Paracrine Hormonal Signals From Islet α-Cells Regulate Microtubule Dynamics in β-Cells to Promote Insulin Secretion in Mouse and Human Islets

Abstract

The microtubule network in β-cells attenuates insulin secretion by pulling insulin secretory granules away from the plasma membrane. Thus, high-glucose–induced microtubule remodeling is required for robust glucose-stimulated insulin secretion. We now demonstrate that hormones secreted by α-cells regulate microtubule dynamics in β-cells through receptors for glucagon (GcgR) and glucagon-like peptide 1 (GLP-1R). Activation of GcgR or GLP-1R destabilizes microtubules in β-cells, accompanied by increased insulin secretion. In contrast, inhibiting these receptors attenuates high-glucose–induced microtubule destabilization and decreases secretion. Supporting the physiological significance of this regulation, β-cells in islets with a higher α-cell–to–β-cell ratio exhibit more dynamic microtubules than those with a lower ratio, and a high-fat diet challenge in mice, which can compromise β-cell secretion, attenuates this effect in their islets. Within individual islets, β-cells located near α-cells show faster microtubule remodeling upon glucose stimulation than those more distant from α-cells. Consequently, islets with a higher α-cell–to–β-cell ratio secrete more insulin in response to glucose stimulation and plasma membrane depolarization, results recapitulated by exogenous glucagon stimulation or chemically induced microtubule destabilization in islets with lower α-cell–to–β-cell ratios. These combined results suggest that α-cells use glucagon-mediated and/or GLP-1–mediated paracrine signaling to fine-tune β-cell secretion via microtubule remodeling.

Article Highlights

• Glucagon/glucagon-like peptide 1 sensitizes glucose-induced microtubule remodeling in β-cells.

• Microtubule density in islets inversely correlates with the α-cell–to–β-cell ratio.

• Glucose-stimulated insulin secretion levels in single islets positively correlate with their α-cell–to–β-cell ratio.

• Glucagon and microtubule destabilization mobilize the same granule pool.

Introduction

The levels of insulin secretion from islet β-cells depend primarily on the level of Ca²⁺ influx that triggers secretion and the number of Ca²⁺-responsive insulin secretory granules (ISGs) (1,2). Glucose metabolism regulates both factors: it elevates the ATP-to-ADP ratio in β-cells, triggering membrane depolarization and Ca²⁺ influx to induce secretion. It also fuels ISG biosynthesis and mobilization, thereby increasing the pool of releasable ISGs.

Besides glucose, endocrine hormones also modulate Ca²⁺ influx and the pool of releasable ISGs. For example, glucagon-like peptide 1 (GLP-1) and glucagon from the enteroendocrine system and/or islet cells elevate intracellular cAMP levels (3–5), activating protein kinase A (PKA) and exchange proteins directly activated by cAMP (EPACs) (6). Activated PKA phosphorylates several channel and vesicular proteins to enhance Ca²⁺ influx and increases Ca²⁺ sensitivity of ISGs, thereby augmenting glucose-stimulated insulin secretion (GSIS) (6–8). Activated EPACs were suggested to promote secretion by both inducing cytoplasmic Ca²⁺ influx and facilitating ISG biosynthesis via microtubule (MT) nucleation at the Golgi apparatus (9,10).

MTs are biopolymers composed of α/β-tubulin dimers with distinct dynamics at their plus and minus ends. MTs maintain cell architecture and facilitate cargo transport via the motor proteins kinesin and dynein, which move toward the plus and minus ends, respectively (11,12). In β-cells, most MTs are derived from the Golgi and form a nondirectional meshwork, a feature that allows them to perform both canonical and noncanonical functions (13). Specifically, high-glucose stimulation promotes MT nucleation from the Golgi via a cAMP-dependent pathway in the cell center (10), thereby supporting a canonical function in ISG biosynthesis (14). In contrast, the nondirectional nature of the β-cell MTs prevents them from serving as tracks for directional ISG transport (13,15,16). Instead, the peripheral MTs withdraw ISGs from beneath the plasma membrane and reduce the pool of releasable ISGs (17). Therefore, high-glucose–induced destabilization and remodeling of peripheral MTs are critical for robust GSIS in β-cells (13,18,19). How glucose induces MT destabilization, however, remains incompletely understood.

We recently reported that MT-associated protein tau facilitated glucose-induced MT destabilization (18). Under basal glucose conditions, tau is poorly phosphorylated, allowing it to associate with and stabilize MTs in cells. Elevated glucose levels induce tau phosphorylation, which attenuates its affinity for MTs, leading to its dissociation and subsequent MT destabilization (18). Intriguingly, this process depends on the activities of several kinases, including GSK3, PKA, PKC, and CDK5, all of which can be activated by cAMP (20–22). These findings, combined with the roles of glucagon and GLP-1 in inducing cAMP production in β-cells, led us to test the hypothesis that islet α-cells regulate MT stability in β-cells by secreting glucagon and/or GLP-1.

Research Design and Methods

Animals and Human Donor Islets

The Vanderbilt University Institutional Animal Care and Use Committee (Nashville, TN) approved mouse usage. Both sexes of mice were used at a 1:1 ratio whenever possible. Euthanasia follows procedures recommended by the American Veterinary Medical Association. The Gcg^iCre mice (strain #030663) (23), C57BL6, R26R^eYFP (B6.129X1-Gt(ROSA)26SorTm1(EYFP)Cos/J), and R26^DTR (57BL/6-Gt(ROSA)26Sortm1(HBEGF)Awi/J) were from The Jackson Laboratory. Ins1^Apple mice were described in Brown et al. (24). Intraperitoneal diphtheria toxin (DT) injection was performed in 2-month-old mice at 20 ng each. Results from males and females were combined because they yielded identical results. The high-fat diet challenge (HFD, ∼60% of calories from fat) lasted 2 months in 2-month-old C57BL/6J mice. The control diet contained ∼13% of its calories in fat.

Human islets were from the Integrated Islet Distribution Program or A.N.B.’s islet cell laboratory in the University of Louisville, Louisville, KY (see Checklist for Reporting Human Islet Preparations Used in Research in the Supplementary Material), where informed consent was obtained from donors. Only islets with >80% purity, >95% viability, and a stimulation index above 2.0 were used for functional tests. Islets with dark cores composed of dead cells, verified via DAPI or trypan blue staining, were excluded.

Islet Isolation, Culture, Insulin Secretion, and Hormone Assays

Mouse pancreata were perfused with collagenase IV (Sigma) dissolved in Hank’s balanced salt solution, digested at 37°C, washed with RPMI 1640 (Gibco) containing 11 mmol/L glucose (G11) and 10% FBS (complete RPMI media), and hand-picked. Islets were then allowed to recover overnight in complete RPMI media.

For insulin secretion, islets were equilibrated in HEPES-buffered Krebs-Ringer buffer with 2.8 mmol/L glucose (G2.8) for 1 h at 37°C. Ten to 12 islets were then transferred into individual wells of 12-well plates, stimulated by G2.8 (basal low secretion, allowing more stable secretion than G5.5), G20 (20 mmol/L, supraphysiologically stimulatory but allows β-cell response monitor, similar to G16.5 in mouse or G11 in human islets), and G20K (20 mmol/L glucose plus 30 mmol/L KCl, ensuring high Ca²⁺ influx and ISG mobilization). Forty-five–minute windows were used to capture both the first and second phases of secretion (25). We have reported that osmolarity change did not affect secretion (18). Single-islet secretion assays used a similar approach with 96-well plates. Total insulin or glucagon was assayed after ethanol-HCl extraction with ELISA kits from Alpco and Mercodia, respectively. Glucagon assays in the entire pancreas were done similarly after ethanol-HCl extraction. Glucose tolerance tests were performed after overnight fasting and an intraperitoneal glucose injection of 2 g/kg body weight, with blood glucose read via a tail vein snip.

Pseudoislet Production

For mouse pseudoislet (PSI) production, FACS-purified mouse β-cells and non-β islet cells were mixed and aliquoted as 30-μL hang-drops (2,000 cells in each drop) in complete RPMI-1640 media for 4 days (24). For human PSI, dissociated islet cells were resuspended in Vanderbilt Pseudoislet Media or VPM (50% CMRL1066 +50% Vasculife Basal Media, 10% heat-inactivated FBS, 1× antibiotics, 0.5× Glutamax, 2.5 mmol/L HEPES, 0.5× sodium pyruvate plus recombinant human vascular endothelial growth factor, recombinant human epidermal growth factor, recombinant human fibroblast growth factor basic, recombinant human IGF-1, ascorbic acid, hydrocortisone hemisuccinate, heparin sulfate, gentamicin/amphotericin, and iCell Endothelial Cells Medium Supplement, from Lifefactor and MDsystems). Roughly 200,000 cells were aliquoted into each prepared Aggrewell of a 96-well plate, which was centrifuged at 200g for 5 min and left undisturbed in a cell culture incubator for 5 days.

Immunofluorescence and Microscopy in Single Islets

Islets were fixed in 4% paraformaldehyde +0.1% saponin (MilliporeSigma) and permeabilized with 50% DMSO + 0.2% Triton X–100 + 1× PBS for immunofluorescence (IF) staining. Antibodies used were mouse anti-E-Cadherin (BD), guinea pig anti-insulin (Dako), mouse antiglucagon (Dako), rabbit antiglucagon (MilliporeSigma), rabbit anti-somatostatin (anti-SST) (Jackson ImmunoResearch), goat anti-SST (Dako), and rabbit antidetyrosinated tubulin (DeTyr-tubulin) (MilliporeSigma). Primary antibodies were used at dilutions of 1:500 to 1:2,000, depending on the concentration. Secondary antibodies (1:1,000) were from Jackson ImmunoResearch. Image capture used Nikon Eclipse A1R, Zeiss LSM 880 Airyscan, or Olympus FV1000 laser scanning confocal microscopes (40×, 63×, or 100× objectives). For each islet, multiple Z-stacked images were taken. Different cell types or MT levels in several slices of each islet (two to five, depending on islet size or whether cell type or DeTyr-tubulin was quantified) were quantified with Image J.

MT Dynamic Time Course Assays

Islets were stimulated with 16.5 mmol/L glucose and fixed at 0, 15, 30, or 60 min poststimulation. Insulin, glucagon, and DeTyr-tubulin were stained with whole-mount IF. Imaging and quantification were performed as described above.

Ca²⁺ Recording

Islets isolated from 2- to 3-month-old Ins1^Apple;R26R^eYFP;Gcg^iCre mice were attached to glass-bottomed plates (D35-14-1.5P; Cellvis) for 3 days in RPMI 1640 complete media, loaded with 20 μmol/L Cal-630-AM for 45 min at 37°C, and equilibrated in Krebs-Ringer buffer. Cal-630 fluorescence was measured (excitation at 640 nm, emission at 684 nm) under G2.8, G20, and 30 mmol/L KCl every 10 s using a Nikon Crest V3 spinning disk confocal system with a Ti2-E inverted microscope (20× magnification), Sona 4.2B-11 camera (Andor), and Celesta light engine (Lumencor).

Chemical/Agonist/Antagonist Treatment for MT Imaging

Islets were pretreated in complete RPMI media containing G2.8 plus chemicals (MT-modifier [nocodazole (Noc), 10 ng/mL], agonists [glucagon or liraglutide, 100 nmol/L each], or Adomeglivant [1 µmol/L]/exendin 9 [20 µmol/L]) for 1 h. Islets were split into two groups: one group was stimulated with 20 mmol/L glucose and the other with G2.8. Fixation and IF staining then followed.

Statistical Analysis

Student t test, one-way or two-way ANOVA, and Pearson correlation analyses were used. P values <0.05 were considered significant.

Data and Resource Availability

G.Q.G. will fulfill reasonable data/reagent requests.

Results

GcgR/GLP-1R Activation Induces MT Destabilization in β-Cells

To test the effects of glucagon or GLP-1 signaling on β-cell MT, mouse islets were treated with either agonists or antagonists of GLP-1R or GcgR (Fig. 1A). MT stabilities were then compared after 2 h of incubation with either G2.8 or G20 (Fig. 1A). Using DeTyr-tubulin as a marker for long-lived MTs (18,26), we found that G20 induced substantial MT destabilization in mock-treated β-cells (Fig. 1B–E and V). A GLP-1R agonist, liraglutide, destabilized MTs under basal glucose, while a GLP-1R antagonist, exendin-9, attenuated G20-induced MT disassembly (Fig. 1F–M and V). Similarly, exogenous glucagon destabilized MTs in β-cells at basal glucose levels; the GcgR-specific antagonist adomeglivant attenuated G20-induced MT destabilization (Fig. 1N–U and W). Combining G20 with liraglutide or glucagon did not induce more MT destabilization than each single reagent, whereas exendin-9 and adomeglivant significantly reduced G20-induced MT destabilization compared with DMSO plus G20 (Fig. 1V and W).

Figure 1.

Glucagon and GLP-1 signaling regulate high-glucose–induced MT disassembly in β-cells. A: The experimental design. Isolated wild-type adult mouse islets (four mice) were allowed to recover overnight (O/N), then pretreated with compounds in media containing G2.8 for 2 h. Glucose was elevated to 20 mmol/L (G20) or remained at 2.8 mmol/L and incubated for two more hours before characterization. B–U: IF images of DeTyr-tubulin, E-cadherin, and insulin in mouse islet cells incubated with G2.8 or G20 and 0.05% DMSO control (B–E), 100 nmol/L liraglutide (Lir) (F–I), 1 μmol/L exendin-9 (Ex9) (J–M), 100 nmol/L glucagon (Gcg) (N–Q), or 20 μmol/L adomeglivant (Ado) (R–U). Blue, DAPI; magenta, insulin; cyan, E-cadherin; red, DeTyr-tubulin. V and W: Quantification of DeTyr-tubulin IF intensity in β-cells in arbitrary units (A. U.). Each dot represents the mean DeTyr-tubulin intensity in the cytoplasm of one β-cell (n = 48–116 in V; n = 45–115 in W). Three independent stainings were done (one or two animals each). P values are from two-way ANOVA with Tukey multiple comparison test. **P < 0.01, ***P < 0.001. Only P values below 0.05 are presented.

Consistent with the established roles of glucagon and GLP-1 in insulin secretion, both hormones enhanced GSIS, while their antagonists inhibited it (Supplementary Fig. 1). These findings suggest that GcgR and GLP-1R signaling are key regulators of glucose-induced MT destabilization in β-cells. Notably, the MT destabilizing effect of glucagon or liraglutide was not limited to prolonged (2 h) G2.8 conditioning, which induces higher levels of glucagon secretion than G5.5 (the approximate resting glucose level in mice): preincubating islets at G5.5 followed by acute treatment with glucagon or liraglutide (together with G5.5 or G20) destabilized MTs within 15 or 30 min after hormone addition, with the combined hormone and G20 treatment producing the most significant effects within 30 min (Supplementary Fig. 1). These studies establish glucagon and GLP-1 signaling as critical regulators of MT dynamics in β-cells.

Mouse Islets with Higher α-Cell–to–β-Cell Ratios Have Less Stable MT in Their β-Cells

We next tested whether α-cell abundance in islets affected β-cell MT stability, postulating that a higher α-cell–to–β-cell ratio would lead to greater glucagon and GLP-1 signaling in β-cells, thereby promoting MT destabilization and insulin secretion. We injected DT into adult Rosa^DTR;Gcg^iCre mice that expressed DT receptor (DTR) exclusively in proglucagon-expressing cells (23). This reduced α-cells by ∼70% in islets (Supplementary Fig. 2), accompanied by significantly higher levels of DeTyr-tubulin in β-cells compared with mock-injected controls (Fig. 2A–I). Within individual islets, α-cell–to–β-cell ratios inversely correlated with MT stability in both mock- and DT-injected mice (Fig. 2J and K). In contrast, MT stability in the primary cilia of β-cells remained unchanged, indicating that α-cell ablation specifically affected cytoplasmic MTs (Fig. 2L).

Figure 2.

DeTyr-tubulin density anticorrelates with the α-cell–to–β-cell ratio in mouse islets. Adult R26^DTR (Con) and R26^DTR;Gcg^iCre (DTRΔ) mice were injected with 20 ng DT. Their islets were isolated at least 2 months after DT injection to assess the long-term effects of α-cell ablation using MT studies. A–H: IF images showing the levels of DeTyr-tubulin (A and E) in β-cells without α-cell ablation (control R26^DTR mice) (A–D) or with α-cell ablation (R26^DTR;Gcg^iCre, noted as DTRΔ) (E–H). Insulin (B and F) and glucagon (C and G) staining were used to identify α- and β-cells, with merged images shown (D and H). DAPI was used to identify nuclei (blue). I: Quantification of DeTyr-tubulin intensity (arbitrary units [A. U.]) in islets of control and DTRΔ animals. Each dot represents one islet (six mice were used, with n = 45 islets examined). J and K: Scatter plots showing a negative correlation between the mean DeTyr-tubulin intensity (A. U.) in β-cells and the relative abundance of α-cells (represented by the ratio of α-cell–to–β-cell areas) in control (J, n = 40 islets) and DTRΔ conditions (K, n = 39–43 islets). L: The DeTyr-tubulin intensity (A. U.) in cilia (n = 210–275 cells). P values in J and K are from the Pearson correlation coefficient test, tailed with a 95% CI. P value in L is from a t test, with a two-tailed type 2 error. **P < 0.01, ***P < 0.001. Only P values below 0.05 are presented.

Note that we did not detect altered glucose tolerance in the DT-injected mice (Supplementary Fig. 2), likely because the reduced GSIS was insufficient to disrupt systemic glucose homeostasis under nonstressful conditions. Neither did we detect altered insulin levels in β-cells of wild-type or Rosa^DTR;Gcg^iCre mice with or without DT injection (Supplementary Fig. 2). We therefore focused on how α-cells modulate β-cell MT dynamics and GSIS in purified islets, a more sensitive assay of islet secretory function.

Islet α-Cells Preferentially Affect MT Stability in Neighboring β-Cells

We determined whether α-cells preferentially affect MT stability in neighboring β-cells. Islets were treated with G20 for 0, 15, or 30 min. MT stability was compared in β-cells that directly contacted α-cells (“C”) and in those one or two cells away (“C+1” and “C+2”) from the closest α-cells, identified by inspecting multiple adjacent optical sections (Fig. 3A–D). Under G2.8, MT stability was comparable across all three β-cell subclasses (Fig. 3A, B, and E). In contrast, MT stability in C β-cells was significantly lower than in C+1 and C+2 cells after high-glucose treatment (Fig. 3C–E). Consistently, MTs in C+1 and C+2 β-cells were significantly more resistant to Noc-induced depolymerization than those in C-type β-cells after glucose stimulation (Supplementary Fig. 3). In addition, HFD treatment, which can cause β-cell dysfunction (27), eliminated these differences (Supplementary Fig. 3), consistent with the increased MT stability in β-cells of diabetic mice (13). These data, together with the effects of GcgR and GLP-1R manipulation on MT stability, support a model that α-cells use paracrine-like signaling (e.g., via glucagon and/or GLP-1 from α-cells and their receptors in neighboring β-cells) to modulate β-cell MT dynamics.

Figure 3.

Mouse α-cells had bigger effects on their neighboring β-cells in MT disassembly. A–D: DeTyr-tubulin IF in mouse β-cells before or 30 min after high-glucose (G20) stimulation. Isolated whole wild-type islets were used. The few β-cells that directly contact (C), are one cell away from (+1), or are two cells away from (+2) an α-cell are circled in A and C. E: Quantification of MT stability in β-cells according to their relative distance from α-cells after 0, 15, or 30 min of treatment by G20. Each dot represents one cell (n = 44–52 for each condition). Two batches of islets from different mice (a total of six) were used. *P < 0.05, **P < 0.01, ****P < 0.0001, all from t test, with unpaired two-tailed type 2 errors. Only P values below 0.05 are presented.

Because cAMP has been shown to affect transmembrane ion transport (28), we compared Ca²⁺ influx in the α-contacting and noncontacting β-cells. There was no difference between these two β-cell subclasses in response to high glucose or KCl-induced depolarization (Supplementary Fig. 4), consistent with a recent study reporting how islet α-cells impact β-cell Ca²⁺ influx (29). Nor did we observe different Ca²⁺ response when comparing islets with high or low α-cell–to–β-cell ratios (Supplementary Fig. 4). These findings support a lack of Gcg/GLP-1 effect on Ca²⁺ influx in β-cells while leading us to explore how β-cells integrate α-cell signals and glucose metabolism to regulate insulin secretion. Note that, for unknown reasons, we did not observe the effect of α-cells on the timing of Ca²⁺ response in β-cells (Supplementary Fig. 4) reported in Tong et al. (29), which should be pursued in future studies.

MT Destabilization Mediates the GSIS-Promoting Activity of α-Cells

Several studies reported that mouse islet α-cells stimulated β-cell secretion (30–32), while others reported otherwise (33,34). To resolve this discrepancy, we compared insulin secretion between islets with different α-cell abundance in R26R^eYFP;Gcg^iCre;Ins1^Apple mice, which allowed identification of α-cells (eYFP⁺) and β-cells (Apple⁺) in live cells (Fig. 4A–C). Islets with higher α-cell–to–β-cell ratios exhibited greater insulin secretion under G2.8, G20, and G20K conditions (Fig. 4D). Additionally, we compared GSIS in PSIs reconstituted from purified β-cells and non–β-cells. PSIs with only β-cells had compromised GSIS compared with those containing all islet cell types, which was rescued by GLP-1R activation (Fig. 4E–I).

Figure 4.

α-Cells and MTs regulate an overlapping pool of ISGs. A–C: Preseparation of islets into α-cell proportion low (A) and α-cell proportion high (B) pools, with quantification shown in one mouse (C, n = 40–47). Here, islets from adult R26R^eYFP;Gcg^iCre;Ins1^Apple mice were manually separated into α-cell proportion high and low pools. D: Insulin (Ins) secretion under G2.8, G20, and G20K (green and black dots: α-cell proportion high or low islets, respectively). Each dot represented a secretion assay of one mouse (n = 4 or four mice used, with results averaged from two to four technical repeats each); Gcg, glucagon. E: FACS gating used for separation of β-cells and non–β-cells from Ins1^Apple islets. Here, adult islets from Ins1^Apple mice were dissociated and sorted for PSI prep. F: PSIs, made from β-only cells, β-cells + non-β-cells, and non–β-cells. G–I: Insulin secretion in response to different levels of glucose, with or without liraglutide. Each dot represents one GSIS assay, done with different PSIs (n = 8–20). For each GSIS assay, ∼10 PSIs were used. P values are from two-way ANOVA with Tukey multiple comparison test. Four mice were used for β-cell purification and PSI production. J: A scheme showing the experimental design to test the interaction of glucagon (Gcg) and MT destabilization. K–N: Insulin secretion that was induced by G2.8 alone (K), G20 alone (L), G20K (M), and total secretion (G2.8 + G20 + G20K) (N). Each dot represents data from one biological replicate (one or two wild-type mice to obtain enough islets for all assays, with a total of eight mice to produce the five data points at each condition: n = 5). O and P: Scheme and results of testing Gcg-induced secretion after degranulation with Noc and G20K. Each dot represents the results from one wild-type mouse (n = 4), with two or three independent secretion assays performed and averaged for each. Presented in K–N and P are (mean + SEM). P values are from t tests (paired with islets from each mouse), with two-tailed type 2 error. *P < 0.05, **P < 0.01, ***P < 0.001. Only P values below 0.05 are indicated. IEQ, islet equivalents.

We next investigated how MT interacts with α-cell–driven signals to regulate insulin secretion. Exogenous glucagon and Noc were used to activate GcgR and GLP-1R signaling and to induce MT destabilization, respectively (Fig. 4J) Neither treatment alone altered basal insulin secretion compared with DMSO-treated controls, whereas their combination significantly increased basal secretion (Fig. 4K). Under G20, either glucagon or Noc alone increased insulin secretion compared with the controls, and cotreatment further augmented secretion beyond Noc treatment but not beyond glucagon alone (Fig. 4L). Under G20 plus 30 mmol/L KCl (G20K), glucagon and Noc treatment did not further enhance secretion (Fig. 4M). When total insulin release was compared (including those induced by G2.8, G20, and G20K), we found that Noc, glucagon, and Noc plus glucagon each induced greater insulin secretion than the mock samples (Fig. 4N). However, the Noc-plus-glucagon cotreatment did not significantly elevate insulin secretion compared with either treatment alone (Fig. 4N). Furthermore, degranulation by G20K plus MT depolymerization eliminated the glucagon-enhanced insulin secretion (Fig. 4O and P). These findings suggest that glucagon signaling and MT destabilization act on the same pool of ISGs of different levels of Ca²⁺ sensitivity. Therefore, maximizing cytoplasmic Ca²⁺ influx and MT destabilization leads to the secretion of all releasable ISGs. We next tested whether this mechanism is conserved in human islets.

The α-Cell–to–β-Cell Ratio Positively Correlates With Insulin Secretion in Human Islets

We correlated insulin secretion with α-cell–to–β-cell ratios in individual islets. Optical sectioning of IF-stained individual islets using confocal microscopy (Supplementary Fig. 5) showed that the α-cell–to–β-cell ratio ranged from 0.00 to 0.84 (Fig. 5A–C), and the δ-cell–to–β-cell ratio ranged from 0.00 to 0.77 (Fig. 5D). There was a positive correlation between the ratios of α-cells to β-cells and of δ-cells to β-cells. The functional significance of this finding was not pursued.

Figure 5.

Human islet α-cell–to–β-cell ratio positively correlates with insulin secretion. A and B: IF images of two human islets with low (A) and high (B) α-cell–to–β-cell ratios. Four panels (single and merged channels) are shown for each islet section. Gcg, glucagon; Ins, insulin. C and D: Histograms of islet distribution with different α-cell–to–β-cell (C) or δ-cell–to–β-cell (D) ratios. Human primary islets were stained as whole mounts for IF and imaged with confocal microscopy at multiple focal planes. The IF⁺ areas for each hormone were assayed in multiple focal planes and used as surrogates for cell numbers. Islets from four donors were used (with a total of n = 291 islets examined). E–G: Correlation of insulin secretion with α-cell-to–β-cell ratio, induced by G2.8 (E), G20 (F), and G20K (G). In these results, hormone ratios (in pg) were used as surrogates for α-cell–to–β-cell ratios; n = 90 islets from three donors were used. H–J: Correlation of insulin secretion in individual human islets with glucagon-to-insulin ratios after Noc treatment at G2.8, G11, and G20K; n = 74 islets from three donors were used. K: The Noc-induced insulin secretion under different stimuli. The top plot shows islets with a glucagon-to-insulin (in pg) ratio below 0.10. The bottom plot shows those with a glucagon-to-insulin ratio above 0.10 (n = 30–48 for each condition). L and M: IF staining showing the cellular compositions of two human PSIs, with single and merged channels shown. N: α-cell–to–β-cell or δ-cell–to–β-cell ratios (ratios of hormone IF⁺ areas) in single human islets or PSIs (a total of n = 76 islets or PSIs were used, from three donors). O: Insulin secretion of single islets or PSIs (n = 73, from three donors). In N and O, each dot represents one islet or PSI. The P values are from the F-test (two-tailed type 2 errors), which measures the degree of data spread. *P < 0.05, **P < 0.01, ***P < 0.001. Only P values below 0.05 are presented.

We examined insulin secretion at G2.8, G11, and G20K in individual human islets. The α-cell–to–β-cell ratio positively correlated with insulin secretion under all conditions (Fig. 5E–G). Depolymerization of MTs by Noc did not eliminate this correlation at G2.8 or G11 (Fig. 5H and I) but eliminated the correlation at G20K (Fig. 5J), even though Noc enhanced insulin secretion under G20K as G11 (Fig. 5K). These findings support a model in which α-cells regulate the Ca²⁺ sensitivity of ISGs and MT stability. Noc, which only influences MT stability or G11 alone, did not completely mask α-cell–mediated enhancement of insulin secretion, because α-cell–secreted hormones increase ISG Ca²⁺ sensitivity. However, maximizing Ca²⁺ influx and MT destabilization with G20K masked the contribution of α-cell–secreted hormones because all releasable ISGs have been mobilized to the cell periphery under G20K.

We next tested whether altering the proportions of islet cell types was sufficient to modify insulin secretion. PSIs were made from dissociated human islet cells, thereby significantly reducing the variation in the proportions of α-, β-, and δ-cells within individual PSIs (Fig. 5L–N). Consequently, these PSIs exhibited reduced variation in insulin secretion at G2.8, G11, and G20K (Fig. 5O). These findings are consistent with the conclusion that differential α-cell–to–β-cell ratios within individual islets contribute to differential secretion.

Human Islet α-Cells Regulate MT Stability in Neighboring β-Cells in a Paracrine Manner

To test whether α-cells in human islets regulate MT dynamics in neighboring β-cells, we examined whether the rate of glucose-induced MT disassembly in β-cells correlates with their proximity to neighboring α-cells. Within 15 min of glucose stimulation, the level of DeTyr-tubulin in β-cells directly contacting α-cells is lowered more rapidly than in those not contacting α-cells (Fig. 6A–E). These results suggest that α-cells can regulate the MT dynamics of neighboring β-cells in a paracrine manner, a process conserved in both mouse and human cells.

Figure 6.

MT stability in human β-cells anticorrelates with their distance from α-cells. A–D: IF images of DeTyr-tubulin, glucagon (Gcg), and insulin (Ins) in human islet cells. Islets were conditioned in media with 5.5 mmol/L glucose overnight (A and B) and stimulated with 16.5 mmol/L glucose for 15 min (C and D). Inset in B is cropped from the image at the same x-y position but 2.8 microns deeper in the z-axis in the same magnification to show α-cell G2 that is underneath the surrounding β-cells and α-cell G1. Asterisks and “G” mark the position of α-cells (delineated by white dashed lines). “C” indicates β-cells (delineated by red dashed lines) in direct contact with α-cells; “+1” indicates β-cells (delineated by cyan dashed lines) that have one other β-cell between them and the closest α-cell; “+2” indicates β-cells (delineated by magenta dashed line) that have two other β-cells between them and the closest α-cell. E: Quantification of DeTyr-tubulin IF intensity in β-cells stimulated by 16.5 mmol/L glucose for 0, 15, 30, or 60 min. Each dot represents the value of one β-cell, from a total of three donors. Between 35 and 44 cells were counted in each condition. Heights of boxes represent means. Error bars represent SDs. P values comparing “C” and “+1” are shown in cyan; P values comparing “C” and “+2” are shown in magenta, calculated with t tests, with two-tailed type 2 errors. *P < 0.05, **P < 0.01, ***P < 0.001. Only P values below 0.05 are presented. F: A proposed model to explain how ISGs in β-cells respond to different stimuli. Under high glucose (e.g., G20), medium levels of Ca²⁺ influx and partial MT destabilization were induced, allowing the release of the portion of the ISGs with high sensitivity to Ca²⁺ (the type of ISG marked with “1”). A complete MT depolymerization with Noc increased the pool of releasable ISGs to allow more secretion, including ISG type “2”. Gcg induced stronger MT destabilization (but not complete depolymerization) and also increased the sensitivity of ISGs to Ca²⁺. These two effects could enhance secretion (including ISG types “3” and “4” if Noc is used). Under G20K, the high levels of Ca²⁺ influx allowed the secretion of ISGs with both low and high Ca²⁺ sensitivity (ISG types “1” and “2”). Similarly, Noc and Gcg would boost the secretion, releasing those freed by MT destabilization (ISG types “3” and “4”) or those made releasable by increasing Ca²⁺ sensitivity (ISG type “5”). Yet cotreatment of Noc and Gcg under KCl could not further boost secretion over Gcg treatment alone, because Gcg alone could induce the secretion of the majority of the releasable ISGs.

Discussion

Balanced glucagon and insulin secretion from endocrine islet α- or β-cells, respectively, regulate blood glucose homeostasis (35,36). In this study, we show that α-cells regulate insulin secretion by modulating MT dynamics in β-cells, likely via paracrine signals, a mechanism attenuated in mouse islets under metabolic stress that can lead to β-cell failure (27). These findings underscore the importance of α-cell–to–β-cell cross talk via an MT destabilization-based mechanism in β-cell function and diabetes.

Paracrine signaling between islet cells coordinates hormone secretion. To this end, Ucn3 and insulin secreted from β-cells can induce SST release from δ-cells, which represses β- or α-cell secretion (37,38). However, whether α-cells regulate insulin secretion remains controversial. Specifically, exogenous glucagon could stimulate insulin secretion (39). Several in vitro studies suggest that α-cells enhance insulin secretion by elevating cAMP production in β-cells (3–5,30–32,40). However, others reported that ablating α-cells from islets did not affect insulin secretion or glucose homeostasis in mice (33,34). We examined insulin secretion from mouse and human islets with different α-cell–to–β-cell ratios. We showed that higher α-cell proportions in islets facilitate greater insulin secretion. These results underscore the importance of islet α-cells for robust GSIS. They also suggest that increasing the proportions of α-cells in islets will benefit GSIS, a possibility that could be explored for producing human embryonic stem cell–derived islet-like organoids for transplantation therapy of diabetes (41). Note that this latter finding differs slightly from that reported in Fouque et al. (40), which showed that 1% of α-cells in PSIs could restore GSIS to the levels found in PSIs with 15% α-cells. This discrepancy may be related to the presence of other islet cell types (e.g., δ-cells) in islets, which could be explored in future studies.

Most agree that α-cell–secreted glucagon and/or GLP-1 facilitate cAMP production in β-cells, which promotes insulin secretion by modulating Ca²⁺ influx (8), Ca²⁺ sensitivity of ISGs (42), and ISG biosynthesis (10). Surprisingly, we found that α-cells had little effect on glucose- or depolarization-induced Ca²⁺ influx, similar to a recent report (29). Instead, α-cells significantly destabilize β-cell MTs at single-cell or islet levels, requiring GcgR and GLP-1R activation, with the β-cells that contact α-cells being affected the most. These results, together with our findings that GLP-1 can replace α-cells to enhance GSIS in mouse PSIs, suggest that the α-cell–to–β-cell cross talk is paracrine, as α-cell–β-cell contact can be replaced by chemical activation of GLP-1R/GcgR. This conclusion is particularly important for understanding human islet function, because human islets usually have a higher proportion of α-cells than mouse islets, and α-cell–β-cell contacts occur throughout human islets (43) rather than mainly near the islet periphery, where most α-cells were found in mouse islets (44). In addition, these results provide a mechanism for how high glucose induces MT destabilization in β-cells. Specifically, we have shown that MT destabilization is required for GSIS, but we do not know the mechanism of this destabilization (13,18,19). Now we propose that GcgR and GLP-1R signaling is a key trigger of this process, and that β-cell MTs constitute a hub that integrates signals from glucose, islet cell composition, and incretin hormones to fine-tune GSIS for homeostasis. The fact that HFD challenge deactivates this α-cell–induced MT destabilization (results herein) and diabetes enhances MT stability (13) underscores the physiological significance of this mechanism. Therefore, ways to sustain this α-cell–β-cell cross talk could be explored in the future to delay and/or prevent β-cell failure and diabetes.

β-Cell functions are heterogeneous, a property that is crucial for regulated secretion in response to different types and levels of stimuli (45). Whether islet secretion is heterogeneous is unknown, although its structural heterogeneity has been documented (46). Here, we identify apparent differences in secretory activity among individual islets. We further showed that the varied α-cell–to–β-cell ratio contributes to this secretory heterogeneity, likely by modulating the MT network of β-cells. Thus, testing the fitness of islets (secretory function, proliferation, and viability) with different α-cell–to–β-cell ratios under different metabolic stress (e.g., conditions with high glucose, free fatty acids, or both, which require different stress responses [47]) will be an interesting future study.

Several questions remain unanswered. First, low glucose did not cause detectable MT destabilization despite induction of glucagon secretion. How high glucose interacts with physiological glucagon and GLP-1 levels to regulate MT is unclear, and how quickly the glucagon/GLP-1 effect on MT can be reversed is also unknown. Second, the isolated islets have altered architecture (cell type rearrangement) and lack proper vascularization/directional blood flow. In vitro islet studies also excluded the effects of gut-secreted GLP-1 and glucose-dependent insulinotropic polypeptide during feeding. Thus, it remains unclear how β-cells respond to α-cell signaling in vivo. This question is particularly intriguing under HFD conditions, where inflammatory factors, in combination with glucose-dependent insulinotropic polypeptide, regulate GLP-1 production (48). Third, it is unclear whether glucagon or GLP-1 plays the predominant role in this paracrine signaling, and we did not compare the cAMP levels in β-cells with or without contact to α-cells. Addressing these questions will be critical for understanding β-cell dysfunction in diabetes, a condition characterized by changes in glucagon and GLP-1 production (49,50). Fourth, it is unclear why α-cell–regulated MT destabilization is attenuated under HFD challenge. In this regard, a recent report showed that GLP-1–mediated signaling was attenuated by metabolic stress via GLP-1R mislocalization in β-cells (29), providing a possible answer. Alternatively, metabolic stress could induce inflammatory responses in mice, which can deregulate GLP-1 expression via interleukin 6 (IL-6) and IL-1β (48).

Despite these questions, others’ reports and our new findings suggest that α-cell signaling can 1) increase the Ca²⁺sensitivity of ISGs (42), 2) induce new MT polymerization for ISG biosynthesis (10), and 3) destabilize existing MTs for ISG secretion. These activities can explain all findings by far. Our model is that β-cells contain a defined pool of releasable ISGs with variable Ca²⁺ sensitivity (Fig. 6F). MTs inhibit their secretion via physical association. G20 induces a moderate Ca²⁺influx and partial MT depolymerization in the cell periphery to release ISGs with high Ca²⁺ sensitivity. Noc treatment increases secretion by eliminating the MT inhibition. Glucagon/GLP-1 from α-cells, via GcgR/GLP-1R signaling, induces moderate MT destabilization and increases the Ca²⁺ sensitivity of ISGs, thereby promoting insulin secretion. Noc and glucagon together produce additive effects due to increased ISG Ca²⁺ sensitivity and MT depolymerization. G20K induces maximal Ca²⁺ influx, masking the glucagon-mediated ISG sensitization. Thus, the combination of Noc, glucagon, and G20K results in near-complete secretion of the releasable ISG pool.

This article contains supplementary material online at https://doi.org/10.2337/figshare.30861590.

Funding Statement

This study is supported by National Institute of Diabetes and Digestive and Kidney Diseases (DK106228 for G.G. and I.K.; DK125696 and DK128710 for G.G.). Imaging was performed with Vanderbilt University Medical Center Cell Imaging Shared Resource (CA68485, DK20593, DK58404, DK59637, and EY08126).