Basal GABA Regulates GABABR Conformation and
Release Probability at Single Hippocampal Synapses
Tal Laviv,1,3Inbal Riven,1,3,4Iftach Dolev,1Irena Vertkin,1Bartosz Balana,2Paul A. Slesinger,2and Inna Slutsky1,*
1Department of Physiology and Pharmacology, Sackler Faculty of Medicine, Tel Aviv University, 69978 Tel Aviv, Israel
2Peptide Biology Laboratory, The Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, CA 92037, USA
3These authors contributed equally to this work
4Present address: Department of Chemical Physics, Weizmann Institute of Science, 76100 Rehovot, Israel
are composed of GB1a/GB2subunits and critically
influence synaptic and cognitive functions. Here, we
explored local GABABR activation by integrating
optical tools for monitoring receptor conformation
and synaptic vesicle release at individual presynaptic
boutons of hippocampal neurons. Utilizing fluores-
cence resonance energy transfer (FRET) spectros-
copy, we detected a wide range of FRET values for
CFP/YFP-tagged GB1a/GB2receptors that negatively
synapses. High FRET of GABABRs associated with
low release probability. Notably, pharmacological
manipulations that either reduced or increased basal
receptor activation decreased intersynapse vari-
ability of GB1a/GB2receptor conformation. Despite
variability along axons, presynaptic GABABR tone
was dendrite specific, having a greater impact on
synapses at highly innervated proximal branches.
Prolonged neuronal inactivity reduced basal receptor
activation, leading to homeostatic augmentation of
release probability. Our findings suggest that local
variations in basal GABA concentration are a major
determinant of GB1a/GB2conformational variability,
which contributes to heterogeneity of neurotrans-
mitter release at hippocampal synapses.
Probability of neurotransmitter release (Pr) is a critical determi-
nant of synaptic strength in neuronal connections. Presynaptic
inhibition mediated via G protein-coupled receptors (GPCRs)
constitutes a widespread, evolutionary conserved, regulatory
root of neurotransmitter release (Wu and Saggau, 1997). Among
diverse presynaptic GPCR types, GABABRs are known to inhibit
basal synaptic transmission in a wide range of central synapses.
GABABRs are obligatory heterodimers, requiring two homolo-
gous subunits, GB1and GB2, for functional expression (Jones
et al.,1998; Kaupmann et al., 1998;White et al.,1998). Molecular
guishable GB1isoforms, 1a and 1b (Bettler et al., 2004) that
couple to the pertussis toxin-sensitive G proteins, Giand Go.
Recent generation of isoform-specific GB1knockout mice re-
vealed a critical role of GB1a/GB2receptors, mainly localizing at
presynaptic boutons, in regulation of long-term potentiation and
hippocampus-dependent memory function (Vigot et al., 2006).
Although existence of GABABR heterodimers as molecular
entities has been recently accepted, physiological patterns of
neuronal activity that induce local receptor activation at
presynaptic boutons and, consequently, inhibition of neuro-
transmitter release, remain controversial. Electrophysiological
studies in hippocampal slices provided conflicting observations,
ranging from tonic inhibition of glutamate release mediated by
GABABRs during spontaneous synaptic activity (Jensen et al.,
2003), to a necessity of high-level interneuron activity to activate
presynaptic heteroreceptors (Isaacson et al., 1993). Conversely,
only postsynaptic receptor activation was detected during phar-
macologically induced rhythmic oscillations (Scanziani, 2000).
These studies typically estimated average effects of endoge-
nous GABA on neurotransmitter release over populations of
synapses. Given a profound heterogeneity in release properties
across hippocampal synapses(Branco etal.,2008;Dobrunzand
Stevens, 1997; Murthy et al., 1997; Rosenmund et al., 1993;
Slutsky et al., 2004), averaging might obscure the influence of
local extracellular GABA concentration ([GABA]o) at the single-
In attempt to explore local GABABR activation at individual
presynaptic boutons, weintegrated optical techniques for simul-
taneous monitoring of protein-protein interactions and synaptic
vesicle recycling. Fluorescence resonance energy transfer
(FRET) spectroscopy has been utilized to estimate inter-
molecular associations between CFP/YFP-tagged receptor
subunits, whereas activity-dependent FM styryl dyes have
been used to detect presynaptic activity at single hippocampal
boutons. We have used this system to address the following
questions: does basal GABA activate the GB1a/GB2receptor
and decrease synaptic vesicle release probability? If so, does
this basal receptor activation vary across synapses, contributing
to heterogeneity of presynaptic strength? Do presynaptic
boutons adapt to chronic changes in ongoing neuronal activity
by regulating the extent of basal GABABR activation?
Our results reveal that basal synaptic activity promotes GB1a/
GB2association by a variable degree across boutons along
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 253
a given axon, contributing to a nonuniform inhibition of vesicle
release inhippocampal neurons.
activation arises from local differences in basal [GABA]oin the
vicinity of synapses. The presynaptic GABABR tone is differen-
tially regulated along the dendritic tree and homeostatically
controlled by prolong changes in neuronal activity. These data
suggest a critical role for GB1a/GB2receptor in local regulation
of release probability and its adaptation to changes in ongoing
GABABR FRET Reveals Variability at Single Excitatory
Binding of GABA to the GABABR and subsequent receptor acti-
vation is the first step toward producing presynaptic inhibition.
To explore the regulation of transmitter release by the presyn-
aptic GB1a/GB2receptors at individual hippocampal synapses,
we searched for an optical detector of GABABR activation by
endogenous GABA. We utilized FRET spectroscopy to study
conformational changes between fluorescently tagged GB1a
(GB1aCFP) and GB2(GB2YFP) receptor subunits (Fowler et al.,
2007). FRET provides an accurate measure of possible changes
in the relative distance (<100 A˚) and/or orientation between the
fluorophores. First, we explored interactions between GB1aCFP
and GB2YFPproteins expressed at excitatory boutons in cultured
Presynaptic localization of the tagged GABABRs in boutons
was confirmed by colocalization of GB1aYFPreceptor subunit
with CFP-tagged synapsin Ia protein (Syn1aCFP; see Figure S1A
available online) and with FM4-64 dye (Figure 1D), a marker of
presynaptic vesicle turnover. We measured the steady-state
(Figures 1C, 1D, and S1B–S1E), which dequenches the donor
(CFP). FRET efficiency was measured under miniature synaptic
activity (Emini) in the presence of tetrodotoxin (+TTX), which
inhibits spike-dependent neurotransmitter release.
To estimate Eminiusing a spectroscopic approach, CFP/YFP
emission were collected in the range of 400–700 nm under
405 nm excitation wavelength, at single presynaptic boutons
expressing GB1aCFPand GB2YFPproteins. A typical emission
spectrum (Figure 1C) and high-magnification confocal images
(Figure 1D) at the presynaptic bouton marked by arrowhead in
Figures 1A2and 1A3show an increase in CFP fluorescence after
YFP photobleaching, indicating dequenching of the donor and
the presence of FRET. The photobleaching was specific for the
selected region of interest and did not produce any detectable
damage of presynaptic function (see technical details in
Figure S1 and Experimental Procedures). The bouton activity
was confirmed by its staining with FM4-64 under maximal elec-
trical stimulation (600 action potentials at a rate of 10 Hz) or high
potassium solution (50 mM HiK, 1 min) (Figure 1D). Quantitative
analysis of FRET signals between GB1aCFPand GB2YFPproteins
across different boutons along a single axon under miniature
synaptic activity revealed a surprising heterogeneity in Emini,
with a range of 0.12 to 0.33 and a coefficient of variation
(CV = SD/mean) of 34% (Figure S2A). On average, Eminiwas
0.16 ± 0.01 (CV = 40%) across 38 functional boutons from
7 neurons (p < 0.0001; Figure 1E). The observed FRET signals
did not depend on the expression level of the acceptor
Emission wavelength (nm)
Figure 1. Monitoring GB1a/GB2Interactions
by FRET at Hippocampal Boutons
(A) Representative confocal images of pyramidal
neuron in hippocampal culture that was cotrans-
fected with GB2YFPand GB1aCFP. White box in
(A1) corresponds to the blow-ups in (A2) and (A3).
Arrowheads: the bouton that was bleached for
calculation of FRET efficiency in (C) and (D).
constructs used for FRET detection: GB1aCFP/
GB2YFP, GB1aCFP/TNFR2YFP, Gb1N’-YFP/Gg2N’-CFP,
(C) Detection of FRET efficiency (E) by spectral
approach under 405 nm excitation in a single
(FD) photobleaching. E = 0.24.
(D) Pseudo-color coded fluorescent images of
a functional boutonbefore
photobleaching. Note increase of CFP fluores-
cence (ex: 405 nm; em: 476–496 nm) after YFP
(ex: 514 nm; em: 530–550 nm) photobleaching.
FM4-64: DF image of synaptic vesicle turnover
during maximal stimulation (600 APs @ 10 Hz).
Arrowhead: FM+punctum which is colocalized
(E) Summary of Eminidata for GB1aCFPonly (n = 18,
N =5), GB1aCFP/TNFR2YFP(n=11, N= 4),GB1aCFP/
proteins expressed in
GB2YFP(n = 38, N = 7), and Gg2N’-CFP/Gb1N’-YFP(n = 13, N = 4), Gg2N’-CFP/TNFR2YFP(n = 9, N = 3). Eminifor GB1aCFP/GB2YFPand Gg2N’-CFP/Gb1N’-YFPwere
significantly higher (p < 0.0001) than nonspecific GB1aCFP/TNFR2YFPand Gg2N’-CFP/TNFR2YFPinteractions. Scale bars: 10 mm (A1), 2 mm (A3, D).
Error bars represent SEM. See also Figure S2.
GABABR Conformation and Release Probability
254 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
molecules per bouton (as determined by total YFP fluorescence
intensity), on the expression level of the donor molecules per
bouton (as determined by CFP fluorescence intensity), and on
the donor-to-acceptor ratio (Figures S2B–S2D), excluding these
parameters as a potential source of variability. To rule out non
specific association due to overexpression of the tagged
proteins, we cotransfected GB1aCFPand a nonrelated tumor
necrosis factor receptor 2, C terminally tagged with YFP
(TNFR2YFP). These two CFP/YFP tagged proteins showed
negligible FRET (0.023 ± 0.004, n = 11; Figure 1E). Furthermore,
background enhancement of CFP emission, assessed by
photobleaching at 514 nm in neurons expressing only GB1aCFP,
was<3% (0.027±0.003,n=18;Figure 1E). Therefore, thehigher
(p < 0.0001) are due to specific association of GB1aCFPand
GB2YFPproteins in functional boutons.
GB1a/GB2heterodimer, we measured the FRET between another
known heterodimer, the Gb1g2G protein subunits (Figure 1B).
In contrast to the variable FRET between GB1aCFPand GB2YFP
receptor subunits, the FRET between b1YFPand g2CFPproteins
across boutons was uniform (CV = 8%) and higher (0.23 ±
0.005, n = 13; Figures 1E and S2E–S2G). Background FRET was
negligible between TNFR2YFPand Gg2N’-CFP(0.02 ± 0.005, n = 9;
Figure 1E). Taken together, these results suggest that the vari-
ability in FRET is a specific feature of the GB1a/GB2association.
Basal [GABA]oRegulates Local GB1a/GB2Association
at Excitatory Boutons
We next investigated the possible source of inter-synapse
GB1a/GB2FRET variability at excitatory boutons. We hypothe-
sized that variations in local extracellular GABA concentration
([GABA]o) in the vicinity of synapses during miniature synaptic
activity might account for the variability. To examine this possi-
bility, we compared FRET efficiency between GB1aCFPand
GB2YFPmeasured under miniature synaptic activity (+TTX, Emini)
and FRET efficiency measured at nonfunctional boutons lacking
synaptic vesicle release (Erest). We utilized two different strate-
gies to create inactive boutons (Figure S3): (1) block of
SNARE-mediated vesicle release by tetanus toxin (TeTx); (2)
investigating immature boutons of young neurons (4–5 DIV). In
contrast to the high variability of Emini(CV = 39%), Erestwas
less variable between GB1aCFPand GB2YFPproteins in boutons
of neurons lacking synaptic vesicle release. The CV decreased
(Figure 2A). In addition, mean Erest was significantly lower
(p < 0.01) in either TeTx-treated (0.12 ± 0.03, n = 24) or immature
neurons (0.11 ± 0.005, n = 15), compared to Eminiunder basal
conditions (0.16 ± 0.02; n = 20). These results indicate that in
the absence of SNARE-mediated neurotransmitter release,
GB1aCFP/GB2YFPFRET signals do not significantly vary across
boutons. We propose that miniature vesicle release promotes
FRET between GB1aCFPand GB2YFPproteins by a variable
degree according to local GABA levels in the vicinity of boutons.
Increase in FRET between GB1aCFPand GB2YFPproteins could
be due to either change in the conformation of pre-existing
dimers or to the equilibrium between monomers and dimers.
As several studies indicate that heterodimerization of GB1a/GB2
in both TeTx-treatedand immature neurons
is required for functional receptor expression (Jones et al., 1998;
Kaupmann et al., 1998; White et al., 1998), we favor the interpre-
tation that activity-dependent conformational changes occur in
GB1a/GB2heterodimers. Altogether, these results confirm that
formation of GB1aCFP/GB2YFPheterodimers does not require
synaptic activity and suggest that intrinsic GB1aCFP/GB2YFP
associations do not contribute to the observed inter-synapse
FRET variability under miniature synaptic activity.
To explore whether levels of GABA were involved in deter-
mining the extent of FRET between GB1aCFPand GB2YFPat indi-
vidual boutons, we assessed the effect of different pharmaco-
logical manipulations. We first measured Eminis before and
CGP54626 for different population of boutons in the same axon.
Interestingly, CGP54626 (1 mM) reduced both the CV (from 41%
to 19%) and the mean Eminifor GB1aCFPand GB2YFP(from 0.17 ±
0.016 to 0.12 ± 0.007, n = 12–15, p < 0.01; Figure 2B, left graph).
Notably, CGP54626 did not affect Erestin the absence of GABA
release (n = 14–16, p > 0.4; Figure 2B, right graph), suggesting
a specific effect of the antagonist on released GABA. To test
the effect of antagonist at the same population of boutons (i.e.,
without photobleaching), we monitored the ratio of YFP to CFP
emission (FYFP/FCFP) under CFP excitation before and after
ratio and decreased its variability (p < 0.001; Figures S4A–S4C),
similar to the donor dequenching method.
Because antagonizing basal GABA reduced the FRET
between GB1aCFPand GB2YFP, we hypothesized that maximal
receptor activation would increase the FRET efficiency but also
reduce the variability. To investigate this possibility, we exam-
ined the effect of a saturating dose of baclofen, a competitive
selective agonist of GABABRs. Baclofen (10 mM) increased Emini
between GB1aCFPand GB2YFPsubunits to 0.29 ± 0.007 (n = 12,
p < 0.0001; Figure 2C, left graph) with a concomitant ?5-fold
reduction of CV to 8%. Notably, baclofen induced strong FRET
in immature neurons (n = 9, p < 0.0001; Figure 2C, right graph).
Baclofen also increased the FYFP/FCFPratio and reduced signal
variability at the same population of boutons (n = 9, p < 0.001;
Figure S4D). The observed baclofen-induced increase in FRET
reversed upon washout (Figures S4D and S4E). The effect of ba-
clofen on GB1aCFP/GB2YFPFRET was specific since it did not
induce FRET betweenGB1aCFP
(Figure S4F). These results complement recent FRET study on
agonist-induced rearrangements between GB1a and GB2
subunits tagged at intracellular loops (Matsushita et al., 2010).
Next, we examined whether GABA uptake affects GB1a/GB2
interactions through regulation of [GABA]o under miniature
synaptic activity. The specific GABA transporter inhibitor
SKF-89976A (25 mM) was used to raise [GABA]o. Blocking GABA
uptake induced ?2-fold increase in the mean level of Eminifrom
0.13 ± 0.01 to 0.27 ± 0.02, while decrease in CV from 43% to
24%(n = 14–15, p < 0.0001; Figure 2D). Thus, similar to baclofen,
elevating [GABA]odrives GABABRs to an activated state (high
FRET). Furthermore, these results suggest that GABA transporter
Finally, we probed the pattern of GB1aCFP/GB2YFPinteractions
under GABA release evoked by action potentials (Figure 2E).
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 255
In these experiments, recurrent network activity was blocked by
a nonselective antagonist of ionotropic glutamate receptors
(kynurenic acid). Activity-induced FRET changes were reversible
(Figure S4G), enabling FRET monitoring under miniature
(no stimulation) versus evoked (by 5 Hz field stimulation) release
at the same population of boutons. Interestingly, evoked release
produced ?20% increase in the mean E and preserved the vari-
ability (CV = 38%, n = 16; Figure 2E, left graph). Importantly, 5 Hz
stimulation did not change FRET in nonreleasing TeTx-treated
neurons (n = 11, p > 0.4; Figure 2E, right graph), indicating the
requirement for synaptic release of GABA.
Taken together, these data suggest that (1) GB1aCFP/GB2YFP
form stable dimers in the absence of synaptic activity and
change their conformation in response to agonist (Figure 2F);
(2) GB1a/GB2 FRET variability decreases in the absence of
assembly; (3) [GABA]onear synaptic contacts induces partial
GABABR activation per excitatory bouton and underlies vari-
ability of GB1a/GB2associations under basal synaptic activity.
Therefore, C terminally tagged GB1a/GB2 receptor subunits
provide highly sensitive detectors for [GABA]oin the vicinity of
excitatory synapses and for GABABR activation.
Although presynaptic GB1a/GB2receptors are predominantly
localized at excitatory boutons as heteroreceptors, a fraction
of GABAergic terminals expresses GB1a/GB2 autoreceptors
(Vigotetal.,2006).Wedetected only 7%(3/41)of inhibitorycells,
which expressed (GB1aCFP
+ GFP) proteins from whole
population of the transfected cells, comparing to 30% (15/49)
of inhibitory cells that expressed GFP alone. Inhibitory cells
were identified by the lack of spines and by immunostaining
with GAD-65 antibody after FRET measurements. To compare
the properties of auto- versus heteroreceptors, we analyzed
FRET efficiencies at a fraction of inhibitory boutons expressing
GB1aCFP/GB2YFPautoreceptors (Figure 3A). Under miniature
synaptic activity, FRET efficiencies at inhibitory boutons were
higher (0.21 ± 0.01, n = 12; Figure 3B, control) and less variable
(CV =16.6%) than at excitatory ones (Figure 2). Inthe absence of
synaptic activity, FRET efficiencies within the autoreceptors
reduced to ?10% (0.11 ± 0.005 and 0.10 ± 0.006, n = 9–13,
p < 0.001, for immature and TeTx-treated neurons, respectively;
Figure 3B), similarly to the Erestlevels measured within the heter-
oreceptors. Application of CGP54626 antagonist also resulted in
a reduction of Eminito the Erestlevel (0.10 ± 0.008, n = 9–11,
p < 0.0001; Figure 3C), suggesting higher basal [GABA]oin the
+/- miniature release
t s i nog a + t s i nog a t na +
+ GAT-1 inhibitor
Figure 2. Analysis of Intersynapse Heterogeneity of GB1a/GB2Associations at Excitatory Boutons
(A) FRET efficiency between GB1aCFPand GB2YFPin functional boutons under miniature activity (+TTX, control group, 0.16 ± 0.02, CV = 39%, n = 20, N = 5) and
nonfunctional boutons: TeTx-pretreated (0.12 ± 0.03, CV = 14%, n = 24, N = 6) and immature (4–5 DIV, 0.11 ± 0.005, CV = 14%, n = 15, N = 4) neurons. Mean E is
lower and less variable in nonfunctional boutons.
(B) CGP54626 (1 mM) reduced mean Eminito 0.12 ± 0.007 and its variability to CV = 19% in control (left, n = 12–15, N = 4), but not in TeTx-treated neurons (right,
n = 14–16, N = 4).
(C) Baclofen (10 mM) increased mean Eminiand reduced its variability in control (left, 0.29 ± 0.007, CV = 8%, n = 12–17, N = 4, p < 0.0001) and in immature (right,
0.28 ± 0.004, CV = 5%, n = 9, N = 3 p < 0.0001) neurons.
(D) SKF-89976A (25 mM) increased mean Emini(0.27 ± 0.02, n = 14–15, N = 4, p < 0.0001) and decreased CV from 43% to 24%.
(E) Left: Comparison of E monitored under miniature activity and during 5 Hz stimulation (both conditions in the presence of kynurenic acid) over the same
population of synapses in control neurons (n = 16, N = 4). Stimulation increased mean E (p < 0.001) but did not affect CV (44% and 38% for mini and 5 Hz, respec-
tively). Right: In TeTx-treated neurons, 5 Hz did not affect E (n = 11, N = 3, p > 0.4).
(F) Schematic illustration of increase in FRET within the GB1aCFP/GB2YFPheterodimer by endogenous [GABA]o.
Error bars represent SEM. See also Figure S4.
GABABR Conformation and Release Probability
256 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
vicinity of GB1aCFP/GB2YFPautoreceptors. Conversely, Eminis
were increased by baclofen (0.28 ± 0.007, n = 10–13, p < 0.01;
Figure 3D) and GABA transporter blocker SKF-89976A (0.32 ±
0.014, n = 9–11, p < 0.0001; Figure 3E).
In summary, FRET measurements at GABAergic boutons
suggest that (1) in the absence of synaptic activity, assembly
of GB1a/GB2autoreceptors is similar to those of heterorecep-
tors; (2) activation of GB1a/GB2autoreceptors is stronger and
less variable due to higher local [GABA]o. Therefore, GB1a/GB2
GABAergic boutons, may induce autoinhibition of quantal
GABA release, and, therefore, contribute to intersynapse vari-
ability of basal [GABA]o.
Agonist-Induced Uniform GABABR Activation
We then examined the potency of baclofen to induce conforma-
tional changes in the GB1a/GB2heterodimer. First, we used
HEK293 cells as a heterologous expression system to examine
possible agonist-induced FRET changes between the tagged
GB1aand GB2receptor subunits (Figure S5A). Utilizing donor-
dequenching method, we measured FRET efficiency of 0.095 ±
0.004 under resting conditions (Figures S5B and S5C), similarly
to previous results (Fowler et al., 2007). FRET did not depend
on the expression levels of acceptor, donor, and donor/acceptor
ratio (Figures S5D–S5F). To test FRET signals before and after
baclofen application, weutilized nondestructive sensitizedemis-
increase in FRET (n = 10, p < 0.01; Figure S5G), accompanied by
an increase in G protein-gated inwardly rectifying potassium
currents (n = 10; Figure S5H) as a readout of receptor activation.
These data suggest that agonist-induced increase in FRET
between the C terminally tagged GB1a/GB2receptor subunits
reflects receptor activation.
Then, we measured the potency of baclofen to induce confor-
mational changes in the GB1aCFP/GB2YFPreceptor expressed in
presynaptic boutons and compared to changes in presynaptic
Ca2+transients and in basal vesicle release. Baclofen dose-
response curve for FRET efficiency between GB1aCFPand
GB2YFPrevealed an ED50 of 0.83 ± 0.006 mM (n = 8–10;
Figure 4A), which is comparable to the affinity for native
GB1a/GB2receptors (?0.6 mM; Malitschek et al., 1998). Next,
we constructed a dose-response curve for baclofen inhibition
on presynaptic Ca2+transients. Functional boutons were identi-
fied by FM4-64 marker (Figure 4B). Presynaptic Ca2+transients
evoked by low-frequency stimulation were measured by high-
affinity fluorescent calcium indicator Oregon Green 488
fluorescence, while decreased the size of action-potential
dependent fluorescence transients (DF/F; Figure 4B). Baclofen
induced concentration-dependent inhibition of presynaptic
of 0.46 ± 0.01 mM (n = 10–15; Figure 4A), similarly to its effect
in hippocampal slices (Wu and Saggau, 1995). Finally, we esti-
mated the potency of baclofen on inhibition of synaptic vesicle
turnover. Activity-dependent FM1-43 dye has been used to
estimate changes in basal synaptic vesicle turnover at hippo-
campal synapses (Abramov et al., 2009). We quantified the total
amount of releasable fluorescence (DF) at each bouton following
stimulation by 30 action potentials at a rate of 0.2 Hz in the pres-
ence of 10 mM FM1-43 and subsequent destaining (Figures S6A
and S6B). Analysis of baclofen (1 mM) effect revealed an average
decrease of DF across synapses (Figures 4C and 4D). The effect
of baclofen was uniform across boutons, no correlation was
found between presynaptic changes and initial DF values
(Spearman r = –0.1, p > 0.2; Figure 4E). Given that baclofen
inhibits FM destaining rate (Figure S7) without affecting synaptic
2+transients with maximal inhibition level of ?20% and ED50
Figure 3. Analysis
(A) Representative confocal images of an inhibitory neuron in
hippocampal culture that was cotransfected with GB2YFP
and GB1aCFP. White box in (A1) corresponds to the blow-ups
in (A2) and (A3). Scale bars: 10 mm (A1), 2 mm (A3).
(B) FRET efficiency between GB1aCFPand GB2YFPin functional
boutons under miniature activity (+TTX, control group,
0.21 ± 0.01, CV = 16%, n = 12, N = 4) and nonfunctional bou-
tons: TeTx-pretreated (0.10 ± 0.006, CV = 20%, n = 13, N = 3)
and immature (0.11 ± 0.005, CV = 15%, n = 9, N = 3) neurons.
(C) The GABABR competitive antagonist CGP54626 (1 mM)
reduced mean Eminito 0.10 ± 0.008 (CV = 23%, n = 9–11,
N = 2).
(D) The GABABR competitive agonist baclofen (10 mM)
increased mean Eminito 0.28 ± 0.07, (CV = 9%, n = 10–13,
N = 3, p < 0.01).
(E) GABA uptake inhibitor SKF-89976A (25 mM) increased
mean Eminito 0.32 ± 0.01 (CV = 15%, n = 9–11, N = 2,
p < 0.0001).
Error bars represent SEM.
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 257
vesicle endocytosis (Isaacson and Hille, 1997), the observed ba-
clofen-induced reduction in DF signals primarily reflects inhibi-
tion of vesicle exocytosis. Applying this method for construction
of a full dose-response curve, we estimated an ED50of 0.40 ±
0.06 mM for baclofen-induced presynaptic inhibition in our
experimental system (Figure 4A). The similarity in potency of ba-
clofen to increase GB1a/GB2FRET and to inhibit both presyn-
aptic Ca2+transients and vesicle release indicates a tight
coupling between the GB1a/GB2conformational dynamics and
functional activation of presynaptic GABABreceptors.
Variability of Basal GABABR-Mediated
To assess the role of GABABRs in presynaptic inhibition induced
by basal GABA release, we quantified synaptic vesicle recycling
at hippocampal synapses before and after application of the
GABABRantagonist. First, wetested CGP54626 effectson basal
vesicle release evoked by low frequency (0.2 Hz) stimulation.
Under these conditions, CGP54626 nonuniformly increased
fluorescence intensity of FM1-43 signals at individual boutons
(Figure 5A). Analysis of the CGP54626effect across 236boutons
revealed anaverage increase of DF across synapses (Figure 5B).
However, in contrast to uniform effect induced by agonist
(Figure 4E), antagonist predominantly affected boutons exhibit-
ing low initial DF values (Figure 5C; Spearman r = ?0.9,
p < 0.0001). At the level of synaptic population, CGP54626
increased the total presynaptic strength, defined as S = DF 3
D, whereas D is the density of FM+boutons per image area
(1.66 ± 0.92, N = 6, p < 0.0001; Figure 5D). In addition to
enhancement of evoked basal presynaptic strength (0.2 and
1 Hz, Figure 5D), CGP54626 increased miniature presynaptic
strength by similar extent (1.56 ± 0.16, N = 6, p > 0.5;
Figure 5D) in a nonuniform manner (Spearman r = –0.6,
p < 0.0001). Taken together, these results suggest that under
both miniature and evoked basal synaptic activity GABABRs
are activated by GABA, leading to a synapse-specific presyn-
GB1a/GB2Associations Correlate to Pr
at Single Synapses
To assess the influence of GB1a/GB2conformational state on
tonic inhibition of synaptic vesicle release, we simultaneously
monitored the effect of GABABR antagonist on GB1aCFP/GB2YFP
FRET ratio and on basal vesicle recycling at individual
synapses of pyramidal neurons. Neuronal cultures expressing
GB1aCFPand GB2YFPfusion proteins were subjected to FM4-
64 staining under evoked (by 0.2 Hz stimulation) or miniature
(+TTX) synaptic activity. Ectopic expression of GB1aCFP/GB2YFP
fusion proteins did not affect the level of baclofen-induced
presynaptic inhibition (Figures S6C and S6D). Figure 5E
demonstrates differential effect of CGP54626 antagonist at
two boutons under 0.2 Hz stimulation. At the bouton #1,
CGP54626 induced a ?25% decrease in FYFP/FCFP (due to
increase in FCFPwith concomitant reduction in FYFP) and an
?96% increase in DFFM4-64. In contrast, CGP54626 did not
significantly alter either FYFP/FCFPor DFFM4-64at the bouton #2.
The pooled data (Figure 5F) demonstrate a correlation between
the degree of antagonist-induced decrease in FYFP/FCFPratio
within the GB1aCFPand GB2YFPpair and increase in basal
synaptic vesicle recycling under evoked (n = 13, Spearman
r = -0.1
FBac / FCnt
slope = 1
log [baclofen] (M)
Percentage of E
CFP + GB2
Percentage of [Ca]i ( )
and Pr ( ) inhibition
Figure 4. Baclofen DisplaysSimilarPotency
for GB1a/GB2Conformational Changes and
Presynaptic Inhibition of [Ca2+]iand Basal
(A) Dose-response curves of baclofen on E
between GB1aCFPand GB2YFPproteins (n = 8–10,
sients (n = 10–15, ED50= 0.46 ± 0.01 mM), and on
presynaptic activity measured by FM1-43 dye
(n = 4–7, ED50= 0.40 ± 0.06 mM). E at 100 mM
baclofen was set as 100%.
(B) Activation of GABABRs with baclofen reduced
Functional presynaptic boutons were detected
by FM4-64 dye staining (B1, dotted line shows
position of line scan). Ca2+transients were evoked
by 0.2 Hz stimulation during 500 Hz line scan of
boutons before and after baclofen application
(B2and B3). Ca2+transients were quantified as
DF/F before (black) and after (green) baclofen
application (average of 10 traces). Scale bar:
150 ms (B3), 2 mm (B1).
(C) Representative DFFM1-43images before and
Stimulation during FM1-43 staining: 30 action
potentials at 0.2 Hz (see Figures S6A and 6B
for details). Scale bar: 2 mm. Fluorescence intensities (arbitrary units) are coded using a pseudocolor transformation.
(D) Single-synapse analysis of baclofen effect across 118 boutons (slope of linear fit is 0.47 ± 0.02). Dotted line designates no change (slope = 1).
(E) Baclofen-induced inhibition did not correlate to initial presynaptic strength (Spearman r = –0.1, p > 0.2, the same data as in D).
Error bars represent SEM. See also Figure S6.
GABABR Conformation and Release Probability
258 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
r = –0.95, p < 0.0001) or miniature (n = 10, Spearman r = –0.93,
p < 0.0001) activity at individual boutons. These data indicate
that GB1a/GB2conformation may predict the level of tonic inhi-
bition at the single-bouton level.
The above relationship suggests that basal vesicle release
might be fine-tuned through dynamic regulation of the
GB1a/GB2receptor conformation. To directly examine the rela-
tionship between vesicle exocytosis per se and GB1a/GB2asso-
ciations, we monitored FM destaining kinetics (Figure S7A),
a reliable indicator of release probability, together with FRET
proteins at individual boutons. For this purpose, the total pool
of recycling vesicles has been stained by 600 action potentials
at a rate of 10 Hz and colocalized FM4-64/CFP/YFP puncta
were subjected to FRET analysis. The FM4-64 destaining rate
and FRET efficiency between GB1aCFPand GB2YFPsubunits
FCGP / FCnt
slope = 1
CGP effect (% SCnt)
#1 (E = 0.08)
#2 (E = 0.21)
r = -0.95
(GB1aCFP + GB2YFP)
FR (% change by CGP)
r = -0.93r = -0.95
(% change by CGP)
Figure 5. Basal GABABR Activation Negatively Correlates to Release Probability at the Single-Bouton Level
(A) Representative DFFM1-43images before and 10 min after application of 1 mM CGP54626. Stimulation during FM1-43 staining: 30 action potentials at 0.2 Hz.
Destaining: 1000 action potentials at 2 Hz. Fluorescence intensities (arbitrary units) are coded using a pseudocolor transformation. Scale bar: 2 mm.
(B and C) CGP54626 effect at the level of single synapses. Slope of linear fit is 1.35 ± 0.03 (n = 236, B). Dotted line denotes no change (slope = 1). CGP54626-
induced augmentation is inversely correlated to initial presynaptic strength (Spearman r = - 0.9, p < 0.0001, C).
(D) Comparison of the GABABR-mediated presynaptic inhibition under miniature versus evoked (stimulation frequency of 0.2 and 1 Hz) basal release. CGP54626
displayed similar degree of presynaptic enhancement on miniature and evoked basal presynaptic strength (6–9 experiments per group, p > 0.5).
(E) Representative data demonstrating effects of 1 mM CGP54626 on DFFM4-64and FYFP/FCFPratio during 0.2 Hz stimulation at two boutons expressing GB1aCFP
and GB2YFPproteins. CGP54626 significantly changed both FYFP/FCFPand DFFM4-64in the bouton #1, while produced only minor changes on FYFP/FCFPand
DFFM4-64in the bouton #2.
r = –0.95, p < 0.0001, red symbols) or miniature (10 boutons, Spearman r = –0.93, p < 0.0001, gray symbols) synaptic vesicle turnover (DF = (DFCGP/DFCnt– 1) 3
(G) Representative data demonstrating correlation between GB1aCFP/GB2YFPFRET efficiency (E) and FM4-64 destining rate constant during 1 Hz stimulation (k)
in two boutons expressing GB1aCFPand GB2YFP.
(H) Pooled data across 16 boutons reveal negative correlation between E within the GB1aCFP/GB2YFPheterodimer and FM4-64 decay rate constant (Spearman
r = –0.95, p < 0.0001).
Error bars represent SEM. See also Figure S7.
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 259
have been measured during 1 Hz stimulation. The FM destaining
rate constant (k = 1/tdecay, whereas tdecayis an exponential time
course) is proportional to probability of vesicle release. The de-
staining rate constant was plotted against GB1aCFP/GB2YFP
FRET efficiency per bouton. Figure 5G demonstrates two bou-
tons with different FRET levels: the bouton #1 displays smaller
increase in FCFPfollowing YFP photobleaching than the bouton
#2 (left graph), resulting in lower FRET (0.085 versus 0.21).
Notably, the bouton #2 is characterized by higher FRET and
slower detaining rate (k was 0.0024 versus 0.0052 s-1, for #2
and #1, respectively; Figure 5G). The pooled data from 16 bou-
tons demonstrate an inverse correlation between the initial
GB1aCFP/GB2YFPFRET efficiency and the rate of FM destaining
per bouton (Spearman r = –0.95, p < 0.0001; Figure 5H), where
greater FRET correlates with slower release of synaptic vesi-
cles. These results suggest variable levels of basal [GABA]o
GABABR signaling, which is detected by GB1a/GB2conforma-
log [ baclofen] (M)
ΔE (% from Emax)
Inhibition by SKF
Figure 6. Reduction in the Presynaptic GABABR Tone
by Prolong Activity Blockade
(A) Activity blockadeby TTX (48 h) increased miniature presyn-
N = 10–13, p < 0.01).
(B) Activity blockade reduced the effectiveness of CGP54626
to increase miniature presynaptic strength (p < 0.0001).
CGP54626 (1 mM) increased SFM1-43by (1.6 ± 0.09)-fold in
control (left, N = 11, p < 0.0001), but only by (1.06 ± 0.04)-
fold in TTX-treated neurons (right, N = 10, p > 0.05).
(C) Activity blockade reduced the effectiveness of CGP54626
to decrease Emini(p < 0.0001). CGP54626 (1 mM) decreased
Eminifrom 0.15 ± 0.01 to 0.09 ± 0.006 (left, n = 9–16, N = 4,
p < 0.01) with concurrent reduction in CV from 38% to 19%
in control cultures. In TTX-treated cultures, CGP54626
changed neither themean level nor the variability of Emini(right,
n = 13–17, N = 4, p > 0.2).
(D) Dose-response curve of baclofen on Emini between
GB1aCFPand GB2YFPproteins in control (n = 7–10, N = 4,
ED50 = 0.7 ± 0.005 mM) versus TTX-treated (n = 7–13,
N = 4–5, ED50= 0.6 ± 0.004 mM) cultures.
(E) Reduction of basal presynaptic strength by GABA uptake
inhibitor SKF-89976A (25mM)incontrol (N= 5,p< 0.01) versus
TTX-treated (N = 5, p < 0.01) cultures.
(F) Activity blockade did not alter the effectiveness of
SKF-89976A to decrease presynaptic strength in control
versus TTX-treated cultures (N = 5, p > 0.9)
Error bars represent SEM. See also Figure S8.
Inactivity Reduces Presynaptic
Release probability of individual synapses can be
lated activity and by prolonged changes in ongoing
activity. In hippocampal cultures, strong evidence
exists on augmentation of release probabilities
(Branco et al., 2008; Murthy et al., 2001; Thiagara-
jan et al., 2005) with concomitant reduction in
release heterogeneity (Branco et al., 2008) as
a homeostatic response to prolong synaptic
disuse. However, molecular mechanisms that underlie presyn-
aptic disuse hypersensitivity remain largely unknown. If basal
activation of GABABRs regulates vesicle release probability at
a rapid timescale, then changes in the GABABR tone at a slow
timescale might probably underlie homeostatic presynaptic
plasticity. To examine this hypothesis, we estimated the
GABABR tone following pharmacological neuronal silencing by
TTX for 48 hr. Effects of the GABABR antagonist were quantified
on miniature presynaptic strength measured by FM1-43 and on
GB1a/GB2activation detected by FRET. As expected, inactivity
induced an increase in miniature presynaptic strength across
populationof synapses(N=10–13,p<0.01;Figure 6A). Notably,
CGP54626 lost its effectiveness on miniature presynaptic
activity in TTX-treated cultures (N = 10, p > 0.05; Figure 6B; right
graph), comparingto control ones(N=11,p<0.0001; Figure6B;
left graph). Moreover, inactivity induced reduction in GB1aCFP/
GB2YFPFRET efficiency: Eminidecreased from 0.15 to 0.10 and
its variability from 38 to 17% (n = 16–17, p < 0.001; Figure 6C).
Consequently, CGP54626 did not affect GB1aCFP/GB2YFPFRET
GABABR Conformation and Release Probability
260 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
efficiency in TTX-treated cultures (n = 13–17, p > 0.2; Figure 6C,
right graph), while efficiently reduced Eminiin control cultures
(n = 9–16, p < 0.05; Figure 6C, left graph). Two days of TTX treat-
ment did not induce significant changes in the expression levels
of GB1aCFP, GB2YFPor their ratio (Figure S8). Furthermore, the
potency of baclofen to increase GB1aCFP/GB2YFPFRET effi-
ciency was not altered by TTX pretreatment (n = 8–13;
Figure 6D), suggesting no changes in responsiveness of the
GB1a/GB2receptor to agonist. These results suggest decrease
in the ambient [GABA]oas the major determinant of inactivity-in-
duced reduction in the GABABR tone. To test if acceleration of
GABA clearance might be involved in reduction of basal GABA
on inhibition ofpresynaptic
TTX-treated cultures. SKF-89976A equally inhibited presynaptic
strength by ?60% in control and TTX-treated cultures (N = 5,
p < 0.01; Figures 6E and 6F). Therefore, we concluded that inac-
tivity-induced lost of the GABABR tone was highly-likely due to
reduction in the amount of GABA release (see Discussion).
strength was increasedin
Presynaptic GABABR Tone Is Dendrite Specific
Although release probabilities exhibit high heterogeneity along
an axon and within a whole population of hippocampal
synapses, this variability is significantly reduced across
individual dendritic branches (Branco et al., 2008; Murthy
et al., 1997). Notably, release probabilities negatively correlate
to the number of synapses on the branch (Branco et al., 2008),
suggesting that presynaptic mechanisms might be involved in
down-scaling of quantal synaptic weight at heavily innervated
dendrites along dendritic tree (Liu and Tsien, 1995). Therefore,
we asked whether basal activation of presynaptic GABABRs
might be regulated at the level of individual dendritic branches.
Thus, we compared the effect of GABABR antagonist at thick
proximal dendrites versus its effect at thin distal dendritic
mated evoked basal vesicle exocytosis by analyzing the FM1-43
destaining rates separately at proximal dendrites with high
synapse density (Figure 7A) versus distal dendrites with low
synapse density of FM+puncta (Figure 7B). CGP54626 induced
2-fold increase in destaining rates of 67 boutons at highly inner-
vated proximal dendrites (destaining rate constant, k = 1/tdecay,
was (1.7 ± 0.06) 3 10?3and (3.7 ± 0.13) 3 10?3s-1, before and
afterantagonistapplication, respectively; Figure7C).Incontrast,
CGP54626 did not produce significant change in destaining
rates of 71 boutons at distal dendrites (k was (2.4 ± 0.06) 3
10?3and (2.6 ± 0.09) 3 10?3s-1, before and after antagonist
application, respectively; Figure 7D). Notably, FM+boutons at
distal dendrites were characterized by higher destaining rates
than at proximal one (p < 0.001; Figure 7E, white bars), suggest-
ing higher Pr. Both, Pr dynamic range from inhibited to noninhi-
bited state (p < 0.0001; Figure 7E, white versus black bars), and
intrinsic Pr in the absence of GABABR-mediated inhibition
(p < 0.0001; Figure 7E, difference between black bars) were
higher at proximal thick dendrites.
To compare the extent of tonic activation of GABABRs along
dendritic tree under miniature synaptic activity, we measured
CGP54626 effects on both presynaptic strength and GB1aCFP/
GB2YFPFRET ratio at boutons contacting proximal versus distal
dendrites in the presence of TTX. At proximal branches,
CGP54626 increased DFFM1-43(n = 20, p < 0.0001; Figure 7F,
left), while decreasing FRET ratio (n = 8, p < 0.05; Figure 7G,
left). However, at distal branches CGP54626 was effective
neither at the level of vesicle recycling (n = 20, p > 0.5,
Figure 7F, right) nor protein-protein interactions (n = 6, p > 0.6;
Figure 7G, right). Moreover, basal FRET ratio was higher at
proximal synapses (n = 6 – 8, p < 0.01; Figure 7H), suggesting
aptic GABABR tone is differentially regulated along dendritic
tree, predominantly impacting proximal synapses.
Next, we examined whether diffusion distance between
excitatory and inhibitory synapses is different at proximal thick
versus distal thin dendritic branches. As FM dye staining cannot
distinguish excitatory from inhibitory synapses, we used both
excitatory and inhibitory synapse-specific marker proteins to
distinguish between these two types of synapses (Figure S9A).
Excitatory synapses were labeled with the vesicular glutamate
transporter VGLUT1 antibody, whereas inhibitory synapses
were labeled with GABA synthetic enzyme GAD65 antibody.
We found that the total number of synapses along a fixed length
of dendrite positively correlated to the diameter of the dendritic
branch, with higher synapse number at thick proximal dendrites
(Spearman r = 0.75, p < 0.0001; Figure S9B), as has been
reported in earlier study (Liu, 2004). Interestingly, the ratio of
to dendritic diameter, reaching constant ?0.8 ratio at dendrites
with diameter >4 mm (Figure S9C). Altogether, these results
suggest that diffusion distance for GABA between inhibitory
and excitatory synapses is shorter at thick proximal dendrites.
Finally, we explored location dependency of basal GABABR-
mediated presynaptic inhibition in a more intact preparation,
namely in acute hippocampal slices. We compared the presyn-
aptic GABABR activation at the most proximal dendrites of CA1
pyramidal neurons in stratum radiatum versus the most distal
apical dendrites in stratum lacunosum-moleculare (Figure 8A).
Electron microscopy study suggests higher NI/NEratio at prox-
imal dendrites of pyramidal CA1 hippocampal neurons in rat
prevent direct intracellular recordings, we relied on electrophys-
iological measurements of extracellular field EPSPs (fEPSPs).
lular levels of transmitters, special care was taken to maintain
physiological inhibition-excitation balance and, consequently,
underthe followingconditions: (1)inintact slices withoutsurgical
isolation of CA3 region; (2) without blocking of GABAAreceptors;
(3) under physiological concentration of [Ca2+]o and [Mg2+]o
(1.2 mM for both ions) for storage and recording extracellular
solutions in order to maintain Pr.
First, we assessed the strength of basal synaptic transmission
in proximal CA3-CA1 synapses pathway by measuring fEPSPs
evoked by low frequency stimulation (0.1 Hz). To test whether
presynaptic GABABRs are activated under these conditions, we
tested the effect of CGP54626 antagonist on the slope of input
(amplitude of fiber volley)/output (slope of fEPSP) curve.
CGP54626 increased the slope of input/output curve by 40%
(from 0.6 ± 0.06 to 1.0 ± 0.05, n = 5, p < 0.01; Figure 8B).
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 261
changes, we estimated CGP54626 effect on paired-pulse facili-
tation. At proximal synapses, CGP54626 reduced paired-pulse
facilitation by 20% (from 211% ± 11% to 169% ± 9%, n = 7,
p < 0.01; Figures 8C and 8D). Next, we assessed the strength
of basal synaptic transmission at the most distal synapses,
receiving direct input from perforant pathway of entorhinal
affected neither basal synaptic transmission (1.02 ± 0.1 versus
1.06 ± 0.07 slope for control and CGP54626, respectively, n =
9, p > 0.8; Figure 8E), nor synaptic facilitation (n = 9, p > 0.7;
was higher (0.6 ± 0.06 versus 1.02 ± 0.1), while paired-pulse
facilitation was lower (142% ± 14% versus 221% ± 12%, n = 9,
ities. These results suggest differential basal GABABR activation
along apical dendrites of CA1 pyramidal neurons under evoked
conditions, with predominant inhibition of glutamate release at
GABABR Conformational Diversity
Presynaptic GABABreceptors mediate GABA-dependent inhibi-
tion of glutamate release, impacting plasticity of hippocampal
synapses and hippocampus-dependent memory function (Vigot
et al., 2006). Whether presynaptic GABABRs are activated under
basal synaptic activity and impact the probability of neurotrans-
mitter release has been unclear. To explore GABABR activation
at individual hippocampal boutons, we took advantage of the
Figure 7. Differential Presynaptic GABABR Tone at Proximal versus Distal Dendrites of Pyramidal Neurons in Hippocampal Culture
(A and B) FM1-43 images of maximal stimulation (600 action potentials at 10 Hz) at proximal (A) and distal (B) dendritic branch. Fluorescence intensities (arbitrary
units) are coded using a pseudocolor transformation and are merged with DIC image. Scale bar: 10 mm.
(C and D) Effect of CGP54626 (1 mM) on the destaining rate constant (k) during 1 Hz stimulation at proximal high-density (C) and distal low-density (D) dendrites.
CGP54626 increased k from (1.7 ± 0.06) 3 10?3to (3.7 ± 0.13) 3 10?3s?1at 67 synapses contacting proximal dendrites, but did not affect it at 71 synapses
contacting distal dendrites (k were (2.4 ± 0.06) 3 10?3and (2.6 ± 0.09) 3 10?3s?1in control and after CGP54626 application, respectively).
(E) On average (n = 4), CGP54626 mainly increased FM1-43 destaining rate in boutons at proximal dendrites (p < 0.0001) that display lower basal destaining rate
than distal ones (p < 0.001).
(F) Effect of CGP54626 (1 mM) on DFFM1-43at single boutons under miniature synaptic activity (+TTX) at proximal (left, n = 20, p < 0.0001) and distal (right, n = 20,
p > 0.5) dendrites.
(G) Effect of CGP54626 (1 mM) on FYFP/FCFP(FR) at boutons expressing GB1aCFPand GB2YFPunder miniature synaptic activity (+TTX) at proximal (left, n = 8,
p < 0.05) and distal (right, n = 6, p > 0.6) dendrites.
(H) On average, FR between GB1aCFPand GB2YFPwas higher at boutons contacting proximal dendrites than the distal ones (n = 6–8, p < 0.01) under miniature
Error bars represent SEM. See also Figure S9.
GABABR Conformation and Release Probability
262 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
selective targeting of GB1ato presynaptic boutons (Vigot et al.,
a FRET-based detector of intermolecular associations between
GB1aand GB2receptor subunits, and FM dye-based tracking of
synaptic vesicle release. We discovered an inter-synapse vari-
ability of GB1a/GB2activity that correlates with variations in basal
levels of GABA. The basal activation of GABABRs was dendrite
specific and homeostatically regulated by ongoing neuronal
activity. These findings demonstrate a critical link between
GABABR heterodimer conformational dynamics and local regula-
tion of release probability at hippocampal synapses.
A prominent feature of the FRET signal detected between
GB1aand GB2subunits is the high intersynapse heterogeneity
that occurs under basal synaptic activity. Two potential sources
of FRET variability may be considered: a biological one due to
and/or in the number of presynaptic GABABRs or a methodolog-
ical one due to instrumentation variability or nonspecific associ-
ations. The latter seems unlikely since negative controls did not
reveal specific FRET. On the other hand, several experiments
support the explanation of variations in basal [GABA]oas a major
determinant of FRET variability. The variability in FRET efficiency
decreased significantly with inhibiting SNARE-mediated syn-
aptic vesicle exocytosis and interfering with binding of ambient
GABA to the receptor with a selective GABABR antagonist.
Furthermore, because [GABA]o depends on the balance of
GABA release, up-take and diffusion, we examined the influence
of these parameters on heterogeneity of FRET signals. Increas-
ing [GABA]o by inhibiting GABA up-take resulted in ?2-fold
mean Emini(Figure 2D), while addition of a saturating concentra-
tion of baclofen reduced variability by ?5-fold and increased the
mean Emini (Figure 2C). Moreover, stronger basal GABABR
activation was detected at proximal thick dendrites (Figures 7
and8).Given thehighersynapse densityandlarger ratioof inhib-
itory to excitatory synapses at proximal versus distal dendrites
(Figure S9), diffusion distance from inhibitory to excitatory bou-
tons might be a significant determinant of local [GABA]oin our
Fiber volley (μV)
fEPSP slope (μV/ms)
Fiber volley (μV)
fEPSP slope (μV/ms)
Figure 8. Differential Basal GABABR Activation at Proximal versus Distal CA1 Synapses in Hippocampal Slices
(A) Experimental set-up and CA1 organization. For recordings at proximal apical CA1 dendrites in stratum radiatum (SR), the stimulating electrode at Schaffer
collateral (SC) input was placed near cell bodies ?200 mm from the recording electrode. For recording at the most distal dendrites in stratum lacunosum molec-
ulare (SLM), the stimulating electrode at perforant path (PP) was placed ?200 mm from the recording electrode.
(Band E) Effect of 10 mMCGP54626 on input-output relationship between the amplitudeof fiber volleyand the slope of fEPSPfor gradually increasing stimulation
intensities at proximal (B, n = 5, p < 0.01) versus distal (E, n = 9, p > 0.8) synapses. Inserts: representative recordings of fEPSPs before (black) and 10 min after
(gray) application of CGP54626.
(C and F) Representative recordings of paired-pulse facilitation of fEPSPs (2 action potentials; interspike interval, 20 ms; interpair interval,30 s)before (black) and
10 min after (gray) application of CGP54626 at proximal (C) versus distal (F) synapses.
(D and G) Paired-pulse facilitation (PPF) of fEPSPs, normalized to the first peak at proximal (D) and distal (G) synapses. CGP54626 reduced PPF at proximal
synapses (n = 7, p < 0.01), but did not affect it at distal ones (n = 9, p > 0.7). Note reduction in the initial PPF at distal synapses (p < 0.001).
Scale bars: (B, C, E, and F) 0.1 mV, 10 ms. Error bars represent SEM.
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 263
a fraction of GABAergic neurons (Figure 3) triggers negative
feedback loop, inhibiting GABA release and, therefore, contrib-
uting to spatial differences in [GABA]o(Figure 9B). Collectively,
these data suggest that differences in the local [GABA]onear
synaptic contacts, but not in the assembly of GB1a/GB2receptor
heterodimers, is the major determinant of the FRET variability.
Differences in the expression levels of endogenous GB1a/GB2
receptors at individual synapses may provide an additional
element of variability in vivo.
How does the synapse-specific GB1aCFP/GB2YFPFRET effi-
ciency relate to receptor activation? The FRET efficiency is
determined by the proximity and orientation of the CFP and
YFP fluorophores. Notably, the mean Eminiin the presence of
GABABR antagonist is low but significantly higher than back-
ground. This result is consistent with the known formation of
an obligate heterodimer (Jones et al., 1998; Kaupmann et al.,
1998; White et al., 1998). The increase in mean Eminiwith agonist
suggests the GABABR undergoes a conformational change,
bringing the two fluorophores closer (or into a more favorable
orientation). Remarkably, the EC50for FRET superimposes with
the IC50for baclofen-dependent inhibition of presynaptic Ca2+
transients and of vesicle release. Previous spectroscopic FRET
studies with CFP/YFP tagged G-proteins also found overlap in
the EC50for G protein activation and FRET. The variable Emini
that is measured under basal synaptic activity likely reflects
different activation states of the GABABheterodimer. Nikolaev
et al. (2006) examined the change in FRET with tagged a2A
adrenergic receptor and different agonists and found the rate
of receptor activation correlated with the ability of agonists to
stabilize different receptor conformational changes (Nikolaev
et al., 2006). Similarly, local [GABA]omight affect the degree of
FRET between closely associated molecular pairs. In this case,
activation of the GB1a/GB2heterodimer occurs through series
of conformational intermediates, suggesting general conforma-
tional flexibility similarly to adrenoreceptors (Kobilka and Deupi,
2007). Irrespective of the exact activation mechanism, our data
suggest a partial GABABR activation by ambient GABA during
basal synaptic activity. Thus, C terminally tagged GB1a/GB2en-
gineered FRET pair constitutes highly sensitive sensor for
receptor activation by endogenous [GABA]o.
Functional Diversity of GABABR Tone
be a universal property of synaptic transmission in a wide range
of species (Atwood and Karunanithi, 2002). In hippocampal
cultures and slices, highly variable Pr have been recorded
utilizing electrophysiological and optical tools (Branco et al.,
2008; Dobrunz and Stevens, 1997; Murthy et al., 1997;
Rosenmund et al., 1993; Slutsky et al., 2004). The origin of this
variability, although central to our understanding of synaptic
transmission and plasticity, is not fully understood. While various
morphological correlates of Pr have been proposed (Schikorski
and Stevens, 1997), molecular mechanisms underlying func-
tional diversity of presynaptic boutons remain obscure.
Our study suggests that the level of basal activation of presyn-
aptic GABABRs regulates Pr of individual hippocampal boutons,
regulatory mechanism of intersynapse Pr variability was
explored previously, but a uniform reduction in vesicle exocy-
tosis by baclofen was observed across individual boutons
(Isaacson and Hille, 1997). While our results confirm a homoge-
(Figure 4E), we observed a remarkable variability in the positive
effect of GABABR antagonist on basal presynaptic activity.
An inverse correlation between basal presynaptic strength and
the GB1a/GB2conformational state at the level of individual bou-
tons (Figure 5H) suggests that the probability of glutamate
Figure 9. Illustration of the GABABR Activation at Presynaptic Boutons as Function of [GABA]o
are closely associated (FRET efficiency, solid black line), while intersynapse variability of these associations (dashed black line) is low. Increase in [GABA]ounder
quantal transmitter release (light blue background), enhances GB1a/GB2associations and increases their intersynapse variability, promoting GB1aR-mediated
inhibition of vesicle release (solid red line).
(B) Dendrite-specific GABABR-mediated tonic inhibition of basal vesicle release. At highly innervated proximal dendrites, GABA activates GB1a/GB2heterore-
ceptors, resulting in inhibition of glutamate release and reduction in Pr (upper panel). A low fraction of inhibitory synapses expressing GB1a/GB2autoreceptors
triggerautoinhibition ofGABArelease, resulting inreduction ofbasal[GABA]oatneighboring excitatory synapses.Atdistaldendriticbrancheswithlowersynapse
density, [GABA]onear glutamatergic boutons is insufficient to activate GB1a/GB2receptors, leaving intrinsic Pr unaltered and, therefore, higher than at proximal
branches. Excitatory synapses (black), Inhibitory synapses (blue), GABA (blue circles), GB1a/GB2receptors (turquoise rectangle), GABABR-mediated autoinhi-
bition of GABA release (blue arrows), GABABR-mediated inhibition of glutamate release (black arrows).
GABABR Conformation and Release Probability
264 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
releasecan beregulated throughsynapse-specific fine-tuning of
the GABABR activation. Based on these data, following scenario
of endogenous receptor activation can be proposed (Figure 9A).
During periods of synaptic inactivity which can occur before
synapse maturation or as a consequence of experience-depen-
dent synaptic depression at GABAergic boutons, local [GABA]o
approximates its lowest boundary. As a result, GB1a/GB2heter-
odimers at neighboring glutamatergic boutons switch to inactive
cascade and, consequently, leaving intrinsic probability of gluta-
mate release unaltered. Increase in ambient [GABA]oduring
basal activity strengthens GB1a/GB2associations by variable
tion of glutamate release. Synchronous activation of multiple
inhibitory boutons should increase local [GABA]o and might
further strengthen GB1a/GB2 associations with concurrent
decrease in glutamate release.
Homeostatic Regulation of GABABR Tone
Homeostatic regulation of synaptic strength represents a basic
mechanism of neuronal adaptation to chronic changes in
ongoing activity levels. There is a wide repertoire of synaptic
modifications at both sides of the synaptic cleft that enable
neuronal homeostasis. At the presynaptic side of hippocampal
synapses, strong evidence exists on augmentation of release
probabilities as a homeostatic response to prolong synaptic
disuse (Branco et al., 2008; Murthy et al., 2001; Thiagarajan
et al., 2005). Functionally, activity-dependent presynaptic
homeostasis, in contrast to the postsynaptic one, may
profoundly impact short-term synaptic dynamics by altering
the filtering properties of synapses. However, molecular mecha-
unknown. Our data propose reduction in the extent and vari-
ability of the GABABR tone as a homeostatic mechanism under-
lying inactivity-induced increase in release probabilities and
reduction in release heterogeneity (Branco et al., 2008). As
activity blockade did not trigger changes in the GB1a/GB2
receptor assembly, expression level and responsiveness to
agonist, we conclude that reduction in the ambient [GABA]o
might cause the lost of tonic GABABR activation. Indeed, inac-
tivity has been demonstrated to induce reduction in the amount
of GABA released by individual vesicles (De Gois et al., 2005;
Hartman et al., 2006) that should reduce the ambient [GABA]o.
Alternatively, prolong synaptic disuse might trigger upregulation
presynaptic responsiveness to GABA transporter blocker
following activity blockade (Figures 6E and 6F) makes this possi-
bility less likely. Given a predominant localization of GB1a/GB2
receptors at excitatory boutons (Vigot et al., 2006), homeostatic
regulation of the GABABR tone may preferentially adjust release
probabilities of glutamatergic vesicles. Interestingly, inhibitory
neurons expressing CB1Rs have been recently shown to
increase their release probabilities in response to chronic inac-
tivity through reduction in endocannabinoid tone (Kim and Alger,
2010). Future studies are warranted to examine whether tuning
basal activation of high-affinity presynaptic GPCRs represents
a general mechanism of presynaptic homeostasis in other
Dendrite-Specific Scaling of the GABABR Tone
While high heterogeneity of release probability and Ca2+flux
dynamics have been observed at boutons along axons (Koester
and Sakmann, 2000; Murthy et al., 1997), release properties of
neighboring synapses tend to be correlated at spatially
thy et al., 1997). Postsynaptic target specificity of transmitter
signaling (Davis et al., 1998), and local postsynaptic depolariza-
tion levels (Branco et al., 2008) might contribute to presynaptic
differentiation. Our results demonstrate that GABABR-mediated
presynapticinhibition is dendrite
pronounced at proximal than at distal dendritic branches of
CA1 hippocampal neurons in hippocampal cultures and acute
slices. Our single-synapse analysis of presynaptic activity in
cultures suggests different synaptic configuration of proximal
versus the distal dendrites (Figure 9B): proximal branches
display higher ratio of inhibitory-to-excitatory synapses, higher
total synapse density and lower Pr due to higher extent of
presynaptic GABABR tone. In slices, the most proximal SC
synapses along apical dendrites of CA1 pyramidal cells, with
higher inhibitory-to-excitatory synapse ratio (Megı ´as et al.,
2001), displayed higher paired-pulse facilitation, lower input-
output slope, and higher GABABR activation under low-
frequency stimulation than the most distal perforant synapses.
At first glance, our results seem to differ from the data by Ahmed
and Siegelbaum (2009), suggesting higher Pr and lower paired-
pulse ratio at Schaffer collateral than at perforant CA1 synapses.
This discrepancy can be attributed to differences in the experi-
mental design. Ahmed and Siegelbaum performed recordings
from medial Shaffer collateral synapses while our study focused
on proximal synapses. In addition, their ACSF included GABABR
and GABAAR blockers, and [Ca2+]otwice as high as in our study,
making the direct comparison problematic.
Local fine-tuning of presynaptic GABABR tone may be
involved in synaptic scaling along dendritic compartment,
tive correlation between the number of functional boutons and
unitary synaptic strength (Liu and Tsien, 1995) or its presynaptic
determinant, Pr (Branco et al., 2008). Moreover, differential
presynaptic GABABR tone along dendritic tree might enable
higher dynamic range of synapses at highly innervated dendrites
during temporospatially correlated synaptic activity, local inhibi-
tion-excitation balance (Liu, 2004), and normalization of synaptic
inputs (Magee and Cook, 2000) in hippocampal networks.
It remains to be seen whether tuning of GB1a/GB2heterodimer
conformation emerges as a possible molecular basis for feed-
back regulation of transmitter release and its plasticity by local
dendritic activity in vivo.
Hippocampal Cell Culture
Primary cultures of CA3-CA1 hippocampal neurons were prepared from
newborn Wistar rats on postnatal days 0–2, as described (Slutsky et al.,
2004). The experiments were performed in 12–18 DIV cultures (except those
performed in 4–5 DIV young neurons as mentioned in the text). All animal
experiments were approved by the Tel Aviv University Committee on Animal
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 265
Hippocampal neurons were imaged using a Zeiss LSM510 META and FV1000
spectral confocal microscopes. The experiments were conducted at room
temperature in extracellular Tyrode solution that contained (in mM): NaCl,
145; KCl, 3; glucose, 15; HEPES, 10; MgCl2, 1.2; CaCl2, 1.2 (pH adjusted to
7.4 with NaOH). To isolate miniature synaptic activity, TTX (1 mM) has been
added to extracellular solution. FRET imaging was carried out as above.
For spectral analysis, CFP was excited at 405 nm (Zeiss) or at 442 nm
(Olympus) and fluorescence emission was measured between 400 and
700 nm, with a 10 nm l step size. In order to reduce phototoxicity and photo-
bleaching, most of the FRET experiments were performed using a narrowed
emission spectrum (470–550 nm, 20 nm l step size) composed of CFP peak
(486 ± 10 nm) and a YFP peak (534 ± 10 nm) containing YFP emission due
to FRET, direct YFP excitation at 405 nm, and CFP emission tail. YFP was
imaged and specifically photobleached at 514 nm (excitation, 100% laser
power, 30 ms duration) and 534 ± 5 nm (emission). Notably, photobleaching
at 514 nm was specific and did not significantly affect YFP signal at neigh-
boring boutons (Figure S1B). YFP fluorescence recovery after photobleaching
was ?18% and developed within ?40 s following photobleaching (Figures
S1C and S1D). Thus, images of CFP were captured within ?30 s from photo-
bleaching. We verified that this procedure did not affect functionality of
presynaptic boutons (Figure S1E).
The emission spectral properties of FM4-64 allow simultaneous imaging
with CFP and YFP fluorophores (Figure S1F). Images were 512 3 512 pixels,
with a pixel width of 92–110 nm. Z stacks were collected from 3–4 mm optical
slice, at 0.6–0.8 mm steps.
Calculation of FRET Efficiency
emission (486 ± 10 nm) before and after the acceptor photobleaching. FRET
efficiency, E, was then calculated using the equation E = 1 ? IDA/ID, where
IDAis the peak of donor emission in the presence of the acceptor and IDis
the peak after acceptor photobleaching (Riven et al., 2003). Detection of
CFP/YFP signals has been done using custom-written scripts in MATLAB
(see Supplemental Experimental Procedures).
Activity-dependent FM1-43 and FM4-64 styryl dye have been used to detect
functional boutons. FM loading and unloading were done using protocols
described previously (Abramovet al., 2009,and in Supplemental Experimental
Procedures). The fluorescence of individual synapses was determined from
the difference between images obtained after staining and after destining
(DF). To stain all the functional boutons capable of vesicle recycling, we
used maximal stimulation protocol: 600 action potentials at a rate of 10 Hz
or high KCl solution (50 mM, 1 min). To estimate vesicle recycling/release
during low-frequency stimulation, we quantified (1) DF signal for staining by
1 Hz stimulation following staining of boutons by maximal stimulation. For
FM+puncta detection, DF images have been analyzed (only the puncta exhib-
iting R90% destining were subjected to analysis). For DF analysis, see
Supplemental Experimental Procedures.
Detecting Presynaptic Calcium Transients
Fluorescent calcium indicator Calcium Green 488 BAPTA-1 AM was dissolved
in DMSO to yield a concentration of 1 mM. For cell loading, cultures were
incubated at 37?C for 30 min with 3 mM of this solution diluted in a standard
extracellular solution. Imaging was performed using FV300 ad FV1000
Olympus confocal microscopes, under 488 nm (excitation) and 510–570 nm
(emission), using 500 Hz line scanning.
Electrophysiology in Acute Hippocampal Slices
Coronal hippocampal slices (400 mm) from 5- to 7-week-old Wistar rats were
prepared using standard procedures as described (Abramov et al., 2009;
see Supplemental Experimental Procedures). The ACSF contained, in mM:
NaCl, 125; KCl, 2.5; CaCl2, 1.2; MgCl2, 1.2; NaHCO3, 25; NaH2PO4, 1.25;
glucose, 25. Extracellular fEPSPs were recorded at room temperature with
a glass pipette containing ACSF (1–2 MU) from proximal synapses in the
CA1 stratum radiatum or distal synapses in stratum lacunosum moleculare
using a MultiClamp700A amplifier (Molecular Devices). Stimulation of the SC
or PP pathway was delivered through a glass suction electrode (10–20 mm
tip) filled with ACSF. fEPSPs were analyzed by pClamp10 software (Molecular
All error bars represent standard error. For experiments in cultures, n is desig-
nated for the number of boutons and N for the number of imaged neurons
(or number of experiments for population FM analysis). All the experiments
were repeated at least in 3 different batches of cultures. For slice experiments
(Figure 8), n marks the number of slices. One-way ANOWA Kruskal-Wallis
nonparametrictestandStudent’s ttestswereused.Nonparametric Spearman
test was used for correlation analysis.
Experimental Procedures and can be found with this article online at doi:10.
Informationincludes nine figuresand Supplemental
We thank Dr. Bernhard Bettler for critical comments on the manuscript;
Dr. Nathan Dascal and Dr. Bernard Attali for discussions; Dr. Eitan Reuveny
for providing TNFR2YFP, Gb1YFPand Gg2CFPcDNAs; Dr. Daniel Gitler for
Syn1aCFPand Hilla Fogel for the help with image processing algorithms. This
work was supported by Binational Science Foundation (I.S. and P.A.S., grant
No. 2007199) and Israel Science Foundation (I.S., grants No. 993/08 and
Accepted: June 14, 2010
Published: July 28, 2010
Abramov, E., Dolev, I., Fogel, H., Ciccotosto, G.D., Ruff, E., and Slutsky, I.
(2009). Amyloid-beta as a positive endogenous regulator of release probability
at hippocampal synapses. Nat. Neurosci. 12, 1567–1576.
Ahmed, M.S., and Siegelbaum, S.A. (2009). Recruitment of N-Type Ca(2+)
channels during LTP enhances low release efficacy of hippocampal CA1 per-
forant path synapses. Neuron 63, 372–385.
Atwood, H.L., and Karunanithi, S. (2002). Diversification of synaptic strength:
presynaptic elements. Nat. Rev. Neurosci. 3, 497–516.
Bettler, B., Kaupmann, K., Mosbacher, J., and Gassmann, M. (2004). Molec-
ular structure and physiological functions of GABA(B) receptors. Physiol.
Rev. 84, 835–867.
Branco, T., Staras, K., Darcy,K.J., and Goda, Y. (2008).Local dendritic activity
sets release probability at hippocampal synapses. Neuron 59, 475–485.
Davis, G.W., DiAntonio, A., Petersen, S.A., and Goodman, C.S. (1998). Post-
synaptic PKA controls quantal size and reveals a retrograde signal that regu-
lates presynaptic transmitter release in Drosophila. Neuron 20, 305–315.
De Gois, S., Scha ¨fer, M.K.H., Defamie, N., Chen, C., Ricci, A., Weihe, E.,
Varoqui, H., and Erickson, J.D. (2005). Homeostatic scaling of vesicular
glutamate and GABA transporter expression in rat neocortical circuits. J. Neu-
rosci. 25, 7121–7133.
Dobrunz, L.E., and Stevens, C.F. (1997). Heterogeneity of release probability,
facilitation, and depletion at central synapses. Neuron 18, 995–1008.
Fowler, C.E., Aryal, P., Suen, K.F., and Slesinger, P.A. (2007). Evidence for
association of GABA(B) receptors with Kir3 channels and regulators of G
protein signalling (RGS4) proteins. J. Physiol. 580, 51–65.
Hartman, K.N., Pal, S.K., Burrone, J., and Murthy, V.N. (2006). Activity-depen-
dent regulation of inhibitory synaptic transmission in hippocampal neurons.
Nat. Neurosci. 9, 642–649.
GABABR Conformation and Release Probability
266 Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc.
Isaacson, J.S., and Hille, B. (1997). GABA(B)-mediated presynaptic inhibition Download full-text
of excitatory transmission and synaptic vesicle dynamics in cultured hippo-
campal neurons. Neuron 18, 143–152.
Isaacson, J.S., Solı ´s, J.M., and Nicoll, R.A. (1993). Local and diffuse synaptic
actions of GABA in the hippocampus. Neuron 10, 165–175.
Jensen, K., Chiu, C.-S., Sokolova, I., Lester, H.A., and Mody, I. (2003). GABA
transporter-1 (GAT1)-deficient mice: differential tonic activation of GABAA
versus GABAB receptors in the hippocampus. J. Neurophysiol. 90,
W.J., Johnson, M., Gunwaldsen, C., Huang, L.Y., et al. (1998). GABA(B)recep-
torsfunctionasaheteromericassembly ofthesubunitsGABA(B)R1and GABA
(B)R2. Nature 396, 674–679.
Kaupmann, K., Malitschek, B., Schuler, V., Heid, J., Froestl, W., Beck, P.,
Mosbacher, J., Bischoff, S., Kulik, A., Shigemoto, R., et al. (1998). GABA(B)-
receptor subtypes assemble into functional heteromeric complexes. Nature
static mechanism for specific inhibitory synapses. Nat. Neurosci. 13, 592–600.
Kobilka, B.K., and Deupi, X. (2007). Conformational complexity of G-protein-
coupled receptors. Trends Pharmacol. Sci. 28, 397–406.
Koester, H.J., and Sakmann, B. (2000). Calcium dynamics associated with
action potentials in single nerve terminals of pyramidal cells in layer 2/3 of
the young rat neocortex. J. Physiol. 529, 625–646.
Liu, G. (2004). Local structural balance and functional interaction of excitatory
and inhibitory synapses in hippocampal dendrites. Nat. Neurosci. 7, 373–379.
Liu, G., and Tsien, R.W. (1995). Properties of synaptic transmission at single
hippocampal synaptic boutons. Nature 375, 404–408.
Magee, J.C., and Cook, E.P. (2000). Somatic EPSP amplitude is independent
of synapse location in hippocampal pyramidal neurons. Nat. Neurosci. 3,
Malitschek, B., Ru ¨egg, D., Heid, J., Kaupmann, K., Bittiger, H., Fro ¨stl, W.,
Bettler, B., and Kuhn, R. (1998). Developmental changes of agonist affinity
at GABABR1 receptor variants in rat brain. Mol. Cell. Neurosci. 12, 56–64.
Matsushita, S., Nakata, H., Kubo, Y., and Tateyama, M. (2010). Ligand-
induced rearrangements of the GABA(B) receptor revealed by fluorescence
resonance energy transfer. J. Biol. Chem. 285, 10291–10299.
Megı ´as, M., Emri, Z., Freund, T.F., and Gulya ´s, A.I. (2001). Total number and
distribution of inhibitory and excitatory synapses on hippocampal CA1 pyra-
midal cells. Neuroscience 102, 527–540.
Murthy, V.N., Sejnowski, T.J., and Stevens, C.F. (1997). Heterogeneous
release properties of visualized individual hippocampal synapses. Neuron
Murthy, V.N., Schikorski, T., Stevens, C.F., and Zhu, Y. (2001). Inactivity
produces increases in neurotransmitter release and synapse size. Neuron
Nikolaev, V.O., Hoffmann, C., Bu ¨nemann, M., Lohse, M.J., and Vilardaga, J.P.
(2006). Molecular basis of partial agonism at the neurotransmitter alpha2A-
adrenergic receptor and Gi-protein heterotrimer. J. Biol. Chem. 281,
Riven, I., Kalmanzon, E., Segev, L., and Reuveny, E. (2003). Conformational
rearrangements associated with the gating of the G protein-coupled potas-
sium channel revealed by FRET microscopy. Neuron 38, 225–235.
Rosenmund, C., Clements, J.D., and Westbrook, G.L. (1993). Nonuniform
probability of glutamate release at a hippocampal synapse. Science 262,
Scanziani, M. (2000). GABA spillover activates postsynaptic GABA(B) recep-
tors to control rhythmic hippocampal activity. Neuron 25, 673–681.
Schikorski, T., and Stevens, C.F. (1997). Quantitative ultrastructural analysisof
hippocampal excitatory synapses. J. Neurosci. 17, 5858–5867.
Slutsky, I., Sadeghpour, S., Li, B., and Liu, G. (2004). Enhancement of synaptic
plasticity through chronically reduced Ca2+ flux during uncorrelated activity.
Neuron 44, 835–849.
Thiagarajan, T.C., Lindskog, M., and Tsien, R.W. (2005). Adaptation to
synaptic inactivity in hippocampal neurons. Neuron 47, 725–737.
Vigot, R., Barbieri, S., Bra ¨uner-Osborne, H., Turecek, R., Shigemoto, R.,
Zhang, Y.P., Luja ´n, R., Jacobson, L.H., Biermann, B., Fritschy, J.M., et al.
(2006). Differential compartmentalization and distinct functions of GABAB
receptor variants. Neuron 50, 589–601.
White, J.H., Wise, A., Main, M.J., Green, A., Fraser, N.J., Disney, G.H., Barnes,
A.A., Emson, P., Foord, S.M., and Marshall, F.H. (1998). Heterodimerization is
required for the formation of a functional GABA(B) receptor. Nature 396,
Wu, L.G., and Saggau, P. (1995). GABAB receptor-mediated presynaptic inhi-
bition in guinea-pig hippocampus is caused by reduction of presynaptic Ca2+
influx. J. Physiol. 485, 649–657.
Wu, L.G., and Saggau, P. (1997). Presynaptic inhibition of elicited neurotrans-
mitter release. Trends Neurosci. 20, 204–212.
GABABR Conformation and Release Probability
Neuron 67, 253–267, July 29, 2010 ª2010 Elsevier Inc. 267