European Journal of Neuroscience, Vol. 27, pp. 2402–2412, 2008
doi:10.1111/j.1460-9568.2008.06194.x
Coincident pre- and postsynaptic activity downregulates NKCC1 to hyperpolarize ECl during development Trevor Balena and Melanie A. Woodin Department of Cell & Systems Biology, University of Toronto, Toronto, Ontario, Canada Keywords: cation-chloride cotransporter, GABA, inhibitory synaptic plasticity, inhibitory synaptic transmission, Sprague–Dawley
Abstract In the mature CNS, coincident pre- and postsynaptic activity decreases the strength of c-aminobutyric acid (GABA)A-mediated inhibition through a Ca2+-dependent decrease in the activity of the neuron-specific K+-Cl– cotransporter KCC2. In the present study we examined whether coincident pre- and postsynaptic activity can also modulate immature GABAergic synapses, where the Na+K+-2Cl– (NKCC1) cotransporter maintains a relatively high level of intracellular chloride ([Cl–]i). Dual perforated patch-clamp recordings were made from cultured hippocampal neurons prepared from embryonic Sprague–Dawley rats. These recordings were used to identify GABAergic synapses where the reversal potential for Cl– (ECl) was hyperpolarized with respect to the action potential threshold but depolarized with respect to the resting membrane potential. At these synapses, repetitive postsynaptic spiking within ± 5 ms of GABAergic synaptic transmission resulted in a hyperpolarizing shift of ECl by 10.03 ± 1.64 mV, increasing the strength of synaptic inhibition. Blocking the inward transport of Cl– by NKCC1 with bumetanide (10 lm) hyperpolarized ECl by 16.14 ± 4.8 mV, and prevented this coincident activity-induced shift of ECl. The bumetanide-induced hyperpolarization of ECl occluded furosemide, a K+-Cl– cotransporter antagonist, from producing further shifts in ECl. Together, this indicates that brief coincident pre- and postsynaptic activity strengthens inhibition through a regulation of NKCC1. This study further demonstrates ionic plasticity as a mechanism underlying inhibitory synaptic plasticity.
Introduction Early in the development of the mammalian nervous system, the concentration of intracellular chloride ([Cl–]i) is relatively high, due to the electrically neutral inward transport of Cl– by Na+-K+-2Cl– cotransporter (NKCC1; Xu et al., 1994; Plotkin et al., 1997; Delpire, 2000; Mercado et al., 2004; Dzhala et al., 2005). Under physiological conditions, this cation-chloride cotransporter (CCC) uses energy from the inward Na+ gradient to transport K+ and Cl– into the cell. Because c-aminobutyric acid (GABA)A receptors have a high Cl– permeability (Kaila, 1994), the elevated [Cl–]i of immature neurons renders the reversal potential for GABA (EGABA) more depolarized than the resting membrane potential (RMP). In this scenario, GABAA receptor activation leads to membrane depolarization and neuronal excitation (Mueller et al., 1984; Ben-Ari et al., 1989; Luhmann & Prince, 1991; Zhang et al., 1991). Shortly after birth in rodents there is a change in the expression level of the CCCs maintaining EGABA in the hippocampus; NKCC1 is downregulated, while the neuron-specific K+-Cl– cotransporter (KCC2) is upregulated (Mercado et al., 2004; Rivera et al., 2005). KCC2 derives energy from the K+ gradient to extrude Cl–. The increased upregulation of KCC2 results in EGABA hyperpolarization, which renders GABAergic transmission inhibitory (Rivera et al., 1999; Hubner et al., 2001; Payne et al., 2003). Disrupting the expression or regulation of NKCC1 or KCC2 during development can change the normal inhibition–excitation balance,
Correspondence: Dr M. A. Woodin, as above. E-mail:
[email protected] Received 10 January 2008, revised 18 February 2008, accepted 5 March 2008
which is critical for proper neuronal circuit development and function (Turrigiano & Nelson, 2004; Dzhala et al., 2005; Fukuda, 2005; Hensch & Fagiolini, 2005; Tao & Poo, 2005; Akerman & Cline, 2006; Kanold & Shatz, 2006). Moreover, alterations in neuronal activity can disrupt normal CCC expression and regulation (Fiumelli & Woodin, 2007). For example, chronic blockade of GABAA receptors during development can prevent the normal hyperpolarization of EGABA both in cultured hippocampal neurons (Ganguly et al., 2001) and in vivo in the turtle retina (Leitch et al., 2005). Neuronal activity can also alter CCC function in the mature CNS (Fiumelli & Woodin, 2007). In vitro, brief coincident pre- and postsynaptic activity at mature GABAergic synapses weakens inhibition through a KCC2-mediated increase in [Cl–]i, which depolarizes EGABA (Woodin et al., 2003). This raises the question, are immature GABAergic synapses also modified by spiketiming-dependent plasticity (STDP)? We examined whether coincident pre- and postsynaptic activity could also regulate the strength of GABAergic transmission during development, when it is in transition from excitation to inhibition, which normally occurs during the second postnatal week of hippocampal development (Rivera et al., 1999). Using dual perforated patch-clamp recordings, we examined whether brief coincident preand postsynaptic activity could regulate the strength of shunting inhibitory synapses [action potential (AP) threshold > EGABA > RMP). Shunting inhibition decreases the input resistance leading to a short-circuiting of neighboring excitatory currents (Alger & Nicoll, 1979; Andersen et al., 1980; Stuart et al., 1997; Banke & McBain, 2006), and occurs when GABAergic transmission is hyperpolarizing as well as when it is depolarizing during immature GABAergic transmission. We found that coincident activity led to a
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
Activity-induced hyperpolarization of ECl 2403 NKCC1-mediated hyperpolarizing shift of chloride reversal potential (ECl), which effectively increases synaptic inhibition.
Materials and methods Hippocampal cultures Low-density cultures of dissociated embryonic rat hippocampal neurons were prepared as previously described (Bi & Poo, 1998; Woodin et al., 2003). In brief, E18–19 pregnant Sprague–Dawley rats were briefly exposed to carbon dioxide and cervically dislocated in accordance with guidelines from the Canadian Council on Animal Care. Hippocampi were then removed from E18–19 embryonic rats and treated with trypsin for 15 min at 37 C, followed by gentle trituration. The dissociated cells were plated at a density of 50 000 cells ⁄ mL on poly-l-lysine-coated 25-mm glass coverslips in 35-mm Petri dishes. Cells were plated in Neurobasal medium (Invitrogen, Carlsbad, California, USA), supplemented with 2% B-27 (Invitrogen). Twenty-four hours after plating, half of the medium was replaced with the original plating medium containing 20 mm KCl. Every three days following, one-third of the medium was replaced with the same KCl-supplemented solution. Both glia and neurons were present under these culture conditions. Cells were recorded from after 10–16 days in culture.
Electrophysiology Whole-cell perforated patch recordings using either amphotericin B (150 lg ⁄ mL; Sigma-Aldrich, Oakville, Ontario, Canada) or gramicidin (50 lg ⁄ mL; Sigma-Aldrich) were performed on pairs of synaptically connected cultured hippocampal neurons. Specifically, gramicidin was used for all experiments that required the addition of bumetanide or furosemide (Figs 2 and 5); amphotericin was used in all remaining experiments (Figs 1 and 3). The recording pipettes were made from glass capillaries (World Precision Instruments, Sarasota, Florida, USA), with a resistance of 4–12 MW. The pipettes were filled with an internal solution containing (in mm): K-gluconate, 154; NaCl, 9; MgCl2, 1; HEPES, 1; EGTA, 0.2; and either amphotericin or gramicidin, pH 7.4, osmolarity ¼ 300 mOsmol. The cultures were continuously perfused (approximately 1 mL ⁄ min) with extracellular recording solution containing (in mm): NaCl, 150; KCl, 3; CaCl2Æ2H2O, 3; MgCl2Æ6H2O, 2; HEPES, 10; glucose, 5; pH 7.4, osmolarity ¼ 307–315 mOsmol. Recordings were performed with a MultiClamp 700B (Molecular Devices, Sunnyvale, California, USA) patch-clamp amplifier. Signals were filtered at 5 kHz using amplifier circuitry. Data were acquired and analysed using Clampfit 9 (Molecular Devices). Recordings started after the series resistance had dropped below 30 MW, and were only continued if it did not change by more than 5%. For assaying synaptic connectivity, each neuron was stimulated at a low frequency (0.05 Hz) by a 1-ms step depolarization from )70 to +20 mV in voltage-clamp mode. GABAergic postsynaptic currents (GPSCs) were distinguishable from excitatory postsynaptic currents (EPSCs) by longer decay times and sensitivity to 10 lm of the GABAA receptor antagonist gabazine (tested at the end of the experiment). Upon occasion we did detect autaptic GABAergic synapses in our cultures, however, we did not examine these synapses in the present study. In all recordings, the postsynaptic neuron was voltage clamped at )70 mV resulting in inward GPSCs. During the STDP induction protocol both neurons were switched to current-clamp mode and injected with current (2 nA, 2 ms) both pre- and postsynaptically to generate an AP in each cell, at a frequency of 5 Hz for 30 s. This protocol resulted in 150 pairs of
pre- and postsynaptic APs. For coincident STDP protocols, there was a ± 5 ms delay between pre- and postsynaptic APs; for non-coincident protocols the interval was increased to ± 100 ms. All recordings were performed at room temperature (25 C). The RMP and AP threshold were determined in current-clamp mode; the RMP was determined in the absence of current injection or synaptic activity, while the AP threshold was determined as the point of inflection in the upward AP waveform. ECl was determined by varying the holding potential of the postsynaptic cell in 10-mV increments and measuring the resulting GPSC amplitude; each set of current–voltage (I–V) measurements was repeated after a 5-min interval. A linear regression of both sets of GPSC amplitude measurements was then used to calculate the voltage dependence of GPSCs. The intercept of this line with the abscissa was taken as ECl. The slope of the same line was taken as GPSC conductance. Values have been corrected for the liquid junction potential of 7 mV. The liquid junction potential was calculated experimentally by filling a recording pipette and the bath with a solution containing (in mm): K-gluconate, 154; NaCl, 9; MgCl2, 1; HEPES, 1; EGTA, 0.2; pH 7.4. In current-clamp mode (with no commands) the pipette offset potentiometer was used to null the voltage. The bath solution was then replaced with the extracellular solution used for recordings and the voltage was noted. The liquid junction potential was taken as the inverse of this voltage reading. GABAA receptors are permeable to both HCO3– and Cl– (0.2–0.4 ratio; Kaila, 1994). Due to the relatively positive HCO3– equilibrium potential ()10 mV), which is set by mechanisms that control intracellular pH regulation (Kaila & Voipio, 1987), HCO3– mediates an inward, depolarizing current (Kaila & Voipio, 1987; Kaila et al., 1993; Gulledge & Stuart, 2003). However, our experiments were performed in bicarbonate-free solution buffered with HEPES, and thus GABAA receptor activation was solely mediating a Cl– current. For this reason we report ECl and not EGABA.
Chemicals When bumetanide (10 lm; Sigma-Aldrich) and furosemide (25 and 100 lm; Sigma-Aldrich) were required throughout the entire experiment they were added to the perfusion. When they were required acutely, the perfusion was terminated and the antagonists were added directly to the bath. Bumetanide and furosemide were prepared by making 100 mm stock solutions in anhydrous dimethylsulfoxide, which were then diluted in extracellular solution. Gabazine (10 lm; Tocris, Ellisville, Missouri, USA) was also added directly to the bath in the absence of perfusion.
Statistical analysis All data are presented as mean ± SEM. The P-value reported for control average GPSC amplitude (Fig. 3D) was obtained by comparing the current amplitude in three 10-min bins using a oneway anova. The P-values reported for the CCC-inhibitor-induced changes in the driving force (DF; Fig. 2D) were obtained using a oneway repeated-measures anova followed by a Tukey test. Unless otherwise noted all other P-values reported were obtained using unpaired t-tests. We performed linear regression analyses and obtained the correlation coefficients in order to determine the relationship between: (1) the coincident activity-induced hyperpolarization of ECl and the initial GPSC amplitude; and (2) the coincident activityinduced hyperpolarization of ECl and the amplitude variability. All statistical analysis was performed using SigmaStat 2.03.
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
2404 T. Balena and M. A. Woodin
Fig. 1. Cultured hippocampal GABAergic synapses are shunting during development. (A) Stimulation of presynaptic GABAergic neurons produced GPSCs that were inward (top panel) and GPSPs that were depolarizing (middle panel). GPSCs and GPSPs were both blocked by the GABAA receptor antagonist gabazine (10 lm). EPSCs had relatively short decay time constants (bottom panel) compared with GPSCs. (B) A diagram illustrating the normal hyperpolarization of ECl in hippocampal neurons during development. Left: early in development ECl is more depolarized than AP threshold, rendering GABAergic transmission excitatory. Middle: when ECl is more hyperpolarized than AP threshold, but more depolarized than RMP, GABAergic transmission is largely shunting in nature. This is the category of GABAergic synapses that was examined in the present study. Right: at mature GABAergic synapses, ECl is often hyperpolarized with respect to the RMP, rendering GABAergic transmission hyperpolarizing inhibitory. (C) Example of how shunting inhibition was characterized at each GABAergic synapse. The RMP and the AP threshold were determined in current-clamp mode (i), while ECl was determined in voltage-clamp mode (ii) by stepping the postsynaptic membrane potential in 10-mV increments from )80 mV to )20 mV while stimulating GABAergic synapses in order to generate an I–V curve. Inset: sample traces of GPSCs recorded during the construction of this I–V curve. Scale bars: 50 ms and 100 pA. Only synapses where ECl was more depolarized than the RMP, and hyperpolarized with respect to the AP threshold, were characterized as having shunting inhibition and examined further.
Results Cultured hippocampal GABAergic synapses are shunting during development Dual perforated patch-clamp recordings were made from pairs of hippocampal neurons cultured at a low density. Hippocampal cultures contain glia, pyramidal neurons and GABAergic interneurons. GPSCs and GABAergic postsynaptic potentials (GPSPs) were characterized by their relatively long decay time constants (35.50 ± 3.37 ms; n ¼ 6; Fig. 1A), in comparison to EPSCs (5.76 ± 0.64 ms; n ¼ 5; Fig. 1A), and by their sensitivity to the GABAA receptor antagonist gabazine
(10 lm; Fig. 1A). For each synapse we characterized the RMP and AP threshold in current-clamp mode (Fig. 1Ci). We then determined the ECl in voltage-clamp mode by constructing an I–V curve (Fig. 1Cii). Based on the relation of ECl to RMP, we characterized GABAergic synapses as one of two populations: (1) shunting inhibition, defined by an ECl more depolarized than the RMP and more hyperpolarized than the AP threshold (Fig. 1B middle and 1C); or (2) hyperpolarizing inhibition, defined when ECl was hyperpolarized with respect to the RMP. Because we were interested in examining whether activity during development could regulate Cl– homeostasis and thus synaptic inhibition, we selected GABAergic synapses that were from the
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
Activity-induced hyperpolarization of ECl 2405
Fig. 2. Shunting inhibition is maintained by NKCC1. (A) Representative experiment showing that the acute application of 10 lm bumetanide hyperpolarized ECl (black line: I–V curve obtained in antagonist-free extracellular solution; gray line: I–V curve obtained with extracellular solution containing 10 lm bumetanide). Inset: GPSC recorded in antagonist-free extracellular solution (black trace), and GPSC recorded in extracellular solution containing 10 lm bumetanide (gray trace). Voltage-clamped at )70 mV. Scale bars: 15 ms and 30 pA. (B) Representative experiment showing that the acute application of 25 lm furosemide also hyperpolarized ECl (black line: I–V curve obtained in antagonist-free extracellular solution; solid gray line: I–V curve obtained with extracellular solution containing furosemide; dashed gray line: I–V curve obtained following washout of bumetanide with drug-free extracellular solution). Inset: GPSC recorded in antagonist-free extracellular solution (black trace), and GPSC recorded in extracellular solution containing 25 lm furosemide (gray trace). Voltage-clamped at )70 mV. Scale bars: 15 ms and 20 pA. (C) Summary of antagonist-induced changes in ECl. Both bumetanide and furosemide significantly hyperpolarized ECl (bumetanide, n ¼ 5, P ¼ 0.035; 25 lm furosemide, n ¼ 4, P ¼ 0.04; 100 lm furosemide, n ¼ 7, P ¼ 0.006). There was no significant difference in the magnitude of the ECl shift induced by bumetanide, 25 lm furosemide, or 100 lm furosemide (one-way anova P ¼ 0.599). (D) Acute application of bumetanide hyperpolarized ECl, regardless of whether ECl was initially shunting or hyperpolarizing (black or gray lines, respectively; upper panel). This bumetanide-induced shift in ECl resulted in a significant decrease in the DF for shunting inhibition synapses (lower panel), with no significant change in the DF for hyperpolarizing inhibitory synapses (lower panel). Subsequent application of furosemide produced no further shifts in ECl for synapses that were initially shunting (top panel), and no further change in the DF (bottom panel). In contrast, subsequent application of furosemide significantly depolarized ECl (top panel), and significantly decreased the DF (bottom panel), for synapses that were initially hyperpolarizing. Upper panel: black solid lines represent the results of individual GABAergic synapses that were initially shunting in nature, gray lines represent individual GABAergic synapses that were initially hyperpolarizing inhibitory. The dashed black line is the mean ± SEM of all GABAergic synapses that were initially shunting in nature, the dashed gray line is the mean ± SEM of all GABAergic synapses that were initially hyperpolarizing inhibitory. *Indicates statistical significance.
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
2406 T. Balena and M. A. Woodin shunting inhibition population (Fig. 1B middle panel and 1C). On average, these synapses had an ECl of )48.50 ± 3.08 mV (n ¼ 28), which was more depolarized than the RMP )67.36 ± 1.13 mV (n ¼ 69; P < 0.001), but more hyperpolarized than AP threshold )29.98 ± 1.16 mV (n ¼ 51; P < 0.001). Under these conditions, GPSCs were inward and GPSPs were depolarizing (Fig. 1A), thus these GABAergic synapses did not exhibit any hyperpolarizing inhibition.
Shunting inhibition is maintained by NKCC1 EGABA is largely determined by the electrochemical gradient for Cl– (Kaila, 1994), which is dependent upon the balance of NKCC1 and KCC2 (Rivera et al., 1999; Mercado et al., 2004; Rivera et al., 2005). The normal hyperpolarization of EGABA in development is due to a combination of downregulated NKCC1 and upregulated KCC2 (Rivera et al., 1999; Mercado et al., 2004; Rivera et al., 2005). In order to determine the relative contributions of NKCC1 and KCC2 to Cl– homeostasis in our cultured hippocampal neurons, we recorded from presynaptic GABAergic neurons and their postsynaptic partners, and examined the shift in ECl induced by acute application of CCC antagonists. Gramicidin was used as the perforating agent in this set of experiments because it forms pores that are permeable to monovalent cations and small uncharged molecules but not to Cl–, permitting reliable measurements of ECl (Kyrozis & Reichling, 1995; Owens et al., 1996). Addition of the NKCC1 antagonist bumetanide (10 lm; Payne, 1997; Hannaert et al., 2002; Dzhala et al., 2005) produced a hyperpolarization of ECl by 16.14 ± 4.8 mV (n ¼ 5; P ¼ 0.035; Fig. 2A and C), with no significant change in GPSC conductance (n ¼ 5; P ¼ 0.868). Similarly, application of the non-specific CCC antagonist furosemide also hyperpolarized ECl (25 lm furosemide: 13.9 ± 3.86 mV, n ¼ 4, P ¼ 0.04; 100 lm furosemide: 11.13 ± 2.46 mV, n ¼ 7, P ¼ 0.006; Fig. 2B and C). While furosemide is routinely used at a concentration of 100 lm (Woodin et al., 2003), we found in the present study that this concentration produced a 37.08 ± 25.24% decrease in GABAergic conductance (corresponding to a decrease of 0.52 ± 0.17 pS; n ¼ 7; P ¼ 0.02). Furosemide at a concentration of 100 lm has previously been reported to decrease GABAA conductance (Wafford et al., 1996). For this reason we used a lower concentration of furosemide (25 lm), which in our experiments significantly hyperpolarized ECl without changing GPSC conductance (n ¼ 4; P ¼ 0.134). There was no difference between the magnitude of ECl hyperpolarization induced by bumetanide, 25 lm furosemide or 100 lm furosemide (one-way anova P ¼ 0.599; Fig. 2C). These results suggest that shunting inhibition is being largely maintained by NKCC1. However, because furosemide is not a specific antagonist for KCC2 it is difficult to determine if there is a KCC2-mediated regulation of Cl– in these neurons. In order to gain further insight into the effects of bumetanide and furosemide on Cl– homeostasis we asked whether the bumetanide-
induced hyperpolarization of ECl occluded any further furosemideinduced shifts in ECl. In this experiment we examined both populations of neurons: those that exhibited shunting inhibition and those with hyperpolarizing inhibition. In addition to measuring the inhibitor-induced regulation of ECl, we also quantified the resulting change in the DF for Cl–, which is the difference between the RMP and ECl (a + value indicates an outward DF for Cl–). For GABAergic synapses that were shunting in nature (AP threshold > ECl > RMP), acute bumetanide application (10 lm) induced a significant hyperpolarization of ECl from )46.5 ± 3.6 mV to )65.01 ± 5.15 mV (n ¼ 5; P ¼ 0.025; Fig. 2D top), and effectively eliminated the DF by shifting it from 20.85 ± 3.6 mV to 2.35 ± 5.15 mV (P ¼ 0.004; Fig. 2D bottom). Subsequent addition of 25 lm furosemide produced no further shift in ECl (n ¼ 3; P ¼ 0.424; Fig. 2D) or DF (P ¼ 0.795). Thus, the bumetanide-induced hyperpolarization of ECl and decrease in DF occluded any additional shifts in ECl or DF induced by furosemide. Because both bumetanide and furosemide act on NKCC1, these results suggest that this subpopulation of neurons has low Cl– extrusion and high Cl– uptake largely mediated by NKCC1. We then asked whether bumetanide also occluded the effects of furosemide when we examined synapses from the population characterized by hyperpolarizing inhibition (RMP > ECl). There was a significant difference in ECl and DF between the shunting and hyperpolarizing inhibition groups under control conditions (ECl: )46.5 ± 3.6 mV vs )77.07 ± 3.38 mV, P < 0.001; DF: 20.85 ± 3.6 vs )9.71 ± 3.38, P < 0.001; Fig. 2D). Acute application of bumetanide to hyperpolarizing inhibitory synapses induced a hyperpolarization of ECl to )83.53 ± 4.091 mV (n ¼ 5; P ¼ 0.016; Fig. 2D) and an increase in the DF from )9.71 ± 3.38 to )16.170 ± 4.091 mV (n ¼ 5; Fig. 2D). Unlike with shunting inhibition, these bumetanideinduced shifts in ECl and DF did not occlude further shifts by furosemide. When furosemide was added to hyperpolarizing inhibitory synapses already in the presence of bumetanide, ECl depolarized to )65.89 ± 3.63 mV (n ¼ 5; P ¼ 0.009; Fig. 2D) and the DF decreased to 1.471 ± 3.636 (n ¼ 5; P < 0.001; Fig. 2D), effectively eliminating the DF for Cl–. Thus, the bumetanide-induced hyperpolarization of ECl and increase in DF does not occlude a furosemideinduced depolarization of ECl at GABAergic synapses that are initially hyperpolarizing inhibitory. These results would suggest that the bumetanide-induced block of NKCC1 at hyperpolarizing inhibitory synapses isolates the contribution of KCC2 to ECl regulation. This contribution by KCC2 was then abolished by the addition of furosemide.
Coincident activity hyperpolarizes ECl in developing neurons Mature inhibitory GABAergic synapses in both hippocampal cultures and slices are sensitive to coincident pre- and postsynaptic activity (Woodin et al., 2003; Fiumelli & Woodin, 2007), and repetitive postsynaptic spiking (Fiumelli et al., 2005). STDP occurs when the postsynaptic neuron repetitively fires APs within 20 ms before or after
Fig. 3. Coincident activity hyperpolarizes ECl during development. (Ai) Coincident pre- and postsynaptic activity (5 Hz for 30 s, +5 ms STDP interval) hyperpolarized ECl (black line, I–V curve obtained during the control period; gray line: I–V curve obtained 10 min after STDP induction). Inset: the current traces of GPSCs recorded during the construction of this I–V curve. Scale bars: 50 ms and 120 pA. (Aii) STDP induction (+5 ms, given at arrow) decreased GPSC amplitude due to a decrease in the DF for Cl–. (Bi) Non-coincident activity (5 Hz for 30 s, +100 ms STDP interval) produced a decrease in GPSC conductance with no change in ECl. (black line, I–V curve obtained during the control period; gray line: I–V curve obtained 10 min after non-coincident STDP induction). Inset: the current traces of GPSCs recorded during the construction of this I–V curve. Scale bars: 50 ms and 350 pA. (Bii) STDP induction (+100 ms, given at arrow) decreased GPSC amplitude due to a decrease in GPSC conductance. (C) Summary of the activity-induced changes in ECl and GPSC conductance induced by coincident and noncoincident STDP protocols. Coincident activity induced a significant hyperpolarization of ECl (P ¼ 0.01). Non-coincident activity significantly decreased GPSC conductance (P ¼ 0.05). (D) Using the perforated patch-clamp configuration GABAergic synapses could be recorded for over 30 min with no significant change in GPSC amplitude (n ¼ 6; P ¼ 0.34). ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
Activity-induced hyperpolarization of ECl 2407
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
2408 T. Balena and M. A. Woodin
Fig. 4. STDP-induced ECl hyperpolarization does not depend on initial GPSC amplitude. (A) The relationship between initial GPSC amplitude and the change in ECl induced by coincident pre- and postsynaptic activity. The solid line is a linear regression (r ¼ 0.05; P ¼ 0.901). (B) The relationship between the amplitude variability and the change in ECl induced by coincident pre- and postsynaptic activity. The amplitude variability was taken as the standard deviation (SD) of the GPSC amplitude throughout the duration of the recording. The solid line is a linear regression (r ¼ 0.12; P ¼ 0.776).
the activation of a GABAergic synapse (Woodin et al., 2003). GABAergic STDP is due to a Ca2+-dependent decrease in KCC2 activity, which increases intracellular Cl–. The resulting depolarization of ECl effectively reduces the strength of inhibition. In the present study we wanted to know whether similar physiological patterns of activity can also regulate the strength of inhibition during development. STDP was induced by injecting current pulses into the postsynaptic neuron to fire APs in synchrony with repetitive presynaptic stimulation at a frequency of 5 Hz for 30 s. The spiketiming interval was termed coincident when pre- and postsynaptic spiking was ± 5 ms; when the interval was increased to ± 100 ms the activity was referred to as non-coincident. Coincident activity resulted in an ECl hyperpolarization of )10.03 ± 1.64 mV (n ¼ 8; P ¼ 0.011; Fig. 3A and C), with no change in synaptic conductance (n ¼ 8; P ¼ 0.796; Fig. 3Ai and C). The hyperpolarization of ECl decreased the driving force for Cl– (as ECl approached Vclamp), accounting for the
significant decrease in GPSC amplitude of 36% (n ¼ 8; P ¼ 0.001; Fig. 3Aii). ECl remained hyperpolarized throughout the recordings, which at a maximum were maintained for 40 min following STDP induction. Whether these activity-induced changes in [Cl–]i are maintained longer than 40 min was not examined in the present study. Moreover, because these experiments were performed with dual patch-clamp recordings we can not identify whether the activityinduced plasticity was input specific. Non-coincident activity produced no significant change in ECl ()3.24 ± 1.08 mV; n ¼ 5; P ¼ 0.479; Fig. 3Bi and C), but did decrease the GPSC conductance by 34.8 ± 10.45% (n ¼ 5; P ¼ 0.05; Fig. 3B and C). The decrease in conductance can account for a 17% decrease in GPSC amplitude (n ¼ 5; Fig. 3Bii). The use of amphotericin as the perforating agent did not affect the GPSC amplitude. When synaptic amplitudes were recorded for over 30 min in the absence of STDP, there was no significant change in
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
Activity-induced hyperpolarization of ECl 2409 GPSC amplitude (n ¼ 6; P ¼ 0.34; Fig. 3D). Moreover, stable control recordings of GPSC amplitudes and the ability to induce GABAergic STDP have both been previously demonstrated using amphotericin as the perforating agent (Woodin et al., 2003). The magnitude of the STDP-induced modification of glutamatergic synapses between cultured hippocampal neurons depends upon the initial synaptic strength (Bi & Poo, 1998). By performing a linear regression we determined that at GABAergic synapses the magnitude of the STDP-induced ECl hyperpolarization did not depend on either
the initial GPSC amplitude (r ¼ 0.05; P ¼ 0.901; Fig. 4A) or the amplitude variability during the control recording (r ¼ 0.12; P ¼ 0.776; Fig. 4B). NKCC1 is required for activity-induced hyperpolarization of ECl Based on the predominant action of NKCC1 in maintaining shunting inhibition in our neurons (Fig. 2), we rationalized that coincident
Fig. 5. NKCC1 is required for activity-induced hyperpolarization of ECl. (Ai) Example of I–V curves obtained in the presence of 10 lm bumetanide; pre-STDP (black), post-STDP (gray). (Aii) The current traces of GPSCs recorded during the construction of the I–V curves in (Ai). Scale bars: 10 ms and 100 pA. (B) When coincident pre- and postsynaptic activity was given in the presence of 10 lm bumetanide, there was no significant change in ECl (P ¼ 0.84; dashed line). Solid black lines represent ECl before and after coincident STDP for individual experiments (n ¼ 5). (C) Following STDP, ECl hyperpolarized by )10.03 ± 1.54 mV; this shift was significantly larger than the change in ECl following STDP performed in the presence of bumetanide (n ¼ 5; *P ¼ 0.013). ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
2410 T. Balena and M. A. Woodin activity may be hyperpolarizing ECl through a regulation of this cotransporter. In order to determine whether the coincident activityinduced hyperpolarizing ECl required NKCC1, we repeated the above STDP experiments in the presence of bumetanide, again using gramicidin as the perforating agent. Bumetanide prevented the coincident activity-induced hyperpolarization of ECl (pre-STDP: )90.06 ± 2.71 mV; post-STDP )93.11 ± 3.26 mV; n ¼ 5; P ¼ 0.492; Fig. 5A and B), and left GPSC conductance unchanged (P ¼ 0.841). This non-significant change in ECl following STDP performed in the presence of bumetanide was significantly different than the )10.03 ± 1.54 mV hyperpolarization of ECl that occurred in the absence of bumetanide (P ¼ 0.013; Fig. 5C). Thus, coincident activity hyperpolarizes ECl during development in an NKCC1dependent manner, effectively increasing the strength of inhibition.
Discussion At mature inhibitory GABAergic synapses coincident pre- and postsynaptic spiking alters the activity of KCC2, resulting in a depolarization of EGABA, which decreases the strength of inhibition (Woodin et al., 2003). Our present results demonstrate that immature GABAergic synapses, which are largely shunting in nature, are also sensitive to the temporal pattern of pre- and postsynaptic activity. At these developing synapses, coincident activity acted via NKCC1 to hyperpolarize ECl, which effectively increased synaptic inhibition. When pre- and postsynaptic activity were non-coincident, synaptic inhibition was decreased through a decrease in GPSC conductance, with no change in ECl. These results are particularly important in light of recent findings that suggest that spike-timing is critical in neural circuit information processing and storage (Dan & Poo, 2006). The ability of activity to regulate the strength of synaptic inhibition through a shift in ECl is supported by results from a number of other studies that have also demonstrated activity-induced ionic plasticity (Fiumelli & Woodin, 2007). Spontaneous activity in the isolated spinal cord of the chick embryo induces a depression of GABAergic transmission through a hyperpolarization of EGABA (Chub & O’Donovan, 2001). At more mature synapses, repetitive postsynaptic spiking (Fiumelli et al., 2005) and coincident pre- and postsynaptic activity (Woodin et al., 2003) resulted in a weakening of GABAA-mediated transmission through a decrease in KCC2 activity. Furthermore, when GABAergic transmission is antagonized during development synaptic inhibition fails to develop (Leitch et al., 2005). This present study builds on these previous demonstrations of ionic plasticity by providing the first evidence that activity-induced plasticity of synaptic inhibition can occur at immature synapses via NKCC1 regulation. However, to fully understand how activity regulates GABAergic transmission, we still need to examine whether excitatory GABAergic transmission between hippocampal neurons is also modified by coincident pre- and postsynaptic activity. As mentioned previously, repeated postsynaptic spiking (10 Hz, 5 min) produces a KCC2-mediated hyperpolarization of ECl in mature neurons (Fiumelli et al., 2005). We did not examine whether repeated postsynaptic spiking regulated ECl in immature neurons with shunting inhibition because we were more interested in examining the importance of the temporal order of spikes, which is known to be critical for information processing in the CNS (Froemke & Dan, 2002; Dan & Poo, 2006). Moreover, the mechanisms underlying the coincident pre- and postsynaptic activity-induced hyperpolarization of ECl differ from the postsynaptic spiking-induced hyperpolarization of ECl. In particular, the postsynaptic spiking-induced hyperpolarization of ECl requires the release of Ca2+ from internal stores (Fiumelli
et al., 2005), while the coincident activity-induced regulation of ECl does not (Woodin et al., 2003). Shunting inhibition is independent of the polarity of GABAergic transmission; that is to say, it occurs when GPSCs are both depolarizing or hyperpolarizing (as long as EGABA is hyperpolarized with respect to the AP threshold). When GPSCs are depolarizing (but subthreshold), the nature of the inhibition is largely shunting (Alger & Nicoll, 1979; Andersen et al., 1980; Stuart et al., 1997). In contrast, when GPSCs are hyperpolarizing the nature of the inhibition is mediated by both hyperpolarization of the postsynaptic membrane and shunting inhibition. While we did not examine shunting inhibition at mature GABAergic synapses in the present study, Woodin et al. (2003) previously demonstrated that coincident pre- and postsynaptic activity at GABAergic synapses with hyperpolarizing and shunting inhibition produced a KCC2-mediated hyperpolarization of ECl. Moreover, we did not demonstrate in the present study that the immature GABAergic transmission produces a decrease in input resistance or a short-circuiting of neighboring excitatory synapses; we simply defined synapses as having shunting inhibition based on the relationship between ECl, RMP and AP threshold. During development of the hippocampus, shunting inhibition is observed transiently onto pyramidal cells, as GABAA-mediated currents pass from excitatory to hyperpolarizing inhibitory (Rivera et al., 1999; Ben-Ari, 2002; Rivera et al., 2005). However, shunting inhibition has recently been identified onto interneurons in the CA3 region of the more mature hippocampus (Banke & McBain, 2006; Szabadics et al., 2006). In our cultures we recorded from a mixed population of hippocampal neurons and thus can not distinguish whether coincident activity-induced ECl hyperpolarization was celltype specific. In the future it will also be interesting to determine whether coincident activity can also regulate GABAergic synapses that have reverted to an immature state as a result of trauma (van den Pol et al., 1996; Nabekura et al., 2002), pain (Coull et al., 2005) or epileptic activity (Rivera et al., 2002; Payne et al., 2003; Rivera et al., 2005; Fiumelli & Woodin, 2007). Through the use of the NKCC1-specific antagonist bumetanide (Gillen et al., 1996; Holtzman et al., 1998; Race et al., 1999; Hannaert et al., 2002), we demonstrated that depolarizing GABAA-mediated responses are largely maintained by Na+-K+-Cl– cotransport. Similar to reports from other studies (Yamada et al., 2004; Dzhala et al., 2005; Nakanishi et al., 2007), bumetanide hyperpolarized ECl. We then wanted to know whether there was any additional CCC-mediated Cl– regulation in these neurons. Unfortunately there is no specific antagonist to the neuron-specific KCC2. As an alternative we used the non-specific CCC antagonist furosemide, which has similar Ki values for NKCC1 and KCC2 (Payne, 1997; Payne et al., 2003). Our rational was that if after the bumetanide-induced hyperpolarization of ECl furosemide produced no further shift in ECl, then we could conclude that there was no significant KCC2 regulation in these immature neurons. Moreover, in the presence of both antagonists there should be no Cl– regulation and thus no driving force for Cl–. What we found was that for neurons from the shunting inhibition population, the use of bumetanide occluded any further furosemide-induced shifts in ECl. Moreover, in the presence of either bumetanide, or bumetanide and furosemide, there was effectively no driving force for Cl–. Thus, we can conclude that shunting inhibition at immature GABAergic synapses was largely maintained by NKCC1. As a control experiment, we repeated the bumetanide–furosemide occlusion experiments on mature neurons with hyperpolarizing inhibition. We note that furosemide is used routinely to antagonize KCC2 in more mature neurons when there is a high KCC2 : NKCC1 expression ratio (Woodin et al., 2003). We hypothesized that if
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
Activity-induced hyperpolarization of ECl 2411 furosemide produced a further change in ECl, following the bumetanide shift, this change could be attributable to KCC2. We found that bumetanide alone produced an initial hyperpolarizing shift in ECl. The ability of bumetanide to hyperpolarize ECl in mature neurons suggests that when GABAergic transmission is hyperpolarizing inhibitory, these neurons still have some functional NKCC1mediated Cl– regulation. However, bumetanide did not occlude additional furosemide-induced shifts in ECl at these mature synapses. Addition of furosemide to mature synapses already in the presence of bumetanide depolarized ECl. Moreover, in the presence of bumetanide and furosemide there was no effective driving force for Cl–. Collectively these results demonstrate that shunting inhibition in immature neurons is largely maintained by NKCC1, while hyperpolarizing inhibition in mature neurons is maintained by both NKCC1 and KCC2. Furthermore, we demonstrated that the presence of bumetanide occluded the hyperpolarization of ECl induced by coincident activity, suggesting that coincident activity mediates a decrease in NKCC1 activity during development. In contrast, at mature inhibitory synapses, coincident activity depolarizes EGABA through a decrease in KCC2 activity (Woodin et al., 2003). However, in the present study no such depolarization of ECl was observed in the presence of bumetanide, suggesting that during development coincident activity is primarily regulating NKCC1. One similarity between the present findings on immature neurons and the findings by Woodin et al. (2003) on mature neurons is that non-coincident activity produced no change in ECl, but did produce a decrease in GABAergic conductance. The decrease in conductance resulted in a decrease in GPSC amplitude. The mechanism underlying the non-coincident activityinduced change in conductance has not been examined in either study. Phosphorylation and dephosphorylation are firmly established in the regulation of NKCC1 transport in non-neuronal cells (Payne & Forbush, 1995). For example, in secretory tubules, decreasing [Cl–]i leads to phosphorylation of NKCC1, which activates the transporter (Lytle & Forbush, 1996). Based on the time-course of NKCC1 downregulation in the present study (< 5 min; Fig. 3A and B), we suggest that activity leads to the dephosphorylation of NKCC1, which in turn hyperpolarizes ECl. This suggested mechanism does not preclude changes in NKCC1 membrane expression, which may be required to maintain the activityinduced hyperpolarization of ECl in the long term. Our pharmacological evidence suggests that at the shunting inhibitory synapses in our cultures there was little KCC2-mediated Cl– regulation. However, because KCC2 expression levels do not always correspond with predicted EGABA values, the present results do not preclude the possibility that KCC2 was expressed in our cultures. In the lateral superior olive, where KCC2 is expressed early in development, there is ineffective KCC2-mediated Cl– extrusion (Blaesse et al., 2006). In cultured hippocampal neurons, the developmental expression of KCC2 also fails to parallel the functional activity of KCC2, due to a kinasedependent rate-limiting step that is required for KCC2 transport (Khirug et al., 2005). Moreover, protein kinase C-dependent phosphorylation of KCC2 has been demonstrated to increase the targeting of KCC2 to the neuronal cell surface (Lee et al., 2007). Taken together with previous work, a model emerges where coincident activity can induce bi-directional plasticity of GABAergic transmission, with the direction of the modification dependent upon the maturity of the synapse. When GABAergic synapses have matured, the same pattern of activity weakens inhibition through a depolarization of EGABA (Woodin et al., 2003). However, at immature GABAergic synapses, where a relatively depolarized ECl is maintained by NKCC1, coincident activity hyperpolarizes ECl, increasing the strength of inhibition.
Acknowledgements We acknowledge John Ormond, Hubert Fiumelli and MuMing Poo for their helpful comments on the manuscript. This work was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery grant to M.A.W.
Abbreviations AP, action potential; CCC, cation-chloride cotransporter; DF, driving force; ECl, chloride reversal potential; EGABA, GABA reversal potential; EPSC, excitatory postsynaptic current; GABA, c-aminobutyric acid; GPSC, GABAergic postsynaptic current; GPSP, GABAergic postsynaptic potential; KCC2, neuron-specific K+-Cl– cotransporter; NKCC1, Na+-K+-2Cl– cotransporter; RMP, resting membrane potential; STDP, spike-timing-dependent plasticity; [Cl–]i, concentration of intracellular chloride.
References Akerman, C.J. & Cline, H.T. (2006) Depolarizing GABAergic conductances regulate the balance of excitation to inhibition in the developing retinotectal circuit in vivo. J. Neurosci., 26, 5117–5130. Alger, B.E. & Nicoll, R.A. (1979) GABA-mediated biphasic inhibitory responses in hippocampus. Nature, 281, 315–317. Andersen, P., Dingledine, R., Gjerstad, L., Langmoen, I.A. & Laursen, A.M. (1980) Two different responses of hippocampal pyramidal cells to application of gamma-amino butyric acid. J. Physiol., 305, 279–296. Banke, T.G. & McBain, C.J. (2006) GABAergic input onto CA3 hippocampal interneurons remains shunting throughout development. J. Neurosci., 26, 11720–11725. Ben-Ari, Y. (2002) Excitatory actions of gaba during development: the nature of the nurture. Nat. Rev. Neurosci., 3, 728–739. Ben-Ari, Y., Cherubini, E., Corradetti, R. & Gaiarsa, J.L. (1989) Giant synaptic potentials in immature rat CA3 hippocampal neurones. J. Physiol., 416, 303– 325. Bi, G.Q. & Poo, M.M. (1998) Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type. J. Neurosci., 18, 10464–10472. Blaesse, P., Guillemin, I., Schindler, J., Schweizer, M., Delpire, E., Khiroug, L., Friauf, E. & Nothwang, H.G. (2006) Oligomerization of KCC2 correlates with development of inhibitory neurotransmission. J. Neurosci., 26, 10407– 10419. Chub, N. & O’Donovan, M.J. (2001) Post-episode depression of GABAergic transmission in spinal neurons of the chick embryo. J. Neurophysiol., 85, 2166–2176. Coull, J.A., Beggs, S., Boudreau, D., Boivin, D., Tsuda, M., Inoue, K., Gravel, C., Salter, M.W. & De Koninck, Y. (2005) BDNF from microglia causes the shift in neuronal anion gradient underlying neuropathic pain. Nature, 438, 1017–1021. Dan, Y. & Poo, M.M. (2006) Spike timing-dependent plasticity: from synapse to perception. Physiol. Rev., 86, 1033–1048. Delpire, E. (2000) Cation-chloride cotransporters in neuronal communication. News Physiol. Sci., 15, 309–312. Dzhala, V.I., Talos, D.M., Sdrulla, D.A., Brumback, A.C., Mathews, G.C., Benke, T.A., Delpire, E., Jensen, F.E. & Staley, K.J. (2005) NKCC1 transporter facilitates seizures in the developing brain. Nat. Med., 11, 1205– 1213. Fiumelli, H., Cancedda, L. & Poo, M.M. (2005) Modulation of GABAergic transmission by activity via postsynaptic Ca2+-dependent regulation of KCC2 function. Neuron, 48, 773–786. Fiumelli, H. & Woodin, M.A. (2007) Role of activity-dependent regulation of neuronal chloride homeostasis in development. Curr. Opin. Neurobiol., 17, 81–86. Froemke, R.C. & Dan, Y. (2002) Spike-timing-dependent synaptic modification induced by natural spike trains. Nature, 416, 433–438. Fukuda, A. (2005) Diuretic soothes seizures in newborns. Nat. Med., 11, 1153– 1154. Ganguly, K., Schinder, A.F., Wong, S.T. & Poo, M. (2001) GABA itself promotes the developmental switch of neuronal GABAergic responses from excitation to inhibition. Cell, 105, 521–532. Gillen, C.M., Brill, S., Payne, J.A. & Forbush 3rd, B. (1996) Molecular cloning and functional expression of the K-Cl cotransporter from rabbit, rat, and human. A new member of the cation-chloride cotransporter family. J. Biol. Chem., 271, 16237–16244.
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
2412 T. Balena and M. A. Woodin Gulledge, A.T. & Stuart, G.J. (2003) Excitatory actions of GABA in the cortex. Neuron, 37, 299–309. Hannaert, P., Alvarez-Guerra, M., Pirot, D., Nazaret, C. & Garay, R.P. (2002) Rat NKCC2 ⁄ NKCC1 cotransporter selectivity for loop diuretic drugs. Naunyn Schmiedebergs Arch. Pharmacol., 365, 193–199. Hensch, T.K. & Fagiolini, M. (2005) Excitatory-inhibitory balance and critical period plasticity in developing visual cortex. Prog. Brain Res., 147, 115– 124. Holtzman, E.J., Kumar, S., Faaland, C.A., Warner, F., Logue, P.J., Erickson, S.J., Ricken, G., Waldman, J. & Dunham, P.B. (1998) Cloning, characterization, and gene organization of K-Cl cotransporter from pig and human kidney and C. elegans. Am. J. Physiol., 275, F550–F564. Hubner, C.A., Stein, V., Hermans-Borgmeyer, I., Meyer, T., Ballanyi, K. & Jentsch, T.J. (2001) Disruption of KCC2 reveals an essential role of K-Cl cotransport already in early synaptic inhibition. Neuron, 30, 515–524. Kaila, K. (1994) Ionic basis of GABAA receptor channel function in the nervous system. Prog. Neurobiol., 42, 489–537. Kaila, K. & Voipio, J. (1987) Postsynaptic fall in intracellular pH induced by GABA-activated bicarbonate conductance. Nature, 330, 163–165. Kaila, K., Voipio, J., Paalasmaa, P., Pasternack, M. & Deisz, R.A. (1993) The role of bicarbonate in GABAA receptor-mediated IPSPs of rat neocortical neurones. J. Physiol., 464, 273–289. Kanold, P.O. & Shatz, C.J. (2006) Subplate neurons regulate maturation of cortical inhibition and outcome of ocular dominance plasticity. Neuron, 51, 627–638. Khirug, S., Huttu, K., Ludwig, A., Smirnov, S., Voipio, J., Rivera, C., Kaila, K. & Khiroug, L. (2005) Distinct properties of functional KCC2 expression in immature mouse hippocampal neurons in culture and in acute slices. Eur. J. Neurosci., 21, 899–904. Kyrozis, A. & Reichling, D.B. (1995) Perforated-patch recording with gramicidin avoids artifactual changes in intracellular chloride concentration. J. Neurosci. Meth., 57, 27–35. Lee, H.H., Walker, J.A., Williams, J.R., Goodier, R.J., Payne, J.A. & Moss, S.J. (2007) Direct protein kinase C-dependent phosphorylation regulates the cell surface stability and activity of the potassium chloride cotransporter KCC2. J. Biol. Chem., 282, 29777–29784. Leitch, E., Coaker, J., Young, C., Mehta, V. & Sernagor, E. (2005) GABA typeA activity controls its own developmental polarity switch in the maturing retina. J. Neurosci., 25, 4801–4805. Luhmann, H.J. & Prince, D.A. (1991) Postnatal maturation of the GABAergic system in rat neocortex. J. Neurophysiol., 65, 247–263. Lytle, C. & Forbush 3rd, B. (1996) Regulatory phosphorylation of the secretory Na-K-Cl cotransporter: modulation by cytoplasmic Cl. Am. J. Physiol., 270, C437–C448. Mercado, A., Mount, D.B. & Gamba, G. (2004) Electroneutral cation-chloride cotransporters in the central nervous system. Neurochem. Res., 29, 17–25. Mueller, A.L., Taube, J.S. & Schwartzkroin, P.A. (1984) Development of hyperpolarizing inhibitory postsynaptic potentials and hyperpolarizing response to gamma-aminobutyric acid in rabbit hippocampus studied in vitro. J. Neurosci., 4, 860–867. Nabekura, J., Ueno, T., Okabe, A., Furuta, A., Iwaki, T., Shimizu-Okabe, C., Fukuda, A. & Akaike, N. (2002) Reduction of KCC2 expression and GABAA receptor-mediated excitation after in vivo axonal injury. J. Neurosci., 22, 4412–4417. Nakanishi, K., Yamada, J., Takayama, C., Oohira, A. & Fukuda, A. (2007) NKCC1 activity modulates formation of functional inhibitory synapses in cultured neocortical neurons. Synapse, 61, 138–149. Owens, D.F., Boyce, L.H., Davis, M.B. & Kriegstein, A.R. (1996) Excitatory GABA responses in embryonic and neonatal cortical slices demonstrated by
gramicidin perforated-patch recordings and calcium imaging. J. Neurosci., 16, 6414–6423. Payne, J.A. (1997) Functional characterization of the neuronal-specific K-Cl cotransporter: implications for [K+]o regulation. Am. J. Physiol., 273, C1516–C1525. Payne, J.A. & Forbush 3rd, B. (1995) Molecular characterization of the epithelial Na-K-Cl cotransporter isoforms. Curr. Opin. Cell Biol., 7, 493– 503. Payne, J.A., Rivera, C., Voipio, J. & Kaila, K. (2003) Cation-chloride co-transporters in neuronal communication, development and trauma. Trends Neurosci., 26, 199–206. Plotkin, M.D., Snyder, E.Y., Hebert, S.C. & Delpire, E. (1997) Expression of the Na-K-2Cl cotransporter is developmentally regulated in postnatal rat brains: a possible mechanism underlying GABA’s excitatory role in immature brain. J. Neurobiol., 33, 781–795. van den Pol, A.N., Obrietan, K. & Chen, G. (1996) Excitatory actions of GABA after neuronal trauma. J. Neurosci., 16, 4283–4292. Race, J.E., Makhlouf, F.N., Logue, P.J., Wilson, F.H., Dunham, P.B. & Holtzman, E.J. (1999) Molecular cloning and functional characterization of KCC3, a new K-Cl cotransporter. Am. J. Physiol., 277, C1210–C1219. Rivera, C., Li, H., Thomas-Crusells, J., Lahtinen, H., Viitanen, T., Nanobashvili, A., Kokaia, Z., Airaksinen, M.S., Voipio, J., Kaila, K. & Saarma, M. (2002) BDNF-induced TrkB activation down-regulates the K+Cl) cotransporter KCC2 and impairs neuronal Cl) extrusion. J. Cell Biol., 159, 747–752. Rivera, C., Voipio, J. & Kaila, K. (2005) Two developmental switches in GABAergic signalling: the K+-Cl) cotransporter KCC2 and carbonic anhydrase CAVII. J. Physiol., 562, 27–36. Rivera, C., Voipio, J., Payne, J.A., Ruusuvuori, E., Lahtinen, H., Lamsa, K., Pirvola, U., Saarma, M. & Kaila, K. (1999) The K+ ⁄ Cl) co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature, 397, 251–255. Stuart, G., Spruston, N., Sakmann, B. & Hausser, M. (1997) Action potential initiation and backpropagation in neurons of the mammalian CNS. Trends Neurosci., 20, 125–131. Szabadics, J., Varga, C., Molnar, G., Olah, S., Barzo, P. & Tamas, G. (2006) Excitatory effect of GABAergic axo-axonic cells in cortical microcircuits. Science, 311, 233–235. Tao, H.W. & Poo, M.M. (2005) Activity-dependent matching of excitatory and inhibitory inputs during refinement of visual receptive fields. Neuron, 45, 829–836. Turrigiano, G.G. & Nelson, S.B. (2004) Homeostatic plasticity in the developing nervous system. Nat. Rev. Neurosci., 5, 97–107. Wafford, K.A., Thompson, S.A., Thomas, D., Sikela, J., Wilcox, A.S. & Whiting, P.J. (1996) Functional characterization of human gammaaminobutyric acidA receptors containing the alpha 4 subunit. Mol. Pharmacol., 50, 670–678. Woodin, M.A., Ganguly, K. & Poo, M.M. (2003) Coincident pre- and postsynaptic activity modifies GABAergic synapses by postsynaptic changes in Cl- transporter activity. Neuron, 39, 807–820. Xu, J.C., Lytle, C., Zhu, T.T., Payne, J.A., Benz, E. Jr & Forbush 3rd, B. (1994) Molecular cloning and functional expression of the bumetanide-sensitive Na-K-Cl cotransporter. Proc. Natl Acad. Sci. USA, 91, 2201–2205. Yamada, J., Okabe, A., Toyoda, H., Kilb, W., Luhmann, H.J. & Fukuda, A. (2004) Cl- uptake promoting depolarizing GABA actions in immature rat neocortical neurones is mediated by NKCC1. J. Physiol., 557, 829–841. Zhang, L., Spigelman, I. & Carlen, P.L. (1991) Development of GABAmediated, chloride-dependent inhibition in CA1 pyramidal neurones of immature rat hippocampal slices. J. Physiol., 444, 25–49.
ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd European Journal of Neuroscience, 27, 2402–2412
