Functional properties of spontaneous EPSCs and non-NMDA receptors in rod amacrine (AII) cells in the rat retina

Abstract

The functional properties of spontaneous, glutamatergic EPSCs and non-NMDA receptors in AII amacrine cells were studied in whole cells and patches from slices of the rat retina using single and dual electrode voltage clamp recording. Pharmacological analysis verified that the EPSCs (E_rev∼0 mV) were mediated exclusively by AMPA-type receptors. EPSCs displayed a wide range of waveforms, ranging from simple monophasic events to more complex multiphasic events. Amplitude distributions of EPSCs were moderately skewed towards larger amplitudes (modal peak 23 pA). Interevent interval histograms were best fitted with a double exponential function. Monophasic, monotonically rising EPSCs displayed very fast kinetics with an average 10–90 % rise time of ∼340 μs and a decay phase well fitted by a single exponential (τ_decay∼760 μs). The specific AMPA receptor modulator cyclothiazide markedly slowed the decay phase of spontaneous EPSCs (τ_decay∼3 ms). An increase in temperature decreased both 10–90 % rise time and τ_decay with Q₁₀ values of 1.3 and 1.5, respectively. The decay kinetics were slower at positive membrane potentials compared to negative membrane potentials (205 mV/e-fold change in τ_decay). Step depolarization of individual presynaptic rod bipolar cells or OFF-cone bipolar cells evoked transient, CNQX-sensitive responses in AII amacrine cells with average peak amplitudes of ∼330 pA. Ultrafast application of brief (∼1 ms) or long (∼500 ms) pulses of glutamate to outside-out patches evoked strongly desensitizing responses with very fast deactivation and desensitization kinetics, well fitted by single (τ_decay∼1.1 ms) and double exponential (τ₁∼3.5 ms; τ₂∼21 ms) functions, respectively. Double-pulse experiments indicated fast recovery from desensitization (τ∼12.4 ms). Our results indicate that spontaneous, AMPA receptor-mediated EPSCs in AII amacrine cells have very fast, voltage-dependent kinetics that can be well accounted for by the kinetic properties of the AMPA receptors themselves.

Fast excitatory synaptic transmission in the central nervous system (CNS) is predominantly mediated by ionotropic glutamate receptors of the AMPA type (reviewed in Dingledine et al. 1999; Jonas & Monyer, 1999). The time course and amplitude of the postsynaptic conductance change can differ substantially between different synapses and it is generally thought that these properties are major determinants of the signal transformation and processing characteristics of specific synapses (reviewed in Geiger et al. 1999). The main postsynaptic factors involved in determining the time course of the postsynaptic conductance change seem to be the deactivation and desensitization kinetics of the AMPA receptors. Both these kinetic parameters are determined by the specific subunit composition of the receptors (reviewed in Dingledine et al. 1999; Jonas & Monyer, 1999). There is strong evidence that the AMPA receptors expressed by interneurons in the neocortex and hippocampus display very fast gating characteristics compared with receptors in principal neurons and, corresponding to this, the postsynaptic conductance change is considerably faster in interneurons compared with principal neurons in the same locations (reviewed in Geiger et al. 1999).

In the mammalian retina, AII amacrine cells constitute an important population of local-circuit interneurons in the network of neurons thought to convey visual signals during scotopic vision (the rod pathway). They receive glutamatergic synaptic input from bipolar cell axon terminals and contact their postsynaptic targets via either electrical or chemical (glycinergic) synapses (reviewed by Massey & Maguire, 1995). Rod bipolar cell axon terminals contact the AII amacrine cells at the arboreal dendrites proximally in the inner plexiform layer and axon terminals of OFF-cone bipolar cells contact the lobular appendages in the distal part of the inner plexiform layer (Kolb & Famiglietti, 1974; Strettoi et al. 1992; Chun et al. 1993). The bipolar input to AII amacrine cells is of particular interest because the synapses made by both rod bipolar cells and OFF-cone bipolar cells are ribbon synapses, where transmitter release is presumed to reach high rates and to be modulated in a continuous manner (von Gersdorff & Matthews, 1999).

Previous work has shown that AII amacrine cells express functional ionotropic glutamate receptors of both the non-NMDA and the NMDA type (Boos et al. 1993; Hartveit & Veruki, 1997). Mørkve et al. (2002) recently demonstrated that the non-NMDA receptors expressed by AII amacrine cells are high-affinity AMPA receptors and found no evidence for high-affinity kainate receptors. The AMPA receptors were linked to channels with significant Ca²⁺ permeability. In order to further our understanding of the synaptic transmission between bipolar cells and AII amacrine cells, we have extended the investigation of the functional properties of non-NMDA-type receptor channels on AII amacrine cells by characterizing spontaneous postsynaptic currents (PSCs). We were particularly interested in determining the time course and amplitude of the corresponding postsynaptic conductance change. For comparison, we have also characterized the kinetic properties of non-NMDA-type glutamate receptor channels in somatic patches, with ultrafast agonist application.

METHODS

General aspects of the methods have previously been described in detail (Hartveit, 1996). Albino rats (4–7 weeks postnatal) were deeply anaesthetized with halothane in oxygen and killed by cervical dislocation (procedure approved under the surveillance of the Norwegian Animal Research Authority). Retinal slices were visualized with a ×40 water immersion objective and infrared differential interference contrast (IR-DIC) videomicroscopy (Axioskop FS; Zeiss, Germany). When filled with intracellular solution, recording pipettes typically had resistances of 4–7 MΩ for recordings in the whole-cell configuration and 6–10 MΩ for recordings from outside-out patches. For most outside-out patch recordings, pipettes were coated with dental wax and fire-polished immediately before use. We also recorded from two nucleated outside-out patches (Mørkve et al. 2002), but the results were not included in the quantitative analysis. The reference electrode (Ag–AgCl wire) was connected to the recording chamber via a solution bridge.

Recordings were carried out either at room temperature (24–27 °C) or at an elevated temperature of 34.0 ± 0.5 °C. In the latter case, an automatic temperature control unit continuously monitored and regulated the temperature at the recording site by heating the external solution and the recording chamber (ATR-4, Quest Scientific, North Vancouver, BC, Canada). The temperature was continuously monitored by thermistors positioned at the inflow and in the recording chamber, as close as possible to the recording site.

Solutions and drugs

The extracellular perfusing solution was continuously bubbled with 95 % O₂–5 % CO₂ and had the following composition (mm): 125 NaCl, 25 NaHCO₃, 2.5 KCl, 2.5 CaCl₂, 1 MgCl₂, 10 glucose, pH 7.4. For recordings of spontaneous PSCs we used an external solution with low Ca²⁺ (0.15 mm) and high Mg²⁺ (3.35 mm). In single electrode recording experiments in the whole-cell configuration, recording pipettes were filled with (mm): 125 potassium gluconate, 8 KCl, 5 Hepes, 1 CaCl₂, 1 MgCl₂, 5 ethylene glycol-O,O′-bis(2-aminoethyl)-N,N,N′,N′-tetraacetic acid (EGTA), 4 disodium adenosine 5′-triphosphate (Na₂ATP), 2 N-(2, 6-dimethylphenylcarbamoylmethyl)triethylammonium bromide (QX-314; Tocris Cookson, Bristol, UK). pH was adjusted to 7.3 with KOH. At the standard holding potential of −60 mV, this solution virtually eliminated the driving force for chloride currents. In experiments where the holding potential was varied in order to examine the reversal potential (E_rev) of currents (whole-cell and outside-out patches), the intracellular solution contained (mm): 125 CsOH, 125 gluconic acid, 15 tetraethylammonium chloride, 5 Hepes, 1 CaCl₂, 1 MgCl₂, 4 NaCl, 5 EGTA, 4 Na₂ATP. In some experiments, 25 μm spermine was added to this intracellular solution. pH was adjusted to 7.3 with CsOH. In experiments with simultaneous dual recording from pairs of electrically coupled cells, the intracellular solution contained (mm): 140 potassium gluconate, 5 Hepes, 1 CaCl₂, 1 MgCl₂, 5 EGTA, 4 Na₂ATP). This solution was also used in some recordings from pairs of cells coupled via chemical synapses. Alternatively, for such recordings the intracellular solution contained (mm): 125 CsCl, 15 tetraethylammonium chloride, 5 Hepes, 1 CaCl₂, 1 MgCl₂, 4 NaCl, 5 EGTA, 4 Na₂ATP. Lucifer yellow was added at a concentration of 1 mg ml⁻¹ to all intracellular solutions for visualization of cells at the end of the recording.

Theoretical liquid junction potentials were calculated with the computer program JPCalcW (Axon Instruments, Union City, CA, USA) and membrane holding potentials were automatically corrected for the liquid junction potentials on-line.

Drugs were added directly to the extracellular solution used to perfuse the slices. All whole-cell recordings were performed in the presence of 10 μm bicuculline methchloride (Tocris Cookson) and 1 μm strychnine (Research Biochemicals, Natick, MA, USA). Although NMDA-evoked responses in AII amacrine cells display fast rundown (Hartveit & Veruki, 1997), the external solution also contained 20–30 μm 3-((RS)-2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP; Tocris Cookson) in most experiments. The concentrations of the other drugs were as follows (supplier in parenthesis): 0.05–100 μm 1-(4-aminophenyl)-3-methylcarbamyl-4-methyl-7,8-methylenedioxy-3,4-dihydro-5H-2,3-benzodiazepine (GYKI 53655; Eli Lilly, Indianapolis, IN, USA), 10 μm 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; Tocris Cookson), 100 μm cyclothiazide (Eli Lilly), 2 μm (2S,4R)-4-methylglutamic acid (SYM 2081; Tocris Cookson), 0.3 μm tetrodotoxin (TTX; Tocris Cookson). Solutions were either made up freshly for each experiment or prepared from concentrated aliquots stored at −20 °C. CNQX and cyclothiazide were first dissolved at 100 mm in dimethylsulfoxide (Sigma, St Louis, MO, USA) and then diluted to the final concentration by sonication.

Fast drug application

Ultrafast application of glutamate was performed according to the description of Jonas (1995; see also Colquhoun et al. 1992). Glutamate was applied from a theta-tube application pipette (nominal septum thickness ∼117 μm; final tip diameter ∼300 μm; Hilgenberg, Malsfeld, Germany). The pipette tip with the outside-out patch was positioned near the interface between control solution and agonist-containing solution continuously flowing out of each barrel, about 100 μm away from the tip of the application pipette. The solution flow rate was maintained by syringe pumps (KDS220; KD Scientific, Boston, MA, USA) controlled by the data acquisition software (see below). Concentration jumps of agonist to the patch were performed by rapidly moving the position of the application pipette and thus the interface between the two solutions. The application pipette was mounted on a piezo actuator (LSS-3100/PZS-100HS; Burleigh Instruments Inc., Fishers, NY, USA) operated by an amplifier (PZ-150M; Burleigh Instruments Inc.) that received square wave voltage pulses from the ITC-16 interface (see below). Before being fed to the amplifier, the voltage pulses were conditioned by an electronic circuit consisting of an RC-filter and an inductive element. Glutamate was dissolved in Hepes-buffered solution containing (mm): 145 NaCl, 2.5 KCl, 2.5 CaCl₂, 1 MgCl₂, 5 hemisodium-Hepes, 10 glucose, pH adjusted to 7.4 with HCl. Agonist pulses were applied every 3–5 s.

The solution exchange time was measured as the change in liquid junction current of an open-tip patch pipette upon a change from normal, Hepes-buffered extracellular solution to the same solution diluted to 10 % with distilled water. The rise time was calculated as the interval between 10 % and 90 % of the peak amplitude relative to baseline and, under optimal conditions, ranged between 150 and 400 μs. To facilitate comparisons with previously published results from other laboratories, we have in some places reported the 20–80 % rise time. The 20–80 % rise time of the solution exchange ranged between 100 and 300 μs. The optimal position of the patch pipette in relation to the application pipette was mapped out at the start of each experiment. For several experiments, the patch was blown away at the end of the recording and the solution exchange time was verified after switching to the 10 % dilute solution.

Data acquisition

Voltage clamp recordings were made with an EPC9/2 amplifier (HEKA Elektronik, Lambrecht, Germany) controlled by Pulse software (HEKA Elektronik). For voltage clamp recordings, cells and patches were generally held at a membrane potential of −60 mV. In current clamp recordings, the fast current clamp feedback circuitry of the EPC9/2 was used to better follow rapid changes in membrane potential. Application of voltage protocols and digital sampling of the analog signals were performed via an ITC-16 interface (Instrutech, Port Washington, NY, USA). For some experiments with simultaneous dual recording, the DAC-stimulus template corresponded to the digitization of a previously recorded action potential (Veruki & Hartveit, 2002). Before sampling, the signal was low-pass filtered (analog 3- and 4-pole Bessel filters in series) with a corner frequency (−3 dB) between 1/12.5 and 1/3 of the inverse of the sampling interval (10–100 μs). Capacitative currents caused by the recording pipette capacitance (C_fast) and the cell membrane capacitance (C_slow) were measured with the automatic capacitance neutralization network feature of the EPC9/2. For each such measurement during dual recording, the test pulse stimuli were sent simultaneously to both amplifiers. The average capacitance in the whole-cell recordings was 14.6 ± 0.4 (s.e.m.) pF (n = 89). Throughout every recording, the series resistance was regularly monitored by applying a series of 10 mV hyperpolarizing voltage pulses (16 ms duration). The C_slow neutralization circuitry of the EPC9/2 was transiently disabled and during dual recordings the stimulus was sent to both amplifiers simultaneously in order to eliminate junctional currents. The average series resistance was 16.5 ± 0.8 MΩ (n = 89). In single recordings, compensation of series resistance was sometimes used. Cells with uncompensated series resistance above 40 MΩ were excluded from analysis.

For establishment of current–voltage (I–V) relationships, sampling was repeated at a series of membrane command potentials. The potential was stepped to a new, constant value before the start of sampling (1–5 s for whole-cell recordings, 250 ms for patch recordings) in order to allow the membrane current to relax to a plateau level.

Analysis

Data were analysed with PulseFit (HEKA Elektronik), PulseTools (HEKA Elektronik), Igor Pro (WaveMetrics, Lake Oswego, OR, USA), DataView (Dr W. J. Heitler, University of St. Andrews, UK), and AxoGraph (Axon Instruments). Spontaneous PSCs were detected with a threshold of 3–8 pA depending on the noise level (MiniAnalysis; Synaptosoft, Decatur, GA, USA) and confirmed by eye. For amplitude and interevent interval analysis, complex (multiphasic; see below) PSCs were considered to consist of overlapping events. In order to determine the amplitude of an individual event riding on the decay phase of a preceding event, the true baseline was estimated by extrapolating the decay phase of the preceding event with a single exponential function. For kinetic analysis, however, we only included well separated, monophasic (see below) PSCs which appeared to rise in a monotonic fashion without visible deviation of the rising phase (cf. Traynelis et al. 1993) and which decayed exponentially. For analysis of averaged PSCs, the number of individual events for each cell ranged between 150 and 950.

The decay time course of individual and averaged PSCs, as well as responses evoked by ultrafast application of agonist, was estimated by curve fitting with exponential functions. For single exponential functions we used the function:

(1)

where I(t) is the current as a function of time, A is the amplitude at time 0, τ is the time constant, and I_ss is the steady-state current amplitude (typically zero). For double exponential functions we used the function:

(2)

where I(t) is the current as a function of time, A₁ and A₂ are the amplitudes at time 0 of the first and second exponential components, τ₁ and τ₂ are the time constants of the first and second exponential components, and I_ss is the steady-state current amplitude (typically zero). Fitting was started 250–450 μs after the peak amplitude (typically 300 μs for PSCs). For double exponential functions the amplitude contribution was calculated as 100 % × (A_x/(A₁ + A₂)). The relative amplitude of fitted exponentials depends on where time 0 is taken to be (e.g. Silver et al. 1996). For both spontaneous PSCs and evoked responses in patches we defined time 0 as the start of the response (determined by eye as the point in time at which the current rose from the baseline noise) (Otis et al. 1996). Temperature coefficients (Q₁₀ values) for decay time constants were calculated from (τ_TL/τ_TH)^10/ΔT, where τ_TL and τ_TH are the decay time constants at lower and higher temperatures, respectively, and ΔT is the temperature difference. For 10–90 % rise time and peak amplitude (A), the respective Q₁₀ values were calculated from (r_10–90%TL/r_10–90%TH)^10/ΔT and (A_TH/A_TL)^10/ΔT.

For measurement of E_rev values, data points of I–V relationships were fitted by fourth- to seventh-order polynomial functions. E_rev was then calculated by interpolation. In order to estimate ion channel rectification, the first derivatives of the polynomial functions obtained by curve-fitting of I–V relationships were used to generate slope conductance (G_slope) vs. voltage relationships. The ratio between the slope conductances at +40 mV and −60 mV (G_{slope, + 40 mV}/G_{slope, −60 mV}) was calculated as a rectification index.

For analysis of voltage-dependent kinetics (PSCs, agonist-evoked responses), plots of the decay time constant vs. membrane potential were fitted with the function:

(3)

where τ is the decay time constant, τ₀ is the time constant at 0 mV, E_m is the membrane potential and H is the change in membrane potential (in mV) for an e-fold increase in τ (a measure of voltage sensitivity; Magleby & Stevens, 1972; Colquhoun et al. 1992).

Data are presented as means ± s.e.m. (n = number of cells or events) and percentages are presented as percentage of control. Statistical analyses were performed using Student's two-tailed t tests (unpaired, unless otherwise stated) and differences were considered significant at the P < 0.05 level. For illustration purposes, most raw data records were low-pass filtered (−3 dB; digital non-lagging Gaussian filter; −3 dB at 1–4 kHz). Unless otherwise noted, the current traces shown in the figures represent individual traces.

RESULTS

Identification of AII amacrine cells in retinal slices

We recorded from over 140 AII amacrine cells in retinal slices (whole-cell and outside-out patch recordings). Cells were visually targeted for recording according to the following criteria: (1) location of the cell body at and across the border between the inner nuclear layer and the inner plexiform layer; (2) medium size of the cell body; and (3) a thick primary dendrite that tapers as it descends into the inner plexiform layer (Fig. 1A; Mørkve et al. 2002; Veruki & Hartveit, 2002). All cells were filled with Lucifer yellow and fluorescence microscopy subsequently allowed visualization of each cell's complete morphology, including lobular appendages in sublamina a and arboreal dendrites in sublamina b of the inner plexiform layer (Mørkve et al. 2002; Veruki & Hartveit, 2002). Characteristic unclamped action currents (Mørkve et al. 2002), occurring both spontaneously (Fig. 1B) and when evoked by 5 mV depolarizing voltage steps (Fig. 1C; 5 ms duration; holding potential −60 mV), were blocked by TTX (300 nm; data not shown), indicating that they depended on activation of voltage-gated Na⁺-channels.

Figure 1. AII (rod) amacrine cells in the rat retinal slice preparation.

A, infrared differential interference contrast videomicrograph of an AII amacrine cell in a retinal slice. Arrow points to the cell body. Scale bar, 10 μm. B, spontaneous activity displaying action currents (the two leftmost of which are marked by arrows) escaping from voltage clamp and smaller, spontaneous postsynaptic currents (PSCs). C, electrophysiological ‘signature’ of an AII amacrine cell whereby brief 5 mV depolarizing steps (upper panel) from holding potential (−60 mV) evoke action currents escaping from voltage clamp (middle panel), recorded shortly after establishing the whole-cell configuration (Control). After a short delay, action currents are completely blocked by the presence of QX-314 in the intracellular solution (lower panel; QX-314). Two overlaid traces in each panel.

To record spontaneous PSCs in AII amacrine cells in the absence of spontaneous postsynaptic action currents mediated by voltage-gated I_Na, we added QX-314 to the internal solution. In this way, spontaneous and evoked action currents disappeared completely within 1–2 min after establishing the whole-cell recording configuration (Fig. 1C).

Pharmacological properties of spontaneous PSCs

All cells that were subsequently identified as AII amacrine cells displayed spontaneous PSCs. As glutamate is the prime neurotransmitter candidate of bipolar cells (reviewed by Massey & Maguire, 1995), we first examined the effect of CNQX on the spontaneous PSCs. This antagonist is selective for non-NMDA receptors, but has limited ability to discriminate between AMPA and kainate receptors (e.g. Wilding & Huettner, 1996). At a concentration of 10 μm, CNQX blocked nearly all spontaneous PSCs (Fig. 2A). The unblocked PSCs displayed small amplitudes and slow kinetics (Fig. 2A; lower right). The block of spontaneous PSCs with faster kinetics was completely reversible upon removal and washout of CNQX (not shown). We consider the PSCs blocked by CNQX to be glutamatergic, excitatory PSCs (EPSCs). Similar results were seen in six other cells.

Figure 2. CNQX and GYKI 53655 block spontaneous PSCs with fast kinetics, but not smaller amplitude currents with slow kinetics.

A, bath application of CNQX (10 μm) blocks spontaneous PSCs with fast kinetics. Here and in Fig. 3, activity is displayed at both a slow (upper panel) and a fast (lower panels) time scale. Smaller amplitude currents with slower kinetics (lower left panel, arrow) are not blocked by CNQX (lower right panel). B, bath application of GYKI 53655 (100 μm) blocks spontaneous PSCs with fast kinetics. Smaller amplitude currents with slower kinetics (lower left panel, arrow) are not blocked by GYKI 53655 (lower right panel). The gap in the trace (upper panel) corresponds to a 5 s period during which stimuli for monitoring series resistance were applied. For illustration purposes, we have chosen examples with a relatively high frequency of smaller amplitude currents with slow kinetics (here and in Fig. 3).

We next examined the effect of the selective AMPA receptor antagonist GYKI 53655, a non-competitive 2,3-benzodiazepine AMPA receptor antagonist with no or minimal effect on kainate receptors (Wilding & Huettner, 1995). At a concentration of 100 μm (previously shown to block the AMPA receptors expressed by AII amacrine cells; Mørkve et al. 2002), GYKI 53655 blocked almost all spontaneous PSCs (Fig. 2B, n = 24 cells). As with CNQX, the unblocked PSCs had relatively low amplitudes and slow kinetics (Fig. 2B; lower right). Postsynaptic currents mediated by either NMDA or kainate receptors display relatively slow kinetics (for review see Lester et al. 1994 and Lerma et al. 2001). Addition of CPP (30 μm), a selective NMDA receptor antagonist, did not block the slower spontaneous PSCs that were not blocked by GYKI 53655 (100 μm; Fig. 3A, middle panels; n = 5 cells). The possible involvement of specific kainate receptors was investigated by examining the effect of the selective kainate receptor agonist SYM 2081, which causes potent receptor desensitization (Zhou et al. 1997). SYM 2081 (2 μm) did not block the slower spontaneous PSCs that were not blocked by GYKI 53655 or CPP (Fig. 3A, right panels; n = 3 cells). These results suggested that the spontaneous PSCs with low amplitude and slow kinetics were not mediated by activation of glutamate receptors.

Figure 3. Small amplitude currents with slow kinetics are blocked by TTX, but not by CPP or SYM 2081.

A, CPP (30 μm; middle panels) and SYM 2081 (2 μm; right panels) do not block small amplitude currents with slower kinetics, present both in the control condition (left panels) and in the presence of GYKI 53655 (100 μm; middle and right panels). Here and in B, the dotted lines indicate the position of the current traces displayed at a fast time scale (lower panels) on the corresponding current traces displayed at a slow time scale (upper panels). B, TTX (300 nm; right panels) blocks small amplitude currents with slower kinetics, present both in the control condition (left panels) and in the presence of GYKI 53655 (100 μm; middle and right panels).

To further investigate the spontaneous currents with low amplitude and slow kinetics, we added TTX (300 nm) to the bath solution in order to block presynaptic action potentials. All PSCs that were not blocked by 100 μm GYKI 53655 (Fig. 3B, middle panels) were completely blocked by TTX (Fig. 3B, right panels; n = 6 cells). To investigate the possibility that the PSCs were electrical PSCs evoked by action potentials in electrically coupled AII amacrine cells, we performed simultaneous dual recordings from pairs of AII amacrine cells (Veruki & Hartveit, 2002). Presynaptic spikes, either spontaneous spikes in current clamp or simulated spikes in voltage clamp (with I_Na blocked by TTX), evoked slow, inward currents in the postsynaptic cell with amplitude and kinetic parameters very similar to those of spontaneously occurring non-glutamatergic PSCs (data not shown). In subsequent recordings, electrical PSCs were excluded, either by adding 300 nm TTX to the extracellular bath solution or by excluding events with 10–90 % rise times greater than 1.1 ms from analysis.

Frequency and amplitude distribution of spontaneous, glutamatergic EPSCs

Spontaneous EPSCs occured at a high frequency and often in bursts of overlapping events (Fig. 4A) in standard extracellular solution with 2.5 mm Ca²⁺ and 1 mm Mg²⁺. The average frequency was 90 ± 16 Hz (n = 12). When the release probability was reduced by lowering the concentration of Ca²⁺ to 0.15 mm and increasing the concentration of Mg²⁺ to 3.35 mm (‘low Ca²⁺’; Fig. 4B), the frequency of spontaneous EPSCs was markedly lower (27 ± 6 Hz, n = 19). In low Ca²⁺, several EPSCs tended to cluster together, while other events seemed to occur randomly. This pattern of synaptic activity could suggest the presence of coordinated neurotransmitter release from one or more presynaptic cells.

Figure 4. Temporal distribution, waveforms and amplitude distribution of spontaneous, glutamatergic EPSCs in AII amacrine cells.

A, spontaneous activity with marked bursting in an AII amacrine cell recorded in extracellular solution with 2.5 mm Ca²⁺. Here, and in B, activity is displayed both at a slow (left panels) and fast (right panels) time scale. B, spontaneous activity with moderate bursting (‘clustering’) in an AII amacrine cell recorded in extracellular solution with low Ca²⁺ (0.15 mm; 3.35 mm Mg²⁺). C, wide range of simple monophasic and complex multiphasic waveforms observed for spontaneous EPSCs in AII amacrine cells. Monophasic EPSCs have a fast rise time and a slower exponential decay (a, e and f). Multiphasic EPSCs (b, c, d and g) seem to encompass a cluster of individual, monophasic events with variable temporal dispersion such that peaks of individual events can be clearly resolved (b, c and d) or only form inflections on the rising or falling phase (g). D, amplitude distribution of spontaneous EPSCs recorded in extracellular solution with low Ca²⁺. Peak (mode) at ∼17 pA; notice skew toward larger amplitudes; bin width 2 pA. Here, and in Fig. 5C, the noise distribution is shown as an unfilled histogram (peak scaled to the peak of the EPSC amplitude distribution). E, interevent interval histogram of events in D; double exponential fit indicated by white line (τ₁ = 6.3 ms, A₁ = 95 %; τ₂ = 32.7 ms, A₂ = 5 %); bin width 0.02 ms. The entire histogram is shown for each graph.

Many spontaneous EPSCs appeared as individual events with a fast rise time and a slower, exponential decay (‘monophasic’; Fig. 4Ca, e and f). Additionally, we observed spontaneous EPSCs with more complex waveforms (‘multiphasic’; Fig. 4Cb, c, d and g). These complex EPSCs included EPSCs with multiple, separate peaks (Fig. 4Cb, c and d) and EPSCs with multiple, overlapping peaks that could not be clearly resolved (Fig. 4Cg). In extracellular solution with low Ca²⁺, approximately 90 % of the spontaneous EPSCs were monophasic events. With 2.5 mm Ca²⁺ in the extracellular solution, the proportion of monophasic events was lower, but difficult to estimate quantitatively because of the high frequency of EPSCs. In order to increase the relative frequency of monophasic events for quantitative analysis, most recordings were performed in low Ca²⁺.

Amplitude histograms were constructed from all EPSCs in each of 10 cells recorded in low Ca²⁺ (Fig. 4D). In each case, the amplitude distribution was skewed towards larger amplitude values. The peak (mode) ranged between 15 and 37 pA (average 23.4 ± 2.6 pA). The mean amplitude of the distributions varied between 21.4 and 47.8 pA (average 29.1 ± 2.5 pA). The coefficient of variation varied between 0.35 and 0.54 (average 0.45 ± 0.02). We also constructed interevent interval histograms from the same cells (Fig. 4E). The histograms were best fitted with a double exponential function with a slow time constant between 7.0 and 131.1 ms (average = 50.0 ± 12.4 ms) and a faster time constant between 1.4 and 14.2 ms (average = 7.0 ± 1.3 ms). The presence of the faster time constant is consistent with the presence of clusters of EPSCs in this condition.

Time course of spontaneous, glutamatergic EPSCs

To determine the kinetic properties of the EPSCs, we analysed continuous recordings (20–60 s duration) by selecting well-separated, spontaneous inward currents with simple, monophasic waveforms. For the cell illustrated in Fig. 5(A–H), recorded in external solution with low Ca²⁺, the peak amplitude of individual EPSCs (n = 514) ranged from 7.8 to 129.6 pA (Fig. 5C; average 35.4 ± 7.1 pA). The 10–90 % rise time of individual EPSCs ranged from 178 to 812 μs (Fig. 5D; average 320 ± 4 μs). The decay phase was well fitted by a single exponential function, both for individual EPSCs (Fig. 5A; τ_decay = 246–1877 μs; average 695 ± 9 μs) and for the averaged EPSC (Fig. 5B; τ_decay = 693 μs). Kinetic properties of monophasic EPSCs were measured for a total of 10 cells recorded with low Ca²⁺ in the extracellular solution. The mean 10–90 % rise time was 339 ± 17 μs (range 266–416 μs). The mean decay time constant was 762 ± 38 μs (range 595–980 μs).

Figure 5. Kinetics of monophasic, spontaneous EPSCs; EPSC amplitude as a function of τ_decay and 10–90 % rise time.

A, five overlaid spontaneous EPSCs in an AII amacrine cell (dotted lines), aligned by 50 % rise time. A single exponential fit (continuous lines) has been overlaid on two EPSCs. Data from same AII amacrine cell in A–H. B, average waveform of spontaneous EPSCs (dotted line); single exponential fit indicated by continuous line. C, distribution of peak amplitude for spontaneous EPSCs; bin width 2 pA. D, distribution of 10–90 % rise time for spontaneous EPSCs; bin width 0.01 ms. E, distribution of τ_decay (single exponential) for spontaneous EPSCs; bin width 0.025 ms. F, the relation between EPSC peak amplitude and 10–90 % rise time. G, the relation between EPSC peak amplitude and τ_decay. H, no correlation between 10–90 % rise time and τ_decay; linear fit indicated by continuous line; r (linear correlation coefficient) = 0.18. I, spontaneous EPSCs recorded at ambient temperature (26 °C; top panel) and after heating to 34 °C (bottom panel). J, average waveforms of spontaneous EPSCs recorded at 26 °C and 34 °C (same cell as I), aligned at onset. Waveforms plotted at same scale (top panel) and after normalization of peak amplitudes (bottom panel). Notice reduced rise and decay time, as well as increased peak amplitude, after increasing the temperature.

When peak amplitude was plotted against the 10–90 % rise time (Fig. 5F), the smallest events had rise times that spanned the full range of observed values. The largest events had faster rise times, typically below 500 μs (Fig. 5F). A similar pattern was seen when the peak amplitude was plotted against the time constant of decay (Fig. 5G). The smallest events had decay time constants that spanned the whole range of observations, while the largest amplitudes typically displayed values below 1 ms (Fig. 5G). Similar results were observed for the other nine cells. Importantly, there was no correlation between rise time and decay time (Fig. 5H), suggesting that electrotonic filtering does not have a large effect on the kinetic properties of the spontaneous synaptic events, or, alternatively, that all events are generated at a similar electrotonic distance.

We next examined the kinetic and amplitude parameters of simple, monophasic spontaneous EPSCs at a more physiologically relevant temperature. The temperature dependence was determined by increasing the temperature from ambient (∼26 °C) to 34 °C. Figure 5I illustrates EPSCs from an AII amacrine cell recorded in low Ca²⁺ at 26 °C (upper panel) and at 34 °C (lower panel). In Fig. 5J the average waveforms of the EPSCs in the two conditions are shown superimposed, both with the same scaling (upper panel) and after normalization to facilitate comparison (lower panel; 309 events at 26 °C, 263 events at 34 °C). When the temperature was increased from 26 °C to 34 °C, the 10–90 % rise time of the average waveform decreased from 322 μs to 263 μs, the decay time constant decreased from 746 μs to 595 μs and the peak amplitude increased from 23.2 pA to 32.3 pA. Similar results were observed for five other cells. The average 10–90 % rise time decreased from 304 ± 10 μs to 259 ± 7 μs, the average decay time constant decreased from 685 ± 32 μs to 516 ± 24 μs and the average peak amplitude increased from 31.5 ± 3.4 pA to 39.7 ± 3.4 pA (P < 0.01 for all parameters; paired t test). There was no statistically significant change in the average frequency of events (P = 0.19; paired t test). The corresponding Q₁₀ values were 1.25 ± 0.04 for the 10–90 % rise time, 1.50 ± 0.09 for the decay time constant (cf. Silver et al. 1996) and 1.44 ± 0.15 for the peak amplitude.

Voltage dependence of spontaneous, glutamatergic EPSCs

We next examined the voltage dependence of the spontaneous glutamatergic EPSCs by varying the membrane holding potential in steps of 10 mV increments from −80 to +70 mV (25 μm spermine in the internal solution, low Ca²⁺ and 20 μm CPP in the external solution). The example traces illustrated in Fig. 6A show that the spontaneous EPSCs reversed from inward to outward currents at a membrane potential very close to 0 mV. In order to analyse the current–voltage (I–V) relationship quantitatively, we averaged well-isolated, spontaneous EPSCs occurring during a 5–10 s period (n = 50–120) at each holding potential (Fig. 6B) and plotted the peak amplitude of each average EPSC against the holding potential (Fig. 6C; filled circles). For this cell, E_rev was estimated to be 0.1 mV. Similar results were observed for three other cells (E_rev = 0.7 ± 1.2 mV). The I–V curve of the peak response was approximately linear (Fig. 6C, filled circles). We analysed rectification quantitatively by calculating an index of rectification (RI) as the relation between the slope conductances at +40 mV and −60 mV (G_{slope,+40 mV}/G_{slope,−60 mV}). For the cell illustrated in Fig. 6A–C, the RI was 1.4 (average 0.8 ± 0.2; n = 4). From the individual (Fig. 6A) and averaged (Fig. 6B) EPSCs, it can be seen that the time course of the decay phase was voltage dependent, with slower decay at more positive membrane potentials. Corresponding to this, the I–V relation displayed marked outward rectification when it was calculated from response values during the decay phase (Fig. 6C; open circles; average RI = 23.9 ± 19.8). From a polynomial curve fit to the points of the I–V relation, the E_rev was calculated as 2.5 mV (average 2.3 ± 1.1 mV).

Figure 6. Voltage dependence of spontaneous EPSCs in AII amacrine cells.

A, traces with spontaneous EPSCs in the same AII amacrine cell at three different holding potentials; notice reversal of current and slower decay at positive holding potential (top trace). B, averaged spontaneous EPSCs at a series of holding potentials (from −80 to +70 mV; 10 mV steps; EPSCs aligned at 50 % rise time before and after averaging). Notice slower decay at positive compared to negative holding potentials. Filled and open circle (peak and decay phase, respectively) indicate time points used for I–V relationships in C. Same cell as in A. C, I–V relationships of averaged spontaneous EPSCs, measured for peak (•) and decay phase (○) of responses in B. Data points are fitted with 5th order polynomial functions. D, decay time constant (single exponential function; plotted as mean ± s.e.m.; n = 4 cells) of averaged, spontaneous EPSCs as a function of holding potential. Data points have been fitted with eqn (3).

We further examined the voltage dependence of the EPSC decay time course by fitting each decay phase with a single exponential function and plotting the decay time constant vs. the membrane potential (Fig. 6D; data pooled for the same four cells as above). The data points could be well fitted with eqn (3), giving a value for H of 209 mV (the change in membrane potential for an e-fold increase in τ).

Pharmacological modulation of spontaneous, glutamatergic EPSCs

Cyclothiazide is a benzothiadiazine which selectively modulates AMPA receptors by reducing agonist-evoked desensitization (Partin et al. 1993; Wong & Mayer, 1993). We have previously reported modulation of AMPA-evoked responses by cyclothiazide in patches from AII amacrine cells (Mørkve et al. 2002). We tested the effect of cyclothiazide on the spontaneous, glutamatergic EPSCs in AII amacrine cells by first recording EPSCs in the control condition (Fig. 7A; low Ca²⁺) and then in the presence of 100 μm cyclothiazide (Fig. 7B). In Fig. 7C the average EPSCs from the two conditions are shown overlaid to facilitate comparison. For this cell, cyclothiazide increased the decay time constant of the averaged EPSC from 0.95 ms to 2.92 ms and the peak amplitude from 21.5 pA to 23.0 pA. After washout of cyclothiazide, the effect was reversed (not shown). Similar results were seen for a total of six cells. Cyclothiazide increased the average decay time constant from 0.83 ± 0.07 ms to 3.11 ± 0.34 ms (P = 0.0005; paired t test), but there was no significant change of peak amplitude (29.6 ± 5.4 pA vs. 36.4 ± 7.4 pA; P = 0.053; paired t test). Cyclothiazide had no effect on the average frequency of spontaneous EPSCs (22 ± 8 Hz, control; 23 ± 6 Hz, cyclothiazide; P = 0.7, paired t test), suggesting that in this system any presynaptic effect of cyclothiazide is less pronounced than for synaptic transmission in hippocampal cell cultures (Diamond & Jahr, 1995).

Figure 7. Cyclothiazide modulates the kinetics of spontaneous EPSCs in AII amacrine cells.

A and B, traces with spontaneous EPSCs in the same AII amacrine cell in the control condition (A) and in the presence of 100 μm cyclothiazide (B). C, averaged spontaneous EPSCs in the control condition (1) and in the presence of cyclothiazide (2). Notice slower decay in the presence of cyclothiazide.

Presynaptic glutamatergic inputs to AII amacrine cells

The results so far presented demonstrate spontaneous, glutamatergic synaptic input to AII amacrine cells, but do not indicate the presynaptic source(s). In order to cast light on the integrity of the two potential sources of glutamatergic inputs to AII amacrine cells, we performed simultaneous, dual recordings from either a rod bipolar cell or an OFF-cone bipolar cell and an AII amacrine cell.

Figure 8A shows an example of the response in an AII amacrine cell evoked by stepping the membrane potential of a rod bipolar cell from −60 mV to −30 mV for 100 ms. The response shows a large amplitude, fast initial transient response followed by a decay to approximately baseline. The response in the AII amacrine cell could be completely and reversibly blocked by 10 μm CNQX (not shown). Similar responses were seen in 28 of 34 cell pairs examined (average peak amplitude 326 ± 32 pA; range 85–732 pA).

Figure 8. Synaptic input from bipolar cells to AII amacrine cells verified by dual recording.

A, simultaneous, dual recording of a rod bipolar cell and an AII amacrine cell synaptically connected to each other. Diagram (left) indicates recording configuration with both cells in voltage clamp and stimulus applied to rod bipolar cell. Traces (right) illustrate voltage clamp stimulus waveform (top) and corresponding presynaptic response in rod bipolar cell (middle) and postsynaptic response in AII amacrine cell (bottom). B, simultaneous, dual recording of an OFF-cone bipolar cell and an AII amacrine cell synaptically connected to each other. Diagram (left) indicates recording configuration with both cells in voltage clamp and stimulus applied to OFF-cone bipolar cell. Traces (right) illustrate voltage clamp stimulus waveform (top) and corresponding presynaptic response in OFF-cone bipolar cell (middle) and postsynaptic response in AII amacrine cell (bottom). Dashed lines (in A and B) indicate baseline.

Figure 8B shows an example of the response in an AII amacrine cell evoked by stepping the membrane potential of an OFF-cone bipolar cell from −60 mV to −30 mV for 100 ms. Similar to the response evoked by depolarizing a rod bipolar cell, the response encompassed a large amplitude, fast initial transient response and a decay to approximately baseline. The response in the AII amacrine cell could be completely and reversibly blocked by 10 μm CNQX (not shown). Similar responses were seen in 9 of 32 cell pairs examined (average peak amplitude 327 ± 61 pA; range 87–691 pA). These results demonstrate that many of the synaptic contacts between rod bipolar cells and AII amacrine cells, and between OFF-cone bipolar cells and AII amacrine cells, remained intact in our in vitro slice preparation. Accordingly, spontaneous EPSCs in AII amacrine cells could, in principle, arise from both sources of glutamatergic input.

Deactivation and desensitization kinetics of AMPA receptors in AII amacrine cells

The kinetic properties of AMPA receptors in AII amacrine cells were studied by ultrafast application of glutamate (3 mm) to somatic outside-out patches. Both short (∼1 ms) and long (∼500 ms) pulses of glutamate evoked responses that rose rapidly to a peak followed by a slower decay. Figure 9A illustrates the time course of the response evoked by a short pulse of glutamate. The response rose rapidly to a peak, with a 20–80 % rise time of 269 μs (10–90 % rise time 384 μs; patches which gave 20–80 % rise times longer than 400 μs were excluded from analysis). For eight patches, the average 20–80 % rise time of responses evoked by brief pulses was 274 ± 24 μs. At the end of the pulse, the response rapidly decayed. This deactivation, reflecting the closure of channels after removal of agonist (e.g. Koh et al. 1995), was well fitted with a single exponential function with a time constant of 0.71 ms (average τ = 1.07 ± 0.12 ms; range 0.71–1.78 ms). We never observed that a double exponential function provided a better fit of the deactivation time course (as determined by eye).

Figure 9. Deactivation and desensitization kinetics of AMPA receptors in AII amacrine cells.

A, response (bottom trace; dotted line; average of 25 trials) of outside-out patch from an AII amacrine cell to brief (∼1 ms), ultrafast application of glutamate (3 mm), overlaid with single exponential fit to decay phase (continuous line). Here, and in subsequent panels, the initial deflection of the baseline corresponds to stimulus artifact and the amplifier stimulus output is illustrated by the top trace. B, response (bottom trace; dotted line; average of 25 trials) of outside-out patch (same as in A) to long (500 ms), ultrafast application of glutamate (3 mm), overlaid with double exponential fit to decay phase (continuous line). Dashed line indicates baseline. C, responses and curve fits from initial periods in A and B overlaid to illustrate marked difference between rapid deactivation (A) and slower desensitization kinetics (B). D, the difference between deactivation and desensitization kinetics is particularly well illustrated by the overlaid responses of an outside-out patch (nucleated) to a series of applications (3 mm glutamate) of variable duration (∼1, 5, 9, 17, and 32 ms). Each trace is the average of 5 trials. E, overlaid responses of outside-out patch evoked by two brief (∼1 ms) pulses of glutamate (3 mm), pulses separated by recovery intervals from 4 to 68 ms). Each trace is the average of 4 trials. The dashed line indicates the mean peak current activated by the first pulse. F, depression of the response to the second pulse of glutamate (as in E) as a function of the recovery interval between the first and the second pulse (plotted as mean ± s.e.m.; n = 6 patches). Data points have been fitted with a single exponential function. Inset shows data points for shorter intervals at higher magnification. G, overlaid responses of outside-out patch to brief (∼1 ms), ultrafast application of glutamate (3 mm) at a series of holding potentials (from −80 to +70 mV; 10 mV steps; each trace is the average of 4 trials). Notice slower decay at positive compared to negative holding potentials. Filled and open circle (peak and decay phase, respectively) indicate time points used for I–V relationships in H. H, I–V relationships of responses in G, measured for peak (•) and decay phase (○). Peak and decay phase data points are fitted with 5th and 6th order polynomial functions, respectively. I, deactivation time constant (as in G; single exponential function) as a function of holding potential (plotted as mean ± s.e.m.; n = 8 patches). Data points have been fitted with eqn (3).

With 500 ms pulses of glutamate, the rise time and peak response amplitude were very similar to the corresponding values obtained for short pulses. However, the time course of desensitization, reflecting the closure of channels in the maintained presence of 3 mm glutamate, was considerably slower than the time course of deactivation (Fig. 9B and C). For the example illustrated in Fig. 9B, fitting with a single exponential function gave a time constant of 5.75 ms (not illustrated; average τ = 7.09 ± 0.73 ms; range 4.59–10.60 ms; n = 7 patches at −60 mV). In contrast to the deactivation kinetics, however, the time course of desensitization was better fitted with a double exponential function with a fast (τ₁ = 5.01 ms) and a slow component (τ₂ = 56.5 ms). For the seven patches tested, the average values were τ₁ = 3.5 ± 0.5 ms (74 ± 5.9 % amplitude contribution) and τ₂ = 20.8 ± 3.9 ms (26 ± 5.9 % amplitude contribution). The equilibrium response to 3 mm glutamate was measured as the current at the end of the 500 ms pulses and was on average 4.2 ± 0.9 % of the peak response (range 1.4–9.2 %).

The difference between deactivation and desensitization kinetics was particularly well illustrated by application of intermediate-duration glutamate pulses. In the example illustrated in Fig. 9D, it can be seen that the decay time course changed instantly from the relatively slow desensitization to the relatively fast deactivation upon removal of agonist.

In order to study the time course of recovery from desensitization, we employed a double-pulse protocol with application of two brief (∼1 ms) pulses of glutamate (3 mm) separated by a series of increasing recovery intervals (4–2052 ms; interval measured from the end of the first to the beginning of the second pulse). For each patch we averaged up to four responses at each interval. Figure 9E shows a representative example from one patch with a series of superimposed traces with inter-pulse intervals ranging from 4 ms to 68 ms. For short intervals, the first pulse strongly depressed the response to the second pulse, but there was a rapid reduction of the depression with increasing recovery intervals. When we plotted the average reduction of the response to the second pulse as a function of the recovery interval between the two pulses (Fig. 9F), the time course of response recovery was well fitted by a single exponential function (τ_recovery = 12.4 ms). Extrapolation of the fitted curve to time 0 indicated that 64 % of the channels were desensitized by the first pulse (cf. Colquhoun et al. 1992). Fitting with a double exponential function only slightly improved the fit and only ∼4 % of the amplitude at time 0 was associated with the slow time constant.

Voltage dependence of deactivation kinetics of AMPA receptor mediated currents activated by glutamate in AII amacrine cells

We investigated the I–V relation of the current activated by ∼1 ms pulses of 3 mm glutamate by testing the response at holding potentials from −80 mV to +70 mV in 10 mV increments. The outside-out patches were held at −60 mV and the membrane potential was stepped to the new value 100–250 ms before applying the concentration jump. For each patch we averaged up to four responses at each membrane potential. A representative example of a series of responses obtained from an AII amacrine cell patch is shown in Fig. 9E. The I–V relationship for the peak response was approximately linear (Fig. 9F; E_rev = −8.3 mV; average −5.5 ± 0.8 mV; n = 8 patches) while the I–V relationship for the decay phase displayed clear outward rectification (E_rev = −0.9 mV; average −2.7 ± 1.0 mV). The difference between the two I–V curves reflects the fact that the time course of deactivation (after removal of glutamate) was voltage-dependent, being slower at positive than at negative membrane potentials. For each membrane potential value, the deactivation time course was well fitted with a single exponential function. Figure 9G shows the deactivation time constant plotted against the membrane potential (data pooled for 8 patches). The relation was well fitted by eqn (3), giving a value for H of 234 mV (the change in membrane potential for an e-fold increase in τ).

DISCUSSION

In this study we have investigated the functional properties of spontaneous EPSCs in AII amacrine cells and compared them with the functional properties of extrasynaptic receptors in somatic patches as studied with ultrafast application of glutamate. One population of spontaneous synaptic events with relatively slow rise and decay times were blocked by TTX and were demonstrated to correspond to electrical PSCs evoked by action potential firing in other AII amacrine cells coupled via electrical synapses to the recorded cell. Another population of events with much faster rise and decay times were blocked by the non-NMDA receptor antagonist CNQX and by the AMPA receptor-specific antagonist GYKI 53655. The E_rev was very close to 0 mV. Furthermore, the decay time course of these spontaneous EPSCs was modulated by the AMPA receptor-specific drug cyclothiazide. These results indicate that the spontaneous EPSCs in AII amacrine cells are mediated by specific AMPA receptors. We found no evidence for a contribution of specific kainate receptors to spontaneous synaptic currents, consistent with our previously reported lack of these receptors in AII amacrine cells (Mørkve et al. 2002) and with the pharmacology of evoked EPSCs in amacrine cells of tiger salamander (Tran et al. 1999).

Kinetic properties of EPSCs in AII amacrine cells

The spontaneous EPSCs in AII amacrine cells occurred in a variety of waveforms, both as simple monophasic EPSCs with a single peak and an exponential decay and as complex multiphasic EPSCs with more than one peak and a non-exponential decay. Monophasic EPSCs recorded at room temperature were characterized by very fast kinetics with a 10–90 % rise time of ∼340 μs and a decay that was well fitted by a single exponential function (τ_decay∼760 μs). Spontaneous glutamatergic EPSCs with fast decay kinetics have previously been observed in unspecified amacrine cells of mouse retina (Frech et al. 2001). At an elevated temperature of 34 °C, the 10–90 % rise time decreased to ∼260 μs and τ_decay decreased to ∼520 μs. The time course of decay was voltage dependent with slower decay at positive than at negative membrane potentials. The voltage sensitivity was low, with ∼209 mV change in membrane potential required for an e-fold change in τ_decay. Similar voltage-dependent decay has previously been reported for AMPA receptor-mediated EPSCs in the auditory pathway, specifically for calyceal synapses on neurons in the avian nucleus magnocellularis (Zhang & Trussell, 1994; Otis et al. 1996) and for ribbon synapses between hair cells and afferent fibres in the mammalian auditory nerve (Glowatzki & Fuchs, 2002). Although it is unlikely that this weak voltage dependence is of functional importance for synaptic transmission in these pathways, it is interesting that several synaptic connections at early stages of sensory pathways might employ AMPA receptor channels with similar functional properties. The voltage-dependent decay of EPSCs in AII amacrine cells is probably explained by voltage sensitivity of the receptors themselves (cf. Mørkve et al. 2002). Voltage-dependent decay was first observed at the neuromuscular junction and has been interpreted to mean that the kinetic properties of the channels play an important role in determining the PSC kinetics (Magleby & Stevens, 1972). To our knowledge, it is not known which AMPA receptor subunit(s) might be responsible for assembling channels with voltage-dependent kinetics as observed in the present study.

Analysis of the parameters of spontaneous, glutamatergic EPSCs (amplitude and frequency distribution, kinetics) did not reveal any heterogeneity that would suggest two distinct populations of EPSCs corresponding to synapses made by rod bipolar cells and synapses made by OFF-cone bipolar cells. This could suggest that, with respect to the receptor properties investigated here, there is no functional distinction between AMPA receptors in these two sets of synapses. An alternative hypothesis is that only one of the two presynaptic sources makes a significant contribution to the total number of spontaneous EPSCs, either due to a difference in the average number of synapses or release sites made by each type of presynaptic terminal or a difference in the rate of release. In our light-adapted preparation, one could speculate that the rate of release from rod bipolar terminals would be higher than that from OFF-cone bipolar terminals. Experiments with simultaneous dual recordings from a presynaptic bipolar cell (either a rod bipolar or an OFF-cone bipolar) and a postsynaptic AII amacrine cell verified the integrity of both sources of glutamatergic input in our preparation. In principle, each pathway could contribute to the population of spontaneous EPSCs. An important topic for future work will be to resolve the functional properties of the receptors in each of these pathways. The observed difference in effective connectivity between rod bipolar cells and OFF-cone bipolar cells could suggest a higher degree of selectivity in the connections between OFF-cone bipolar cells and AII amacrine cells. Alternatively, it is possible that some OFF-cone bipolar cells are not presynaptic to AII amacrine cells.

Molecular identity of AMPA receptors expressed by AII amacrine cells

In the present study, we observed spontaneous EPSCs in AII amacrine cells with very fast kinetics. It is likely that the kinetic properties of the receptors themselves play an important role in determining the kinetics of the EPSCs. We examined the receptor kinetics by ultrafast application of glutamate to outside-out patches taken from the somata of AII amacrine cells. Like the EPSCs, the AMPA receptor channels in patches displayed very fast kinetics with a deactivation time course well fitted by a single exponential function with τ_decay∼1.1 ms. Additionally, deactivation was voltage dependent with slower decay at positive than at negative membrane potentials (∼234 mV/e-fold change in τ_decay). The desensitization time course was slower and best fitted by a double exponential function with τ₁∼3.5 ms and τ₂∼21 ms. Desensitization was nearly complete in the presence of maintained application of glutamate, with a steady-state component ∼4 % of the peak response. Activation of receptors with brief pulses evoked substantial desensitization (Colquhoun et al. 1992; Hestrin, 1993), but the receptors rapidly recovered from desensitization with a time constant of ∼12 ms. These results indicate that the decay time course of spontaneous EPSCs in AII amacrine cells is relatively similar to the deactivation time course of responses in outside-out patches. Additionally, both processes display similar voltage sensitivity. The difference in τ_decay (∼0.8 ms vs.∼1.1 ms) could be explained by a difference in the exact molecular composition between synaptic and somatic (extrasynaptic) receptors or by a difference in the temporal profile of agonist concentration between the two conditions. Nevertheless, the kinetic parameters are surprisingly similar and suggest that the properties of the extrasynaptic receptors are fairly representative of the properties of the synaptic receptors. The EPSC decay kinetics are clearly different from the desensitization kinetics, suggesting that desensitization plays only a minor role in determining the time course of spontaneous EPSCs. This conclusion is similar to that for several other CNS synapses (reviewed by Geiger et al. 1999).

From studies in expression systems, both GluR-A and GluR-D homomeric receptors display deactivation kinetics similar to those observed for glutamate responses in outside-out patches from AII amacrine cells. For GluR-D receptors, Mosbacher et al. (1994) found similar deactivation time constants for GluR-D flip and flop (τ_decay∼0.6 ms). With regard to desensitization, however, GluR-D flop is considerably faster (τ_decay∼0.9 ms) than GluR-D flip (τ_decay∼3.6 ms). Accordingly, the marked difference between deactivation and desensitization kinetics for glutamate responses in outside-out patches from AII amacrine cells could suggest a predominant involvement of GluR-D flip, but not GluR-D flop. Interestingly, the desensitization time constant for GluR-D flip is similar to the dominant (faster) time constant for desensitization that we found for glutamate receptors in patches from AII amacrine cells (∼3.5 ms). For GluR-A receptors (flip and flop), Mosbacher et al. (1994) found a deactivation time constant of ∼1.1 ms. This is also very similar to the deactivation time constant we found for glutamate responses in patches from AII amacrine cells and only slightly longer than the decay time constant of spontaneous EPSCs. In addition, the desensitization time constants for GluR-A receptors are ∼3.4 ms (flip) and ∼3.7 ms (flop) (Mosbacher et al. 1994), similar to the desensitization time constant of glutamate responses in patches from AII amacrine cells. However, with regard to the rate of recovery from desensitization, GluR-A receptors display a τ_recovery of ∼147 ms (Partin et al. 1996). This is markedly longer than the τ_recovery for GluR-D receptors (∼6–43 ms; Lomeli et al. 1994) and for glutamate responses in patches from AII amacrine cells (∼12 ms), suggesting that GluR-A is not involved. Interestingly, τ_recovery for AMPA receptors on AII amacrine cells is much more similar to τ_recovery for GluR-D flip (6–14 ms) than for GluR-D flop (31–43 ms), further suggesting that the AMPA receptors on AII amacrine cells are dominated by GluR-D flip. Consistent with our functional analysis, morphological studies have found evidence for the presence of GluR-D (and GluR-C), but not GluR-A, in synapses between rod bipolar cells and AII amacrine cells (Qin & Pourcho, 1999; Ghosh et al. 2001; Li et al. 2002). The rectification properties of spontaneous EPSCs and glutamate-evoked responses in patches are similar to those reported for AMPA-evoked responses in nucleated patches from AII amacrine cells (Mørkve et al. 2002). The implications of these properties for the subunit composition of the receptors were discussed in our previous study.

Presynaptic mechanisms and glutamatergic synaptic input to AII amacrine cells

We investigated the amplitude distribution of spontaneous, glutamatergic EPSCs in the presence of a low concentration of Ca²⁺ to reduce the release probability and thereby enhance the relative frequency of monophasic EPSCs. It is tempting to speculate that the modal peak of the amplitude distribution (∼23 pA) could correspond to the amplitude of presumed quantal EPSCs in AII amacrine cells (cf. Glowatzki & Fuchs, 2002). Assuming an E_rev of ∼0 mV, the quantal (chord) conductance would be ∼380 pS. With a single channel conductance of ∼23 pS (Mørkve et al. 2002), this would correspond to approximately 17 open channels at the peak of a single quantal EPSC. The estimated quantal conductance is within the range of values estimated for several other glutamatergic synapses in the CNS (e.g. 138 pS for synapses between retinal ganglion cells and thalamic relay cells: Paulsen & Heggelund, 1994; 180 pS for synapses between mossy fibres and cerebellar granule cells: Silver et al. 1992; 400 pS for synapses between hair cells and cochlear afferent fibres: Glowatzki & Fuchs, 2002; 450 pS for synapses between calyx of Held terminals and cells in the medial nucleus of the trapezoid body: Sahara & Takahashi, 2001).

In a study of synaptic transmission at ribbon synapses between cochlear hair cells and afferent auditory nerve fibres, Glowatzki & Fuchs (2002) found strong evidence for multivesicular release. The results suggested that the amplitude of spontaneous EPSCs at a single site can vary, depending on the fraction of vesicles (docked at one presynaptic ribbon) that are released simultaneously. If similar multivesicular release can occur from retinal bipolar cells, the largest spontaneous EPSCs might be evoked through simultaneous release of all docked vesicles at a single ribbon in a bipolar terminal. Sterling (1998) has reviewed evidence indicating that a single ribbon in a rod bipolar terminal docks ∼20 vesicles and tethers ∼100 vesicles. With a quantal amplitude of ∼23 pA, perfectly synchronous release, and linear summation, this could correspond to a maximum peak amplitude of spontaneous EPSCs of ∼460 pA. In rabbit retina, the vitreal dendrites of an AII amacrine cell have been electronmicroscopically reconstructed through serial sections (Strettoi et al. 1992). The cell received 47 ribbon synapses from a total of nine rod bipolar cells. Each rod bipolar cell contributed from 1 to 15 ribbon synapses. If these numbers are roughly relevant for synapses between rod bipolar cells and AII amacrine cells in the rat retina, evoked release with perfectly synchronous release from all ribbons of a rod bipolar axon terminal could theoretically evoke compound EPSCs with a peak amplitude ranging from ∼460 pA to ∼6.9 nA. In simultaneous dual recordings of rod bipolar cells and AII amacrine cells, responses were evoked by step depolarization of the rod bipolar cell. The peak amplitude of responses in the AII amacrine cells was lower, ranging from 85 to 732 pA. It will be interesting to see how quantitative morphological data from the rat retina compares with these numbers.