GABAC responses were recorded in cultured cone-driven horizontal cells from the catfish retina using the patch clamp technique. At a holding potential of −49 mV, a bicuculline-resistant inward current (IGABA) was observed when 10 μM GABA was applied. The amplitude of IGABA increased as the extracellular Ca2+ ([Ca2+]o) was increased. Concentration–response curves of IGABA at 2.5 and 10 mM [Ca2+]o had similar EC50 (3.0 and 3.1 μM) and Hill coefficients (1.54 and 1.24). However, the maximal response estimated at 10 mM [Ca2+]o was larger than the maximal response at 2.5 mM [Ca2+]o. Increasing Ca influx through voltage-gated Ca channels and the resulting rise in the intracellular Ca2+ concentration had no effects on IGABA. However, IGABA was inhibited by extracellular divalent cations, with the following order of the inhibitory potency: Zn2+ > Ni2+ > Cd2+ > Co2+. The inhibitory action of Zn2+ on the [Ca2+]o-dependent IGABA increase was noncompetitive. The action of [Ca2+]o on IGABA was mimicked by Ba2+ or Sr2+. These results demonstrate that the extracellular domain of GABAC receptors has two functionally distinct binding sites represented by Ca2+ (facilitation) and Zn2+ (inhibition). Since [Ca2+]o and [Zn2+]o change into the opposite direction by light, it seems likely that they modify cooperatively the efficacy of the positive feedback consisting of the GABAC receptor.
The outer plexiform layer, the first synaptic layer of the vertebrate retina, is where signals transduced in photoreceptors are transmitted to bipolar cells. Photoreceptor-bipolar transmission is modulated by the feedback synapse from horizontal cells to photoreceptors. The input–output relation of these synapses has been studied in the tiger salamander and other animals (see Wu, 1994). The transmission at these synapses may be more complex than originally thought, based on the modulatory actions of divalent cations. For example, Zn2+ has been shown to be coreleased with transmitters from the synaptic vesicles of photoreceptors (Wu et al., 1993; Dong and Werblin, 1995).
It has been reported that the Ca2+ concentrations in the inner and outer plexiform layers are changed by illumination (Livsey et al., 1990; Gallemore et al., 1994). These Ca2+ concentration changes are thought to result primarily from Ca influx accompanying transmitter release. We have shown that the extracellular Ca2+ concentration in the inner plexiform layer is important for determining the activities of nicotinic ACh receptors in retinal ganglion cells (Kaneda et al., 1995). Modulation of known chemical receptors by the extracellular Ca2+ concentration in the outer plexiform layer has not been reported.
The GABAC receptor was initially identified in horizontal cells of the catfish retina (Qian and Dowling, 1993). Since the resting potential of the horizontal cell is ∼−80 mV ([K+]o = 5 mM), and the reversal potential of the GABAC response (ECl) of the intact horizontal cell is ∼−30 mV (Takahashi et al., 1995a), GABA induces a depolarization that enhances GABA release from horizontal cells (Schwartz, 1987). In this sense, GABAC receptors in catfish horizontal cells make a positive auto-feedback loop, the concept proposed by Stockton and Slaughter (1991) and Kamermans and Werblin (1992) for GABAA receptors of the amphibian horizontal cells. These GABAC auto-receptors are thought to contribute to accelerating light-evoked responses (Takahashi et al., 1995a). In the present experiments, we examined the actions of extracellular Ca2+ and other divalent cations on the GABAC receptor. We found that the extracellular domain of the GABAC receptor had two divalent cation binding sites with contrasting functions; i.e., a facilitative Ca2+ binding site and an inhibitory Zn2+ binding site. Based on the present findings, we speculate as to the functional role of GABAC receptors in horizontal cells.
Materials And Methods
Our methods were previously described in detail (Takahashi et al., 1995a). Briefly, preparations were made from a catfish that had been dark adapted for more than 3 h. Retinas were incubated for 40 min at room temperature in a solution containing (mM): 125 NaCl, 1 NaH2PO4, 2.6 KCl, 1 Na pyruvate, 10 Glucose, 5 EGTA, 10 HEPES, and 3–4 U/ml papain (Worthington Biochemical Co., Freehold, NJ) together with 5 μM of its activator, L-cysteine (pH adjusted to 7.0 with NaOH). Isolated cell preparations contained both rod- and cone-driven horizontal cells, easily distinguished by their characteristic morphology under an inverted microscope (TMD; Nikon, Tokyo, Japan). Dissociated cells were kept in a one-to-one mixture of L-15 medium (Life Technologies, Inc., Grand Island, NY) and a culture medium containing (mM): 56.5 NaCl, 0.5 MgCl2, 0.3 MgSO4, 1.5 CaCl2, 5 glucose, 10 HEPES, and 10 mg/liter BSA, pH adjusted to 7.6 with NaOH, for 4–7 d at 10°C. As the rod-driven horizontal cells died in 2–3 d, only the cone-driven horizontal cells were used for recordings.
Patch pipettes were fabricated from borosilicate capillaries (GC-150F-10; Clark Electromedical Instruments, Reading, UK) using a two-stage electrode puller (PP-83; Narishige, Tokyo, Japan). To block the voltage-gated K conductances, pipettes were filled with a Cs-rich internal solution containing (mM): 120 CsCl, 1 NaCl, 2 MgCl2, 1 CaCl2, 10 EGTA, 10 HEPES, pH adjusted to 7.4 with CsOH. The resistance of the electrode filled with the Cs-rich internal solution was 8–12 MΩ. To reduce stray capacitance, the outer wall of the pipette except for the tip was coated with Apiezon wax (Apiezon Products Ltd., London, UK) and the residual capacitance was compensated for electrically. The reference electrode was a Ag-AgCl wire connected to the dish by a 140 mM NaCl agar bridge. Recordings were made using a patch clamp amplifier (CEZ-2300; Nihon Kohden, Tokyo, Japan). The series resistance was not compensated, but the error due to the series resistance was less than a few millivolts, as the peak current amplitude did not exceed 100 pA. Current and voltage signals were monitored on an oscilloscope, recorded on a chart recorder (WR7700; Graphtech, Yokohama, Japan) and stored on a video tape recorder (HV-S21; Mitsubishi, Tokyo, Japan) through a PCM processor (frequency characteristics 0–20 kHz ± 1 dB, PCM-501ES; Sony, Tokyo, Japan). Signals were digitized directly or from the tape recorder by a 12 bit A/D converter (ADX-98H; Canopus, Kobe, Japan) attached to a personal computer (PC9801RX; NEC, Tokyo, Japan). Command voltages were generated by a D/A converter (DAX-98; Canopus) driven by the same computer. The holding potential was corrected for the liquid junction potentials measured by a previously reported method (Kaneko and Tachibana, 1986a).
To block voltage-gated cation conductance and the current carried via GABA transport (Haugh-Scheidt et al., 1995; Takahashi et al., 1995b), cells were superfused with Na-free solution containing (mM): 127.6 N-methyl-d-glucamine (NMDG), 1 MgCl2, 2.5 CaCl2, 15 glucose, and 10 HEPES, adjusted to pH 7.8 with 1 N HCl (final Cl− concentration of 124.6 mM was calculated from the amount of 1 N HCl, MgCl2, and CaCl2). The Y-tube system (Suzuki et al., 1990) was used for GABA application to expedite (100–200 ms) solution changes. Repeated application of 10 μM GABA every 3 min induced a response augmentation that reached a steady level in ∼10 min. We therefore used the data collected after IGABA had reached the steady state (>10 min). All experiments were carried out at room temperature. Unless otherwise specified, experiments were carried out in at least three different cells.
Effects of the Extracellular Ca Concentration on IGABA
When 10 μM GABA was applied to a cone-driven horizontal cell at a holding potential (Vh) of −49 mV, an inward current (IGABA) was induced. The current began to flow without a detectable delay and peaked within 2 s (Fig. 1,A, left). The GABA-induced response showed a biphasic decay during prolonged application (∼2 min). Within 30 s, the amplitude of IGABA was reduced to 55% of the peak value, while only a slight amplitude reduction was seen after 30 s. IGABA recorded in the present experiment consisted exclusively of current passing through GABAC channels, since the response blocked by 100 μM picrotoxin (not shown) was not affected by addition of 100 μM bicuculline (Fig. 1 D) as reported previously (Takahashi et al., 1995a).
The IGABA amplitude increased monotonically when the extracellular Ca concentration ([Ca2+]o) was increased from 0.1 to 10 mM (Fig. 1,C). At 0.1 mM [Ca2+]o, the amplitude of IGABA was 69 ± 8% (mean ± SD, n = 11) that of the control ([Ca2+]o = 2.5 mM; Fig. 1,A). IGABA recorded in a nominally Ca2+-free solution was identical to that recorded at 0.1 mM [Ca2+]o. However, such a solution may contain Ca2+ in the micromolar range (Kurahashi, 1990). Because of the large increase in holding current and the decrease in the IGABA amplitude, [Ca2+]o lower than that in the nominally Ca2+-free solution was not assessed in this study. On the other hand, when [Ca2+]o was increased from 2.5 to 10 mM, the amplitude of IGABA rose to 129 ± 12% (n = 11) that of the control (Fig. 1 B). The response time course was the same under these two conditions. The [Ca2+]o-dependent IGABA modulation was reversible; IGABA returned to the control value 30–40 s after reverting to the control medium.
The [Ca2+]o-modulated current represented IGABA through the GABAC receptor channel, since the amplitude of IGABA was unaffected by 100 μM bicuculline (Fig. 1 D), confirming that the GABAA receptor was not involved.
Concentration–Response Relationships for IGABA at 2.5 and 10 mM [Ca2+]o
To understand the mechanism of the Ca2+ and GABA receptor interaction, the GABA concentration–response relation was examined in both 2.5 and 10 mM [Ca2+]o (Fig. 2). Under both [Ca2+]o conditions, the least effective concentration of GABA was ∼500 nM. With an increase in the GABA concentration, the response amplitude increased sigmoidally and reached an apparent saturation at 100 μM. The two curves were similar except that the saturated amplitude at 10 mM [Ca2+]o was 36% larger than that at 2.5 mM [Ca2+]o. These curves were fitted by the equation I = Imax/(1 + [EC50/C]n), where Imax is the maximal response, C is the concentration of GABA, n is the Hill coefficient, and EC50 is the concentration of GABA that produced a half-maximal IGABA. The two curves were described by a similar Hill coefficient and EC50; at 2.5 mM [Ca2+]o, the Hill coefficient was 1.54, EC50 3.0 μM (n = 7), while at 10 mM [Ca2+]o the Hill coefficient was 1.24, EC50 3.1 μM (n = 11). As described above, the only difference between the two curves was in the Imax values; 1.22 for 2.5 mM [Ca2+]o and 1.66 for 10 mM [Ca2+]o.
Effects of Ca Influx on IGABA
There is a possibility that GABAC receptor activity is modulated via changes in the intracellular Ca2+ concentration. Therefore, in the present experiments, Ca influx produced by the L-type Ca current (step to −9 mV, Vh = −49 mV, 10 s in duration) (Shingai and Christensen, 1983) was increased to study the effects of the intracellular Ca2+ concentration on the amplitude of IGABA. The IGABA amplitude was not augmented when additional Ca influx was induced before or during an application of 10 μM GABA, indicating that the action of [Ca2+]o on IGABA is due to Ca2+ binding to an extracellular allosteric site on the GABAC receptor.
To clarify the mechanism underlying the facilitation of IGABA by Ca2+, we attempted to record single channel activity in outside-out patches excised from horizontal cells. In 12 successfully excised patches, however, we were unable to detect any single-channel activity. This failure is probably attributable to the extremely low density of GABAC receptor channels. According to a reported noise analysis, the single channel conductance of the GABAC receptor channel is ∼8 pS (Takahashi et al., 1995a), which generates a single channel current of 0.4 pA at −50 mV. The maximal IGABA was ∼100 pA, which is consistent with 250 channels. The surface area of an average horizontal cell (60–100 μm in diameter) is ∼8,000 μm2 (assuming a flat disk 100 μm in diameter), and has an estimated density of 0.03 channels/μm2. This value suggests that the channel density is too low for detection with a patch pipette that has a 1–2 μm opening.
Effects of Divalent Cations on IGABA
It has been shown that [Ca2+]o modulates the activity of the nicotinic ACh receptor in a concentration-dependent manner, and that various divalent cations also affect the activity of this receptor (Mulle et al., 1992; Kaneda et al., 1995). Furthermore, an inhibitory action of divalent cations on the GABAA receptor was also reported in the retina (Kaneko and Tachibana, 1986b). As GABAC receptor activity was found to be dependent upon [Ca2+]o, we also studied the effects of other divalent cations on IGABA. The response to 10 μM GABA was completely blocked when 4 mM CoCl2 was added to the external solution (Fig. 3,A). This inhibitory effect appeared immediately and was reversible. The magnitude of inhibition was sigmoidally dependent upon the Co2+ concentration (Fig. 3 B); the inhibitory effect of Co2+ became apparent at 40 μM and the response to GABA was completely blocked at 4 mM. The concentration– inhibition curve was fitted by the equation I = 1 − Imax/(1 + [IC50/C]n), where I represents the normalized response amplitude in the presence of Co2+, Imax is the control response amplitude (1.0) in the absence of Co2+, C is the concentration of Co2+, n is the Hill coefficient, and IC50 is the concentration of Co2+ that inhibited IGABA to 0.5. Based on the data obtained from 11 cells, we estimated the Hill coefficient to be 1.24 and IC50 to be 284 μM.
We examined the effects of various divalent cations on IGABA and classified the tested divalent cations into three groups: those showing an inhibitory effect like Co2+, those showing a facilitatory action like Ca2+, and those that had no effect. To the first group belonged Zn2+, Ni2+, Cd2+, and Co2+. The order of the inhibitory potency of these divalent cations (at 100 μM) on IGABA was Zn2+ > Ni2+ > Cd2+ > Co2+ (Fig. 4,A). The second group included Ba2+ and Sr2+. When extracellular Ca2+ was replaced with Ba2+ or Sr2+, the amplitude of IGABA was increased in a concentration-dependent manner (Fig. 4 B). Mg2+ and Mn2+ were neither facilitatory nor inhibitory; the amplitude of IGABA did not change when Mg2+ was excluded from the solution or 1 mM Mn2+ was added to the perfusate.
Inhibitory Actions of Zn2+ on [Ca2+]o-dependent Facilitation of IGABA
Of the divalent cations affecting IGABA, Ca2+ and Zn2+ have the physiological potential to affect the outer plexiform layer via changes in their extracellular concentrations (Gallemore et al., 1994; Wu et al., 1993). It has also been demonstrated that the activity of GABAC receptors is inhibited by Zn2+ (Dong and Werblin, 1995). We attempted to determine whether Ca2+ and Zn2+ compete for the same binding site on the GABAC receptor and found that [Ca2+]o-dependent IGABA modulation was observed in the presence of 5 μM Zn2+. However, the IGABA amplitude was reduced by ∼30% at all [Ca2+]o (Fig. 5). When 30 μM Zn2+ was added to the perfusate, the minimal amplitude of IGABA was greatly reduced and became nearly undetectable even in the presence of 1 mM [Ca2+]o. At 100 μM Zn2+, there was no current even at 10 mM [Ca2+]o. These observations demonstrate that the inhibitory action of Zn2+ on the [Ca2+]o-dependent facilitation of GABAC receptors is noncompetitive.
Mechanisms of Ca-dependent GABAC Receptor Modulation
Our experiments demonstrate the activity of GABAC receptors to be regulated by extracellular Ca2+. The concentration–response curve revealed an increase in Imax with no apparent change in either EC50 or the Hill coefficient when [Ca2+]o was increased from 2.5 to 10 mM. Thus, Ca binding to an extracellular site on GABAC receptors increases the number of active GABAC receptors but does not change the affinity of GABA for GABAC receptors. A [Ca2+]o-dependent augmentation of the ACh response in neuronal nicotinic ACh receptors was explained by an increase in the maximal probability of channel opening (Mulle et al., 1992). These investigators showed that Ca binding to an extracellular site increased the opening frequency of ACh channels without affecting the actual channel kinetics. In addition, the effects of extracellular Mg2+, Ba2+, and Sr2+ in GABAC receptors are similar to those exerted by these ions on neuronal nicotinic ACh receptors. Therefore, the increase in active GABAC receptors produced by [Ca2+]o may be explained by an increase in the opening frequency, as proposed for neuronal nicotinic ACh receptors.
Differentiation of Ca2+-binding Site from Zn2+-binding Site
In the present experiments, IGABA was modulated by divalent cations in two different ways. An apparent inhibition of IGABA was observed with Zn2+, Ni2+, Cd2+, and Co2+ while Ba2+ and Sr2+ mimicked the action of Ca2+. We observed that the action of Zn2+ on the [Ca2+]o-dependent facilitation of GABAC receptors was noncompetitive, indicating that the masking action of Zn2+ on the [Ca2+]o-dependent facilitation of GABAC receptors is not due to occupation of the Ca2+-binding site by Zn2+. In GABAA receptors, there are also two distinct types of GABAA receptor modulation, the inhibitory actions of Zn2+ and other divalent cations (Kaneko and Tachibana, 1986b) and the facilitatory actions of lanthanides (Ma and Narahashi, 1993). Based on detailed studies, lanthanides are thought to exert their facilitatory action through a lanthanide-binding site different from the Zn2+-binding site on GABAA receptors. Therefore, it is probable that GABAC receptors have two functionally contrasting divalent cation binding sites within their extracellular domain, as proposed for GABAA receptors.
Physiological Significance of GABAC Receptor Modulation by [Ca2+]o
The activity of GABAC receptors is augmented at a high [Ca2+]o, as shown in the present experiments, but inhibited in the presence of extracellular Zn2+ (Dong and Werblin, 1995). In light of the potential physiological relevance, we have described herein the distribution and dynamics of Ca2+ and Zn2+ in the outer plexiform layer. It has been shown that a long-lasting increase in [Ca2+]o occurs in the outer plexiform layer (∼0.5 mM), when the feline retina is illuminated (Gallemore et al., 1994). In addition, in the tiger salamander retina, Zn2+ reportedly accumulates in photoreceptor terminals, strongly suggesting corelease of glutamate and Zn2+ (Wu et al., 1993). Since the reversal potential of the GABAC response of the intact horizontal cell is ∼−30 mV (Takahashi et al., 1995a), GABA induces a depolarization of horizontal cells that makes a positive auto-feedback loop (Stockton and Slaughter, 1991; Kamermans and Werblin, 1992). To summarize, there are three events going on in the dark, an increase in GABA release, an increase in [Zn2+]o, and a decrease in [Ca2+]o. These events interact in such a way that the increase in [Zn2+]o and the decrease in [Ca2+]o both work to reduce the effect of GABA release on auto-receptors on horizontal cells.
This work was supported by Grants in Aid from the Ministry of Education, Science and Culture of Japan (07558294, 08458272, 08279240, and 09268232 to A. Kaneko, and 08680893 and 09680820 to M. Kaneda), the Paul Kayser Award in Merit of Retinal Research from the Retina Research Foundation, and Grants in Aid from Keio Gijuku Academic Development Funds to M. Kaneda.
Address correspondence to M. Kaneda, MD, Ph.D., Department of Physiology, Keio University School of Medicine, Tokyo 160, Japan. Fax: 81-3-3359-0437; E-mail: email@example.com