How do tonic glutamatergic synapses evade receptor desensitization?
- Corresponding author S. M. Wu: Cullen Eye Institute, Baylor College of Medicine, One Baylor Plaza, NC-205, Houston, TX 77030, USA. Email: swu{at}bcm.tmc.edu
Abstract
Photoreceptor output synapses are the best known tonic chemical synapses in the nervous system, in which glutamate is continuously released in darkness, activating AMPA/kainate receptors in postsynaptic neurons. It has been shown that glutamate receptors in certain types of second-order retinal cells are largely desensitized in darkness, leading to small postsynaptic currents and reduced response dynamic ranges. Here we show that the tonic glutamatergic synapses between photoreceptors and rod-dominated hyperpolarizing bipolar cells (HBCRs) in the salamander retina evade postsynaptic receptor desensitization by using (1) multiple invaginating ribbon junctions as releasing sites for low-frequency, synchronized multiquantal release at each site; and (2) the GluR4 AMPA receptors as the postsynaptic receptors. The multiquantal events exhibit faster decay time than the GluR4 receptor desensitization time constant and therefore self-desensitization is minimized, and the average inter-event duration in darkness is much longer than the GluR4 desensitization recovery time and thus mutual desensitization is avoided. Consequently, the HBCRs are not desensitized in darkness, allowing light signals to be encoded by the full operating range of the glutamate-gated postsynaptic currents. Our study illustrates for the first time how a tonic glutamatergic synapse avoids postsynaptic receptor desensitization, a strategy that may be shared by many other synapses in the nervous system that need extended operation capacity.
Chemical synaptic transmission in the nervous system follows two major modalities. In a phasic synapse, an action potential transiently depolarizes the presynaptic membrane, increases calcium entry that initiates a cascade of molecular events leading to exocytosis of synaptic vesicles and a brief burst of neurotransmitter release to the postsynaptic receptors, resulting in a transient postsynaptic response (Sudhof, 2004). In a tonic synapse, on the other hand, neurotransmitter release, through a similar exocytotic process, is controlled by presynaptic neurons that do not generate action potentials, but instead exhibit steady and graded voltage signals (Prescott & Zenisek, 2005). These synapses continuously release neurotransmitters for the duration of the presynaptic depolarization, resulting in sustained postsynaptic signals (Copenhagen & Jahr, 1989; Heidelberger, 2007).
Recent evidence has revealed that many types of postsynaptic receptors, such as the glutamate and GABA receptors, desensitize within milliseconds following agonist binding (Trussell & Fischbach, 1989; Jones & Westbrook, 1995). Receptor desensitization shapes postsynaptic signals in phasic synapses mainly by shortening the decay time, since the agonist–receptor binding in these synapses is brief. In tonic synapses, however, fidelity of synaptic transmission is much more severely limited, because a sustained transmitter release elicits a much attenuated postsynaptic response (except for the initial several milliseconds before desensitization occurs) as postsynaptic receptors are desensitized by the steady neurotransmitter release. Therefore there are two options for a tonic synapse in dealing with receptor desensitization: accept a narrower operating range of the postsynaptic current mediated by desensitized receptors; or find ways to evade receptor desensitization so that the synapse can use the full receptor operating range.
The best known tonic synapses in the nervous system are the glutamatergic synapses between photoreceptors and second-order retinal neurons, the bipolar cells and horizontal cells (HCs) (Wu, 1994). Photoreceptors are tonically depolarized and release glutamate continuously in darkness (Copenhagen & Jahr, 1989), and the continuous glutamate flow desensitizes postsynaptic AMPA/kainate glutamate receptors in second-order retinal cells (Yang et al. 1998). It has been shown that the kainate receptors in ground squirrel HBCs are largely desensitized when the photoreceptors are held at the dark membrane potential, and thus photoreceptor hyperpolarizing signals, such as light responses, only induce a very small postsynaptic current (< 5% of the un-desensitized postsynaptic current) (Devries & Schwartz, 1999). The AMPA receptors in salamander horizontal cells (HCs) are also desensitized in darkness, as application of cyclothiazide, an AMPA receptor desensitization blocker, depolarized the HCs in darkness and enhanced the light responses (Yang et al. 1998). In this paper, we show that the photoreceptor–rod-dominated hyperpolarizing bipolar cell (HBCR) synapse in the salamander retina evades AMPA receptor desensitization so that light-evoked graded responses in photoreceptors may elicit sustained postsynaptic current responses, spanning the full dynamic range.
Methods
Electrophysiology
Larval tiger salamanders (Ambystoma tigrinum) purchased from Charles D. Sullivan, Co. (Nashville, TN, USA) and Kons Scientific Co. Inc. (Germantown, WI, USA) were used in this study. All animals were handled in accordance with the policies on treatment of laboratory animals of Baylor College of Medicine and the National Institutes of Health. Before each experiment, salamanders were anaesthetized in MS222 (2 g l−1) until the animal gave no visible response to touch or water vibration. The animals were then quickly decapitated and the eyes were enucleated. The procedures of dissection, retinal slicing and recording were described in previous publications (Werblin, 1978; Wu, 1987). Dissection and recording were done under infrared illumination with a dual Nitemare infrared scope (BE Meyers, Redmond, WA, USA). Oxygenated Ringer solution was introduced continuously to the superfusion chamber, and the control Ringer solution contained 108 mm NaCl, 2.5 mm KCl, 1.2 mm MgCl2, 2 mm CaCl2 and 5 mm Hepes (pH 7.7). All chemicals were dissolved in control Ringer solution.
A photostimulator was used to deliver light spots (of diameter 600–1200 μm) to the retina via the epi-illuminator and the objective lens of the microscope. The intensity of unattenuated (log I = 0) 500 nm light was 2.05 × 107 photons μm−2 s−1. Since we delivered an un-collimated stimulus light beam through an objective lens with large numerical aperture (Zeiss 40×/0.75 water), the incident light entered the retinal slice in many directions, and thus the effect of photoreceptor self-screening was minor (Field & Rieke, 2002).
Dual or single voltage-clamp recordings were made with an Axopatch 700A amplifier connected to a DigiData 1200 interface and pCLAMP 10 software. Patch electrodes of 5 MΩ tip resistance (when filled with an internal solution containing 118 mm caesium methanesulphonate, 12 mm CsCl, 5 mm EGTA, 0.5 mm CaCl2, 4 mm ATP, 0.3 mm GTP, 10 mm Tris, 0.8 mm Lucifer yellow or sulphorhodamine, and when adjusted to pH 7.2 with CsOH) were made with Narishige or Sutter patch electrode pullers. The chloride equilibrium potential, ECl, with this internal solution was approximately −60 mV. The equilibrium potential of cation current (EC) was determined by the reversal potential of glutamate-induced current in morphologically identified bipolar cells in Ringer solution containing 2 mm Co2+ (Wu & Maple, 1998). Estimates of the liquid junction potential at the tip of the patch electrode prior to seal formation varied from −9.2 to −9.6 mV. For simplicity, we corrected all holding potentials by 10 mV. Spontaneous and light-evoked current responses were analysed by in-house software and SigmaPlot (Jandel Scientific). The average unitary sEPSCs (I1) in HBCs were analysed by the method described by Katz and Miledi (Katz & Miledi, 1972). In this analysis, we approximated individual sEPSCs as ‘shot’ effects because the rising phase of sEPSCs is very fast (≤ 1.2 ms), and at least some events appeared to have single-exponential decay.
Three-dimensional cell morphology was visualized in living retinal slices (250–300 μm in thickness) through the use of Lucifer yellow fluorescence with a confocal microscope (Zeiss 510). Images were acquired by using a ×40 water immersion objective (n.a. = 0.75), the 458 nm excitation line of an argon laser, and a long-pass 505 nm emission filter. Consecutive optical sections were superimposed to form a single image using the Zeiss LSM-PC software, and these compressed image stacks were further processed in Adobe Photoshop 6.0 to improve the contrast. Since signal intensity values were typically enhanced during processing to improve the visibility of smaller processes, the cell bodies and larger processes of some cells appear saturated due to their larger volume of fluorophore. Although the background images of the retinal slices were acquired simultaneously with the fluorescent cells, they were imaged using transmitted light. The level at which dendritic processes stratified in the inner plexiform layer (IPL) was characterized by the distance from the processes to the distal margin of the IPL. We selected cells in the bipolar cell layers with somas situated beneath the surface of the slice and they usually had relatively intact processes (assessed by rotation of the stacked images).
Immunocytochemistry
Retinas were fixed in 4% paraformaldehye in phosphate-buffered saline (PBS; pH 7.4) for 30–60 min at room temperature, and then extensively rinsed with PBS. Whole-mount retinal tissue was blocked with 3% donkey serum in PBS with 0.5% Triton X-100 and 0.1% sodium azide from 2 h to overnight to reduce non-specific labelling. The tissue was then incubated in primary antibody in the presence of 1% donkey serum–PBS with 0.5% Triton X-100–0.1% sodium azide for 3–10 days at 4°C. Controls lacking primary antibodies were blank. After extensive washing with PBS containing 0.5% Triton X-100–0.1% sodium azide the tissue was incubated overnight with immunofluorescent secondary antibody. After further rinsing, the tissue was mounted with Vectrashield. The specimens were then observed with a confocal laser-scanning microscope (Zeiss LSM510). Stacked images were acquired through the whole retina, using Zeiss LSM-PC software, and cells were counted manually.
Antibodies against GluR4 and GluR2 were obtained from Chemicon International (Temecula, CA, USA) and used at a dilution of 1: 1000. Secondary antibodies were donkey conjugated CY3 (Jackson ImmunoResearch, West Grove, PA, USA) and Alexa 488 (Molecular Probes, Eugene, OR, USA), used at a dilution of 1: 100. TOPRO3 (0.01 μl ml−1), a nuclear dye used to label cell bodies in the outer plexiform layer (OPL), was obtained from Molecular Probes.
Results
Synaptic transmission from photoreceptors to HBCRs is mediated by AMPA receptors
Previous studies have revealed that rod-dominated hyperpolarizing bipolar cells (HBCRs) exhibit spontaneous large excitatory postsynaptic current (LETC) events in darkness and bear axon terminals ramified in the distal 25% of the IPL (Wu et al. 2000; Pang et al. 2004). In order to study photoreceptor inputs to HBCRs, we made dual whole-cell voltage-clamp recordings from rod–HBCR pairs in dark-adapted salamander retinal slices. Figure 1 shows a rod–HBCR pair filled with sulphorhodamine (rod, red) and Lucifer yellow (HBCR, green) (Fig. 1A), the simultaneous current responses to a light step, with the rod voltage held at −40 mV and the HBCR at various holding potentials (Fig. 1B); and the HBCR current responses at various holding potentials to a voltage step (from −40 mV to 0 mV) in the rod (Fig. 1C). In darkness, the zero-current potential of the HBCR was −40 mV, and the dark current was accompanied by many LETC events (thick noisy current traces). Light greatly reduces the LETC event frequency, resulting in a sustained reduction of inward current. The LETC events reverse near 0 mV (close to EC (Maple et al. 1994)), whereas the light-evoked sustained outward current response did not reverse (also see the current–voltage relations in Fig. 1D). The depolarizing voltage step in the rod elicited a large transient inward current in the HBCR that reversed near −10 mV. These results demonstrate that the HBCR LETC events have the same reversal potential (near the reversal potential of the glutamate-elicited current, EC) as the rod-elicited excitatory postsynaptic current, suggesting that these two postsynaptic currents are both mediated by the photoreceptor–HBCR synapses. The light-evoked sustained outward current has no apparent reversal potential because light evokes both excitatory inputs from photoreceptors (associated with a conductance decrease) and inhibitory inputs from amacrine cells (associated with a conductance increase) (Pang et al. 2004).
Figure 2 shows effects of 100 μm cyclothiazide (CTZ) on average rod-elicited responses in a HBCR held at −60 mV (near ECl). Before CTZ application, the average rod-elicited peak current was 252 ± 45 pA with an average decay time constant of 40.6 ± 3.0 ms. CTZ slightly enhanced the response with an average peak of 297 ± 55 pA (not statistically significant from the control value, P > 0.05), and significantly prolonged the response with a decay time constant of 47.1 ± 5.2 ms (P < 0.01). We also carried out the same experiments in the presence of the desensitization blocker for the flop variants of the AMPA receptors PEPA (Sekiguchi et al. 1997, 1998) and the kainate receptor desensitization blocker concanavalin A (Partin et al. 1993), and found that they exerted no effects (results not shown). This suggests that the rod→HBCR synaptic signals are mediated by CTZ-sensitive AMPA receptors (flip variants), consistent with a previous study showing that CTZ enhanced postsynaptic current responses elicited by puff application of glutamate in salamander HBCRs (Maple et al. 1999; Cadetti et al. 2005). Results in Fig. 2 suggest that a depolarizing voltage step in a rod induces a transient burst of glutamate release that activates AMPA receptors with a decay time constant of about 47.1 ms. The glutamate release is transient because the voltage step in a rod results in depolarizations in adjacent rods via rod–rod coupling and Ih in the unclamped adjacent rods shapes the depolarizing currents into transient voltages (Attwell & Wilson, 1980). Consequently, these rods transiently release glutamate on postsynaptic cells (Attwell et al. 1983). Under physiological conditions (in the absence of CTZ), AMPA receptor desensitization reduced the transient postsynaptic response (average peak amplitude 297 to 252 pA) and shortened the decay time (average decay time constant 47.1 to 40.6 ms) of the rod-elicited responses in the HBCRs.
Paired-pulse study suggests receptors are likely to be GluR2 and/or GluR4
Figure 3A (black traces) shows the postsynaptic responses in a HBCR held at −60 mV elicited by pairs of depolarizing voltage steps with various inter-pulse durations in a rod. The first voltage step in the rod evoked a transient inward response of about 250 pA, similar to results in Fig. 2, and a second identical voltage step 150 ms later evoked a much depressed response. The second response became larger as the inter-pulse duration increased, and it reached the same peak value as the first response when the duration was about 3 s. The recovery time course of such paired-pulse depression obtained from four rod–HBCR pairs is given in Fig. 3B (black symbols), and the data can be fitted by a single exponential function with a time constant of 741 ms. We then repeated this experiment in the presence of 100 μm CTZ (red traces and symbols in Fig. 3A and B), and found that the second responses were not much affected, suggesting that the rod→HBCR pair-pulse depression is not mediated by AMPA receptor desensitization, but other mechanisms such as vesicle depletion (Rabl et al. 2005, 2006; Singer & Diamond, 2006). The fact that CTZ did not enhance the second response with inter-pulse duration of 150 ms suggests that the AMPA receptor desensitization induced by the first pulse fully recovered during a time period shorter than 150 ms. Based on the desensitization recovery time courses of various types of AMPA receptors (GluR1–4) revealed by HEK cell expression studies (Grosskreutz et al. 2003), the AMPA receptors in the photoreceptor–HBCR synapse is likely to be GluR2 and/or GluR4 (other AMPA or KA receptors have desensitization full recovery times substantially longer than 150 ms).
Immunocytochemistry suggests that the receptors are GluR4, not GluR2
We then used immunocytochemistry to study localizations of GluR2 and GluR4 glutamate receptors in the salamander retina. Figure 4A shows heavy immunostaining with antibodies against GluR4 subunits in the OPL of a retinal section in which a HBCR was filled with Lucifer yellow. It is evident that GluR4-positive plaques (red) are co-localized on many HBCR dendrites (green), including dendritic processes in the rod invaginations. On the other hand, Fig. 4B illustrates that antibodies against GluR2 subunits did not label anything in the salamander OPL. These results, in conjunction with the paired-pulse–CTZ data in Fig. 3, suggest that the photoreceptor–HBCR synapse is mediated by GluR4 AMPA receptors.
Large excitatory transient current (LETC) events in HBCRs and their frequency in darkness
As shown in Fig. 1, the HBCR LETC events occur at high frequency in darkness, and the frequency decreases in the presence of light. Figure 5A shows a HBCR current trace recorded under voltage-clamp conditions at −40 mV in faster time scale. It is evident that the HBCR dark current is accompanied by many LETC events (arrows) and the sustained inward current in darkness (as compared with the current level in bright light, dashed line) appears primarily composed of the superposition of overlapping LETC events. The reduction in frequency of these events by light makes them distinguishable from one another (arrowheads). This allows amplitude and waveform characterization of individual events. Figure 5B is the amplitude histogram of 237 LETC events recorded from 15 HBCRs in the presence of light. These events were identified according to three criteria: (1) starts from baseline current level in light (e.g. the dashed line in Fig. 5A) with a monotonic rise phase and a time-to-peak shorter than 1.5 ms; (2) decay phase can be roughly fitted by a single exponential function; and (3) peak amplitude larger than 3 times the baseline current fluctuations (≥ 6pA). Figure 5B shows the HBCR LETC event peak amplitude distribution between 6 and 70 pA with an average of 19.72 ± 10.93pA. The average time-to-peak is 1.33 ± 0.42 ms and the decay time constant is 1.54 ± 0.53 ms. A short sample of LETC events in control Ringer solution is shown in Fig. 6Aa. By using these average event parameters, an average LETC event is simulated in Fig. 6Ba. Integrating this average event with respect to time renders the average event charge (q) as 76.9 fC.
By using the same procedures and criteria, we analysed 156 LETC events from four HBCRs in the presence of CTZ. A short sample of LETC events in CTZ is shown in Fig. 6Ab. The average peak amplitude is 20.89 ± 18.11 pA, average time-to-peak 1.35 ± 0.31 ms, average decay time constant 1.63 ± 0.70 ms and average event charge is 84.48 fC (Fig. 6Bb). The close similarity between the average LETC events in the absence and presence of CTZ suggests that AMPA receptor desensitization exerts little action on the LETC event amplitude and decay time. A likely explanation is that the decay time constant of the glutamate in the synaptic cleft during a LETC event (as indicated by the average LETC decay constant in CTZ (1.63 ± 0.7 ms)) is shorter than the desensitization time constant of the glutamate receptors. Our immunocytochemical and paired-pulse depression results suggesting that the photoreceptor–HBCR synapse uses GluR4 receptors (with a desensitization time constant τD = 3.1 ms (Grosskreutz et al. 2003)) support this explanation.
In order to determine whether the HBCR LETC events are uniquantal (mediated by single synaptic vesicles), we analysed 137 LETC events recorded from three HBCRs in low calcium (0.5 mm instead of 2.5 mm) Ringer solution. Figure 6Ac shows a sample of LETC events in low calcium Ringer. The LETC frequency in low calcium was greatly reduced and the average peak amplitude was also reduced to 8.2 ± 2.3pA. The average time-to-peak of LETCs in low calcium was 1.23 ± 0.32 ms and the decay time constant was 1.6 ± 0.3 ms. (Fig. 6Bc). This suggests that the LETC events in HBCRs are likely to be multiquantal because the amplitude of multiquantal, but not uniquantal, events would be lowered by reducing the calcium-dependent probability of release (PR) (Singer et al. 2004; Suryanarayanan & Slaughter, 2006). Possible mechanisms underlying such multiquantal release will be discussed later.
We next estimated the average frequency of LETC events (number of LETC events per second) in HBCRs in darkness, by dividing the total charge carried by the HBCR dark current per second (Q) by the average LETC event charge (q). Q was obtained by integrating the area between the dark current trace and the dashed horizontal line (e.g. in Fig. 5A) for 1 s. From nine HBCRs, we found that on average Q = 56 653 ± 29 852 fC s−1. The ratio Q/q gives the number of LETC events released on a HBCR in darkness per second, which is (56 653 fC s−1)/(76.9 fC) = 737 s−1.
In Fig. 7, we show the frequency of HBCR LETC events in darkness and in light (Fig. 7A), and during a presynaptic depolarizing voltage step in a rod (Fig. 7B), by deconvolving the HBCR current responses with the average LETC event (Neher & Sakaba, 2001, 2003) (Fig. 6Aa). It is evident that within the average lifetime of a single LETC event, multiple LETC events superimpose, resulting in a sustained (DC) inward dark current (about 5 events per 10 ms). Light greatly reduces the LETC frequency and makes LETC events discrete with an outward baseline current shift (light response). In the rod-evoked postsynaptic responses, the presynaptic depolarizing voltage step (−40 mV to 0 mV) elicits a transient asynchronized burst of LETC events with a peak of roughly 50 events per 10 ms, followed by a number of events lasting for about 250 ms. These results indicate that glutamate release in darkness (Vrod = −40 mV) does not saturate postsynaptic receptors in HBCRs, because rod depolarization from −40 mV is still capable of eliciting additional large inward current with an increased number of LETC events. Alternatively, our data suggest rod depolarization beyond −40 mV may be capable of recruiting more glutamate receptors in HBCRs than those activated in darkness.
CTZ does not reduce HBCR dark current or light responses
We next studied the effects of CTZ on HBCR dark current and light responses. Figure 8 shows that 100 μm CTZ exerts no effects on the dark current (difference between the mean current in darkness and the current level in the presence of saturating light, see dashed line in Fig. 5A). We performed this experiment on 17 HBCRs, and none showed any detectable CTZ-induced changes in the dark current. This result was puzzling because we have shown that the photoreceptor–HBCR synapse uses the fast-desensitizing AMPA GluR4 receptors, and the LETC events in darkness are in such high frequency that they should desensitize one another. CTZ, by blocking receptor desensitization, should have substantially increased the inward dark current and the amplitude of the light responses. In order to unravel this riddle, we estimated the rate of synaptic release and the number of releasing sites between photoreceptors and HBCRs based upon photoreceptor terminal morphology.
Invaginating ribbon synapses between photoreceptors and HBCRs
It has been shown by Lasansky that 80% of the HBC dendritic contacts with photoreceptors in the salamander retina are located at the invaginating ribbon junctions (Lasansky, 1978). Since multiple vesicles are lined up by synaptic ribbons at the release site (primed for release) (Paillart et al. 2003), and our low calcium experiment (Fig. 6Cc) suggests that HBCR LETC events are multiquantal, we postulate that each LETC event in a HBCR is mediated by a synchronized release of multiple synaptic vesicles at a photoreceptor ribbon junction. Figure 9A shows two typical ribbon synapses in a thin section of a rod synaptic terminal, illustrating that two columns of synaptic vesicles are lined up at the two sides of the long ribbons. Figure 9B shows low power electron micrographs of two consecutive sections containing three rod terminals and two cone terminals, and Fig. 9C gives an example of four consecutive tangential thin sections of ribbon synapses in a cone and a rod synaptic terminal. By going through such serial thin sections containing a total of 45 rod and 38 cone terminals, we found that on average there are 4.3 ± 2.5 ribbon synapses per rod terminal and 13.4 ± 2.6 ribbon synapses per cone terminal. Additionally, the surfboard-shaped ribbons are 50–60 nm thick and 150–350 nm wide (a ribbon typically extends through 2–5 consecutive 700 Å thin sections) accommodating 4–10 vesicles per row (both sides of the ribbon) (see Fig. 9A). Therefore we believe that a LETC event in HBCRs is mediated by the synchronized release (rise time less than 1.2 ms) of less than 10 synaptic vesicles in a ribbon junction.
We next estimated the number of ribbon synapses from rods and cones to each HBCR. Previous studies have revealed that the average dendritic diameter (d) of salamander HBCRs is 74 ± 11 μm (Wu et al. 2000) and the rod and cone density is 4000 mm−2 (Zhang et al. 2004). Therefore, on average, a HBCR dendritic field (π(d/2)2 = 4300 μm−2) covers an area containing 17 rods and 17 cones, or 302 ribbon synapses (4.3 × 17 + 13.4 × 17). In the tiger salamander retina, it has been shown that about 80% of ribbon contacts goes to HBCs and 20% goes to depolarizing bipolar cells (DBCs) (Lasansky, 1978), and that the HBCR dendritic coverage factor is 3 (Zhang et al. 2004). Moreover, HBCRs receive 75% of their inputs from rods and 25% from cones (Pang et al. 2004). We therefore conclude that the dendrites of each HBCR makes about 30 [80%(75% × 4.3 × 17 + 25% × 13.4 × 17)/3] ribbon synaptic contacts with rod and cone photoreceptors. This number is very close to Lasansky's number of HBC dendritic contacts from the HRP electron microscopic study (Lasansky, 1978).
By dividing the average number of LETC events in darkness obtained in the previous section by the number of ribbon synaptic contacts on HBCR dendrites, we found that there are about 25[(737 s−1)/30] LETC events per ribbon synapse per second. In other words, the dark current in a HBCR is mediated mainly by synchronized multivesicular glutamate release events distributed among 30 ribbon synapses, each releasing 25 events per second (or on average an event every 40 ms). This average release interval (40 ms) is much longer than the desensitization recovery time course of GluR4 receptors (τrec = 3.2 ms (Grosskreutz et al. 2003), or 99% recovery in about 15 ms) and thus mutual desensitization (receptor desensitization in a later event caused by agonist binding in preceding events) rarely occurs in darkness in the photoreceptor–HBCR synapse.
Discussion
HBCRs evade glutamate receptor desensitization by using invaginating ribbon junctions and GluR4 AMPA receptors
In this article, we present evidence showing that although the photoreceptors–HBCR synapse in the salamander retina uses AMPA receptors with fast desensitization kinetics, it evades postsynaptic receptor desensitization during tonic glutamate release in darkness, so that it can encode information for a wider range of light intensities. Our study suggests that this synapse avoids glutamate receptor desensitization by two means: it uses invaginating ribbon junctions as the presynaptic release sites and GluR4 AMPA receptors as the postsynaptic receptors.
The invaginating ribbon junctions allow synchronized release of multiple vesicles (multiquantal release, as suggested by our low calcium experiment), resulting in large excitatory transient current (LETC) events in HBCRs. These ribbon junctions have three advantages over the conventional chemical synapses for evading receptor desensitization: (1) The synchronized multiquantal LETC events are larger than single vesicular events (minis) in conventional chemical synapses (Trussell & Fischbach, 1989), and thus fewer events (or longer inter-event durations) are needed to mediate the HBCR dark current. The longer event interval at the release site helps to avoid mutual desensitization; (2) The LETC events are synchronized multiquantal events which have faster rise and decay kinetics than asynchronized bursts of uniquantal events, thereby minimizing self-desensitization (receptor desensitization caused by agonist binding within a single release event) and mutual desensitization; (3) Each ribbon junction is anatomically isolated from other synaptic release sites (via highly tortuous invaginations), so that one release event in one ribbon junction does not interfere with (mutually desensitize) other events in other ribbon junctions.
Combining these advantages of the ribbon junctions with the fast desensitization and desensitization recovery kinetics of the GluR4 AMPA receptors, HBCR LETC events in darkness are able to avoid self-desensitization and mutual desensitization. Little self-desensitization occurs because the HBCR LETC decay time constant (τD = 1.54 ms) is shorter than the GluR4 desensitization time constant (τD = 3.1 ms) (Grosskreutz et al. 2003), and mutual desensitization is minimized because the average LETC event interval at each ribbon junction (40 ms) is much longer than the desensitization recovery time of the GluR4 receptors (τrec = 3.2 ms (Grosskreutz et al. 2003), 99% recovery in 15 ms).
One may argue that the reason for lack of CTZ actions on HBCR dark current and LETC events is that the postsynaptic receptors in the photoreceptor–HBCR synapse are not AMPA receptors, where the CTZ effects on glutamate-induced currents reported previously (Maple et al. 1999) merely suggest extra-synaptic AMPA receptors in these cells. Two lines of evidence in this study argue against this notion: (1) Our immunocytochemical data clearly demonstrate that GluR4 AMPA receptors are localized in physiologically identified HBCR dendrites, especially in processes residing in photoreceptor synaptic terminal invaginations (Fig. 4), where the ribbon junctions are found (Lasansky, 1973); (2) CTZ prolonged the rod depolarization-evoked HBCR postsynaptic responses, suggesting that the postsynaptic receptors are AMPA receptors. The reason why CTZ affects rod depolarization-evoked responses but not the dark current is that the rod depolarizing voltage step elicits a large burst of LETC events. Since de-convolution analysis (Fig. 7) shows that the rod voltage step elicits many more LETCs than darkness, and since the current step in a single rod only stimulates a subset of the photoreceptor–HBCR release sites, the frequency of LETC events per release site during the voltage-elicited burst (lasting about 250 ms) must be much higher than the LETC frequency in darkness. Consequently, the multiquantal release events mutually desensitize one another, and CTZ, by blocking AMPA receptor desensitization, prolongs the time course of this rod-evoked postsynaptic response.
HBCR LETC events
We propose that the HBCR LETC events are mediated primarily by multiquantal glutamate release at ribbon junctions. Our data show that they are multiquantal because low extracellular calcium reduced the average peak amplitude in addition to decreasing the frequency, whereas for uniquantal mini events calcium affects only the frequency but not the amplitude (Singer et al. 2004). We believe most LETC events occur at the ribbon junctions because it has been shown that 80% of the synaptic contacts made by salamander HBC dendrites with photoreceptors are at the ribbon junctions (the remaining 20% at the basal junctions) (Lasansky, 1978), and that multiple synaptic vesicles are primed in the readily releasable position in ribbon junctions. This does not, however, exclude the possibility that some LETC events are mediated by vesicular glutamate release at basal junctions. If a significant percentage of HBCR LETCs are indeed mediated by basal junctions in addition to the ribbon junctions we estimated in this study (on average 30 per HBCR), then the average number of LETC events per release site per second (25 s−1) would be smaller and the average event interval per release site would be longer than 40 ms, which would make mutual desensitization of LETC events even less likely to occur.
We measured the HBCR LETC event peak amplitude and kinetics in Fig. 6 from individual LETCs in light, because LETC events in darkness are of such high frequency that individual events could not be separately distinguished. We assume that the amplitude and kinetics of individual LETCs in darkness and light are very similar, and thus we used the charge carried by the average LETC in light (q) to estimate the number of LETC events per second in darkness (Q/q = 737 s−1, see Results). This approach may not give a totally accurate estimate of LETC frequency in darkness. Since we have shown that low calcium reduces LETC event amplitude, light may also reduce the LETC amplitude because it hyperpolarizes photoreceptors and reduces calcium influx. However, like the possible addition of basal junctions as LETC release sites discussed in the last section, this factor does not jeopardize our argument that LETC events in individual release sites have long enough inter-event durations to evade mutual desensitization: If the peak amplitude of LETC events in darkness is significantly larger than that in light, then the average charge carried by a LETC will be larger, and thus there would be fewer than 737 events per second per HBCR in darkness. This would lead to fewer events per release site per second, or longer event intervals, which would reduce the probability of mutual desensitization even further.
Glutamate receptors in other second-order neurons in the salamander retina
It is interesting to note that while all second-order cells have dendritic processes postsynaptic to photoreceptor ribbon junctions (Lasansky, 1973), LETC events are only observed in HBCRs. Cone-dominated HBCs (HBCCs) exhibit much smaller events, and HCs and DBCs exhibit no excitatory transient current events (Yang et al. 1998; Pang et al. 2004). The most likely explanation for this is that different second-order neurons use different postsynaptic receptors. It has been shown that DBCs in the retina use a metabotropic glutamate receptor (mGluR6) with slow activation kinetics (Nawy & Jahr, 1991), and thus the fast synchronized vesicular release events in the ribbon junctions are largely filtered. The HBCCs and HCs in the salamander retina have been shown to have CTZ-sensitive AMPA receptors (Maple et al. 1999), but these receptors may contain subunits with much slower desensitization and desensitization recovery kinetics, such as those in GluR1 and GluR3 AMPA receptors (Grosskreutz et al. 2003). It has been shown, for example, that the desensitization recovery time constant of the GluR1 receptors (τrec) is 123.2 ms, so events at ribbon junctions with average intervals of 40 ms would result in substantial mutual desensitization of these receptors. Results reported in a previous study showing that CTZ depolarizes HCs in darkness and enhances HC light responses (Yang et al. 1998) are consistent with the idea that HCs use AMPA receptors with slow desensitization and desensitization recovery kinetics.
Acknowledgements
This work was supported by grants from NIH (EY 04446), NIH Vision Core (EY 02520), the Retina Research Foundation (Houston), and the Research to Prevent Blindness, Inc.
Footnotes
-
(Received 11 January 2008; accepted after revision 15 April 2008; first published online 17 April 2008)
References
Figure 1. Dual whole-cell voltage-clamp recordings from a rod-HBCR pair A, a rod–HBCR pair filled with sulphorhodamine (rod, red) and Lucifer yellow (HBCR, green); B, simultaneous current responses of the rod and HBCR to a light step (500 nm, −3 log unit attenuation), with the rod voltage held at −40 mV and the HBCR voltage at various holding potentials; C, HBCR current responses at various holding potentials to a voltage step (from −40 mV to 0 mV) in the rod; D, current–voltage relations of the light-evoked currents (▴) and the peak values (•) of the rod-elicited responses of the HBCR.
Figure 2. Average rod-elicited responses in a HBCR held at −60 mV (near ECl) before (black) and after application of 100 μm cyclothiazide (CTZ; red) averaged over 8 traces From 4 rod–HBCR pairs, the average rod-elicited peak current was 252 ± 45 pA with an average decay time constant of 40.6 ± 3.0 ms. CTZ slightly enhanced (P > 0.05) the response with an average peak of 297 ± 55 pA, and significantly prolonged the response (P < 0.01) with an average decay time constant of 47.1 ± 5.2 ms.
Figure 3. Paired-pulse study of rod-elicited HBCR responses A, postsynaptic responses of a HBCR held at −60 mV elicited by pairs of depolarizing voltage steps in a rod with various inter-pulse durations. Black traces were recorded in control medium and the red traces were obtained in the presence of 100 μm CTZ. B, the recovery time course of such paired-pulse depression obtained from 4 rod–HBCR pairs in control medium (black symbols) and in 100 μm CTZ (red symbols). The vertical axis is the ratio of the second/first peak responses and the horizontal axis is the inter-pulse duration. Data can be fitted by single exponential functions with a time constant of 741 ms in the control medium and 735 ms in CTZ.
Figure 4. Immunocytochemical localization of GluR4 and GluR2 Aa, Lucifer yellow-filled rod–HBCR pair in a retinal section. The rod has two synaptic terminals. Ab, the same rod–HBCR pair immunostained with antibodies against GluR4 subunits. GluR4-positive plaques (red) are co-localized on many HBCR dendrites, including dendritic processes in the rod invaginations. B, antibodies against GluR2 subunits did not label anything in the salamander OPL.
Figure 5. Large excitatory transient current (LETC) events in HBCRs A, HBCR current trace recorded under voltage-clamp conditions in darkness and in light at −40 mV. The HBCR dark current is accompanied by many LETC events (arrows) and the sustained inward current in darkness (as compared with the current level in bright light, dashed line). The frequency of the LETC events was greatly reduced and the events became discrete (arrowheads). B, amplitude histogram of 237 LETC events recorded from 15 HBCRs in the presence of light.
Figure 6. LETCs in normal Ringer's in CTZ and in low calcium A, samples of LETCs in control Ringer solution (a), in control Ringer solution containing 100 μm CTZ (b), and in low calcium (0.5 mm) Ringer solution (c). B, simulation of average HBCR LETC events in light in control Ringer solution (a), in control Ringer solution containing 100 μm CTZ (b), and in low calcium (0.5 mm) Ringer solution (c).
Figure 7. Frequency of LETC events Frequency of HBCR LETC events in darkness and in light (A), and during presynaptic depolarizing voltage step in a rod (B). Number of LETC events was obtained by deconvolving the HBCR current responses with the average LETC event described in Fig. 6Aa.
Figure 8. Effects of CTZ on HBCR light responses HBCR currents recorded under voltage-clamp conditions in darkness and in light held at −40 mV in control conditions (A) and in the presence of 100 μm cyclothiazide (CTZ) (B). The dark current is defined as the difference between the mean current in darkness and the current level in the presence of saturating light (dashed line).
Figure 9. Electron micrographs of ribbon synapses in rod and cone synaptic terminals A, high power electron micrograph of two ribbon synapses in a rod synaptic terminal. Two columns of synaptic vesicles are lined up at the two sides of the long ribbons. Calibration bar: 0.1 μm B, low power electron micrographs of two consecutive thin sections showing three rod terminals (r), and two cone terminals (c) with over 20 synaptic ribbons. Calibration bar: 2 μm. C, an example of four intermediate-power consecutive thin tangential sections of ribbon synapses at the outer plexiform layer level containing a cone and a rod synaptic terminal. Calibration bar: 0.5 μm.
