Glucose-Dependent Regulation of γ-Aminobutyric Acid (GABA_A) Receptor Expression in Mouse Pancreatic Islet α-Cells

Collagenase digestion was used to isolate intact islets from the pancreas of adult female CD1 mice (12–16 weeks) before culture in RPMI 1640 medium (10 mmol/l glucose) overnight before experiments (5). αTC1-9 cells (passage number 40–43; American Type Culture Collection) were grown in Dulbecco’s modified Eagle’s medium containing 16 mmol/l glucose (5). For measurement of GABA_AR expression, cells were cultured in 16 mmol/l glucose to 60% confluence and switched to the test condition for 24 h.

To identify living islet α-cells, using an analogous approach to that previously used to monitor Ca²⁺ in these cells (5), an adenovirus encoding monomeric red fluorescent protein (mRFP) under the control of the preproglucagon (PPG) (1.6 Kb) promoter was generated. Luciferase cDNA was removed from pShuttle-PPG-Luc (5) by digestion with HindIII and XbaI. cDNA encoding mRFP was then ligated into the same HindIII XbaI site of pShuttle-PPG-Luc to generate pPPG-mRFP. Adenoviruses were prepared as shown by Ainscow and Rutter (36).

Islets (second day of culture) and αTC1-9 cells were preincubated for 1 h at 37°C in Krebs buffer containing (in mmol/l) 120 NaCl, 4.8 KCl, 1.5 CaCl₂, 1.2 MgCl₂, 24 NaHCO₃, 10 glucose, and 1 mg/ml BSA at pH 7.4. αTC1-9 cells in 12-well plates were incubated for 40 min in 700 μl Krebs buffer adjusted for each test condition. Islets were handpicked in groups of seven to eight and incubated for 1 h in 400 μl Krebs buffer. All medium was collected after the incubation, and the cells were lysed in 500 μl acidified ethanol/0.1% triton. Glucagon was assayed in medium and cell contents by radioimmunoassay (Linco Research, St. Charles, MO).

Total RNA was isolated from αTC1-9 cells and islets using TRIzol reagent (Invitrogen Technologies). Total RNA from mouse brain was used as a positive control for GABA_AR expression. One-step reverse transcription PCR (RT-PCR) was performed using Superscript One-Step RTPCR system with Platinum TaqDNA Polymerase (Invitrogen). Published gene-specific primer sequences for GABA_AR α4 (37); α1, 2, 3, 5, 6; β1–3; γ1–3; and δ (38) subunits were used. Additional primers were used for α1 (5′-TTTCGGACCAGTTTCAGACC-3′; 5′-AACGTGACCCATCTTCTGCT-3′; 396 bp) and ε (5′-CGGTTTGGCTTCATTGTCTT-3′; 5′-AGAAGTCCAAAGCCGTGAGA-3′; 202 bp). PCR amplifications were performed under standard conditions, and the identity of all amplicons was expressed in αTC and islet cells confirmed by DNA sequencing. PCR products were electrophoresed on 1% agarose gels with GeneRuler 100 bp DNA Ladder Plus (MBI Fermentas).

Quantitative analysis of the abundance of GABA_AR α4, β3, and γ2 subunit mRNAs was performed on a DNA Engine Opticon System (MJ Research). Reverse transcription was performed using random hexamers with Omniscript reverse transcriptase (Qiagen), and real-time PCR detection used Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen). The reference was18S rRNA (39), and gene-specific primers for GABA_ARs were designed using GenBank sequences (Gabra4 [200 bp]: 5′-TGTCACCAGCTTTGGGCCCGTT-3′; 5′-GGAGCTGTCATGTTATGTGAGAC-3′; Gabrb3 (201 bp): 5′-ATTACCACCGTGCTCACCAT-3′; 5′-TGTCTTCTCCGCAAGCTTCT-3′; Gabrg2 (91 bp): 5′-GGAGCCTGGAGACATGGGA-3′; 5′-TGAACAAGCAAAAGGCGGTA-3′). The identity of each amplicon and specificity of amplification was confirmed by gel electrophoresis and melting curve analysis. All primer combinations were optimized and validated for relative quantification of gene expression using the comparative threshold cycle method (39) described in the PE Applied Biosystems sequence detection system user bulletin #2 (http://www.appliedbiosystems.com). Briefly, data for each target gene were normalized to the endogenous control gene (rRNA), and the fold change in target gene mRNA abundance was determined using the 2^−ΔΔCt method such that levels in treated condition are expressed relative to control (=1) (40).

Freshly isolated islets were dispersed to single cells by mechanical dissociation in calcium-free medium. Single cells were cultured overnight and infected with the pPPG-mRFP adenovirus (6 h) and cultured for a further 36 h before labeling. Cells were fixed in 4% paraformaldehyde, permeabilized in 0.1% Triton X-100, incubated with primary antibodies to GABA_AR α4 (1:50; Santa Cruz Biotechnology) or γ2 (1:50; Calbiochem) subunits overnight at 4°C, and detected with fluorescein isothiocyanate–anti-rabbit IgG (1:500; Jackson ImmunoResearch Laboratories). Images were captured on a Leica TCS NT laser scanning confocal microscope using a 63× oil immersion objective.

αTC1-9 cells were homogenized in buffer containing 10 mmol/l Tris, pH 7.4, 0.32 mol/l sucrose, 5 mmol/l EDTA, and protease inhibitor cocktail (Complete Mini; Roche). Aliquots of crude membrane samples were assayed for protein content using BCA protein assay (Pierce), and increasing amounts of protein (10–40 μg) were separated by SDS-PAGE on 10% (wt/vol) polyacrylamide gels. Proteins were transferred onto polyvinylidene difluoride membranes (Millipore). The blots were blocked for 2 h in Tris-buffered saline/5% nonfat dry milk before incubation with the GABA_AR α4 subunit antibody (1:100; Santa Cruz Biotechnologies) for 1 h at room temperature. Horseradish peroxidase–conjugated donkey anti-rabbit (1:7,500; Amersham Biosciences) and Supersignal West Pico Chemiluminescent Substrate (Pierce) were used for chemiluminescent detection. To ensure an analysis in the linear range, films were exposed for various times, and the quantification of immunoreactive bands was performed using the National Institutes of Health ImageJ program (http://rsb.info.nih.gov/ij).

Data are expressed as the mean ± SE for the number of experiments indicated. Statistical significance and differences between means were evaluated using the Student’s t test.

The effects of GABA on glucagon secretion from mouse islets or αTC1-9 cells are shown in Fig. 1. As expected (1), glucose dose-dependently inhibited glucagon secretion in both cases, with near-maximal inhibition of glucagon secretion in islets achieved at ∼5 mmol/l glucose (46.3 ± 7.8% of 0.5 mmol/l glucose value, n = 5, P < 0.05) and in αTC1-9 cells at ∼1 mmol/l glucose (40.5 ± 5.1% of 0 mmol/l glucose value, n = 4, P < 0.05) (5). At low glucose concentrations, at which glucagon secretion is stimulated, GABA (10 or 100 μmol/l) significantly reduced glucagon secretion (Fig. 1A and B) from both islets (62.1 ± 11.4 or 54.1 ± 5.4% of control, n = 4, P < 0.05) or αTC1-9 cells (60.8 ± 11.6 or 42.7 ± 11.0% of control, n = 4, P < 0.05), while exerting no effect at high glucose concentrations.

To investigate the role of GABA_ARs, we next examined the effects of the selective GABA_AR antagonist, SR95531 (22), on the ability of GABA to suppress glucagon secretion at low glucose concentrations (Fig. 1C and D). In both islets and αTC1-9 cells, GABA (10 μmol/l) significantly suppressed glucagon secretion to ∼50% of control levels, whereas in the presence of SR95531 (10 μmol/l), exogenous GABA exerted no significant effect on glucagon secretion.

To quantify the likely contribution of endogenous GABA release to the control of glucagon secretion by glucose from isolated intact islets, we next examined the effects of the antagonist alone (Fig. 1E). At 0.5 mmol/l glucose, SR95531 had no effect on glucagon secretion, suggesting that endogenous GABA levels were low under these conditions. However, SR95531 significantly increased glucagon secretion to 125.8 ± 8.0 and 195.7 ± 25.2% of control levels at 5 and 10 mmol/l glucose, respectively (n = 7, P < 0.01), thus blunting the inhibitory effect of glucose at either concentration (Fig. 1E) but demonstrating a qualitatively more important role for GABA at 10 mmol/l glucose. Consistent with this selective GABA_AR antagonist having no intrinsic or confounding nonspecific activity, incubation of clonal αTC1-9 cells with SR95531 in the absence of GABA exerted no effect on glucagon secretion stimulated by low glucose or at elevated glucose concentrations (Fig. 1F).

The above findings demonstrated the functional importance of GABA_ARs in suppressing glucagon secretion from mouse islets. To identify the expression pattern of GABA_ARs in mouse islets and αTC1-9 cells, we next used RT-PCR with subunit-specific primers (Fig. 2A). The detection of β-actin was used as a positive control for all samples, whereas PCR was performed in the absence of the reverse transcription step as a negative control. The α4, β3, and γ2 and ε subunits were expressed in both αTC1-9 cells and in islets, whereas mRNAs encoding the α3, β2, and δ subunits were detected only in αTC1-9 cells. α5, α6, β1, β2, γ1, and γ3 subunits were absent from both. α1 mRNA was detected in mouse islets and αTC1-9 cells when RT-PCR was performed with previously published primer sequences (35,38) (data not shown), but sequencing revealed that this product was not related to Gabra1. Use of a novel primer set, specific for mouse Gabra1, produced a product of the expected size in brain, but not in islet cells (Fig. 2A), consistent with the absence of this subunit from the mouse endocrine pancreas.

We confirmed the presence of GABA_AR subunit proteins in mouse pancreatic α-cells using confocal immunocytochemistry. Clear cell surface expression of both α4 and γ2 subunits was apparent in αTC1-9 cells (Fig. 2B). Individual α-cells were identified in dispersed mouse islets by infection with an adenovirus expressing mRFP under the control of the PPG promoter (Fig. 2C). This strategy is analogous to that recently used to record Ca²⁺ changes selectively in α-cells (5), where 92% (54/59 cells) of mRFP-labeled cells were also glucagon positive, while insulin-positive cells were uninfected (Fig. 2C, lower panels). Immunoreactivity against both α4 and γ2 subunits was clearly evident at the cell surface of the mRFP-labeled primary α-cells (Fig. 2C). Thus, αTC1-9 cells and isolated mouse islets express a similar discrete subset of GABA_AR subunits.

We next determined whether changes in glucose concentration might affect the expression of GABA_AR subunits in intact isolated mouse islets. Culture for 24 h at increasing glucose concentrations (2.5, 5.5, and 16 mmol/l) strongly augmented the expression of GABA_AR α4, β3, and γ2 mRNAs (Fig. 3A). Thus, after culture at 2.5 mmol/l glucose, expression of α4 subunit mRNA was 33.9 ± 13.3% of the level at 16 mmol/l (n = 3, P < 0.05), increasing to 73.4 ± 9.6% at 5.5 mmol/l glucose (n = 3, P < 0.05). Similar but less marked effects of glucose on β3 and γ2 mRNA expression were also observed (Fig. 3A).

We next sought to determine the extent to which glucose affected GABA_AR gene expression in α-cells, as opposed to other islet cell types. Because mouse α-cells cannot readily be purified (e.g., by fluorescence-activated cell sorting) (20), GABA_AR expression was measured in clonal αTC1-9 cells. A similar dose-dependent increase in the expression of α4, β3, and γ2 mRNAs was evident in αTC1-9 cells cultured for 24 h at increasing glucose concentrations (1, 3, 10, and 16 mmol/l; Fig. 3B). Thus, after culture at 1 mmol/l glucose, expression of α4 subunit mRNA was 27.9 ± 2.4% of the level at 16 mmol/l (n = 3, P < 0.05), increasing to 52.1 ± 3.9% at 3 mmol/l glucose (n = 3, P < 0.05). As in the islets, the effects of glucose on β3 and γ2 mRNA expression were less marked, but each gene displayed significantly higher expression levels after culture at 16 mmol/l glucose compared with 1.0 mmol/l glucose. Correspondingly, Western (immuno)blot analysis revealed that GABA_AR α4 subunit protein was significantly (∼2.0-fold) higher after culture at 16 mmol/l compared with 1 mmol/l glucose (Fig. 3C).

The above data indicated that glucose is an important regulator of GABA_AR expression in pancreatic α-cells. Because GABA_AR expression has been shown to be modulated by depolarizing concentrations of K⁺ ions in neurons (41), we sought next to determine whether the effects of elevated glucose concentrations on GABA_AR gene expression may be due, at least in part, to membrane hyperpolarization (42) or depolarization (3) and decreases in intracellular free [Ca²⁺]. Culture of αTC1-9 cells for 24 h at elevated (25 mmol/l) versus normal (5 mmol/l) K⁺ led to an inhibition of the expression of each isoform at 16 but not 1 mmol/l glucose, by 33.3 ± 3.7, 38.9 ± 5.9, and 36.8 ± 0.2% of control levels for α4, β3, and γ2 mRNA, respectively (Fig. 4A, n = 3, P < 0.05). By contrast, insulin (20 nmol/l) was without effect on the expression of any of the studied GABA_AR isoforms in αTC1-9 cells (Fig. 4C), consistent with previous observations that insulin has no effect on GABA_A receptor gene expression in neurons (43), while increasing the trafficking of the receptors to the cell surface both in these (43) and α-cells (23). In these experiments, we confirmed that the expression of each GABA_AR mRNA was reduced in 1 mmol/l glucose, as in earlier experiments (Fig. 4B and D vs. Fig. 3).

We next examined whether the inhibitory effects of K⁺ depolarization or low glucose on GABA_AR gene expression might be mediated by enhanced Ca²⁺ entry (Fig. 5), given the presence in α-cells of voltage-gated calcium channels (L-type and T-type) (3). Nimodipine (25 μmol/l), an L-type calcium-channel blocker that inhibits Ca²⁺ oscillations in αTC1-9 cells at low glucose concentrations (M.A.R., G.A.R., unpublished data), was without effect on GABA_AR expression at 16 mmol/l glucose (Fig. 5) but abolished the inhibitory effects of low (1 mmol/l) glucose or high [K⁺]. Thus, diminished Ca²⁺ influx was necessary and sufficient for the enhancement of GABA_AR α4, β3, and γ2 subunit expression at elevated glucose concentrations.

To further investigate the possible mechanisms underlying the glucose-induced regulation of GABA_AR expression, we examined the effects of the calcineurin (Ca²⁺/calmodulin-dependent protein phosphatase 2B) inhibitor cyclosporin A and the adenylate cyclase activator forskolin. In αTC1-9 cells, cyclosporin A (1 μmol/l) significantly increased the expression of β3, but not α4, mRNA at low glucose concentrations (Fig. 6). Elevating cAMP levels by the addition of forskolin (10 μmol/l) significantly increased the expression of β3 mRNA at both glucose concentrations tested, whereas α4 mRNA was only increased at 1 mmol/l glucose (Fig. 6).

The first aim of this study was to reexamine the potential contribution of GABA release to the acute regulation of glucagon secretion from mouse islets. We have recently demonstrated that glucose can act directly on the mouse α-cell to inhibit Ca²⁺ oscillations (5), an effect that may be prompted by increased intracellular free ATP concentrations and closure of ATP-sensitive K⁺ channels—the latter causing cell depolarization and the inactivation of Na⁺-dependent action potentials (3). Although this study suggested that the release of insulin or Zn²⁺ was unlikely to be involved, a potential role for other paracrine mechanisms was not excluded.

Here, we show that release of endogenous GABA from neighboring β-cells is likely to play a role in regulating glucagon secretion from mouse islets. Thus, exogenous GABA inhibited glucagon secretion at low glucose concentrations by 50–60% in both pancreatic mouse islets and murine αTC1-9 cells (Fig. 1A and B). These effects of GABA are thus substantially more than the 18% inhibition of secretion observed at 0 mmol/l glucose by Gilon et al. (20), a difference that may reflect either the use of alternative mouse strains (NMRI mouse in the earlier study) as well as differences in preculture (1 day in the present study) and incubation conditions (0 mmol/l glucose [20] vs. 0.5 mmol/l [present study]). From inspection of Fig. 1E, it is apparent that GABA release may contribute ∼15 or ∼40% toward the overall inhibitory effect of 5 or 10 mmol/l glucose on glucagon release, respectively. The greater apparent contribution of GABA release at higher glucose concentrations may reflect both 1) more substantial release of this neurotransmitter because β-cells are progressively depolarized and 2) the “unmasking,” in the absence of GABA_AR function, of a stimulatory effect of high glucose concentration on glucagon secretion (7,44). Nonetheless, these contributions appear to be less than those estimated using the same approach in rat islets (22). Other observations point to differences in the relative importance of GABAergic regulation of glucagon release in mouse and rat. First, in the rat, blockade of GABA_ARs with SR95531 stimulated glucagon release both at low (1 mmol/l) and elevated (20 mmol/l) glucose concentrations, implying a tonic restraint by endogenous intra-islet GABA even under conditions where release of the neurotransmitter from β-cells is unlikely to be stimulated (15). Second, SR95531 completely eliminated the inhibitory effect of elevated (20 mmol/l) glucose concentrations on glucagon secretion from rat islets (22), implying that enhanced GABA release might be the sole mechanism through which glucose acts to regulate glucagon secretion in this species. In contrast, the present findings in mouse are consistent with the view that the previously described metabolic effects of glucose on isolated mouse α-cells (5) are likely to play a significant role in the control of glucagon secretion in this species.

Our results suggest that, in mouse pancreatic α-cells, the most abundant GABA_AR complexes likely comprise 2α4, 2β3, and γ2/ε subunits. Importantly, α4-containing GABA_ARs display a distinct pharmacology compared with complexes with other α-subunits, with high affinity for GABA (45), slow desensitization kinetics (46), and insensitivity to benzodiazepines (47). The putative paracrine regulator of glucagon secretion, Zn²⁺, co-released from β-cells with insulin, might thus act at α-cell GABA_ARs. Zn²⁺ ions inhibit GABA_AR function by an allosteric mechanism that depends on the subunit composition: α-β subunit combinations show the highest sensitivity, although the incorporation of a γ2 subunit reduces the inhibitory potency (48), as demonstrated in cell lines expressing human α4, β3, and γ2 GABA_ARs (45). In mouse islets, Zn²⁺ ions have no effect on glucagon secretion (5), but Zn²⁺ is reported to inhibit glucagon release from rat islets (9), possibly by opening α-cell K_ATP channels (3,10).

We report here for the first time that glucose regulates GABA_AR gene expression in islets and implicates decreases in intracellular free [Ca²⁺] as a likely mediator of the effects of the sugar. We propose that by inhibiting electrical activity and hence Ca²⁺ oscillations, glucose removes an inhibitory regulator (which may be Ca²⁺ itself) for GABA_AR gene expression. High K⁺ thus mimics the effect of glucose lowering (Fig. 4A), whereas nimodipine reverses this effect (Fig. 5). Calcium influx through N-type Ca²⁺ channels has been shown to be important in controlling glucagon secretion in mouse islets, but the role of the L-type Ca²⁺ channel is less clear (3). However, we have found that, in αTC1-9 cells, nimodipine completely blocked the effects of low glucose on glucagon secretion, whereas the N-type Ca²⁺ channel inhibitor w-conotoxin was without effect (M.A.R., G.A.R., unpublished data), suggesting differing roles for these channels in clonal versus primary α-cells. Nonetheless, our observations leave open the possibility that, in islets, α-cell L- and N-type channels may be somewhat selectively coupled to transcriptional or secretory events, respectively.

The mechanisms regulating GABA_AR gene expression have so far largely been confined to the study of cerebellar granule cell neurons where depolarizing concentrations of K⁺ (41), cAMP (49), and mitogen-activated protein kinase regulation (50) affect GABA_AR abundance. For example, culturing rat cerebellar granule neurons at 25 mmol/l K⁺ reduced β3, as well as α1, α6, and β2 subunit mRNA levels (41). The importance of Ca²⁺ entry-dependent pathways in the control of gene expression is well documented. One such mechanism is the regulation of cAMP response element binding protein (CREB) dephosphorylation by the Ca²⁺/calmodulin-dependent protein phosphatase 2B/calcineurin (51). Consistent with a possible role for calcineurin activation in the effects of low glucose, the calcineurin inhibitor cyclosporine A partially reversed the inhibitory effects of 1 mmol/l glucose on the expression of the β3 receptor subunit (Fig. 6). Likewise, elevation of cAMP levels with forskolin increased the expression of both the α4 and β3 subunits, albeit at both low and high glucose concentrations (see below). Inhibition of calcineurin can prolong CREB phosphorylation and stimulate CRE-mediated gene expression (52), and downregulation of GABA_AR function by a calcineurin-dependent dephosphorylation of subunits has been shown earlier (53). Because CREB binding has recently been demonstrated at the promoters of the GABA_AR β3 and γ2 subunit genes (54), we speculate that calcineurin-mediated dephosphorylation of CREB may suppress GABA_AR gene expression in α-cells. However, the upregulation of a transcriptional repressor such as lost on transformation 1 (Lot1)/zinc finger regulator of apoptosis and cell cycle arrest (ZAC) when calcineurin is activated by low glucose (55) represents a further potential mechanism.

We have considered the possibility that an autocrine effect of glucagon release on GABA_AR may be involved in the observed responses to glucose. Because, in neurons, elevated cAMP increased α1 and β2, but decreased β3 expression (49), loss of glucagon-stimulated increases in cAMP may potentially contribute to the increased expression of GABA_AR β3 and possibly α4 and β3 subunits seen in α-cells at high glucose concentrations. Importantly, however, elevation of cAMP by forskolin did not overcome the decrease of GABA_AR expression seen at low glucose concentrations (Fig. 6), indicating that other cAMP-independent mechanisms must also be involved in the effects of glucose.

This work was supported by grants to G.A.R. from the Wellcome Trust (Programme Grant 067081/Z/02/Z) and the Juvenile Diabetes Research Foundation (#1-2003-235). G.A.R. is a Wellcome Trust Research Leave Fellow.