Activation of the Tonic GABA_C Receptor Current in Retinal Bipolar Cell Terminals by Nonvesicular GABA Release

Abstract

Within the second synaptic layer of the retina, bipolar cell (BC) output to ganglion cells is regulated by inhibitory input to BC axon terminals. GABA_A receptors (GABA_ARs) mediate rapid synaptic currents in BC terminals, whereas GABA_C receptors (GABA_CRs) mediate slow evoked currents and a tonic current, which is strongly regulated by GAT-1 GABA transporters. We have used voltage-clamp recordings from BC terminals in goldfish retinal slices to determine the source of GABA for activation of these currents. Inhibition of vesicular release with concanamycin A or tetanus toxin significantly inhibited GABA_AR inhibitory postsynaptic currents and glutamate-evoked GABA_AR and GABA_CR currents but did not reduce the tonic GABA_CR current, which was also not dependent on extracellular Ca²⁺. The tonic current was strongly potentiated by inhibition of GABA transaminase, under both normal and Ca²⁺-free conditions, and was activated by exogenous taurine; however inhibition of taurine transport had little effect. The tonic current was unaffected by GAT-2/3 inhibition and was potentiated by GAT-1 inhibition even in the absence of vesicular release, indicating that it is unlikely to be evoked by reversal of GABA transporters or by ambient GABA. In addition, GABA release does not appear to occur via hemichannels or P2X₇ receptors. BC terminals therefore exhibit two forms of GABA_CR-mediated inhibition, activated by vesicular and by nonvesicular GABA release, which are likely to have distinct functions in visual signal processing. The tonic GABA_CR current in BC terminals exhibits similar properties to tonic GABA_AR and glutamate receptor currents in the brain.

INTRODUCTION

The retina performs significant processing of visual signals by means of lateral inhibitory connections within its two synaptic layers. In the inner plexiform layer (IPL), amacrine cells modulate ganglion cell activity both directly and by reducing bipolar cell (BC) glutamate release (Lukasiewicz and Werblin 1994; Zhang and Slaughter 1995). Amacrine cells form numerous conventional and reciprocal GABAergic synapses with BC axon terminals (Marc and Liu 2000), which express two subtypes of ionotropic GABA receptor: GABA_A and GABA_C receptors (GABA_ARs and GABA_CRs).

GABA_AR and GABA_CR currents evoked by GABA or by light differ significantly in their kinetics: GABA_AR currents are rapid and transient, whereas GABA_CR currents are slower and more sustained (Eggers and Lukasiewicz 2006a,b; Hull et al. 2006; Lukasiewicz and Shields 1998; Qian and Dowling 1995; Zhang and Slaughter 1995). GABA_CRs also have a higher GABA affinity and lower single-channel conductance than GABA_ARs (Feigenspan and Bormann 1994a; Palmer 2006; Qian and Dowling 1995). The receptor subtypes are therefore likely to have distinct roles in retinal processing and indeed have recently been shown to differentially regulate BC output (Eggers and Lukasiewicz 2006a,b). GABA_CRs appear to be particularly important for limiting postsynaptic N-methyl-d-aspartate (NMDA) receptor activation and generating transient ganglion cell light responses (Dong and Werblin 1998; Matsui et al. 2001; Sagdullaev et al. 2006; Zhang et al. 1997). Consistent with GABA_ARs and GABA_CRs subserving distinct functions in BC terminals, there is evidence that the receptor pathways are independently activated and regulated. For example, GABA_ARs and GABA_CRs do not appear to be co-localized either anatomically (Koulen et al. 1998) or functionally (Palmer 2006), and GABA_CRs but not GABA_ARs are strongly regulated by GAT-1 GABA transporters (Hull et al. 2006; Ichinose and Lukasiewicz 2002; Palmer 2006).

In addition to mediating slow evoked currents in BC terminals, GABA_CRs generate a spontaneous tonic current that regulates the ability of the terminals to fire Ca²⁺-dependent action potentials (Hull et al. 2006). Tonic GABA currents have also been described in brain regions such as the hippocampus, cerebellum, and thalamus, where they are mediated by extrasynaptic GABA_ARs that, like GABA_CRs, have a high GABA affinity and slow rate of desensitization and are strongly regulated by GABA uptake (Glykys and Mody 2007a). In some cases, tonic GABA_AR currents are not dependent on GABA released from synaptic vesicles (Rossi et al. 2003; Wu et al. 2003, 2006). To determine the source of GABA for activation of spontaneous and evoked GABAR currents in BC terminals, we have made voltage-clamp recordings directly from BC terminals in goldfish retinal slices. Our findings reveal novel functional properties of GABA_CR-mediated inhibition in the IPL.

METHODS

Goldfish (Carassius auratus) were dark-adapted for 1 h and killed by decapitation followed immediately by destruction of the brain and spinal cord under Schedule 1 of the UK Animals (Scientific Procedures) Act 1986. The eyeballs were removed and retinae dissected out and treated for 25 min with hyaluronidase to remove vitreous humor. Each retina was quartered, placed ganglion cell layer down on filter paper and kept at 4°C in medium comprising (in mM): 127 NaCl, 2.5 KCl, 1.0 MgCl₂, 0.5 CaCl₂, 5 HEPES, and 12 glucose, adjusted to pH 7.45 with NaOH.

Slices were cut at 250-μm intervals, transferred to the recording chamber, and perfused (1 ml/min) with medium comprising (in mM): 108 NaCl, 2.5 KCl, 1.0 MgCl₂, 2.5 CaCl₂, 24 NaHCO₃, and 12 glucose, gassed with 95% O₂-5% CO₂, pH 7.4. For nominally Ca²⁺-free solution, CaCl₂ was replaced by MgCl₂ (2.5 mM) and EGTA (2.0 mM) was added. For incubation in concanamycin A and tetanus toxin, and for the interleaved control experiments, slices were immersed in HEPES-buffered solution (as in the preceding text but with 2.5 mM CaCl₂) and not perfused to prevent wash-out of the toxins except in experiments requiring the subsequent application of drugs. Slices treated with concanamycin A/tetanus toxin and interleaved control slices were sequentially prepared from retinal quarters from the same fish. Slice preparation and recordings were performed at room temperature in daylight conditions. Drugs were bath-applied via the extracellular solution and locally-applied via pressure application from a low-resistance glass micropipette, positioned 20–50 μm from the recorded terminal, using a Picospritzer II (Intracell, Royston, UK). Drugs and salts were obtained from Tocris (Bristol, UK), Sigma-Aldrich (Gillingham, UK), Fisher Scientific (Loughborough, UK), and Toronto Research Chemicals.

Whole cell voltage-clamp recordings were obtained from large Mb-type BC terminals as described previously (Palmer et al. 2003). Most recordings were made from axon-severed terminals (determined by their capacitive current) (see Palmer et al. 2003) to eliminate currents arising from somatodendritic receptors and reduce the leak current. However, no differences in GABAR currents were observed between isolated terminals and the terminals of intact BCs. Patch pipettes (5–8 MΩ) were pulled from borosilicate glass and filled with solution comprising (in mM) 115 CsCl, 25 HEPES, 10 TEA-Cl, 3 Mg-ATP, 0.5 Na-GTP, 0.5 EGTA, pH 7.2. CsCl-based intracellular solution was used to increase the driving force through GABARs at a holding potential of −60 mV. Membrane current (I_M) was recorded via an EPC-10 patch-clamp amplifier controlled by Patchmaster software (HEKA, Lambrecht/Pfalz, Germany). Series resistance (R_S) was monitored and recordings were not used for analysis if I_M changes were accompanied by changes in R_S.

Off-line analysis was performed using IgorPro software (WaveMetrics, Lake Oswego, OR). Inhibitory postsynaptic currents (IPSCs) were identified and analyzed using macros kindly provided by Dr H. Taschenberger. The tonic current was measured as the average I_M during recording periods lacking IPSCs. Current variance was measured following subtraction of a straight-line fit of 3-s current traces selected to be free from IPSCs. Pooled data are expressed as means ± SE; statistical significance was assessed using Student's t-test with P < 0.05 considered significant.

RESULTS

To investigate the source of GABA that activates spontaneous and evoked GABAR currents in BC terminals, we tested the effect of inhibiting vesicular GABA release. Spontaneous GABA_AR and GABA_CR currents, recorded using CsCl-based intracellular solution at −60 mV, are readily distinguished: GABA_ARs mediate fast, transient IPSCs that are sensitive to bicuculline, whereas GABA_CRs mediate a tonic current that is sensitive to (1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA) (Palmer 2006) (Fig. 1 A). In BC terminals, GABA_ARs do not contribute to the tonic current (Hull et al. 2006). Thus TPMPA (50–100 μΜ) reduced the tonic current from −57 ± 5 to −13 ± 2 pA (n = 3, P < 0.05), but subsequent application of bicuculline (50 μΜ) caused no further reduction (−17 ± 4 pA, n = 3; Fig. 1B).

FIG. 1.

The tonic GABA_CR current in bipolar cell (BC) terminals is activated by GABA released via a nonvesicular mechanism. A: spontaneous GABAR-mediated membrane current (I_M) recorded from BC terminals with CsCl-based intracellular solution at −60 mV. The GABA_AR antagonist bicuculline (50 μM) inhibits the fast inhibitory postsynaptic currents (IPSCs), whereas the GABA_CR antagonist TPMPA (150 μM) inhibits the tonic current component (bic wo = bicuculline washed-out). B: example plot of I_M against time showing that TPMPA (50 μM) inhibits the tonic current but subsequent application of bicuculline (50 μM) has no further effect. C: example I_M recordings from BC terminals in control slices and slices incubated in concanamycin A (3.5 μΜ), an inhibitor of vesicular release. Concanamycin A eliminated the fast IPSCs but did not reduce the tonic current. D: tonic current vs. IPSC frequency for individual BC terminal recordings in control slices and slices incubated in concanamycin A (3.5 μΜ; circles), and mean data for these recordings (squares). E: application of TPMPA (150 μM) to a slice incubated in concanamycin A (3.5 μΜ) caused a reversible reduction in the tonic current. F: tonic current vs. IPSC frequency for individual BC terminal recordings in control slices and slices incubated in tetanus toxin (1–4 μg/ml; circles), and mean data for these recordings (squares). Tetanus toxin significantly reduced the IPSC frequency but did not inhibit the tonic current. In this and subsequent figures, error bars represent SE.

Spontaneous GABAR currents were recorded from BC terminals in retinal slices incubated in concanamycin A (3.5 μΜ for ∼50 min), an inhibitor of the proton-translocating ATPase that powers GABA uptake into synaptic vesicles (Drose and Altendorf 1997; Zhou et al. 2000), and compared with GABAR currents in interleaved control slices (incubated in the same concentration of the solvent DMSO). BC terminals in concanamycin A-treated slices exhibited a greatly reduced frequency of spontaneous IPSCs (control: 11 ± 2 Hz, n = 8; concanamycin A: 0.1 ± 0.1 Hz, n = 9, P < 0.01) but no reduction in the amplitude of the tonic current (control: −29 ± 3 pA, n = 8; concanamycin A: −37 ± 6 pA, n = 9; Fig. 1, C and D). Application of TPMPA (100–150 μΜ) to concanamycin A-treated slices reduced the tonic current in a reversible manner (from −36 ± 9 to −11 ± 3 pA, n = 5, P < 0.05; Fig. 1E), confirming that it was mediated by GABA_CRs.

Vesicular exocytosis was also inhibited by incubating slices in tetanus toxin (1–4 μg/ml for ∼50 min), which acts by cleaving the synaptic vesicle protein synaptobrevin (Grumelli et al. 2005). GABAR currents in tetanus toxin-treated slices were compared with GABAR currents in interleaved control slices. On average, spontaneous IPSC frequency was significantly reduced in tetanus toxin-treated slices (control: 17 ± 2 Hz, n = 20; tetanus toxin: 5 ± 1 Hz, n = 19, P < 0.01; Fig. 1F). Tetanus toxin was less effective at blocking vesicular release than concanamycin A, and the extent of the inhibition was variable between recordings. However, the amplitude of the tonic current was not reduced by tetanus toxin (control: −52 ± 7 pA, n = 20; tetanus toxin: −48 ± 8 pA, n = 19), and there was no correlation between the size of the tonic current and IPSC frequency (Fig. 1F). Application of TPMPA (100 μM) to tetanus toxin-treated BC terminals reduced the tonic current from −60 ± 6 to −17 ± 2 pA (n = 6, P < 0.01). These results strongly suggest that the tonic GABA_CR current is not activated by vesicular GABA release.

In addition to mediating a tonic current in BC terminals, GABA_CRs mediate more transient currents that can be evoked by activation of reciprocal amacrine cell synapses (Chavez et al. 2006; Vigh et al. 2005). To determine the source of GABA for evoked GABA_CR currents, we activated reciprocal synapses by local pressure-application of glutamate. Glutamate application (100 μM, 10 ms, 10 psi) evoked a large inward current that was almost completely inhibited by the GABA_AR and GABA_CR antagonist picrotoxin (50 μM; reduced from −56 ± 17 to −3 ± 1 pC, n = 5, P < 0.05; Fig. 2 A). A small inward current remained that was not observed in the presence of the glutamate transporter inhibitor dl-threo-β-benzyloxyaspartic acid (TBOA) (50 μM) (Palmer et al. 2003). However, as TBOA significantly increased both the tonic current and the size of evoked GABA responses (data not shown) and as these glutamate application parameters evoked only small glutamate transporter currents, TBOA was not used in subsequent experiments.

FIG. 2.

Glutamate-evoked GABAR currents are dependent on vesicular GABA release. A: example BC terminal current responses to focal application of glutamate (glu; 100 μM, 10 ms, 10 psi) in control conditions and following perfusion of the GABA_AR and GABA_CR antagonist picrotoxin (50 μM). B: example glu-evoked responses mediated by GABA_ARs (in 150 μM TPMPA), and GABA_CRs (in 50 μM bicuculline). C: example glu-evoked responses from slices incubated in concanamycin A (3.5 μΜ) and tetanus toxin (4 μg/ml). The glu-evoked response was virtually abolished by concanamycin A and reduced by tetanus toxin. D: the degree of inhibition of the glu-evoked response by concanamycin A and tetanus toxin was similar to the degree of inhibition of spontaneous IPSCs [tonic current and IPSC frequency data from Fig. 1; glu-evoked response charge data: concanamycin A (n = 9) compared with interleaved controls (n = 8), tetanus toxin (n = 12) compared with interleaved controls (n = 12)].

To determine the contribution of GABA_ARs and GABA_CRs to the glutamate-evoked response, recordings were made in the presence of TPMPA (100–150 μΜ) or bicuculline (50 μΜ). Glutamate-evoked responses were observed under both conditions, but their kinetics differed. Evoked GABA_AR responses contained many fast IPSC-like events, whereas GABA_CR responses exhibited a much smoother time course (Fig. 2B). Although we were specifically interested in the release mechanism of GABA mediating the GABA_CR component, mixed GABA_AR and GABA_CR responses were recorded to avoid the potentiating effects of bicuculline on GABA_CR currents that result from amacrine cell disinhibition (Palmer 2006; Zhang et al. 1997).

Glutamate-evoked GABA responses were compared between slices incubated in concanamycin A or tetanus toxin and interleaved control slices. Concanamycin A (3.5 μΜ) virtually abolished the glutamate-evoked response (control: −124 ± 58 pC, n = 8; concanamycin A: −7 ± 2 pC, n = 9; P < 0.05; Fig. 2, C and D). Tetanus toxin (4 μg/ml) also significantly inhibited the glutamate-evoked response (control: −137 ± 29 pC, n = 12; tetanus toxin: −49 ± 15 pC, n = 12; P < 0.05; Fig. 2, C and D), although the size of the response was variable between recordings. On average, the amount of inhibition of the glutamate-evoked GABA response by tetanus toxin or concanamycin A was very similar to the amount of inhibition of spontaneous IPSCs (Fig. 2D). These results are consistent with glutamate-evoked GABA release occurring via a conventional vesicular mechanism and do not support the existence of an evoked GABA_CR component of nonvesicular origin. Tonic and evoked GABA_CR currents in BC terminals therefore appear to be activated via distinct mechanisms.

To further investigate the source of GABA for activation of tonic GABA_CR currents, we examined whether the tonic current is Ca²⁺ dependent. Application of nominally Ca²⁺-free extracellular solution (containing 3.5 mM Mg²⁺) virtually abolished GABA_AR-mediated IPSCs (frequency reduced from 20 ± 5 to 0.1 ± 0.03 Hz, n = 5, P < 0.05) and initially caused a small but significant reduction in the tonic current (from −50 ± 10 to −32 ± 6 pA, n = 5, P < 0.05; Fig. 3, A and B). However, in the continued absence of Ca²⁺, while the IPSCs remained inhibited (0.2 ± 0.06 Hz, n = 5), the tonic current returned to baseline levels (−49 ± 5 pA, n = 5; Fig. 3, A and B). TPMPA (100 μM) inhibited the tonic current in Ca²⁺-free solution (reduced to −12 ± 2 pA, n = 6, P < 0.05; Fig. 3A). The tonic GABA_CR current in BC terminals is therefore not dependent on extracellular Ca²⁺ although it is transiently reduced when Ca²⁺ is removed.

FIG. 3.

The tonic GABA_CR current is Ca²⁺ independent and is not activated by GABA transporter reversal. A: example traces from an experiment showing the effect of perfusion of Ca²⁺-free extracellular solution (containing 3.5 mM Mg²⁺) and the sensitivity of the tonic current in Ca²⁺-free solution to TPMPA. B: mean changes in IPSC frequency and tonic current for BC terminals exposed to Ca²⁺-free solution (n = 5). The data for Ca²⁺-free(1) and Ca²⁺-free(2) were collected at the peak of the inhibitory effect and after the tonic current had stabilized, respectively. In this and subsequent figures, *, P < 0.05 compared with baseline. C: application of the GAT-1 antagonist NO-711 (3 μM) causes a large increase in the tonic current that is fully reversed by addition of TPMPA (100 μM). D: an example experiment showing that application of the GAT-2/3 antagonist SNAP-5114 (50 μM) has no effect on the tonic current.

Nonvesicular, Ca²⁺-independent GABA release can occur via reversal of the GABA transporter GAT-1 under normal, nonpathological conditions (Wu et al. 2007). GAT-1 is localized to amacrine cell processes surrounding BC terminals (S. M. Jones, D. N. Furness, and M. J. Palmer, unpublished observations); however, inhibition of GAT-1 with the nontransportable antagonist NO-711 significantly enhances rather than reduces the tonic GABA_CR current (Hull et al. 2006) (Fig. 3C). To investigate a possible role for GAT-2 or GAT-3 in GABA release, we applied 1-[2-[tris(4-methoxyphenyl)methoxy]ethyl]-(s)-3-piperidinecarboxylic acid (SNAP)-5114 (50 μM), but this had no significant effect on the amplitude of the tonic current (baseline: −50 ± 7 pA; SNAP-5114: −65 ± 13 pA, n = 6; Fig. 3D). GAT-2 and GAT-3 therefore do not appear to be involved in regulating the tonic GABA_CR current, either via GABA uptake or release.

To confirm that the tonic GABA_CR current is activated by GABA rather than another amino-acid such as taurine, we examined the effect of increasing the intracellular GABA concentration by inhibiting GABA transaminase with vigabatrin (γ-vinyl GABA). Vigabatrin (500 μM) caused a significant increase in the tonic current (from −45 ± 5 to −179 ± 35 pA, n = 5, P < 0.05) that was fully reversed by subsequent application of picrotoxin (100 μM) or TPMPA (100 μM; −23 ± 6 pA, n = 4, P < 0.05; Fig. 4, A and D). Vigabatrin potentiated the tonic current without affecting the kinetics of GABA_AR-mediated IPSCs (n = 5; Fig. 4B), a similar effect to that of NO-711 (Palmer 2006). Furthermore, in the presence of Ca²⁺-free solution to inhibit vesicular release, vigabatrin (500 μM) caused a similarly large increase in the tonic current (from −36 ± 7 to −176 ± 39 pA, n = 5, P < 0.05) that was again fully reversed by picrotoxin (100 μM; −16 ± 2 pA, n = 4, P < 0.05; Fig. 4, C and D). In cultured hippocampal neurons, vigabatrin increases the tonic GABA_AR current by inducing reversal of GABA transporters (Wu et al. 2003). We therefore investigated the effect of inhibiting GAT-1 following potentiation of the tonic current by vigabatrin (500 μM). NO-711 (3 μM) caused a further potentiation of the tonic current (from −165 ± 28 to −245 ± 32 pA, n = 5, P < 0.05) that was returned to baseline by picrotoxin (200 μM; −52 ± 8 pA, n = 5, P < 0.05; Fig. 4E). Therefore even under conditions of increased intracellular GABA, GABA release does not occur via reversal of GAT-1.

FIG. 4.

Inhibition of GABA transaminase strongly potentiates the tonic GABA_CR current. A: an example experiment showing that under normal conditions, application of the GABA transaminase inhibitor vigabatrin (500 μM) causes a large increase in the tonic current that is fully reversed by the addition of picrotoxin (100 μM). B, left: peak-scaled average IPSCs from the recording in A, generated from spontaneous events occurring before and during vigabatrin application. Right: vigabatrin did not affect the kinetics of the IPSCs (IPSC decay expressed as the weighted time constant of a biexponential fit; n = 5). C: an example experiment showing that vigabatrin (500 μM) also caused a large increase in the tonic current in the presence of Ca²⁺-free extracellular solution, which was reversed by picrotoxin (100 μM). D: average data for vigabatrin application in normal (n = 5) and Ca²⁺-free (n = 5) conditions. E: an example experiment showing that application of NO-711 (3 μM) in the presence of vigabatrin (500 μM) further potentiates the tonic current, which is subsequently inhibited by picrotoxin (200 μM).

The tonic current in BC terminals is therefore highly sensitive to both the intra- and extracellular GABA concentration. However, as taurine is known to act as an agonist at GABA_CRs (Ochoa-de la Paz et al. 2008; Pan et al. 2005) and to be present in the retina (Lombardini 1991), we investigated the effect of taurine application on the tonic current. Taurine (1 mM) evoked an increase in the tonic current (from −49 ± 10 to −128 ± 18 pA, n = 5, P < 0.01) that was fully reversed by addition of TPMPA (100 μM; −24 ± 6 pA, n = 3, P < 0.05; Fig. 5, A and C). To determine the potential contribution of endogenous taurine to the tonic current, we applied guanidinoethylsulphonate (GES) to inhibit taurine uptake. GES (100 μM) evoked a very small increase in the tonic current (from −29 ± 2 to −41 ± 5 pA, n = 5, P < 0.05; Fig. 5, B and C). A higher concentration of GES (1 mM) evoked a slightly larger increase in the tonic current (from −35 ± 6 to −68 ± 9 pA, n = 14, P < 0.01) and also caused a significant reduction in the frequency of IPSCs (from 12.3 ± 2.3 to 1.4 ± 0.5 Hz, n = 14, P < 0.01) and a prominent increase in the current variance (Fig. 5D). The large current variance in GES was reduced by application of bicuculline (50 μM; from 21.8 ± 3.6 to 9.6 ± 1.8 pA², n = 7, P < 0.01; Fig. 5D), consistent with GES acting as a high-affinity, low-efficacy agonist at GABA_ARs (Mellor et al. 2000). The increase in the tonic current evoked by GES (1 mM) was not significantly inhibited by bicuculline but was reduced by picrotoxin (100 μM) or TPMPA (100–500 μM; from −75 ± 12 to −18 ± 4 pA, n = 11, P < 0.01; Fig. 5D). It was noted that an unusually high concentration of TPMPA was required to maximally inhibit the tonic current in the presence of GES, suggesting that GES may also be acting as a high-affinity, low-efficacy agonist at GABA_CRs. To investigate this possibility, GES was applied locally to recorded BC terminals by pressure ejection in the presence of bicuculline (50 μM in the bath and ejection pipette). GES application (1 mM, 0.1–1.0 s) evoked a small inward current (−12.8 ± 3.1 pA) that was inhibited by picrotoxin (100 μM; −1.0 ± 0.3 pA, n = 3; Fig. 5E). This rapid, localized effect is consistent with a direct action of GES on GABA_CRs in BC terminals, which is likely to account for the increase in the tonic current evoked by bath application of GES. The limited effect of GES (100 μM) indicates that endogenous taurine is unlikely to be a significant activator of the tonic current. In addition, it is inconsistent with the tonic current being activated via reversal of taurine transporters, which can also transport GABA (Tomi et al. 2008).

FIG. 5.

The tonic GABA_CR current is activated by exogenous taurine. A: application of taurine (1 mM) evokes an increase in the tonic current that is reversed by addition of TPMPA (100 μM). B: application of guanidinoethylsulphonate (GES, 100 μM) evokes a very small increase in the tonic current, which is inhibited by picrotoxin (100 μM). C: mean data for taurine (1 mM, n = 5) and GES (100 μM; n = 5) application. D: example traces from an experiment showing the effect of GES (1 mM) application. The large current variance in GES was inhibited by bicuculline (50 μM), and the tonic current was inhibited by TPMPA (100 μM). E: local pressure application of GES (1 mM, 1.0 s, 10 psi) in the presence of bicuculline (50 μM) evoked a small inward current (average of 10 GES applications at 30-s intervals). Both the tonic curent and the GES-evoked current were inhibited by picrotoxin (100 μM; average of 7 GES applications at 30-s intervals).

It has been proposed that high-affinity extrasynaptic GABA receptors may be activated by ambient GABA as GABA transporters cannot lower the extracellular GABA concentration below a value determined by their ionic stoichiometry (Cavelier et al. 2005; Richerson and Wu 2003). If the GABA_CR tonic current in BC terminals is activated by ambient GABA, we reasoned that inhibition of GAT-1 in the absence of vesicular release would be unlikely to produce the large increase in the tonic current that occurs under normal conditions (Fig. 3C). We therefore tested the effect of applying NO-711 (3 μM) following removal of extracellular Ca²⁺. NO-711 caused a significant increase in the tonic current (from −49 ± 5 to −178 ± 23 pA, n = 5, P < 0.01) that was fully reversed by the addition of TPMPA (100 μM; −33 ± 1 pA, n = 3; Fig. 6, A and B). The magnitude of the tonic current increase evoked by NO-711 in Ca²⁺-free solution (−129 ± 26 pA) was very similar to that evoked in interleaved experiments in normal Ca²⁺ solution (−122 ± 7 pA, n = 4; Fig. 6B). In addition, in two BC terminals in which vesicular release had been blocked by concanamycin A (4.6 μM; IPSC frequency <0.2 Hz), the tonic current was strongly potentiated by NO-711 (Fig. 6C). These results are inconsistent with activation of the tonic GABA_CR current by ambient GABA and instead indicate an active, nonvesicular mechanism for GABA release.

FIG. 6.

The tonic GABA_CR current is unlikely to be activated by ambient GABA. A: I_M vs.time (top) and example current traces (bottom) from an experiment showing the potentiating effect of NO-711 (3 μM) in the presence of Ca²⁺-free extracellular solution. The increase in the tonic current was reversed by TPMPA (100 μM). B: The increase in the tonic current evoked by NO-711 (3 μM) was similar in control (n = 4) and Ca²⁺-free (n = 5) conditions. C: Application of NO-711 (3 μM) also caused a large increase in the tonic current in BC terminals incubated in concanamycin A (4.6 μΜ), which was reversed by addition of TPMPA (100 μM).

We therefore investigated two potential nonvesicular, Ca²⁺-independent mechanisms that have been shown to mediate glutamate release from astrocytes: hemichannels (uncoupled gap junctions) (Ye et al. 2003) and large-pore P2X₇ receptors (Duan et al. 2003). Under normal conditions, application of the hemichannel inhibitor carbenoxolone (100–200 μM; n = 4), the nonselective P2 receptor antagonist pyridoxalphosphate-6-azophenyl-2′,4′-disulphonic acid (PPADS, 50–100 μM; n = 3), or the P2X₇ receptor antagonist Brilliant Blue G (BBG, 1 μM; n = 4) had no significant effect on the amplitude of the tonic current in BC terminals (Fig. 7, A and B). Subsequent application of TPMPA (50–100 μM) or picrotoxin (50–100 μM) confirmed that tonic GABA_CR current was present in these recordings (Fig. 7, A and B). The hemichannel and P2X₇ receptor antagonists were also tested in Ca²⁺-free extracellular solution (with 3.5 mM Mg²⁺) to inhibit vesicular release. Application of carbenoxolone (100–200 μM; n = 4) or BBG (1 μM; n = 4) alone, or in combination with PPADS (50–100 μM; n = 4), had no effect on the tonic current in Ca²⁺-free conditions (Fig. 7, C and D). GABA release therefore appears to occur via a nonvesicular, Ca²⁺-independent but as yet unidentified mechanism to activate the tonic GABA_CR current in BC terminals.

FIG. 7.

GABA is not released via hemichannels or P2X₇ receptors. A: an example experiment showing the lack of effect of the P2X₇ receptor antagonist Brilliant Blue G (BBG, 1 μM) on the tonic current, which was subsequently reduced by picrotoxin (100 μM). B: mean data for the application of carbenoxolone (Carb, 100–200 μM, n = 4), BBG (1 μM, n = 4), and PPADS (50–100 μM, n = 3) under normal conditions. C: an example experiment showing the lack of effect of carbenoxolone (100 μM) in the presence of Ca²⁺-free extracellular solution, with subsequent inhibition of the tonic current by picrotoxin (100 μM). D: mean data for the application of carbenoxolone (100 μM, n = 4), BBG (1 μM, n = 4), and carbenoxolone plus BBG and PPADS (PP, 50–100 μM, n = 4) under Ca²⁺-free conditions.

DISCUSSION

To determine the source of GABA that activates the tonic GABA_CR current in goldfish BC terminals, we investigated the effect of inhibiting vesicular release. As expected, this strongly reduced the frequency of spontaneous GABA_AR IPSCs but, somewhat surprisingly, did not inhibit the tonic GABA_CR current. By contrast, glutamate-evoked GABAR currents, comprising both GABA_AR and GABA_CR components, were dependent on the vesicular release of GABA. These results strongly suggest that evoked and tonic GABA_CR currents in BC terminals are activated by GABA originating from vesicular and nonvesicular sources, respectively. BC terminals also exhibit GABA_CR-mediated miniature IPSCs (Palmer 2006), which were not apparent following inhibition of vesicular release. However, as it is difficult to reliably discriminate these small, slow events from the tonic current noise, it was not possible to quantify this effect. The partial, transient reduction in the tonic GABA_CR current on exposure to Ca²⁺-free solution may reflect a contribution of vesicular GABA release to the tonic current that, when removed, is compensated for by upregulation of the nonvesicular mechanism.

By comparison, tonic GABA_AR currents in the brain are, in some cases, dependent on vesicular GABA release but in other cases are activated by GABA from nonvesicular sources including ambient GABA, GABA transporter reversal, and unidentified nonvesicular mechanisms (Brickley et al. 1996; Bright et al. 2007; Glykys and Mody 2007b; Rossi et al. 2003; Wall and Usowicz 1997; Wu et al. 2006, 2007). Furthermore, tonic NMDA receptor (NMDAR) currents in hippocampal neurons are activated by glutamate released via a Ca²⁺-independent, nonvesicular mechanism (Cavelier and Attwell 2005; Jabaudon et al. 1999; Le Meur et al. 2007). The activation of tonic currents by nonvesicular neurotransmitter release may therefore be a general feature of CNS function.

It is likely that evoked and tonic GABA_CR currents in BC terminals are mediated by distinct populations of receptors localized to synaptic and extrasynaptic sites, respectively. GABA_CRs have been immunolocalized within synapses in BC terminals (Fletcher et al. 1998; Koulen et al. 1998) but are also present throughout the extrasynaptic membrane (Jones, Furness and Palmer, unpublished observations). Tonic GABA_AR and NMDAR currents are believed to be mediated by peri- or extrasynaptic receptors that have a distinct subunit composition from the synaptic receptors that mediate phasic currents (Farrant and Nusser 2005; Glykys and Mody 2007a; Le Meur et al. 2007). Similarly, synaptic and extrasynaptic GABA_CRs in BC terminals may differ in their ρ subunit composition. For example, perch ρ1 homomers have a higher GABA sensitivity, slower deactivation kinetics, and lower single-channel conductance than receptors containing ρ2 (Pan et al. 2006; Qian et al. 1998; Zhu et al. 2007), properties that would make them particularly suitable for mediating the tonic GABA_CR current.

A high concentration of exogenous taurine was capable of activating the tonic GABA_CR current in BC terminals, similar to the activation by taurine of a tonic GABA_AR current in the thalamus (Jia et al. 2008). However, it is likely that GABA is the major activator of the spontaneous tonic current in BC terminals due to the large potentiation of the tonic current that results from inhibiting either the uptake or breakdown of GABA and the very minor effect of inhibiting taurine uptake. The observed action of the taurine transporter inhibitor GES as a GABA_AR and GABA_CR agonist means that it should be used with caution in experiments where GABAR activation could cause confounding results (Mellor et al. 2000). Taurine has been shown to biphasically modulate the GABA responses of human ρ1 receptors expressed in Xenopus oocytes, with inhibition at taurine concentrations exceeding 0.3 mM (Ochoa-de la Paz et al. 2008). Our observed increase in the tonic current evoked by 1 mM taurine may reflect differences in the sensitivity to taurine and/or GABA of heterologously expressed ρ1 receptors and native GABA_CRs.

GABA release via either GABA or taurine transporter reversal does not appear to be involved in generating the tonic GABA_CR current in BC terminals. Even under conditions of increased intracellular GABA, GAT-1 continues to operate in the forward direction to lower the extracellular GABA concentration. Ambient extracellular GABA has been proposed to activate extrasynaptic receptors mediating tonic currents (Cavelier et al. 2005; Richerson and Wu 2003), but we found that inhibition of GAT-1 in the absence of vesicular GABA release produced an increase in the tonic current of similar magnitude to control conditions. These results are most consistent with the tonic GABA_CR current being activated by GABA that is continuously released via a Ca²⁺-independent, nonvesicular mechanism and removed via the activity of GAT-1.

Nonvesicular glutamate release occurring via P2X₇ receptors can evoke a tonic NMDAR current in hippocampal pyramidal neurons (Fellin et al. 2006). However, inhibition of P2X₇ receptors had no effect on the tonic GABA_CR current in BC terminals. Similarly, inhibition of hemichannels was without effect. The mechanism of GABA release, as well as its cellular source, therefore remain to be ascertained. Nonvesicular GABA release may occur from amacrine cells and/or from retinal astrocytes or Müller cells. Müller cells are known to release neurotransmitters such as ATP and possibly glutamate, leading to changes in ganglion cell activity (Newman 2004; Newman and Zahs 1998). Glial cells are also likely to be the source of nonvesicular glutamate release in the hippocampus because tonic NMDAR currents are potentiated by inhibition of the glial enzyme glutamine synthetase (Cavelier and Attwell 2005; Jabaudon et al. 1999; Le Meur et al. 2007).

In addition to determining the source and mechanism of release for GABA, it will be important to establish how the tonic GABA_CR current is regulated. For example, it may be regulated by changes in the rate of GABA release, the rate of GABA uptake by GAT-1, and/or by changes in the number or properties of BC terminal GABA_CRs. When expressed in cell lines, GABA_CR subunits are internalized in response to activation of protein kinase C (PKC) (Filippova et al. 1999; Kusama et al. 2000), which is concentrated in the axon terminals of certain types of BC (Greferath et al. 1990; Suzuki and Kaneko 1990). Activation of PKC or metabotropic receptors linked to the phospholipase C pathway causes a reduction in GABA_CR currents in rat BCs (Feigenspan and Bormann 1994b). Similarly, the activity of GAT-1 can be modulated on a time scale of minutes by phosphorylation-induced insertion or removal of transporters from the cell membrane (Beckman et al. 1999; Deken et al. 2003; Whitworth and Quick 2001).

Tonic GABA_AR currents in the brain regulate a variety of neuronal processes, including synaptic integration and network excitability, via their effects on membrane conductance (Farrant and Nusser 2005). In BC terminals, the tonic GABA_CR current regulates the ability of the terminals to fire Ca²⁺-dependent action potentials (Hull et al. 2006) and is likely to operate on a slow time scale to regulate BC output by setting the responsiveness of the terminal to somatic depolarization and synaptic feedback mechanisms. By contrast, the relatively rapid and transient inhibition resulting from synaptically evoked GABA_CR activation may be important for dynamically limiting BC output during light responses (Lukasiewicz and Werblin 1994; Zhang and Slaughter 1995).

In conclusion, we have found that different forms of GABA_CR-mediated inhibition in BC terminals are evoked by vesicular and by nonvesicular GABA release. These two forms of inhibition are likely to be independently activated and regulated and to have distinct functions, further increasing the diversity of mechanisms that regulate BC terminal output. The specific roles and relative importance of these mechanisms in retinal processing within the IPL remain to be determined.

GRANTS

This work was funded by a Medical Research Council (MRC) Career Development Award, a MRC New Investigator Research Grant, a MRC Studentship, and a Keele University Dandelion Award.