Content uploaded by Matthias Braun
Author content
All content in this area was uploaded by Matthias Braun
Content may be subject to copyright.
Glucose Inhibition of Glucagon Secretion From Rat
␣-Cells Is Mediated by GABA Released From
Neighboring -Cells
Anna Wendt,
1
Bryndis Birnir,
1
Karsten Buschard,
2
Jesper Gromada,
3
Albert Salehi,
1
Sabine Sewing,
3
Patrik Rorsman,
1
and Matthias Braun
1
␥-Aminobutyric acid (GABA) has been proposed to
function as a paracrine signaling molecule in islets of
Langerhans. We have shown that rat -cells release
GABA by Ca
2ⴙ
-dependent exocytosis of synaptic-like
microvesicles. Here we demonstrate that GABA thus
released can diffuse over sufficient distances within the
islet interstitium to activate GABA
A
receptors in neigh-
boring cells. Confocal immunocytochemistry revealed
the presence of GABA
A
receptors in glucagon-secreting
␣-cells but not in - and ␦-cells. RT-PCR analysis de-
tected transcripts of ␣
1
and ␣
4
as well as 
1–3
GABA
A
receptor subunits in purified ␣-cells but not in -cells.
In whole-cell voltage-clamp recordings, exogenous ap-
plication of GABA activated Cl
ⴚ
currents in ␣-cells. The
GABA
A
receptor antagonist SR95531 was used to inves-
tigate the effects of endogenous GABA (released from
-cells) on pancreatic islet hormone secretion. The
antagonist increased glucagon secretion at 1 mmol/l
glucose twofold and completely abolished the inhibitory
action of 20 mmol/l glucose on glucagon release. Basal
and glucose-stimulated secretion of insulin and soma-
tostatin were unaffected by SR95531. The L-type Ca
2ⴙ
channel blocker isradipine evoked a paradoxical stimu-
lation of glucagon secretion. This effect was not ob-
served in the presence of SR95531, and we therefore
conclude that isradipine stimulates glucagon secretion
by inhibition of GABA release. Diabetes 53:1038 –1045,
2004
I
n the mammalian central nervous system (CNS),
␥-aminobutyric acid (GABA) is the most important
inhibitory neurotransmitter. Three types of GABA
receptors have been identified: GABA
A
and GABA
C
receptors are ligand-gated Cl
⫺
channels, while GABA
B
receptors are G protein coupled (1). In the CNS, applica-
tion of GABA reduces excitability by a combination of
GABA
A
and GABA
B
receptor activation, leading to mem-
brane repolarization, reduced Ca
2⫹
influx, and suppres-
sion of neurotransmitter release.
Outside the CNS, GABA and GAD, the enzyme cata-
lyzing the formation of the neurotransmitter, are present
at high levels in the pancreatic -cells (2,3). GAD65 is the
most abundant, if not the only, isoform of GAD in human
islets (4) and has been implicated in the etiology of
autoimmune type 1 diabetes (5). At the time of diagnosis,
⬃80% of type 1 diabetic patients have autoantibodies
against GAD65 (6). The physiological roles of GABA and
GAD65 in pancreatic islets are unknown. GABA can, via
the formation of succinic semialdehyde and succinic acid,
be introduced into the tricarboxylic acid cycle and has
therefore been suggested to serve as an energy source
within the -cell (7). Others have hypothesized that GABA,
by analogy to its role in the CNS, functions as a paracrine
signaling molecule that conveys messages from the -cells
to surrounding cells in the islet (7–9).
In the -cell, GAD65 and GABA are associated with
synaptic-like microvesicles (SLMVs), which are distinct
from the insulin-containing large dense-core vesicles
(LDCVs) (10,11). SLMVs are smaller in diameter (⬃90 nm)
than LDCVs (⬃300 nm) (12) and accumulate GABA by
active transport (11). We have developed a patch clamp–
based technique to investigate GABA secretion from single
-cells with high temporal resolution. This technique in-
volves the expression of GABA
A
receptors at a high
density in the -cell using adenoviral vectors. We have
thus been able to demonstrate that GABA is released by
voltage- and Ca
2⫹
-dependent exocytosis of SLMVs from
-cells (12). In this study we have focused on the influence
of endogenously released GABA on islet hormone secre-
tion. We demonstrate that GABA, following its exocytotic
release, travels over sufficient distances to activate GABA
receptors in neighboring cells, that functional GABA
A
receptors are present in ␣-cells, and that GABA receptor
antagonism selectively affects glucagon release.
RESEARCH DESIGN AND METHODS
Islet preparation. Pancreatic islets and single islet cells were prepared from
Sprague Dawley or Wistar rats as described previously (13). In experiments
involving detection of vesicular GABA release (Fig. 1B–D), cells were coin-
fected with adenoviruses expressing the ␣
1
and 
1
subunits of the GABA
A
receptor (12) and cultured for ⬃24 h. All experimental procedures involving
animals were approved by the ethical committees in Lund, the city of
Hamburg, and the University of Copenhagen.
Electrophysiology. The electrophysiological measurements were conducted
in the standard whole-cell configuration. ␣-Cells were identified on the basis
of their small size (⬍4 pF as compared with ⬎4pFfor-cells) and the
From the
1
Department of Physiological Sciences, Lund University, Lund,
Sweden;
2
Bartholin Instituttet, Kommunehospitalet, Copenhagen, Denmark;
and
3
Lilly Research Laboratories, Hamburg, Germany.
Address correspondence and reprint requests to Matthias Braun, Depart-
ment of Physiological Sciences, BMC B11, SE-221 84 Lund, Sweden. E-mail:
matthias.braun@mphy.lu.se.
Received for publication 18 November 2003 and accepted in revised form 19
January 2004.
CNS, central nervous system; FACS, fluorescence-activated cell sorter;
GABA, ␥-aminobutyric acid; LDCV, large dense-core vesicle; SLMV, synaptic-
like microvesicle.
© 2004 by the American Diabetes Association.
1038 DIABETES, VOL. 53, APRIL 2004
inactivation properties of their voltage-gated Na
⫹
channels (14,15). The
experiments were performed either at 32°C (Fig. 1) or at room temperature
(Fig. 4).
Solutions. The standard extracellular solution consisted of (in mmol/l) 138
NaCl, 5.6 KCl, 2.6 CaCl
2
, 1.2 MgCl
2
, 5 HEPES (pH 7.4 with NaOH), and 1–5
glucose. In the experiment shown in FIg. 1B–C, pentobarbital (0.05) and
forskolin (2 mol/l) were included to increase the amplitude of the GABA-
activated Cl
⫺
currents (16) and to elevate intracellular cAMP levels to
stimulate GABA release (12), respectively. When the cells were depolarized
with 50 mmol/l KCl (Fig. 1), the concentration of NaCl was correspondingly
decreased. For the electrophysiological recordings of endogenous GABA
A
receptor activity (Fig. 4), 20 TEA-Cl was substituted for an equal concentra-
tion of NaCl. The pipette solution (intracellular solution) used for all mea-
surements was composed of (in mmol/l) 125 CsCl, 30 CsOH, 10 EGTA, 1
MgCl
2
, 5 HEPES (pH 7.15 with HCl), and 3 Mg-ATP. With this intracellular
solution, activation of Cl
⫺
channels will result in Cl
⫺
efflux at negative
membrane potentials, thus giving rise to an inward current (downward
deflection). In the experiments displayed in Fig. 1D, 9 mmol/l CaCl
2
was added
to the above intracellular solution to increase the intracellular free concen-
tration of Ca
2⫹
to 3 mol/l.
Immunofluorescence. Intact rat islets were fixed with the pH-shift/formal-
dehyde method and permeabilized with 5% Triton X-100. Normal donkey
serum at 5% was used to block unspecific binding. The islets were coincubated
with antibodies directed against insulin (1:200 dilution; Eurodiagnostica,
Malmo¨ , Sweden), glucagon (1:200; Dako, A
¨
lvsjo¨ , Sweden), somatostatin
(1:200; Biogenesis, Poole, U.K.), and the 
2/3
subunits of the GABA
A
receptor
(1:100, clone 62-3G1; Upstate Biotechnology, Lake Placid, NY) at 4°C over-
night. Cy2-, Cy3-, and Cy5-conjugated secondary antibodies (Jackson Immu-
noResearch Laboratories, West Grove, PA) were used to detect the labeled
sites. Immunofluorescence was then detected using a confocal microscope
(Zeiss 510; Zeiss, Go¨ ttingen, Germany).
RT-PCR. Total RNA was extracted from fluorescence-activated cell sorter
(FACS)-purified (17) rat ␣- and -cells using RNeasy Mini Kit (Qiagen, Hilden,
Germany) and used for reverse transcription using random hexamer primers
and Superscript II (Invitrogen, Karlsruhe, Germany). Total RNA from human
brain (Clontech, Heidelberg, Germany) was used as positive control. The
predicted sizes for the bands in the positive control were ␣
1
363 bp, ␣
2
407 bp,
␣
3
505 bp, ␣
4
246 bp, ␣
5
589 bp, ␣
6
177 bp, 
1
270 bp, 
2
342 bp, and 
3
379 bp;
the ␣s and s correspond to previously identified GABA
A
receptor subunits.
PCR amplifications were performed under standard conditions running 40
cycles. Primer sequences are available on request (sewing@lilly.com).
Western blot. Islet homogenate was separated by SDS-PAGE on 8% acryl-
amide gels and transferred onto polyvinylidine fluoride membranes. The
membranes were incubated overnight with the anti-GABA
A

2/3
antibody (2
g/ml, clone 62-3G1; Upstate Biotechnology). After 1 h incubation with
horseradish-peroxidase– coupled secondary antibody (1:20,000 dilution), the
blots were developed by using SuperSignal West Pico Chemiluminescent
Substrate (Pierce, Cheshire, U.K.) and visualized on X-ray films.
Hormone release measurements. Insulin, glucagon, and somatostatin re-
lease was determined by radioimmunoassay as described elsewhere (18,19).
Briefly, batches of 8 –10 islets were preincubated in 1 ml of Krebs-Ringer
buffer (KRB) supplemented with 1 mmol/l glucose for 30 min followed by 1 h
incubation in 1 ml Krebs-Ringer buffer containing 1 or 20 mmol/l glucose. The
specific GABA
A
receptor antagonist SR95531 (10 mol/l; Sigma, St. Louis, MO)
or the specificCa
2⫹
channel antagonists isradipine (2.5 mol/l; kindly
provided by J. Striessing, Innsbruck, Austria), SNX482 (0.1 mol/l; Peptids
International, Louisville, KY), and -conotoxin GVIA (0.1 mol/l; Alomone
Labs, Jerusalem, Israel) were included as indicated. At the end of the
incubation, duplicate aliquots (25–100 l) of the medium were removed and
frozen pending the radioimmunoassay.
Statistical analysis.Data are given as means ⫾ SE. Statistical significances
were evaluated using Student’s t test.
RESULTS
Intercellular GABA signaling between pancreatic is-
let cells. We have previously demonstrated that GABA
FIG. 1. Ability of GABA to function as a paracrine signaling molecule. A: Schematic picture showing the principles of the experiments. Membrane
currents are recorded from a cell voltage clamped at ⴚ70 mV and dialyzed with 10 mmol/l EGTA. Exocytosis from surrounding cells is induced
by depolarization with elevated K
ⴙ
. B: Whole-cell recording from a voltage-clamped cell in a cluster of rat islet cells infected with adenovirus
expressing GABA
A
␣
1

1
receptors. Addition of K
ⴙ
(50 mmol/l) is indicated by the horizontal bar. Inward current transients triggered by GABA
release are indicated by arrows (n ⴝ 4). C: Examples of current transients from B on an extended time base. D: Examples of current transients
due to exocytosis of GABA-containing SLMVs, recorded from an isolated cell. Exocytosis was elicited by intracellular dialysis with 3 mol/l free
Ca
2ⴙ
.
A. WENDT AND ASSOCIATES
DIABETES, VOL. 53, APRIL 2004 1039
can be released from rat -cells by Ca
2⫹
-dependent exo
-
cytosis of SLMVs (12). Here we investigate whether GABA
released from one cell can activate GABA receptors in
neighboring cells, a requirement for GABA to function as a
paracrine regulator. To this end, patch-clamp recordings
were applied to cells within small islet cell clusters in-
fected with adenovirus expressing GABA
A
receptor Cl
⫺
channels. Secretion was stimulated by adding 50 mmol/l
K
⫹
to the extracellular medium. This leads to membrane
depolarization, opening of voltage-gated Ca
2⫹
channels,
and exocytosis in the unclamped cells. The membrane
potential of the patch-clamped cell was kept at ⫺70 mV
throughout the experiment, which is too negative for Ca
2⫹
channel activation and voltage-dependent GABA release
(12). Furthermore, intracellular free Ca
2⫹
was clamped to
nonexocytotic levels by including 10 mmol/l EGTA in the
pipette-filling solution dialyzing the cell interior. Any re-
corded GABA-activated Cl
⫺
currents must accordingly
result from GABA being released by the surrounding
unclamped cells as illustrated schematically in Fig. 1A.
Figure 1B shows the holding current before and during
stimulation with high extracellular K
⫹
. Before stimulation,
the holding current was stable. Following addition of 50
mmol/l K
⫹
to the extracellular medium, a series of nine
fast current transients were observed (arrows). Examples
of these current transients are shown in Fig. 1C on an
expanded time base. Figure 1D shows examples of GABA-
activated Cl
⫺
currents recorded in a single cell in which
exocytosis was stimulated by dialyzing the interior of the
voltage-clamped cell with a Ca
2⫹
-EGTA buffer containing
3 mol/l free Ca
2⫹
instead of the Ca
2⫹
-free medium. The
events shown in Fig. 1C–D show a great resemblance. We
analyzed the activation of the current spikes and their
durations by measuring the t
10 –90%
(i.e., the time it takes
for the current to increase from 10 to 90% of its maximal
amplitude) and half-width (the time the current exceeds
the half-maximal amplitude). Mean values for t
10 –90%
and
the half-width of 15 ⫾ 2 and 53 ⫾ 5ms(n ⫽ 9),
respectively, were obtained. These values are similar to
those recorded from -cells in which exocytosis was
stimulated by intracellular application of 3 mol/l Ca
2⫹
(t
10 –90%
12.6 ⫾ 0.7 ms, half-width 30 ⫾ 1 ms [12]). We
conclude that GABA released from one cell can diffuse to
neighboring cells and that the concentration of the neuro-
transmitter remains sufficiently high to activate GABA
A
receptors in these cells.
GABA
A
receptor subunits are expressed in ␣-cells
but not in -or␦-cells. Ionotropic GABA
A
receptors
have been reported to be endogenously expressed in rat
(20) and human (21) pancreatic islets. We used RT-PCR to
determine the expression pattern of endogenous GABA
A
receptors in FACS-purified ␣- and -cells from rat, utilizing
specific primers for the ␣
1
-␣
6
and the 
1
-
3
subunits of the
GABA
A
receptor. The ␣
1
, ␣
4
, 
1
, 
2
, and 
3
subunits were
expressed in rat pancreatic ␣-cells, whereas none of the
subunits investigated were detected in -cells (Fig. 2). The
identity of all PCR products was confirmed by sequencing.
We verified that the GABA
A
receptor transcripts detected
with RT-PCR translated into protein by performing West-
ern blot on rat islet homogenate. As shown in Fig. 3A,an
antibody directed against the 
2/3
subunits detected a
protein with a molecular weight of ⬃60 kDa, close to the
FIG. 2. RT-PCR on purified ␣-and -cells with specific primers against
different GABA
A
receptor subunits. Total RNA from FACS-purified ␣-
and -cells was reverse transcribed into cDNA, and PCR analysis was
performed with specific primers for the ␣
1– 6
and 
1–3
subunits (ⴙ). In
the negative control reaction (ⴚ), reverse transcriptase was omitted.
Total RNA from human brain was reverse transcribed into cDNA and
used as a positive control. Molecular standards are shown to the left.
FIG. 3. Western blot and confocal immunocytochemistry with a GABA
A
receptor–specific antibody. A: Western blot on rat islet homogenate (⬃150
islets) using a specific antibody against the 
2/3
subunits of the GABA
A
receptor. B and C: Confocal immunocytochemistry of isolated rat islets
detecting the 
2/3
subunits of the GABA
A
receptor, insulin, and glucagon or somatostatin and overlay as indicated. Scale bars: 10 m.
GABAergic SIGNALING IN ISLETS
1040 DIABETES, VOL. 53, APRIL 2004
expected size (57 kDa) (22). Confocal immunocytochem-
istry using the same antibody revealed the presence of
GABA
A
receptors in the glucagon-containing ␣-cells (Fig.
3B), whereas no 
2/3
immunoreactivity was observed in
-cells (Fig. 3B– C)or␦-cells (Fig. 3C).
Electrophysiological measurements of endogenous
GABA
A
receptors. We next applied whole-cell patch-
clamp measurements to confirm the presence of func-
tional GABA
A
receptors in ␣-cells. Application of 1 mmol/l
GABA to functionally identified (see
RESEARCH DESIGN AND
METHODS
and below) ␣-cells elicited inward Cl
⫺
currents
(Fig. 4A). In this particular experiment, the amplitude of
the current was unusually large, reaching a peak ampli-
tude of approximately ⫺600 pA. The amplitude of the
GABA-evoked Cl
⫺
current was extremely variable and
ranged between ⫺20 pA and ⫺600 pA, with an average of
⫺138 ⫾ 93 pA (n ⫽ 6). The current elicited by GABA was
blocked by the specific GABA
A
receptor antagonist
SR95531 (Fig. 4B). As expected for ␣-cells (23), the
capacitance of the cells containing GABA-activated cur-
rents averaged 3.0 ⫾ 0.3 pA (n ⫽ 6), significantly less than
the 4.7 ⫾ 0.2 pF (n ⫽ 31) obtained for -cells (P ⬍ 0.001).
No inward Cl
⫺
currents could be elicited in -cells (not
shown).
Effect of endogenous GABA release on pancreatic
hormone release. ␣-Cells are electrically active (24) and
secrete glucagon in response to membrane depolarizations
and opening of voltage-dependent Ca
2⫹
channels (17).
Because GABA
A
receptors are present in ␣-cells, it is
expected that exocytosis of GABA-containing SLMVs from
-cells will hyperpolarize the ␣-cells with resultant sup-
pression of action-potential firing and glucagon secretion.
This is not easily studied using electrophysiological tech-
niques, but by using the GABA
A
receptor antagonist
SR95531, some insight into the significance of the intrinsic
GABAergic signaling in the islets can nevertheless be
obtained. Pancreatic hormone secretion was measured in
isolated intact rat pancreatic islets at 1 and 20 mmol/l
glucose in the absence and presence of 10 mol/l SR95531.
It should be noted that by using an antagonist, we explore
the role of endogenous GABA and no exogenous GABA
was applied in these experiments. As seen in Table 1,
including SR95531 in the extracellular medium had no
effect on the release of insulin or somatostatin, irrespec-
tive of the glucose concentration. By contrast, glucagon
secretion at 1 and 20 mmol/l glucose was stimulated
approximately two- to threefold in the presence of the
antagonist. Importantly, whereas glucagon secretion was
reduced by ⬃50% under control conditions when raising
glucose to 20 mmol/l, this effect was abolished in the
presence of SR95531.
Ca
2ⴙ
channel antagonists reveal paracrine interac
-
tions between islet cells. Table 2 summarizes the effects
of the L-type Ca
2⫹
channel blocker isradipine (2.5 mol/l),
the N-type Ca
2⫹
channel antagonist -conotoxin GVIA
(100 nmol/l), and the R-type Ca
2⫹
channel blocker SNX482
(100 nmol/l) on insulin, glucagon, and somatostatin secre-
tion at 1 and 20 mmol/l glucose in isolated rat pancreatic
islets. Isradipine had no effect on insulin, somatostatin, or
glucagon release measured at 1 mmol/l glucose. Glucose-
induced insulin secretion was inhibited by ⬎80% by the
L-type channel blocker, whereas there was no significant
effect on somatostatin release. By contrast, SNX482 failed
to affect insulin secretion, but produced an ⬃50% reduc-
tion in glucose-stimulated somatostatin secretion. SNX482
likewise lacked effect on glucagon release measured at 1
mmol/l glucose.
In the presence of 1 mmol/l glucose, -conotoxin GVIA
(0.1 mol/l) inhibited glucagon secretion by ⬃50%, similar
to the inhibition produced by raising glucose to 20 mmol/l.
Importantly, neither isradipine nor SNX482 affected gluca-
gon secretion at 1 mmol/l glucose. Addition of isradipine in
the presence of 20 mmol/l glucose resulted in a paradox-
ical stimulation of glucagon release. Figure 5 summarizes
glucagon release measured at 20 mmol/l glucose under
control conditions and in the presence of 2.5 mol/l
isradipine, 0.1 mol/l SNX482, and 10 mol/l of the GABA
A
receptor antagonist SR95531. These data have been nor-
malized to glucagon secretion measured at 1 mmol/l
glucose. This analysis reveals that, whereas glucagon
secretion in the presence of SR95531 and 20 mmol/l
glucose exceeds that observed at 1 mmol/l glucose, the
release of the hormone measured in the presence of a
combination of 20 mmol/l glucose and 2.5 mol/l isradip-
ine was ⬃80% of that observed at 1 mmol/l glucose. We
propose that isradipine stimulates glucagon release by
FIG. 4. Electrophysiological measurements of endogenous GABA
A
receptors. Cl
ⴚ
current elicited by application of 1 mmol/l GABA to an
␣-cell under control conditions (A) and in presence of the GABA
A
receptor antagonist SR95531 (B).
TABLE 1
Effect of the GABA
A
receptor antagonist SR95531 on hormone release from isolated rat islets
Glucose (mmol/l)
Insulin (ng 䡠 islet
⫺1
䡠 h
⫺1
)
Somatostatin
(pmol 䡠 islet
⫺1
䡠 h
⫺1
)
Glucagon (pg 䡠 islet
⫺1
䡠 h
⫺1
)
1 20 1 20 1 20
Control 0.37 ⫾ 0.03 6.17 ⫾ 0.3 5.05 ⫾ 0.5 24.48 ⫾ 1.1 27.18 ⫾ 2.1 15.31 ⫾ 1.2
10 mol/l SR95531 0.32 ⫾ 0.06 6.01 ⫾ 0.2 5.50 ⫾ 0.2 26.1 ⫾ 1.6 46.72 ⫾ 3.8* 51.04 ⫾ 4.0*†
Data are means ⫾ SE of 10 experiments. *P ⬍ 0.001 vs. control; †NS vs. 10 mol/l SR95531 at 1 mmol/l glucose.
A. WENDT AND ASSOCIATES
DIABETES, VOL. 53, APRIL 2004 1041
inhibiting the release of a paracrine regulator from the
-cell. Given the observation that isradipine was ineffec-
tive when added to islets already exposed to SR95531, it
appears that the compound mediating the effect is GABA.
DISCUSSION
We have recently developed an electrophysiological assay
based on the overexpression of GABA
A
receptor Cl
⫺
channels that allows us to detect exocytosis of single
GABA-containing vesicles from -cells (12). In this study
we have applied the assay to clusters of islet cells (Fig.
1A). As the experimental conditions were designed to
prevent any exocytosis from the patch-clamped cell, the
Cl
⫺
current transients recorded must result from GABA
exocytosis from neighboring cells. The experiment there-
fore suggests that GABA is indeed capable of traveling in
the interstitium between islets cells. Close inspection of
the observed current spikes revealed that they rose with
the same velocity as those recorded in single cells, where
the current transients result from GABA being released
from the same cell. This indicates that the receptor cell
must be sitting very close to the cell in which GABA
release occurred. Indeed, electron microscopy reveals that
the interstitial space between neighboring islet cells is
very small (25), which facilitates paracrine regulation.
GABA
A
receptors are heteromultimeric channels typi
-
cally composed of two ␣, two , and a varying third
subunit. In RT-PCR experiments performed on FACS-
purified islet cells, we found that rat ␣-cells express the ␣
1
and ␣
4
subunit as well as 
1⫺3
(Fig. 2). Although RT-PCR
is not a quantitative method, the intensity of the band for
the ␣
4
subunit suggests that this is the predominant ␣
subunit in ␣-cells. Receptor combinations containing the
␣
4
subunit are found extrasynaptically in hippocampus
and thalamus (26) and characteristically exhibit a higher
affinity for GABA (27). This argues that release of GABA
may activate GABA
A
receptors even beyond the immediate
vicinity of the release sites, which facilitates GABAergic
signaling within the islet. The experiment shown in Fig. 1B
was performed in cells transfected with ␣
1

1
receptors,
which have a lower GABA sensitivity than receptors
containing the ␣
4
subunit (11 mol/l [28] vs. 0.5–2.6 mol/l
[27]). The extent of intercellular GABAergic signaling
might accordingly be even stronger than suggested by the
present data.
To date, most functional experiments on GABAergic
signaling in pancreatic islets have been conducted using
application of exogenous GABA (29,30) or agonists such
as muscimol (7). GABA
A
receptor Cl
⫺
channels inactivate
(i.e., the channels enter a nonconducting state) within
seconds when exposed to GABA concentrations of ⱖ30
mol/l (31). Thus, in studies using application of high
concentrations of GABA to affect hormone release, a
current could flow through the channels only for a brief
period immediately after the addition of the agonist.
Accordingly, during most of the experiment, both mem-
brane potential and secretion are likely to be unaffected. A
way around this problem is to use an antagonist against
the GABA
A
receptor instead and study the influence of
endogenous GABA on pancreatic hormone release. How-
ever, bicuculline, which has been used to this end in the
past has promiscuous pharmacological properties and
inhibits, for example, Ca
2⫹
-activated K
⫹
channels (32) in
addition to the GABA
A
receptors, which may lead to
unspecific effects (M.B., A.W., unpublished data). In this
study we have instead used the selective antagonist
SR95531 and thereby found that, at least in rat islets, only
glucagon secretion is modulated by GABA
A
receptor acti
-
vation. This observation is corroborated by the finding that
GABA
A
receptors are only detected in pancreatic ␣-cells.
The fact that neither somatostatin nor insulin secretion
were affected by the antagonist, although the electrophys-
iology of ␦- and -cells is similar to that of the ␣-cells,
confirms that the action of this compound is selective.
Surprisingly, addition of SR95531 stimulated glucagon
secretion measured at 1 mmol/l glucose. This might indi-
cate that GABA is present at biological concentrations
FIG. 5. Effect of the Ca
2ⴙ
channel blockers SNX482 and isradipine and
the GABA
A
receptor antagonist SR95531 on glucagon secretion from
isolated rat islets. Glucagon secretion was measured in the presence of
20 mmol/l glucose alone or after including 0.1 mol/l SNX482, 2.5
mol/l isradipine, or 10 mol/l SR95531 as indicated. Data are normal-
ized to glucagon release at 1 mmol/l glucose and are means ⴞ SE of
8 –10 experiments.
TABLE 2
Effect of Ca
2⫹
channel antagonists on hormone release from rat islets
Glucose (mmol/l)
Insulin (ng 䡠 islet
⫺1
䡠 h
⫺1
)
Somatostatin
(pmol 䡠 islet
⫺1
䡠 h
⫺1
)
Glucagon (pg 䡠 islet
⫺1
䡠 h
⫺1
)
1 20120 1 20
Control 0.49 ⫾ 0.04 10.39 ⫾ 0.7 4.30 ⫾ 0.6 17.82 ⫾ 2.3 36.85 ⫾ 2.3 19.79 ⫾ 1.2
2.5 mol/l isradipine 0.47 ⫾ 0.05 1.72 ⫾ 0.2* 3.65 ⫾ 0.5 14.06 ⫾ 1.3 34.10 ⫾ 2.5‡ 28.98 ⫾ 3.6§
0.1 mol/l SNX482 0.42 ⫾ 0.04 8.87 ⫾ 0.9 2.92 ⫾ 0.3 9.37 ⫾ 1.3† 39.75 ⫾ 3.9 23.91 ⫾ 3.3
0.1 mol/l -conotoxin GVIA 0.44 ⫾ 0.04 9.40 ⫾ 0.6 3.44 ⫾ 0.5 14.13 ⫾ 2.6 17.03 ⫾ 2.9* 16.06 ⫾ 1.1
Data are means ⫾ SE of 8 experiments. Hormone release was measured at 1 and 20 mmol/l glucose in the absence and presence of isradipine,
SNX482, and -conoxotin GVIA as indicated. Statistical significance is evaluated against the control at the same glucose concentration. *P ⬍
0.001; †P ⬍ 0.01; ‡NS; §P ⬍ 0.05.
GABAergic SIGNALING IN ISLETS
1042 DIABETES, VOL. 53, APRIL 2004
already at low glucose concentrations. Indeed, it has been
estimated that 25% of the total -cell content of GABA (21
pmol per 1,000 -cells) is released per hour regardless of
the glucose concentration (33). This corresponds to a net
release rate of 0.4 amol of GABA, equivalent to approxi-
mately one GABA-containing SLMV, per second and cell.
Our studies on exocytotic GABA release performed on
single cells did not provide evidence for this relatively high
basal release rate (12). However, additional signaling
cascades may be operative in intact islets. For example, it
has been proposed that GABA release from the -cells is
controlled by glutamate, cosecreted with glucagon from
the ␣-cells, rather than by glucose directly (25). It should
be pointed out, however, that even millimolar amounts of
glutamate have no detectable effect of GABA release in our
hands (12). Alternatively, basal GABA release may reflect
passive leakage mediated by an uncharacterized trans-
porter. The relative contributions of basal and stimulated
release to GABAergic signaling remain to be elucidated.
In this study we demonstrate that hormone release from
␣-, -, and ␦-cells depends differentially on Ca
2⫹
influx
through different types of Ca
2⫹
channels. Thus glucose-
induced insulin secretion is almost completely blocked by
the L-type Ca
2⫹
channel blocker isradipine. Glucose-in
-
duced somatostatin release is principally due to SNX482-
sensitive R-type Ca
2⫹
channels, while glucagon secretion
is dependent on -conotoxin–sensitive N-type Ca
2⫹
chan
-
nels (Table 2). Interestingly, isradipine stimulated gluca-
gon secretion in the presence of a maximally inhibitory
concentration of glucose, although this Ca
2⫹
channel
antagonist had no effect when release of the hormone was
stimulated by hypoglycemia. This observation is easiest to
explain if glucagon secretion was suppressed at high
glucose concentrations by a compound released from the
-cells. Theoretically, any substance released from the
-cell could exert such an inhibitory action; possible
candidates include insulin itself, zinc, ATP, islet amyloid
polypeptide, etc. (34). Indeed, both insulin (35) and zinc
(36) have been suggested to suppress glucagon release. It
is pertinent that isradipine had no additional stimulatory
effect in the presence of SR95531. The data therefore
suggest that a significant part of the inhibitory action of
glucose in isolated rat islets is mediated by GABAergic
mechanisms. They also argue that exocytosis of GABA-
containing SLMVs, like insulin-containing LDCVs, depends
on Ca
2⫹
entry via L-type Ca
2⫹
channels. This could either
reflect a requirement of such channels for glucose-induced
electrical activity (37) or exocytosis of SLMVs being
directly coupled to Ca
2⫹
entry via these channels. Our
single-cell measurements of GABA release using -cells
infected with GABA
A
receptors have clearly established
that exocytosis of SLMVs is controlled by Ca
2⫹
entry via
voltage-gated Ca
2⫹
channels (12), but the molecular iden
-
tity of the channel remains to be determined. However,
given that L-type Ca
2⫹
channels account for the largest
part (60%) of the whole-cell Ca
2⫹
current with little
(⬍10%) contribution by both R- and N-type Ca
2⫹
channels
(A.W., M.B., P.R., unpublished results), it seems reason-
able to conclude that the L-type Ca
2⫹
channels are indeed
important for exocytosis of the GABA-containing SLMVs.
Figure 6 summarizes schematically the molecular play-
ers and nature of interactions involved. Glucose-induced
electrical activity of the -cell leads to opening of voltage-
gated L-type Ca
2⫹
channels with subsequent Ca
2⫹
influx
and Ca
2⫹
-dependent exocytosis of both insulin-containing
LDCVs and GABA-containing SLMVs. GABA thus released
diffuses in the narrow space between the release site and
the neighboring ␣-cells, where it activates GABA
A
receptor
Cl
⫺
channels. The rate of release is one vesicle per second
in the absence of glucose but increases 5- to 10-fold during
electrical activity. Each event elicits a rapidly activating
current that lasts ⬎100 ms. Because the ␣-cells have a high
input resistance (few active channels), the activation of
the GABA
A
receptor Cl
⫺
channels will clamp the ␣-cell
membrane potential to the Cl
⫺
equilibrium potential (E
Cl
).
Depending on the intracellular Cl
⫺
concentration, opening
of Cl
⫺
channels will suppress electrical activity by either
hyperpolarizing or depolarizing the ␣-cell. In the former
case, the membrane potential becomes too negative for
action potential generation. In the latter case, the mem-
brane potential becomes so positive that the ion currents
involved in action-potential generation undergo voltage-
dependent steady-state inactivation. The reduction in ac-
tion-potential firing leads to reduced activation of N-type
Ca
2⫹
channels and reduced exocytosis of the glucagon-
containing secretory granules in the ␣-cell. In addition to
N-type Ca
2⫹
channels, ␣-cells are also equipped with
L-type Ca
2⫹
channels (17) that open during depolarization
and contribute to the increases in cytoplasmic free Ca
2⫹
concentration (38,39). However, these channels contribute
little to exocytosis under basal conditions (17,40), but
their importance increases dramatically under certain
physiological situations. For example, L-type Ca
2⫹
chan
-
nels account for 80% of glucagon secretion under experi-
mental conditions associated with enhanced protein
kinase A activity (17).
We acknowledge that paracrine regulation by GABA
may not be the only mechanism influencing glucagon
FIG. 6. Model for GABAergic regula-
tion of glucagon secretion. See text
for details. [Ca
2ⴙ
]
i
, cytoplasmic free
Ca
2ⴙ
concentration; GABA
R
, GABA
A
receptor Cl
ⴚ
channels; VGLC, voltage-
gated L-type Ca
2ⴙ
channels; VGNC,
voltage-gated N-type Ca
2ⴙ
channels;
, membrane potential; 8, an increase
or decrease in ion concentration or
membrane potential (depolarization/
hyperpolarization).
A. WENDT AND ASSOCIATES
DIABETES, VOL. 53, APRIL 2004 1043
release. Release of Zn
2⫹
from the -cells has recently been
proposed to suppress glucagon release in neighboring
␣-cells in rat islets (36). Studies using a novel somatostatin
receptor antagonist have shown that somatostatin inhibits
glucagon secretion in rat islets (41). We point out that
these mechanisms are not mutually exclusive, and it is
indeed possible that all these mechanisms operate in
parallel in ␣-cells in situ. The finding that SR95531 com-
pletely reverses the inhibitory effect of glucose on gluca-
gon secretion strongly suggests, however, that GABAergic
signaling plays an important role in the metabolic control
of glucagon secretion in intact rat pancreatic islets. In
addition, pancreatic -cells contain metabotropic GABA
B
receptors (42), thus providing yet another GABAergic
regulatory mechanism within the islet. This aspect will be
addressed at length elsewhere (M.B., A.W., K.B., A.S., S.S.,
J.G., P.R., unpublished observations). Interestingly, mouse
islets contain much less GABA (2), and GABAergic mech-
anisms such as those we discuss here are consequently
likely to be less important in such islets. This argues that
other mechanisms regulate glucagon secretion in mouse
islets (see for example ref. 43). It is now essential to
establish whether human ␣-cells also are equipped with
GABA
A
receptor Cl
⫺
channels and, if so, to clarify their
role in the control of glucagon secretion. Circumstantial
evidence in support of an important role for GABAergic
mechanisms in human islets comes from the finding that
these islets contain high levels of both GABA and GAD65.
ACKNOWLEDGMENTS
This study was supported by the Swedish Research Coun-
cil (Medicine, grant no. 8647), the Juvenile Diabetes Re-
search Foundation, the Swedish Diabetes Association, the
NovoNordisk Foundation, and the European Commission
(projects Growbeta and Neuronal Ca
2⫹
-channels). We
thank Kristina Borglid and Britt-Marie Nilsson for excel-
lent technical assistance.
REFERENCES
1. Chebib M, Johnston GA: The ‘ABC’ of GABA receptors: a brief review. Clin
Exp Pharmacol Physiol 26:937–940, 1999
2. Michalik M, Nelson J, Erecinska M: GABA production in rat islets of
Langerhans. Diabetes 42:1506 –1513, 1993
3. Taniguchi H, Okada Y, Seguchi H, Shimada C, Seki M, Tsutou A, Baba S:
High concentration of ␥-aminobutyric acid in pancreatic -cells. Diabetes
28:629 – 633, 1979
4. Petersen JS, Russel S, Marshall MO, Kofod H, Buschard K, Cambon N,
Karlsen AE, Boel E, Hagopian WA, Hejnaes KR, Moody A, Dryberg T,
Lernmark A, Madsen O, Michelson BK: Differential expression of glutamic
acid decarboxylase in rat and human islets. Diabetes 42:484 – 495, 1993
5. Lernmark AF: Immunology of Insulin-Dependent Diabetes Mellitus. Ox-
ford, Blackwell Science, 1997
6. Schmidli RS, Colman PG, Bonifacio E, Bottazzo GF, Harrison LC: High
level of concordance between assays for glutamic acid decarboxylase
antibodies: the First International Glutamic Acid Decarboxylase Antibody
Workshop. Diabetes 43:1005–1009, 1994
7. Sorenson RL, Garry DG, Brelje TC: Structural and functional consider-
ations of GABA in islets of Langerhans: -cells and nerves. Diabetes
40:1365–1374, 1991
8. Shi Y, Kanaani J, Menard-Rose V, Ma YH, Chang PY, Hanahan D, Tobin A,
Grodsky G, Baekkeskov S: Increased expression of GAD65 and GABA in
pancreatic -cells impairs first-phase insulin secretion. Am J Physiol
Endocrinol Metab 279:E684 –E694, 2000
9. Rorsman P, Berggren PO, Bokvist K, Ericson H, Mo¨ hler H, O
¨
stenson CG,
Smith PA: Glucose-inhibition of glucagon secretion involves activation of
GABA
A
-receptor chloride channels. Nature 341:233–236, 1989
10. Reetz A, Solimena M, Matteoli M, Folli F, Takei K, De Camilli P: GABA and
pancreatic -cells: colocalization of glutamic acid decarboxylase (GAD)
and GABA with synaptic-like microvesicles suggests their role in GABA
storage and secretion. EMBO J 10:1275–1284, 1991
11. Thomas-Reetz A, Hell JW, During MJ, Walch-Solimena C, Jahn R, De
Camilli P: A ␥-aminobutyric acid transporter driven by a proton pump is
present in synaptic-like microvesicles of pancreatic -cells. Proc Natl Acad
SciUSA90:5317–5321, 1993
12. Braun M, Wendt A, Birnir B, Broman J, Eliasson L, Galvanovskis J,
Gromada J, Mulder H, Rorsman P: Regulated exocytosis of GABA-
containing synaptic-like microvesicles in pancreatic -cells. J Gen Physiol.
In press
13. Olofsson CS, Go¨ pel SO, Barg S, Galvanovskis J, Ma X, Salehi A, Rorsman
P, Eliasson L: Fast insulin secretion reflects exocytosis of docked granules
in mouse pancreatic -cells. Pflugers Arch 444:43–51, 2002
14. Go¨ pel S, Kanno T, Barg S, Galvanovskis J, Rorsman P: Voltage-gated and
resting membrane currents recorded from -cells in intact mouse pancre-
atic islets. J Physiol 521:717–728, 1999
15. Hiriart M, Matteson DR: Na channels and two types of Ca channels in rat
pancreatic -cells identified with the reverse hemolytic plaque assay.
J Gen Physiol 91:617– 639, 1988
16. Birnir B, Tierney ML, Dalziel JE, Cox GB, Gage PW: A structural determi-
nant of desensitization and allosteric regulation by pentobarbitone of the
GABA
A
receptor. J Membr Biol 155:157–166, 1997
17. Gromada J, Bokvist K, Ding WG, Barg S, Buschard K, Renstro¨ mE,
Rorsman P: Adrenaline stimulates glucagon secretion in pancreatic ␣-cells
by increasing the Ca
2⫹
current and the number of granules close to the
L-type Ca
2⫹
channels. J Gen Physiol 110:217–228, 1997
18. Vonen B, Florholmen J, Giaever AK, Burhol P: Radio-immunoassay of
somatostatin from isolated rat pancreatic islets. Scand J Clin Lab Invest
49:135–138, 1989
19. Salehi A, Chen D, Hakanson R, Nordin G, Lundquist I: Gastrectomy induces
impaired insulin and glucagon secretion: evidence for a gastro-insular axis
in mice. J Physiol 514:579 –591, 1999
20. Borboni P, Porzio O, Fusco A, Sesti G, Lauro R, Marlier LN: Molecular and
cellular characterization of the GABA
A
receptor in the rat pancreas. Mol
Cell Endocrinol 103:157–163, 1994
21. Yang W, Reyes AA, Lan NC: Identification of the GABA
A
receptor subtype
mRNA in human pancreatic tissue. FEBS Lett 346:257–262, 1994
22. Li M, De Blas AL: Coexistence of two -subunit isoforms in the same
␥-aminobutyric acid type A receptor. J Biol Chem 272:16564 –16569, 1997
23. Pipeleers DG, in’t Veld PA, Van de Winkel M, Maes E, Schuit FC, Gepts W:
A new in vitro model for the study of pancreatic ␣ and  cells. Endocri-
nology 117:806 – 816, 1985
24. Rorsman P, Hellman B: Voltage-activated currents in guinea pig pancreatic
␣
2
cells: evidence for Ca
2⫹
-dependent action potentials. J Gen Physiol
91:223–242, 1988
25. Hayashi M, Yamada H, Uehara S, Morimoto R, Muroyama A, Yatsushiro S,
Takeda J, Yamamoto A, Moriyama Y: Secretory granule-mediated co-
secretion of L-glutamate and glucagon triggers glutamatergic signal trans-
mission in islets of Langerhans. J Biol Chem 278:1966 –1974, 2003
26. Mody I: Distinguishing between GABA
A
receptors responsible for tonic
and phasic conductances. Neurochem Res 26:907–913, 2001
27. Brown N, Kerby J, Bonnert TP, Whiting PJ, Wafford KA: Pharmacological
characterization of a novel cell line expressing human ␣
4

3
␦ GABA
A
receptors. Br J Pharmacol 136:965–974, 2002
28. Birnir B, Tierney ML, Pillai NP, Cox GB, Gage PW: Rapid desensitization of
␣
1

1
GABA
A
receptors expressed in Sf9 cells under optimized conditions.
J Membr Biol 148:193–202, 1995
29. Gu XH, Kurose T, Kato S, Masuda K, Tsuda K, Ishida H, Seino Y:
Suppressive effect of GABA on insulin secretion from the pancreatic
-cells in the rat. Life Sci 52:687– 694, 1993
30. Gilon P, Bertrand G, Loubatieres-Mariani MM, Remacle C, Henquin JC: The
influence of ␥-aminobutyric acid on hormone release by the mouse and rat
endocrine pancreas. Endocrinology 129:2521–2529, 1991
31. Frosch MP, Lipton SA, Dichter MA: Desensitization of GABA-activated
currents and channels in cultured cortical neurons. J Neurosci 12:3042–
3053, 1992
32. Seutin V, Johnson SW: Recent advances in the pharmacology of quaternary
salts of bicuculline. Trends Pharmacol Sci 20:268 –270, 1999
33. Smismans A, Schuit F, Pipeleers D: Nutrient regulation of ␥-aminobutyric
acid release from islet -cells. Diabetologia 40:1411–1415, 1997
34. Rorsman P, Renstro¨ m E: Insulin granule dynamics in pancreatic beta cells.
Diabetologia 46:1029 –1045, 2003
35. Greenbaum CJ, Havel PJ, Taborsky GJ, Jr, Klaff LJ: Intra-islet insulin
permits glucose to directly suppress pancreatic ␣-cell function. J Clin
Invest 88:767–773, 1991
GABAergic SIGNALING IN ISLETS
1044 DIABETES, VOL. 53, APRIL 2004
36. Ishihara H, Maechler P, Gjinovci A, Herrera PL, Wollheim CB: Islet -cell
secretion determines glucagon release from neighbouring ␣-cells. Nat Cell
Biol 5:330 –335, 2003
37. Antunes CM, Salgado AP, Rosario LM, Santos RM: Differential patterns of
glucose-induced electrical activity and intracellular calcium responses in
single mouse and rat pancreatic islets. Diabetes 49:2028 –2038, 2000
38. Berts A, Gylfe E, Hellman B: Ca
2⫹
oscillations in pancreatic islet cells
secreting glucagon and somatostatin. Biochem Biophys Res Commun
208:644 – 649, 1995
39. Berts A, Gylfe E, Hellman B: Cytoplasmic Ca
2⫹
in glucagon-producing
pancreatic ␣-cells exposed to carbachol and agents affecting Na
⫹
fluxes.
Endocrine 6:79 – 83, 1997
40. Gromada J, Hoy M, Buschard K, Salehi A, Rorsman P: Somatostatin
inhibits exocytosis in rat pancreatic ␣-cells by G
i2
-dependent activation of
calcineurin and depriming of secretory granules. J Physiol 535:519 –532,
2001
41. Cejvan K, Coy DH, Efendic S: Intra-islet somatostatin regulates glucagon
release via type 2 somatostatin receptors in rats. Diabetes 52:1176 –1181,
2003
42. Brice NL, Varadi A, Ashcroft SJ, Molnar E: Metabotropic glutamate
and GABA
B
receptors contribute to the modulation of glucose-stimu
-
lated insulin secretion in pancreatic -cells. Diabetologia 45:242–252,
2002
43. Go¨ pel SO, Kanno T, Barg S, Weng XG, Gromada J, Rorsman P: Regulation
of glucagon release in mouse ␣-cells by K
ATP
channels and inactivation of
TTX-sensitive Na
⫹
channels. J Physiol 528:509 –520, 2000
A. WENDT AND ASSOCIATES
DIABETES, VOL. 53, APRIL 2004 1045