Synaptically induced sodium signals in hippocampal astrocytes in situ

Tissue preparation and identification of astrocytes

Experiments were performed on acute tissue slices (250 μm) of mouse hippocampus (Mus musculus, Balb/C; postnatal days 12–21). Animals were decapitated following anaesthesia with CO₂ and their brains were rapidly removed. Transverse hippocampal slices were prepared as described previously (Edwards et al. 1989). Sectioning was performed in standard saline at 4–6°C. The standard saline was composed of (in mm): 125 NaCl, 2.5 or 4 KCl, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃ and 20 glucose, bubbled with 95% O₂ and 5% CO₂, resulting in a pH of 7.4. After sectioning, slices were kept at 34°C for 20 min in standard saline that contained 0.5–1 μm sulforhodamine 101 (SR101), followed by a 10 min incubation in saline without SR101 at 34°C. This procedure results in the selective staining of passive astrocytes as described earlier (Nimmerjahn et al. 2004; Kafitz et al. 2008). Afterwards, slices were kept in standard saline at room temperature (19–22°C) until they were used for experiments, which were also performed at room temperature.

Electrical stimulation of Schaffer collaterals was performed by square pulses (150 μs duration) delivered at 50 Hz via a thin borosilicate glass pipette (Hilgenberg, Waldkappel, Germany) filled with saline (pipette resistance: 2–4 MΩ). The stimulation pipette was placed at a distance of 50–200 μm from the somata of selected SR101-positive astrocytes in the stratum radiatum of the CA1 area, unless stated otherwise. d-Aspartate was applied by a pressure application device (PDES-02D, NPI Electronic GmbH, Tamm, Germany) coupled to standard micropipettes (Hilgenberg, Waldkappel, Germany) placed 20–100 μm from cell bodies of selected cells in the presence of 0.5 μm tetrodotoxin. All other substances were applied with the bath perfusion system. Chemicals were purchased from Sigma, except for tetrodotoxin (Alomone Labs, Jerusalem, Israel or Biotrend Chemicals, Cologne, Germany) and dl-threo-ß-benzyloxyaspartatic acid (TBOA; Tocris/BIOZOL Diagnostica Vertrieb GmbH, Eching, Germany).

Electrophysiology

Standard whole-cell recordings were obtained from the somata of SR101-positive astrocytes with an EPC10 amplifier (HEKA Elektronik, Lambrecht, Germany); PatchMaster software (HEKA Elektronik, Lambrecht, Germany) was used for data acquisition. The pipette solution contained (in mm): 120 KMeSO₃, 32 KCl, 10 N-(2-hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid (Hepes), 4 NaCl, 4 Mg-ATP, 0.4 Na₃-GTP and 0.5–1 sodium-binding benzofuran isophthalate (SBFI); the pH was adjusted to pH 7.3. Astrocytes were generally held at a membrane potential of −85 mV; liquid junction potential was not corrected. To ensure sufficient diffusion of SBFI into distal branches, cells were filled for at least 20 min before starting the sodium imaging.

Na⁺ imaging and calibration of SBFI fluorescence

Conventional, wide-field fluorescence imaging was performed using a variable scan digital imaging system (TILL Photonics, Martinsried, Germany) attached to an upright microscope (Olympus BX51Wi, 40×/60× water immersion objective, NA 0.80/0.90, Olympus Europe, Hamburg, Germany) and a CCD camera as sensor (TILL Imago VGA, TILL Photonics, Martinsried, Germany). Cells were dye-loaded by bolus injection of the membrane-permeable form of SBFI (SBFI-AM (the acetoxymethyl ester); Molecular Probes/Invitrogen, Karlsruhe, Germany) into the CA1 region of the hippocampus as described earlier (Meier et al. 2006). When performing patch-clamp experiments, individual cells were loaded via the patch pipette that contained 0.5 or 1 mm of the membrane-impermeable form of SBFI.

In situ calibration of intracellular SBFI fluorescence in bolus-loaded cells was performed as described earlier (Meier et al. 2006). Following loading with SBFI-AM, slices were perfused with calibration solutions containing different sodium concentrations as well as ionophores (3 μm gramicidin, 10 μm monensin) to equilibrate extra- and intracellular sodium concentrations. Excitation spectra, recorded from astrocyte somata in bolus-loaded slices at sodium concentrations of 0 mm and 30 mm, indicated an isosbestic point around or slightly below 340 nm, while at longer wavelengths, fluorescence emission decreased with increasing sodium concentration (n= 9; Fig. 1B). This ‘in situ’ spectrum is very similar to those obtained by others (e.g. Borzak et al. 1992) and confirms earlier reports showing that the spectral properties of SBFI inside cells differ from those in vitro (Harootunian et al. 1989; Rose & Ransom, 1996a; Diarra et al. 2001; Meier et al. 2006). Proper determination of the isosbestic point in the ‘in situ’ condition was not feasible because the performance of our imaging equipment (optical parts, lenses, monochromator) quickly drops below 340 nm.

To validate the contribution of autofluorescence, that is the fluorescence of unstained tissue, to the fluorescence generated in SBFI-loaded slices, we obtained excitation spectra in slices before and after loading with SBFI. These spectra revealed that at excitation wavelengths between 330 and 390 nm, autofluorescence contributed only about 10% to the fluorescence signal obtained from the stratum radiatum of SBFI-loaded slices.

Based on this, we choose 345 nm, at which SBFI emission is only weakly sodium sensitive, and 385 nm, at which SBFI is strongly sodium sensitive, as alternate excitation wavelengths for ratio calculation (see also Fig. 2B).

For wide-field imaging with SBFI, fluorescence emission (>440 nm) was collected at 3–4 Hz from defined regions of interest (ROI) containing cellular structures following alternate excitation at 345 nm and at 385 nm. After background correction, the fluorescence ratio (F₃₄₅/F₃₈₅) was calculated for individual ROIs and analysed off-line by employing OriginPro Software (OriginLab Corporation, Northhampton, MA, USA).

Standard background correction was performed as described earlier (Garaschuk et al. 1998). In brief, for each excitation wavelength (345 and 385 nm), fluorescence emission was collected from a region within the field of view which was free of clearly visible SR101-positive cellular processes (cf. region marked as ‘bg’ in Fig. 2A and B) and the determined background values were then subtracted from the fluorescence intensity of the ROIs for each frame. Background fluorescence is generated by both cellular autofluorescence (see above) and extracellular dye. Thus, background fluorescence is significantly larger than autofluorescence. Only cells with fluorescence values that were at least 30% higher than the background fluorescence were included in the study.

When bolus-loaded slices were subjected to Na⁺-free calibration saline containing ionophores (see above), ratio levels in the ROIs dropped rapidly, reaching a new steady state within about 20 min (n= 6 neurons, n= 51 astrocytes; not shown), as reported before (Rose & Ransom, 1996a; Meier et al. 2006; Bennay et al. 2008). Subsequent stepwise changes in the sodium concentration of the calibration saline then caused stepwise changes in the SFBI ratio, indicating stepwise changes in the intracellular sodium concentration (n= 44 astrocytes; Fig. 1C). When normalized to the fluorescence ratio obtained in sodium-free saline, the relationship between changes in SBFI fluorescence and intracellular sodium concentration was virtually linear for sodium concentrations between 10 and 30 mm (n= 44; Fig. 1D). Within this range, a change in the fluorescence ratio by 10% corresponded to a change in the intracellular sodium concentration by 7.1 mm.

Baseline intracellular sodium was determined by a direct comparison between the fluorescence ratios of single ROIs in physiological saline with that following perfusion with calibration saline at different sodium concentrations (Fig. 1D). Activity-induced changes in SBFI ratio, ranging from 0.4% to 33%, were generally expressed as changes relative to baseline in per cent. Because of the linearity of the ratio signal between 10 and 30 mm sodium, ratio changes in per cent were then multiplied by a factor of 0.71 (see above) to reflect absolute changes in the sodium concentration. Calibration parameters obtained with the bolus-loading technique were also used for quantification of SBFI signals obtained during patch-clamp experiments, because earlier studies in our laboratory have shown that, for a given imaging set-up, calibration parameters of intracellular SBFI fluorescence are comparable for bolus-loading of SBFI-AM and loading of SBFI through a patch pipette (Meier et al. 2006; Bennay et al. 2008).

Unless otherwise specified, data are presented as means ±s.e.m. and were statistically analysed by Student's paired or unpaired t tests, as appropriate. If not stated otherwise, n represents the number of analysed cells. Each set of experiment was performed on at least three different slice preparations.

Synaptically induced sodium transients in neurons and astrocytes

In situ calibration experiments (see Methods; Fig. 1D) revealed that baseline intracellular sodium concentration was 12.0 ± 1.1 mm (n= 6) in somata of CA1 pyramidal neurons and 10.8 ± 1.2 mm in SR101-positive astrocytes (n= 8), which is in good agreement with values reported earlier (e.g. Rose & Ransom, 1996a; Chatton et al. 2001; Sheldon et al. 2004). To evoke synaptic activity, afferent Schaffer collaterals were stimulated by current pulses delivered at 50 Hz through a fine micropipette positioned in the stratum radiatum. Short-burst stimulation of Schaffer collaterals (50 Hz, 200 ms) induced a significant transient reduction in fluorescence emission at 385 nm in both neurons and astrocytes, while fluorescence emission at 345 nm only slightly decreased (Fig. 2A and B). Fluorescence in background regions was virtually unchanged (Fig. 2A and B). Thus, synaptic activity caused a transient increase in the SBFI fluorescence ratio (F_345/F₃₈₅), indicating an increase in the sodium concentration in the somata of pyramidal neurons as well as of SR101-positive astrocytes (Fig. 2C).

The amplitude of sodium transients detected from ROIs positioned over neuronal somata ranged from 0.9 to 23.6 mm (average 5.1 ± 0.3 mm; n= 223). Peak sodium concentrations were reached only 5–20 s after the stimulation (Fig. 2C). Sodium transients detected from ROIs positioned over the somata of SR101-positive astrocytes ranged from 0.6 to 7.8 mm (average 2.1 ± 0.1 mm; n= 122) and peak concentrations were reached within 0.6–10 s (Fig. 2C).

In cultured astrocytes, intercellular waves of sodium, propagating with a speed of about 15 μm s⁻¹, can be evoked in response to mechanical or direct electrical stimulation of single cells (Bernardinelli et al. 2004).

In the present study, stimulation pipettes were generally positioned 50–200 μm from the somata of SR101-positive, SBFI-loaded astrocytes. Under these conditions and with the given imaging frequency of 3–4 Hz, we did not observe sodium signals propagating in a wave-like manner from an individual astrocyte to its neighbours within the field of view (222 μm × 168 μm), indicating simultaneous activation of these astrocytes in response to synaptic stimulation. To reduce the number of activated cells in the field of view, we positioned the stimulation pipette close (∼20 μm) to SBFI-loaded astrocyte cell bodies. Under these conditions, detectable synaptically induced sodium signals were restricted to astrocytes located within a radius of about 50 μm around the stimulation pipette (n= 6 slices, Fig. 3A). While the amplitude of these sodium transients typically decreased with increasing distance from the stimulation pipette, no difference in their delay or rise time, indicative of a wave-like propagation from cell to cell, was visible (Fig. 3A). Thus, these data suggest that the observed sodium signals were again primarily caused by simultaneous synaptic activation of the involved astrocytes. To selectively induce a sodium elevation in a single astrocyte, we then placed the stimulation pipette in direct physical contact with their soma. Electrical stimulation resulted in an immediate, strong and long-lasting increase in the intracellular sodium concentration in this cell (n= 8; Fig. 3B). In addition, we observed sodium elevations in neighbouring astrocytes within a radius of about 40–50 μm. Both the peak amplitudes and the slopes of the rising phase of sodium elevations in surrounding cells typically decreased with increasing distance from the stimulated cell, suggesting propagation in a wave-like manner.

Specificity of cellular sodium signals

In a widefield fluorescence imaging system, fluorescence emission is generated (and collected) not only in the entire plane of focus, but also above and below. In addition, light scattering causes significant distortion of stained structures, and small astrocyte processes and neuronal dendrites cannot be resolved. As our bolus-loading technique stains both astrocytes and neurons at the same time, we carried out several sets of experiments to ensure that fluorescence signals obtained from ROIs positioned over cellular somata indeed primarily reflect intracellular signals from the chosen cell without significant distortions by fluorescence emission from out-of-focus structures.

To evaluate a possible crosstalk between neuronal and astrocyte SBFI fluorescence signals in bolus-loaded slices, we applied d-aspartate. d-Aspartate is a non-metabolized substrate of sodium-dependent glutamate uptake which is also transported (Danbolt, 2001). Its application thus results in sodium transients and inward currents in astrocytes if functional transporters are expressed (Rose & Ransom, 1996b; Chatton et al. 2001; Bennay et al. 2008; Kelly et al. 2009). Moreover, d-aspartate is an agonist of NMDA receptors, the activation of which causes sodium transients in CA1 pyramidal neurons (Rose & Konnerth, 2001). Pressure application of d-aspartate (1 mm) evoked transient increases in the SBFI fluorescence ratio in ROIs positioned over neuronal and astrocyte cell bodies, indicating intracellular sodium transients in both cell types (Fig. 4A). For a puff duration of 400 ms, these transients amounted to 13.4 ± 0.9 mm in neurons (n= 79) and 4.8 ± 0.3 mm in astrocytes (n= 95). d-Aspartate-induced sodium transients could be evoked repetitively in both cell types, while their amplitude decayed by 10–30% with five applications within 30 min (n= 15 neurons and 14 astrocytes; Fig. 4A). When APV (dl-2-amino-5-phosphonopentanoic acid, 50 μm), a selective blocker of NMDA receptors was applied, neuronal sodium transients were strongly reduced (reduction by 88.8 ± 3.2%; n= 10; Fig. 4B). At the same time, APV only slightly reduced the amplitude of glial sodium transients (reduction by 16.8 ± 3.6%; n= 13; Fig. 4B). While this could indicate some overlay of glial with dendritic signals under control conditions, this apparent effect of APV on astrocytes might also – at least partly – be explained by the observed slight rundown of the response under control conditions. Furthermore, it could arise from activation of NMDA receptors on astrocytes themselves, as recently described (Serrano et al. 2008).

To further test the cell type specificity of fluorescence signals, we performed a bolus injection of SBFI-AM into the stratum oriens. This resulted in the loading of somata of CA1 pyramidal neurons of the stratum pyramidale as well as their apical dendrites in the stratum radiatum, whereas astrocytes in the stratum radiatum did not visibly load with SBFI. Upon application of d-aspartate, SBFI-stained neuronal cell bodies showed transient changes in SBFI fluorescence emission, indicating an increase in sodium concentration. Fluorescence emission from the somata of SR101-stained astrocytes in the stratum radiatum, in contrast, was not different from background and did not change in response to stimulation (n= 11; not shown). This indicates that fluorescence emission from apical CA1 pyramidal cell dendrites located above or below the plane of focus of astrocyte somata had no detectable influence on glial fluorescence emission under these conditions.

We also carried out widefield imaging experiments on astrocytes individually loaded with SBFI. To this end, the membrane-impermeable form of SBFI was loaded into SR101-positive astrocytes through a patch pipette in the whole-cell patch-clamp mode. Twenty to thirty minutes after breakthrough, SBFI fluorescence was visible in the entire cell, including primary and secondary processes as well as fine processes arising from them, generating out-of-focus light of a cloud-like appearance (Figs 5A and 6A). As reported before, burst stimulation of Schaffer collaterals (50 Hz, 200 ms) induced long-lasting inward currents in astrocytes (n= 12; Fig. 5B), that largely reflect the activation of electrogenic glutamate transport (Brew & Attwell, 1987; Bergles & Jahr, 1997; Diamond et al. 1998; Luscher et al. 1998). Inward currents were accompanied by transient changes in SBFI emission in astrocyte somata, indicating sodium increases ranging from 0.4 to 4.3 mm (average: 1.4 ± 0.3; n= 12; see ROI 3 (r3) in Fig. 5B), which is similar to values obtained from bolus-loaded slices (see above). Fluorescence emission (at excitation with 345 and 385 nm) from surrounding SR101-positive astrocytes (Fig. 5C: ROIs r8, r9, r10) as well as from background regions (Fig. 5C: bg) did not change, indicating the absence of activity-dependent changes in autofluorescence. No significant differences in the time course and kinetics of sodium transients detected from astrocyte somata were observed when comparing signals from bolus-loaded and pipette-loaded astrocytes.

Taken together, these experiments demonstrate that synaptic activity generates sodium signals in astrocytes. Furthermore, they show that under our experimental conditions, fluorescence signals from ROIs positioned over CA1 pyramidal cell bodies in the stratum pyramidale and astrocyte cell bodies in the stratum radiatum primarily represent sodium signals from these cell types in bolus-loaded slices (cf. also later experiments and Figs 8B and 9Aa).

Sodium transients in cellular subdomains of astrocytes

Loading of astrocytes through a patch pipette enabled the visualization not only of astrocyte somata, but also of their primary and secondary processes and surrounding structures with a low background. Using this loading technique, we asked whether sodium transients differ in different cellular domains by analysing the amplitude and time course of fluorescence signals from the soma, and from several primary processes together with the diffuse fluorescence surrounding them, individually (Figs 5A and 6A).

We found large differences in the amplitude and kinetics of activity-induced sodium transients between the different cellular domains in single astrocytes (n= 12; Fig. 5B). The sodium signals were largest in one to two cellular domains that represented a primary branch with its fine processes (r1 and r2, Fig. 5). Maximal sodium transients were characterized by a fast rise time and a defined peak, which reached up to 6.0 mm. In other cellular domains (r4–r7, Fig. 5), the amplitudes of the sodium transients were smaller and their peaks only reached many seconds later (see arrows, Fig. 5B). In the soma (r3, Fig. 5), intermediate sodium signals in terms of amplitude and time to peak were recorded. We also found that, with low stimulation intensities, sodium signals were largely restricted to a primary branch and its associated fine processes and only weakly invaded the soma and other cellular domains (compare r1 and r2–7, Fig. 5).

Differences in amplitude, rise time and decay time of sodium transients between different cellular domains of astrocytes were also observed upon selective activation of electrogenic glutamate uptake. Local pressure application of d-aspartate (1 mm) evoked a somatic inward current in whole-cell patch-clamped, SBFI-loaded astrocytes (n= 30 applications, 8 cells; Fig. 6B). The average peak amplitude of this inward current was dependent on the duration of the d-aspartate application and amounted to 1.0 ± 0.4 nA when d-aspartate was applied for 200 ms (n= 13 applications, 8 cells). Amplitude and time course of evoked sodium transients depended not only on the duration of the application, but also on the spatial relationship between application pipette and analysed region of interest. We consistently found that sodium signals were largest and their peaks reached fastest in cellular domains located close to the application pipette (n= 8 cells, Fig. 6B). Again, no signals were detectable in surrounding SR101-positive, SBFI-unstained astrocytes nor in background regions (Fig. 6C).

Pharmacology of synaptically induced sodium transients

Our results obtained so far demonstrate that Schaffer collateral stimulation evokes sodium transients in the millimolar range in CA1 pyramidal neurons and SR101-positive astrocytes. We next characterized the basic pharmacological properties of these activity-induced sodium transients in bolus-loaded slices to elucidate the mechanisms of their generation. To this end, we first established that sodium signals can be evoked repetitively in neurons and astrocytes in the course of an experiment (n= 37 neurons and 32 astrocytes; Fig. 7A). Increasing the stimulation intensity increased the amplitude of synaptically induced sodium transients in both cell types (n= 53 neurons and 25 astrocytes; Fig. 7B; cf. Fig. 5).

Both neuronal and glial sodium transients induced by synaptic stimulation were blocked during perfusion with 0.5–1 μm tetrodotoxin, confirming their dependence on action potential generation (n= 42 neurons, 14 astrocytes; Fig. 8A). During perfusion with bicuculline methiodide (20 μm), a specific antagonist of GABA_A receptors, the amplitudes of synaptically evoked neuronal and glial signals were increased to about 2- and 1.3-fold in neurons (n= 42) and astrocytes (n= 14), respectively (Fig. 8A). Astrocytes express GABA_A receptors and their activation depolarizes the cells (e.g. Meier et al. 2008), thereby reducing the driving force for sodium-dependent glutamate uptake. Blocking this depolarization by bicuculline might thus result in increased activation of glutamate uptake and sodium influx. In addition, bicuculline methiodide might increase neuronal excitability due to its effect on calcium-activated potassium currents (Johansson et al. 2001).

Earlier work has established that suprathreshold short-burst Schaffer collateral stimulation induces sodium transients in dendrites and spines of CA1 pyramidal neurons that are dependent on AMPA receptor activation (Rose & Konnerth, 2001). While hippocampal astrocytes are characterized by a strong expression of sodium-dependent glutamate transport (Matthias et al. 2003), there is also some evidence for expression of AMPA receptors (Zhou & Kimelberg, 2001). We hence tested the influence of AMPA receptor activation on synaptically induced sodium signals in neurons and astrocytes. In CA1 pyramidal neurons, inhibition of AMPA receptors by 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 μm) diminished sodium transients by 76.4 ± 3.7% (n= 31; Fig. 8B). At the same time, the amplitude of sodium transients in astrocytes was reduced by only 17.5 ± 5.9% (n= 16; Fig. 8B).

To study the involvement of sodium-dependent glutamate transport, we applied the non-transported, competitive glutamate-uptake blocker TBOA (200 μm). We found that, within minutes after application of TBOA, neuronal sodium concentrations began to rise to values of about 30 mm (n= 19; Fig. 9Aa). Synaptic stimulation resulted in further persistent sodium increases in only part of the neurons (n= 10; Fig. 9Aa). TBOA-induced neuronal sodium loads did not recover within 60 min after removal of TBOA from the saline (not shown). In contrast, TBOA did not alter the steady-state intracellular sodium concentration of astrocytes (n= 25; Fig. 9Aa). Schaffer collateral stimulation only triggered a minor increase in the sodium concentration in astrocytes in the presence of TBOA (n= 20; Fig. 9Aa). Compared to control, the amplitude of synaptically induced glial sodium transients was reduced by 93.2 ± 3.2% (n= 20; Fig. 9Aa). It recovered to about 65% of control values upon removal of TBOA (n= 15; Fig. 9Aa).

The persistent neuronal sodium elevation in the presence of TBOA was probably the consequence of an activation of ionotropic glutamate receptors following failure of glutamate uptake and accumulation of extracellular glutamate, as shown earlier (Rothstein et al. 1996; Jabaudon et al. 1999). To prevent sodium influx through ionotropic glutamate receptors, the experiment was repeated in the presence of the receptor blockers CNQX (10 μm) and APV (50 μm). Under these conditions, addition of TBOA did not significantly influence baseline intracellular sodium of neurons (n= 38 neurons in 6 slices).

As reported before (Rose & Konnerth, 2001); see also Fig. 8B), postsynaptic neuronal sodium transients were nearly completely diminished during combined application of both receptor blockers (reduction by 88.3 ± 1.7%; n= 42; not shown). In the presence of CNQX and APV, the amplitude of glial sodium signals was reduced by 35.8 ± 6.9% (n= 24; Fig. 9Ab). After addition of TBOA, stimulation-induced sodium transients in astrocytes were reduced by 86.8 ± 2.8% as compared to control without blockers (n= 24, Fig. 9Ab). In contrast to the recovery seen in experiments in which just TBOA was used, activity-induced sodium transients did not recover within a period of 60 min following removal of the three blockers.

To further establish the influence of an activation of electrogenic glutamate transport on sodium concentration in astrocytes in situ, we repeated experiments with puff application of d-aspartate. In the sole presence of TBOA, the amplitude of d-aspartate-induced sodium transients in astrocytes was reduced by 78.0 ± 4.3% (n= 12, Fig. 9Ba), confirming that they were largely mediated by activation of glutamate transport (cf. Fig. 4B). Following removal of TBOA, transients recovered to about 90% of control values (n= 12; Fig. 9Ba). In the presence of APV and CNQX, the amplitude of d-aspartate-induced sodium transients was reduced by 10.4 ± 3.2% (n= 37; Fig. 9Bb), which is not different from the reduction seen during APV alone reported above (see Fig. 4B). After addition of TBOA, signals were reduced by 82.0 ± 3.2% of control values without blockers (n= 24, Fig. 9Bb). As observed with synaptic stimulation, we did not obtain a recovery of the signals within 60 min after washout of the three blockers.

Mechanisms of activity-induced neuronal sodium transients

Our data confirm earlier work which demonstrated that suprathreshold short-burst Schaffer collateral stimulation induces detectable sodium transients in CA1 pyramidal neurons (Rose & Konnerth, 2001). Their study found that while these synaptically induced neuronal sodium transients are mainly caused by influx of sodium through NMDA receptor channels, they are strongly sensitive to CNQX, a blocker of ionotropic AMPA receptor channels. This is because AMPA receptor activation by synaptically released glutamate is necessary to provide sufficient depolarisation to relieve the magnesium block of NMDA receptors (Rose & Konnerth, 2001). In the present study, we observed a reduction of neuronal sodium transients by about 80% in the presence of CNQX, again confirming the earlier data.

In apical CA1 pyramidal cell dendrites, sodium transients determined from dendritic regions of interest close to the stimulation pipette reached an amplitude of about 10 mm and peaked within 1–2 s following short-burst stimulation of Schaffer collaterals (5 pulses at 50 Hz; Rose & Konnerth, 2001). The peak amplitude of these dendritic sodium transients decayed monoexponentially with increasing distance from the region of maximal response and their rise time was prolonged. In the present study, the amplitude of sodium transients detected in neuronal somata averaged 5 mm with 10 stimulation pulses and peak concentrations were reached only 5–20 s after the stimulation. Based on the earlier observations mentioned above (Rose & Konnerth, 2001), this indicates that somatic sodium signals might be largely mediated by diffusion of sodium from the apical dendritic tree and active spines following activation of postsynaptic cells.

The present study shows that TBOA, a widely used competitive transport blocker of sodium-dependent glutamate transport, caused a strong and persistent rise in the intracellular sodium concentration of CA1 pyramidal neurons, emphasizing the vital role of glutamate uptake for glutamate clearance and neuronal function (Maragakis & Rothstein, 2001; Marcaggi & Attwell, 2004; Schousboe et al. 2004; Seifert et al. 2006). In line with this, earlier work has established that addition of TBOA causes glutamate accumulation and activation of ionotropic glutamate receptors in the intact tissue (Rothstein et al. 1996; Jabaudon et al. 1999). This is supported by the fact that blocking ionotropic receptors prevented the strong rise in intracellular sodium evoked by TBOA in the present study.

Activity-induced sodium transients in astrocytes

Our experiments demonstrate that glutamatergic synaptic activity induces sodium transients in astrocytes in a hippocampal slice preparation. Only recently, two reports established the occurrence of activity-induced sodium signals in cell bodies (Kirischuk et al. 2007) and in processes (Bennay et al. 2008) of Bergmann glial cells of the cerebellum. Synaptically induced sodium transients in hippocampal astrocytes reached amplitudes of up to 8 mm, a value which is similar to that determined in Bergmann glial processes upon short-burst stimulation (Bennay et al. 2008).

The time course of intracellular sodium transients as compared to calcium transients is surprisingly slow, a phenomenon observed and described by several other authors (e.g. Kirischuk et al. 1997; Regehr, 1997). The present study shows that sodium signals in astrocytes are strongly determined by the spatial relationship between the region of influx and the area where measurements are taken (see Figs 5 and 6). A similar dependence was described for synaptically induced sodium transients in CA1 pyramidal cell dendrites and dendritic spines (Rose & Konnerth, 2001). In addition, time constants of activity-induced intracellular sodium signals are dependent on the temperature (Rose et al. 1999; Bennay et al. 2008). In Bergmann glial cells, a Q₁₀ of about 1.7 was determined for rise and decay times, while the peak amplitudes of sodium signals were similar at room temperature and at 32°C (Bennay et al. 2008). This indicates the dominating role of membrane transport in their generation and recovery and suggests that, at physiological temperature, sodium signals in astrocytes will be accelerated by a factor of about 2.

Earlier work had shown that bath application of glutamate evokes an elevation of sodium in cultured astrocytes (Kimelberg et al. 1989; Rose & Ransom, 1996b; Chatton et al. 2000). In such cultures, intracellular sodium concentrations are equalized between astrocytes by diffusion of sodium through gap junctions (Rose & Ransom, 1997) and intercellular waves of sodium, propagating with a speed of 15 μm s⁻¹, can be evoked in response to mechanical or direct electrical stimulation of single cells (Bernardinelli et al. 2004). It has been proposed that these sodium waves induce metabolic waves by increasing glucose uptake within the astrocytic network (Bernardinelli et al. 2004). In the present study, we could not detect differences in the rise time of astrocytic sodium signals upon Schaffer collateral stimulation, indicating simultaneous activation of these astrocytes in response to synaptic stimulation. In contrast, when a single astrocyte was directly stimulated by current injection through a pipette positioned in physical contact to its soma, we not only observed a rapid increase in the intracellular sodium concentration in the stimulated cell, but also sodium elevations in neighbouring astrocytes. In this case, the slope of the rising phase in surrounding cells decreased with increasing distance from the stimulated cell, a behaviour similar to that described from intercellular sodium waves in culture (Bernardinelli et al. 2004). This suggests that, in the intact tissue, sodium can indeed travel in a wave-like manner within the astrocytic syncytium from regions of high to regions of low activity.

When single astrocytes were dye-filled through a patch-pipette, we could detect differences in both amplitude and time course of sodium signals induced by Schaffer collateral stimulation as well as following focal application of d-aspartate between different cellular regions of astrocytes represented by the somata and the main primary branches and their adjacent fine processes. For synaptic stimulation, these differences were most evident with low stimulation intensities. Here, sodium signals could be largely restricted to single domains, with only weak invasion of the cell body, indicating that these regions were close to synaptic release sites for glutamate, and the main regions of sodium influx. Similar observations were reported from branches of Bergmann glial cells upon local stimulation of parallel fibres (Bennay et al. 2008). These results suggest that in cellular domains close to activated synapses, sodium-dependent processes could be regulated locally and independent from other cellular regions.

In cultured astrocytes, sodium transients upon application of glutamate are primarily generated by activation of glutamate uptake (Rose & Ransom, 1996b; Chatton et al. 2000, 2001). Pharmacological inhibition with TBOA results in a reduction of these signals by 70–80% (Chatton et al. 2001). In addition, glutamate-induced sodium signals in cultured astrocytes are partly sensitive to CNQX, suggesting that AMPA receptor activation also plays a role (Rose & Ransom, 1996b; Chatton et al. 2000, 2001). Our data show a similar pharmacological profile of sodium transients evoked by synaptic stimulation in hippocampal astrocytes in acute tissue slices. Synaptically induced sodium signals were partly reduced in the presence of CNQX, indicating that activation of ionotropic AMPA receptors following synaptic release of glutamate is also a pathway for sodium influx into astrocytes in situ. Application of TBOA to the slice preparation nearly completely blocked synaptically induced sodium signals in astrocytes, while it did not influence baseline sodium concentrations. This suggests that sodium-dependent glutamate uptake mediates the majority of sodium influx into astrocytes in situ in response to Schaffer collateral stimulation.

These conclusions are supported by the experiments in which TBOA was added in the presence of the ionotropic glutamate receptor blockers CNQX and APV. Again, TBOA largely blocked glial sodium transients. While we could not obtain a recovery under these conditions, earlier experiments had demonstrated that synaptically induced sodium signals can be evoked repetitively if no blockers are present (Fig. 7), and can – at least partly – recover after application and washout of TBOA alone (Fig. 8Ba). A similar failure of recovery was observed upon application of d-aspartate under these conditions: signals induced by d-aspartate could be evoked repetitively in the absence of blockers (Fig. 4A), and showed a nearly complete recovery after washout of either APV (Fig. 4B) or TBOA (Fig. 9Ba) alone, while no effect of washout was obtained after combined application of the blockers. Thus, the failure of recovery was most probably caused by a failure of sufficient washout of the blockers from the slice tissue.

Taken together, our results suggest that sodium-dependent glutamate uptake mediates the majority of sodium influx into astrocytes in situ in response to Schaffer collateral stimulation. Furthermore, they indicate that activation of AMPA receptors is an additional relevant factor contributing to sodium signals in astrocytes in the intact tissue.

Possible consequences of intracellular sodium transients in astrocytes

The inwardly directed sodium gradient energizes many transport processes across the plasma membrane, and increases in the intracellular sodium concentration reduce the driving force for these transporters. Thus, sodium transients influence the recovery of calcium transients mediated by Na⁺/Ca²⁺ exchange (Kirischuk et al. 1997; Golovina et al. 2003). Furthermore, intracellular sodium transients will influence sodium-dependent reuptake of transmitters. Even minor elevations in the sodium concentration will reduce the driving force for sodium-dependent GABA uptake, and might promote the release of GABA by transport reversal (Wu et al. 2007). As glutamate transport is coupled to the inward transport of three sodium ions, the electrochemical sodium gradient provides its major driving force (Nicholls & Attwell, 1990; Levy et al. 1998; Danbolt, 2001; Maragakis & Rothstein, 2001). Together with a membrane depolarization induced by synaptic stimulation, sodium transients will significantly reduce the activity of glutamate uptake (Zerangue & Kavanaugh, 1996). In addition, a recent study has provided evidence that intracellular sodium elevations following prolonged activation of the glutamate transporter GLT-1 trigger a clustering and a decrease of the surface expression of this transporter (Nakagawa et al. 2008).

Another well-documented consequence of glial sodium elevations is increased ATP hydrolysis by the Na⁺/K⁺-ATPase. In cultured astrocytes, this has been suggested to promote aerobic glycolysis, glucose uptake and glycogen breakdown (Pellerin & Magistretti, 1994; Chatton et al. 2000; Voutsinos-Porche et al. 2003; Magistretti & Chatton, 2005). Interestingly, activation of GABA uptake, in contrast to glutamate, is not associated with significant sodium loads, nor with a measurable metabolic response (Chatton et al. 2003). Recent studies have demonstrated that the stimulation of GLUT-1-mediated glucose uptake by glutamate requires the concurrence of both sodium and calcium signals in astrocytes (Loaiza et al. 2003; Porras et al. 2008). Our finding that long-lasting sodium signals are present in astrocytes following neuronal release of glutamate supports the proposed tight link between excitatory neuronal activity, glutamate transport, sodium signals and glucose utilization by astrocytes (Pellerin & Magistretti, 1994; Tsacopoulos & Magistretti, 1996).