Activation of glutamate transport evokes rapid glutamine release from perisynaptic astrocytes

Abstract

Stimulation of astrocytes by neuronal activity and the subsequent release of neuromodulators is thought to be an important regulator of synaptic communication. In this study we show that astrocytes juxtaposed to the glutamatergic calyx of Held synapse in the rat medial nucleus of the trapezoid body (MNTB) are stimulated by the activation of glutamate transporters and consequently release glutamine on a very rapid timescale. MNTB principal neurones express electrogenic system A glutamine transporters, and were exploited as glutamine sensors in this study. By simultaneous whole-cell voltage clamping astrocytes and neighbouring MNTB neurones in brainstem slices, we show that application of the excitatory amino acid transporter (EAAT) substrate d-aspartate stimulates astrocytes to rapidly release glutamine, which is detected by nearby MNTB neurones. This release is significantly reduced by the toxins l-methionine sulfoximine and fluoroacetate, which reduce glutamine concentrations specifically in glial cells. Similarly, glutamine release was also inhibited by localised inactivation of EAATs in individual astrocytes, using internal dl-threo-β-benzyloxyaspartic acid (TBOA) or dissipating the driving force by modifying the patch-pipette solution. These results demonstrate that astrocytes adjacent to glutamatergic synapses can release glutamine in a temporally precise, controlled manner in response to glial glutamate transporter activation. Since glutamine can be used by neurones as a precursor for glutamate and GABA synthesis, this represents a potential feedback mechanism by which astrocytes can respond to synaptic activation and react in a way that sustains or enhances further communication. This would therefore represent an additional manifestation of the tripartite relationship between synapses and astrocytes.

Key points

• Following release of glutamate from excitatory synapses, excitatory amino acid transporters (EAATs) sequester this glutamate into neighbouring astrocytes.

• The signalling effects on the astrocyte and the mechanisms by which this glutamate is recycled back to the synapse are currently unclear.

• In this study we use electrophysiological recording from neurones and astrocytes to show that a surge of the neurotransmitter glutamate, as usually occurs during neuronal activity, activates astrocytes and causes them to rapidly release the amino acid glutamine.

• This glutamine mediates a fast signal back to the neurones, where it is sequestered and is available for the biosynthesis of further neurotransmitters.

• Our data demonstrate a novel feedback mechanism by which astrocytes can potentially modulate neuronal function, and pave the way for development of new therapeutic approaches to treat neurological disorders.

Introduction

Ultrastructual analysis of mammalian synapses has shown that the astrocytic membrane is located immediately adjacent to glutamatergic synapses in the cerebellum, cortex, hippocampus and brainstem (Spacek, 1985; Ventura & Harris, 1999; Satzler et al. 2002). This glial association is highly dynamic, with increases in synaptic activity due to long-term potentiation, kindling, sensory activity, motor skills learning and synaptic maturity causing hypertrophy of astrocytic processes, increased astrocytic volume and enhanced ensheathment of synapses (Wenzel et al. 1991; Hawrylak et al. 1993; Anderson et al. 1994; Jones & Greenough, 1996; Genoud et al. 2006; Witcher et al. 2007). More recently it has been established that astrocytes play a central role in regulating synaptic communication through their ability to sequester and release neurotransmitters and neuromodulators. This tripartite relationship between presynaptic, postsynaptic and astrocytic elements has been clearly demonstrated for neuroactive compounds such as adenosine, ATP, glutamate and d-serine (Perea & Araque, 2007; Bardoni et al. 2010; Henneberger et al. 2010). Accordingly, it is evident that astrocytes respond in a localised manner to specific neurotransmitters, in a rapid and controlled fashion (Perea et al. 2009).

Astrocytes are also proposed to be mediators of the glutamate–glutamine cycle, where synaptically released glutamate is recycled via conversion to glutamine in glial cells (van den Berg & Garfinkel, 1971; Hertz et al. 1999; Bak et al. 2006). Essential components of this cycle include: uptake by high-affinity excitatory amino acid transporters (EAATs; Danbolt, 2001), which are known to be concentrated on astrocytic membranes adjacent to synapses (Chaudhry et al. 1995); conversion of glutamate to glutamine by the astrocytic enzyme glutamine synthetase (Norenberg & Martinez-Hernandez, 1979); release of glutamine from astrocytes; sequestration of this glutamine into neurones (Chaudhry et al. 2002a); and subsequent conversion back to glutamate by enzymes such as phosphate-activated glutaminase (Kvamme et al. 2000). The importance of this cycle in maintaining neurotransmitter supply has been demonstrated in vitro (Pow & Robinson, 1994; Laake et al. 1995) and in vivo (Gibbs et al. 1996; Barnett et al. 2000; Sibson et al. 2001).

Consequently, the release of astrocytic glutamine is a vital step in the maintenance of neurotransmitter levels at glutamatergic synapses. Furthermore, since glutamate is the main precursor of GABA, the continued supply of glutamine has also been shown to be crucial for sustaining inhibitory neurotransmission (Sonnewald et al. 1993; Liang et al. 2006; Fricke et al. 2007). However, despite the physiological importance of this process, rapid release of glutamine from individually identified astrocytes in the vicinity of a synapse has not been directly demonstrated.

In this present study, we investigate astrocytic glutamine release at the calyx of Held synapse in the auditory brainstem. This glutamatergic synapse surrounds the principal cells of the medial nucleus of the trapezoid body (MNTB) and is large enough to be visually identified. In acutely isolated brain slices, the astrocyte immediately adjacent to the synapse can be easily located and the close proximity of the astrocytic cell body to the neuronal synapse is advantageous for patch-clamp recording. We have previously shown that the postsynaptic MNTB cells express system A glutamine transporters, the electrogenic nature of which results in a membrane current upon activation by glutamine. Taking advantage of this, we have used the MNTB principal neurones as glutamine detectors to sense glutamine release from glial cells immediately apposing the synapse. We find that stimulation of EAAT glutamate transporters on astrocytes results in a very rapid, tightly coupled release of glutamine, which would be available to the presynaptic terminal to replenish the glutamate supply for synaptic transmission.

Methods

Ethical approval

All procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986, and approved by the Local Ethical Review Panel at the University of Cambridge, UK.

Slice preparation

Wistar rats 10–15 days old were killed by decapitation and brains quickly removed and immediately transferred to an ice-cold, low-sodium solution, composed of (in mm): 250 sucrose, 2.5 KCl, 10 glucose, 1.25 NaH₂PO₄, 26 NaHCO₃, 4 MgCl₂, 0.1 CaCl₂, and gassed with 95% O₂–5% CO₂ to pH 7.4. Transverse brain slices, 110 to 130 μm thick, containing MNTB were obtained, using an Integraslice 7550PSDS (Campden Instruments; Loughborough, UK). Slices were placed in an incubation chamber containing artificial cerebrospinal fluid (aCSF) with composition (in mm): 125 NaCl, 2.5 KCl, 10 glucose, 1.25 NaH₂PO₄, 26 NaHCO₃, 1 MgCl₂ and 2 CaCl₂ which was gassed with 95% O₂–5% CO₂ (pH 7.4) and kept at 35°C for 1 h prior to storage at room temperature for up to 8 h.

Electrophysiological recording

For recording, brain slices were continually perfused with aCSF (composition as above; 31–35°C), at a rate of 1 ml min⁻¹ and visualised with infrared differential interference contrast optics. MNTB principal cells were identified by their characteristically large (∼20 μm) spherical soma and astrocytes adjacent to principal cells were identified by their smaller soma (<10 μm). Astrocytes exhibited characteristic electrophysiological properties (membrane resistance <10 MΩ; resting membrane potential < −78 mV; lack of voltage-activated currents) and identification was confirmed in a separate series of experiments by labelling slices with 1 μm sulforhodamine 101 for 20 min at 35°C, immediately following slicing (Nimmerjahn et al. 2004; Kafitz et al. 2008). Furthermore, Lucifer yellow (<0.05%) was included in the patch pipette solution to enable fluorescent identification of cell type at the end of the experiment.

Membrane currents from MNTB principal neurones and astrocytes were obtained by whole-cell voltage clamping at −70 and −80 mV, respectively, using a HEKA EPC-10 double amplifier (HEKA Elektronik Dr Schulze GmbH; Lambrecht/Pfalz, Germany), filtered at 10 and 2.9 kHz, and digitized at 25 kHz with Patchmaster software (HEKA). Recordings were made using thick-walled glass pipettes (GC150F-7.5; Harvard Apparatus; Edenbridge, Kent, UK) with open-tip resistances of 5–8 MΩ for neurones and 6–9 MΩ for astrocytes. Whole-cell access resistances were <30 MΩ and series resistance compensation of 50% (100 μs lag time) was applied to neuronal recordings. For neuronal recordings, pipettes were filled with a solution containing (in mm): 110 caesium methanesulfonate, 40 HEPES, 10 TEA chloride, 5 Na₂-phosphocreatine, 20 sucrose, 0.2 EGTA, 2 MgATP, 0.5 Na₂GTP and 0.008 CaCl₂ (pH 7.2 with CsOH). For astrocytes the patch pipette contained (in mm): 130 KCl, 2 NaCl, 4 glucose, 10 HEPES, 0.1 EGTA, 1 MgATP, 0.5 Na₂GTP and 0.025 CaCl₂ (pH 7.2 with KOH). In experiments where the effects of internal Cs⁺ or K⁺ removal were assessed, Cs⁺ and K⁺ were replaced by N-methyl-d-glucamine (NMDG).

All recordings were made in the presence of (in μm): 40 dl-2-amino-5-phospohonopentanoic acid (APV), 10 dizocilpine maleate (MK801), 10 (–)-bicuculline methochloride, 1 strychnine, 1 TTX, 20 2,3,dioxo-6-nitro-1,2,3,4-tetrahydrobenzoquinoxaline-7- sulfonamide (NBQX) and 10 mm TEA chloride. Transporter substrates (d-aspartate and l-glutamine) were dissolved in the external solution and applied by pressurized ejection (2–8 psi; 13.8-68.9 kPa) from a pipette (open tip resistance 4–6 MΩ) using a Picospritzer II (General Valve; Fairtrade, NJ, USA). In neurones, d-aspartate puffs were 2 s in duration and repeated every 30 s. In astrocytes, d-aspartate puffs were of 5 s duration and repeated every minute. Antagonists were applied by addition to the bathing solution and by inclusion in a second puff pipette, along with the agonist. The two puffer pipettes were of equal tip diameter (pulled from the same glass capillary), connected to the same pressure source and placed equidistant from the cells. For experiments investigating the proportional contribution of EAAT1 and EAAT2 to the glia d-aspartate response, 10 μm UCPH-101 and 450 μm dihydrokainate (DHK) were used. The percentage inhibitions were corrected for the fact that these competitive inhibitors do not block 100% of the EAAT1 or EAAT2 currents, respectively. UCPH-101 inhibition of EAAT1 was calculated to be 85%, using K_i = 0.41 μm (calculated from the data of Jensen et al. 2009) and a K_m for d-aspartate of 60 μm (Arriza et al. 1994). DHK inhibition of EAAT2 was calculated to be 81%, using K_i = 23 μm and K_m for d-aspartate of 54 μm (Arriza et al. 1994). Similarly 200 μm dl-threo-β-benzyloxyaspartic acid (TBOA) would be expected to inhibit EAAT1 by 52%, using K_i = 42 μm and inhibit EAAT2 by 88%, using K_i = 5.7 μm (Shimamoto et al. 1998). Using the relative proportions of EAAT1 and EAAT2 present (see Results), 200 μm TBOA would be expected to inhibit 64% of the total EAAT current in astrocytes. Furthermore, 20 μm of the higher affinity substrate inhibitor (3S)-3-[[3-[[4-(trifluoromethyl)benzoyl]amino]phenyl]methoxy]-l-aspartic acid (TFB-TBOA; Shimamoto et al. 2004) would be expected to inhibit 99.5% of the astrocytic EAAT current.

All chemicals were obtained from Sigma Aldrich (Gillingham, Dorset, UK) except TTX, bicuculline, NBQX, APV, MK801, UCPH-101 and DHK (Ascent Scientific; Bristol, UK) and TBOA and TFB-TBOA (Tocris BioScience; Bristol, UK).

Sodium imaging

For measuring sodium concentration changes in astrocytes, 0.5 mm SBFI (Minta & Tsien, 1989) was included in the patch-pipette solution, and NaCl omitted. Slices were imaged on a Nikon FN1 microscope using a 60× NA 1.0 fluorite lens (Nikon Corporation, Tokyo, Japan). Cells were illuminated at 350 and 380 nm for 100 ms at each wavelength (Optoscan monochromator, Carin Research, Faversham, UK) and emitted light separated by a 400 nm dichroic mirror and filtered through a 420 nm long-pass filter. Fluorescence was detected using a Cascade 512B electron-multiplying CCD camera (Photometrics, Tucson, AZ, USA), and 350:380 nm ratio images were analysed using Metafluor software (Molecular Devices, Sunnyvale, CA, USA). Calibration was performed by construction of a linear calibration curve using imaging with pipette solutions of known sodium concentrations. A 10% change in background-subtracted 350:380 nm ratio corresponded to a change in sodium concentration of 4.0 mm (R² = 0.99).

Data analysis

Data are presented as mean ± SEM and regarded as statistically significant when P < 0.05 using one-way analysis of variance (ANOVA) with Dunnett's post hoc test (GraphPad Prism 5.01; GraphPad Software, San Diego, CA, USA), unless otherwise stated. Significance levels indicated on figures are 0.05 (*), 0.01 (**) and 0.001 (***).

Results

MNTB cells can act as glutamine sensors

Astrocytic glutamine release was assessed by recording system A glutamine transporter currents on adjacent neurones. To confirm the suitability of the postsynaptic MNTB cells as glutamine sensors in the region of the synapse, whole-cell patch-clamp recordings from these cells were made and their sensitivity to glutamine investigated. To ensure that trace contamination of glutamine with glutamate does not activate glutamate receptors on the MNTB cells and generate a current, all recordings were performed in the presence of ionotropic glutamate receptor antagonists APV, MK801 and NBQX. Puff-application of 10 mm glutamine from a glass pipette adjacent to the cell (Fig. 1A) induced a membrane current (I_Gln) of −20.8 ± 0.8 pA (Fig. 1B; mean ± SEM; n = 81). Removal of external Na⁺ (replaced with choline) abolished I_Gln (Fig. 1C; 99.5 ± 3.6% inhibition; n = 8; P < 0.001), demonstrating the sodium-dependent nature of the current. To further investigate the nature of I_Gln, the system A substrate α-(methylamino)isobutyric acid (MeAIB) was bath applied prior to and during the glutamine puff. MeAIB is a competitive substrate with specificity for system A transport over other glutamine transporters (Christensen et al. 1965), and 20 mm MeAIB reduced I_Gln by 92.9 ± 2.8% (Fig. 1C; n = 10; P < 0.001). These results show that glutamine induces a membrane current in MNTB neurones, which is mediated by the system A glutamine transporters (Blot et al. 2009). Any possible role of high-affinity glutamate transporters (EAATs) in generating this postsynaptic current was excluded by showing a lack of effect of the non-transportable EAAT inhibitor TBOA (200 μm) on the glutamine-evoked current (Fig. 1C; 2.6 ± 3.0% inhibition; n = 6; P > 0.05). We also tested the effects of the EAAT activator d-aspartate on I_Gln. Bath application of 200 μm d-aspartate did not have a significant effect on I_Gln (Fig. 1C; 110.4 ± 5.8% of control; n = 3; P > 0.05), confirming that this current is not mediated by EAATs and demonstrating that d-aspartate does not affect glutamine transport on the MNTB cell, which is an important control for later experiments.

Figure 1. The postsynaptic glutamine response is mediated by system A transporters.

A, a differential interference contrast (DIC) image of an MNTB neurone (asterisk) in a brainstem slice, which is whole-cell voltage-clamped at −70 mV via the patch pipette (arrow). Glutamine or glutamine + antagonist were applied by pressure ejection from one of two puffer pipettes placed equidistant from the cell (arrowheads). The scale bar is 20 μm. B, 1 s puff application of 10 mm glutamine induced a membrane current of −20.8 ± 0.8 pA (n = 81). C, the glutamine-induced current was abolished by the removal of external sodium ions and very significantly inhibited by the system A transporter substrate MeAIB (20 mm). An EAAT inhibitor (TBOA; 200 μm) or EAAT substrate (d-aspartate; 200 μm) had no effect on the glutamine-induced current.

Perisynaptic astrocytes express functional glutamate transporters

Astrocytes are known to express EAAT1 and EAAT2 transporters, which are responsible for removing glutamate from the synaptic cleft following neurotransmission (Bergles et al. 1999; Danbolt, 2001). To investigate EAAT-mediated glutamate transport on astrocytes in the MNTB, we whole-cell voltage-clamped astrocytes and puff-applied d-aspartate, an EAAT substrate that does not activate ionotropic glutamate receptors. Astrocytes adjacent to the calyx of Held synapse could be identified by bathing slices in sulforhodamine 101 (Fig. 2A–C; Nimmerjahn et al. 2004; Kafitz et al. 2008). Additionally, electrical membrane properties were used to confirm astrocytic recordings (Muller et al. 2009): the resting membrane potential was <−78 mV, the current–voltage relationship was passive (Fig. 2E) and the membrane resistance was <10 MΩ. Puff-application of 200 μm glutamate (in the presence of glutamate receptor inhibitors APV, MK801 and NBQX) induced a current of –31.3 ± 10.5 pA (n = 5), which was inhibited 86.7 ± 11.8% by addition of the EAAT inhibitor TBOA (n = 5; P < 0.05; paired Student's two-tailed t test). To avoid complications resulting from activation of glutamate receptors, subsequent experiments used d-aspartate to stimulate EAATs. Puff-application of 200 μm d-aspartate induced a current (glial I_D-asp) of −29.0 ± 1.5 pA (n = 46) which was inhibited by external application of TBOA (Fig. 2G; 72.5 ± 2.6% inhibition; n = 5; P < 0.001). In addition, glial I_D-asp was eliminated by application of the high-affinity EAAT substrate inhibitor TFB-TBOA (Fig. 2F–G; 99.1 ± 0.9% inhibition; n = 5; P < 0.001). Similarly, inclusion of 5 mm TBOA in the patch-pipette solution also significantly reduced glial I_D-asp (Fig. 2G; 62.2 ± 12.4% inhibition; n = 3; P < 0.05), further indicating that it is mediated by astrocytic EAAT glutamate transporters. The subtypes of EAATs responsible for the astrocytic current were investigated using the isoform-specific inhibitors UCPH-101 and dihydrokainate (DHK). UCPH-101 exhibits a high degree of selectivity for EAAT1 (Jensen et al. 2009) and 10 μm inhibited glial I_D-asp by 60.2 ± 4.6% (corrected for incomplete inhibition of EAAT1 – see Methods; n = 3). DHK is selective for EAAT2 (Arriza et al. 1994) and 450 μm inhibited glial I_D-asp by 30.0 ± 3.9% (corrected; n = 3). A combination of UCPH-101 and DHK resulted in 9.7 ± 7.4% of glial I_D-asp remaining uninhibited. To ensure that puff-application did not evoke a membrane current as the result of mechanical stimulation, external solution lacking any agonists was puff-applied to astrocytes. A very slight outward current was observed (0.96 ± 0.53 pA; n = 5) which, combined with the observation that TFB-TBOA eliminates glial I_D-asp, demonstrates that our results are not significantly contaminated by puff artefacts.

Figure 2. Activation of perisynaptic astrocytes by d. aspartate application.

A, a DIC image of the MNTB showing astrocyte and neuronal somata. B, fluorescent image of sulforhodamine-101 labelled astrocytes in the same field of cells. C, overlay image of panels A and B showing the position of the astrocyte somata adjacent to the MNTB neurones. Scale bar in A, B and C is 10 μm. D, a fluorescent image of a whole-cell patch-clamped astrocyte (from a different slice to A) with Lucifer yellow included in the internal solution. The astrocyte processes can be seen surrounding the postsynaptic neurone (hash), in the same location as the presynaptic calyx of Held terminal (unlabelled). The scale bar is 20 μm. E, the linear current–voltage relationship of the astrocyte indicates a lack of significant voltage-activated currents, low membrane resistance and negative resting membrane potential. F, puff application of 200 μm d-aspartate at a holding potential of −80 mV induced an inward glutamate transport current in voltage-clamped astrocytes of −29.0 ± 1.5 pA (n = 46). Inclusion of 20 μm TFB-TBOA in the external solution eliminated the d-aspartate-induced current, indicating that it is mediated by activation of EAAT glutamate transporters. G, averaged data showing that the glial d-aspartate-induced current was inhibited by TBOA in the external and internal solutions and also by removal of K⁺ from the internal solution, demonstrating the involvement of EAATs. Inclusion of the system A substrate MeAIB (20 mm) had no effect on the current. H, fluorescent imaging of the astrocytic [Na⁺] by inclusion of SBFI in the patch-pipette solution. A 5 s puff application of 1 mm d-aspartate induced a [Na⁺] rise in the soma of 3.4 ± 0.3 mm (n = 7), which was inhibited by 50.9 ± 6.7% (n = 3) by the inclusion of 200 μm TBOA in the external solution, indicating that the [Na⁺] rise was mediated by EAAT transporters.

Glutamate transport is linked to the co-transport of three sodium ions, one proton and the counter transport of a potassium ion. Consistent with this stoichiometry, removal of intracellular K⁺ from the patch-pipette solution (replaced with NMDG) substantially inhibited the glutamate transporter current in the astrocyte (Barbour et al. 1988): glial I_D-asp was reduced by 89.2 ± 3.9% of control values by removal of (Fig 2G; n = 5; P < 0.001). Co-transport of sodium would be expected to cause an increase in [Na⁺]_i upon stimulation of glutamate transporters. This was investigated by fluorescent imaging of [Na⁺]_i using 0.5 mm SBFI in the patch-pipette solution (Minta & Tsien, 1989). Application of d-aspartate induced a rapid rise in the somatic [Na⁺]_i, by 3.4 ± 0.3 mm (n = 7), which was reduced by 50.9 ± 6.7% (Fig. 2H; n = 3; P < 0.05; paired Student's 2-tailed t test) by 200 μm TBOA. MeAIB (20 mm) had no significant effect on the glial I_D-asp (0.5 ± 5.9% inhibition; n = 5; P > 0.05), indicating that system A glutamine transport is not involved in this astrocytic activation and that MeAIB does not affect EAATs. These data confirm that d-aspartate activates glutamate transporters on perisynaptic astrocytes, causing an inward membrane current and increase in [Na⁺]_i.

Activation of astrocytic glutamate transporters causes glutamine release

Glutamine release into the region of the synapse was detected by whole-cell patch-clamping the MNTB cell. We find that activation of glial glutamate transporters using puff-application of 200 μm d-aspartate induces the release of glutamine, which subsequently stimulates the MNTB cell. This d-aspartate-induced glutamine transporter current recorded in MNTB cells (MNTB I_D-asp) was observed in 113 of 125 MNTB cells (90.4%) and averaged −8.7 ± 0.6 pA (Fig. 3; n = 125). As these neurones express a high density of ionotropic glutamate receptors and exhibit glutamatergic EPSCs of >10 nA (Forsythe & Barnes-Davies, 1993; Borst et al. 1995), the orders of magnitude difference between the receptor and transporter currents would prevent adequate pharmacological separation of glutamate-induced currents and preclude the use of glutamate as an EAAT substrate in these experiments. Nevertheless, the role of glutamate transporters in initiating the d-aspartate-induced current is confirmed by the application of TBOA, which inhibited the MNTB I_D-asp by 67.9 ± 9.1% (Fig. 3A; n = 6; P < 0.01). Furthermore, TFB-TBOA (20 μm) inhibited MNTB I_D-asp by 94.2 ± 9.5% (Fig. 3D; n = 3; P < 0.01). It is possible that d-aspartate directly activates neuronal EAATs on MNTB principal cells, as immunohistochemical experiments reveal low levels of the neuronal EAAT3 transporter in the MNTB (Renden et al. 2005). However, to be certain that d-aspartate does not directly activate any putative neuronal EAATs, an intracellular solution lacking Cs⁺ (which substitutes for internal K⁺ on EAAT transporters; Barbour et al. 1991) was used. Upon removal of (substituted by NMDG), MNTB I_D-asp was unchanged, at 113.0 ± 28.5% of control (Fig. 3B; n = 3; P > 0.05). In contrast to effects of removing internal K⁺ on the EAAT current in glial cells (Fig. 2G), this result demonstrates that EAATs are not functional in the postsynaptic MNTB cells and do not contribute to MNTB I_D-asp. If glutamine acting on system A transporters is creating the observed MNTB current, then MeAIB or an excess of glutamine in the external solution would both be expected to abolish the current. Accordingly, 20 mm MeAIB inhibited MNTB I_D-asp completely (Fig. 3C; n = 6; P < 0.001) as did 5 mm glutamine (Fig. 3D; n = 5; P < 0.001) consistent with MNTB I_D-asp being mediated by released glutamine acting on MNTB cells.

Figure 3. Stimulating perisynaptic astrocytes results in glutamine release, detected by neighbouring postsynaptic neurones.

A, application of 200 μm d-aspartate to the brain slice resulted in a current in whole-cell voltage-clamped postsynaptic MNTB neurones, which was inhibited by adding 200 μm TBOA to the external solution, demonstrating the involvement of EAAT activation. B, the postsynaptic d-aspartate-evoked current was unaffected by the removal of Cs⁺ from the internal solution, highlighting the lack of involvement of postsynaptic EAATs in the d-aspartate response. C, application of the system A substrate MeAIB (20 mm) occluded the effect of d-aspartate application, indicating that system A is involved in the generation of the postsynaptic current. D, averaged data showing the effects of 200 μm TBOA, 20 μm TFB-TBOA, removal of Cs⁺, 20 mm MeAIB and 5 mm glutamine on the d-aspartate-induced neuronal current.

Release of glutamine is mediated by perisynaptic astrocytes

Two separate pharmacological approaches were used to demonstrate the involvement of astrocytes in releasing the glutamine that we detect. l-methionine sulfoximine (MSO) is an inhibitor of glutamine synthetase, and prevents the conversion of glutamate to glutamine in astrocytes. It specifically reduces glutamine production in comparison with other amino acids (Fonnum et al. 1997) and the subsequent reduction in the astrocytic glutamine concentration (Tani et al. 2010) would be expected to reduce glutamine release and thus inhibit MNTB I_D-asp. Alternatively, fluoroacetate (FAc) is a glial-specific toxin that is taken up into astrocytes, converted to fluorocitrate and inhibits aconitase. The resulting inhibition of the astrocytic tricarboxylic acid (TCA) cycle causes astrocytic glutamate to be metabolised via α-ketoglutarate, and subsequently reduces glutamine production (Swanson & Graham, 1994; Fonnum et al. 1997), which would reduce MNTB I_D-asp. Incubation of the brain slices in 10 mm MSO for at least 120 min prior to recording inhibited MNTB I_D-asp by 64.0 ± 24.5% (Fig. 4A and C; n = 9; P < 0.01). Correspondingly, application of 10 mm FAc for 30–120 min inhibited MNTB I_D-asp by 66.5 ± 7.4% (Fig. 4B and C; n = 8; P < 0.01) without any observable effects on the viability of the MNTB cells. These pharmacological data strongly implicate the astrocytes as being the source of the released glutamine. There are three times as many MNTB principal cells as astrocytes in this brain region (Reyes-Haro et al. 2010), suggesting that a single perisynaptic astrocyte may ensheathe more than one calyx of Held synapse and that each synapse may be ensheathed by only one astrocyte, which is coupled to neighbouring astrocytes via gap junctions (Muller et al. 2009). To investigate this relationship, we simultaneously whole-cell patch-clamped MNTB cells and neighbouring astrocytes, and activated glial glutamate transport (Fig. 5A). Puff-application of d-aspartate (Fig. 5B) demonstrated that MNTB I_D-asp was activated very rapidly, within a few milliseconds of glial I_D-asp. Two different techniques were used to investigate the role of an identified astrocyte in mediating the glutamine release. First, the astrocyte was whole-cell patch-clamped with a pipette containing extracellular solution and at a membrane potential of 0 mV, which would prevent astrocytic d-aspartate transport by removal of the driving force powering the EAATs. The glutamine release (as assessed by MNTB I_D-asp) was recorded immediately before and after inactivating the astrocytic EAATs in this way. Our data show that the MNTB I_D-asp was instantly inhibited by 81.6 ± 15.5% (Fig. 5C and D; n = 4; P < 0.01) by inactivation of the astrocyte. The second method was to include 5 mm TBOA in the astrocytic patch-pipette solution. This high concentration of TBOA was used to ensure that sufficient TBOA would diffuse down the astrocytic processes to inhibit glutamate transporters, which are located adjacent to the synapse. Internal TBOA reduced glial I _D-asp by 62.2 ± 12.4% (n = 3; P < 0.05) and reduced MNTB I_D-asp by 75.7 ± 10.3% (Fig. 5C; n = 3; P < 0.05). These data suggest that a perisynaptic astrocyte mediates the glutamate transporter-evoked glutamine release, with a high degree of temporal precision.

Figure 4. The stimulated release of glutamine is mediated by glial cells.

A, postsynaptic currents induced by the application of d-aspartate to the synapse are significantly inhibited by incubating the tissue in the glutamine synthetase inhibitor l-MSO (10 mm), which depletes astrocytes of glutamine. B, incubating the tissue with the glial-specific metabolic toxin fluoroacetate (FAc; 10 mm) also severely compromised the d-aspartate-induced postsynaptic current, further demonstrating the role of glial cells in the activated pathway. C, averaged data indicating the similar levels of inhibition of the postsynaptic d-aspartate-induced current by l-MSO (64%) and FAc (67%).

Figure 5. A perisynaptic astrocyte mediates the d. aspartate-induced release of glutamine.

A, a DIC image showing a voltage-clamped astrocyte (asterisk) adjacent to a voltage-clamped postsynaptic MNTB neurone (hash). A puffer pipette (arrowhead) is positioned nearby to apply d-aspartate. The scale bar is 20 μm. B, puff application of 200 μm d-aspartate induces an EAAT-mediated current in the astrocyte (black trace) and an almost instantaneous system A-mediated current in the adjacent neurone (grey trace). C, the neuronal current was significantly inhibited by inactivating the perisynaptic astrocyte by continual depolarization to 0 mV, or by including TBOA in the astrocyte patch-pipette solution. D, averaged data (from 4 cells) showing the time dependence of the reduction in postsynaptic d-aspartate-induced current when the astrocyte is inactivated by voltage-clamping the cell at 0 mV.

Discussion

The release of glutamine from astrocytes is a key component of the glutamate–glutamine cycle (Chaudhry et al. 2002a). However, the time scale on which this cycle may occur has not been established. Here, we directly demonstrate that activation of glutamate transport on perisynaptic astrocytes causes a rapid release of glutamine from the astrocyte, with tight temporal coupling. This suggests that astrocytic glutamine may be an immediate modulator of neuronal function.

We have studied glutamine release from astrocytes in the MNTB, adjacent to the calyx of Held synapse. This brain area has several key advantages for these experiments: the MNTB principal cells are spherical and have only one or two short dendrites (Leao et al. 2008), making them electrically compact and allowing recording of the relatively small transporter current without significant dendritic filtering; the main synaptic input to the MNTB principal cell is somatic and can be visually identified; the astrocyte adjacent to the synapse can be located; and the small number of astrocytes compared with principal cells (one astrocyte for three principal cells) suggests that only one astrocyte may be associated with the majority of the synaptic input to a single cell (Reyes-Haro et al. 2010). In addition, the ability to perform simultaneous patch-clamp experiments of the identified perisynaptic astrocyte and the postsynaptic neurone allows direct manipulation of this glial element in the synapse. Since we have demonstrated that the postsynaptic MNTB cell is sensitive to glutamine levels in the external solution, we have used the membrane current resulting from system A glutamine transporter activation as an indicator of glutamine release from the astrocyte when glutamate transport is stimulated in the immediate vicinity of the synapse.

Astrocytic glutamate uptake

Our data demonstrate that MNTB astrocytes possess functional glutamate transporters, making them likely candidates to take up synaptically released glutamate. Immunohistological studies reveal that MNTB astrocytes express both the EAAT1 (GLAST) and the EAAT2 (GLT-1) isoforms of glutamate transporters (Renden et al. 2005; Ford et al. 2009), and our data using isoform-specific antagonists, UCPH-101 and DHK, reveal that EAAT1 is the main functional isoform present, contributing approximately twice the current of EAAT2. This finding is consistent with the expression pattern observed in the cerebellum and in contrast to the forebrain where EAAT2 predominates (Chaudhry et al. 1995; Lehre & Danbolt, 1998). A negligible amount of current remains when both EAAT1 and EAAT2 are inhibited, which, in combination with the almost complete inhibition of current by removal of internal K⁺ and its elimination by TFB-TBOA, suggests an insignificant contribution of non-EAAT transport in mediating the d-aspartate current in astrocytes. Astrocytes are known to sequester synaptically released glutamate via EAATs and a direct activation of glial glutamate transporters following synaptic stimulation has been observed in the cerebellum (Clark & Barbour, 1997) and hippocampus (Bergles & Jahr, 1997), but not in the MNTB (Reyes-Haro et al. 2010). This difficulty in observing synaptically induced astrocytic transporter currents could be a result of the high conductance of the astrocyte membrane or as a consequence of the small number (<10%) of MNTB cells that retain active synaptic connections following the brainstem slice procedure (Billups et al. 2002). Moreover, inhibiting astrocytic glutamate transport influences the levels of glutamate in and around the synapse and alters the properties of synaptic transmission at the calyx of Held synapse (Renden et al. 2005), as well as at other calyceal synapses (Otis et al. 1996), in the cerebellum (Barbour et al. 1994; Overstreet et al. 1999; Carter & Regehr, 2000; Marcaggi et al. 2003) and in the hippocampus (Scanziani et al. 1997; Huang et al. 2004). This demonstrates that glutamate sequestration by astrocytic EAATs plays an important role in synaptic physiology. As location of EAATs adjacent to active synapses is a highly dynamic process and an increase in synaptic activity causes a rapid clustering of transporters adjacent to regions of higher glutamate release (Zhou & Sutherland, 2004; Yang et al. 2009; Benediktsson et al. 2012), astrocytic glutamate uptake mediates a key neuronal–glial communication pathway.

Astrocytic glutamine release

Our data show that stimulation of astrocytes by EAAT activation causes the release of glutamine. Glutamine release is detected by its subsequent activation of the system A amino acid transporter, and while it is possible that astrocytes release a system A substrate other than glutamine, this is not supported by our data. MSO is a relatively specific inhibitor of glutamine synthetase and inhibiting this pathway reduces astrocytic glutamine content but does not reduce glutamate, aspartate or alanine concentrations (Fonnum et al. 1997). Further support for glutamine mediating the glial–neuronal communication comes from the use of FAc. Inhibition of the glial TCA cycle by FAc (or its active metabolite fluorocitrate) has been shown to substantially reduce glutamine production, but not that of other amino acids or ATP, and does not inhibit glial glutamate or d-aspartate uptake (Paulsen et al. 1988; Swanson & Graham, 1994; Fonnum et al. 1997). The only other system A substrate thought capable of mediating transfer of neurotransmitter precursors from glia to neurones is alanine (Broer et al. 2007). However, both FAc and MSO have been shown to actually increase the alanine content of astrocytes by 2- to 3-fold (Fonnum et al. 1997), presumably as a consequence of diverting astrocytic glutamate metabolism away from glutamine production. As FAc and MSO inhibit the glial–neuronal communication in our experiments, it cannot be mediated by release of alanine or any of its metabolites from astrocytes. In addition to its action as a system A substrate, alanine also acts as an agonist at glycine receptors (Young & Snyder, 1973), which are expressed at the calyx of Held and have a depolarizing action (Price & Trussell, 2006; Huang & Trussell, 2008) that enhances neurotransmitter release (Turecek & Trussell, 2001; Awatramani et al. 2005). However, any contribution of alanine-activated currents to our results is eliminated by the strychnine present in the external solution.

It has been proposed that glutamine release from astrocytes is mediated by the system N transporter family, which is comprised of two main transporters: system N 1 (named SN1, NAT or SNAT3) and system N 2 (SN2 or SNAT5) (Kilberg et al. 1980; Chaudhry et al. 1999; Gu et al. 2000; Nakanishi et al. 2001). Immunohistochemical staining has shown that both isoforms are expressed in astrocytes of the MNTB, where they are located in astrocytic processes adjacent to presynaptic terminals (Boulland et al. 2002; Cubelos et al. 2005). The homologous transporter SNAT7 has also been ascribed to system N, although this transporter is expressed in neurons and not in astrocytes (Hagglund et al. 2011). Glutamine transport by system N is linked to the co-transport of a sodium ion and the counter-transport of a proton, resulting in electroneutral, pH-dependent transport (Broer et al. 2002; but see Fei et al. 2000). Compared with system A, which does not counter-transport a proton (Chaudhry et al. 2002b), system N operates near to thermodynamic equilibrium under physiological conditions and is therefore expected to reverse more readily to mediate glutamine efflux from astrocytes. In addition to systems A and N, astrocytes express systems L, y+L and ASC, which can also transport glutamine (Nagaraja & Brookes, 1996; Su et al. 1997; Broer et al. 1999; Heckel et al. 2003). These transporters are thought to be capable of mediating glutamine release and are obligate exchangers, requiring another amino acid to be transported in the opposite direction. Importantly, systems L and y+L are not sensitive to aspartate (Kanai et al. 1998; Segawa et al. 1999; Broer et al. 2000), and ASC is not influenced by d-aspartate (Heckel et al. 2003), excluding direct d-aspartate–glutamine exchange as a mechanism for glutamine release in our experiments.

Triggering glutamine release

Our data indicate that EAAT activation is required for glutamine release. However, it is important to establish that there is no direct activation of EAATs on the postsynaptic MNTB neurones themselves, and that the d-aspartate-induced currents were entirely mediated by astrocytic glutamine release. This was evident from their complete inhibition by MeAIB (Fig. 3C), which does not inhibit EAATs (Fig. 2G), and the lack of inhibition by removal of internal Cs⁺ (Fig. 3B). Therefore, we conclude that there is no EAAT transporter activity in postsynaptic MNTB neurones. While it is clear that d-aspartate application is activating astrocytes and not neurones, the exact mechanism linking glutamate transport to glutamine release is unclear. It has previously been shown that incubating cultured astrocytes with glutamate increases the internal glutamine concentration and can stimulate glutamine release by enhancing the affinity of system N transport for glutamine (Broer et al. 2004). However, this effect is only apparent after 10 min and the glutamine release that we observe occurs on the millisecond time scale. While we cannot rule out an effect of d-aspartate on the substrate affinity of the glutamine release mechanism, it is unlikely as d-aspartate is not metabolised to glutamine and would not be expected to have the same direct effects on system N transport as glutamate. A more likely mechanism underlying the incredibly rapid glutamate transporter-induced glutamine release we observe might be due to the properties of EAATs. As transport of d-aspartate by EAATs is driven by the co-transport of three sodium ions and a proton, and counter transport of a potassium ion (Zerangue & Kavanaugh, 1996), its application causes depolarization of the astrocytes. These ion and voltage changes may stimulate subsequent glutamine release. However, our experiments were performed under voltage-clamp, ruling out any possible effect of depolarization in mediating the stimulation of glutamine release. It is also unlikely that membrane voltage changes could drive the glutamine release physiologically, as (1) the high conductance of the astrocyte plasma membrane reduces any voltage change by a current-shunting mechanism, and (2) the electroneutrality of system N transport would suggest insensitivity to voltage changes. Similarly, the pH change associated with EAAT-mediated transport (internal acidification; Bouvier et al. 1992) would also not be expected to stimulate glutamine release, as the stoichiometry of system N transport indicates that internal acidification would inhibit release. In contrast, we show that activation of EAATs results in a significant and rapid rise in internal sodium ion concentration (Fig. 2H), consistent with previous reports of large sodium concentration rises in response to glutamate transporter activation in Bergmann glia (Kirischuk et al. 2007; Bennay et al. 2008) and astrocytes in cortical cultures (Chatton et al. 2000) and hippocampal slices (Langer & Rose, 2009; Langer et al. 2011). The low millimolar increase in sodium concentration that we observe in the astrocytic soma probably reflects a much larger increase in the fine processes (Langer & Rose, 2009; Langer et al. 2011), which will significantly alter the equilibrium of system N transport to one where a higher external glutamine concentration will be established, resulting in the release of glutamine to the extracellular space (Fig. 6). As there is a 3:1 relationship between sodium and glutamate influx on EAATs (Zerangue & Kavanaugh, 1996) and a 1:1 coupling of sodium and glutamine release by system N (Broer et al. 2002), amplification between glutamate uptake and glutamine release could occur. The concentration of glutamine in the extracellular space is thought to be 0.4 mm (Kanamori & Ross, 2004), which is below the level that would saturate the system A transporter isoforms thought to be responsible for the import of glutamine into neurones (K_ms = 0.3–2.3 mm; Yao et al. 2000; Chaudhry et al. 2002b; Mackenzie et al. 2003). Therefore, any increase in glutamine release would enhance neuronal glutamine uptake.

Figure 6. A proposed mechanism for astrocytic glutamine release.

Activation of glial glutamate transporters (EAATs) causes a rise in glial [Na⁺]_i, which favours the efflux of glutamine via glial system N transporters (SN). Released glutamine is subsequently detected by system A transporters (SA) located on adjacent MNTB neurones, resulting in a neuronal glutamine transport current.

Physiological implications

Glutamine is used by neurones in the generation of glutamate for excitatory neurotransmission (Lieth et al. 2001; Sibson et al. 2001; Rothman et al. 2003), and it is therefore likely that glutamine released from astrocytes is sequestered into presynaptic terminals for this purpose. The high levels of neurotransmission that are observed in brain-slice models of epilepsy are dependent on glutamine (Bacci et al. 2002; Tani et al. 2007, 2010), and the high firing frequencies of auditory synapses such as the calyx of Held (Hermann et al. 2007; Kopp-Scheinpflug et al. 2008) suggest that a similar mechanism also operates in this brain region under physiological conditions. The tight temporal coupling between sensing glutamate and releasing glutamine infers that astrocytic glutamine release may reflect a mechanism by which excitatory transmission may be rapidly modulated. In addition to glutamate production, glutamine is also required for the production of GABA (Liang et al. 2006; Fricke et al. 2007), indicating that astrocytic glutamine release may similarly enhance inhibitory neurotransmission. Hence, it is possible that glutamine release mediates the modulation of GABAergic transmission by astrocytes in the barrel cortex (Benedetti et al. 2011). These observations suggest that alteration of astrocytic glutamine release, such as by phosphorylation of system N transporters (Balkrishna et al. 2010; Nissen-Meyer et al. 2011) would be expected to have far-reaching effects on neuronal communication. Hence, our study demonstrates the potential for astrocytes to modulate neurotransmission on a very rapid time scale by the release of glutamine. This would represent a further manifestation of the tripartite relationship between astrocytes and synapses.

Glossary

Abbreviations

aCSF

artificial cerebrospinal fluid

DHK

dihydrokainate

EAAT

excitatory amino acid transporter

FAc

fluoroacetate

MeAIB

α-(methylamino)isobutyric acid

MK801

dizocilpine maleate

MNTB

medial nucleus of the trapezoid body

MSO

l-methionine sulfoximine

Author contributions

N.M.U. and B.B. conceived and designed the experiments. N.M.U., M.-C.M., Q.C. and B.B. collected, analysed and interpreted the data. N.M.U. and B.B. drafted the manuscript and N.M.U., M.-C.M., Q.C. and B.B. critically revised the manuscript. All authors approved the final version of the manuscript. All experiments were performed in the Department of Pharmacology, University of Cambridge.