| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1 Department of Ophthalmology and Visual Sciences, Department of Anatomy and Neurobiology, Washington University School of Medicine, 660 S. Euclid Avenue, St Louis, MO 63110, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
(Received 18 January 2005;
accepted after revision 9 March 2005;
first published online 10 March 2005)
Corresponding author P. D. Lukasiewicz: Department of Ophthalmology/Campus Box 8096, Washington University School of Medicine, 660 S. Euclid Avenue, St Louis, MO 63110, USA. Email: lukasiewicz{at}vision.wustl.edu
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Recent work suggests that the contributions of OPL- and IPL-mediated lateral signalling are not this straightforward, and indicate that lateral signalling in the IPL mediates a component of the steady surround response in ganglion cells (Cook & McReynolds, 1998a; Taylor, 1999; Flores-Herr et al. 2001). Thus, lateral interactions in both the OPL and IPL could contribute to the ganglion cell steady surround, as suggested by Naka (1977).
In this study, we investigated how steady, presynaptic lateral inhibition in the OPL and IPL contributes to ganglion cell surround responses. Dim surround stimuli reduced the sensitivity of the ganglion cells to central illumination. GABAC receptor antagonists, which act primarily at bipolar cell axon terminals in salamander (Lukasiewicz et al. 1994; Cook & McReynolds, 1998a) and do not affect horizontal cell signalling to cones (Kamermans et al. 2001; Verweij et al. 2003) or bipolar cells (Hare & Owen, 1996; Cook & McReynolds, 1998a), blocked the reduction in light sensitivity. Brighter steady surround illumination caused a further reduction in ganglion cell light sensitivity that was unaffected by GABA receptor antagonists. Carbenoxolone and cobalt (< 10 µM), which reduce horizontal cell to cone signalling (Kamermans et al. 2001; Verweij et al. 2003), decreased the effect of the bright, but not dim, surround illumination, suggesting that this component was mediated by horizontal cells. We also assessed the role of the direct inhibitory pathway from amacrine cells to ganglion cells by recording light-evoked IPSCs. Unlike the IPL pathway to bipolar cells, this pathway was activated primarily by bright, and less by dim, illumination. Thus, lateral inhibition in both plexiform layers affects ganglion cell light sensitivity.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Larval tiger salamanders obtained from Charles Sullivan (Nashville, TN, USA) were kept in aquaria at 5°C on a 12 h light–dark cycle. Salamanders were anaesthetized in an ice bath until they were immobilized, and then decapitated and pithed before the eyes were enucleated. The experimental protocol was approved by the Washington University's Institutional Animal Care and Use Committee. Eyes were dark-adapted for at least 1 h before dissection. Dissection and recording procedures were performed under dim red illumination. Retinal slices were prepared as described by Lukasiewicz et al. (1994) and sliced at 300-µm intervals. For whole mount preparations, the isolated retina was cut into two pieces, which were placed in a chamber, ganglion cell side up, and were immobilized with nylon netting. Individual slices or retinal pieces were transferred to a recording chamber and viewed through an upright, fixed-stage microscope (Eclipse E-600-FN; Nikon, Tokyo, Japan) with a 40 x water-immersion lens and Hoffman modulation contrast optics. Using a gravity-fed perfusion system, the preparation was continuously superfused with a Ringer solution containing (mM): NaCl 112, KCl 2, CaCl2 2, MgCl2 1, glucose 5 and Hepes 5; pH adjusted to 7.8 with NaOH.
Whole-cell recording
Electrodes were pulled from borosilicate glass (IB150F-4; World Precision Instruments, Sarasota, FL, USA) with a P-97 Flaming/Brown micropipette puller (Sutter Instruments, Novato, CA, USA). Whole-cell recordings were made from ganglion cells by using 5-M
pipettes filled with solution containing (mM) caesium gluconate 95.25, TEA-Cl 8, MgCl2 0.4, EGTA 1 and Na-Hepes 10; pH adjusted to 7.5 with HCl. Lucifer yellow (0.1%, Aldrich Chemical Co, Milwaukee, WI, USA) was added to the pipette solution to identify ganglion-cell morphologies after electrophysiological recordings. Membrane potentials were corrected for junction potentials (–14.9 mV).
The voltage-clamp recordings were made with either 3900A Integrating Patch Clamp (Dagan, Minneapolis, MN, USA) or Axopatch 200A (Axon Instruments, Union City, CA, USA). Data were digitized and stored with a personal computer using TL-1 data acquisition system (Axon Instruments). Patchit software (White Perch Software, Somerville, MA, USA) was used to generate voltage command outputs, acquire data, control the light stimuli, and operate the drug superfusion system. Data were filtered at 1 kHz with a four-pole Bessel filter and were sampled at 0.5–10 kHz.
Drugs
The entire recording chamber was superfused by a gravity-fed perfusion system. A computer-controlled superfusion system was used to apply drug-containing Ringer solutions. For all experiments, glycine receptors were blocked with strychnine (5 µM). The GABAB receptor antagonist CGP55845was obtained from Tocris Cookson (Ballwin, MO, USA). All other chemicals were obtained from Sigma (St Louis, MO, USA).
Light stimulation
Centre and surround illumination were controlled separately. Red centre illumination (diameter, 180 µm) was obtained by placing a band-pass filter (peak wave length (
max) = 633 nm) in front of a tungsten–halogen light source. The centre stimulus was projected through the objective lens via a light pipe attached to the microscope camera port and presented every 30–60 s. The intensity of the unattenuated light stimulus in the plane of the preparation was equivalent to 4.52 x 108 photons µm–2 s–1 of 633-nm monochromatic light. Centre spot intensity was varied with neutral density filters. The surround stimulus was an annulus that was projected through the microscope condenser. The annulus pattern (for slice preparations: i.d., 300 µm; o.d. 2300 µm; and for isolated retina preparations: i.d., 800 µm; o.d., 3000 µm) was generated by masking the centre of the light path. A red light-emitting diode (LED) (max = 635 nm, HLMP3750A; Digi-Key, Thief River Falls, MN, USA) was used for the surround light source. Surround light intensity was controlled by varying the LED current. The maximum, unattenuated intensity of the LED was 1.53 x 106 photons µm2 sec1 at 635 nm.
Intensity–response functions were generated by plotting the EPSC at light onset (ON -EPSC) amplitude as a function of centre light intensity. Intensity–response relations were generated in either the absence or presence of surround illumination. Surround illumination was presented from 5 to 10 s prior to, and until the end of, centre stimulation. Similar stimulus durations used in previous studies did not result in desensitization of surround responses (Sakmann & Creutzfeldt, 1969; Thibos & Werblin, 1978; Cook & McReynolds, 1998a). Surround light intensity was of dim (–3 log) or bright (–1 to 0 log) intensity. After obtaining responses in control Ringer solution, responses were then obtained in the presence of drug-containing solutions.
For experiments with carbenoxolone we needed to obtain the light intensity–response relations rapidly. To accomplish this, we ramped the centre light stimulus intensity (diameter, 180 µm; max, 635 nm) by increasing the current to a red LED in incremental steps (each step was 0.5–1.0 log unit, 100–150 ms) (see Fig. 6A) (Werblin, 1971). The intensity of the unattenuated light stimulus in the plane of the preparation was similar to that of the tungsten–halogen light source (4.54 x 108 photons µm–2 s–1 at 635 nm). The threshold level to evoke an EPSC was first determined using conventional light-intensity steps. The ramp stimulus was varied from background response threshold intensity to saturation in less than a second. The ramp stimulus was presented every 60 s. Bright surround illumination was presented in combination with centre illumination as described above.
|
To ensure that surround illumination was not scattering into the centre and desensitizing responses to centre illumination, we only analysed cells in which dim surround illumination alone did not elicit an EPSC. Brighter surround illumination often elicited small, transient EPSCs (
13% of maximal centre response), which decayed completely before the receptive-field centre was activated. Increasing EPSC amplitudes in response to brighter centre illumination indicated that bright surround illumination did not saturate the centre response. Although bright surround light, which scattered into the centre, may have partially desensitized centre pathways, we still were able to determine the effects of antagonists on dim and bright surrounds signalling, as described below.
Analysis
For the conventional light-step stimulus experiments, peak ON -EPSC amplitude was measured by using Tack software (White Perch Software). Data were normalized to maximum response of the centre-evoked EPSC in control solution. SigmaPlot (SPSS, Chicago, IL, USA) curve-fitting routines were used to fit light intensity–response curves to the Hill equation:
|
|
For the ramp light-stimulus experiments, we calculated cumulative photons at a given time of the ramp by integrating photons per unit area over the time course of the ramp light stimulus. The light-evoked EPSC was re-plotted as a function of cumulative photons µm–2 (see Fig. 6B) to obtain the intensity–response relation. Response amplitude and cumulative photons were normalized to maximum amplitude and to the photon number that evoked half-maximum response, respectively.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
To determine the presynaptic effects of steady surround illumination, we measured the amplitudes of EPSCs evoked by illumination of the receptive-field centre (L-EPSCs) in the absence or presence of surround illumination. While previous studies have assessed salamander ganglion cell responses by using extracellular and intracellular voltage recordings (Thibos & Werblin, 1978; Cook & McReynolds, 1998a), we assessed their excitatory inputs directly by voltage clamping the ganglion cell to the reversal potential for the major inhibitory synaptic inputs (ECl). By measuring the L-EPSCs at ECl, we confined our analysis to the two lateral signalling pathways that were presynaptic to the ganglion cell dendrites. All recordings were obtained from ganglion cells that responded at light onset and offset (ON-OFF), the most frequently encountered subtype in the salamander slice preparation (Mittman et al. 1990). A single morphological class of ON-OFF ganglion cell was recorded from, which had dendritic processes that stratified in the mid-IPL, where cone-dominant bipolar cell inputs co-stratify (Wu et al. 2000). Because EPSCs at light offset (OFF L-EPSCs) showed significant variability in amplitude, we limited our analysis to the ON L-EPSCs. Steady surround illumination has similar effects on sustained and transient ON responses in salamander retina (Thibos & Werblin, 1978). Strychnine was included in the bath to block the change-sensitive surround inhibitory inputs (Cook et al. 1998) and it had no effect on steady surround ganglion cell responses (Cook et al. 1998; Cook & McReynolds, 1998a).
Illumination of the ganglion cell receptive-field centre elicited EPSCs that increased in amplitude with increasing light intensity (Fig. 1A, upper row). The L-EPSC was maximal at 0 log relative attenuation. In a few cells, at 0 log relative attenuation, the L-EPSC was reduced in amplitude, most probably attributable to scattered light stimulating the surround, which reduces the centre response amplitude (Cook & McReynolds, 1998a). The L-EPSC amplitudes were plotted as a function of light intensity in Fig. 1B (•).
|
). Increasing the surround illumination further reduced the L-EPSCs (Fig. 1A, lower row), increased the shift of the intensity–response function and suppressed the maximal response (Fig. 1B, 
). While the suppressive effect on the maximal response by bright surround illumination has also been reported by others (Thibos & Werblin, 1978), the circuitry underlying this effect is unknown. Figure 1C shows mean spot intensity–response curves obtained from 10 ganglion cells with centre or centre plus surround illumination. Maximal L-EPSC amplitude and light sensitivity (L50, the luminance eliciting a half-maximal response) varied from ganglion cell to ganglion cell, making comparisons across the population difficult. The L50 values varied from 3.7 x 10–4 to 1.2 x 10–2, with a mean value of 6.5 x 10–3 relative attenuation, which, after conversion to photon fluxes, were comparable to L50 values reported by Yang & Wu (1997) for salamander, cone-driven bipolar cells. To more easily compare intensity–response functions across different neurones, intensity–response curves were normalized to the maximal L-EPSC amplitude and the half-maximal light intensity of the spot function (Fig. 1C). Dim surround light shifted the control curve to the right, but there was no significant decrease in maximum L-EPSC amplitude (Fig. 1C; Table 1). Brighter surround light produced a further shifting of the curve to the right and also reduced the maximum L-EPSC amplitude of the average response (Fig. 1C; Table 1). These results are consistent with earlier extracellular- and intracellular-recording studies of mammalian and salamander ganglion cells (Sakmann & Creutzfeldt, 1969; Thibos & Werblin, 1978; Cook & McReynolds, 1998b).
|
GABA receptor antagonists alter ganglion cell EPSCs evoked by centre illumination
The receptive-field centre illumination evokes both the excitation and the surround inhibition in ganglion cells because the steady surround component extends through the centre component of the receptive field (Rodieck & Stone, 1965; Cook & McReynolds, 1998a). In the inner retina, there are amacrine cells with both narrow- and wide-field processes. The relative contributions of these two morphological classes of amacrine cells, and the types of inhibitory receptors that they use to signal, are unclear. To assess the effects of GABA antagonists on the inhibition activated by centre stimuli and attributed to amacrine cells, we first determined the effects of GABA receptor blockers on EPSCs evoked by centre illumination alone. L-EPSCs evoked by sub-saturating centre illumination and recorded in the absence or presence of GABA receptor antagonists are shown in Fig. 2A. The blockade of GABAA receptors by bicuculline reduced the amplitude of the L-EPSC, but prolonged the response. Blocking GABAC and GABAA receptors, by the subsequent addition of I4AA and picrotoxin, increased the amplitude and prolonged the L-EPSCs beyond both control conditions and in the presence of bicuculline.
|
GABAC receptor antagonists reduced the dim surround inhibition of ganglion cell L-EPSCs
As centre light-evoked EPSCs in ganglion cells were affected by a component of the surround, we assessed the effects of annular illumination on the GABAergic surround by comparing EPSCs elicited by centre and dim surround illumination with EPSCs in response to centre illumination alone. This allowed us to detect the direct actions of GABA receptor antagonists on wider-field surround signalling and control for any antagonist effects on central pathway signalling. When the GABAA receptor antagonist bicuculline was present, dim surround illumination still reduced the L-EPSCs (Fig. 3A). In the presence of bicuculline, dim surround shifted the intensity–response curve to brighter intensities and reduced the maximum amplitude of the L-EPSCs (Fig. 3B, Table 1). The inability of bicuculline to block the effects of dim surround illumination indicated that GABAA receptors did not mediate the predominant lateral inhibitory signals. The subsequent addition of the GABAC receptor antagonists, I4AA and picrotoxin to the bath blocked the effects of dim surround illumination, indicating that GABAC receptors mediated its effects (Fig. 3C and D, and Table 1). These data suggest that GABAC receptors, located primarily on bipolar cell terminals, mediate a large component of the lateral inhibitory signal elicited by dim surround illumination. Because GABAA receptors are present on salamander bipolar cell terminals, albeit to a lesser extent than GABAC receptors (Lukasiewicz et al. 1994), they may make a small contribution to surround signalling.
|
Activation of the receptive-field surround with brighter annular illumination produced a further reduction of the L-EPSC amplitude and caused an additional shift of the spot intensity–response curve to higher intensities, as well as reducing the maximum amplitude (Fig. 1, and Table 1). Although GABA receptor blockers were effective in blocking the effects of dim surround illumination, they had little effect on bright surround inhibition. Figure 4 shows L-EPSCs in either the absence or presence of bright surround illumination in the presence of antagonists. Neither bicuculline alone (Fig. 4A and B) nor bicuculline in combination with GABAC receptor blockers (Fig. 4C and D, and Table 1) reversed the effects of bright surround inhibition. These data suggest that different lateral inhibitory pathways mediated the effects of dim and bright surround illumination observed in ganglion cells.
|
To ensure that our results were not attributed to truncated lateral signal pathways, which may exist in the slice preparation, we repeated these experiments, and obtained the same results (data not shown), in the more intact retinal whole mount preparation. The same combination of GABAA and GABAC blockers that were used in the slice experiments, eliminated dim to moderate surround inhibition (n = 3), but did not affect the inhibition mediated by bright surround illumination (n = 5).
Bright surround inhibition of ganglion cell L-EPSCs was not mediated by GABAB receptors
Although ionotropic GABA receptors did not mediate the bright surround response, it is possible that inhibition by bright surround illumination was mediated through GABAB receptors located in the IPL. GABAB receptors were shown to mediate presynaptic inhibition at bipolar cells (Maguire et al. 1989), which reduced light-evoked ganglion cell EPSCs (Shen & Slaughter, 2001). We tested whether GABAB receptors were involved in the suppression of L-EPSCs mediated by bright surround illumination by comparing the effects of centre and centre plus surround illumination in the presence of a GABAB receptor antagonist. Bath application of CGP55845(100 µM) did not reverse the effects of bright surround illumination (Fig. 5A and B, and Table 1), indicating that GABAB receptors did not play a role in mediating this lateral inhibition. Even when the GABAB blocker was applied in combination with bicuculline, picrotoxin and I4AA (Fig. 5C and D, and Table 1), bright surround illumination was still effective, indicating that neither ionotropic nor metabotropic GABA receptors mediated bright surround inhibition. Instead of reversing the suppression attributed to bright surround inhibition, GABAB receptor blockers enhanced the surround suppressive effects (Table 1). This effect, while interesting, was beyond the scope of the present study and not investigated further.
|
Although GABA receptors are present in the OPL, steady surround inhibition in the outer retina is not mediated by GABA receptors on cones or bipolar cells (Hare & Owen, 1996; Verweij et al. 1996, 2003; Cook & McReynolds, 1998a; McMahon et al. 2004), but has been attributed to the modulation of calcium channels on cone photoreceptor terminals by horizontal cells (Verweij et al. 1996; Kamermans et al. 2001). Horizontal cell feedback to cones was blocked by carbenoxolone or cobalt and was not reduced by GABA receptor blockers (Verweij et al. 1996, 2003; Kamermans et al. 2001; Fahrenfort et al. 2004).
We tested whether signalling from horizontal cells to cone photoreceptors mediated the bright surround inhibition observed in ganglion cells by recording responses in the absence or presence of carbenoxolone. As carbenoxolone also caused a time-dependent suppression of the cone output (Kamermans et al. 2001; Vessey et al. 2004), we obtained complete light intensity–response functions rapidly by ramping the light intensity in a staircase fashion with increments of 0.5 or 1.0 log intensity steps with durations of 100–150 ms (Fig. 6A). The blocking effects of carbenoxolone on horizontal cell feedback are best observed when central cones were maximally activated by light (Kamermans et al. 2002; Verweij et al. 2003). We used a modified version of this protocol where we partially activated the receptive-field centre with dim illumination and then measured the effects of further increases of centre illumination, in either the absence or presence of bright surround illumination (Fig. 6A).
In response to a ramp of light intensity, we recorded a slowly rising ganglion cell L-EPSC whose amplitude increased as a function of centre illumination intensity (Fig. 6A). GABAA, GABAC and glycine receptor antagonists were included in the bath to block lateral inhibition at the IPL. Intensity–response relationships were constructed by re-plotting the L-EPSCs as a function of cumulative photons µm–2 (Fig. 6B, see Methods). In the presence of bright surround illumination, the centre intensity–response relationship was shifted to brighter intensities and the maximum amplitude was reduced (Fig. 6C). These ramp-generated intensity–response relationships were similar to the intensity–response relationship generated by increasing centre-illumination steps illustrated above (Fig. 4D, and Tables 1 and 2).
|
Although bright surround signalling was blocked by carbenoxolone, the effects of dim surround illumination on the responses to centre illumination were not affected by carbenoxolone. To observe the effects of dim surround illumination mediated by inner retinal pathways, GABA receptor antagonists were not present in the bath. Because we could not obtain reliable responses to ramp-intensity stimuli, in the absence of GABA receptor blockers, we used steps of sub-saturating centre illumination to assess the effects of surround inhibition. As noted above, L-EPSC run-down precluded us from obtaining intensity–response relationships using steps of illumination. Dim surround illumination reduced the response to centre illumination steps in either the absence (Fig. 7A) or presence (Fig. 7B) of carbenoxolone. As expected, carbenoxolone also reduced the effects of centre illumination, but was ineffective in reducing the effects of dim surround illumination (Fig. 7B and C). The results with carbenoxolone suggest that the pathways that mediate dim and bright surround inhibition are distinct.
|
While our findings with carbenoxolone are consistent with its blockade of horizontal cell signalling to cones, carbenoxolone has also been reported to reduce calcium currents in cones (Vessey et al. 2004). The latter effect could explain the time-dependent decrease in cone output that we observed, but it is unclear whether this effect contributed to the blockade of surround signals that we observed (Kamermans & Fahrenfort, 2004). Therefore we used another method to reduce horizontal cell signalling to cones. Low concentrations of cobalt, which do not eliminate synaptic transmission, have been shown to block horizontal cell feedback to cones in turtle (Thoreson & Burkhardt, 1990; Vigh & Witkovsky, 1999) and macaque retina (Verweij et al. 2003). In goldfish retina, this cobalt blockade has been attributed to block of hemichannels on horizontal cells (Fahrenfort et al. 2004). As cobalt blocks surround antagonism in ganglion cells at photopic light levels (Vigh & Witkovsky, 1999; McMahon et al. 2004), we tested whether low concentrations of cobalt blocked bright surround inhibition in our preparation.
Because cobalt can also reduce synaptic transmission, we first determined the optimal concentration of cobalt needed to block surround inhibition with minimal effects on transmitter release. To do this, we measured the effects of cobalt on synaptic transmission by recording L-EPSCs and comparing these effects with the ability of cobalt to block bright surround inhibition. As with the carbenoxolone experiments, the centre pathway was partially activated with red illumination (–1 log attenuation). The cobalt concentration–response curve for the centre light response is shown in Fig. 8. The fraction of inhibition of the centre response by bright surround illumination is also plotted in Fig. 8. Complete surround inhibition of the centre responses is indicated by 1.0, and no inhibition is indicated by 0.0. At a concentration that preferentially reduced cone surround inhibition in goldfish and macaque retina, cobalt (50 µM) abolished the ganglion cell EPSCs evoked by centre illumination (n = 4). When 0.3 µM cobalt was used, some surround inhibition still remained. Bright surround inhibition was dramatically reduced by 3–10 µM cobalt, but the L-EPSCs were only decreased by half. Thus we determined whether 3–10 µM cobalt blocked the effect of bright surround illumination on light-sensitivity curves.
|
|
We also tested whether cobalt (10 µM) affected dim surround inhibition, which we attributed to a GABA-dependent pathway in the IPL (Fig. 3). L-EPSCs were recorded in the absence or presence of dim surround illumination from four ganglion cells in control solution or in cobalt-containing solution. As shown above, dim surround illumination reduced the L-EPSCs (Fig. 10A), and shifted the light intensity–response curve to the right and compressed it slightly (Fig. 10B, and Table 2). Cobalt reduced the L-EPSCs (Fig. 10C) and shifted and compressed the spot intensity–response curves from their control values (Fig. 10D ‘centre only’), as noted above. In contrast to its effects on bright surround, cobalt did not block the effects of dim surround illumination. In the presence of cobalt, dim surround illumination still reduced the L-EPSC amplitude (Fig. 10C right) and shifted the spot intensity–response curve (Fig. 10D, and Table 2). After washing out cobalt, we tested the effects of GABAA and GABAC receptor blockers on dim surround illumination in the same set of cells. As expected, the GABA receptor blockers eliminated the effect of dim surround illumination (Fig. 10E and F, and Table 2).
|
Bright surround inhibition of ganglion cell L-EPSCs was not mediated by a presynaptic, GABAergic pathway
The experiments with carbenoxolone and cobalt suggested that the bright surround inhibition is mediated by a different pathway from the dim surround signalling pathway. We examined whether the bright surround inhibition also uses a GABA-dependent pathway by determining whether GABA transporter (GAT) inhibition affected bright surround signalling (Ichinose & Lukasiewicz, 2002). The effect of GABA signalling on the surround inhibition is different at the inner and the outer retina. In the IPL, we showed that presynaptic surround inhibition was mediated by GABA signalling. When GABA signalling is enhanced in the IPL by GAT inhibitors, IPL surround inhibition is enhanced and this enhancement can be reversed with GABA receptor blockers (Ichinose & Lukasiewicz, 2002). In the OPL, GABA does not mediate surround signalling (Verweij et al. 1996, 2003), but it modulates surround inhibition (Kamermans et al. 2002). The horizontal cell feedback to cones is reduced by the activation of ionotropic GABA receptors and enhanced after GABA signalling is reduced by blocking either GABA release with GAT inhibitors or GABA receptors with antagonists (Kamermans et al. 2002).
To determine whether bright surround inhibition was also mediated by a GABA-dependent, presynaptic pathway, we examined the effects of a GAT inhibitor, NO-711, and GABAA and GABAC receptor blockers on the bright surround inhibition of ganglion cell L-EPSCs. Figure 11A and B shows the L-EPSCs and light-sensitivity curves in the absence and presence of bright surround illumination. NO-711 reduced the L-EPSCs and shifted the centre intensity–response relation to the right (Fig. 11C and D, and Table 2) in agreement with our previous results (Ichinose & Lukasiewicz, 2002). The shift of the ‘centre only’ intensity–response curve by NO-711 can be attributed to the enhancement of the GABAergic inhibition in the IPL.
|
|
In addition to the presynaptic surround signalling pathways described above, there is also a GABAergic amacrine cell pathway that directly inhibits ganglion cells (Cook & McReynolds, 1998a). We assessed the contribution of this pathway by recording light-evoked inhibitory postsynaptic currents (L-IPSCs) in ganglion cells that were voltage-clamped to 0 mV, the reversal potential for L-EPSCs. Strychnine was included in the bath to isolate the GABAergic L-IPSCs. The standard dim surround stimulus used throughout this study evoked small amplitude, asynchronous L-IPSCs (Fig. 12B). Dimmer surround stimuli, which did not inhibit L-EPSCs in ganglion cells, evoked smaller L-IPSCs; however, they were not significantly different from those evoked by the standard dim surround (Fig. 12A and D). By contrast, standard bright surround stimuli evoked large L-IPSCs composed of synchronous and asynchronous events (Fig. 12C and D). Comparisons of the L-IPSC charge transfer evoked by different surround illumination intensities show that bright surround illumination was more effective than dim surround illumination in activating the feedforward IPL inhibitory surround pathway. These findings are in contrast to our results for presynaptic, IPL inhibition that showed that dim surround illumination was most effective in suppressing ganglion cell L-EPSCs. In agreement with previous studies (Dong & Werblin, 1998; Lukasiewicz & Shields, 1998; Flores-Herr et al. 2001), the L-IPSCs evoked by bright surround illumination were blocked by bicuculline (Fig. 12E and F), indicating that they were mediated by GABAA receptors. Our results illustrate that surround signalling in the IPL was mediated by two amacrine cell pathways. Dim surround signalling utilized a presynaptic pathway that reduced bipolar cell transmitter release and bright surround signalling utilized a pathway that directly inhibited ganglion cells.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
|
Blockade of GABAC receptors located mainly on bipolar cell axon terminals (Lukasiewicz et al. 1994; Enz et al. 1996; Koulen et al. 1997) eliminated the dim surround component of ganglion cell responses (Fig. 13). Although GABA receptors are present on cone photoreceptors, as well as on bipolar cell dendrites, in mammals and in lower vertebrates (Vardi et al. 1992; Picaud et al. 1998; Wu & Maple, 1998; Shields et al. 2000), they do not mediate lateral inhibition in the outer retina in either lower vertebrates or primates as surround inhibition was still observed in bipolar cells and cone photoreceptors in the presence of GABA receptor antagonists (Hare & Owen, 1996; Verweij et al. 1996, 2003; Cook & McReynolds, 1998a). In agreement with our findings, GABA blockers eliminated mesopic surround inhibition to rabbit ganglion cells (Flores-Herr et al. 2001) and to rat bipolar cell terminals (Euler & Masland, 2000). These findings suggest that GABA blockers acted primarily at bipolar cell axon terminals to block presynaptic surround inhibition of ganglion cells.
Both wide- and narrow-field GABAergic amacrine cells are present in salamander retina (Yang et al. 1991; Deng et al. 2001). The contributions of these two classes of amacrine cells to the surround receptive-field organization are unclear. Our results show that inhibitory signals mediated by both narrow- and wide-field amacrine cell inputs had the same properties. Both spot and annulus stimuli activated GABAC receptor-mediated pathways (Figs 2 and 3) and blockade of GABAA receptors enhanced GABAC signalling in both cases. These findings suggest that either both narrow- and wide-field amacrine cells contribute to the ganglion cell surround or that wide-field amacrine cells can mediate both lateral and local signalling (Cook & McReynolds, 1998a).
Glycinergic amacrine cells in salamander also have wide-field processes (Yang et al. 1991) and mediate a change-sensitive, wide-field inhibition of ganglion cells (Cook et al. 1998). Studies in salamander, rabbit and primate retina showed that strychnine, which blocks transient inhibition (Belgum et al. 1984; Cook et al. 1998), had no effect on the sustained surround component of ganglion cells (Cook et al. 1996; Cook & McReynolds, 1998a; Flores-Herr et al. 2001; McMahon et al. 2004). These studies suggest that both glycinergic amacrine and interplexiform cells (Maple & Wu, 1998) do not contribute to steady surround inhibition. In our study, the effects of steady surround inhibition cannot be attributed to glycinergic neurones because we always included strychnine in the bath.
We found that carbenoxolone and cobalt, agents reported to block feedback to cones (Kamermans, 2001; Verweij et al. 2003; Fahrenfort et al. 2004), had little effect on dim surround inhibition, suggesting that dim surround illumination does not strongly activate lateral inhibitory pathways in the outer retina. Consistent with our findings, others showed that horizontal cell feedback to cones in goldfish and primate retinas is most effective when activated by bright illumination (Kraaij et al. 2000; McMahon et al. 2004). Although we cannot rule out a role of the outer retina in mediating a component of dim surround inhibition, under our experimental conditions it did not play a major role.
Dim surround effects were not mediated by rod photoreceptor pathways
Rod and cone signalling pathways have different light sensitivities; however, our observations cannot be attributed to the differential activation of these pathways by dim or bright surround illumination. We used red light in all of our experiments, which should only minimally activate salamander rods at dim light intensities (Fain & Dowling, 1973). We also performed our dissections using dim red light that eliminated the rod-mediated responses that were observed if dissections were performed using IR illumination (see Results, Hensley et al. 1993). All of the ON-OFF ganglion cells recorded had dendritic proccesses that stratified in the mid-IPL and received inputs from cone-dominant bipolar cells (Wu et al. 2000). Thus, the dim and bright surround signalling that we recorded were most probably mediated primarily by cone signalling pathways.
Presynaptic inhibition evoked by bright surround illumination was primarily in the OPL
Bright illumination of the receptive-field surround reduced the sensitivity of the ganglion cell to centre illumination further and also reduced the maximum L-EPSC amplitude. Thibos & Werblin (1978) reported similar effects on ganglion cell spike output. We found that the effects of bright surround illumination on the L-EPSCs, unlike dim illumination, were not reversed by GABA receptor antagonists, suggesting that they were not GABA-mediated. The effects of bright surround illumination were not mediated by glycine because strychnine was always present in the bath solution. Our inability to reduce the effects of bright surround illumination with GABA and glycine receptor antagonists suggests that the lateral inhibition was either mediated in the outer retina or by amacrine cells that release an inhibitory transmitter other than GABA or glycine.
Bright surround inhibition was reduced by carbenoxolone and cobalt, suggesting that it was largely mediated in the outer retina (Fig. 13). In support of this notion, bright surrounds were found to most effectively activate lateral feedback to cones in goldfish (Kraaij et al. 2000) and in primates (McMahon et al. 2004). Carbenoxolone and cobalt, which have been shown to block feedback inhibition to cone photoreceptors (Thoreson & Burkhardt, 1990; Kamermans et al. 2001; Verweij et al. 2003), blocked the effects of bright surround inhibition on ganglion cells. Consistent with our findings, McMahon et al. (2004) showed that photopic surround inhibition to primate parasol ganglion cells was reduced by carbenoxolone and by cobalt. Vessey et al. (2004) showed that carbenoxolone also reduces calcium currents in cones. This may have contributed to the time-dependent decrease of the centre response that we observed with carbenoxolone, but these actions cannot account for the effects of carbenoxolone on feedback to cones (Kamermans & Fahrenfort, 2004). In our experiments, carbenoxolone still reduced the effects of bright surrounds, consistent with the blockade of feedback. The simplest interpretation of our results is that these blockers reduced signalling from horizontal cells to cones, diminishing the effects of bright surround illumination. Carbenoxolone may have also reduced electrical coupling between horizontal cells, decreasing lateral signalling in the OPL. However, calculations by Verweij et al. (2003) indicate that reducing horizontal cell coupling does not affect the strength of horizontal cell feedback signalling.
Carbenoxolone could block gap junction communication between amacrine cells, reducing the effectiveness of IPL surround signalling. However, the main effects of carbenoxolone were not on amacrine cells because it was ineffective in blocking lateral inhibition elicited by dim surround illumination, which was attributed to GABAC receptor-mediated signalling between amacrine and bipolar cells. Also cobalt, which should not block gap junctional signalling, mimicked the effects of carbenoxolone, suggesting that the main effects of carbenoxolone were not in the IPL. Low concentrations of cobalt have been shown to reduce the GABA-evoked current in turtle cones and fish horizontal cells (Kaneko & Tachibana, 1986; Kaneda et al. 1997). However in our studies, cobalt (10 µM) did not affect GABA-dependent dim surround signalling pathways (Fig. 9), suggesting that cobalt, at this concentration, did not affect GABAC receptors in our preparation. The simplest interpretation of our cobalt and carbenoxolone data is that they reduced bright surround inhibition by attenuating horizontal cell feedback to cones.
Is bright surround inhibition also mediated by presynaptic IPL pathways?
Our findings may not rule out a role of the inner retina in bright surround inhibition because GABA may act as a slow modulator of horizontal cell feedback to cones (Kamermans et al. 2002). In this model, the activation of GABA receptors on cone terminals reduces feedback from horizontal cells. GABA receptor blockers should increase this feedback signal, enhancing lateral signalling in the OPL. In agreement with McMahon et al. (2004), we found that GABA receptor blockers did not reduce bright, presynaptic surround signalling, suggesting that it was not mediated by the IPL. It is unclear why bright surround signals are not mediated by the presynaptic IPL pathway. One possibility is that the presynaptic IPL pathway saturates at dim to moderate light intensities, precluding bright intensity signalling. Alternatively, an IPL component of bright surround inhibition might be masked as GABA receptor antagonists block an IPL component while enhancing an OPL component of the steady surround. However, our GAT inhibitor experiments suggested that bright surround inhibition was not mediated by a GABA-dependent, inner retinal feedback pathway (Fig. 11).
Our studies of the effects of IPL-mediated surround inhibition on ganglion cell EPSCs focused on the component mediated by signalling to bipolar cell axon terminals. An additional component of inner retinal surround inhibition is attributed to amacrine cell signalling to ganglion cells (Cook & McReynolds, 1998a; Roska et al. 2000; Flores-Herr et al. 2001). Our results suggest that surround inhibition to bipolar cell axon terminals and ganglion cell dendrites are different. Dim surround intensities, which activated inhibitory inputs to bipolar cell terminals, were less effective in activating direct inhibition of ganglion cells. By contrast, brighter surrounds activated GABAergic inputs to ganglion cells. These findings suggest that different amacrine cell circuits that utilize different complements of GABA receptors were involved in presynaptic versus direct surround inhibition in the
) reversed the effects of bicuculline and shifted the light–response curve back to control levels (L50, P = 0.07; max, P = 0.35; n = 10).
