The cation-chloride cotransporter NKCC1 promotes sharp waves in the neonatal rat hippocampus

Authors


Corresponding author K. Kaila: Department of Biological and Environmental Sciences, PO Box 65 (Viikinkaari 1), University of Helsinki, FIN-00014 Helsinki, Finland. Email: kai.kaila@helsinki.fi

Abstract

Earlier studies indicate a crucial role for the interconnected network of intrinsically bursting CA3 pyramidal neurons in the generation of in vivo hippocampal sharp waves (SPWs) and their proposed neonatal in vitro counterparts, the giant depolarizing potentials (GDPs). While mechanisms involving ligand- and voltage-gated channels have received lots of attention in the generation of CA3 network events in the immature hippocampus, the contribution of ion-transport mechanisms has not been extensively studied. Here, we show that bumetanide, a selective inhibitor of neuronal Cl uptake mediated by the Na+–K+–2Cl cotransporter isoform 1 (NKCC1), completely and reversibly blocks SPWs in the neonate (postnatal days 7–9) rat hippocampus in vivo, an action also seen on GDPs in slices (postnatal days 1–8). These findings strengthen the view that GDPs and early SPWs are homologous events. Gramicidin-perforated patch recordings indicated that NKCC1 accounts for a large (∼10 mV) depolarizing driving force for the GABAA current in the immature CA3 pyramids. Consistent with a reduction in the depolarization mediated by endogenous GABAA-receptor activation, bumetanide inhibited the spontaneous bursts of individual neonatal CA3 pyramids, but it slightly increased the interneuronal activity as seen in the frequency of spontaneous GABAergic currents. An inhibitory effect of bumetanide was seen on the in vitro population events in the absence of synaptic GABAA receptor-mediated transmission, provided that a tonic GABAA receptor-mediated current was present. Our work indicates that NKCC1 expressed in CA3 pyramidal neurons promotes network activity in the developing hippocampus.

During slow-wave sleep, awake immobility, and consummatory behaviours such as eating, drinking and grooming, the interconnected network (Lebovitz et al. 1971; MacVicar & Dudek, 1980) of intrinsically bursting CA3 pyramidal neurons (Kandel & Spencer, 1961) generates endogenous network events known as hippocampal sharp waves (SPWs; Jouvet et al. 1959; Vanderwolf, 1969; O'Keefe & Nadel, 1978; Buzsaki, 1986; Suzuki & Smith, 1987). SPWs are the first patterned type of network activity generated by the hippocampus during development, but the associated ripple activity (O'Keefe & Nadel, 1978; Ylinen et al. 1995) is not seen until around postnatal day 14 in the rat (Leinekugel et al. 2002; Karlsson & Blumberg, 2003; Buhl & Buzsaki, 2005). Spontaneous network events known as ‘giant depolarizing potentials’ (GDPs; Ben Ari et al. 1989), are thought to be the in vitro counterpart of early SPWs (Leinekugel et al. 2002).

The CA3 region acts as a pacemaker for GDP initiation (Ben Ari, 2001), which is associated with a simultaneous build-up of neuronal excitation in pyramidal neurons and interneurons (Menendez de la Prida & Sanchez-Andres, 2000; Sipiläet al. 2005). These network events are completely blocked by selective AMPA-receptor antagonists (Bolea et al. 1999; see also Ben Ari et al. 1989; Lamsa et al. 2000) indicating a crucial role for glutamatergic transmission in neuronal synchronization. Moreover, the temporal patterns of GDP activity are shaped by the intrinsic bursting properties of neonatal CA3 pyramidal neurons (Sipiläet al. 2005). Hence, the mechanism of GDP initiation, based on bursting CA3 pyramidal neurons mutually coupled by excitatory connections, is similar to that of adult SPWs (see Traub & Wong, 1982; Buzsaki, 1986; Suzuki & Smith, 1987).

During the two first postnatal weeks in the rat hippocampus, GABAA receptor-mediated responses are depolarizing (Ben Ari et al. 1989) and after excitator (Dzhala & Staley, 2003; Khazipov et al. 2004). While the interneuronal network does not generate network activity in the absence of glutamatergic transmission (Bolea et al. 1999), endogenous GABAergic signalling facilitates the voltage-dependent intrinsic bursting of the immature CA3 pyramids and permits their synchronization during GDPs (Sipiläet al. 2005). However, the reversal potential of GABAergic responses (EGABA) has not been measured in neonatal CA3 pyramidal neurons with methods that leave the intracellular Cl concentration intact.

Uptake of Cl by the Na+–K+–2Cl cotransporter isoform 1 (NKCC1) has been shown to provide the driving force for depolarizing GABAA receptor-mediated responses in various types of immature neurons (Rohrbough & Spitzer, 1996; Plotkin et al. 1997; Fukuda et al. 1998; Sun & Murali, 1999; Li et al. 2002; Yamada et al. 2004; Rivera et al. 2005; Chub et al. 2006). The ontogenetic shift to hyperpolarizing GABA action is caused by a concomitant developmental down-regulation of NKCC1 and an up-regulation of the K+–Cl cotransporter isoform 2 (KCC2; Rivera et al. 1999; Yamada et al. 2004; Lee et al. 2005). Although the NKCC1 protein is expressed at high levels in the neonatal CA3 pyramids (Marty et al. 2002), its functional significance has not been studied in these cells.

Consistent with the facilitatory role of depolarizing GABA, GDPs were recently shown to be blocked by bumetanide (Dzhala et al. 2005), a specific inhibitor of NKCC1 (Isenring et al. 1998; Payne et al. 2003). In this work, we show for the first time that blocking NKCC1 by bumetanide inhibits SPWs in the neonate hippocampus in vivo. Furthermore, our data demonstrate that NKCC1 provides a large (∼10 mV) depolarizing driving force for GABAA receptor-mediated Cl currents in immature CA3 pyramidal neurons and increases the general level of excitability by facilitating the spontaneous burst activity of these cells. Taken together, our data point to a key role for NKCC1 in CA3 pyramidal neurons, but not in interneurons, in the facilitation of in vivo SPWs in the neonatalrat hippocampus.

Methods

In vivo electrophysiological recordings

All animal experiments were approved by the local Ethics Committee for Animal Research at the University of Helsinki. Wistar rat pups (postnatal day P5–P6, where P0 refers to the day of birth) were anaesthetized using hypothermia (Cunningham & McKay, 1993; Lahtinen et al. 2002). The depth of hypothermia was assessed by observing reactions to tail or paw pinch. Rat pups were kept together with their parents and littermates except during operation and recording sessions. Topical application of lidocaine (1%; Braun, Melsungen, Germany) was used to cause analgesia of the skull during operation and postoperatively. Craniectomies were performed without damaging the underlying dura using a miniature drill equipped with a 0.7 mm diameter carbide dental burr (ELA, Engelskirchen, Germany). A stereotaxic instrument (Stoelting, Wood Dale, IL, USA) was used to place the electrode tips into the hippocampal CA3 region: 1.6–2.7 mm posterior from bregma, 1.6–2.2 mm lateral from midline, 1.8–2.4 mm below dura. The tips of the Teflon-coated silver wire electrodes (uncoated diameter 0.125 mm; Advent Research Materials Ltd, Halesworth, UK) were chlorided and then implanted in the hippocampus. A reference electrode was placed subdurally over the cerebellum. The implanted electrodes were connected to a microconnector (GM-4; Microtech, Conchester, PA, USA), which was fixed to the skull using dental acrylic. To verify the electrode positions, dye injections into hippocampus were made with the electrode coordinates. The brain was removed and coronal sections were cut (slice thickness 200 μm) with subsequent light-microscopic examination. After the operation, the incision made for electrode implantation was sutured using 6–0 monofilament. After complete recovery from hypothermic anaesthesia, the pups were returned to their original litter and the recordings were performed at an age of P7–9.

The rectal body temperature of the pups was measured using a thermocouple (K101; Voltcraft, Hirschau, Germany) and it was maintained at ∼33–34°C by controlling the ambient temperature during the in vivo electrophysiological recordings. Direct-current (DC) recordings of the hippocampal activity were made with Ag–AgCl electrodes and a custom-designed DC amplifier (Tallgren et al. 2005). The amplified hippocampal signals were sampled at 0.5–3 kHz using a 12-bit data acquisition AD-board (National Instruments, Austin, TX, USA).

Bumetanide (50 mg ml−1 in DMSO diluted to 0.5 mg ml−1 with 0.9% NaCl) was applied intraperitoneally at a dose of 5 μmol kg−1. Control recordings with the vehicle only showed no effect on SPWs. SPW duration was taken from time points where the signal amplitude during rise and decay showed a deviation of 3 s.d. from the baseline ‘noise’.

In vitro electrophysiological recordings

Wistar rat pups (P1–8) were decapitated, and the brains were dissected in cold (0–4°C) oxygenated (95% O2–5% CO2) standard solution containing (mm): 124 NaCl, 3.0 KCl, 2.0 CaCl2, 25 NaHCO3, 1.1 NaH2PO4, 1.3 MgSO4, and 10 d-glucose, pH 7.4 at 32°C. Coronal brain slices (350–600 μm) were cut with a vibrating-blade microtome (VT1000S; Leica, Nussloch, Germany) and allowed to recover at 32°C for > 1 h before use.

Individual slices were transferred into a submersion-type recording chamber perfused with the standard solution (32–33°C). The CA3 pyramidal neurons were visually identified using infrared video microscopy (Stuart et al. 1993). An Axopatch 200A amplifier was used for whole-cell recordings. Patch pipettes had a resistance of 5–8 MΩ when filled with (mm): 140 caesium methanesulphonate (CsMs), 2 MgCl2 and 10 Hepes, pH 7.2 with CsOH; 2 mm EGTA and 5 mm MgATP were included in the pipette filling solution in some experiments. Only those recordings were analysed where the access resistance was less than 12% of the input resistance of the neuron. EGABA was measured with gramicidin-perforated patch recordings using a pipette filling solution containing 150 mm KCl, 10 mm Hepes (pH 7.2 with KOH) and 100–250 μg ml−1 gramicidin D (Sigma, St Louis, MO, USA). Gramicidin was dissolved in a 50 mg ml−1 DMSO stock solution. The CA3 cells were held at their resting membrane potential (RMP), and current–voltage (I–V) relations were obtained from peak responses elicited by laser-flash photolysis of caged GABA (O-(CNB-caged) γ-aminobutyric acid, a-carboxy-2-nitrobenzyl ester, trifluoroacetic acid salt; Molecular Probes, Eugene, OR, USA) during 1 s steps to different membrane voltages (Khirug et al. 2005). Caged GABA (2 mm) was delivered to the vicinity of the recorded cell using an UltraMicroPump II syringe pump (WPI, Sarasota, FL, USA). For local photolysis of caged GABA, 15 ms UV laser flashes (Enterprise 653; Coherent, Santa Clara, CA, USA) were delivered via a multimode optical fibre through the objective. Focusing the UV beam yielded an uncaging spot of 10 μm in diameter (Khirug et al. 2005). The recorded intracellular voltage was corrected for a calculated −13 mV and −3.6 mV liquid-junction potential in whole-cell and perforated-patch recordings, respectively (Barry, 1994). Extracellular field potential (FP) recordings were performed with conventional NaCl filled (150 mm) glass capillary electrodes (tip diameter 5–10 μm) placed in the CA3 stratum pyramidale.

2,3-Dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinox-aline-7-sulphonoamide (NBQX), DL-2-amino-5-phosphonovalecic acid (D,L-AP-5), 6-imino-3-(4-methoxyphenyl)-1(6H)-pyridazinebutanoic acid hydrobromide (SR 95531, gabazine), and isoguvacine hydrochloride were from Tocris Cookson (Bristol, UK). Picrotoxin and 3-(aminosulphonyl)-5-(butylamino)-4-phenoxybenzoic acid (bumetanide) were from Sigma. The concentrations of NBQX, d,l-AP-5 and picrotoxin were 10 μm, 40 μm and 100 μm, respectively. In some experiments, the extracellular K+ concentration ([K+]o) was raised up to 11 mm by adding KCl. For detailed analysis of the effects of [K+]o elevation on spontaneous unit and network activity of CA3 pyramidal neurons, see Sipiläet al. (2005).

Data analysis

The in vitro recordings were low-pass filtered at 1.6 kHz and digitized at 5 kHz and analyzed using the Clampfit (Molecular Devices, Union City, CA, USA) and Strathclyde Electrophysiology WinWCP and WinEDR (John Dempster, Glasgow, UK) programs.

Extracellular events in slices were analysed as described before (Sipiläet al. 2004). Spontaneous network events (‘field GDPs’, fGDPs), were detected with an amplitude threshold set at a fixed level (∼25–100 μV) in each experiment. Field potential recordings were also used to examine spontaneous unit activity of intact neurons.

Spontaneous GABAergic postsynaptic currents (GABA-PSCs) were detected using Strathclyde Electrophysiology WinEDR program with a fixed amplitude threshold for each experiment (typically ∼6 pA). Events that appeared to consist of unitary GABA-PSCs were chosen for analysis of the decay time constant, which was obtained from a fit of a single exponential function to the averaged GABA-PSCs.

Unless otherwise stated, data are presented as mean ±s.d. Quantitative comparisons were based on Student's t test, and P-values < 0.05 were considered statistically significant.

Results

NKCC1 promotes early SPWs in vivo

In DC field potential recordings from the hippocampi of freely-moving rat pups, SPWs occurred as positive deflections with an amplitude of 104 ± 32 μV, duration of 320 ± 60 ms and a frequency of 0.24 ± 0.15 Hz (n= 6 pups, ∼120 events in each recording; Fig. 1A and B). Intraperitoneal application of the NKCC1-specific inhibitor, bumetanide (5 μmol kg−1; Isenring et al. 1998; Payne et al. 2003), blocked these events within 6–10 min in 6 out of 6 recordings (Fig. 1B). The pups were then returned to their mothers and a subsequent control recording at ∼3 h after bumetanide application showed a full recovery of the SPWs (102 ± 22 μV, 290 ± 70 ms, 0.28 ± 0.17 Hz, n= 6 pups, ∼120 events in each recording during recovery from the bumetanide dose; Fig. 1B). In DC recordings with Ag–AgCl electrodes (Fig. 1A), the shape of SPWs was different from those detected with polarizable metal electrodes (e.g. tungsten or stainless steel) that provide AC-coupling only (cf. Tallgren et al. 2005). When we high-pass filtered the DC-recordings at 0.5 Hz to mimic the bandwidth of conventional AC recordings, the SPWs had a biphasic shape (Fig. 1A), which is similar to previously published recordings of early SPWs (Leinekugel et al. 2002; Karlsson & Blumberg, 2003).

Figure 1.

Inhibition of NKCC1 blocks hippocampal SPWs in neonatal rats
A, a direct current (DC) recording (left) of a typical SPW in a P8 rat pup hippocampus in vivo and the same event after high-pass filtering at 0.5 Hz (right). B, intraperitoneal application of 5 μmol kg−1 bumetanide (bume) blocks early SPWs in a reversible manner (high-pass 0.5 Hz).

NKCC1 is required for the depolarizing action of GABA in immature CA3 pyramidal neurons

In a recent study, using whole-cell recordings in immature CA1 pyramidal neurons, Dzhala et al. (2005) reported a small negative shift in EGABA with bumetanide suggesting Cl accumulation via NKCC1.

However, the magnitude (∼3 mV) of the reported EGABA shift is likely to be much too small to explain the robust inhibitory effect of bumetanide on the network events. Given the role of immature CA3 pyramidal neurons in the initiation of GDPs and SPWs (see Introduction), we studied the influence of NKCC1 on EGABA in the immature CA3 pyramids in vitro using gramicidin-perforated patch recordings, which leaves the intracellular Cl concentration intact (Kyrozis & Reichling, 1995). The RMP of P2–4 CA3 pyramidal neurons was −53.3 ± 5.6 mV (n= 10 cells) in the presence of TTX (0.3–0.5 μm). The intracellular Cl concentration was calculated taking into account the bicarbonate permeability of GABAA receptors as described before (Kaila et al. 1993). The EGABA obtained by uncaging GABA at the soma was −44.4 ± 6.8 mV (n= 10 cells; Fig. 2AC) which corresponds to an intracellular Cl concentration of ∼21 mm and indicates a driving force (defined here as EGABA− RMP) of 9.0 ± 3.0 mV for the depolarizing GABAergic currents (Fig. 2D). In neurons that were incubated in 10 μm bumetanide (30–50 min) prior to the recordings, the mean EGABA (−61.9 ± 3.8 mV, P= 0.0004 versus control) was slightly hyperpolarizing (by −3.5 ± 3.4 mV, P= 0.00002 versus control, n= 4 cells; RMP −58.4 ± 2.2 mV, P= 0.11 versus control; Fig. 2AD) corresponding to an intracellular Cl concentration of ∼9 mm. These data demonstrate that Cl accumulation by NKCC1 generates a remarkably large positive shift in ECl (∼23 mV) resulting in a comparable shift in EGABA (∼18 mV) in the immature neurons.

Figure 2.

NKCC1 is required for the depolarizing action of GABA in immature CA3 pyramidal neurons
A, currents evoked by somatic GABA uncaging (indicated by the dots above the traces) at various membrane voltages (from −68 mV, 10 mV steps) in control and in bumetanide. B, peak GABAA receptor-mediated current versus holding potential from the recordings in A. C, resting membrane potential (RMP) and reversal potential of GABAA-mediated currents (EGABA) from individual recordings in control (n= 10) and in bumetanide (n= 4). D, driving force of GABAA-mediated currents (EGABA− RMP) from individual recordings (small circles) and the mean ±s.d. (large circles with error bars). Filled and open circles indicate data in control and in bumetanide, respectively, in B–D.

Distinct roles of NKCC1 in the generation of spontaneous activity of CA3 pyramidal neurons and interneurons

Next, we assessed the role of NKCC1 at the level of single-unit activity of intact CA3 pyramidal neurons in vitro under conditions where network events were blocked by NBQX and AP-5. The overall FP unit spike frequency was 0.66 ± 0.46 Hz (n= 4 recordings). Addition of 10 μm bumetanide increased transiently (for ∼5–10 min) the spike frequency to 161 ± 8% (n= 4; P= 0.0006), while a progressive decrease was seen thereafter to 49 ± 35% (P= 0.05, 15 min) and 32 ± 22% (P= 0.009, 30 min) of the control value (Fig. 3A upper traces and Fig. 3B). In order to examine whether the inhibitory effect of bumetanide on unit activity was dependent on endogenous GABAA receptor activation, a set of experiments was carried out in the presence of picrotoxin (and NBQX, AP-5). Since picrotoxin hyperpolarizes and, consequently, blocks the spontaneous activity of the immature pyramidal neurons, [K+]o was elevated (to 8–9 mm) to depolarize the neurons back to their burst-generating voltage window (see Sipiläet al. 2005). With GABAA receptors blocked, spike frequency (0.80 ± 0.53 Hz in control, n= 7 recordings) did not decrease even during prolonged (30 min) application of 20 μm bumetanide but a slight increase to 124 ± 11% was seen within ∼5 min (n= 7, P= 0.0007; Fig. 3A lower traces and Fig. 3B). The main conclusion from the above results is that, under physiological conditions, the spontaneous activity of individual immature CA3 pyramidal neurons is facilitated by NKCC1 and that this action is mediated by the depolarizing action of endogenous GABA.

Figure 3.

NKCC1 facilitates spontaneous activity of individual neonatal CA3 pyramidal cells
A, field potential recordings showing unit activity of CA3 pyramidal neurons and the effect of bumetanide (upper traces). Lower traces show unit activity in the presence of picrotoxin (PiTX) in 9 mm extracellular K+ and the effect of bumetanide. B, time course of the effect of bumetanide on the mean frequency (+s.e.m.) of unit activity from 4 and 7 experiments in the absence and presence of PiTX, respectively (bin 5 min). In all recordings, GDPs were blocked by NBQX and AP-5.

In order to study whether the endogenous tonic and phasic GABAA receptor-mediated conductances are influenced by bumetanide, we used whole-cell recordings (at 0 mV) to maintain the Cl equilibrium potential largely constant with a Cs+-based, low-Cl pipette solution (see Methods). In the presence of NBQX and AP-5, 10 μm bumetanide caused a minor increase in spontaneous GABA-PSC frequency (5.6 ± 5.2 Hz and 6.1 ± 5.1 Hz in control versus bumetanide; P= 0.04; n= 5 cells; Fig. 4A and B). Bumetanide had no significant effect on GABA-PSC peak amplitude (Fig. 4C), GABA-PSC decay time constant (26 ± 8 ms and 29 ± 10 ms in control versus bumetanide; n= 4 cells; Fig. 4D), baseline holding current (−0.3 ± 2.0 pA change by bumetanide, n= 4 cells) or input resistance of the neurons (430 ± 170 MΩ and 410 ± 190 MΩ in control versus bumetanide, n= 5 cells). The lack of effect on GABAA current amplitude indicates that 10 μm bumetanide did not affect the GABAA-receptor conductance or driving force in the whole-cell recordings. A lack of effect on driving force under whole-cell clamp is expected, especially as NKCC1 is mainly localized to the somata of immature CA3 pyramidal neurons (Marty et al. 2002) and the whole-cell pipette clamps the somatic Cl. Importantly, the minor increase in spontaneous GABA-PSC frequency indicates that interneuronal activity is rather inhibited than promoted by NKCC1.

Figure 4.

Bumetanide slightly increases the frequency of spontaneous postsynaptic GABAA currents
A, voltage-clamp recording at 0 mV with a low-chloride pipette filling solution shows a minor increase in spontaneous GABA-PSC frequency but no effect on the baseline holding current. B, a bar graph showing mean (+s.e.m.) spontaneous GABA-PSC frequency relative to control (n= 5 cells). Data obtained within 15–20 min after the beginning of bumetanide application were used for calculation of spontaneous GABA-PSC frequency. C, cumulative probability histogram of spontaneous GABA-PSC amplitudes from the recording in A (black line – control, grey line – bumetanide). D, averaged spontaneous GABA-PSCs from 40 single events in control and in bumetanide from the recording in A. In all recordings, GDPs were blocked by NBQX and AP-5.

NKCC1 facilitates fGDP occurrence in the presence and absence of phasic GABAA receptor-mediated transmission

Consistent with the findings of Dzhala et al. (2005), bath application of bumetanide (2–20 μm) blocked GDPs as seen in field potential recordings (fGDPs; n= 10 slices; Fig. 5Aa) and voltage-clamp recordings (n= 4 cells; Fig. 5Ab). A recovery of fGDPs was seen after washout of the drug (Fig. 5Aa).

Figure 5.

Bumetanide blocks spontaneous network events driven by CA3 pyramidal neurons in vitro in the presence and absence of interneuronal input
Aa, a field potential (FP) recording showing a block of field giant depolarizing potentials (fGDPs) by 10 μm bumetanide (band-pass 0.2–5 Hz; inset shows a DC recording of a single fGDP). b, voltage-clamp recording showing GDPs as bursts of postsynaptic GABAA receptor-mediated currents that are blocked by bumetanide. B, in the presence of the competitive GABAA receptor antagonist SR 95531, application of 32 μm isoguvacine (isog) increases fGDP frequency. Further addition of bumetanide blocks fGDPs (band-pass 0.2–5 Hz; inset shows a DC recording of a single fGDP). Note that SR 95531 completely blocks synaptic GABAA responses (interneuronal input) and a subsequent application of isoguvacine facilitates the tonic mode of GABAA activation (see Results).

It might be argued that the minor increase in GABA-PSC frequency (see Fig. 4A and B) could underlie the inhibitory effect of bumetanide on fGDPs and SPWs. A competitive GABAA-receptor antagonist, SR 95531 (gabazine), has a higher efficacy on GABA-PSCs (phasic interneuronal input) versus tonic GABAA conductance, a property that is attributable to the higher affinity of extrasynaptic versus synaptic GABAA receptors to GABA (see Stell & Mody, 2002). Hence, GABA-PSCs are completely blocked but the tonic GABAA current is only partially inhibited by a moderate concentration of SR 95531 and a subsequent addition of the GABAA agonist, isoguvacine, imposes a GABAergic current in a purely tonic manner (see Sipiläet al. 2005; Stell & Mody, 2002). As shown before (Sipiläet al. 2005), 3 μm SR 95531 reduced fGDP frequency (n= 4 slices). Subsequently, isoguvacine was applied with increasing concentrations (4–32 μm) until a steady increase in fGDP frequency was observed (n= 4 slices; Fig. 5B). A further addition of bumetanide blocked fGDPs in 4 out of 4 recordings in a reversible manner (Fig. 5B). Importantly, this shows that bumetanide blocks fGDPs when the interneurnal input to immature CA3 pyramidal neurons is blocked but a tonic GABAA current is present. Hence, this effect is fully attributable to NKCC1 expressed by pyramidal neurons, and the activity of interneurons is not a crucial mechanism by which NKCC1 promotes the early network events.

Discussion

The roles of ligand- and voltage-gated channels have been widely studied in the generation of CA3-driven network events in the immature hippocampus (see Introduction), while the contribution of ion-transport mechanisms has not been systematically investigated. However, such information is clearly needed in view of the dramatic developmental changes that take place in neuronal anion regulation during early postnatal development in the rat hippocampus (e.g. Payne et al. 2003; Rivera et al. 2005). The present work shows that SPWs are blocked by the specific inhibitor of NKCC1, bumetanide, in the neonatal rat hippocampus in vivo. Since this drug also blocks in vitro GDPs (see Fig. 5A and Dzhala et al. 2005), our in vivo findings strengthen the view that SPWs and GDPs are largely homologous network events (see Leinekugel et al. 2002). In particular, NKCC1 is responsible for a large (∼10 mV) depolarizing driving force for GABA in neonatal CA3 pyramidal neurons and, thereby, promotes the spontaneous burst activity of individual immature CA3 pyramids. Taken together, the present data indicate that inhibition of NKCC1 located in CA3 pyramidal neurons is a key mechanism by which bumetanide suppresses the network events.

In the immature CA3 pyramidal neurons, bumetanide caused a ∼13 mV and a ∼18 mV negative shift in the driving force and the reversal potential of GABAA receptor-mediated currents, respectively (Fig. 2). The difference between these two values is due to the more hyperpolarized RMP in the presence of bumetanide, which is expected on the basis of the tonic GABAA current that has a depolarizing action under control conditions (see Ben Ari et al. 1989; Sipiläet al. 2005).

Endogenous GABAA-receptor activation depolarizes immature CA3 pyramidal neurons (Ben Ari et al. 1989) and, hence, promotes their voltage-dependent intrinsic bursting (Sipiläet al. 2005). In the present work, bumetanide inhibited the spontaneous activity of the CA3 pyramids in the presence but not in the complete absence of endogenous ionotropic GABAergic signalling (Fig. 3). On the other hand, bumetanide caused only a slight increase in spontaneous GABA-PSC frequency without changing GABA-PSC kinetics or the tonic GABAA conductance (Fig. 4). Hence, the inhibitory effect of this drug on the spontaneous activity of single pyramidal neurons is fully explained by a reduction in the driving force of endogenous tonic and synaptic GABAA currents. The available data indicate that the following mechanisms govern the burst activity of immature CA3 pyramidal cells: NKCC1 accumulates Cl and makes GABAergic actions depolarizing. This GABAergic depolarization is often sufficient to activate a persistent Na+ current that further depolarizes the membrane to action potential threshold and leads to bursting. Ca2+ influx, caused by the bursts of spikes, activates a slow K+-mediated afterhyperpolarization that terminates the bursts and accounts for the subsequent refractory period which sets a lower limit for interburst intervals (Sipiläet al. 2006).

An important finding of the present study is that bumetanide blocked early SPWs in vivo (Fig. 1). In support of the idea that GDPs are the in vitro counterpart of SPWs (Leinekugel et al. 2002), bumetanide also blocks GDPs (Fig. 5A and Dzhala et al. 2005). Given the key role of the interconnected network of bursting CA3 pyramidal neurons in the generation of SPWs (Buzsaki, 1986; Suzuki & Smith, 1987) and GDPs (Sipiläet al. 2005), the inhibitory effect of bumetanide on these early network events is readily explained by its GABA-dependent inhibitory effect on the burst activity of the individual neonatal CA3 pyramidal neurons (Fig. 3). Our work also demonstrates that NKCC1 has a direct role in the spontaneous population activity of pyramidal neurons, which is independent of phasic (synaptic) interneuronal inputs. This conclusion is based on the experiments where GABA-PSCs were blocked by the competitive GABAA receptor antagonist SR 95531, and the tonic GABAergic current component and fGDP occurrence were enhanced by isoguvacine. Under these conditions, bumetanide blocks fGDPs (Fig. 5B) by reducing the driving force (Fig. 2) of the tonic GABAA conductance.

Inhibition of interneuronal activity is not likely to explain the blockade of SPWs or GDPs by bumetanide since the drug slightly enhanced rather than reduced the spontaneous GABA-PSC frequency (Fig. 4). Furthermore, bumetanide inhibited fGDPs in the complete absence of spontaneous GABA-PSCs (phasic GABA), with a tonic GABAA conductance only present (Fig. 5B). Finally, the interneuronal network does not generate recurrent network activity in the absence of glutamatergic transmission (see Fig. 4A and Leinekugel et al. 1998; Hollrigel et al. 1998; Bolea et al. 1999), which is consistent with the view that GDPs are paced by the network of bursting CA3 pyramidal neurons and that GABAergic signalling (tonic and phasic) has a temporally non-patterned permissive role (Sipiläet al. 2005).

The increase in spontaneous GABA-PSC frequency by bumetanide is not readily explained by a reduction in the driving force of GABAA currents in interneurons. Interestingly, the frequency of unit activity of pyramidal neurons was increased by bymetadine in the absence of GABAA receptormediated transmission, which indicates that this excitatory effect of the drug is not specific to interneurons. The increase in spontaneous activity caused by bumetanide was not studied further, but our data show that it is not dependent on ionotropic GABAergic transmission.

The present work show for the first time that CA3-driven SPWs are promoted by NKCC1 in the immature hippocampus in vivo. The in vitro experiments indicate that bumetanide acts by reducing the depolarizing driving force for GABAA currents. In conclusion, the present data imply that NKCC1-mediated depolarizing GABAergic signalling has a strong facilitatory action on in vivo SPWs in the immature hippocampus. Given that these events are thought to be involved in the activity-dependent development of the neuronal circuitry (cf. Katz & Crowley, 2002), our results suggest that the loop-diuretics, e.g. when used to treat heart failures of newborn human babies, might have unwanted side-effects on the proper development of the brain.

Appendix

Acknowledgements

This work was supported by the Academy of Finland and by the Sigrid Juselius Foundation.

Ancillary