Cell, Vol. 105, 521���532, May 18, 2001, Copyright ���2001 by Cell Press GABA Itself Promotes the Developmental Switch of Neuronal GABAergic Responses from Excitation to Inhibition voltage at which GABAergic currents change their direc- tion). Recent work has shown that the developmental transformation of GABAergic synaptic transmission from depolarizing to hyperpolarizing is due to a shift in EGABA toward a more hyperpolarized potential, which is Karunesh Ganguly,1,2,4 Alejandro F. Schinder,2,4 Scott T. Wong,2,5 and Mu-ming Poo1,2,3,5 1 Program in Neuroscience 2 Section of Neurobiology, Division of Biology University of California, San Diego likely the result of an ontogenetic decrease in the intra- La Jolla, California 92093 cellular Cl2 concentration ([Cl2]i Cherubini et al., 1990 Luhmann and Prince, 1991 Chen et al., 1996 Owens et al., 1996). Indeed, changes in the mRNA level for the Summary K1-coupled Cl2 transporter KCC2 have been shown to correlate with the modification of GABAergic transmis- GABA is the main inhibitory neurotransmitter in the sion (Lu et al., 1999 Rivera et al., 1999 Vu et al., 2000). adult brain. Early in development, however, GABAergic KCC2 increases the rate of Cl2 extrusion, thus leading synaptic transmission is excitatory and can exert to a reduction in [Cl2]i and a consequent shift in EGABA widespread trophic effects. During the postnatal pe- toward more hyperpolarized potentials (Jarolimek et al., riod, GABAergic responses undergo a switch from be- 1999 Kakazu et al., 1999 Rivera et al., 1999). ing excitatory to inhibitory. Here, we show that the This conversion of GABAergic transmission from de- switch is delayed by chronic blockade of GABAA recep- polarizing to hyperpolarizing is also accompanied by a tors, and accelerated by increased GABAA receptor change in GABA-mediated biochemical signaling. Only activation. In contrast, blockade of glutamatergic during this early developmental period, depolarizing transmission or action potentials has no effect. Fur- GABAergic potentials activate voltage-dependent Ca21 thermore, GABAergic activity modulated the mRNA channels (VDCCs) and elevate [Ca21]i (Connor et al., levels of KCC2, a K1-Cl2 cotransporter whose expres- 1987 Yuste and Katz, 1991 Wang et al., 1994). Such sion correlates with the switch. Finally, we report that GABA-induced elevation of [Ca21]i is likely to play a GABA can alter the properties of depolarization- critical role in the maturation of the nervous system. induced Ca21 influx. Thus, GABA acts as a self-limiting For instance, GABA-mediated increases in [Ca21]i can trophic factor during neural development. induce BDNF expression (Berninger et al., 1995) and promote neuronal survival and differentiation (LoTurco In the adult central nervous system, g-amino-butyric et al., 1995 Marty et al., 1996 Ikeda et al., 1997). GABA- acid (GABA) is the primary inhibitory neurotransmitter. induced elevation of [Ca21]i may also be required to It regulates a neuron���s ability to fire action potentials form, stabilize, and strengthen synaptic connections either through hyperpolarization of the membrane po- (Kirsch and Betz, 1998 Caillard et al., 1999 Kneussel tential or through shunting of excitatory inputs. During and Betz, 2000). early postnatal development, however, GABAergic syn- While the developmental transformation of GABAer- aptic transmission is excitatory, able to elevate the intra- gic transmission is well documented, little is known cellular Ca21 concentration ([Ca21]i), and even capable about signals that induce this transformation. Since neu- of triggering action potentials (Mueller et al., 1984 Luh- ronal activity is known to increase during development, mann and Prince, 1991 Yuste and Katz, 1991 Reichling we examined in the present study whether synaptic ac- et al., 1994 Wang et al., 1994 Leinekugel et al., 1995 tivity can regulate the switch of GABAergic transmis- Obrietan and van den Pol, 1995 Chen et al., 1996 Owens sion. We found that the change in GABA signaling was et al., 1996 Khazipov et al., 1997). Over a limited postna- largely prevented by chronic blockade of GABAA recep- tal period, in the hippocampus, neocortex, and hypo- tors, and was accelerated by increased GABA receptor thalamus, as well as other regions of the brain, there is activation. Changes in the level of KCC2 mRNA tightly a switch of the electrophysiological (depolarization to correlated with the observed changes in GABA signal- hyperpolarization) and biochemical (Ca21-mediated sig- ing. In addition, we found that spontaneous GABAergic naling) properties of GABAergic transmission (Mueller activity regulated the activation of voltage-dependent et al., 1984 Ben-Ari et al., 1989 Cherubini et al., 1991 Ca21 currents. These findings point to GABA as a critical Luhmann and Prince, 1991 Owens et al., 1996). maturation factor for the switch of the physiological and The GABAA receptor channel predominantly conducts biochemical properties of GABA signaling. Cl2 ions. Consequently, the nature of GABAergic trans- mission, excitatory versus inhibitory, is determined pri- Results marily by the electrochemical gradient for Cl2, which depends on the intra- and extracellular concentrations Switch of GABAergic Transmission from Depolarizing of Cl2. This electrochemical gradient sets the reversal to Hyperpolarizing potential for GABAergic currents (EGABA the membrane To study the change in GABA signaling, we first moni- tored GABA-induced elevations of [Ca21]i over develop- 3 Correspondence: mpoo@uclink.berkeley.edu ment. GABA-mediated depolarization was reflected by 4 These authors contributed equally to this work. an increase in [Ca21]i. Cultures of hippocampal neurons 5 Present address: Division of Neurobiology, Department of Molecu- were loaded with the Ca21-sensitive dye Fluo-4 AM and lar and Cell Biology, University of California, Berkeley, Berkeley California 94720. changes in fluorescence were measured using confocal

Cell 522 Figure 1. Developmental Changes in GABA-Induced Responses (A) Pharmacological profile of GABA-induced elevations of [Ca21]i (BMI 1 PTX, 10 and 50 mM nimodipine, 10 mM thapsigargin 1 BHQ, 2 and 10 mM and baclofen, 10 mM). Time course of changes in [Ca21]i, assessed by Fluo-4 fluorescence intensity. Traces are average fluorescence intensity (% change shown by the scale) recorded from 30���40 randomly sampled neurons in response to pulses of GABA (open bar, 10 mM, 15 s). The elevation of [Ca21]i during application of thapsigargin 1 BHQ was attributed to emptying of intracellular Ca21 stores. (B) Time course of the GABA switch. The percentage of neurons exhibiting Ca21 elevation in response to GABA is shown for different days after cell plating. Each point represents mean 6 sem (n 5 3 to 21 experiments, 121 in total). Each experiment involved recording from 30���40 neurons. Insets: Sample recordings from 20 randomly selected neurons at the time points indicated by the arrows. (C) Representative recordings of Ca21 imaging of young (7 day) and old (13 day) neurons. First panels show bright-field images. The fluorescence images on the right represent typical [Ca21]i in neurons before (���baseline���), during (���GABA���), and after (���recovery���) application of a pulse of GABA (higher intensity represents higher [Ca21]i). Image field 5 604 3 604 mm. (D) Left: Representative examples of peak GABA-induced currents versus membrane voltages (I-V relationship). The cells were voltage clamped at 270 mV and stepped to different potentials (290 to 140 mV, 10 mV steps). EGABA was calculated by fitting the I-V curve to a second-order polynomial function. Right: Averaged I-V curves (n 5 11 in both cases).