Adrenaline Stimulates Glucagon Secretion by Tpc2-Dependent Ca²⁺ Mobilization From Acidic Stores in Pancreatic α-Cells

The ability of the “fight-or-flight” hormone adrenaline to increase plasma glucose levels by stimulating liver gluconeogenesis is in part mediated by glucagon, the body’s principal hyperglycemic hormone (1). Glucagon is secreted by the α-cells of the pancreas (2). Reduced autonomic stimulation of glucagon secretion may result in hypoglycemia, a serious and potentially fatal complication of diabetes (3). It has been estimated that up to 10% of insulin-treated patients die of hypoglycemia (4). Understanding the mechanism by which adrenaline stimulates glucagon secretion and how it becomes perturbed in patients with diabetes is therefore essential.

The mechanism by which adrenaline stimulates glucagon secretion has only partially been elucidated. It is clear, however, that it involves activation of β-adrenergic receptors (5–7), leading to elevated intracellular cAMP ([cAMP]_i), and culminates in elevation of the cytoplasmic free Ca²⁺ concentration ([Ca²⁺]_i) with resultant Ca²⁺-dependent stimulation of glucagon secretion (6,8,9). However, our understanding of the spatiotemporal interrelationship between different intracellular Ca²⁺ stores involved in α-cell adrenaline signaling remains patchy.

Here, we have explored how adrenaline triggers glucagon release by a combination of hormone secretion measurements, electrophysiology, and imaging of cytoplasmic Ca²⁺, cAMP, and PKA activity in α-cells within intact mouse pancreatic islets. We have extended the measurements to human islets and also tested how the responsiveness to adrenaline is affected when islets are cultured under hyperglycemic conditions to establish whether and how this regulation becomes impaired in diabetes. Our data indicate that Ca²⁺ release from acidic (lysosomal) stores plays a critical and unexpected role in adrenaline-induced glucagon secretion.

The following substances were used (source given in parentheses): the L-type Ca²⁺ channel blocker isradipine and the P/Q Ca²⁺ channel blocker ω-agatoxin (Alomone Laboratories, Jerusalem, Israel); the α₁-antagonist prazosin (Abcam, Cambridge, U.K.); the membrane-permeable PKA inhibitor myr-PKI, the inositol triphosphate receptor inhibitor Xestospongin C, the nicotinic acid adenine dinucleotide phosphate (NAADP) antagonist Ned-19, the V-ATPase inhibitor bafilomycin, the sarco/endoplasmic reticulum (sER) ATPase inhibitor thapsigargin, and noradrenaline (Tocris Bioscience, Bristol, U.K.); the EPAC2 inhibitor ESI-05 (BioLog, Bremen, Germany); and the insulin receptor antagonist (S961; Novo Nordisk, Måløv, Denmark). All other compounds were obtained from Sigma-Aldrich (Dorset, U.K.).

The extracellular solution (EC1) used for imaging and current-clamp electrophysiology (Fig. 2C) experiments contained (in millimoles per liter ) 140 NaCl, 4.6 KCl, 2.6 CaCl₂, 1.2 MgCl₂, 1 NaH₂PO₄, 5 NaHCO₃, and 10 HEPES (pH 7.4 with NaOH). The pipette solution (IC1) consisted of 76 K₂SO₄, 10 NaCl, 10 KCl, 1 MgCl₂, and 5 HEPES (pH 7.35 with KOH). The extracellular solution (EC2) for recordings of the depolarization-induced exocytosis (Fig. 3C and D) contained 118 NaCl, 5.6 KCl, 2.6 CaCl₂, 1.2 MgCl₂, 5 HEPES, 20 TEA-Cl, and 6 glucose (pH 7.4 with NaOH). In the latter experiments, the pipette-filling medium (IC2) consisted of 76 Cs₂SO₄, 10 NaCl, 10 KCl, 1 MgCl₂, and 5 HEPES (pH 7.35 with CsOH).

NMRI mice (Charles River) were used throughout the study except for the experiments involving the genetic manipulation of TPC1 and -2 (Fig. 5). Tpcn1^−/− and Tpcn2^−/− mice were developed on C57Bl/6N background as previously described (10). Cd38^−/− mice were developed on C57Bl/6J background (11). Age- and sex-matched C57Bl/6N and C57Bl/6J wild-type animals were used as controls. Mice were kept in a conventional vivarium with a 12-h-dark/12-h-light cycle with free access to food and water and were killed by cervical dislocation. All experiments were conducted in accordance with the United Kingdom Animals (Scientific Procedures) Act (1986) and the University of Oxford ethics guidelines.

Pancreatic islets were isolated from mice by injecting collagenase solution into the bile duct, with subsequent digestion of the connective and exocrine pancreatic tissue.

Human pancreatic islets were isolated at the Oxford Diabetes Research & Wellness Foundation Human Islet Isolation Facility according to published protocols (12,13).

Unless otherwise stated, experiments were performed on acutely isolated islets. For experiments aiming to emulate chronic hyperglycemia, islets were cultured for 48 h in RPMI medium containing 30 mmol/L or the standard 11 mmol/L glucose, supplemented with 10% FBS, 100 IU/mL penicillin, and 100 μg/mL streptomycin (all reagents from Life Technologies, Paisley, U.K.). Recombinant sensors were delivered via adenoviral (GCaMP6f, AKAR3) or BacMam (cADDis) vectors at 10⁵ infectious units per islet, followed by 24–36 h culturing at 11 mmol/L glucose (as described above) to express the sensors.

Time-lapse imaging of [Ca²⁺]_i in intact freshly isolated mouse islets was performed on an inverted Zeiss AxioVert 200 microscope equipped with the Zeiss 510-META laser confocal scanning system. Prior to Ca²⁺ imaging, mouse islets were loaded with 6 μmol/L Fluo-4 at room temperature or 10 μmol/L Fluo4FF (Molecular Probes) at 37°C for 90 min. Both dyes were excited at 488 nm and emission was collected at 530 nm, using the 512 × 512 frame scanning mode with the pixel dwell time of 6 μs (image frequency: 0.25 Hz). Time-lapse imaging of [Ca²⁺]_i in groups of mouse islets was done on an Axiozoom.V16 microscope (0.1 Hz) using GCaMP6f (14).

PKA activity was reported in pancreatic islet cells using recombinant fluorescence resonance energy transfer (FRET) probe AKAR3 (15). AKAR3 CFP fluorescence was excited at 458 nm using an Axiozoom.V16 microscope, which has been shown previously to excite CFP selectively (16); the emitted light was collected at 485 nm (CFP) and 515 nm (YFP) (image frequency: 0.0 Hz). Time-lapse imaging of [cAMP]_i was performed using a recombinant Green Downward cADDis sensor (Montana Molecular, Bozeman, MT). The cADDis fluorescence was excited at 485 nm and the emission recorded at 515 nm using an Axiozoom.V16 microscope. PKA and cAMP were imaged at a frequency of 16 mHz.

All imaging at 34°C was performed using an open chamber. The bath solution containing various stimuli was perifused continuously at a rate of 200 μL/min. In experiments involving PKI, ESI-05, ryanodine, Xestospongin C, Ned-19 thapsigargin, and the Ca²⁺ channel blockers ω-agatoxin and isradipine, cells were preincubated in the solution of the respective agent for 90 min. For the bafilomycin experiments, the preincubation time was 20 min. By preincubating the islets in the agonist/antagonist solutions, we aimed to reach the saturation of the effect by the start of the recording. α-Cells were identified as those exhibiting a response to 1 mmol/L glutamate (Fig. 1C–E) (17) and/or adrenaline (known to inhibit secretion of both insulin and somatostatin) (18).

Images were acquired using ZenBlack or ZenBlue software (Carl Zeiss).

Batches of 10–20 size-matched freshly isolated mouse islets were preincubated in 1 mL Krebs-Ringer buffer containing 1 mmol/L glucose and 0.2% BSA for 30 min at 37°C, followed by a 1-h test incubation in 1 mL of the same medium supplemented as indicated. Glucagon was determined by radioimmunoassay (Euro Diagnostica, Malmö, Sweden) (19). The experiments on the Tpcn1, Tpcn2, or Cd38 knockout mice (Fig. 5) and control mice were assayed using the Meso Scale Discovery glucagon assay (Rockville, MD).

The electrophysiological measurements were performed on α-cells within freshly isolated intact islets (from NMRI or C57Bl/6 mice), using an EPC-10 patch-clamp amplifier (HEKA Electronics, Lambrecht/Pfalz, Germany) and Pulse software. All electrophysiological experiments were performed at 34°C. α-Cells were identified as those active at low (3 mmol/L) glucose and were differentiated from δ-cells (some of which fire action potentials, albeit at low frequency at this glucose concentration) by the distinct appearance of action potentials (Supplementary Fig. 2A). For the membrane potential recordings (Fig. 2C), the perforated patch configuration was used as previously described (20) using solutions IC1 and EC1. Exocytosis was measured as increases in membrane capacitance in α-cells in intact islets as described previously using the standard whole-cell configuration and IC2 and EC2.

Image sequences were analyzed (registration, background subtraction, region of interest intensity vs. time analysis, F/F₀ calculation) using open-source FIJI software (http://fiji.sc/Fiji). The numerical data were analyzed using IgorPro package (WaveMetrics). For calculation of partial areas under the curve (pAUCs), the recording was split into 30-s intervals, and area under the curve was computed for each individual interval (Supplementary Fig. 1C), using the trapezoidal integration. Numbers of measurements/cells are specified in figure legends; the experiments on human islets were performed on islets isolated from three donors. Statistical analysis was performed using R (21). Data are presented as mean ± SEM. The Mann-Whitney U test or Wilcoxon paired test was used to compute the significance of difference between independent and dependent samples, respectively. Multiple comparisons within one experiment were performed using the Kruskall-Wallis test with Nemenyi post hoc analysis (independent samples) or the Friedman test with Nemenyi post hoc analysis (dependent samples).

We tested the effect of adrenaline on glucagon secretion at a glucose concentration that roughly approximates hypoglycemia in vivo (3 mmol/L) (22) and minimizes the activity of β- and δ-cells (see ref. 23 and Supplementary Fig. 1B). Adrenaline stimulated glucagon secretion from isolated mouse pancreatic islets by 3.8 ± 0.8-fold (Fig. 1A), in line with previously reported results (5).

Glucagon secretion is a Ca²⁺-dependent process and is stimulated by an elevation of [Ca²⁺]_i (5). We quantified the adrenaline effect on [Ca²⁺]_i in α-cells within intact islets using time-lapse laser scanning confocal microscopy. At 3 mmol/L glucose, <20% of the cells in pancreatic islets isolated from NMRI mice were active and generated [Ca²⁺]_i oscillations (Fig. 1B). Of the spontaneously active cells, >70% responded to glutamate (Supplementary Movie 1) and were thus identified as α-cells (17,24). In α-cells thus identified, adrenaline induced a rapid and reversible increase in [Ca²⁺]_i (Fig. 1C–E). Similar effects of adrenaline were observed at 1 mmol/L glucose (Supplementary Fig. 1A and E).

The majority of the islet cells (>80%) were inactive at 3 mmol/L glucose but were stimulated when glucose was elevated to 20 mmol/L, as expected for β- or δ-cells (Fig. 1E). At 3 mmol/L glucose, adrenaline did not affect [Ca²⁺]_i in any of these cells (Supplementary Fig. 1B) and at 20 mmol/L actually reduced [Ca²⁺]_i (not shown). Assessed as pAUC (see research design and methods and Supplementary Fig. 1C), responsiveness to adrenaline strongly correlated with spontaneous [Ca²⁺]_i oscillations at 3 mmol/L glucose (Pearson r = 0.78) and responsiveness to glutamate (r = 0.81) (Supplementary Fig. 1D). Similar responses to adrenaline and glutamate were observed in human islets (Fig. 1D and E) and islets of C57Bl/6N mice (Fig. 1F). These data make it unlikely that paracrine effects influence the [Ca²⁺]_i responses to adrenaline in α-cells. Indeed, neither addition of exogenous insulin nor inhibition of insulin receptor with S961 significantly modified the adrenaline signaling in α-cells (Supplementary Fig. 5B).

Glucagon-secreting α-cells are equipped with several types of voltage-gated Ca²⁺ channels (25). The effect of adrenaline on glucagon secretion was abolished by inhibition of L-type (with isradipine) but not P/Q-type (with ω-agatoxin) Ca²⁺ channels (Fig. 2A). The changes in adrenaline-induced [Ca²⁺]_i dynamics produced by isradipine and ω-agatoxin (Fig. 2B) correlated with the effects of these blockers on glucagon secretion (Fig. 2A). When isradipine was applied just before adrenaline, the response to adrenaline was attenuated but not abolished (Fig. 2B). The effect of adrenaline on [Ca²⁺]_i was mimicked by noradrenaline (released by intra-islet adrenergic nerve endings [26]) and by the β-adrenergic agonist isoprenaline but abolished by β-antagonist propranolol. By contrast, the α₁-adrenergic inhibitor prazosin, reported to inhibit the adrenaline-induced [Ca²⁺]_i increases in single α- cells (6), had no detectable effect (Fig. 2B). Thus, the effect of adrenaline on α-cell [Ca²⁺]_i dynamics principally reflects activation of β-receptors.

We tested whether the adrenaline-induced increase in [Ca²⁺]_i can be attributed to stimulation of α-cell action potential firing by whole-cell perforated-patch measurements (20). In agreement with the [Ca²⁺]_i measurements, α-cells were electrically active and fired action potentials at a frequency of ∼1 Hz when exposed to 3 mmol/L glucose. Adrenaline depolarized α-cells by 3 mV, produced a 10-mV reduction of spike height (Fig. 2C and Supplementary Fig. 2F), and increased the duration of the action potential from 4 to 8 ms but did not (except for a moderate stimulation during the initial 4 s) increase spike frequency (Supplementary Fig. 2F and G). These small effects of adrenaline on α-cell electrical activity, possibly reflecting transient activation of a depolarizing store-operated membrane conductance after depletion of the sER Ca²⁺ stores (27), cannot account for the pronounced and sustained (several minutes) increases in [Ca²⁺]_i (Fig. 1E and F). Indeed, neither hyperpolarizing the plasma membrane with K_ATP channel opener diazoxide nor chelating the extracellular Ca²⁺ with EGTA prior to the addition of adrenaline diminished its capacity to increase [Ca²⁺]_i in α-cells (Fig. 2B and Supplementary Fig. 2C–E).

β-Adrenergic signaling results in the Gs-mediated activation of adenylyl cyclase and, hence, increases the cytosolic concentration of cAMP ([cAMP]_i) (28) (Supplementary Fig. 3), which in turn activates the downstream targets PKA and exchange protein directly activated by cAMP (EPAC2). We determined the significance of PKA and EPAC in the regulation of [Ca²⁺]_i and glucagon secretion using the inhibitors myr-PKI (for PKA) and ESI-05 (for EPAC2). Both myr-PKI and ESI-05 inhibited adrenaline-induced glucagon secretion (Fig. 3A) and significantly reduced the adrenaline-induced increase in [Ca²⁺]_i (Fig. 3B), with ESI-05 exerting a much stronger effect than myr-PKI on glucagon secretion. The finding that EPAC2 and/or PKA, despite having similar effects on [Ca²⁺]_i, differentially affect adrenaline-induced glucagon secretion (P < 0.05) suggests that cAMP-dependent activation of EPAC2 may be more tightly linked to exocytosis.

The effects of adrenaline itself on glucagon secretion have previously been examined (5). Here, we explored the roles of PKA and EPAC2 on the cAMP-dependent stimulation of the late-stage depolarization-evoked exocytosis (monitored as increases in membrane capacitance). Figure 3C shows the increase in plasma membrane electrical capacitance (ΔC_m) evoked by a 20-ms depolarization from −70 to 0 mV in the absence or presence of cAMP in the patch pipette and after inhibition of PKA or EPAC2; this pulse duration is comparable with that of the α-cell action potential (Fig. 2C). Under control conditions, exocytotic responses were small (<5 fF). When cAMP was included in the pipette, exocytosis was stimulated by ∼10-fold (Fig. 3C). The stimulatory effect of cAMP on depolarization-evoked exocytosis was resistant to PKI but reduced by ESI-05 (Fig. 3C and D). Longer depolarizations evoked larger exocytotic responses, but the effects of cAMP and the inhibitors were the same (Supplementary Fig. 4).

We explored the effect of adrenaline on [cAMP]_i and its relationship to [Ca²⁺]_i further by imaging the activity of cAMP’s downstream target PKA using the FRET sensor AKAR3 (15). In most islet cells, the application of adrenaline reduced [cAMP]_i (Supplementary Fig. 3), probably reflecting the effect of α2-adrenergic receptors in β- (29) and δ-cells (30). However, in a subset of islet cells (∼10%, corresponding to α-cells), adrenaline increased [cAMP]_i (Supplementary Fig. 3C). These changes in [cAMP]_i correlated with increased α-cell activity and reduced β/δ-cell activity of PKA (Fig. 3E). We compared the adrenaline-induced increases in PKA activity in α-cells with those produced by increasing concentrations of the adenylyl cyclase activator forskolin (Fig. 3F and G). Forskolin dose-dependently increased PKA activity and [Ca²⁺]_i; half-maximal effects were observed at 0.19 ± 0.05 and 5.3 ± 1.5 μmol/L, respectively. The increase in PKA activity induced by 5 μmol/L adrenaline was equivalent to the effect of 4.2 ± 1.1 μmol/L forskolin (Fig. 3H), in agreement with the observation that stimulation of glucagon secretion is only seen at micromolar concentrations of adrenaline (5). The observation that the adrenaline-induced increase in [Ca²⁺]_i requires high [cAMP]_i (= high forskolin) is consistent with the involvement of EPAC2 (K_d for cAMP ≈ 10⁻⁵ mol/L (31)) and the previous report that the stimulatory effect of adrenaline on glucagon secretion is dramatically reduced in EPAC2 knockout islets (5).

The findings that adrenaline produces a marked increase in [Ca²⁺]_i despite having only a marginal effect on α-cell electrical activity and that it retains the capacity to stimulate glucagon secretion in the absence of extracellular Ca²⁺ and in the presence of a high concentration of the K_ATP channel activator diazoxide (to suppress electrical activity [20]) suggest that it, via activation of β-receptors and cAMP-dependent stimulation of PKA and EPAC2, acts by mobilizing Ca²⁺ from intracellular stores. We tested for the capacity of adrenaline to induce glucagon secretion and increase [Ca²⁺]_i in the presence of inhibitors of inositol triphosphate receptor (Xestospongin C [32]), sER Ca²⁺ ATPase (thapsigargin or cyclopiazonic acid [CPA]), and Ca²⁺-induced Ca²⁺ release (ryanodine). Ryanodine, Xestospongin C, and thapsigargin all inhibited the stimulatory effect of adrenaline on glucagon secretion (Fig. 4A) but—unexpectedly—did not appear to affect the adrenaline-induced increase in [Ca²⁺]_i (Supplementary Fig. 5A). We reasoned that there might be multiple Ca²⁺-release mechanisms operating in parallel in the α-cell and that Ca²⁺ release via any one of them might suffice to saturate the high-affinity Ca²⁺ dye Fluo-4 (which has a K_d for Ca²⁺ of 300 nmol/L). Indeed, when the response to adrenaline was instead assayed using the low-affinity indicator Fluo-4FF (K_d = 9.7 μmol/L), ryanodine as well as xestospongin C and CPA produced marked and statistically significant diminution of the [Ca²⁺]_i-increasing effect (Fig. 4B and C).

Prompted by the observations indicating the presence of a ryanodine-independent component in adrenaline-induced Ca²⁺ mobilization, we explored the possible involvement of acidic (lysosomal) Ca²⁺ stores (10) in this phenomenon. Bafilomycin, an inhibitor of the vacuolar type-H⁺ ATPase, abolished the adrenaline effect on [Ca²⁺]_i in α-cells (Fig. 4D).

Ca²⁺ release from lysosomal stores is mediated by activation of Ca²⁺- and Na⁺-permeable (33,34) two-pore channels (TPCs) (10). We analyzed the role of these channels in the adrenaline-induced [Ca²⁺]_i increase and glucagon secretion by generation of Tpcn1 or Tpcn2 knockout mice. The mRNA of both Tpcn1 and Tpcn2 is expressed in mouse and human α-cells (35–37). While adrenaline increased [Ca²⁺]_i in α-cells in islets isolated from Tpcn1^−/− mice (Fig. 5A and D), it had very little effect in islets from Tpcn2^−/− mice (Fig. 5B and D). The effects of ablating TPC1 and TPC2 on [Ca²⁺]_i correlated with those on glucagon secretion: adrenaline retained a stimulatory effect on glucagon secretion in Tpcn1^−/− but not Tpcn2^−/− islets (Fig. 5C). The effect of adrenaline was seemingly smaller in Tpcn1^−/− than in wild-type islets (but this difference did not attain statistical significance). Ryanodine had no inhibitory effect on glucagon secretion in Tpcn2^−/− islets (Fig. 5C). Likewise, pharmacological inhibition of NAADP (the intracellular messenger mobilizing lysosomal Ca²⁺ via TPC2 channels) (38,39), which inhibited glucagon secretion in wild-type islets, had no effect in Tpcn2^−/− islets. Genetic ablation of ADP-ribosyl cyclase 1 (CD38), the enzyme producing NAADP, significantly reduced the increase in [Ca²⁺]_i produced by adrenaline (Figs. 4D and 5B) and abolished adrenaline-induced glucagon secretion (Fig. 5C).

Finally, we tested the effects of culturing islets at 30 mmol/L glucose for 72 h (as an experimental paradigm of poorly controlled diabetes) on adrenaline’s capacity to elevate [Ca²⁺]_i and stimulate glucagon secretion. “Hyperglycemia” abolished the stimulatory action of adrenaline on both [Ca²⁺]_i and glucagon secretion (Fig. 6A, B, and E). By contrast, islets cultured at 11 mmol/L glucose (close to the fed plasma glucose levels in mice) responded robustly to adrenaline with both a [Ca²⁺]_i increase and stimulation of glucagon secretion. “Hyperglycemia” did not affect the responsiveness to glutamate (Fig. 6B) or effects of adrenaline on [cAMP]_i and PKA activity (Fig. 6C and D).

Adrenaline-induced glucagon secretion involves both PKA- and EPAC2-dependent mechanisms (5). Here, we have explored the downstream mechanisms with a focus on intracellular Ca²⁺ dynamics. Our data are suggestive of a hierarchy of intracellular Ca²⁺ signaling events and highlight a decisive role of lysosomal stores and NAADP-activated Ca²⁺ release via TPC2 channels.

Pharmacological inhibition of EPAC2 nearly abolished the stimulatory effect of adrenaline on both glucagon secretion and [Ca²⁺]_i. By contrast, inhibition of PKA, while lowering [Ca²⁺]_i as strongly as inhibition of EPAC2, had only a moderate effect on glucagon secretion. This suggests that a full response to adrenaline requires the activation of both PKA-dependent and -independent mechanisms and that EPAC2 may act downstream of PKA (Fig. 3C and D).

Given the weak effect of adrenaline on α-cell electrical activity, the high sensitivity of both adrenaline-induced glucagon secretion and [Ca²⁺]_i increase after inhibition of L-type Ca²⁺ channels may seem paradoxical. However, this effect may be secondary to depletion of intracellular Ca²⁺ stores when Ca²⁺ entry is blocked. Indeed, adrenaline retained most of its effect on [Ca²⁺]_i when isradipine was added just before adrenaline (Fig. 2B).

The mechanism by which adrenaline increases [Ca²⁺]_i in α-cells has variably been attributed to increased Ca²⁺ channel activity (8) or activation of α1-adrenoreceptors (6). Our study unveils a previously unrecognized role of acidic Ca²⁺ stores in regulation of glucagon secretion. Ca²⁺ liberated by adrenaline-induced PKA and EPAC2 signaling from the acidic stores then triggers further Ca²⁺ increase by activating ryanodine-sensitive Ca²⁺-induced Ca²⁺ release (40). This hierarchy of intracellular [Ca²⁺]_i signaling is supported by the fact that ryanodine exerts no additive inhibitory effect on glucagon secretion in Tpc2 knockout islets.

Our data suggest a link between PKA/EPAC2 signaling and the Tpc2 channel activity, with the NAADP antagonist Ned19 having no additive inhibitory effect on the adrenaline-induced [Ca²⁺]_i increase when PKA and EPAC2 activity was pharmacologically suppressed (Fig. 4D). The production of NAADP is influenced by high concentrations of cAMP—as expected for an effect mediated by low-affinity cAMP sensor EPAC2—and PKA-dependent phosphorylation of S66 in Tpc2 increases the channel’s open probability (41). The involvement of NAADP production is also illustrated by the ability of the genetic knockdown of the NAADP-generating enzyme CD38 to prevent the effects of adrenaline on both [Ca²⁺]_i and glucagon secretion. NAADP-dependent adrenaline-induced Ca²⁺ release via Tpc2 channels was recently implicated in the transduction of β-adrenergic signals in cardiomyocytes (42) but not pancreatic β-cells (43,44). The observation that Tpcn2^−/− mice have normal fasting blood glucose (43) is not in conflict with the idea that adrenaline stimulates glucagon secretion by a TPC2-dependent mechanism, as adrenaline itself will stimulate hepatic glucose production even in the absence of glucagon, and low glucose also stimulates glucagon secretion by a direct (intrinsic) mechanism independent of systemic signals (20,45).

The innervation of mouse and human islets is rather different (46). The finding that adrenaline’s effects on [Ca²⁺]_i are essentially the same in mouse and human α-cells may seem at odds with the observation that adrenaline is without effect on glucagon secretion in human islets (23). It is likely that adrenaline produces a transient stimulation of glucagon secretion that escapes detection during a 1-h incubation in human islets. We can exclude though that this discrepancy stems from different glucose concentrations used for secretion measurements (1 mmol/L) and [Ca²⁺]_i imaging (3 mmol/L). In a series of initial experiments, adrenaline also increased [Ca²⁺]_i in human islets exposed to 1 mmol/L glucose (Supplementary Fig. 1A).

Circulating levels of adrenaline are below 1 nmol/L, and they remain below 100 nmol/L even during exercise (47). Interestingly, these low concentrations of adrenaline inhibit glucagon secretion, while the stimulatory effect becomes detectable at concentrations >1 μmol/L and depends on activation of EPAC2 (5). Such high levels of agonist are unlikely to occur anywhere except close to nerve terminals (48). Thus, our data suggest that the sympathetic signal mediated by locally released noradrenaline (Fig. 2B) plays a key role in linking the physiological “flight-or-fight” response to the α-cell glucagon release. The schematic in Fig. 7 presents a model for adrenaline signaling in pancreatic α-cells. Hypoglycemia elevates systemic adrenaline (released from the adrenal glands) and/or triggers a local release of noradrenaline from sympathetic nerve endings within or adjacent to pancreatic islets. The catecholamine then binds to a β-adrenoreceptor on the α-cell, resulting in increased cytosolic cAMP levels, which activate PKA (which interacts with the Tpc2 channel residing in the membrane of the acidic vesicles) and EPAC2 (which facilitates the liberation of NAADP by CD38). The two signals converge into a release of Ca²⁺ from the acidic stores into the cytosol, which in turn triggers further liberation of Ca²⁺ from ER. Altogether, this leads to a fourfold stimulation of the glucagon release and, via stimulation of hepatic gluconeogenesis, rapid restoration of blood glucose levels. This process depends on background electrical activity of the α-cell and plasmalemmal Ca²⁺ entry (explaining the effect of isradipine) to refill intracellular Ca²⁺ stores.

The mechanism underlying the attenuation of the sympathoadrenal response in diabetes remains debated. Diabetes is associated with the loss of sympathetic islet innervation, and this may, via reduced glucagon secretion, account for the increased risk of hypoglycemia (49). It is therefore of interest that our in vitro model of chronic hyperglycemia lowers the α-cell’s intrinsic responsiveness to adrenaline (Fig. 6) and thus can itself result in the sympathetic islet neuropathy. The exact mechanism remains to be identified but is likely to be downstream of cAMP and activation of PKA, which were both unaffected. Novel therapeutic strategies that bypass the innervation may help to restore normal counterregulation in patients with diabetes.

Acknowledgments. The authors thank Dr. Jin Zhang (Department of Pharmacology and Molecular Sciences, The Johns Hopkins University School of Medicine, Baltimore, MD) for the gift of AKAR3.

Funding. This work was supported by Medical Research Council program grant G0901521. A.H. is a recipient of a Diabetes UK PhD Studentship. Q.Z. is a Diabetes UK RD Lawrence Fellow. G.A.R., A.G., and P.R. hold Wellcome Trust Senior Investigator awards (102828, 098424, and 095531). During the initial stages of the project, A.I.T. held an Oxford Biomedical Research Council postdoctoral fellowship.

Duality of Interest. A.K.R. is a current employee of Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.S., M.W., J.G.K., A.K.R., C.E.C., and R.R. performed the experiments. A.H., Q.Z., A.G.-A., and A.I.T. performed the experiments and analyzed data. N.C.S., M.Z., D.B., and A.G. provided reagents. M.Z., G.A.R., A.G., P.R., and A.I.T. contributed to data interpretation. G.A.R., A.G., P.R., and A.I.T. wrote the manuscript. P.R.V.J. provided human islets. P.R. and A.I.T. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.