determination of V_m revealed that AG even slightly depolarized V_m from −37.0 ± 1.8 mV in the presence of 10 mmol/L glucose to −34.6 ± 1.9 mV after addition of AG (Fig. 2C and D). AG had no influence on steady-state GSIS in whole islets of SUR1-KO mice [2.0 ± 0.4 ng insulin/(islet∗h) at 10 mmol/L glucose vs. 2.0 ± 0.3 ng insulin/(islet∗h) with AG] (Fig. 2E). These results point to a K_ATP channel–mediated pathway of AG.

To elucidate if AG directly targets K_ATP channels, recordings with excised patches in the inside-out patch configuration were made (Fig. 3A and B). As expected, K_ATP channels were closed by ATP (0.1–1 mmol/L). The addition of AG in the presence of ATP did not open K_ATP channels and thus did not affect single-channel activity (24.1 ± 8.2 pA without ATP vs. 1.1 ± 0.3 pA with ATP vs. 2.7 ± 1.1 pA with ATP and AG), which renders a direct interaction of AG with K_ATP channels unlikely. To evaluate, whether AG modulates K_ATP-channel activity indirectly (e.g., via second messenger pathways), we measured the current in the perforated-patch configuration. AG increased the current from 9.2 ± 1.9 pA under control conditions to 165.0 ± 24.1 pA with AG (Fig. 3C and D). The data indicate AG requires intact β-cell metabolism to be effective.

UAG (20 nmol/L) alone increased mean [Ca²⁺]_c from 0.69 ± 0.03 to 0.75 ± 0.03 and prevented the AG-induced decrease of [Ca²⁺]_c (Fig. 4A and B). UAG had no impact on cell V_m. The addition of AG in the presence of UAG did not alter V_m (−44.5 ± 2.5 mV at 10 mmol/L glucose vs. −45.5 ± 2.3 mV with UAG vs. −45.3 ± 2.1 mV with UAG + AG) (Fig. 4C and D).

UAG (10 and 20 nmol/L) tended to increase GSIS in whole islets, but this effect did not reach significance [1.8 ± 0.2 ng insulin/(islet∗h) at 10 mmol/L glucose vs. 2.1 ± 0.3 ng insulin/(islet∗h) in the presence of 10 nmol/L UAG and 2.2 ± 0.4 ng insulin/(islet∗h) with 20 nmol/L UAG, respectively] (Fig. 4E). The addition of AG in the presence of UAG did not affect GSIS (Fig. 4F) 2.6 ± 0.3 ng insulin/(islet∗h) at 10 mmol/L glucose and 2.3 ± 0.3 ng insulin/(islet∗h) with UAG and AG], whereas AG alone elicited the described inhibitory action. Taken together, the data point to an AG-counteracting effect of UAG in β-cells.

The receptor of AG, GHSR1a, is not only expressed in the arcuate nucleus of the hypothalamus (24) but also in the pancreas (25); however, the data about the cellular localization in the pancreas are conflicting. We used the peptidyl antagonist [D-Lys³]-GHRP-6 to evaluate whether the effect of AG is mediated via GHSR1a in β-cells. Measurements of [Ca²⁺]_c revealed that the [Ca²⁺]_c-diminishing effect of AG was not averted in all experiments by [D-Lys³]-GHRP-6 in β-cells of WT mice (Fig. 5A, compare upper and lower panel). In the presence of the antagonist, 10 cell clusters responded to AG, and 11 were nonresponsive. The average of all experiments revealed that AG still exerted a significant effect after GHSR1a blockade. Measurements showed [D-Lys³]-GHRP-6 alone increased mean [Ca²⁺]_c from 0.64 ± 0.02 in the presence of 10 mmol/L glucose to 0.69 ± 0.02, whereas the subsequent addition of AG decreased the mean [Ca²⁺]_c of all experiments to 0.65 ± 0.02 (Fig. 5B). This dual effect was also evident in cell-V_m measurements (Fig. 5C): in 5 of 10 recordings, AG still hyperpolarized the V_m and decreased AP frequency in the presence of the GHSR1a antagonist, whereas in the other five cells, the effect was blocked (Fig. 5C, compare upper and lower panel).

In the summary of all experiments under this condition, [D-Lys³]-GHRP-6 alone had no effect on V_m, but AG still elicited a significant hyperpolarization (−38.1 ± 1.6 mV at 10 mmol/L glucose vs. −38.0 ± 1.6 mV with [D-Lys³]-GHRP-6 vs. −48.2 ± 3.5 mV with [D-Lys³]-GHRP-6 + AG) (Fig. 5D). These results were reproducible with the non-peptidyl GHSR1a antagonist YIL 781 (n = 8; data not shown) excluding a specific [D-Lys³]-GHRP-6 phenomenon.

GHSR1a exhibits a high constitutive activity, which is approximately 50% of the activity achieved by AG binding to the GHSR1a (26). The inverse agonist K-(D1-NaI)-FwLL-NH₂ stabilizes the inactive conformation of the GHSR1a but still allows AG to bind to the receptor (26,27). Patch-clamp experiments revealed that K-(D1-NaI)-FwLL-NH₂ alone had no influence on V_m (n = 6; data not shown). The combination of the inverse agonist and the peptidyl antagonist finally averted the effect of AG on V_m in every recording (−40.6 mV ± 2.7 at 10 mmol/L glucose vs. −39.0 mV ± 3.3 with K-(D1-NaI)-FwLL-NH₂ and [D-Lys3]-GHRP-6 vs. −40.3 ± 3.0 mV with K-(D1-NaI)-FwLL-NH₂, [D-Lys3]-GHRP-6 and AG). One typical recording of this series is shown in Fig. 5E and the summary of all data of this series is shown in Fig. 5F.

These results were reflected by the GSIS. As expected, AG attenuated GSIS in the presence of 10 mmol/L glucose, and this reduction was only partially inhibited by the GHSR1a antagonist [D-Lys³]-GHRP-6 (Fig. 5G). A complete inhibition of the effect on insulin secretion was achieved by the combination of the inverse agonist K-(D1-NaI)-FwLL-NH₂ and the GHSR1a antagonist [D-Lys³]-GHRP-6 (Fig. 5G). [D-Lys³]-GHRP-6 alone did not affect insulin secretion [2.5 ± 0.3 ng insulin/(islet∗h) in the presence of 10 mmol/L glucose vs. 2.2 ± 0.2 ng insulin/(islet∗h) with [D-Lys³]-GHRP-6; n = 23; P = NS]. In contrast, the inverse agonist K-(D1-NaI)-FwLL-NH₂ reduced the GSIS from 2.7 ± 0.5 ng insulin/(islet∗h) in the presence of 10 mmol/L glucose to 2.1 ± 0.4 ng insulin/(islet∗h) (Fig. 6A).

As shown in Fig. 6B, K-(D1-NaI)-FwLL-NH₂ completely lost its inhibitory effect when the cells were depolarized by 60 mmol/L K⁺ [5.2 ± 0.3 ng insulin/(islet∗h) before addition of the inverse agonist vs. 5.0 ± 0.3 ng insulin/(islet∗h) after application of K-(D1-NaI)-FwLL-NH₂]. This points to an interference of the inverse agonist with the amplifying pathway (28). An influence on the amplifying pathway also agrees with the observation that K-(D1-NaI)-FwLL-NH₂ affected GSIS but not V_m, because the latter parameter predominantly reflects changes in the triggering pathway. The results with 3-isobutyl-1-methylxanthine (IBMX) demonstrate that the K⁺ depolarization did not bring the amplifying pathway to its limits, thereby masking possible changes, and that even a further increase in GSIS via cAMP is possible. Figure 6C reveals that AG administered in the presence of the inverse agonist enhanced GSIS from 2.1 ± 0.4 to 2.7 ± 0.5 ng insulin/(islet∗h) (Fig. 6C, columns 3 and 4), in contrast to the normally observed inhibitory effect of AG (Fig. 6C, columns 1 and 2). This unexpected observation provides evidence that AG can exert stimulatory effects under certain conditions, possibly via the canonical coupling of the GHSR1a to the G_q pathway. It is tempting to speculate that K-(D1-NaI)-FwLL-NH₂ not only inhibits the intrinsic activity of the receptor but concurrently the inhibitory pathway of AG that prevails in β-cells under physiological conditions.

According to the hypothesis of DiGruccio et al. (12) and Adriaenssens et al. (13), AG does not act on β-cells but, rather, on δ-cells, and SST mediates the AG effect by its paracrine influence on β-cells. Although our observation that K_ATP channels in β-cells are critical for the action of AG does not support this theory, we repeated the [Ca²⁺]_c experiments shown in Fig. 2A and B with SST (10 nmol/L) in β-cells of SUR1-KO mice (Fig. 7). In glucose-stimulated β-cells of SUR1-KO mice, [Ca²⁺]_c was either at a plateau or oscillated (Fig. 7A). The upper trace of Fig. 7A shows a spontaneous transition from oscillations to a plateau, which occasionally occurs in β-cells from SUR1-KO mice (29). In β-cell clusters, which exhibited a plateau phase, the addition of SST interrupted the plateau phase and showed ongoing oscillations of [Ca²⁺]_c (Fig. 7A, upper panel). In β-cell clusters, which displayed oscillatory activity, SST led at least to a transient inhibition of the oscillations (Fig. 7A, lower panel). Overall, SST decreased mean [Ca²⁺]_c from 0.47 ± 0.03 at 10 mmol/L glucose to 0.39 ± 0.03 with SST in the first 5 min and to 0.42 ± 0.04 in the 5–20 min interval (Fig. 7B). This observation points to different mode of action of AG and SST in β-cells.

H6056 is a specific antagonist for SSTR2 but can act as antagonist on SSTR2–5 when used in a higher concentration range, as in the following experiment. Unexpectedly, the inhibiting effect of AG on GSIS was abolished in the presence of H6056 (Fig. 7C). The findings make the participation of SST as the sole mode of action of AG unlikely but raise the question of a possible involvement of the SSTRs.

The GHSR1a is a G-protein coupled receptor, canonically linked to G_αq. The binding of AG to GHSR1a activates phospholipase C and, therefore, increases [Ca²⁺]_c (for review, see Yin et al. [8]), which is in contrast to the observed reduction of insulin secretion. Dezaki et al. (9) and Kurashina et al. (10) presented evidence for the coupling of the GHSR1a to G_αi in β-cells and, thus, cAMP as a downstream target of AG. To evaluate cAMP-dependent pathways, we induced insulin secretion in the presence of the membrane-permeable cAMP analog dibutyryl cyclic–cAMP (db-cAMP), as well as with the phosphodiesterase inhibitor IBMX in isolated islets of WT mice. AG no longer decreased GSIS when a cAMP reduction was counteracted by excess cAMP due to addition of db-cAMP or IBMX (Fig. 8A and B). cAMP could influence β-cell function via a cAMP-dependent protein kinase (PKA)-dependent pathway or a PKA-independent pathway (i.e., exchange protein activated directly by cAMP [Epac]), whereby Epac2 represents the most abundant form in the pancreas (30). AG still diminished GSIS in isolated islets of Epac2-KO mice (Fig. 8C).

The [Ca²⁺]_c and V_m experiments performed with cells from SUR1-KO mice and K_ATP current measurements with cells of WT mice demonstrate, to our knowledge for the first time, that AG mediates its effect via K_ATP channels. AG was not able to modulate K_ATP single-channel activity in excised patches, making a direct interaction of AG with K_ATP channels unlikely. However, AG modulated K_ATP currents in cells with intact metabolism and second messenger pathways. Measurements of GSIS in the presence of the phosphodiesterase inhibitor IBMX or the cAMP-analog db-cAMP suggest AG interferes with the amplifying cAMP pathway. This is supported by Dezaki et al. (9), who postulated a PTX-sensitive pathway. They observed a decrease in glucose-induced cAMP production with 10 nmol/L AG in MIN6 cells as well as in isolated rat islets. PKA and Epac are downstream targets of cAMP, both present in β-cells. Our results suggest the interference of AG is primarily PKA dependent because AG attenuated GSIS in isolated islets of Epac2-KO mice. It is well established that increased PKA activity due to activation of the adenylate cyclase and increased cAMP concentration closes K_ATP channels (31). Evidently, AG negatively influences this pathway in β-cells.

Dezaki et al. (7,32) averted the AG-evoked diminution of [Ca²⁺]_c and decrease of blood glucose concentration by the receptor antagonist [D-Lys³]-GHRP-6. Our V_m and [Ca²⁺]_c measurements and determination of GSIS revealed that the effect of AG cannot be completely blocked in the presence of a nonpeptidyl or peptidyl GHSR1a antagonist alone (Fig. 5A–D and G). Our data clearly show that a complete block of the AG effects on different parameters of SSC can only be achieved by suppressing the basal activity of the GHSR1a together with a receptor antagonist. The data prove AG mediates its effects by binding to GHSR1a. The surprising finding that an antagonist and an inverse agonist must be combined seems not to be due to a too-low concentration of the antagonist. In measurements of [Ca²⁺]_c, we increased the concentration of [D-Lys³]-GHRP-6 from 1 to 2 μmol/L and still observed an effect. The Fura-2-AM ratio (340 nm to 380 nm) amounted to 0.66 ± 0.03 in the presence of 10 mmol/L glucose versus 0.62 ± 0.04 after addition of 10 nmol/L AG (P ≤ 0.05; n = 6; data not shown). Moreover, the effect of [D-Lys³]-GHRP-6 on V_m was reproducible with the structurally unrelated antagonist YIL781. One possible explanation for the requirement of both substances may be that as long as the receptor exhibits constitutive activity, AG can activate the receptor—at least in a subpopulation of cells—by a second binding site that is not blocked by the antagonist.

Our data on the SSTR antagonist H6056 point to an involvement of SSTRs in the action of AG. The data are conflicting in regard to the subtypes of SSTR expressed in murine β-cells. In some reports, SSTR3 and SSTR5 have been discussed as dominant subtypes in mouse β-cells (12,13,33). DiGruccio et al. (12) and Adriaenssens et al. (13) reported that SSTR5 expression is low in β-cells in contrast to SSTR3, whereas Ludvigsen et al. (34) found a high expression of SSTR5 in mouse β-cells. Of note, studies with SSTR5-KO mice revealed the importance of this receptor in mouse β-cells. Wang et al. (35) demonstrated that β-cell–specific SSTR5 ablation provoked glucose intolerance and lack of adequate insulin response to glucose stimulation. Another study showed that SST was less potent in inhibiting insulin secretion in islets from SSTR5-KO mice compared with WT islets, suggesting a pivotal role of the receptor in the inhibiting effect of SST on GSIS (33).

Heteromerization of GHSR1a has been shown with a couple of other receptors (36). Park et al. (11) provided an elegant model of heteromerization of the canonically G_αq coupled GHSR1a and the canonically G_αi coupled SSTR5 in β-cells. According to their results, receptor heteromerization depends on the AG-to-SST ratio. A high AG-to-SST ratio forces heteromerization of GHSR1a and SSTR5 and induced G_αi coupling, thus suppressing GSIS and cAMP formation. A low ratio destabilizes the complex. This model explains the involvement of both receptors in the action of AG on β-cells and also elucidates how GHSR1a activation can result in inhibition of GSIS. Our data perfectly support this model, providing at least one plausible explanation for our observations.

According to the hypothesis of DiGruccio et al. (12) and Adriaenssens et al. (13), circulating AG binds to GHSR1a in δ-cells, promoting SST secretion, which binds to SSTR3 in β-cells. Our data revealed that SST, but not AG, affects [Ca²⁺]_c in β-cells isolated from SUR1-KO mice, which refutes an indirect effect of AG mediated by SST released from δ-cells as the only explanation. Evidently, the effect of SST on β-cells, in contrast to that of AG, is not mediated by K_ATP channels. The results of our [Ca²⁺]_c experiments in SUR1-KO mice are in agreement with studies by Kailey et al. (37) and Abel et al. (38), who showed that SST still hyperpolarized V_m and abolished AP firing in human β-cells and INS-1 cells in the presence of the K_ATP-channel inhibitor tolbutamide.

We cannot rule out that other endocrine cells, besides β-cells, remain in the cell clusters after trypsin treatment of isolated islets. Evidently, β-cells represent the dominant cell population in islets of Langerhans of different species, including humans and mice, whereas δ-cells are only a minority (39). In mouse islets, the area stained for SST varies between 3.37% and 5.71%, depending on islet size (40). Taking this into account, it seems unlikely that δ-cells were present in every cell cluster under investigation, but the effects of AG were consistently observable without exception.

Dezaki et al. (9) and our group have shown that AG lowers the cAMP concentration and inhibits insulin secretion, respectively, in MIN6 cells. These findings, together with the different effects of AG and SST on [Ca²⁺]_c in β-cells of SUR1-KO mice, favor a β-cell–specific interaction of AG.

Taken together, our data do not support the hypothesis that SST primarily contributes to the effect of AG, but they do not exclude that GHSR1a is expressed in δ-cells as well, and establish the signaling cascade postulated by DiGruccio et al. (12) and Adriaenssens et al. (13).

Degradation of AG to UAG by carboxypeptidase and butyrylcholinesterase occurs in the plasma (15), so it is likely that AG is the main form that influences β-cell function in the endocrine pancreas. UAG alone did not exhibit significant effects on SSC but antagonized the influence of AG on [Ca²⁺]_c, V_m, and GSIS (Fig. 4A–D and F). So far, no specific UAG receptor is known, but there is evidence that the UAG-mediated enhancement of insulin sensitivity and decrease in glucose output by primary hepatocytes are independent of GHSR1a (41). It was questioned in some studies whether UAG acts as an antagonist at the GHSR1a (42,43), yet it was shown that UAG can operate as an agonist on GHSR1a (42). According to Gauna et al. (42), binding of UAG to the GHSR1a seems unlikely in a low concentration range but was observed at higher concentrations. Our own observation clearly indicates an antagonism between AG and UAG, although it remains unclear at which step of the signaling pathway this antagonism occurs.

Evidently, the UAG-to-AG ratio plays an important role concerning insulin resistance and hyperglycemia (for review, see Delhanty et al. [16]). It is suggested that obesity and T2DM result in a relative UAG deficiency that decreases the UAG-to-AG ratio and favors the aforementioned pathophysiological conditions (6,16,17). These findings led to first therapeutic approaches: AZP-531 (livoletide) was the first UAG analog tested in a clinical phase IIb/III study on food-related behavior in patients with Prader-Willi syndrome, a disease characterized by hyperphagia (44). In 2016, Allas et al. (45) conducted a phase I/II study with AZP-531 in obese participants and participants with T2DM, as well, which revealed that AZP-531 showed a tendency to be beneficial in participants with impaired glucose tolerance.

Moreover, antagonism at the GHSR1a led to some promising results in animal models of obesity-related metabolic diseases. The nonpeptidyl GHSR1a antagonist YIL781 and the more central nervous system–permeable YIL780 both improved glucose tolerance and increased insulin secretion in mice with diet-induced obesity and lean rats (46). In a mouse model of postmenopausal obesity, the peptidyl GHSR1a antagonist [D-Lys³]-GHRP-6 displayed anorexigenic effects and reduced blood glucose levels during acute and longer treatment (47). However, detrimental findings were reported by Mosa et al. (48) in their long-term study, in which [D-Lys³]-GHRP-6 increased blood glucose levels, decreased plasma insulin levels, and worsened glucose tolerance as well as insulin tolerance in nonobese, diabetic MKR mice. Beneficial results were observed with the use of two novel inverse agonists of GHSR1a. Injection of the inverse agonists reduced food intake and visceral adiposity, and improved glucose tolerance in ZDF rats and mice with diet-induced obesity (49). Overall, AG antagonism seems to improve glycemic control even in disease conditions with unchanged AG plasma concentration.

In summary, to our knowledge, this is the first study showing the involvement of K_ATP channels in the action of AG on SSC in β-cells. The effect is indirect, probably mediated by reduced PKA activity. Our data favor a direct action of AG on β-cells but do not exclude additional indirect pathways proposed by other studies.

Acknowledgments. The authors acknowledge the excellent and skillful technical assistance of Isolde Breuning, Department of Pharmacology, Institute of Pharmacy, University of Tübingen, Tübingen, Germany.

Funding. This work was supported by a grant from Deutsche Forschungsgemeinschaft (DFG) (DR225/11-1 to G.D.).

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

Author Contributions. J.K. researched data and wrote and edited the manuscript. P.K.-D. contributed to discussion and study design and edited the manuscript. G.D. designed the study, wrote and edited the manuscript, contributed to discussion, and is the guarantor of this study.