Functional NMDA receptors are expressed by both AII and A17 amacrine cells in the rod pathway of the mammalian retina

Abstract

At many glutamatergic synapses, non-N-methyl-d-aspartate (NMDA) and NMDA receptors are coexpressed postsynaptically. In the mammalian retina, glutamatergic rod bipolar cells are presynaptic to two rod amacrine cells (AII and A17) that constitute dyad postsynaptic partners opposite each presynaptic active zone. Whereas there is strong evidence for expression of non-NMDA receptors by both AII and A17 amacrines, the expression of NMDA receptors by the pre- and postsynaptic neurons in this microcircuit has not been resolved. In this study, using patch-clamp recording from visually identified cells in rat retinal slices, we investigated the expression and functional properties of NMDA receptors in these cells with a combination of pharmacological and biophysical methods. Pressure application of NMDA did not evoke a response in rod bipolar cells, but for both AII and A17 amacrines, NMDA evoked responses that were blocked by a competitive antagonist (CPP) applied extracellularly and an open channel blocker (MK-801) applied intracellularly. NMDA-evoked responses also displayed strong Mg²⁺-dependent voltage block and were independent of gap junction coupling. With low-frequency application (60-s intervals), NMDA-evoked responses remained stable for up to 50 min, but with higher-frequency stimulation (10- to 20-s intervals), NMDA responses were strongly and reversibly suppressed. We observed strong potentiation when NMDA was applied in nominally Ca²⁺-free extracellular solution, potentially reflecting Ca²⁺-dependent NMDA receptor inactivation. These results indicate that expression of functional (i.e., conductance-increasing) NMDA receptors is common to both AII and A17 amacrine cells and suggest that these receptors could play an important role for synaptic signaling, integration, or plasticity in the rod pathway.

the vast majority of excitatory synaptic transmission in the central nervous system is mediated by the amino acid glutamate (Hassel and Dingledine 2012). After synaptic release, glutamate diffuses across the synaptic cleft and binds to different types of receptors in the postsynaptic membrane. There are two main types of ionotropic glutamate receptors (iGluRs), termed N-methyl-d-aspartate (NMDA) receptors and non-NMDA receptors. These two types were originally defined on the basis of their sensitivity to exogenous agonists, and non-NMDA receptors are further classified into AMPA and kainate receptors (reviewed by Hassel and Dingledine 2012; Smart and Paoletti 2012). More recently, molecular investigations have revealed that the pharmacologically defined types of receptors are composed of different subunits. NMDA receptors are heteromeric tetramers composed of two obligatory GluN1 subunits and two accessory GluN2 subunits (GluN2A–D). AMPA receptors are homo- or heteromeric tetramers composed of GluA1-4 subunits, and kainate receptors are homo- or heteromeric receptors composed of GluK1-5 subunits. At many glutamatergic synapses, the postsynaptic density contains both non-NMDA and NMDA receptors such that presynaptic release of glutamate activates both types of receptors, giving rise to a dual-component excitatory postsynaptic current (EPSC; Hassel and Dingledine 2012). At some synapses, however, either NMDA or non-NMDA receptors seem to be missing and the EPSCs are mediated solely by one type of receptor, e.g., by NMDA receptors at “silent synapses” (Malinow et al. 2000) and non-NMDA receptors at climbing- and parallel-fiber synapses on Purkinje cells (Perkel et al. 1990). Because of their special functional properties, including high Ca²⁺ permeability, Mg²⁺-dependent voltage block at negative membrane potentials, and slow kinetics (Traynelis et al. 2010), NMDA receptors and their role in synaptic signaling and plasticity have attracted intense investigation (Paoletti et al. 2013).

In the retina, glutamate is used as a neurotransmitter by both photoreceptors and bipolar cells, mediating input to horizontal cells and bipolar cells and to amacrine cells and ganglion cells, respectively (Massey and Maguire 1995). In the cone pathway, the synaptic transmission between cone bipolar cells and ganglion cells can involve both non-NMDA and NMDA receptors, in both the ON and OFF pathways (Copenhagen et al. 1993), and NMDA receptors can contribute to contrast coding and temporal processing in ganglion cells (Manookin et al. 2010; Stafford et al. 2014). In the rod pathway, rod bipolar cells do not output their signals directly to ganglion cells (Strettoi et al. 1990, 1992). Instead, they contact AII amacrine cells, which are presynaptic to ON-cone bipolar cells via electrical synapses and to OFF-cone bipolar cells and ganglion cells via glycinergic synapses (Strettoi et al. 1992, 1994). Through these connections, the AII amacrine is crucial for retinal signal processing not only under scotopic conditions (reviewed by Bloomfield and Dacheux 2001), but also under mesopic and photopic conditions (Manookin et al. 2008; Münch et al. 2009). AMPA-type non-NMDA receptors make a substantial contribution to the EPSC evoked in AII amacrine cells by depolarization of presynaptic rod bipolar cells (Singer and Diamond 2003), and AII amacrine cells were thought not to express NMDA receptors (Boos et al. 1993). There is increasing evidence, however, from both physiological (Bloomfield and Xin 2000; Hartveit and Veruki 1997; Zhou and Dacheux 2004) and immunocytochemical (Kothmann et al. 2012) investigations, that AII amacrine cells do express NMDA receptors, suggesting that glutamatergic neurotransmission in these cells is more complex than originally believed. In addition to being presynaptic to AII amacrine cells, rod bipolar cells are also presynaptic to a second type of rod amacrine cell, termed AI or A17 (Kolb and Famiglietti 1974). A17 amacrines are wide-field amacrines that provide inhibitory GABAergic feedback to the rod bipolar cell (Chávez et al. 2006; Hartveit 1999; Nelson and Kolb 1985; Raviola and Dacheux 1987), and it has been suggested that they do not express NMDA receptors (Menger and Wässle 2000). At each specialized presynaptic release site (ribbon) of the axon terminal of a rod bipolar cell, two postsynaptic processes from an AII and an A17 amacrine cell collectively constitute a postsynaptic dyad (Kolb and Famiglietti 1974; Raviola and Dacheux 1987). On this basis, we decided to examine in more detail the properties of NMDA receptors expressed by AII amacrine cells and to investigate the potential expression of NMDA receptors by A17 amacrine cells. We also wanted to reexamine the presence of NMDA receptors in rod bipolar cells, for which there is conflicting evidence suggesting both the presence (Karschin and Wässle 1990; Lo et al. 1998; Wenzel et al. 1997) and absence (Fletcher et al. 2000; Hartveit 1996) of NMDA receptors on these cells. In this article, we provide direct evidence that not only AII, but also A17 amacrine cells express functional (i.e., conductance-increasing) NMDA receptors, detected by recording whole cell current responses evoked by application of NMDA. No such responses were observed in rod bipolar cells. Our results suggest that NMDA receptors could play a significant role in synaptic signaling or plasticity in the rod pathway.

METHODS

Retinal slice preparation and visual targeting of neurons.

General aspects of the methods have previously been described in detail (Veruki et al. 2003). Female albino rats (Wistar HanTac; 4–7 wk postnatal) were deeply anaesthetized with isoflurane in oxygen and killed by cervical dislocation (procedure approved under the surveillance of the Norwegian Animal Research Authority). Vertical retinal slices were visualized using an Axioskop 2 FS (Zeiss) with a ×40 water-immersion objective and infrared differential interference contrast (IR-DIC) videomicroscopy.

Solutions and drug application.

The standard extracellular perfusing solution was continuously bubbled with 95% O₂-5% CO₂ and had the following composition (in mM): 125 NaCl, 25 NaHCO₃, 2.5 KCl, 2.5 CaCl₂, 1 MgCl₂, and 10 glucose, pH 7.4. In some experiments, MgCl₂ was omitted from the extracellular solution (with no replacement of divalent cations) to relieve the voltage-dependent block of NMDA receptors. For these experiments, cells were held in the Mg²⁺-free bath solution for at least 10 min before NMDA was applied to ensure a complete washout of the divalent cations. In a few experiments, both MgCl₂ and CaCl₂ were omitted from the extracellular solution (with no replacement of divalent cations). Recording pipettes were filled with (in mM) 125 potassium gluconate, 8 KCl, 5 HEPES, 1 CaCl₂, 1 MgCl₂, 5 EGTA, 4 Na₂ATP, and 2 N-(2, 6-dimethylphenylcarbamoylmethyl)triethylammonium bromide (QX-314; Tocris Bioscience, Bristol, UK) or 125 potassium gluconate, 8 NaCl, 10 HEPES, 1 CaCl₂, 5 EGTA, 4 MgATP, and 2 QX-314; pH was adjusted to 7.3 with KOH. For the acquisition of current-voltage (I–V) relationships, recording pipettes were filled with (in mM) 125 CsCH₃SO₃, 8 NaCl, 10 HEPES, 1 CaCl₂, 5 EGTA, 15 tetraethylammonium chloride (TEA-Cl), and 4 MgATP; pH was adjusted to 7.3 with CsOH. For most nucleated patch experiments, cells were filled with (in mM) 125 CsCl, 8 NaCl, 10 HEPES, 1 CaCl₂, 5 EGTA, 15 TEA-Cl, and 4 MgATP; pH was adjusted to 7.3 with CsOH. For some nucleated patch experiments, recording pipettes were filled with (in mM) 125 potassium gluconate, 8 NaCl, 10 HEPES, 1 CaCl₂, 5 EGTA, 4 MgATP, and 2 QX-314; pH was adjusted to 7.3 with KOH. For visualization of complete cellular morphologies with wide-field fluorescence microscopy after the recording, Lucifer yellow (1 mg/ml; Sigma-Aldrich), Alexa Fluor 488 (50 μM; Invitrogen), or Alexa Fluor 594 (40 μM; Invitrogen) was included in the intracellular solutions.

For pressure application from glass pipettes, drugs were dissolved in a vehicle solution containing (in mM) 145 NaCl, 2.5 KCl, 2.5 CaCl₂, 1 MgCl₂, 5 Na-HEPES, and 10 glucose. MgCl₂ and CaCl₂ were eliminated from this solution to match the bath solution as necessary. We used either a single-barrel puffer pipette (similar in size and shape to the patch pipettes described below) filled with NMDA or a multibarrel puffer pipette with NMDA in one or more barrels and NMDA with the specific NMDA receptor antagonist (R,S)-3-(2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP; 400 μM) in one or more separate barrels (7 barrels in total; for details, see Hartveit 1996). In all cases, the concentration of NMDA in the pipette was 1 mM. In these recordings, an NMDA receptor co-agonist (Kleckner and Dingledine 1988; Traynelis et al. 2010), either 10 μM glycine in the presence of 10 μM strychnine and 100 μM picrotoxin or 200 μM d-serine, was always included in the application pipette. For the nucleated patch experiments, individual barrels of the multibarrel pipettes were filled with 1 mM NMDA (with co-agonist), 1 mM GABA, or 1 mM glycine. For some experiments, 1 or 2 mM (5S,10R)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine maleate (MK-801; Tocris Bioscience) was added to the intracellular solution. Other neurotransmitter receptor antagonists and ion channel blockers were added directly to the extracellular solution at the following concentrations (supplied by Tocris Bioscience, unless otherwise indicated): 1 μM strychnine (Research Biochemicals, Natick, MA) to block glycine receptors; 10 μM (−)-bicuculline methochloride to block GABA_A receptors; 50 μM (1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA) to block GABA_C receptors, 10 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) to block non-NMDA receptors; 0.3 μM tetrodotoxin (TTX) to block voltage-gated Na⁺ channels; and 100 μM 2-[(2,6-dichloro-3-methylphenyl)amino]benzoic acid sodium salt [meclofenamic acid (MFA) sodium salt; Sigma-Aldrich]. Solutions were either made up freshly for each experiment or prepared from aliquots stored at −20°C and diluted to the final concentration on the day of the experiment.

Electrophysiological recording and data acquisition.

Patch pipettes were pulled from thick-walled borosilicate glass (outer diameter, 1.5 mm; inner diameter, 0.86 mm) to obtain an open-tip resistance that ranged from 5 to 7 MΩ when filled with intracellular solution. Whole cell voltage-clamp recordings were performed with an EPC9-dual amplifier (HEKA Elektronik, Lambrecht/Pfalz, Germany) and controlled with PatchMaster software (HEKA Elektronik). After a gigaohm seal was established, currents caused by the recording electrode capacitance (fast capacitative current; “C_fast” function in PatchMaster) were automatically measured and neutralized by the amplifier. After the cell was broken into, currents caused by the cell membrane capacitance (slow capacitative current; “C_slow” function in PatchMaster) were partially neutralized by the amplifier. For whole cell recordings, series resistance (R_s) was monitored throughout the recording, at intervals of 1 min for most experiments, by applying a series of 20-mV hyperpolarizing voltage pulses (16-ms duration) after transiently disabling the C_slow neutralization circuitry of the amplifier. The charging transients were analyzed offline with FitMaster software (HEKA Elektronik) by fitting the decay phase during the voltage pulse with a double exponential function, and R_s was calculated from the amplitude of the voltage pulse and the peak current amplitude extrapolated to the onset of the pulse. R_s was not compensated. Cells with R_s > 40 MΩ were not included in the final material.

To establish a nucleated patch recording, the pipette was slowly withdrawn after the whole cell recording configuration had been established, while continuous light suction (∼50 mbar) was applied to the pipette. When a nucleated patch was successfully isolated, the reduced membrane capacitance resulted in capacitative current transients of the opposite polarity that were cancelled by readjustment of the C_slow neutralization circuitry.

The sampling interval was set to either 50 or 100 μs, and before sampling, signals were low-pass filtered (analog 3- and 4-pole Bessel filters in series) with a corner frequency (−3 dB) of 4 kHz. All recordings were carried out at room temperature (22–25°C). The data acquisition software (PatchMaster) corrected all holding potentials for liquid junction potentials online. Theoretical liquid junction potentials were calculated with JPCalcW (Axon Instruments/Molecular Devices).

Wide-field fluorescence microscopy.

All cells were inspected with wide-field fluorescence microscopy after recording. By visual observation through the microscope oculars, we inspected the morphology of each dye-filled neuron and its processes and the relationship between the branching pattern and the strata of the inner plexiform layer. For documentation, every dye-filled neuron was sketched by hand. In addition, for some cells we acquired digital image stacks using a TILLvisION system (TILL Photonics). Image acquisition and postacquisition deconvolution to remove noise and increase resolution were performed as described previously (Castilho et al. 2015a).

General data analysis.

Data were analyzed with FitMaster (HEKA Elektronik), IGOR Pro (WaveMetrics, Lake Oswego, OR), and Excel (Microsoft, Redmond, WA). The peak amplitude of NMDA-evoked currents was measured as the mean amplitude between two vertical cursors positioned close to the peak response as identified by eye. For acquisition of I–V relationships, the holding potential was incremented by 10-mV intervals. Data points of I–V relationships were connected by straight lines, and reversal potentials (E_rev) were determined by the intersection with the abscissa.

Data are means ± SE (n = number of cells). Statistical analyses with comparisons between or within groups were performed using Student's two-tailed t-test (paired or unpaired, as indicated). Differences were considered statistically significant at the P < 0.05 level. For illustration purposes, most raw data records were low-pass filtered (digital Gaussian filter, −3 dB at 500 Hz to 2 kHz).

RESULTS

Targeting and identification of neurons in the rod bipolar-AII-A17 microcircuit in rat retinal slices.

The cell bodies of AII and A17 amacrine cells each have a characteristic shape and location at the border between the inner nuclear layer and the inner plexiform layer, and as such, they can be readily targeted in an acute retinal slice preparation (Fig. 1A). AII amacrine cells have cell bodies that span the border between the inner nuclear layer and the inner plexiform layer and are bistratified narrow-field amacrine cells, whereas A17 amacrine cells have dome-shaped cell bodies with long, thin processes that carry distinct varicosities and terminate near the ganglion cell layer of the retina. Rod bipolar cells tend to have their cell bodies in the distal part of the inner nuclear layer, in close apposition to the outer plexiform layer (Fig. 1A). The intracellular solutions contained fluorescent dyes that diffused into the cells during recording and allowed unequivocal identification of the complete morphology after recording (Fig. 1B). For all three types of retinal neurons, NMDA and other drugs were applied from single-barrel or multibarrel pipettes positioned close to the location of the neuron in the inner plexiform or the inner nuclear layer (Fig. 1A). Nucleated patches were isolated from AII and A17 amacrine cells by pulling the cell body out of the slice after the whole cell configuration had been established and were used with drug application from multibarrel or single-barrel puffer pipettes (Fig. 1C). For this study, we recorded from a total of 70 AII amacrine cells, 63 A17 amacrine cells, and 6 rod bipolar cells.

Fig. 1.

Identification of rod amacrine cells (AII and A17) and rod bipolar cells in the rat retinal slice preparation. A, left: infrared differential interference contrast (IR-DIC) videomicrograph of an A17 amacrine cell in a retinal slice. Whole cell recording was performed with top pipette (located at cell body) and drug application with bottom pipette. Middle, IR-DIC videomicrograph of an AII amacrine cell in a retinal slice. Whole cell recording was performed with top pipette (located at cell body) and drug application with bottom pipette. Right, IR-DIC videomicrograph of a rod bipolar cell in a retinal slice. Whole cell recording was performed with pipette located at cell body. B: maximum intensity projections (along Z-axis) generated from wide-field fluorescence image stacks (after deconvolution) for cells recorded in retinal slices and filled with Alexa Fluor 594 via patch pipettes: A17 amacrine cell (left), AII amacrine cell (middle), and rod bipolar cell (right). C, left: IR-DIC videomicrograph of a nucleated patch pulled from an AII amacrine cell in the retinal slice and a multibarrel pipette for drug application. Right, IR-DIC videomicrograph of a nucleated patch pulled from an AII amacrine cell in the retinal slice and a single-barrel pipette for drug application. Scale bars A–C: 10 μm.

Both AII and A17 amacrine cells, but not rod bipolar cells, in rat retinal slices respond to application of NMDA.

AII and A17 amacrine cells form the dyad postsynaptic targets of ribbon synapses of glutamatergic rod bipolar cell axon terminals (Fig. 2A). There is evidence that NMDA receptors in glutamatergic synapses can have not only a conventional postsynaptic (including extrasynaptic) localization (Sheng and Kim 2012) to mediate synaptic transmission and control postsynaptic excitability, but also a presynaptic localization, potentially to control transmitter release (Engelman and MacDermott 2004). Accordingly, NMDA receptors could be expressed by any or all of the three neuronal elements of this synaptic microcircuit. In principle, NMDA receptors could also be localized postsynaptically at rod bipolar cell dendrites in the outer plexiform layer, where they receive glutamatergic input from rod photoreceptors (Massey and Maguire 1995).

Fig. 2.

AII and A17 amacrine cells, but not rod bipolar cells, express functional (conductance-increasing) N-methyl-d-aspartate (NMDA) receptors. A: schematic diagram of retinal microcircuit consisting of a rod bipolar cell (RBC) and AII and A17 amacrine cells. Arrows indicate direction and sign of synaptic transmission between the cells: feedforward glutamatergic transmission from RBCs to AII and A17 amacrines (solid arrows) and feedback GABAergic transmission from A17 amacrine to RBC (open arrow). B: current activated in an AII amacrine cell [holding potential (V_hold) = −60 mV] by application of NMDA (1 mM, 1 s) from a puffer pipette. Except where indicated, NMDA was always applied together with the co-agonist d-serine (200 μM). Here and later, recording configuration is indicated by the schematic (right). In this and subsequent figures, the duration of drug application is indicated by the horizontal bar above the current trace, and unless otherwise noted, recordings were performed in nominally Mg²⁺-free extracellular solution (both in the bath and in the drug pipette). C: current activated in an A17 amacrine cell (V_hold = −60 mV) by application (1 s) of NMDA (1 mM) from a puffer pipette. D: no current activated in a rod bipolar cell (V_hold = −60 mV) by application of NMDA (1 mM, 1 s) from a multibarrel puffer pipette. The 2 traces correspond to application directed toward the soma-dendritic region (1) and the axon terminal region (2), obtained by moving the multibarrel pipette as indicated in the schematic (right). To avoid confounding effects caused by application of the vehicle solution itself, application of NMDA was preceded and followed by application of vehicle solution without agonist from a different barrel of the multibarrel pipette (broken horizontal lines). The extracellular solution contained strychnine, bicuculline, (1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and tetrodotoxin (TTX) (see methods).

We tested for the presence of functional (i.e., conductance-increasing) NMDA receptors in AII and A17 amacrine cells and in rod bipolar cells by pressure application of NMDA (1 mM, 1- to 2-s duration) from a puffer pipette with nominally Mg²⁺-free solution in both the bath and the pipette (hereafter referred to as “Mg²⁺-free solution”). In both AII and A17 amacrine cells, NMDA evoked large inward currents at holding potentials of −60 and −70 mV, respectively (close to the resting potentials of the 2 cell types; Fig. 2, B and C). The inward currents displayed relatively fast rise and decay times. All the AII and A17 amacrine cells tested under these conditions responded to NMDA. For AII amacrine cells, the average peak response was 64 ± 7 pA (n = 17, range 28–152 pA), and for A17 amacrine cells, the average peak response was 85 ± 8 pA (n = 22, range 36–182 pA).

In contrast to the two types of rod amacrine cells, NMDA evoked no response in any of the rod bipolar cells tested (n = 6 cells). The traces illustrated for a rod bipolar cell in Fig. 2D demonstrate how we examined two positions of the puffer pipette, with application directed either toward the axon terminal in the inner plexiform layer or toward the dendrites in the outer plexiform layer. In a typical recording, NMDA was first applied within 1–3 min after breaking into the cell and establishing the whole cell recording configuration. To minimize the likelihood that fast rundown of NMDA receptor channels (Horn and Korn 1992) could take place before the first application, we tested three rod bipolar cells with pressure application of NMDA within 20 s after breaking into the cells but still did not observe any responses. In some recordings, we observed small sustained shifts in the current that were tightly synchronized to the duration of drug application. These shifts were not accompanied by changes in noise, as expected for channel gating (cf. Fig. 2, B and C). When the pressure application of NMDA was preceded and followed by application of the vehicle solution (without agonist) from another barrel in the multibarrel complex (Fig. 2D), no shift in current was observed, suggesting that it was caused by a small difference in liquid junction potential between the bath solution and the solution in the puffer pipette. These results strongly suggest that rod bipolar cells do not express functional (conductance-increasing) NMDA receptor channels.

Functional properties of NMDA receptors in AII and A17 amacrine cells.

To examine whether the responses evoked by NMDA in AII and A17 amacrine cells were indeed mediated by NMDA receptors, we applied NMDA together with the specific NMDA receptor antagonist CPP (400 μM), with Mg²⁺-free solution in both the bath and the puffer pipette. For these experiments, we used a multibarrel pipette and first applied NMDA alone from one barrel, followed by coapplication of NMDA and CPP from another barrel (the antagonist was included in the same barrel as NMDA). By directing the tip of the multibarrel pipette toward photoreceptors at the edge of the slice, we used pressure-evoked movements of photoreceptor outer segments as a visual control to verify that fluid was adequately ejected from all barrels used during the pharmacological testing. In the presence of CPP, the NMDA responses were completely blocked for both AII (n = 8 cells) and A17 (n = 5 cells) amacrine cells (Fig. 3, A and B), and for both cell types the responses to NMDA recovered quickly following washout of CPP (Fig. 3, A and B).

Fig. 3.

AII and A17 amacrine cells express NMDA receptors blocked by the specific antagonist (R,S)-3-(2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP). A: currents activated in an AII amacrine cell (V_hold = −60 mV) by application (1.5 s) of NMDA (1 mM) from a multibarrel puffer pipette. Top trace, response evoked by NMDA in the control condition. Middle trace, no response to NMDA when coapplied with CPP (400 μM) in the same pipette barrel. Bottom trace, recovery of response to NMDA after washout of CPP. Here and in B, the extracellular solution contained strychnine, bicuculline, CNQX, and TTX (see Methods). B: currents activated in an A17 amacrine cell (V_hold = −70 mV) by application (1.5 s) of NMDA (1 mM) from a multibarrel puffer pipette. Top trace, response evoked by NMDA in the control condition. Middle trace, no response to NMDA when coapplied with CPP (400 μM) in the same pipette barrel. Bottom trace, recovery of response to NMDA after washout of CPP.

These experiments strongly suggested that the responses to NMDA were mediated by NMDA receptors, but they do not by themselves demonstrate conclusively that the responses were mediated by receptors located on the cells that we recorded from. To rule out the possibility that the NMDA-evoked responses were mediated by transsynaptic network effects, we performed three sets of experiments. In the first set we applied an antagonist intracellularly to block NMDA-evoked responses, in the second set we verified the characteristic I–V relationship expected for NMDA receptor-mediated currents, and in the third set we tested for the presence of NMDA receptor-mediated responses after blocking gap junction-mediated coupling pharmacologically. We first repeated the recordings with application of NMDA (in Mg²⁺-free extracellular solution) after including the NMDA receptor open-channel blocker MK-801 in the recording pipette solution (2 mM). In an attempt to use the cells as their own controls, we applied NMDA repeatedly (approximately every 60 s), starting as soon as possible after the whole cell recording condition had been established. For AII amacrine cells (n = 8 cells), there was no response to NMDA, even during the very first application of NMDA, which for the cell represented in Fig. 4A was obtained within 1 min after breaking into the cell. This is most likely explained by the small cell size and a relatively short diffusion distance from the tip of the pipette and cell body to the location of the NMDA receptors. As a positive control, AII amacrine cells in the same slices recorded without MK-801 added to the intracellular solution displayed the expected inward currents evoked by application of NMDA (data not shown).

Fig. 4.

AII and A17 amacrine cells express NMDA receptors blocked by intracellular application of the specific noncompetitive antagonist (open-channel blocker) (5S,10R)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine maleate (MK-801). A: currents recorded in an AII amacrine cell (V_hold = −60 mV) with application (1 s) of NMDA (1 mM) from a puffer pipette. MK-801 (2 mM) added to the intracellular solution in the recording pipette. Here and in B, time points at the right of each trace indicate the approximate time after the cell was broken into and the whole cell recording configuration established. Notice lack of response to NMDA even during the first application. Here and in B, the extracellular solution contained strychnine, bicuculline, CNQX, and TTX (see methods). B: currents recorded in an A17 amacrine cell (V_hold = −70 mV) with application (1 s) of NMDA (1 mM) from a puffer pipette. MK-801 (2 mM) was added to the intracellular solution in the recording pipette. Notice response to NMDA during the first application and gradual reduction of response amplitude during subsequent applications reflecting increasing block of NMDA receptors by MK-801.

In contrast to AII amacrine cells, we typically observed a small, but clear, response in A17 amacrine cells recorded with MK-801 intracellularly when NMDA was applied within 1–2 min after the whole cell recording configuration had been established. This is consistent with the larger length of the processes of these cells and the presumably correspondingly longer diffusion distance from the tip of the recording pipette and cell body to the location of the NMDA receptors. For the cell represented in Fig. 4B, the first NMDA-evoked response obtained after breaking into the cell displayed a peak amplitude of ∼20 pA. With repeated application of NMDA (once every 60 s), we observed a gradual reduction of the response amplitude such that at ∼4 min of recording, the response was almost completely abolished (Fig. 4B). For A17 amacrine cells recorded with MK-801 in the pipette solution, the average NMDA-evoked response was an inward current of 3 ± 0.4 pA (n = 4 cells, range 2.6–4.2 pA) after ∼4 min of recording.

Voltage-dependent block of NMDA receptors in AII and A17 amacrine cells.

NMDA receptors display a characteristic Mg²⁺-dependent voltage block (Nowak et al. 1984). To investigate this property for the NMDA receptors expressed by AII and A17 amacrine cells, we measured the I–V relationships of the NMDA-evoked responses in the presence and absence of Mg²⁺ in the extracellular solution. We recorded NMDA-evoked currents at a series of holding potentials between −80 and +60 mV. To block K⁺ conductances in the cells, including voltage-dependent conductances, the recording pipette solution contained Cs⁺ and TEA⁺ (see methods). In the presence of extracellular Mg²⁺, the I–V curves for AII amacrines displayed a characteristic J-shape, with a negative slope conductance between −70 and −30 mV (Fig. 5, A and C, filled circles), as is expected for NMDA receptors. When Mg²⁺ was omitted from the bath and puffer pipette solution, the NMDA-evoked responses were markedly enhanced at negative holding potentials (Fig. 5B) and the corresponding I–V curves became considerably more linear (Fig. 5, B and C, open circles). In the presence of Mg²⁺, the E_rev for AII amacrine cells was 5.9 ± 3.6 mV (range −2.0 to 16.1 mV; n = 6 cells). and in the absence of Mg²⁺, it was 4.8 ± 4.0 mV (range −6.1 to 18.0 mV; n = 5 cells). There was no statistically significant difference between these values (P = 0.8377; unpaired t-test).

Fig. 5.

Current-voltage (I–V) relationships of NMDA-evoked whole cell current responses in AII and A17 amacrine cells. A: NMDA-evoked (1 mM, 1 s) responses in an AII amacrine cell, normal (1 mM) Mg²⁺ extracellularly ([Mg²⁺]_o). Here and in subsequent figures, voltages next to the current traces indicate holding potential of the cell (−80 to +60 mV), and the circle below the traces indicates position of the time window for averaging the response amplitude. Here and in B, the extracellular solution contained CNQX and TTX (see methods). B: NMDA-evoked (1 mM, 1 s) responses in an AII amacrine cell, nominally Mg²⁺-free extracellular solution. Notice larger NMDA-evoked responses at negative holding potentials compared with A. C, left: I–V relationship for NMDA-evoked peak responses in 6 AII amacrines in normal Mg²⁺ extracellularly (as in A; filled circles) and 5 AII amacrines in nominally Mg²⁺-free extracellular solution (as in B; open circles). Here and in subsequent figures, the NMDA-evoked current at each holding potential is plotted as mean ± SE (n = 4–6 cells for each data point). Data points are connected by straight lines. Right, I–V relationship for NMDA-evoked peak responses. Data are the same as at left, but with data points normalized (norm) to the current at +40 mV. Data points are connected by straight lines. D: NMDA-evoked (1 mM, 1 s) responses in an A17 amacrine cell, normal (1 mM) Mg²⁺ extracellularly. Here and in E, the extracellular solution contained strychnine, bicuculline, CNQX, and TTX (see methods). E: NMDA-evoked (1 mM, 1.5 s) responses in an A17 amacrine cell, nominally Mg²⁺-free extracellular solution. Notice larger NMDA-evoked responses at negative holding potentials compared with D. F, left: I–V relationship for NMDA-evoked peak responses in seven A17 amacrines in normal Mg²⁺ extracellularly (as in D; filled circles) and 6 A17 amacrines in nominally Mg²⁺-free extracellular solution (as in E; open circles; n = 4–7 cells for each data point). Data points are connected by straight lines. Right, I–V relationship for NMDA-evoked peak responses. Data are the same as at left, but with data points normalized to the current at +40 mV. Data points are connected by straight lines.

When A17 amacrine cells were tested with NMDA at different holding potentials, we noticed that for some cells (with normal extracellular Mg²⁺), the rise time was longer for responses at negative potentials between −80 and −30 mV (Fig. 5D). Even with good voltage and space clamp control, such as when recording from round cells without processes, the inward current through NMDA receptor channels is not zero at these negative potentials. Because of the negative slope conductance, this current could rise to a regenerative depolarization with gradual release from the Mg²⁺-dependent voltage block when recording NMDA-evoked currents in A17 amacrine cells with long, thin processes and suboptimal conditions for good space clamp. This could contribute to the slow increase in the inward current at negative potentials observed for some A17 amacrines (Fig. 5D).

Because of these problems with A17 amacrine cells, we measured the current responses at the negative holding potentials at the same point in time where the peak responses at other holding potentials occurred. The I–V curve displayed a slight J-shape with strong outward rectification (Fig. 5, D and F, filled circles). Similar to the result in AII amacrine cells, when we measured NMDA-evoked responses in the absence of Mg²⁺ in the extracellular solution, the responses were markedly enhanced at negative holding potentials (Fig. 5E), consistent with alleviation of a Mg²⁺-dependent voltage block. In this condition, the I–V curves were considerably more linear (Fig. 5F, open circles). In the presence of Mg²⁺, the E_rev for A17 amacrine cells was 4.5 ± 6.2 mV (range −22 to 22 mV, n = 7 cells), and in the absence of Mg²⁺, it was 8.0 ± 4.8 mV (range −5.6 to 27 mV; n = 6 cells). The difference between these values was not statistically significant (P = 0.67749; unpaired t-test).

Blocking gap junction-mediated coupling with MFA does not block NMDA-evoked responses in AII or A17 amacrine cells.

We have previously demonstrated that MFA completely blocks the electrical synapses between AII amacrine cells and between AII amacrine cells and ON-cone bipolar cells (Veruki and Hartveit 2009). To exclude the possibility that the NMDA-evoked responses observed in either AII or A17 amacrine cells were mediated indirectly by gap junction coupling to other cells, we repeated application of NMDA while we blocked gap junction coupling pharmacologically with MFA. NMDA (1 mM, 1-s duration) was applied once every 60 s, starting immediately after the whole cell recording configuration had been established, when we also added MFA (100 μM) to the extracellular solution to block electrical coupling. Because it takes ∼30 min before the electrical coupling is completely blocked (Veruki and Hartveit 2009), the application of NMDA was continued for 40–45 min in the maintained presence of MFA. For the AII amacrine cell represented in Fig. 6A and the A17 amacrine cell represented in Fig. 6B, there was virtually no change in the responses over time. Similar results were seen for two other AII amacrine cells. These results strongly suggest that NMDA-evoked responses in AII or A17 amacrine cells were mediated by receptors on the cells themselves and not indirectly by gap junction coupling to other neurons.

Fig. 6.

Blocking gap junction coupling pharmacologically with meclofenamic acid (MFA) does not block NMDA-evoked responses in AII or A17 amacrine cells. A: currents activated in an AII amacrine cell (V_hold = −60 mV) by application of NMDA (1 mM, 1 s) from a puffer pipette at 7 (top), 27 (middle), and 45 min (bottom) of whole cell recording with MFA (100 μM) in the extracellular solution. B: currents activated in an A17 amacrine cell (V_hold = −60 mV) by application of NMDA (1 mM, 1 s) from a puffer pipette at 7 (top), 23 (middle), and 38 min (bottom) of whole cell recording with MFA (100 μM) in the extracellular solution. NMDA was applied every 60 s (A and B). Notice how NMDA-evoked responses are maintained in the presence of MFA (A and B).

Neurotransmitter receptors in nucleated patches from AII and A17 amacrine cells.

For a more detailed biophysical analysis of the functional properties of the NMDA responses expressed by AII and A17 amacrine cells, whole cell recording and drug application from puffer pipettes is not an ideal method. The voltage control, including space clamp, is suboptimal in recordings from both AII and A17 amacrine cells, both because of the presence of homologous and heterologous electrical coupling mediated by gap junctions (reviewed by Hartveit and Veruki 2012) and because of the presence of long and thin processes, particularly on the A17s (Fig. 1B). In addition, the speed of drug application obtainable with puffer pipettes and in vitro slices is too slow and variable for adequately measuring the kinetic properties of neurotransmitter receptors. On this background, we attempted a more detailed kinetic and biophysical investigation of the I–V properties of NMDA receptors expressed by AII and A17 amacrine cells by recording NMDA-evoked responses from nucleated patches. These are comparable to small round cells that offer a much better voltage- and space-clamp control and can be tested with faster drug application. In our laboratory, we have previously observed responses of conventional outside-out or nucleated patches from AII amacrine cells to application of several neurotransmitters and receptor agonists, including glutamate (Veruki et al. 2003), kainate (Mørkve et al. 2002), AMPA (Castilho et al. 2015b; Mørkve et al. 2002), and glycine (Gill et al. 2006).

When we tested nucleated patches from AII and A17 amacrine cells, we applied NMDA (1 mM), GABA (1 mM), and glycine (1 mM) from separate barrels of a multibarrel pipette. For AII amacrine cells, 11/13 nucleated patches responded to GABA with robust currents and 13/13 patches responded to glycine with robust currents (Fig. 7A). We did not perform a detailed investigation of the I–V properties of the GABA and glycine receptors but observed inward and outward currents at holding potentials of −60 and +20 mV, respectively (Fig. 7A), as expected for chloride-permeable channels and E_Cl ∼0 mV. When the same 13 nucleated patches were tested with NMDA (at holding potential of −60 and +20 mV), 10/13 patches displayed no response and 3/13 patches responded with minimal currents (≤3 pA) and a small increase in membrane noise (Fig. 7A).

Fig. 7.

Nucleated patches from AII amacrine cells do not respond to NMDA. A: no response of nucleated patch to application of NMDA (1 mM, 3 s; top right traces) at −60 or +20 mV (as indicated), normal (1 mM) Mg²⁺ extracellularly. Here and in B, NMDA was applied together with the co-agonist glycine (10 μM) and strychnine (10 μM) and picrotoxin (100 μM) to block conventional glycine receptors. Robust responses of same nucleated patch to application of glycine (1 mM, 100 ms; bottom left traces) and GABA (1 mM, 500 ms; bottom right traces) at −60 mV (inward currents) and +20 mV (outward currents). B: clear response to NMDA of AII amacrine cell (voltage clamped at +40 mV, normal Mg²⁺ extracellularly) during whole cell recording in retinal slice (top trace; recording configuration indicated in top schematic), but no response when tested with NMDA after the nucleated patch was pulled (bottom trace; recording configuration indicated in bottom schematic).

For A17 amacrine cells, 15 nucleated patches were examined with drug application. All 15 patches were tested with GABA and responded with robust currents (Fig. 8A). Of the same 15 patches, 11 were tested with glycine and all responded with robust currents (Fig. 8A). When we tested these same patches with NMDA (in Mg²⁺-free extracellular solution and a holding potential of −60 mV), 12/15 patches displayed no response and 3/15 patches responded with minimal currents (≤3 pA) and an increase in membrane noise (Fig. 8A).

Fig. 8.

Nucleated patches from A17 amacrine cells show no or only minimal response to NMDA. A: no response of nucleated patch to application of NMDA (1 mM, 1 s; top right traces) at −60 or +20 mV (as indicated). Robust responses of same nucleated patch to application of glycine (1 mM, 500 ms; bottom left traces) and GABA (1 mM, 500 ms; bottom right traces) at −60 mV (inward currents) and +20 mV (outward currents). B: clear response to NMDA of A17 amacrine cell (voltage clamped at −60 mV) during whole cell recording in retinal slice (top trace; recording configuration indicated in top schematic), but only minimal response with increase in membrane noise when tested with NMDA after the nucleated patch was pulled (bottom trace; recording configuration indicated in bottom schematic).

These results suggested that the general lack of NMDA-evoked responses in nucleated patches from AII and A17 amacrine cells reflects a lack of receptors and is not due to technical problems. As an additional control, we performed experiments where we first verified the presence of an NMDA-evoked response in the whole cell configuration, isolated a nucleated patch from the same cell, and tested it with application of NMDA. For AII amacrine cells tested in this way (n = 3), all displayed robust responses to NMDA in the whole cell configuration, but only one cell responded after isolating a nucleated patch and the amplitude was <3 pA (Fig. 7B). Similarly, when we tested A17 amacrine cells (n = 3), all displayed robust whole cell responses to NMDA. When the corresponding nucleated patches were tested, all displayed a visible response to NMDA, but the responses were small (3.8 ± 0.8 pA, range 2.6–5.4 pA; Fig. 8B). Taken together, these experiments strongly suggested that there are very few NMDA receptors located in the cell membrane of the cell bodies of both AII and A17 amacrine cells, unfortunately precluding a rigorous kinetic and biophysical analysis.

Agonist-dependent suppression of NMDA receptor-mediated responses in AII and A17 amacrine cells.

In a previous study of NMDA receptors in AII amacrine cells, we observed a relatively fast time- and agonist-dependent reduction of NMDA-evoked responses, with few cells maintaining responses after being stimulated with repeated application of NMDA for 10–15 min (Hartveit and Veruki 1997). Because the phenomenon seemed to depend on the duration of whole cell recording, it was suggested that it was a consequence of receptor rundown, potentially reflecting the perturbation of neuromodulatory control. Such control is often mediated by phosphorylation of specific amino acids in receptor proteins, and there is strong evidence that NMDA receptors in different types of neurons are subject to such control (reviewed by Salter et al. 2009). There is also evidence, however, that NMDA receptors are influenced by more than one mechanism of desensitization (reviewed by Gibb 2010). On this background, we hypothesized that if the observed response reduction corresponded to classical rundown, it should be irreversible over time, whereas if it was caused, for example, by desensitization following agonist-dependent receptor activation, it should reverse, after reducing the frequency or intensity of stimulation.

We tested both AII and A17 amacrine cells by applying NMDA (1-s duration) at intervals of ∼60 s for recording periods of 20–50 min (Fig. 9). The recordings were performed in Mg²⁺-free extracellular solution at negative holding potentials, near the resting membrane potentials of the cells. With this paradigm, we sometimes observed a sequential increase or decrease in response to the first three or four NMDA-evoked responses, but after that initial change in response, we observed no consistent change of the amplitude of NMDA-evoked responses in either AII (Fig. 9A) or A17 (Fig. 9B) amacrine cells within our recording periods.

Fig. 9.

NMDA-evoked responses in AII and A17 amacrine cells display temporal stability without rundown. A, top: time series of peak response amplitude of an AII amacrine cell (V_hold = −60 mV) to repeated application of NMDA (1 mM, 1-s duration, 60-s interval) during 50 min. Notice stability of response amplitude. Bottom, time series of peak response amplitude of AII amacrine cells (as in top panel; n = 4 cells) to repeated application of NMDA (1 mM, 1 s). Here and in B, time 0 indicates the time of breaking into the cell and establishing the whole cell recording configuration; data points are plotted as means ± SE and normalized to the current evoked by the first application for each cell (n = 1–4 cells for each data point, depending on duration of recording). B, top: time series of peak response amplitude of an A17 amacrine cell (V_hold = −70 mV) to repeated application of NMDA (1 mM, 1-s duration, 60-s interval) during 26 min. Notice stability of response amplitude. Bottom, time series of peak response amplitude of A17 amacrine cells (as in top panel; n = 6 cells) to repeated application of NMDA (1 mM, 1 s; n = 3–6 cells for each data point, depending on duration of recording).

To investigate the conditions for the response reduction previously observed in our laboratory (Hartveit and Veruki 1997), we repeated the application of NMDA at more frequent intervals. For all cells, we first verified a stable baseline for ∼5 min with an application interval of 60 s, before switching to more frequent applications of 10- to 15-s intervals. For the AII amacrine cell represented in Fig. 10A, the response to low-frequency application of NMDA was relatively stable at around 250 pA. When the application interval was reduced from 60 s to 10 s, this led to an immediate reduction in the response amplitude that plateaued at around 170 pA within 1–2 min, corresponding to a suppression of about 34%. In all cases, we made sure that the stimulation interval was sufficiently long for the membrane current to have recovered to the baseline level between subsequent applications of NMDA. After 5 min, the application interval was increased again to 60 s and the response amplitude increased over ∼2 min, with an almost complete reversal of the suppression seen during the 10-s-interval application period (Fig. 10A). During a second period of 10-s-interval applications, the suppression was slightly stronger (43%). When the application interval was increased to 60 s again, the responses recovered almost fully to the previous level (Fig. 10A). For the whole recording period, the series resistance varied slightly, but the change was gradual and unrelated to the temporally distinct suppression seen during the periods with increased stimulus frequency (Fig. 10A). For seven AII amacrine cells tested in this way, the NMDA-evoked response decreased in all cells, from 112 ± 29 pA (range 42–254 pA) with an application interval of 60 s to 72 ± 24 pA (range 19–168 pA) with an application interval of 10–15 s, corresponding to an average decrease of 43 ± 9% (Fig. 10B; P = 0.0067, paired t-test). For five of the cells, we were able to maintain the recordings sufficiently long that recovery could be observed (Fig. 10B).

Fig. 10.

Frequency-dependent suppression of NMDA-evoked responses in AII and A17 amacrine cells. A, top: time series of peak response amplitude of an AII amacrine cell (V_hold = −60 mV) to repeated application of NMDA (1 mM, 1-s duration). For the cell represented, testing started after about 20 min of recording. Interval between pulses: 60 s (22–29 min), 10 s (29–33 min), 60 s (34–43 min), 10 s (43–46 min), and 60 s (47–56 min). Notice response reduction during higher-frequency application of NMDA. Numbers (1, 2, and 3) correspond to raw data traces (bottom) displaying responses evoked by NMDA application during lower- (1, 3) and higher-frequency (2) stimulation periods. Here and in C, time 0 indicates the time of breaking into the cell and establishing the whole cell recording configuration. Middle, time series of corresponding estimates of series resistance (R_s) for whole cell recording at top. R_s estimate was updated between each application of NMDA. Notice relative stability of R_s. B: peak amplitude of NMDA-evoked responses (bars represent means ± SE) in AII amacrine cells (n = 7 cells) in control condition with 60-s intervals between NMDA applications (Control; left bar), during higher-frequency stimulation with 10- to 15-s intervals between NMDA applications (High-freq.; middle bar), and after return to the control condition (Recovery; right bar). Data points for the same cell are connected by lines, and the results from statistical comparisons between averages are indicated (*P < 0.05, statistically significant difference). C, top: time series of peak response amplitude of an A17 amacrine cell (V_hold = −70 mV) to repeated application of NMDA (1 mM, 1-s duration). Interval between pulses: 60 s (3–9 min), 12 s (9–15 min), 60 s (17–25 min), and 12 s (26–32 min). Notice response reduction during higher-frequency application of NMDA. Numbers (1, 2, and 3) correspond to raw data traces (bottom) displaying responses evoked by NMDA application during lower- (1, 3) and higher-frequency (2) stimulation periods. Middle, time series of corresponding estimates of R_s for whole cell recording at top. R_s estimate was updated between each application of NMDA. Notice relative stability of R_s. D: peak amplitude of NMDA-evoked responses in A17 amacrine cells (n = 8 cells) in control condition with 60-s intervals between NMDA-applications (Control; left bar), during higher-frequency stimulation with 10- to 15-s intervals between NMDA applications (High-freq.; middle bar), and after return to the control condition (Recovery; right bar).

The results for A17 amacrine cells were very similar to those for AII amacrine cells. For the A17 amacrine cell represented in Fig. 10C, the NMDA-evoked response was suppressed by about 60% when the application interval was reduced from 60 s to 12 s. The suppression was reversible and could be repeated (Fig. 10C). The series resistance increased slightly during the recording period but was unrelated to the suppression evoked by increased stimulus frequency (Fig. 10C). For eight A17 amacrine cells tested in this way, the NMDA-evoked response decreased in all cells, from 106 ± 19 pA (range 27–182 pA) with an application interval of 60 s to 57 ± 13 pA (range 12–107 pA) with an application interval of 10–15 s, corresponding to an average decrease of 49 ± 3% (Fig. 10D; P = 0.00042, paired t-test). The suppression reversed almost completely when the application interval was increased again (Fig. 10D; 1 cell was lost before recovery was observed).

Reducing extracellular Ca²⁺ increases NMDA responses in AII and A17 amacrine cells.

One possible mechanism that can explain the response suppression observed when the frequency of application was increased is a negative feedback mediated by Ca²⁺ influx through the NMDA receptor channel itself (Rosenmund and Westbrook 1993). Influx of Ca²⁺, acting via calmodulin, constitutes a system for negative feedback of the Ca²⁺-permeable NMDA receptor channels (Ehlers et al. 1996; Zhang et al. 1998). To examine the potential Ca²⁺-sensitivity of NMDA receptors in AII and A17 amacrine cells, we applied NMDA (1-s duration) at 60-s intervals, and after a stable baseline had been established in Mg²⁺-free extracellular solution, we switched to a Mg²⁺- and Ca²⁺-free extracellular solution. For the AII amacrine cell represented in Fig. 11A, the reduced extracellular Ca²⁺ immediately led to an increased response amplitude from a baseline response of ∼250 pA to a level of ∼430 pA, corresponding to an increase of 72%. When we switched back to the original extracellular solution and Ca²⁺ was washed back in, the amplitude of the NMDA-evoked responses returned to ∼275 pA (Fig. 11A). For five AII amacrine cells tested in this way, the NMDA-evoked response increased in all cells, from 142 ± 33 pA (range 67–252 pA) in control (Mg²⁺-free) bath solution to 250 ± 52 pA (range 118–433 pA) in Mg²⁺- and Ca²⁺-free bath solution, corresponding to an average increase of 81 ± 13% (Fig. 11B; P = 0.0067, paired t-test). For three cells, we observed a reversal of the increase when we switched back to normal Ca²⁺; for the other two cells, the recording was lost before we could observe recovery.

Fig. 11.

Ca²⁺-dependent suppression of NMDA-evoked responses in AII and A17 amacrine cells. A, top: time series of peak response amplitude of an AII amacrine cell (V_hold = −60 mV) to repeated application of NMDA (1 mM, 1-s duration) at intervals of 60 s, first in normal extracellular solution ([Ca²⁺]_o = 2.5 mM), followed by nominally Ca²⁺-free extracellular solution (continuous line) and return to normal extracellular solution. Notice response increase in Ca²⁺-free solution. Numbers (1, 2, and 3) correspond to raw data traces (bottom) displaying responses evoked by NMDA application in normal (1, 3) and Ca²⁺-free (2) bath solution. B: peak amplitude of NMDA-evoked responses in AII amacrine cells (n = 5 cells) in control solution with [Ca²⁺]_o = 2.5 mM (Control; left bar), in nominally Ca²⁺-free extracellular solution ([Ca²⁺]_o = 0; middle bar), and after return to control solution with [Ca²⁺]_o = 2.5 mM (Recovery; right bar). C, top: time series of peak response amplitude of an A17 amacrine cell (V_hold = −60 mV) to repeated application of NMDA (1 mM, 1-s duration) at intervals of 60 s, first in normal extracellular solution ([Ca²⁺]_o = 2.5 mM), followed by nominally Ca²⁺-free extracellular solution (continuous line) and return to normal extracellular solution. Notice response increase in Ca²⁺-free solution. Numbers (1, 2, and 3) correspond to raw data traces (bottom) displaying responses evoked by NMDA application in normal (1, 3) and Ca²⁺-free (2) bath solution. D: peak amplitude of NMDA-evoked responses in A17 amacrine cells (n = 5 cells) in control solution with [Ca²⁺]_o = 2.5 mM (Control; left bar), in nominally Ca²⁺-free extracellular solution ([Ca²⁺]_o = 0; middle bar), and after return to control solution with [Ca²⁺]_o = 2.5 mM (Recovery; right bar).

When A17 amacrine cells were tested in the same way, the results were very similar to those observed for AII amacrine cells. For the A17 amacrine cell represented in Fig. 11C, the response increased by 110%. For five A17 amacrine cells, the NMDA-evoked response increased in all cells, from 74 ± 22 pA (range 17–133) in control (Mg²⁺-free) bath solution to 118 ± 32 pA (range 36–201) in Mg²⁺- and Ca²⁺-free bath solution, corresponding to an average increase of 71 ± 11% (Fig. 11D; P = 0.0108, paired t-test). Reversal was observed for three cells, with two cells lost before recovery could be observed.

DISCUSSION

In this study we have used a combination of physiological, pharmacological, and biophysical methods to investigate the potential expression of functional (i.e., conductance-increasing) NMDA receptors by AII and A17 rod amacrine cells, as well as rod bipolar cells, in the rat retina. This information is important for understanding the neurotransmission between glutamatergic bipolar cells and rod amacrine cells, including possible mechanisms of synaptic integration and plasticity. The present results provide strong evidence that NMDA receptors are expressed by both AII and A17 amacrine cells, suggesting that these receptors can have a more important role in mediating glutamatergic signals to both types of amacrines than previously suspected. In the following, we discuss the basis for our conclusion, discuss possible implications with respect to functional roles for NMDA receptors in these cells, and suggest how these questions might be investigated in future work.

For rod bipolar cells, we found no evidence for expression of functional (conductance-increasing) NMDA receptors. This is consistent with a previous report from our laboratory (Hartveit 1996), but we refined our investigation in the current study by increasing the spatial area over which the puffer pipette was positioned to ensure that we would not miss any region of the cell, and we made sure that fast rundown of receptor channels could not explain the absence of responses by minimizing the interval between breaking into the cells and applying NMDA. Earlier physiological evidence for functional NMDA receptors on rod bipolar cells is relatively weak, with no investigation of I–V properties and antagonist specificity (Karschin and Wässle 1990). The current results cannot by themselves explain previous evidence for immunoreactivity of rod bipolar cells for GluN1 (Lo et al. 1998) and GluN2D (Wenzel et al. 1997) subunits. Assuming that the immunolabeling cannot be explained by antibody cross-reactivity, the physiological relevance of the expression is questioned by the absence of any evidence for a conductance that can be activated by NMDA.

Evidence for expression of NMDA receptors by AII and A17 amacrine cells.

Whether a specific type of neuron expresses a specific type of neurotransmitter receptor is apparently a simple and well-defined question, but strong and unequivocal evidence can often be difficult to obtain. First, the type of neuron needs to be well defined, and it is necessary to be able to uniquely identify and target individual neurons of the right type. In our case, both AII and A17 amacrine cells are well-defined types of amacrine cells with unique cellular morphologies (reviewed by Masland 2001) and reasonably well-characterized synaptic inputs and outputs (reviewed by Hartveit and Veruki 2012). In addition, as demonstrated in this study and several previous studies, both from our laboratory (Castilho et al. 2015a; Mørkve et al. 2002; Veruki and Hartveit 2002a, 2002b, 2009; Veruki et al. 2003) and from others (e.g., Boos et al. 1993; Eggers and Lukasiewicz 2006; Grimes et al. 2010, 2014; Menger and Wässle 2000; Singer and Diamond 2003), both AII and A17 amacrine cells can be reliably identified and targeted in the acute retinal slice preparation (Fig. 1A). The morphology observed with IR video microscopy and contrast enhancement techniques (e.g., DIC) is not sufficient, however, for a secure identification. Therefore, all cells we recorded from were also examined with fluorescence microscopy after having been filled with a fluorescent dye in the intracellular solution, allowing us to visually inspect the complete neuronal morphology as preserved in the acute retinal slice preparation (Fig. 1B).

In our physiological experiments, we stimulated NMDA receptors by puffer pipette application of the specific agonist NMDA. Binding of agonist to the receptor triggers channel opening and the resulting current can be recorded as an inward current. When such currents are measured, a series of control experiments are necessary before one can conclude that specific receptors are expressed by the neuron being recorded. First, it is necessary to demonstrate that the response is mediated by specific receptors. Second, it is necessary to verify that the activated receptors display functional properties that are characteristic for the specific receptor type. Third, it is important to provide evidence that the responses are mediated by receptors expressed by the cells recorded from, as opposed to being mediated by specific receptors expressed by other types of neurons and conveyed to the cells recorded from through synaptic network mechanisms.

With respect to pharmacological specificity, two sets of experiments argue for the involvement of specific NMDA receptors. First, the responses evoked by application of NMDA could be completely blocked by CPP, a competitive antagonist that is specific for NMDA receptors (Davies et al. 1986). Second, the responses were also abolished (for AII amacrines) and strongly suppressed (for A17 amacrines) when the noncompetitive NMDA receptor antagonist and open channel blocker MK-801 (Huettner and Bean 1988; Wong et al. 1986) was applied intracellularly. The difference between AII and A17 amacrine cells is most likely due to lower concentrations of MK-801 that reach the location of the receptors in A17 amacrines because these cells have very long and thin processes that restrict and delay spread by diffusion intracellularly.

With respect to functional properties, we demonstrated that the responses evoked by NMDA in whole cell recordings from both AII and A17 amacrine cells displayed I–V properties that are unique for NMDA receptors. First, the I–V curves displayed the classical J-shape, with negative slope conductance between −80 and −30 mV (Nowak et al. 1984). From detailed biophysical investigations, we know that this reflects a Mg²⁺-dependent voltage block of the ion channel associated with the NMDA receptor complex (Nowak et al. 1984). Second, when Mg²⁺ was removed from the extracellular solution, the voltage block disappeared and the shape of the I–V curves became more linear for both AII and A17 amacrine cells. For both AII and A17 amacrine cells, the E_rev of the NMDA-evoked current was close to 0 mV, consistent with activation of nonselective cation channels.

The I–V relationship of the NMDA response and the results with MK-801 also provide strong evidence that the NMDA receptors must be localized on the AII and A17 amacrine cells themselves. First, if the NMDA receptors were not localized on the cell from which the whole cell recording was performed, it is unlikely that changing the holding potential, as was done in the experiments investigating I–V properties, would have changed the driving force to generate the observed I–V curves. For example, AII amacrine cells are coupled to ON-cone bipolar cells via electrical synapses mediated by gap junctions (Strettoi et al. 1992, 1994; Veruki and Hartveit 2002b), and when recording from ON-cone bipolar cells that do not express non-NMDA receptors, it is possible to measure responses evoked by non-NMDA receptor agonists, presumably mediated by activating receptors on AII amacrines (Hartveit 1997). Importantly, however, it is not possible to reverse the direction of the evoked currents, presumably because the resistance of the electrical synapses is too high to adequately control the voltage at the location of the ion channels being activated.

Second, because the NMDA receptor antagonist MK-801 blocked responses mediated by NMDA in both AII and A17 amacrine cells after being applied intracellularly, the most likely interpretation is that the responses were mediated by receptors located in the AII and A17 amacrine cells themselves, where MK-801 could diffuse to the receptors within the processes of these cells. It has to be taken into account, however, that with puffer pipette application of drugs to cells in in vitro slice preparations, it is difficult to control the spatial extent of application, and we do not know the extent to which pressure-applied NMDA, directed at a specific cell, could also reach neighboring cells. Whereas MK-801 strongly suppressed NMDA-evoked responses in both AII and A17 amacrine cells, we cannot exclude the possibility that MK-801 could diffuse across gap junctions that couple these cells to other neurons. On the other hand, current knowledge of the possibilities available for such coupling suggests that if such diffusion of MK-801 should occur, it would not change the conclusion. For homologous coupling, i.e., gap junctions that couple AII amacrines to other AII amacrines (Strettoi et al. 1992; Veruki and Hartveit 2002a) and A17 amacrines to other A17 amacrines (Grimes et al. 2014; Li et al. 2002), any indirectly mediated NMDA receptor-mediated responses blocked by diffusion of MK-801 would originate from NMDA receptors in cells of the same type as recorded from. There is also the possibility that heterologous coupling between AII amacrine cells and ON-cone bipolar cells (Strettoi et al. 1992, 1994; Veruki and Hartveit 2002b) could mediate indirect responses, but existing evidence suggests that ON-cone bipolar cells do not express NMDA receptors (Hartveit 1997). Irrespective of the possibilities for indirect effects mediated via gap junction coupling, NMDA-evoked responses in both AII and A17 amacrine cells were resistant to the gap junction blocker MFA, strongly suggesting that the NMDA-evoked responses of these neurons were mediated by receptors expressed on the cells themselves.

Because neurons are branched structures with dendritic processes that can extend over hundreds of micrometers, it can be challenging and even impossible to obtain the degree of experimental control that is required for rigorous biophysical investigations of the functional properties of ion channels and receptors expressed along these processes. Whereas investigating the functional properties in isolated membrane patches, either conventional outside-out patches or nucleated patches, can to a large extent compensate for this (e.g., Hartveit and Veruki 2007), we were unfortunately unsuccessful in using this approach for more detailed investigations of the NMDA receptors expressed by AII and A17 amacrine cells. Apart from the apical dendrites of AII amacrines, the processes of both types of cells are too thin for excising outside-out patches. Nucleated patches can be isolated for both types of cells, but the NMDA-evoked responses were unfortunately too small and infrequent to allow adequate analysis. As a control that the lack of responses was not due to technical or methodological problems, we performed experiments where we applied the receptor ligands GABA and glycine to the same patches. These ligands evoked robust responses, confirming previous results for non-NMDA-type glutamate receptors (Mørkve et al. 2002; Veruki et al. 2003) and glycine receptors (Gill et al. 2006) at the cell bodies of AII amacrine cells and indicating that the lack of responses obtained with NMDA was due to lack of the receptors themselves and not technical problems. The results for A17 amacrine cells suggest that functional properties of GABA and glycine receptors in these cells can be fruitfully investigated by this technique in future studies.

Activity-dependent suppression of NMDA receptors in AII and A17 amacrine cells.

The reduction of NMDA-evoked responses in AII amacrine cells previously observed in our laboratory was interpreted as reflecting rundown (Hartveit and Veruki 1997). Rundown is a phenomenon observed for several types of receptors and ion channels and is considered to be a consequence of washout of the intracellular environment after establishment of the whole cell recording configuration (Horn and Korn 1992). Accordingly, the response reduction will occur simply as a function of time. This kind of mechanism is not consistent with the observations made in the current study. First, by restricting the frequency of application of NMDA to once every 60 s, the response was very stable for recording periods up to 25–50 min. The ability to maintain stable NMDA responses for such extended time periods did not depend on recording with high series resistance or other measures (e.g., perforated patch) to slow or prevent rundown. This suggests that the NMDA receptors in AII and A17 amacrine cells are not particularly susceptible to rundown as such.

Second, by increasing the frequency of stimulation, corresponding to intervals of 10–15 s, we observed an immediate reduction in the response to NMDA that reversed quickly after reducing the frequency of stimulation. Thus our results suggest that the suppression of NMDA-evoked responses is an expression of an activity-dependent feedback mechanism. Although our results do not provide a mechanistic explanation, a possible explanation is Ca²⁺-dependent inactivation, which constitutes one form of desensitization of NMDA receptors (for review, see Gibb 2010) and has been observed for both native receptors and receptors in heterologous expression systems (Ehlers et al. 1996; Rosenmund and Westbrook 1993). This is supported by our observation of an increased response to NMDA in nominally Ca²⁺-free extracellular solution for both AII and A17 amacrine cells. Although we have not directly verified Ca²⁺ influx through NMDA receptor channels in either AII or A17 amacrine cells, NMDA receptors have been found to be the predominant source of Ca²⁺ signals in several types of neurons, because of both their high Ca²⁺ permeability and their slow kinetics (reviewed by Higley and Sabatini 2012). Further work is required to decide if this mechanism is operative for the NMDA receptors in AII and A17 amacrine cells and whether it plays a specific functional role for signal processing in these cells.

Functional role of NMDA receptors in AII and A17 amacrine cells.

Despite what we believe is strong evidence for the expression of NMDA receptors by both AII and A17 amacrine cells, very little is known with respect to any functional role of these receptors for the signal processing that takes place in these cells. For AII amacrine cells, there is no evidence for an NMDA receptor-mediated component in the response evoked by depolarization of a presynaptic rod bipolar cell (Singer and Diamond 2003). To our knowledge, nothing is known with respect to the receptors mediating input from (some) OFF-cone bipolar cells to the lobular appendages of AII amacrine cells (Strettoi et al. 1992, 1994; Veruki et al. 2003). Immunocytochemical investigations at the ultrastructural level have not found evidence for NMDA receptors postsynaptic to rod bipolar cells (Fletcher et al. 2000), but there is evidence for a presumed extrasynaptic location of the GluN1 subunit of the NMDA receptor at arboreal dendrites of AII amacrine cells (Kothmann et al. 2012). Interestingly, GluN1 was colocalized with Cx36, which mediates gap junction coupling between AII amacrine cells, and Kothmann et al. (2012) found evidence that NMDA receptors could be involved in regulating the strength of coupling between these cells.

For A17 amacrine cells, the reciprocal feedback inhibition that is triggered by depolarization and glutamate release from rod bipolar cells is not blocked in the presence of the NMDA receptor antagonist CPP (Hartveit 1999), although similar experiments with Mb1 bipolar cells in the goldfish retina suggest that NMDA receptors can contribute to the feedforward excitation of inhibitory amacrine cells, at least under some conditions (Vigh and von Gersdorff 2005). Rod bipolar cells seem to constitute the only source of glutamatergic, bipolar cell input to A17 amacrine cells (Nelson and Kolb 1985). If NMDA receptors in the A17 are not localized postsynaptic to rod bipolars, the only alternative left open would seem to be an extrasynaptic location.

Given that expression of NMDA receptors now seems to be a property that is shared by both types of amacrine cells postsynaptic to rod bipolar cells, it is possible that information about the subunit composition of the NMDA receptors expressed by the two cell types can provide valuable information. For many regions and cell types in the central nervous system, it is known that the NMDA receptor subunit expression and composition displays striking variation, not only between regions, but also between cell types and even between different subcellular regions within single neurons (Nusser 2008; Paoletti et al. 2013). From in situ hybridization studies, we know that the GluN2A, GluN2B, and GluN2C subunits of the NMDA receptor are expressed in the rat retina, with considerable heterogeneity among amacrine cells in the inner nuclear layer, but the resolution is not sufficient to decide the specific identity of the various cells (Brandstätter et al. 1994). This, together with the mechanisms and conditions for activation of NMDA receptors on AII and A17 amacrine cells, remains an important question for future investigations. NMDA receptors might contribute directly to signaling in these amacrine cells, or their function could be important for modulation and plasticity of other ion channels that are more directly responsible for signal processing and integration.

GRANTS

This study was supported by Research Council of Norway Grants NFR 213776 (to M. L. Veruki) and NFR 214216 (to E. Hartveit).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Y.Z. and B.T. performed experiments; Y.Z., B.T., and M.L.V. analyzed data; Y.Z., B.T., E.H., and M.L.V. approved final version of manuscript; E.H. conception and design of research; E.H. and M.L.V. interpreted results of experiments; E.H. and M.L.V. drafted manuscript; E.H. and M.L.V. edited and revised manuscript; M.L.V. prepared figures.