Leucine Suppresses α-Cell cAMP and Glucagon Secretion via a Combination of Cell-Intrinsic and Islet Paracrine Signaling Free

Glucagon secretion from pancreatic islet α-cells plays a crucial role in maintaining blood glucose under fasting and hypoglycemic conditions, and supports insulin secretion in the fed state via signaling to neighboring β-cells (1). Glucagon is released through Ca²⁺-dependent exocytosis that is regulated by the electrical activity of the α-cell and the endoplasmic reticulum, as well as by cAMP, which acts by directly stimulating the exocytotic machinery and recruiting additional glucagon granules into the readily releasable pool (2). Although many studies establish the respective roles of α-cell Ca²⁺ and cAMP in regulating the timing and magnitude of glucagon secretion, nutrient regulation of α-cell activity remains poorly understood (2,3).

Glucose is an inhibitor of glucagon secretion that regulates α-cell activity through both paracrine and intrinsic mechanisms. At elevated glucose concentrations, β-cell activation suppresses glucagon secretion by releasing several inhibitory factors that include insulin, GABA, Zn²⁺, urocortin 3, and serotonin/5-HT (4–8). Insulin, despite reducing α-cell cAMP immunofluorescence (4), was later found to reduce glucagon secretion independently of cAMP (9,10), whereas urocortin 3 is coreleased with insulin and suppresses α-cell cAMP by stimulating somatostatin release from δ-cells (4,11,12,13). Antagonism of somatostatin receptor 2 (Sstr2) increases glucagon secretion at both low and high glucose concentrations, establishing the importance of both tonic- and glucose-stimulated paracrine inhibition by δ-cells (4,9,14). However, glucagon secretion is maximally inhibited by glucose at concentrations that do not stimulate insulin release, including in somatostatin knockout mice (15–17), highlighting the importance of α-cell–intrinsic regulation by glucose (18).

Substantial research on α-cell metabolism has focused on the ability of glycolysis to suppress glucagon secretion (17,19–22). Glucose reinforces its own metabolism by switching off mitochondrial β-oxidation via inhibition of CPT1 fatty acid transporters (23,24). However, the high level of lactate dehydrogenase in α-cells shunts glucose-derived pyruvate away from mitochondria (25–27) such that glycolysis is poorly coupled to oxidative phosphorylation (24,26,28). Part of the glucagonostatic effect of glucose may be due to ATP/ADP or lactate-dependent regulation of the K_ATP channel (15,19,24,25,27), although glucose also suppresses glucagon secretion in the presence of K_ATP channel blockers (16,17,29). Consistently, glucose intrinsically suppresses α-cell cAMP (9) and glucagon exocytosis (30,31) independently of Ca²⁺ regulation. Although these studies establish a crucial role for intermediary metabolism in the regulation of glucagon secretion, a majority have been conducted with glucose as the only substrate.

In addition to glucose, α-cells are sensitive to most amino acids (32,33). Glucagon release is potently stimulated by the electrogenic fuels arginine and alanine, which are thought to depolarize the α-cell plasma membrane and stimulate Ca²⁺ influx (34–36). Glutamine is not a potent secretagogue but is important for α-cell proliferation (32), whereas glutamate and glycine stimulate glucagon secretion through α-cell membrane receptors (37,38). One amino acid that has not received significant attention is leucine, which is primarily considered a β-cell fuel. Leucine is the only amino acid that is sufficient to stimulate insulin secretion at low glucose. As an oxidative and anaplerotic mitochondrial fuel, leucine metabolism generates acetyl-CoA for oxidation in the TCA cycle and allosterically activates the anaplerotic enzyme glutamate dehydrogenase (GDH) (39,40). Although gain-of-function mutations in GDH cause β-cell hypersecretion and protein-induced hypoglycemia (41), it is not clear whether leucine regulates α-cell activity.

Here, we show that physiologic concentrations of leucine suppress glucagon secretion from both mouse and human islets. Leucine lowers α-cell cAMP, independently of α-cell Ca²⁺, ATP/ADP ratio, or mitochondrial fatty acid oxidation. Although activation of β/δ-cells contributes to its glucagonostatic effect, leucine also acts directly on the α-cell. The suppressive effect of leucine on cAMP is mimicked by several different mitochondrial fuels and blocked by the mitochondrial poison cyanide. Although glucose also suppresses cAMP, the higher potency of leucine allows it to work more effectively under hypoglycemic and euglycemic conditions. Together, our data suggest a central role of mitochondria in the regulation of glucagon secretion through the α-cell–intrinsic suppression of cAMP.

Human islets were obtained from the Alberta Diabetes Institute Islet Core, Prodo Laboratories, and the Integrated Islet Distribution Program. The age, sex, BMI, and HbA_1c percentage of each donor are provided in Supplementary Table 1. Islets were cultured in Pim(R) Islet Media (Prodo Laboratories) for a period of 24–48 h after being received. This study used Integrated Islet Distribution Program data from the Organ Procurement and Transplantation Network. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy of or interpretation by the Organ Procurement and Transplantation Network or the U.S. Government.

Wild-type C57BL/6J mice (age 12–15 weeks) were obtained from The Jackson Laboratory. GcgCre^ERT mice (42) (Jax 030346) were crossed with GCaMP6s mice (Jax 028866), a Cre-dependent Ca²⁺ indicator strain, and CAMPER mice (Jax 032205), a Cre-dependent cAMP indicator strain. As in our previous study (34), the mice were not treated with tamoxifen (additional details are provided in the next section). Islets were isolated and cultured as in the study by Foster et al. (43) and dispersed as in that by Ho et al. (25). All procedures involving animals were approved by the Institutional Animal Care and Use Committee of the William S. Middleton Memorial Veterans Hospital and followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Reagents were obtained from Sigma-Aldrich unless indicated otherwise. For imaging of α-cell Ca²⁺, islets from GcgCre^ERT:GCaMP6s mice were cultured in the presence of 100 nmol/L 4-hydroxytamoxifen and imaged 3 days postisolation. However, we found that 4-hydroxytamoxifen was unnecessary for GCaMP6s expression, provided that GcgCre^ERT was present in the dam (Supplementary Table 2). Leveraging this breeding strategy to include GcgCre^ERT in both the sire and dam, islets isolated from GcgCre^ERT:CAMPER mice were not treated with 4-hydroxytamoxifen. Expression of GCaMP6s and CAMPER was not observed in the absence GcgCre^ERT. For imaging of α-cell cAMP, islets were isolated from GcgCre^ERT:CAMPER mice and imaged the next day, except in circumstances shown in Figs. 3G and 6G and Supplementary Fig. 4, where islets from GcgCre^ERT mice were cultured for 2 h with Cre-dependent CMV-FlexOn-CAMPER adenovirus (VectorBuilder), transferred to media containing 100 nmol/L 4-hydroxytamoxifen, and imaged 2 days postisolation. For imaging of α-cell ATP/ADP, islets from GcgCre^ERT mice were cultured for 2 h with Cre-dependent CMV-FlexOn-Perceval-HR adenovirus (VectorBuilder), transferred to media containing 100 nmol/L 4-hydroxytamoxifen, and imaged 3 days post-isolation. Islets were imaged in an RC-41LP imaging chamber (Warner Instruments) mounted on a Nikon Ti2 microscope equipped with a 10×/0.5 NA SuperFluor objective (Nikon). The imaging solution contained 137 mmol/L NaCl, 5.6 mmol/L KCl, 1.2 mmol/L MgCl₂, 0.5 mmol/L NaH₂PO₄, 4.2 mmol/L NaHCO₃, 10 mmol/L HEPES, and 2.6 mmol/L CaCl₂ (pH 7.4). Metabolites, mouse GIP (Pheonix Pharmaceuticals), CYN154806 (Tocris Bioscience), and cyanide were added as indicated. The flow rate was maintained at 0.25 mL/min using a feedback-controlled flow cell (Fluigent MCFS-EZ), and the temperature was maintained at 33°C using solution and chamber heaters (Warner). Filters sets (excitation, dichroic, emission) were as follows: Perceval-HR (511/16, 434/21, FF459/526/596-Di01, FF01-542/27; Semrock), CAMPER (434/21, FF459/526/596-Di01, FF02-470/30 and FF01-542/27; Semrock), and GCaMP6s (ET500/20x, ET-Fura2-GFP, ET535/30m; Chroma). Fluorescence emission was collected with a Photometrics Prime 95B or Hamamatsu ORCA-Flash4.0 CMOS camera every 6 s. A single region of interest was used to quantify the average response of each islet using NIS-Elements software (Nikon), with cAMP levels reported as the CAMPER emission ratio R470/542 normalized to the initial condition (R/R_o), ATP/ADP reported as the Perceval-HR excitation ratio R511/434 (R/R_o), and Ca²⁺ reported as the fluorescence intensity normalized to the initial condition (F/F_o). Single α-cell quantification of Perceval-HR, CAMPER, and GCaMP6s was performed with the Imaris Spots tool (Andor).

For each assay, islets from 12 mice or from a single human donor were pooled and then divided into a 12-chamber perifusion system (BioRep PERI5) with 78–100 islets per chamber. Each independent experiment is presented on a separate graph, and the number of technical replicates (corresponding to the number of chambers) is indicated in the figure legend for each experimental condition. Islets were equilibrated at 2 mmol/L glucose for 48 min at 37°C in KRPH buffer (140 mmol/L NaCl, 4.7 mmol/L KCl, 1.5 mmol/L CaCl₂, 1 mmol/L NaH₂PO₄, 1 mmol/L MgSO₄, 2 mmol/L NaHCO₃, 5 mmol/L HEPES, and 0.1% BSA [pH 7.4]) with 100 μL Bio-Gel P-4 (cat. no. 1504124; Bio-Rad) before stimulation, as described in the figure legends. An auxiliary peristaltic pump (Golander Pump), which bypasses the PERI5 pump and manifold, was used in circumstances shown in Figs. 11, 1J, 2E and F, and 7A–FSupplementary Figs. 1 and 5A and B. The perifusion flow rate was 200 μL/min, except in circumstances shown in Figs. 1I and J, 2E and F, and 7A–F and Supplementary Figs. 1 and 5A and B, when the flow rate was reduced to 110 μL/min to resolve basal glucagon release. Insulin and glucagon immunoassays (Promega CS3037A01 and W8022) were used with a TECAN Spark plate reader.

Fatty acid oxidation was measured as in the study by Armour et al. (24), with the addition of physiologic levels of glutamine and leucine as indicated in the figure legend.

Both males and females were analyzed as indicated in Supplementary Table 1; however, sex was not considered a factor in the statistical analysis. All data are presented as mean ± SEM and were analyzed by area under the curve (AUC). For perifusion assays, AUCs were analyzed using an unpaired t test for two conditions, whereas one-way ANOVA was used when comparing more than two conditions. Imaging data were normalized to the control condition (R/R_o or F/F_o). Unpaired t tests were used when comparing the AUCs between control and experimental conditions from two treatment groups. Paired t tests were used when comparing each imaging condition to the preceding (control) condition (note that ANOVA is inappropriate in this case because only two groups can be compared in a controlled fashion).

The data sets generated and analyzed during the current study are available from the corresponding author on reasonable request.

Islet perifusion was used to simultaneously assess insulin and glucagon secretion from mouse islets stimulated with glucose (2 and 10 mmol/L) and mixed amino acids provided at three times their physiologic concentrations in the portal circulation (LQAR; i.e., L, 1.5 mmol/L leucine; Q, 1.8 mmol/L glutamine; A, mmol/L 6.3 alanine; R, mmol/L 0.6 arginine) (34). At low glucose, LQAR induced a large biphasic rise in glucagon secretion that was further augmented by 10 nmol/L glucose-dependent insulinotropic peptide (GIP) (Fig. 1A), consistent with the glucagonotropic effects of arginine/alanine and stimulation of the α-cell GIP receptor (34). LQAR- and GIP-stimulated glucagon release was dramatically reduced by elevated glucose (Fig. 1A). The LQAR amino acid mixture was sufficient to stimulate insulin secretion at low glucose and amplified insulin secretion at high glucose (Fig. 1B).

Similarly to glucose, leucine strongly suppressed glucagon secretion stimulated by the glucagonotropic amino acid mixture QAR, including when GIP was present (Fig. 1C). As expected, leucine alone was sufficient to stimulate insulin secretion, which was further amplified by QAR and GIP (Fig. 1D). Additionally, at 10 mmol/L glucose, leucine suppressed glucagon secretion in the presence of QAR and GIP (Fig. 1E), while potentiating insulin secretion (Fig. 1F). We further tested leucine in the presence of glutamine, because it could be important for maintaining the glutamate pool, which in β-cells feeds the TCA cycle (39) and in α-cells boosts glucagon secretion via autocrine feedback on glutamate receptors (37). Glutamine had no significant effect on glucagon secretion by itself or in the presence of LQAR (Fig. 1G); however, it boosted insulin secretion approximately fourfold in the presence of LQAR (Fig. 1H). Leucine was effective at suppressing glucagon secretion in response to alanine and arginine, regardless of whether glutamine was present (Fig. 1G). Taken together, these findings suggest an important role of leucine in the negative regulation of α-cells at both low and high glucose concentrations.

In the above assays, it was surprising that we did not observe a glucose-dependent reduction in glucagon secretion, an effect reported by many laboratories (e.g., Lai et al. [18]). Because glucagon release is near the limit of detection in the absence of amino acids, we reduced the perifusion flow rate (from 200 to 110 μL/min) to increase the effective concentration of glucagon measured by the immunoassay. At this reduced flow rate, glucagon secretion was reduced by glucose alone (Fig. 1I). A similar glucagonostatic effect was observed in response to leucine (Fig. 1J). Both glucose and leucine increased insulin secretion in these assays (Supplementary Fig. 1). The experiments using this reduced flow rate are listed in Research Design and Methods.

We next used perifusion assays to test the effect of leucine on primary human islets. Leucine stimulated insulin secretion and suppressed QAR-stimulated glucagon secretion in each of the first four preparations tested (Fig. 2A–D). On a technical note, we found that α-cells from human islets were more sensitive than mouse islets to flow changes, which in three of four preparations (Fig. 2A–C but not 2 D) resulted in artifactual increases in glucagon secretion during fraction collector plate changes. This spurious effect was most apparent when the black control traces, which did not receive leucine, exhibited an apparent increase in glucagon release (denoted by arrows in Fig. 2A–C). Using an auxiliary pump to ensure a constant flow rate, we observed that leucine suppressed QAR-stimulated glucagon secretion in two additional human islet preparations without any technical artifacts (Fig. 2E and F). Because all six experiments were controlled, we conclude that leucine suppresses glucagon secretion in human as well as mouse islets.

To determine whether leucine suppresses glucagon secretion through α-cell Ca²⁺, mouse islets were isolated from GcgCre^ERT:GCaMP6s mice. At low glucose, leucine had only a minor effect, whereas QAR strongly increased α-cell Ca²⁺ (Fig. 3A). Separating the amino acids, glutamine reduced Ca²⁺, whereas alanine and arginine strongly increased Ca²⁺ (Supplementary Fig. 2). These findings are consistent with prior reports stating that arginine and alanine depolarize the α-cell plasma membrane (34–36). Although the effect was small, leucine potentiated the glutamine-stimulated reduction in Ca²⁺, while increasing the Ca²⁺ response to alanine and arginine (Supplementary Fig. 2), the condition in which glucagon secretion was maximally suppressed by leucine (Fig. 1). It is therefore unlikely that a Ca²⁺-dependent mechanism explains the glucagonostatic effect of leucine.

In islets isolated from GcgCre^ERT:CAMPER mice, α-cell cAMP was strongly suppressed by LQAR when delivered at three times the physiologic concentration (Fig. 3B). LQAR also reduced cAMP at physiologic concentrations (Supplementary Fig. 3A). Single-cell analysis (displayed as a heatmap with one cell per row in Fig. 3B, left) revealed a heterogenous response to LQAR, even within α-cells of the same islet. Pulses of cAMP were observed in a subset of α-cells in the presence of 2 mmol/L glucose alone; these pulses were blocked by leucine (Fig. 3B, right). Mouse GIP, which acts directly on the α-cell via GIP receptors (34), was used as a positive control that raises cAMP (Fig. 3B).

Next, we separated the amino acids to determine their individual versus combined effects on α-cell cAMP. Alanine/arginine caused a transient decrease in α-cell cAMP, whereas leucine/glutamine resulted in a more pronounced cAMP reduction (Fig. 3C and Supplementary Fig. 3B). Leucine alone was sufficient to decrease α-cell cAMP, and the addition of QAR had no further effect (Fig. 3D). The provision of glutamine before leucine did not potentiate the drop in cAMP or change the kinetics of the drop (Fig. 3E and F), consistent with the islet perifusion assays (Fig. 1). Leucine did not affect the magnitude of the α-cell GIP response, suggesting separable mechanisms for cAMP regulation; however, the absolute level of cAMP was reduced in the presence of leucine and GIP (Fig. 3G). These data are consistent with the ability of leucine to reduce glucagon secretion in the presence of QAR and GIP (Fig. 1C).

We next performed dose-response curves to determine the IC₅₀ of leucine at varying levels of glucose. Leucine was applied to intact islets between 15 μmol/L and 1.5 mmol/L, which encompassed the fasting and nonfasting concentrations of leucine as measured from the tail vein of mice (108 and 167 μmol/L, respectively) (18,44), as well as the higher postprandial concentration found in the portal blood (501 μmol/L) (34). As glucose was elevated from 2 to 6 mmol/L, the IC₅₀ of leucine increased from 57 ± 19 μmol/L (n = 68 islets; n = 3 mice) to 440 ± 112 μmol/L (n = 73 islets; n = 3 mice). In the presence of 10 mmol/L glucose, leucine suppressed α-cell cAMP starting at 150 μmol/L, and the IC₅₀ increased to 1,162 ± 872 μmol/L (n = 71 islets; n = 3 mice) (Fig. 3H). Combined, these data emphasize that leucine suppresses α-cell cAMP at physiologic levels and across a range of glucose concentrations.

In mouse islets, prior studies with the Sstr2 antagonist CYN154806 (CYN) demonstrated an important role of somatostatin regulation of the α-cell at both low and high glucose concentrations (9,18). Both effects were confirmed in our experiments with glucose, which was used as a positive control. The addition of 500 nmol/L CYN increased glucagon secretion at low glucose (Fig. 4A). Also in the presence of CYN, glucose elevation stimulated insulin secretion and reduced LQAR-stimulated glucagon secretion (Fig. 4A and B). Therefore, our control data indicate Sstr2-dependent inhibition of glucagon release at low glucose, as well as Sstr2-independent suppression of glucagon secretion at high glucose. Similar to glucose, leucine effectively suppressed alanine- and arginine-stimulated glucagon secretion in the presence of CYN (Fig. 4C, left). A comparable glucagonostatic effect of leucine was observed in the added presence of glutamine (Fig. 4C, right). As in earlier experiments, leucine enabled QAR-stimulated insulin release (Fig. 4D).

In parallel with the glucagon secretion measurements, Sstr2 inhibition strongly increased α-cell cAMP in the presence of 2 mmol/L glucose (Fig. 4E). The further addition of glucose effectively lowered α-cell cAMP in the presence of CYN, consistent with δ-cell independent regulation of α-cell cAMP (9). Similarly, LQAR and leucine decreased α-cell cAMP with CYN present at both low and high glucose (Fig. 4F–H). Collectively, these data suggest both Sstr2-dependent and Sstr2-independent mechanisms for leucine suppression of α-cell cAMP and glucagon secretion.

To test for an α-cell–intrinsic effect of leucine in intact islets, we used the K_ATP channel opener diazoxide (200 μmol/L) to block Ca²⁺-dependent exocytosis from α-, β-, and δ-cells. Diazoxide increased α-cell cAMP (Fig. 4I–M) to a similar extent as CYN (Fig. 4E–H), which we interpret as a loss of inhibitory paracrine signaling. Sequential application of CYN and diazoxide was used to evaluate this conclusion. In the presence of 2 mmol/L glucose and CYN, diazoxide further increased cAMP, implying that not all paracrine inhibition of α-cell cAMP occurs via Sstr2 (Supplementary Fig. 4). Additionally, in the presence of 2 mmol/L glucose and diazoxide, CYN further increased cAMP levels, indicating that Sstr2 has some tonic activity at low glucose. These results clarify the impact of CYN and diazoxide on α-cell cAMP and confirm the previously reported role of Sstr2 (18).

LQAR effectively lowered α-cell cAMP in the presence of diazoxide, but not to the same level as in its absence (Fig. 4I), consistent with an α-cell–intrinsic effect of LQAR on cAMP in addition to a paracrine effect. GIP was effective at raising α-cell cAMP in the presence of diazoxide (Fig. 4I), as expected for a direct effect on the α-cell GIP receptor (34). In contrast to leucine, alanine/arginine induced only a transient decrease in cAMP, although this effect was reversed by diazoxide (Fig. 4J) and is therefore likely to be Ca^{2 +} or membrane potential-dependent. Similar effects of alanine/arginine on cAMP were observed at physiologic concentrations (A, 2.1 mmol/L; R, 0.2 mmol/L) (34) (Supplementary Fig. 3B). Leucine, alone or in combination with glutamine, strongly suppressed cAMP in the presence of diazoxide (Fig. 4I–M), suggesting that leucine is effective in the absence of paracrine signaling. This effect is also Ca²⁺ independent because diazoxide blocked LQAR-stimulated changes in Ca²⁺ (Fig. 4N). Finally, we used islet dispersion to eliminate paracrine and juxtacrine signaling completely. In isolated α-cells from GcgCre^ERT:CAMPER mice, leucine inhibited α-cell cAMP, with no further effect of QAR (Fig. 4O). Taken together with the CYN and diazoxide experiments in intact islets, our findings in dispersed α-cells confirm an α-cell–intrinsic effect of leucine that suppresses cAMP.

Because leucine and glucose inhibit cAMP and glucagon secretion in a similar manner, we next considered whether leucine suppresses fatty acid oxidation and cytosolic ATP/ADP (ATP/ADP_c), similarly to glucose (Fig. 5A) (24). To do so, β-oxidation of ³H-labeled palmitate was quantified in isolated mouse islets (23,24). Reproducing the experiment performed in these studies, switching islets from 1 to 5 mmol/L glucose in the presence of nonesterified fatty acids (NEFA; 0.36 mmol/L) reduced fatty acid oxidation (Fig. 5B). However, the further addition of glutamine (0.6 mmol/L), alone or in combination with leucine (0.5 mmol/L), had no effect on fatty acid oxidation at either level of glucose.

Next, we tested the effects of leucine and glutamine on ATP/ADP_c at both low and high glucose. Islets isolated from GcgCre^ERT mice and expressing Cre-dependent Perceval-HR were used to investigate ATP/ADP_c selectively in α-cells. Although glucose exhibited a biphasic effect, first increasing and then reducing α-cell ATP/ADP_c, leucine had no significant effect on ATP/ADP_c at 2 or 10 mmol/L glucose (Fig. 5C and D). Glutamine by itself reduced ATP/ADP_c to a similar extent as 10 mmol/L glucose, and the effects of glucose and glutamine were additive (Fig. 5E and F). Overall, the lack of correlation between ATP/ADP_c and cAMP or glucagon secretion argues against ATP/ADP_c as an intermediary signaling mechanism.

To identify α-cell–intrinsic mechanisms linking leucine to cAMP regulation, we tested a series of mitochondrial fuels in the presence of diazoxide to limit paracrine effects from β/δ-cells (Fig. 6A); 10 mmol/L glucose suppressed α-cell cAMP to a similar extent as 1.5 mmol/L leucine, which had no significant effect when added after glucose (Fig. 6B and C). Application of α-ketoisocaproate (KIC; 10 mmol/L), which is transaminated to leucine (45), had a similar effect on α-cell cAMP as leucine and glucose (Fig. 6D). We next tested methyl-succinate, which undergoes a carbon-neutral exchange for malate before directly stimulating succinate dehydrogenase/complex II. Anaplerosis also occurs when the malate converted to pyruvate by malic enzyme cycles back into the mitochondria through pyruvate carboxylase (Fig. 6A); the low malic enzyme activity in mouse islets weakens this effect (46). Methyl-succinate (10 mmol/L) lowered α-cell cAMP, although to a lesser extent than the other fuels (Fig. 6E). These findings are consistent with a common mitochondrial mode of action. To directly test whether stimulation of mitochondrial anaplerosis is sufficient for cAMP regulation, we applied 2-aminobicyclo(2,2,1)heptane-2-carboxylic acid (BCH; 10 mmol/L), a nonmetabolizable leucine analog that activates GDH (40,47). BCH was sufficient to reduce α-cell cAMP, suggesting that GDH activation rather than acetyl-CoA generation underlies leucine suppression of cAMP (Fig. 6F). Although leucine had no effect when applied after BCH, this may be attributed to the off-target inhibition of the β-cell leucine transporter (47), making it difficult to ascertain whether the entire effect of leucine is via GDH. Taken together, these data suggest that mitochondrial metabolism facilitates the α-cell–intrinsic cAMP regulation by leucine, a mechanism that complements extrinsic α-cell regulation via β/δ-cells. Consistent with this interpretation, poisoning mitochondria with 5 mmol/L potassium cyanide (complex IV inhibitor) blocked the suppression of cAMP by leucine (Fig. 6G).

Because our initial hormone secretion experiments were carried out in the presence of 1.5 mmol/L leucine (Figs. 1 and 2), a dose that maximally suppresses cAMP (Fig. 3H), we were motivated to evaluate the islet response at lower concentrations. At low glucose, 0.1 and 0.5 mmol/L leucine dose-dependently inhibited glucagon secretion from mouse islets stimulated with alanine and arginine, respectively (Fig. 7A). Although 0.5 mmol/L leucine stimulated insulin secretion under these conditions, 0.1 mmol/L leucine did not, and neither concentration of leucine stimulated insulin release during the application of alanine and arginine. These effects are consistent with the direct effect of leucine on α-cells (Fig. 4). Similar conclusions were reached using the GDH activator BCH, which dose-dependently suppressed glucagon secretion stimulated by alanine and arginine when applied at 0.5 and 2.5 mmol/L, respectively, without stimulating insulin release (Fig. 7B). At the higher dose of BCH used in Fig. 6 (10 mmol/L), insulin secretion was robustly stimulated, and glucagon was strongly suppressed (Supplementary Fig. 5). Finally, we tested the effect of 0.1 and 0.5 mmol/L leucine on four human islet preparations (Fig. 7C–F), although the responses were not as uniform as the 1.5 mmol/L dose (Fig. 2). In the first preparation, insulin was stimulated by 0.5 mmol/L leucine without any effect on glucagon release (Fig. 7C). In the next two preparations, a glucagonostatic effect was observed only at 0.5 mmol/L leucine, and in these cases, insulin release was also stimulated (Fig. 7D and E). In the final preparation, leucine dose-dependently inhibited glucagon secretion in response to alanine and arginine without affecting insulin secretion (Fig. 7F), matching the results in mouse islets. Overall, these data support the conclusion that physiologically relevant concentrations of leucine suppress glucagon secretion. It is also clear, in mouse islets, that β-cell stimulation is a complementary mechanism to the intrinsic regulation of α-cells by leucine.

Our findings reveal a role of leucine in the regulation of α-cell cAMP and glucagon secretion. Like glucose (9), leucine exerts a direct suppressive effect on α-cell cAMP, in addition to stimulating inhibitory paracrine signals from β/δ-cells. Although the IC₅₀ of leucine increases with glucose, it remains relevant to the level of leucine in the portal circulation (34). While these ex vivo studies may not precisely match the in vivo situation, they suggest that leucine levels in the blood help to set the tone of glucagon secretion in the fasted state and under hypoglycemic conditions, whereas glucose and leucine share control over glucagon secretion in the fed state. Consistently, leucine suppressed glucagon secretion in the presence of GIP, which directly stimulates the α-cell and boosts β-cell function at elevated glucose (34). Most likely, leucine and GIP act on cAMP via separable mechanisms, dependent on metabolic and GPCR signaling, respectively. As discussed below, multiple lines of evidence suggest that leucine works via a mitochondrial mode of action.

Our data indicate that leucine regulates α-cells via a combination of α-cell–intrinsic and islet paracrine signaling, similarly to glucose (4,9,18). Leucine was found to extrinsically regulate α-cells via Sstr2, as expected for a β-cell fuel, whereas mitochondrial fuels, including leucine, BCH, KIC, methyl-succinate, and glucose, were each sufficient to suppress α-cell cAMP in the presence of diazoxide, indicating an intrinsic effect that was confirmed in isolated α-cells. Because diazoxide lowered α-cell Ca²⁺, these experiments argue against Ca²⁺-dependent phosphodiesterases or adenylate cyclases as a central mechanism by which mitochondrial fuels reduce α-cell cAMP. The GDH activator BCH, a purely anaplerotic fuel that is not metabolized to acetyl-CoA, was sufficient to suppress α-cell cAMP in the presence of diazoxide. The fact that glucose, which cannot activate GDH, had the same suppressive effect on cAMP suggests a general role of mitochondria in controlling cAMP production and setting α-cell tone. This interpretation is consistent with the ability of cyanide to block the effect of leucine on cAMP.

Although five different mitochondrial fuels suppressed α-cell cAMP (i.e., leucine, BCH, KIC, methyl-succinate, glucose), the link between fuel metabolism and cAMP remains elusive. From the evolutionary perspective, elevated cAMP is a universal cellular signal for energy deficit. It follows that cAMP levels should be low when fuel is replete. The best-known signaling mechanism for such cAMP regulation is the insulin/glucagon ratio that acts via GPCR signaling. However, as previously reported by the Tengholm laboratory (9), and supported by our studies, there may also be a cell-intrinsic mechanism for fuel surfeit to suppress cAMP that remains undiscovered. Ultimately, the role of mitochondrial metabolism must be clarified by generating α-cell knockouts of critical mitochondrial genes (e.g., Glud1, Pc, Pck2). Pdk4 overexpression in α-cells, used as a method to suppress pyruvate dehydrogenase, blocked glucose suppression of glucagon secretion, but cAMP remains to be examined (24). We note that mitochondria may not be the only mechanism for cAMP regulation by leucine, since mTOR signaling plays an important role in amino acid sensing (3), at least chronically (48).

α-Cells are remarkably sensitive to their environment, highlighting the importance of small differences between studies. It was previously reported that glucose elevation (from 1 to 5 mmol/L) reduced α-cell ATP/ADP_c in the presence of NEFA and increases ATP/ADP_c in its absence (24). However, the same study showed it is possible to avoid endogenous fatty acid depletion by preincubation with >1 mmol/L glucose; in this case, glucose elevation (from 3 to 20 mmol/L) reduced ATP/ADP_c in the absence of exogenous NEFA. Similarly, using 2 mmol/L glucose preincubation, we observed a glucose-dependent reduction in ATP/ADP_c, Ca²⁺ (not shown), and glucagon secretion in the absence of exogenous NEFA. In addition to environmental conditions, there may also be species differences that make it challenging to compare between studies. Prior studies examined the role of leucine in the perfused rat pancreas, with mixed results; a synopsis is provided by Leclercq-Meyer et al. (49). There is general agreement that leucine is inhibitory to glucagon output when β-cells are active; however, in contrast to our study, it was observed in the rat pancreas that 0.2 mmol/L leucine was stimulatory to glucagon output at low glucose. Ultimately, the studies are difficult to compare because rat islets are sensitive to glutamine deprivation, which was withheld by Leclercq-Meyer et al. (49). In mouse islets, leucine suppressed cAMP and glucagon secretion whether or not glutamine was present.

As a technical consideration, we successfully used GcgCre^ERT mice (42) without using tamoxifen. Although it should be possible to perform tamoxifen-inducible studies if GcgCre^ERT is present in the sire, spurious GCaMP6s fluorescence was observed in the absence of tamoxifen (or 4-hydroxytamoxifen) when GcgCre^ERT was present in the dam (Supplementary Table 2). Because tamoxifen itself is not inert (50,51), this feature is potentially advantageous for gene knockout; investigators using GcgCre^ERT mice should tailor the breeding strategy to the experimental goals.

Acknowledgments. The authors thank the organ donors and their families for their kind gifts in support of diabetes research, as well as the Human Organ Procurement and Exchange program and Trillium Gift of Life Network for their work in procuring human donor pancreata for research and Dr. Patrick MacDonald’s team (Alberta Diabetes Institute) for their efforts in human islet isolation. Graphics were created using BioRender.com.

Funding. The Merrins laboratory is supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institutes of Health (NIH; R01DK113103 and R01DK127637), IIDP (BS587P), and U.S. Department of Veterans Affairs Biomedical Laboratory Research and Development Service (I01BX005113) (M.J.M.). E.R.K. received a predoctoral fellowship from the NIH/NIDDK (F31DK134171). The Knudsen laboratory is supported by a Novo Nordisk Foundation Excellence Emerging Investigator Grant in Endocrinology and Metabolism (0054300) (J.G.K.) Human pancreatic islets were provided in part by the NIDDK-funded IIDP (RRID:SCR_014387) at City of Hope (UC4DK098085) and the JDRF-funded IIDP Islet Award Initiative.

This work used facilities and resources from the William S. Middleton Memorial Veterans Hospital and does not represent the views of the Department of Veterans Affairs or the U.S. Government.

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

Author Contributions. E.R.K., H.R.F., E.J., and M.H.E. performed experiments. E.R.K. and M.J.M. conceptualized the studies and drafted the manuscript. J.G.K. contributed resources. All authors reviewed/edited the manuscript. M.J.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 66th Biophysical Society Annual Meeting, San Francisco, CA, 19–23 February 2022.