Paracrine inhibition via G protein inwardly rectifying potassium channels regulates glucagon secretion from human pancreatic alpha cells

SUMMARY

Impaired glucagon secretion from pancreatic alpha cells is a cause of life-threatening hypoglycemia in individuals with type 1 diabetes (T1D). The mechanisms that lead to defective glucagon secretion remain unclear. Here, we show that the human alpha cell’s competence to secrete glucagon depends on paracrine inhibitory input from beta (serotonin [5-HT], γ-aminobutyric acid [GABA]) and delta (somatostatin [SST]) cells. These paracrine signals activate G protein-coupled receptors (GPCRs) that open G protein-gated inwardly rectifying potassium (GIRK) channels, which have a major impact on glucagon secretion. In the absence of this paracrine input, glucagon secretion progressively diminishes until it habituates completely. Strikingly, 5-HT, GABA, and SST restored impaired glucagon secretion in islets from donors with long-duration T1D. These findings indicate that paracrine inhibition is needed to prevent habituation of glucagon secretion. As beta cells are destroyed in T1D, alpha cells lose this inhibition, and consequently, their capacity to adequately secrete glucagon to counter hypoglycemia.

Graphical abstract

In brief

Panzer et al. show that paracrine signals from beta and delta cells inhibit glucagon secretion from alpha cells by opening G protein-gated inwardly rectifying potassium channels. Loss of this inhibition makes alpha cells refractory to glucose, but restoring it reestablishes glucagon secretion in islets from donors with type 1 diabetes.

INTRODUCTION

Pancreatic alpha cells are essential for glucose homeostasis. They secrete glucagon to prevent hypoglycemia by stimulating hepatic gluconeogenesis and glycogenolysis.^1,2 When this regulation is disrupted, as seen in diabetes, improper glucagon secretion contributes to hyperglycemia and increases the risk of hypoglycemia during insulin therapy.^3-5 While it is well established that glucagon secretion is stimulated at low glucose levels and suppressed at high glucose levels,^4,6-8 the dynamic characteristics of this response remain incompletely understood. A deeper understanding of the regulatory mechanisms underlying alpha cell dysfunction is essential for improving glycemic control in diabetes management.

Alpha cells are excitable cells that rely on action potentials to secrete glucagon. The alpha cell’s electrical behavior shows inactivation, adaptation, and desensitization.^9,10 As a result, maintaining alpha cells in a depolarized state does not stimulate but inhibits action-potential firing and glucagon release.^11,12 This habituation can be restored by active mechanisms involving hyperpolarization.¹³ In the pancreatic islet, alpha cells are surrounded by beta cells and delta cells that provide abundant inhibitory paracrine signals.^2,3,14-17 However, for these signals to limit membrane depolarization and reverse reduction of action potentials, they must link to membrane repolarization. Of the beta cell-derived factors that suppress glucagon release, γ-aminobutyric acid (GABA) and serotonin (5-HT) can act on receptors that couple to G protein-gated inwardly rectifying potassium (GIRK) channels that control cellular excitability.^3,18 Delta cells also inhibit alpha cells by releasing somatostatin (SST), a potent inhibitor of glucagon secretion.^6,19-21 Evidence in rodent alpha cells indicates that inhibitory G proteins and SST receptors activate a hyperpolarizing K⁺ channel in alpha cells.^22,23 However, the impact on glucagon secretion was not studied.

We hypothesized that paracrine signals derived from beta and delta cells act on multiple receptors that converge on GIRK channels to allow the alpha cell to maintain its full secretory potential. To this end we performed perifusion studies of dynamic glucagon secretion from isolated human islets. We examined the kinetics of glucagon secretion in response to changes in glucose concentration, paracrine signals, and GIRK channel modulators. Our results demonstrate that prolonged exposure to stimulatory, low-glucose conditions gradually diminishes glucagon secretion and that this habituation is partially reversed by short exposure to high glucose concentrations or paracrine signals. We further show that human alpha cells express the G proteins, G protein-coupled receptors (GPCRs), GIRK channels, and regulatory proteins that form the GIRK macromolecular signaling complex. We recorded bona fide GIRK channel activity in alpha cells by whole-cell patch-clamp electrophysiology and found that modulating these channels strongly impacted glucagon secretion in perifusion studies. Stimulating alpha cells in islets from donors with type 1 diabetes (T1D) by a lowering in glucose concentration did not elicit glucagon secretion, but short exposure to GABA, 5-HT, or SST dramatically increased glucagon secretion. These findings indicate that brief inhibition of alpha cells can restore defective glucagon secretion and suggest that, once beta cell input is lost in T1D, alpha cells cannot respond to changes in glucose levels.

RESULTS

Systematic studies in isolated human islets using the perifusion technique have shown qualitative and quantitative coherence between insulin secretion rates measured in vivo and in vitro during stimulation with physiological glucose concentrations (e.g., Henquin²⁴). Dynamic evaluations of glucagon secretion are less common. Here, we characterized and quantified the kinetics of glucose-induced changes in glucagon secretion from human islets. Given the inherent heterogeneity of human samples, we used islets from multiple donors to ensure a broad representation of biological variability. Isolated human islets from nondiabetic donors were obtained through Prodo Laboratories and from donors with T1D procured by Dr. Alvin Powers (Vanderbilt University) and Dr. Rita Bottino (Director Islet Programs, Imagine Pharma). Detailed donor characteristics can be found in Tables S1 and S2.

Alpha cell responses to lowering glucose concentrations habituate and require a reset mechanism to recover responsiveness

We first examined the dynamics of glucagon secretion in response to step decreases in glucose concentrations from 11 mM to five different concentrations (Figures 1A and 1B). We found that all tested glucose steps, even those to concentrations higher than the human glycemic setpoint (5 mM), increased glucagon secretion (Figures 1A and 1B). Notably, even a modest decrease from 11 mM (11G) to 9 mM (9G) elicited a glucagon response. The glucagon response became progressively stronger as we increased the glucose steps to 7 mM (7G) and 5 mM (5G) and to hypoglycemic conditions (3 mM, 3G, or 1 mM, 1G; Figure 1B). Lowering the glucose concentration further to 1 mM after an initial step to 3, 5, 7, or 9 mM elicited similarly large glucagon responses, independently of the glucose step (Figure S1). These data suggest a saturation of the response at very low glucose levels. Our findings indicate that alpha cells in intact islets do not respond only until they are exposed to hypoglycemic conditions (>4 mM) but also respond to relative reductions in glucose levels even at physiological high levels. Although the alpha cell activation depended on decrements in the glucose concentration, the response magnitude seemed to be related to the absolute glucose concentration reached after the drop. More experimentation and modeling are required to define the complex relationship of alpha cell responses to glucose levels. This complex concentration-response relationship is in line with what has been reported in vivo for glucagon secretion²⁵ and contrasts with what is known for insulin secretion from human islets.^24,26

Figure 1. Dynamic responses in glucagon secretion from human islets to step decreases in glucose concentrations reveal strong habituation.

(A) Perifusion experiments of dynamic glucagon secretion showing responses to step decreases in glucose concentration. Data plotted as fold change over baseline release at 5 mM (5G; n = 5–10 nondiabetic donors).

(B) Quantification of glucagon responses shown in (A) (area under the curve [AUC], min 40–96).

(C) Glucagon secretion in response to prolonged low glucose showing a gradual decline. Data plotted as fold change over baseline (5G; n = 13 nondiabetic donors). For comparison, changes in glucagon secretion elicited by sequential glucose steps from 11G to 7G and 7G to 1G are shown (gray).

(D) Quantification of glucagon secretion shown in (C) at 0, 60, and 90 min in low glucose (1G; plotted as percentage of peak response).

(E) Glucagon secretion in response to specific stimulation with epinephrine (Epi; 10 μM) ± forskolin (Fsk,1 μM) in low glucose (1G) for 90 min, followed by membrane depolarization with KCl (30 mM). Data plotted as fold change over baseline (5G) from a representative nondiabetic donor.

(F) Quantification of glucagon secretion shown in (E) as AUC (minutes 124–154, n = 4–8 nondiabetic donors).

(G) Glucagon secretion in response to opening K_ATP channels with diazoxide (1 or 100 μM; DZ 1 or DZ 100) after prolonged exposure to low glucose (1G, 120 min). Data plotted as fold change over baseline (5G; n = 7 nondiabetic donors).

(H) Quantification of glucagon secretion shown in (G) in response to diazoxide (AUC, min 132–156).

(I and J) Glucagon and insulin secretion during dynamic perifusion in response to a reset period of 5, 15, or 30 min in high glucose (17G). Data normalized to minimum (at 17G) and maximum (at 1G) responses (%, average responses do not reach 100% because peaks do not synchronize; n = 5 nondiabetic donors).

(K) Quantification of peak glucagon response after reset period shown in (I) and (J).

Data are presented as mean ± SEM throughout the paper; *p < 0.05, ANOVA followed by multiple comparisons in (F), (H), and (K) and one-sample t test to compare the mean to 0 (B).

We noticed that, under constant stimulatory conditions, glucagon secretion gradually diminished (Figure 1). To characterize the kinetics of glucagon secretion, we recorded glucagon secretion under prolonged exposure to low glucose (1 mM). After an initial sharp increase upon transitioning from high (7 or 11 mM) to low glucose, glucagon secretion progressively declined (slope = −0.78 ± 0.07, n = 5 islet preparations; Figures 1C and 1D). This pattern was consistent across different starting glucose concentrations and was independent of glucose step size (Figures 1C and S2). After 60 min of continuous exposure to 1 mM glucose, glucagon secretion declined to levels comparable to those observed under inhibitory conditions at high glucose (11 mM; Figure 1D). Using published graphs, we calculated slopes of −0.73, −0.78, and −0.79 for human,²⁵ dog,²⁷ and rat²⁰ glucagon secretion under prolonged hypoglycemic clamps in vivo (Figure S3). However, other in vivo studies show sustained glucagon responses.²⁸ Thus, whether the habituating glucagon response in vitro reflects a pattern seen in in vivo studies may require additional systematic studies.

To exclude that the observed habituation was due to degranulation of alpha cells, we stimulated islets after prolonged exposure to low glucose with epinephrine ± forskolin to increase intracellular cyclic AMP (cAMP) levels (Figures 1E and 1F). This treatment significantly increased glucagon secretion, indicating that granules could still be mobilized for release. At a low concentration (1 μM), diazoxide moderately increased glucagon secretion (Figure 1G), suggesting that a small alpha cell repolarization due to K_ATP channel opening restored glucagon secretion glucose.^10,12 At a high concentration (100 μM), diazoxide inhibited glucagon secretion (Figures 1G and 1H), in agreement with studies showing that alpha cells exposed to this high diazoxide concentration will be permanently hyperpolarized and show no glucagon secretion at either 1 or 6 mM glucose.^10,12 These data suggest that the alpha cell’s membrane potential could still be modulated in both directions to alter glucagon secretion. Thus, in the habituated state, alpha cells generate enough ATP to close ATP-sensitive potassium channels, but the depolarization induced by closing ATP-sensitive potassium channels fails to elicit a full secretory response.

The habituated state of the alpha cell is predicted by electrophysiological studies showing that the size of action potentials diminishes in alpha cells exposed to constant stimulatory conditions, most likely due to inactivation of voltage-gated Na⁺ and Ca²⁺ channels.^9,29 We posited that inhibiting alpha cells by exposing them to high glucose would allow these channels to recover from inactivation and reverse habituation. To this end, we incubated alpha cells in low glucose (1 mM, 1G) for 90 min, followed by a high-glucose phase (17 mM, 17G) for 5, 15, or 30 min before reintroducing low glucose (Figures 1I and 1J). High-glucose exposures lasting 15 min or longer significantly restored glucagon secretion in response to subsequent low-glucose stimulation (Figure 1K). These findings indicate that alpha cells require a resetting inhibitory period after each secretion cycle to regain responsiveness.

Paracrine signaling modulates glucagon secretion and increases alpha cell responsiveness

We next investigated whether the reset effect is driven solely by glucose or by other signaling factors. While high glucose directly inhibits alpha cells, it also activates beta and delta cells, which release paracrine signals that suppress glucagon secretion. Among these, 5-HT, GABA, and SST play key roles in alpha cell regulation. To determine whether these paracrine factors contribute to the reset mechanism, we examined their individual effects on glucagon secretion. Brief exposure (5 min) to either 5-HT or SST under normal glycemic conditions (5.5 mM glucose, 5G) significantly suppressed glucagon secretion compared to baseline levels. Notably, once these inhibitory signals were removed, glucagon secretion exhibited a rebound, surpassing baseline levels (Figures 2A and 2B). A similar pattern was observed when these agonists were applied following a 30-min exposure to low glucose (1 mM, 1G; Figure 2C).

Figure 2. Paracrine signals regulate and restore glucagon secretion from human islets.

(A) Perifusion experiments of dynamic glucagon secretion showing responses to somatostatin (SST; 1 μM) and serotonin (5HT; 10 μM). Data plotted as fold change over baseline release at 5 mM (5G; n = 10 nondiabetic donors).

(B) Quantification of glucagon responses shown in (A). Dotted line indicates baseline glucagon secretion at 5G. Time points indicated in (A) show significant inhibition of glucagon secretion (1 and 3) and rebound response (2 and 4) to the paracrine stimulus.

(C) Glucagon secretion in response to SST (1 μM) or 5HT (10 μM) in low glucose (1G). Data plotted as fold change over baseline release at 5G (n = 6 nondiabetic donors).

(D and E) Glucagon secretion in response to SST (1 μM), 5HT (10 μM), or γ-aminobutyric acid (GABA; 100 μM) after prolonged exposure to low glucose (1G, 90 min). Data normalized to minimum (at 5G) and maximum (at 1G) responses (%).

(F) Quantification of glucagon peak responses during the rebound phase at min 160–180 shown in (D) and (E). Dotted line indicates glucagon levels in the absence of paracrine stimulation.

(G) Glucagon secretion in response to high (17G) and low glucose (1G) of islets from nondiabetic donors and donors with long T1D duration (see Tables S1 and ^S2 for donor characteristics). Data plotted as fold change over baseline release at 5G (n = 5 nondiabetic and 3 T1D donors).

(H) Continuation of glucagon secretion from isolated islets of T1D donors shown in (G) in response to 5HT (100 μM), SST (1 μM), or GABA (100 μM).

Data plotted as fold change over baseline release (5G; n = 4 T1D donors). *p < 0.05; one-sample t test to compare the mean to the mean of basal secretion at 5G (1 in B) and to the mean of control at minutes 160–180 (13.3% in F).

We then tested whether these paracrine molecules could mimic the resetting effect of high glucose after sustained low-glucose stimulation. When normalized to control levels, all tested paracrine antagonists elicited a significant rebound in glucagon secretion following their application (Figures 2D-2F). This rebound was delayed compared to that elicited by exposure to high glucose (Figures 1I-1K), which may reflect the efficacy of individual exogenous agonists versus that of endogenous secretion of paracrine signals as well as differences in the quality of islet preparations. Some islet preparations showed pulsating glucagon responses (e.g., Figures 2A-2D and 2G) that may have masked an earlier response to the agonists. That the rebound appeared later in islets kept at low glucose (Figure 2C) further suggests that it may take alpha cells longer to mount a rebound response when kept at habituating low-glucose conditions.

We tested whether insulin per se could play a role in resetting glucagon secretion by adding exogenous insulin during the prolonged exposure to low glucose (Figure S4). Under conditions during which there was no endogenous insulin secretion (1 mM glucose), insulin did not diminish but, surprisingly, modestly amplified glucagon secretion. Also, exogenous insulin did not produce a rebound response like the one induced by exposure to high glucose or paracrine molecules (Figures 1I-1K, 2D, and 2F).

Inhibitory paracrine signaling paradoxically increases glucagon secretion in islets from long-duration T1D donors

Our findings indicate that the glucagon secretory response habituates to a point where it does not differ from basal levels of secretion. Glucagon secretion, however, could be restored by providing inhibitory input derived from neighboring cells. We hypothesized that, in the absence of beta cells, as is the case in type 1 diabetes, alpha cells remain stuck in the habituated state. This could explain deficient glucagon secretion in individuals with type 1 diabetes. To test this notion, we investigated islets from long-duration T1D donors. Unlike islets from nondiabetic donors, islets from T1D donors lacked a glucagon response to a glucose step from 17 to 1 mM (Figure 2G). However, when we introduced paracrine agonists to islets maintained in 1mM glucose, we observed a striking difference: instead of inhibiting glucagon secretion as seen in nondiabetic islets, the antagonists significantly stimulated glucagon release in T1D islets, leading to a sustained secretory phase (Figures 2H and S5). Glucagon concentratiiabetic donors (increases rons reached levels comparable to those elicited by drops in glucose in islets from nondanged 10–80 pM; Figure S5).

Alpha cells express the molecular framework that links paracrine signaling to GIRK channels

The inhibitory paracrine signals (5HT, SST, and GABA) that suppressed glucagon secretion are known to activate inhibitory GPCRs that signal via Galpha_o/i to reduce intracellular cAMP levels.^30-32 Receptors for these signals can also directly open G protein-coupled inward rectifying K⁺ channels (GIRKs) via Gbeta/gamma subunit signaling (Nguyen et al.³³; Figure 3A). Opening of GIRK channels hyperpolarizes the cell membrane, making GIRKs ideal candidates to repolarize alpha cells and reverse Na⁺ and Ca²⁺ channel inactivation, thus restoring glucagon secretion. We examined publicly available gene expression data for the expression of GPCRs known to couple to GIRK (https://www.gaultonlab.org/islet-genomics/; Elgamal et al.³⁴). We found that alpha cells express Galpha_o/i-linked receptors for GABA, 5-HT, glutamate, and SST known to modulate GIRK channel activity (Figure 3B; Nguyen et al.³³). Alpha cells also express high levels of molecules that are part of the complex macromolecular GIRK signaling framework such as the regulators of G protein signaling, RGS7 and RGS4, the Gbeta5 subunit, and sorting nexin 27 (SNX27) (Figures 3C and 3D). Importantly, we found that alpha cells express the GIRK channel subunits KCNJ3 and KCNJ6 (encoding GIRK1 and GIRK2, respectively). The genes for several of these receptors, signaling molecules, and GIRK subunits, namely GRM4, SSTR1, SSTR2, RGS7, and KCNJ6, are alpha cell-enriched genes (http://www.proteinatlas.org).

Figure 3. Gene expression data of GIRK signaling-related molecules involved in paracrine alpha cell inhibition.

(A) Schematic illustration of GIRK channel activation in alpha cells. Upon activation of inhibitory G protein-coupled receptors (GPCRs) by paracrine signals such as serotonin (5-HT), somatostatin (SST), and γ-aminobutyric acid (GABA), G protein βγ subunits directly bind to and activate GIRK channels. This activation leads to potassium efflux and hyperpolarization of the plasma membrane. GIRK channels are also modulated by endogenous proteins and molecules, including regulator of G protein signaling (RGS) proteins, cholesterol, and sorting nexin 27 (SNX27), which influence their stability and function.

(B) Gene expression levels of inhibitory GPCRs for 5-HT, SST, and GABA in human islet cell subtypes, showing preferential expression in alpha cells.

(C–E) Expression of GIRK channel subunits KCNJ3 (GIRK1) and KCNJ6 (GIRK2, E), along with associated regulatory proteins RGS7, RGS4, SNX27 (C), and the G protein b subunit 5 (GNB5, D), in human alpha cells compared to beta and delta cells and acinar cells.

Data were obtained from publicly available gene expression dataset (https://www.gaultonlab.org/islet-genomics/; Elgamal et al.³⁴). Data are presented as mean ± SEM; n = 27 nondiabetic human donors. * = alpha cell enriched gene (http://www.proteinatlas.org); @ = gene associated with alpha cell electrophysiological properties (humanislets.com); T1D and T2D = expression in alpha cells changes significantly in type 1 or type 2 diabetes (https://www.gaultonlab.org/islet-genomics/).

We searched the HumanIslets data portal (humanislets.com) for associations of the expression of these genes with electrophysiological features of alpha cells. We found the following significant associations: GABBR1 with half inactivation sodium current and sodium current amplitude; GRM4 with sodium current amplitude; RGS4 with half inactivation sodium current; GNB5 with late and late calcium current amplitude; and KCNJ6 with sodium current amplitude. These data indicate that alpha cells are uniquely equipped with the molecular components necessary for GIRK channel-mediated inhibition and support a functional role of these channels in regulating glucagon secretion.

Human alpha cells express functional GIRK

We performed whole-cell patch-clamp experiments on dispersed human islet cells. In voltage clamp, GIRK currents were assessed using a 500-ms voltage ramp (from −120 to 50 mV). We quantified and compared currents elicited at −100 mV. SST induced GIRK currents that were blocked by tertiapin-Q (TPQ), a blocker of GIRK channels (Figure 4). ML-297, an opener of GIRK channels, induced GIRK currents that were blocked by TPQ (Figure 4). Like SST, the GABA_B receptor agonist baclofen (10 μM) produced currents as assessed by the voltage ramp (Figure S6). We could not record GIRK channel activation after exposure to 5-HT, but the islet sample size was small (n = 9 cells from two preparations). We used a patch-clamp protocol to reveal the GIRK current that required artificial conditions. Additional experiments are needed to examine GIRK outward K⁺ currents at physiological voltages and their impact on the electrical behavior of the alpha cell. Nevertheless, our results provide functional evidence for the presence of a GPCR-coupled GIRK conductance in alpha cells.

Figure 4. Functional GIRK currents are present in islet alpha cells.

(A) I-V plot for GIRK currents evoked by somatostatin (SST, 1 μM, gray line). GIRK currents evoked by SST were blocked by the specific GIRK channel antagonist tertiapin-Q (TPQ; 200 nM, black line). A comparison of the SST-evoked current to the basal current (light gray line) is shown in the inset.

(B) Boxplot of GIRK current evoked at −100 mV.

(C) I-V plot for basal GIRK current (light gray line) together with GIRK currents evoked by the GIRK channel specific agonist ML297 (10 μM, dark gray line).

(D) Boxplot of GIRK currents evoked at −100 mV. Cells were held at −40 mV for 50 ms followed by a ramp protocol from −120 to +50 mV for 500 ms before returning to the −40-mV holding potential.

p values are shown for statistical comparisons using one-way ANOVA followed by multiple comparisons in (B), or Student’s t test in (D).

Modulation of GIRK channel activity impacts glucagon secretion

We modulated GIRK channel activity with either TPQ or ML-297 and measured glucagon secretion from human islets using perifusion assays. Notably, TPQ doubled whereas ML-297 significantly suppressed glucagon secretion at baseline glycemia (5.5 mM, 5G; Figures 5A and 5B). These effects were remarkable because they were measured at glucose levels close to the nadir of glucagon secretion as established using static incubation.⁹ Moreover, blocking GIRK channels with TPQ prevented the usual decline in glucagon secretion observed during high glucose (17 mM, 17G; Figures 5C and 5D).

Figure 5. GIRK channel activity regulates glucagon secretion.

(A) Perifusion experiments of dynamic glucagon secretion showing responses to GIRK channel blockade with tertiapin-Q (TPQ; 200 nM) or activation with ML-297 (10 μM). Data plotted as fold change over baseline release at 5 mM (5G; n = 5 nondiabetic donors).

(B) Quantification of glucagon responses shown in (A) (area under the curve [AUC], min 14–47).

(C) Glucagon secretion (relative to glucagon secretion at minute 32) in response to glucose step from 5G to 17G in control conditions and in the presence of TPQ (data are from experiment shown in A).

(D) Quantification of glucagon responses shown in (C) (AUC, min 32–53).

(E) Glucagon secretion in response to GIRK channel blockade with TPQ (200 nM) or activation with ML-297 (10 μM) after a glucose step from 5G to 1G (n = 4–5 nondiabetic donors).

(F) Quantification of glucagon responses shown in (E) (AUC, minutes 40–80).

(G) Glucagon secretion (relative to glucagon secretion at time 62 min) in response to glucose step from 1G to 17G in control conditions and in the presence of TPQ (data are from experiment shown in E).

(H) Glucagon secretion in response to GIRK channel blockade with TPQ (200 nM) or activation with ML-297 (10 μM) after extended low-glucose exposure (1G; 120 min; n = 8 nondiabetic donors).

(I) Quantification of glucagon responses shown in (H) (AUC, min 132–160).

(J) Glucagon secretion (relative to glucagon secretion at minute 130) in response to TPQ and ML-297 (data are from experiment shown in H).

*p < 0.05; one-way ANOVA followed by multiple comparisons in (B), (F), and (I); Student’s t test in (D); two-way ANOVA (mixed-effect analysis) followed by multiple comparisons to control values in (A), (E), (G), and (J).

GIRK channel modulation also had a strong impact on glucagon secretion during the initial response to a lowering in glucose concentration (17–1 mM) and after sustained exposure low glucose (1 mM; Figures 5E-5J). Although a decrease in glucagon secretion was still observed when glucose concentration was transiently raised to 17 mM, the rebound in secretion following the glucose step was significantly attenuated by TPQ compared to glucose alone (Figure 5G). Strikingly, even after prolonged exposure to 1 mM, the GIRK modulators were able to change glucagon secretion in both directions (Figures 5H-5J). These results indicate that GIRK channel activity plays a critical role in regulating alpha cell secretion under basal glucose conditions and in response to changing glucose levels. The data also demonstrate that closing GIRK channels can increase glucagon secretion after habituation.

Measuring insulin secretion in parallel to glucagon secretion in the studies shown in Figure 5 revealed that the GIRK channel modulators did not change insulin secretion (Figure 6), in line with low expression of GIRK channels in beta cells (Figure 3) and published physiological data (Dickerson et al.,³⁵ but see Kailey et al.³⁶).

Figure 6. GIRK channel activity does not regulate insulin secretion.

(A) Perifusion experiments of dynamic insulin secretion showing responses to GIRK channel blockade with tertiapin-Q (TPQ, 200 nM) or activation with ML-297 (10 μM). Data plotted as fold change over baseline release at 5 mM (5G, insulin was measured in the recording shown in Figure 5A; n = 5 nondiabetic donors).

(B and C) Zoom in on the first 30 min of the recording shown in (A) demonstrating that, at 5 mM glucose concentration, insulin secretion does not change (B), while glucagon secretion changes dramatically (C; same as in Figure 5 shown here for illustration).

(D) GIRK channel activity does affect high-glucose (17 mM)-stimulated insulin secretion.

Paracrine signaling couples to GIRK channels to regulate glucagon secretion

To assess whether GIRK channels mediate the inhibitory effects of paracrine signals on glucagon secretion, we challenged glucagon responses to SST or 5-HT with the GIRK channel blocker TPQ (Figure 7). As expected, both SST and 5-HT significantly suppressed glucagon secretion when applied in low-glucose conditions (1 mM, 1G; Figure 7). Blocking GIRK channels with TPQ resulted in an initial increase in glucagon secretion. However, in the presence of TPQ, SST or 5-HT did not reduce glucagon secretion compared to control islets (Figures 7B, 7C, 7E, and 7F). These findings suggest that GIRK channel activation is a key mechanism through which these paracrine signals suppress glucagon secretion, highlighting the role of GIRK in controlling alpha cell function.

Figure 7. GIRK channels mediate paracrine suppression of glucagon secretion.

(A) Glucagon secretion in response to somatostatin (SST, 1 μM) under low-glucose (1G) conditions, with and without prior GIRK channel blockade using TPQ (200 nM, n = 5 nondiabetic donors).

(B) Glucagon secretion as in (A) normalized to largest (100%) and smallest (0%) glucagon levels between minutes 60 and 110. Notice how GIRK channel inhibition attenuates suppression by SST.

(C) Quantification of glucagon responses shown in (B) (area under the curve [AUC], min 60–80).

(D) Glucagon secretion in response to serotonin (5HT, 10 μM) under low-glucose (1G) conditions, with and without prior GIRK channel blockade using TPQ (200 nM).

(E) Glucagon secretion as in (D) normalized to largest (100%) and smallest (0%) glucagon levels between min 60 and 110. Notice how GIRK channel inhibition attenuates suppression by 5HT.

(F) Quantification of glucagon responses shown in (E) (AUC, min 60–80).

* p < 0.05, Student’s t test.

DISCUSSION

Using perifusion of human islets, we determined the dynamics of glucagon secretion in response to varying glucose levels and inhibitory paracrine signals and found that (1) glucagon secretion has a complex concentration-response relationship to glucose levels that contrasts with what has been observed with static incubation studies but is in line with in vivo studies; (2) glucagon secretion habituates under constant stimulatory conditions; (3) short inhibition by paracrine signals reactivates glucagon secretion; and (4) this paracrine inhibition is mediated by hyperpolarizing GIRK channels. Our study demonstrates that glucagon secretion has kinetic properties that cannot be predicted from static incubation studies and cannot be modeled after what is known for insulin secretion. As discussed below, the complex kinetics of glucagon secretion reflect the interplay between glucose stimulation, inhibitory signaling, and GIRK channel activity in alpha cells.

We conducted our studies on multiple donors to ensure a broad representation of the inherent biological heterogeneity of human tissue samples. By exposing islets to a variety of glucose steps, we found that alpha cell secretion does not track glucose levels in a linear manner. Alpha cells respond not only to absolute glucose levels but also to drops in glucose. Even when glucose levels decrease moderately—from 11 to 9 mM, a change of only 2 mM and still within an elevated glucose range—alpha cells exhibited a response. Thus, there was no fixed glucose concentration threshold for glucagon secretion. However, the response was more pronounced when glucose levels dropped below concentrations considered hypoglycemic (<4 mM). This suggests that alpha cells are not only proportional detectors that measure glucose concentration but also differential detectors that measure the rate of change. This behavior is in line with in vivo studies.²⁵

In contrast with results showing that beta cell responses to a constant glucose stimulus are sustained,^26,37,38 we found that alpha cell responses were transient. We calculated a slope of −0.8 for the decrement in secretion with an x intercept at 93 min, which corresponds to a rate of ~10% secretion loss per 9 min. This decline is not due to exhaustion of alpha cells, as they still responded robustly to other stimulants, such as forskolin (which activates adenylyl cyclase and increases cAMP levels), epinephrine (which stimulates beta-adrenergic receptors to enhance glucagon release), and KCl-induced membrane depolarization (which opens voltage-gated calcium channels and triggers exocytosis). The diminished glucagon secretion could be further reduced by diazoxide, a K_ATP channel opener, suggesting that alpha cells still produced enough ATP to close K_ATP channels and elicit glucagon secretion. These findings indicate that the decline in glucagon secretion during prolonged low-glucose exposure is not due to a loss in secretory capacity or energy depletion. Importantly, this decline in secretion is not an in vitro artifact because it mirrors the kinetics of glucagon secretion in hypoglycemic clamps performed in vivo in humans, rats, and dogs.^20,25,27

The decline in glucagon secretion during prolonged low glucose is a physiological state predicted by electrophysiological studies showing that depolarizing alpha cells inhibits rather than stimulates action-potential firing.^11-13 These studies further showed that hyperpolarizing K⁺ currents are needed to limit membrane depolarization and prevent inactivation of alpha cell action potentials, thus re-initiating action-potential firing. Indeed, in our hands, transient inhibition of alpha cells by high glucose reactivated the alpha cell’s potential to secrete glucagon. We therefore propose that (1) habituation of glucagon secretion is caused by inactivation of voltage-gated Ca²⁺ and Na⁺ channels due to prolonged depolarization and (2) high glucose reverses this habituated state by providing inhibitory inputs that restore action-potential firing.

Like high glucose, we found that short exposure to inhibitory paracrine signals such as GABA, 5-HT, and SST reactivated glucagon secretion in islets from nondiabetic donors. When we exposed human islets from donors with long-duration T1D to these paracrine signals, glucagon secretion increased 20- to 100-fold. This effect was the more striking because these islets failed to respond to low-glucose stimulation. That paracrine signals restored glucagon secretion in islets from nondiabetic and diabetic donors indicates that inhibition positively regulates glucagon secretion. Thus, paradoxically, the dysregulation of glucagon secretion in diabetes may stem from a loss in paracrine inhibition. We posit that the loss of beta cell signals keeps alpha cells in a habituated state that prevents adequate glucagon secretion in response to drops in glycemia. This could in part explain the defective glucose counter-regulation seen in patients with type 1 diabetes.

How do SST, 5-HT, and GABA reactivate alpha cells? GABA is known to activate GABA_A receptors that hyperpolarize alpha cells.³⁹ SST and 5-HT activate GPCRs that diminish cAMP levels in alpha cells.^3,36,40 However, receptors for SST, 5-HT, and GABA can also converge on a shared inhibitory mechanism, namely GIRK channels. Using a publicly available gene expression dataset, we found that most of the GIRK-signaling macromolecular complex is expressed in human alpha cells. Several of the genes are alpha cell-enriched genes and are involved in alpha cell physiology. Receptors known to couple to GIRK channels were found for GABA, 5-HT, and SST as well as for glutamate. Of note, Na⁺ current amplitude or half inactivation in alpha cell were physiological traits associated with the expression of the GABA receptor subunit GABBR1, the glutamate receptor GMR4, the GIRK subunit KCNJ6, and the G protein GNB5. Our electrophysiological and secretion studies confirmed that GIRK channels are functional and coupled to inhibitory paracrine signals in alpha cells. Thus, alpha cells are equipped with the molecular complex that links inhibitory paracrine signals to a hyperpolarizing K⁺ conductance that modulates Na⁺ channel activity.

How these paracrine signals affect glucagon secretion in vivo in clamped human subjects has mostly been investigated in T1D patients using receptor antagonists, not agonists (e.g., Farhat et al.⁴¹). A recent study testing the effects of a controlled-release formulation of GABA in subjects with long-standing T1D showed that the treatment re-established a counter-regulatory glucagon response to hypoglycemia.⁴² Among the possible explanations, these effects could be attributed to GABA targeting GIRK channels in alpha cells.

A major finding of our study is that GIRK channels play a key role in regulating glucagon secretion. Previous studies in rodent alpha cells showed that inhibitory G proteins and SST receptors activate a hyperpolarizing K⁺ channel in alpha cells^22,23 (a report mentions in the abstract that SST inhibits secretion of both human beta and alpha cells by activating GIRK channels, but results are only shown for beta cells³⁶). However, SST-induced inhibition of mouse alpha cells is not affected by the GIRK opener TPQ.²¹ Here we demonstrate that GIRK channel activation is an important mechanism through which high glucose and paracrine signals suppress glucagon release. Strikingly, under glucose concentrations that are equivalent to normoglycemic conditions (5 mM), close to the nadir of glucagon secretion in static incubation studies,⁹ GIRK channel modulation produced strong effects on glucagon secretion in both directions. These findings support the notion that GIRK channels not only contribute to regulating glucagon secretion in response to glucose fluctuations but are also responsible for maintaining the alpha cell in a ready state.

According to our hypothesis, blocking GIRK channels and the related depolarization of alpha cells should have accelerated the decay in glucagon secretion. However, this did not happen (Figure 5). One interpretation is that pharmacological blocking or opening of GIRK channels may keep the alpha cell in a clamped state of activation or inhibition, overwhelming any contribution of other membrane conductances or cAMP signaling to glucagon secretion. The pharmacological approach we used may thus not allow us to define the role of GIRK channels in restoring the alpha cell responsiveness to glucose. Alternatively, GIRK activation may not be the only determinant for glucose-mediated restoration of glucagon secretion. Our data using TPQ and low-dose diazoxide suggest that there is enough membrane potential hyperpolarization to allow for depolarization-induced glucagon secretion in the habituated state. The amount of GIRK (and KATP) activity in alpha cells at 1 mM glucose would also be predicted to be low, which suggests that small changes in potassium flux can still have a large impact on glucagon secretion. GIRK channels are clearly potent modulators of glucagon secretion, but loss of paracrine activation of GIRK channels alone cannot explain habituation of glucagon secretion.

The ability to restore glucagon secretion in islets from donors with long-duration T1D by reintroducing paracrine signals underscores the importance of inhibitory pathways in maintaining alpha cell function. Targeting these mechanisms could offer therapeutic strategies to prevent dysregulated glucagon secretion.⁴³ These results also suggest that the loss of glucagon secretion seen in diabetes is not necessarily due to irreversible, intrinsic alpha cell dysfunction but to a changing environment that fails to provide inhibitory priming signals. In the absence of beta cell input, alpha cells remain in a prolonged habituated state that helps explain their glucose blindness in type 1 diabetes.

Limitations of the study

Obtaining human pancreatic islets, especially from T1D donors, is challenging. While we included islets from multiple donors to capture biological variability, our sample size for T1D donors was small. Receiving additional islets for functional studies would require several years. As a result, we could not perform all experiments with T1D islets and prioritized those most relevant for understanding alpha cell regulation. Unfortunately, we could not test the effects of GIRK channel modulators on glucagon secretion in islets from T1D donors.

Recent studies suggest that the inhibitory influence of delta cells may increase in type 1 diabetes,^29,44 which could explain the reduced glucagon response to hypoglycemia. We did not address this possibility in our studies. How the dynamics and magnitude of endogenous secretion of SST and its effects on alpha cells change in islets from donors with T1D could shed light on the contribution of this potential mechanism.

We used a patch-clamp protocol to reveal the GIRK current that required artificial conditions. Additional experiments are needed to examine GIRK outward K⁺ currents at physiological voltages and their impact on the electrical behavior of the alpha cell. A further limitation is that TPQ can block both GIRK and BK channels.⁴⁵ Because BK channels have been shown to play an important role in human glucagon secretion,¹³ it is important to acknowledge that the effects could, at least in part, be mediated by BK or other ion channels.

We could not use mouse models in our studies because murine alpha cells differ in gene expression, electrophysiological properties, and dependence on paracrine signaling. Notably, GIRK channel expression is very low in mouse alpha cells (Figure S7,⁴⁶), making it difficult to investigate their role using the versatile mouse model. Despite these challenges, our study provides insight into how alpha cells integrate paracrine signals to regulate glucagon secretion. By focusing on human islets, we revealed key mechanisms that may serve as therapeutic targets to improve alpha cell function in diabetes.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human organ donors

Human pancreatic islets from non-diabetic human donors were obtained from Prodo Laboratories, Inc. Prodo Labs obtains human pancreases from deceased organ donors with documented informed consent for research purposes, secured properly and legally by the organ recovery organizations. Because the tissues are from brain-dead, de-identified donors and the donation is for research, the University of Miami Institutional Review Board (IRB) considers the use of the tissue as “Exempt” from human studies approval. Pancreatic islets from donors with type 1 diabetic were procured by Dr. Alvin Powers (Vanderbilt University) and Dr. Rita Bottino (Director Islet Programs, Imagine Pharma). Sample sizes are indicated in the Figure Legends. Because of the inherent variability of human samples, we used a minimum of 5 donors for each experiment. Each donor serve as its own control. We obtained only 4 samples from donors with type 1 diabetes. Detailed donor information is listed in Tables S1 and S2.

METHOD DETAILS

Dynamic measurements of hormone secretion

To assess glucose-stimulated hormone release of human pancreatic islets, 120–150 human islets were placed in perifusion columns (Biorep Technologies, Cat# PERI-CHAMBER) and connected to an automated perifusion system. Perifusion was performed with HEPES-buffered solution (125 mM NaCl, 5.9 mM KCl, 2.56 mM CaCl2, 1 mM MgCl2, 25 mM HEPES, 0.1% BSA, pH 7.4) containing different glucose concentrations and 2 mM amino acids (L-alanine, L-arginine and L-glutamine (Sigma)). Islets were perfused at a flow rate of 100 μL/minute and samples were collected every minute in 96-well plates. Prior to each perifusion protocol, islets were flushed for 90 min with baseline HEPES buffer (5.5 mM glucose) to wash out accumulated hormones and enzymes. Perifusion protocols include stimulation with different glucose concentrations, paracrine inhibitors (somatostatin 1 μM, Tocris; GABA 100 μM, Tocris and serotonin 100 μM, Sigma) and/or GIRK modulators (tertiapin-Q 200 nM, Tocris and ML-297 10 μM, Tocris) and KCl (30 mM). Perfusates were stored at −20°C until glucagon/insulin content was measured using commercially available ELISA and Lumit kits (Mercodia, Cat# 10-1113-01 and 10-1281-01; Promega, Cat# W8022 and CS3037A07).

Whole-cell patch clamp electrophysiology

Human islets were dispersed with enzyme free cell dissociation buffer (Gibco, Cat# 13150-016) and plated and incubated overnight (37°C) on coated cover slips (Thermo Fisher, Cat # 174950). Whole-cell patch clamp experiments were performed on pancreatic alpha cells. Alpha cell identity was established by responses to kainate (100 μM, Tocris). Recordings were made using an inverted Nikon microscope (Eclipse TE2000-U) custom-made for fluorescence and equipped with a 20 × 0.8NA and 40 × 0.8NA objectives. Cells were continuously perfused with a bath solution (in mM): 140 NaCl, 20 KCl, 0.5 CaCl₂, 2 MgCl₂,10 HEPES (pH 7.4). Recordings were made at room temperature ~22°C. Patch pipettes were pulled from borosilicate glass (outer diameter, 1.5 mm; inner diameter, 0.86 mm, Sutter Instruments) using a horizontal puller (P-97 Sutter Instruments). Electrode resistances were between 2 and 5 MΩ when filled with the intracellular solution (in mM): 20 NaCl, 130 KCl, 5 EGTA, 5.46 MgCl₂, 2.56 K₂ATP, 0.3 LiGTP, 10 HEPESHepes (pH 7.4). Voltage-clamp recordings were acquired using an Axopatch 200B amplifier (Molecular Devices, San Jose, CA, USA), digitized at 10 kHz (filtered at 5 kHz, Axon Digidata 1550; Molecular Devices), and collected using pClamp 11 (Axon Instruments, San Jose, CA, USA). Cells were held at −40 mV for 50 ms followed by a ramp protocol from −120 to +50 mV for 500 ms before returning to the −40 mV holding potential. This protocol is conventionally used to reveal GIRK channel currents.⁴⁷

QUANTIFICATION AND STATISTICAL ANALYSES

Quantification of hormone measurements

Quantification of hormone release was measured using commercially available ELISA and Lumit kits (Mercodia, Promega). Concentrations of insulin/glucagon were obtained by plotting the absorbance or luminescence of calibrators versus their concentration using sigmoidal four parameter logistic with Prism 9 software (GraphPad software, La Jolla, CA). Data is presented either normalized to average baseline secretion (fold) or as percent of total hormone content (%). Hormone content was measured from lysates obtained from all islets used after the experiment.

Statistical analyses

For statistical comparisons we used Prism 9 (GraphPad software, La Jolla, CA) and performed Student’s t tests (unpaired) or one-way/two-way analysis of variance (ANOVA) corrected for multiple comparisons (using Tukey’s Multiple Comparison Test; each row represented matched data). p-values <0.05 were considered statistically significant. Throughout the manuscript we present data as mean ± SEM.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117068.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Human islets	Prodo Laboratories Inc.	N/A
Human islets from T1D donors	Imagine Pharma	N/A
Chemicals, peptides, and recombinant proteins
Baclofen	Tocris	Cat# 0796
Diazoxide	Sigma	Cat# D9035
Epinephrine	Sigma	Cat# E4250
Forskolin	Sigma	Cat# F6886
GABA	Tocris	Cat# 344
Kainic acid	Tocris	Cat# 7065
Serotonin	Sigma	Cat# H9523
Somatostatin	Tocris	Cat# 1165
Tertiapin-Q	Tocris	Cat# 1316
ML297	Tocris	Cat# 5380
Critical commercial assays
Lumit® Insulin Immunoassay Kit	Promega	Cat# CS3037A07
Lumit® Glucagon Immunoassay Kit	Promega	Cat# W8022
Software and algorithms
Prism	GraphPad	Prism 10
pClamp 11	Axon Instruments	N/A
OriginPro 2021b	OriginLabs Northampton, MA	N/A
Other
Automated perifusion system with collection tray	PG Fluidics LLC	Cat# AuraFlow1
Perifusion system with automated tray handling	Biorep Technologies	Cat# Peri4-02-230-FA
Axopatch 200B amplifier	Molecular Devices, USA	N/A
Axon Digidata 1550	Molecular Devices, USA	N/A
P-97 Sutter Instruments	Sutter Instruments	N/A

Highlights.

• Efficient glucagon secretion from pancreatic alpha cells prevents hypoglycemia

• Signals from beta and delta cells activate GPCRs that open inhibitory GIRK channels

• Loss of this inhibition makes alpha cells refractory to changes in glucose levels

• Restoring these signals in vitro effectively reestablishes impaired glucagon secretion