High-Density Presynaptic Transporters Are Required for Glutamate Removal from the First Visual Synapse

Neuron 50, 63–74, April 6, 2006 ª2006 Elsevier Inc.

DOI 10.1016/j.neuron.2006.02.022

High-Density Presynaptic Transporters Are Required for Glutamate Removal from the First Visual Synapse Jun Hasegawa,1 Takehisa Obara,1 Kohichi Tanaka,2 and Masao Tachibana1,* 1 Department of Psychology Graduate School of Humanities and Sociology The University of Tokyo Tokyo 113-0033 Japan 2 School of Biomedical Science and Medical Research Institute Tokyo Medical and Dental University Tokyo 113-8510 Japan

Summary Reliable synaptic transmission depends not only on the release machinery and the postsynaptic response mechanism but also on removal or degradation of transmitter from the synaptic cleft. Accumulating evidence indicates that postsynaptic and glial excitatory amino acid transporters (EAATs) contribute to glutamate removal. However, the role of presynaptic EAATs is unclear. Here, we show in the mouse retina that glutamate is removed from the synaptic cleft at the rod to rod bipolar cell (RBC) synapse by presynaptic EAATs rather than by postsynaptic or glial EAATs. The RBC currents evoked by electrical stimulation of rods decayed slowly after pharmacological blockade of EAATs. Recordings of the evoked RBC currents from EAAT subtype-deficient mice and the EAATcoupled anion current reveal that functional EAATs are localized to rod terminals. Model simulations suggest that rod EAATs are densely packed near the release site and that rods are equipped with an almost self-sufficient glutamate recollecting system. Introduction Glutamate released from a presynaptic terminal should be removed quickly from the synaptic cleft to ensure reliable synaptic transmission. Diffusion and uptake by EAATs are the major factors that influence the rate of glutamate removal from the synaptic cleft. To date, five subtypes of EAATs have been identified: GLAST1 (EAAT1), GLT1 (EAAT2), EAAC1 (EAAT3), EAAT4, and EAAT5 (for review, see Danbolt [2001]). Although the turnover rates of EAATs are slow, EAATs have high affinities to glutamate, and thus they also serve as a buffer for glutamate (Wadiche et al., 1995; Diamond and Jahr, 1997; Wadiche and Kavanaugh, 1998). Each subtype of EAATs is expressed in specific neurons and/or glial cells (Rothstein et al., 1994; Lehre and Danbolt, 1998). It has been shown in many synapses that postsynaptic and glial EAATs are functional (Clark and Barbour, 1997; Otis et al., 1997; Bergles and Jahr, 1997). On the other hand, contribution of presynaptic EAATs to the synaptic

*Correspondence: [email protected]

transmission is poorly understood. Furthermore, little is known about the detailed spatial arrangement of EAATs in specific synapses. Sensory neurons, such as photoreceptors and hair cells, have synaptic ribbons with halos of synaptic vesicles, and they can release glutamate tonically (von Gersdorff, 2001). In these ribbon synapses, exocytosis of glutamate that is regulated by graded responses to stimuli occurs at high rates (von Gersdorff et al., 1996; Glowatzki and Fuchs, 2002; Thoreson et al., 2004). Thus, glutamate may accumulate in the synaptic cleft more easily in ribbon synapses than in conventional synapses where exocytosis is triggered by spikes. To reset the postsynaptic mechanism, it is expected that EAATs may play an important role for glutamate removal in ribbon synapses. In the present study, we focused on the synaptic transmission from rods to RBCs in the mouse retina. The release of glutamate from rods occurs tonically at a high rate in the dark, and it is reduced by light (Baylor and Fuortes, 1970; Dowling and Ripps, 1973). RBCs are ON-type cells and express mGluR6 (a subtype of group III metabotropic glutamate receptors) on their dendrites (Slaughter and Miller, 1981; Masu et al., 1995; Vardi et al., 2000). Activation of G protein-coupled mGluR6 closes nonselective cation channels, and thus ON-type bipolar cells are hyperpolarized in the dark and depolarized by light (Wilson et al., 1987; Nawy and Jahr, 1990; Shiells and Falk, 1990; Nawy, 1999). Prompt depolarizing responses in ON-type bipolar cells to light stimulation may be achieved by various ways. Presynaptically, glutamate release can be reduced by protons exocytosed together with glutamate from photoreceptors (DeVries, 2001; Hosoi et al., 2005) and by activation of mGluRs in photoreceptor terminals (Koulen et al., 1999; Hosoi et al., 2005). Postsynaptically, a retinal-specific regulator of G protein signaling (Ret-RGS1) may accelerate the deactivation of the mGluR6 cascade in ON-type bipolar cells, resulting in a rapid onset of depolarizing photoresponses (Dhingra et al., 2004). The mGluR6 has a high apparent affinity to glutamate (EC50 = 10 mM) (de la Villa et al., 1995), and almost all nonselective cation channels regulated by mGluR6 are kept closed by tonically released glutamate in the dark (Field and Rieke, 2002; Sampath and Rieke, 2004). Therefore, it is expected that quick removal of glutamate from the synaptic cleft by EAATs may also be an important factor for prompt depolarizing photoresponses in ON-type bipolar cells (Rao-Mirotznik et al., 1998). A rod terminal has a single invagination, into which processes from RBCs and horizontal cells penetrate toward the synaptic ribbon (Rao-Mirotznik et al., 1995). Rod terminals are surrounded by glial Mu¨ller cells (Derouiche, 1996). Immunohistochemical studies in rodent retinas demonstrated the presence of EAATs in rods, RBCs, horizontal cells, and Mu¨ller cells (Rauen et al., 1996, 1998; Harada et al., 1998; Pow and Barnett, 2000; Wiessner et al., 2002). However, it is not clear which EAATs are essential for removing glutamate from the synaptic cleft at the rod-to-RBC synapse.

Neuron 64

Figure 1. Blockade of Glutamate Uptake Slows Photoresponses in RBCs (A) Photoresponses of a mouse RBC. The cell was whole-cell voltage clamped at 273 mV in the absence of inhibitory receptor blockers. Each trace was obtained by averaging four responses and fitted by a sigmoidal function (smooth curve). For comparison, the thick trace (+TBOA, 50 mM) is moved downward to superimpose on the thin trace (control). (B) No significant effect of TBOA on the response amplitude (n = 3). (C) TBOA-induced increase in the time for half maximal amplitude from the light onset (T1/2) (n = 3). (D) TBOA-induced decrease in the slope at T1/2 (n = 3). Error bars indicate 6 SEM. Asterisk, p < 0.05.

Applying the whole-cell recording techniques to RBCs in mouse retinal slices, we recorded the postsynaptic current evoked by electrical stimulation of rods. We examined the effects of pharmacological blockade and genetic deletion of specific subtypes of EAATs on the electrically evoked RBC currents and recorded the EAAT-coupled anion current from rods and RBCs. We found that the decay phase of the evoked RBC currents was significantly truncated by presynaptic EAATs but not by postsynaptic or glial EAATs. Simulation studies of glutamate diffusion at the rod-to-RBC synapse suggest that EAATs are arranged near the release site of the rod terminal at surprisingly high density. These results indicate that the elaborate spatial arrangement of EAATs on rod terminals is an essential factor for reliable synaptic transmission at the first visual synapse. Results Contribution of EAATs to Prompt Light-Induced Responses in RBCs We first examined whether EAATs could affect the transmission of visual information from rods to RBCs in the mouse retina. A step of light induced an inward current in an RBC whole-cell clamped at 273 mV, reflecting the opening of nonselective cation channels via mGluR6 deactivation (Wilson et al., 1987; Nawy and Jahr, 1990; Shiells and Falk, 1990; Nawy, 1999). The rising speed of light-induced currents depends on the removal rate of glutamate that has been released from rods in the dark as well as on the kinetics of the mGluR6 cascade. We compared the rising phase of light-induced currents in the absence and presence of DL-threo-b-benzyloxyaspartate (TBOA), a competitive antagonist of EAATs (Shimamoto et al., 1998). Application of TBOA slightly decreased the basal inward current in the dark (Figure 1A). Inhibition of glutamate uptake by TBOA probably increased basal glutamate concentration in the synaptic cleft, resulting in further mGluR6 activation. TBOA did not affect the amplitude of the light-induced current (Figure 1B) but slowed its rising phase (Figure 1A). We measured the time for half maximal amplitude from the light onset (T1/2) and the slope at T1/2 (Figures 1C and 1D). Both the increase in T1/2 and the decrease in slope were statistically significant. These results indicate that EAATs

at the rod-to-RBC synapse contribute to prompt depolarizing photoresponses in RBCs by facilitating the removal of glutamate from the synaptic cleft. RBC Responses Evoked by Electrical Stimulation of Rods The kinetics of glutamate removal estimated from the light-induced responses in RBCs may be obscured by slow kinetics of rod phototransduction (Stryer, 1986). Thus, we stimulated rods directly by a brief (200 ms) electrical pulse to elicit glutamate release (Higgs and Lukasiewicz, 1999). Tonic release of glutamate from rods in the dark was minimized by continuous application of a strong background light to the retinal slice. An extracellular electrode was positioned at the outer nuclear layer (ONL). When a current pulse (80 mA) was applied to the electrode, nearby rods were depolarized from a resting potential of 263 6 5 (mean 6 SEM) to 222 6 18 mV (n = 3) (Figure 2A). The voltage response reached a peak 5.3 6 0.4 ms after the stimulus onset and then repolarized with a decay time of 6.6 6 0.4 ms. A steady inward current was observed in an RBC voltage clamped at 273 mV, and electrical stimulation evoked an outward current (Figure 2B). As the stimulus intensity was increased, the amplitude of the evoked current was increased and finally saturated (Figures 2B and 2C). The plateau of the saturating response was close to zero (24.9 6 3.0 pA, n = 13), suggesting that the saturation may be ascribed to total closure of the mGluR6-mediated nonselective cation channels (de la Villa et al., 1995; Sampath and Rieke, 2004). This was proved by the observation that L-2-amino-4-phosphonobutyric acid (L-AP4), an agonist of group III mGluRs, decreased the resting inward current by +11.7 6 3.9 pA (n = 6), and completely suppressed the electrically evoked responses (Figures 2D and 2E). These results confirmed that the evoked responses in RBCs were mediated through mGluR6 activation by glutamate that was released from rods by electrical stimulation. The sigmoidal relationship between the stimulus intensity and the amplitude of the evoked postsynaptic current (Figure 2C) is consistent with the observation that the release rate of glutamate is regulated by graded membrane potential changes in rods (Baylor and Fuortes, 1970; Dowling and Ripps, 1973). However, we cannot rule out the possibility that the number of electrically

Presynaptic Glutamate Uptake by Rod Photoreceptors 65

Figure 2. mGluR6-Mediated RBC Responses Are Evoked by Electrical Stimulation of Rods (A) The electrically evoked potential jump recorded from a rod under the current clamp. Constant current pulses (200 ms, 80 mA) were applied through a stimulus electrode positioned near the outer segment of the clamped rod in the presence of 50 mM TBOA. The trace was obtained by averaging three responses. Stimulus artifacts are blanked. The arrowhead indicates the timing of electrical stimulation. (B) Electrically evoked responses recorded from a RBC voltage clamped at 273 mV with a perforated patch electrode. The stimulus intensity was changed as indicated. Current traces (average of more than three responses induced by the same stimulus intensity) were superimposed after baseline alignment. (C) The relationship between the stimulus intensity and the relative amplitude of responses. Data were obtained from 13 RBCs. Each point (mean 6 SEM) contains more than five data and could be fitted by a sigmoidal function. (D) Both the basal inward current and the electrically evoked response in control (thin trace) were suppressed by L-AP4 (thick trace; +L-AP4, 100 mM). Each trace was obtained by averaging more than ten responses. (E) The amplitude (mean 6 SEM) of maximally evoked responses in the absence (control) and presence of L-AP4 (n = 6). Asterisk, p < 0.05.

excited rods is also increased as the stimulus intensity is increased because approximately 20 rods converge on an RBC (Tsukamoto et al., 2001) (but see Discussion).

Prolongation of Evoked Responses by EAAT Antagonist We examined the effects of TBOA on the electrically evoked postsynaptic currents in RBCs to analyze quantitatively how EAATs affected the synaptic transmission from rods to RBCs. Figure 3A illustrates the superimposed current traces recorded from an RBC in the absence (thin traces) and presence (thick traces) of TBOA. The resting current was not affected by TBOA (shift of the resting current by TBOA was +0.2 6 1.4 pA; n = 6), indicating that the background light was strong enough to suppress the glutamate release from rods. Furthermore, the saturating level of the electrically evoked current was not affected by TBOA (shift of the saturating level by TBOA was 20.9 6 1.9 pA; n = 6). These results suggest that the mGluR6 cascade in RBCs remained intact in the presence of TBOA. Relative amplitude in the presence of TBOA was plotted against that in control (Figure 3B). Many points were above the diagonal line, indicating that TBOA slightly shifted the stimulus intensity-response relationship (see Figure 2C) to the left along the stimulus intensity axis. Application of TBOA significantly prolonged both the time to peak and the decay time (the time to e-fold reduction from the peak) of the evoked responses (Figures 3C and 3D). Especially, prolongation of the decay time was prominent. This tendency was observed in full range of the response amplitude. These results suggest that EAATs can efficiently remove glutamate from the synaptic cleft even when the glutamate concentration reaches a level high enough to evoke a saturating response in the RBC. It has been reported that exocytosed glutamate and protons from photoreceptors suppress their presynaptic Ca2+ currents (Koulen et al., 1999; DeVries, 2001; Hosoi et al., 2005). These negative feedback regulations become prominent as the concentration of glutamate or protons at the synaptic cleft is increased. It is expected that these negative feedback signals would truncate the late phase of the electrically evoked response, especially when an intense stimulus was applied in the presence of TBOA. This may lead to underestimation of the

Figure 3. Blockade of Glutamate Uptake Prolongs Electrically Evoked RBC Responses (A) Electrically evoked responses recorded from an RBC held at 273 mV with a perforated patch electrode in the absence (thin trace) and presence of 50 mM TBOA (thick trace). The arrowhead indicates the timing of stimulation. Stimulus intensities were 5, 8, and 20 mA (left to right), respectively. Each trace was obtained by averaging more than 11 responses. Baselines are aligned and artifacts are blanked. (B) The relationship between the response amplitude in control and that in the presence of TBOA evoked by electrical pulses with identical intensity. Data were obtained from 6 RBCs. (C and D) The time to peak and the decay time were increased by TBOA. Data were obtained from 13 cells in control (open circles) and eight cells in the presence of TBOA (filled circles). Each point (mean 6 SEM) contains more than ten data. Double asterisk, p < 0.01.

Neuron 66

Figure 4. Negligible Contribution of GLAST1, EAAC1, or GLT1 to the Rod-to-RBC Synapse (A and B) Electrically evoked RBC responses recorded from a GLAST1-deficient mouse (GLAST1 KO) or from an EAAC1-deficient mouse (EAAC1 KO). Each trace was obtained by averaging more than four responses. The arrowhead indicates the timing of stimulation. Stimulus intensities were 40, 60, and 80 mA for (A) and 20, 30, and 80 mA for (B), respectively. Baselines are aligned, and artifacts are blanked. (C and D) The time to peak and the decay time of electrically evoked RBC responses. Points representing wild-type mice (open circles) are identical to those shown in Figures 3C and 3D. Points for GLAST1-deficient mice (filled diamonds) and those for EAAC1-deficient mice (filled triangles) were obtained from 17 and 13 RBCs, respectively. Each point (mean 6 SEM) contains more than eight data. (E and F) The effects of TBOA (50 mM) on electrically evoked RBC responses in mutant mice. Stimulus intensity was 40 mA. RBCs were held at 273 mV with a perforated patch electrode. (G) No significant effect of DHK (200 mM), a selective blocker of GLT1, on the evoked RBC responses in a wild-type mouse. Stimulus intensity was 80 mA. The RBC was held at 273 mV with a perforated patch electrode.

effects of TBOA. Because this underestimation was not critical for our model simulations (see below), we did not further analyze a possible contribution of these negative feedback signals to the electrically evoked RBC responses. Negligible Contribution of GLAST1, EAAC1, or GLT1 to the Rod-to-RBC Synapse Immunohistochemical studies demonstrated the presence of four subtypes of EAATs in the outer plexiform layer (OPL) of rodent retinas. GLAST1 is selectively expressed on glial Mu¨ller cells (Rauen et al., 1996, 1998; Harada et al., 1998), GLT1 on cones and diverse bipolar cells (Rauen et al., 1996; Harada et al., 1998; Haverkamp and Wa¨ssle, 2000), EAAC1 mainly on horizontal cells (Rauen et al., 1996; Wiessner et al., 2002), and EAAT5 on rods and some bipolar cells (Pow and Barnett, 2000). There is no expression of EAAT4 in the retina. Accumulating evidence indicates that Mu¨ller cells, which express GLAST1 densely, have the ability to absorb a large quantity of glutamate (Derouiche, 1996; Rauen et al., 1998; Rauen, 2000). It has also been shown that cone presynaptic GLT1 shapes the light responses of the postsynaptic neurons (Vandenbranden et al., 1996; Gaal et al., 1998). At the rod-to-RBC synapse, postsynaptic dendrites of RBCs and horizontal cells, where functional EAATs might be expressed, invaginate into a rod terminal (Rao-Mirotznik et al., 1995). We examined which subtype of EAATs was essential for removing glutamate from the synaptic cleft at the rod-to-RBC synapse. To elucidate whether GLAST1 of Mu¨ller cells plays an important role in glutamate removal at the rod-to-RBC synapse, we used the retinal slices obtained from GLAST1-deficient mice (Watase et al., 1998) (Figure 4A). The time to peak and the decay time of the electrically evoked RBC responses in the GLAST1 deficient mice were similar to those in the wild-type mice (Figures 4C and 4D). Application of TBOA prolonged the RBC responses in GLAST1-deficient mice (Figure 4E), indicating that GLAST1 of Mu¨ller cells may not play a predominant role at the rod-to-RBC synapse.

We also examined a possible participation of EAAC1 expressed on horizontal cells. The time to peak and the decay time of the electrically evoked RBC responses recorded from the EAAC1-deficient mice (Peghini et al., 1997) were similar to those from the wild-type mice (Figures 4B–4D). The RBC responses were prolonged by application of TBOA (Figure 4F). Thus, the contribution of EAAC1 of horizontal cells may be negligible at the rodto-RBC synapse. Furthermore, in wild-type mice, application of dihydrokainic acid (DHK), a selective blocker of GLT1, failed to modify the electrically evoked RBC responses (Figure 4G). This is consistent with the observation that GLT1 is the cone-pathway-specific EAAT (Harada et al., 1998; Pow and Barnett, 2000). These results suggest that the remaining subtype of EAATs, namely EAAT5 expressed on rods and/or RBCs, may play a predominant role at the rod-to-RBC synapse. Essential Role of Rod EAATs at the Rod-to-RBC Synapse Immunohistochemical studies have shown in the adult rat retina that EAAT5, a retina-specific glutamate transporter, is present densely in the photoreceptor layer (Pow and Barnett, 2000). To demonstrate physiologically that the rod EAATs play a functional role at the rod-to-RBC synapse in the mouse retina, we monitored the anion current coupled to EAATs (Arriza et al., 1997; Otis and Jahr, 1998; Palmer et al., 2003). A rod was whole-cell voltage clamped with a pipette filled with an NO32 solution to magnify the EAAT-coupled anion current, and glutamate was puff applied to the OPL. Glutamate evoked a prominent inward current, which was suppressed by TBOA (Figures 5A and 5B), indicating the presence of EAATs on rod terminals. We next investigated whether the presynaptic EAATs of rods can recollect glutamate released from their own terminals. In the rod dialyzed with NO32, a short depolarizing pulse immediately induced a large inward current (Figure 5C). This current was suppressed by TBOA, indicating that it was coupled to EAATs. Similar results were

Presynaptic Glutamate Uptake by Rod Photoreceptors 67

Figure 5. EAATs in the OPL Are Located Not on RBC Dendrites but on Rod Terminals (A) Puff application of glutamate (100 mM) to the OPL elicited an inward current in a rod dialyzed with NO32. Application of TBOA (50 mM) suppressed the glutamate-induced inward current. (B) Effects of TBOA on glutamate-induced responses in rods (n = 3). Mean 6 SEM. Asterisk, p < 0.05. (C) A depolarizing pulse (from 284 to 234 mV for 1 ms) elicited an inward current in a rod dialyzed with NO32. TBOA suppressed the inward current. (D) Effects of TBOA on the inward currents elicited by depolarizing pulses in rods (n = 4). Mean 6 SEM. Double asterisk, p < 0.01. (E) Puff application of glutamate to the OPL failed to elicit the response in an RBC dialyzed with NO32. (F) Puff application of glutamate to the IPL elicited an inward current in the RBC shown in (E). TBOA suppressed the glutamate-induced inward current. (G) Summary of glutamate puff experiments in RBCs (n = 5). Dendrites and axon terminals of RBCs are in the OPL and in the IPL, respectively. Cells were held at 269 mV. Mean 6 SEM. Asterisk, p < 0.05.

obtained from three other rods (Figure 5D). These results suggest that the rod EAATs, possibly EAAT5, are located close to the release sites and able to recollect released glutamate efficiently. We also examined a possible existence of EAATs on RBCs. The mGluR6 molecules are located at the base (not the tip) of the invaginating RBC dendrites (Vardi, et al., 2000), and thus EAATs would also be expressed at their base, where puff applied glutamate may reach. However, an RBC dialyzed with NO32 did not respond to glutamate puff applied to its dendrites in the OPL (Figure 5E). On the other hand, the identical RBC could generate an inward current in response to glutamate puff applied to its axon terminal in the inner plexiform layer (IPL) (Figure 5F). This current was suppressed by TBOA, indicating that EAATs are localized to the RBC terminal (Figure 5G). These results indicate that rod presynaptic EAATs but not RBC dendritic EAATs are responsible for removal of glutamate from the synaptic cleft at the rod-to-RBC synapse. Model Simulations of Glutamate Diffusion at the Rod-to-RBC Synapse The above physiological experiments revealed that rod EAATs are essential for rapid removal of glutamate from the synaptic cleft at the rod-to-RBC synapse. However, it is difficult to show immunohistochemically the precise location and number of EAAT molecules on rod terminals. Thus, we performed model simulations of glutamate diffusion in the synaptic cleft of the rod-to-RBC synapse and estimated a possible spatial arrangement of EAAT molecules on rod terminals. To determine the relevant parameters for our model, it is useful to know how long released glutamate stays in the synaptic cleft before diffusing out from the rod-toRBC synapse. We estimated the ‘‘response rate,’’ with which unit responses are elicited. A unit response is de-

fined as an expected mean response induced by closure of nonselective cation channels via activation of one mGluR6 molecule by glutamate. Note that the unit response is different from the quantum response elicited by single vesicular release. Even when the number of glutamate molecules released by a single vesicle fusion is constant, the number of glutamate molecules that can interact with postsynaptic receptors may vary depending on diffusion and uptake in the synaptic cleft. The large evoked RBC responses were deconvoluted with a minimal evoked RBC response in the absence of TBOA (see Figure 3A), which was tentatively assumed as a unit response (see Discussion). Figure 6A illustrates the time course of the ‘‘response rate’’ after application of an intense electrical pulse in the absence (control) and presence of TBOA. The decay time constant of the calculated ‘‘response rate’’ was approximately 15 ms and 85 ms in the absence and presence of TBOA, respectively (Figure 6B). These results suggest that when rod EAATs are not available (+TBOA), the glutamate concentration at the postsynaptic mGluR6 area may be held higher than the mGluR6 sensitivity for no less than 100 ms. On the other hand, when rod EAATs are intact, the glutamate concentration may decay below the mGluR6 sensitivity in approximately 30 ms. It is frequently assumed that glutamate concentration in the synaptic cleft may be decreased very rapidly by diffusion. Although the glutamate sensitivity of mGluR6 is higher than non-NMDA receptors (de la Villa et al., 1995; Hestrin, 1992) and the release rate of glutamate from photoreceptors is very high (Thoreson et al., 2004), it is surprising that glutamate concentration may be kept higher than the mGluR6 sensitivity for over 100 ms after blockade of EAAT functions. We developed a three-dimensional model that mimicked the morphology of invaginating synapse between a rod and an RBC (Figure 6C) (Rao-Mirotznik et al., 1995;

Neuron 68

Figure 6. Prolonged Activation of mGluR6 Requires Numerous Vesicle Fusion (A) The time course of the ‘‘response rate’’ (see text) calculated from the RBC responses evoked by electrical stimulation (10 mA) of rods. Single exponential functions (smooth curves) have the decay time constant of 9.9 ms in control and 90.0 ms in the presence of TBOA (50 mM). (B) Increase in the decay time constant of the ‘‘response rate’’ by TBOA (n = 4). Asterisk, p < 0.05. (C) Geometric structure of the model. A postsynaptic dendritic process of the RBC invaginates into the presynaptic rod terminal. Glutamate molecules are released instantaneously from the Release Site (top view) of the rod terminal. The mGluR6 molecules are located at the base of the invaginating postsynaptic process (mGluR6 Area; side view). (D) Effects of the number of vesicle fusion on the time course of the glutamate concentration experienced by the mGluR6 Area. The decay time constants are 5.9, 13, 16, 17, and 17 ms for fusion of two, eight, 32, 128, and 512 vesicles, respectively. The working range of mGluR6 is shaded. (E) Effects of the diffusion coefficient on the decay of glutamate. The decay time constants are 17, 9.2, and 4.3 ms for the diffusion coefficient of 0.2, 0.4, and 0.8 mm2 $ ms21, respectively. (F) Effects of the width of synaptic cleft on the decay of glutamate. The decay time constants are 17 ms in all conditions. For each trace shown in (D)–(F), calculated values of five trials were averaged.

Vardi et al., 2000), and a Monte Carlo simulation of glutamate diffusion (Bartol et al., 1991) was performed. Firstly, we simulated the situation without functional EAATs. In the mammalian rod terminal, approximately 130 vesicles among 770 vesicles that are tethered to a synaptic ribbon are docked to the active zone (Rao-Mirotznik et al., 1995; Tsukamoto et al., 2001). Accumulating evidence suggests that the entire vesicles tethered to the ribbon may correspond to the readily releasable pool (von Gersdorff and Matthews, 1994; von Gersdorff et al., 1996; Moser and Beutner, 2000). Thus, we changed the number of instantaneous vesicle fusion in the range between 2 and 512. Figure 6D shows the time course of the glutamate concentration experienced by the mGluR6 area. The effective glutamate concentration for mGluR6 activation is ranged between 1 and 100 mM (Figure 6D, shaded area) (de la Villa et al., 1995). To activate the mGluR6 molecules for more than 100 ms, more than 128 vesicles should be fused instantaneously. We also calculated the time course of glutamate concentration changes when 128 vesicles were fused not instanta-

neously but for 12.8 ms because a strong 200 ms pulse depolarized nearby rods for approximately 10 ms (see Figure 2A). However, two decay curves under different release conditions were merged together approximately 30 ms after the start of vesicle fusion, and then there were no significant differences between two curves (data not illustrated). The diffusion coefficient is the most critical parameter to simulate how molecules diffuse. The proposed diffusion coefficient of glutamate widely ranged from 0.2 to 1.0 mm2 $ ms21 (Barbour, 2001; Franks et al., 2002; Rusakov, 2001). When we employed the diffusion coefficient of 0.2 mm2 $ ms21, the fusion of 128 vesicles could activate mGluR6 for more than 100 ms (Figure 6E). In cases of the diffusion coefficient larger than 0.2 mm2 $ ms21, the glutamate concentration rapidly decayed, suggesting that more than 128 vesicles should be fused to maintain the glutamate concentration higher than the mGluR6 sensitivity for more than 100 ms. The width of the synaptic cleft may alter the time for glutamate to diffuse out from the cleft. However, the

Presynaptic Glutamate Uptake by Rod Photoreceptors 69

Figure 7. Presynaptic EAATs Are Densely Expressed Near Active Zone (A and B) Spatial arrangement of EAATs (dark area). (C) Effects of the density of presynaptic EAATs on the decay of glutamate. EAATs were expressed on the 4 mm 3 4 mm area (A). The decay time constants are 17, 10, 7.7, 5.2, and 3.7 ms for the density of 0, 1,250, 2,500, 5,000, and 10,000 mm22, respectively. The working range of mGluR6 is shaded. (D) Effects of spatial arrangement of EAATs on the decay of glutamate. The decay time constants are 3.2 and 6.5 ms for the density of 10,000 mm22 and 2,500 mm22 (8 mm 3 8 mm), respectively. The control trace (10,000 mm22, 4 mm 3 4 mm) is identical to that shown in (C). For each trace shown in (C) and (D), calculated values of five trials were averaged.

decay of glutamate concentration was only slightly affected when the width of the synaptic cleft was changed between 10 and 30 nm (Figure 6F). Based on model simulations without functional presynaptic EAATs, we fixed the parameters of the number of instantaneous vesicle fusion, the diffusion coefficient, and the width of the synaptic cleft as 128 vesicles, 0.2 mm2 $ ms21, and 20 nm, respectively. Spatial Arrangement of Rod EAATs Now we investigate how EAATs are arranged in the rod terminal. EAATs have a high affinity to glutamate and a slow turnover rate of glutamate (Wadiche et al., 1995; Diamond and Jahr, 1997; Wadiche and Kavanaugh, 1998). Their spatial arrangement should satisfy the constraint that the glutamate concentration at the postsynaptic mGluR6 area decays below the mGluR6 sensitivity within approximately 30 ms after 128 vesicles are instantaneously fused to the release site. Immunohistochemical studies have shown that EAATs are not observed in the invaginating region but mostly located at the basal region of photoreceptor terminals (Vandenbranden et al., 2000). Thus, EAATs were modeled to be arranged in the basal region of presynaptic terminal (4 mm 3 4 mm) at various densities (Figure 7A). The more densely EAATs were arranged, the more rapidly the glutamate concentration at the mGluR6 area decayed after the glutamate release (Figure 7C). The glutamate concentration decreased below EC50 of mGluR6 (10 mM) in 30 ms when the density of EAATs was as high as 10,000 mm22 (Figure 7C, bold line). This density is close to an upper limit because an EAAT molecule is assumed to be approximately 10 nm in diameter (Eskandari et al., 2000). This simulation study suggests that a vast number of EAATs must be located on the rod terminal to take up a flood of glutamate quickly. In the mouse retina, approximately 20 rods converge on an RBC (Tsukamoto et al., 2001). As rod terminals

are densely packed in the OPL, it is possible that a part of glutamate released from a rod may be taken up by EAATs of neighboring rod terminals. To test this possibility, extra EAATs were added to the area surrounding the basal region of presynaptic terminal (4 mm 3 4 mm), i.e., EAATs were arranged in the 8 mm 3 8 mm area at the density of 10,000 mm22 (Figure 7B). However, addition of extra EAATs only slightly affected the time course of glutamate decay at the postsynaptic mGluR6 area (Figure 7D), suggesting that the densely arranged EAATs at the basal region of a rod terminal may be sufficient to remove the released glutamate quickly. If the density of EAATs at the basal region of presynaptic terminal were relatively low, the extra EAATs in the surrounding area would contribute to glutamate removal. To test this possibility, we fixed the total number of EAAT molecules, and their density was varied (Figure 7D). The simulation study revealed that EAATs widely arranged at a low density (the 8 mm 3 8 mm region at the density of 2,500 mm22) removed glutamate from the postsynaptic mGluR6 area more slowly than EAATs closely arranged at a high density (the 4 mm 3 4 mm region at the density of 10,000 mm22). Therefore, rod EAATs should be arranged near the release site at high density for efficient removal of glutamate from the synaptic cleft. Simulation of Light-Induced Decrease in Tonic Glutamate Release Rods maintain a high tonic rate of glutamate release in the dark. The tonic fusion rate is estimated to be at least 40 vesicles $ s21, which is much higher than that estimated at conventional synapses (Rao et al., 1994). RBCs are highly sensitive to detect a pause of the tonic glutamate release in response to light stimulation (Rao et al., 1994; Field and Rieke, 2002). To test whether our model can respond appropriately to light stimulation, we simulated the time course of glutamate concentration

Neuron 70

Figure 8. Uptake by EAATs Can Follow the Tonic Release of Glutamate (A–C) The time course of the glutamate concentration experienced by the mGluR6 Area when tonic vesicle fusion was suddenly stopped (top). EAATs were expressed at the density of 10,000 mm22 on the 4 mm 3 4 mm presynaptic area (see Figure 7A). For each trace, calculated values of 20 trials were averaged. The working range of mGluR6 is shaded. (D) Cumulative curves of the released and transported glutamate molecules. Each trace was calculated from a single trial under the condition shown in (C) (EAAT = 10,000 mm22).

changes at the postsynaptic mGluR6 area when the tonic vesicle fusion was suddenly stopped. The tonic fusion rate in the dark was changed in the range between 32 and 512 vesicles $ s21. Under each condition, the glutamate concentration was only slightly lower in the presence of functional EAATs than in their absence (Figures 8A–8C). When the fusion rate was higher than 128 vesicles $ s21, most mGluR6 molecules were maximally activated, resulting in closure of nonselective cation channels (i.e., hyperpolarization in the dark). As soon as the fusion rate was dropped to zero (as if a strong light was turned on), the accumulated glutamate was quickly removed by EAATs even when the fusion rate had been as high as 512 vesicles $ s21. In the absence of functional EAATs, removal of the accumulated glutamate occurred slowly, and this would cause a delay of depolarizing photoresponses in RBCs. It should be noted that in the model the rate of glutamate uptake could follow the rate of tonic glutamate release at 512 vesicles $ s21 (Figure 8D; see slopes of two cumulative curves during tonic vesicle fusion). Approximately 100 ms after the termination of vesicle fusion, the released glutamate molecules were totally transported into the rod (see convergence of two cumulative curves after termination of vesicle fusion). This simulation suggests that rods may be equipped with a self-sufficient recollecting system of glutamate (Winkler et al., 1999). Discussion In the present study, combining electrophysiological experiments and model simulations, we investigated how EAATs functioned at the rod-to-RBC synapse in the mouse retina. Experimental results indicate that EAATs on rod terminals play a predominant role in removing glutamate from the synaptic cleft, whereas EAATs in Mu¨ller cells and postsynaptic neurons make a negligible contribution. Model simulations suggest that densely packed EAATs on the rod terminal are essential for efficient removal of glutamate from the synaptic cleft.

Validity of Parameters for Model Simulations Our model simulations suggest that presynaptic EAATs are densely packed close to the upper limit set by the physical size of an EAAT molecule (Figures 6 and 7). The simulations were based on two main parameters: the glutamate sensitivity of mGluR6 and the decay of ‘‘response rate’’ estimated by deconvolution. The glutamate sensitivity of mGluR6 is not yet determined in mouse RBCs. In the present study, we set the glutamate sensitivity (EC50) of mGluR6 at 10 mM, which was reported by de la Villa et al. (1995) in cat RBCs. If we set the EC50 at 1 mM, 32 vesicles should be fused instantaneously to activate mGluR6 molecules for 100 ms in the absence of functional EAATs, and EAATs should be arranged in the area of 4 mm 3 4 mm at the density of 5,000 mm22 to terminate the mGluR6 activation in 30 ms (data not illustrated). However, it seems difficult to imagine that the instantaneous fusion of only 32 vesicles evokes an almost saturating RBC response because it has been reported that no less than 700 vesicles can be fused to the active zone of a rod (Rao-Mirotznik et al., 1995; Tsukamoto et al., 2001; Thoreson et al., 2004). On the other hand, if we set the EC50 at a higher value than 10 mM, the density of EAATs on the rod terminal should be much higher than 10,000 mm22, which is physically impossible. Therefore, the EC50 of mGluR6 in mouse RBCs is probably close to 10 mM. The expected mean response induced by closure of nonselective cation channels via activation of one mGluR6 molecule by glutamate has not been measured, and the detailed kinetic model of mGluR6-mediated responses is yet to be determined. Furthermore, it is difficult to detect a miniature EPSC-like event in RBCs. Thus, to estimate the ‘‘response rate,’’ we assumed a minimal measurable RBC response evoked by a weak electrical pulse as a unit response induced by activation of a single mGluR6 molecule. Not amplitude but time course of the unit response is critical to estimate the time course of the ‘‘response rate.’’ Although the kinetics of mGluRs is assumed to be slow, it is conceivable

Presynaptic Glutamate Uptake by Rod Photoreceptors 71

that the unit response has much faster kinetics than the minimal measurable response. If this were the case, the duration of mGluR6 activation would be longer than 100 ms in the absence of functional EAATs, and thus much higher density of EAATs would be required to terminate the mGluR6 activation quickly. One may argue against the linearity assumption of deconvolution because mGluR6-mediated responses are desensitized. However, the desensitization of mGluR6mediated responses is mainly induced by Ca2+ entry through nonselective cation channels, and it recovers in hundreds of milliseconds after termination of Ca2+ entry (Shiells and Falk, 2000; Nawy, 2004; Berntson et al., 2004). Thus, mGluR6-mediated pathway may be steadily desensitized by the basal Ca2+ entry and recover during responses. Indeed, we sometimes observed undershoot of the electrically evoked responses (see Figure 2B), which may be ascribed to transient recovery from desensitization. The recovery from desensitization would accelerate the decay of responses, and thus the decay of glutamate concentration is expected to be much slower than 100 ms, i.e., more than 128 vesicles should be fused. Again, it is not plausible to arrange much higher density of EAATs (>10,000 mm22) on the rod terminal to terminate the response quickly. Therefore, the glutamate sensitivity of mGluR6 and the estimated ‘‘response rate’’ seem to be relevant in our model simulations. Contribution of Glial and Postsynaptic EAATs at the Rod-to-RBC Synapse To identify dominant subtypes of EAATs at the rod-toRBC synapse, we measured the electrically evoked RBC responses in GLAST1- or EAAC1-deficient mice (Figure 4). However, response properties in both mutant mice were not significantly different from those in wildtype mice in control (without TBOA). The RBC responses of both mutants were prolonged by application of TBOA similar to those of wild-type mice. Western blotting and immunohistochemical analysis have excluded the possibility of either overexpression of residual EAAT subtypes or emergence of new EAAT subtypes (Peghini et al., 1997; Harada et al., 1998). Moreover, our model simulations indicate that EAATs away from the release site may play a negligible role in removing glutamate from the synaptic cleft (Figure 7D). All of these results strongly support the idea that EAATs of rod terminals but not those of glial Mu¨ller cells or postsynaptic neurons are essential for removing glutamate from the synaptic cleft at the rod-to-RBC synapse. This also implies that strong presynaptic EAATs may serve to confine the diffusion of glutamate to the synaptic cleft, enabling reliable signal transmission without cross talk. Contribution of Extrasynaptic Glutamate Receptors Activation of extrasynaptic receptors by transmitter spilled out from a synaptic cleft is widely known as ‘‘spillover’’ (Diamond, 2002). Many studies have demonstrated that spatially heterogeneous distribution of different types of receptors on postsynaptic membrane elicits a response with multiple components (Rossi and Hamann, 1998; Rusakov and Kullmann, 1998; Matsui et al., 2001; Reichelt and Kno¨pfel, 2002). The ratio of the response components may vary depending on the

amount of transmitter released. Late phase of the response is frequently mediated by the extrasynaptic receptors, which have a higher affinity and slower kinetics than that of the synaptic receptors. However, if postsynaptic receptors are homogeneous, spillover may affect not the decay phase but the rising phase of responses only slightly (Nielsen et al., 2004). Because the rodto-RBC synaptic transmission is mostly mediated by mGluR6 (Figures 2D and 2E), it seems unlikely that extrasynaptic mGluR6s, if any, affected the time course of evoked responses. Moreover, even if other types of glutamate receptors such as non-NMDA receptors were expressed on extrasynaptic membrane of postsynaptic dendrites (Hack et al., 2001; Kamphuis et al., 2003), they could not contribute to the late phase of responses because the affinity of non-NMDA receptors to glutamate (EC50 = 1 mM) (Hestrin, 1992) is much lower than that of mGluR6 (EC50 = 10 mM) (de la Villa et al., 1995). Advantage of Presynaptic EAATs Rods have a large readily releasable pool and the release of glutamate is controlled by graded potential changes (Baylor and Fuortes, 1970; Thoreson et al., 2004). The mGluR6-mediated responses in RBCs are almost saturated in the dark (Sampath and Rieke, 2004), indicating that rods actually release glutamate at a high tonic rate in the dark. To maintain the high tonic rate, depletion of glutamate should be avoided either by newly synthesizing glutamate or by recollecting released glutamate from the synaptic cleft. Glutamate transported into glial cells is converted into nonneuroactive glutamine (Hertz et al., 1999). Glutamine released from glial cells is taken up by neurons, where glutamate is synthesized from glutamine by mitochondrial glutaminase. If the glial EAATs played a dominant role in capturing glutamate, this indirect pathway would be useful. However, our model simulations suggest that almost all glutamate molecules tonically released from a rod are recollected by rod presynaptic EAATs, even when the fusion rate is as high as 512 vesicles $ s21 (Figure 8D). Therefore, it is likely that rods may use a direct, almost self-sufficient recollecting system of glutamate rather than the glutamate-glutamine cycle (Winkler et al., 1999). The estimated diffusion coefficient of glutamate was relatively low (0.2 mm2 $ ms21) (Figure 6E). It has been suggested that the diffusion coefficient of glutamate is lower in the synaptic cleft than in free solution (Nielsen et al., 2004). Slow glutamate diffusion might be apparently inconsistent with the requirement of rapid removal of glutamate from the synaptic cleft. However, slow glutamate diffusion may be favorable for EAATs to capture glutamate efficiently. The rod-to-RBC synapse may have some hindrance of glutamate diffusion (Barbour and Ha¨usser, 1997). Experimental Procedures Retinal Slice Preparation and Recordings Adult (postnatal day >28) C57/BL6 (wild-type) mice were kept in a room maintained at 23ºC on a 12 hr light/dark cycle. The animals were dark adapted for approximately 20 min prior to cervical dislocation. All procedures were performed in accordance with the guidelines of The University of Tokyo and The Physiological Society of Japan and were approved by the Animal Experiment Committee

Neuron 72

of the Faculty of Letters, The University of Tokyo. Retinal slices were prepared as described previously (Matsui et al., 2001). Briefly, under a dim red light the retina was isolated from the pigment epithelium and the ganglion cell layer side was attached to a piece of filter paper. The retina and the filter paper were sliced together into 200 mm sections. The slices were arranged in the recording chamber, which was mounted on the stage of a microscope equipped with infrared differential interference optics (Eclipse E600-FN; Nikon, Tokyo, Japan) in a light-tight Faraday cage. All recordings, except the EAAT-coupled anion current recordings (31.0ºC–32.0ºC) were performed at 34.5ºC–36.5ºC. In some experiments, we used GLAST1deficient mice (Watase et al., 1998) and EAAC1-deficient mice (Peghini et al., 1997). The extracellular control solution consisted of (in mM) 120 NaCl, 3.1 KCl, 2 CaCl2, 1 MgSO4, 23 NaHCO3, 0.5 KH2PO4, and 6 glucose; the pH adjusted to w7.6 with 95% O2/5% CO2 at 36ºC. Unless otherwise mentioned, the solution contained 100 mM (-)-bicuculline methochloride (bicuculline), 200 mM (1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid (TPMPA), and 1 mM strychnine hydrochloride (strychnine) to block GABAA, GABAC, and glycine receptors, respectively. In glutamate puff experiments, Ca2+ was replaced by Co2+. TBOA (50–200 mM) was used to block all EAAT subtypes. DHK (200 mM) and L-AP4 (100 mM) were applied to block GLT1 and mGluR6, respectively. These pharmacological agents were bath applied or pressure applied from a puff pipette (theta tube; one compartment was filled with the control solution and the other with the test solution). The pipette solution for recordings from both RBCs and rods consisted of (in mM) 105 CsCH3SO3, 0.5 CaCl2, 5 EGTA, 20 HEPES, 10 tetraethylammonium chloride, 5.5 MgCl2, 5 ATP disodium salt, 0.5 GTP sodium salt, 0.25% Lucifer yellow dipotassium salt; pH adjusted to 7.6 with CsOH. For recordings of postsynaptic RBC currents evoked by electrical stimulation of rods, 500 mg $ ml21 amphotericin B or 50–100 mg $ ml21 gramicidin D was added to the pipette solution. To record the EAAT-coupled anion current, CH3SO32 was replaced by NO32. The pipette solution for recordings of the rod membrane potential consisted of (in mM) 135 K-gluconate, 3 KCl, 0.2 EGTA, 10 HEPES, 2.5 MgCl2, 2 ATP disodium salt, 0.1 GTP sodium salt, 0.1% Lucifer yellow dipotassium salt; pH adjusted to 7.6 with KOH. Bicuculline, TPMPA, TBOA, DHK, and L-AP4 were purchased from Tocris; and strychnine, amphotericin B, and gramicidin D from Sigma. Stimulation Light stimuli were delivered from a light-emitting diode (LED; emission maximum at 520 nm). The retinal slice was diffusely illuminated from underneath of the recording chamber. The light intensity at the position of the retinal slice was set to 2.3 3 104 photons $ mm22 $ s21 to elicit photoresponses in RBCs. In the electrical-stimulation experiments, a background light with the intensity of 4.0–4.5 3 106 photons $ mm22 $ s21 was continuously applied to suppress the spontaneous glutamate release from rods (Berntson and Taylor, 2000). Short electrical pulses (200 ms in duration) were applied through the pipette filled with the superfusate. The stimulating pipette was inserted into the ONL directly above the cell body of the examined RBC in the inner nuclear layer. The position of the stimulation pipette was critical; a slight lateral shift (<10 mm) along the ONL sharply reduced the amplitude of the evoked current, and thus we searched for a ‘‘hot spot’’ to elicit a maximal response in the RBC by moving the stimulating pipette. Statistical significance was assessed by two-tailed, paired or unpaired Student’s t tests. Differences were considered significant if p was <0.05. Monte Carlo Simulation We performed diffusion simulations by programming the random walk of glutamate molecules in the three-dimensional space (Bartol et al., 1991) and the Markov-style kinetic model of glutamate transporter (Diamond, 2001) with the C language. Briefly, every 1 ms, each glutamate molecule moved along the vector, Cartesian coordinates of which were randomly selected by using the inverse error function. The interaction between glutamate molecules was not assumed. 10 nm grids were drawn on the presynaptic and postsynaptic membrane surfaces. Each unit square of the grids had zero or one mole-

cule (presynaptic EAAT or postsynaptic mGluR6). If a glutamate molecule hit the membrane surface, it bounced off the surface by elastic collision, except when the glutamate molecule was captured by an EAAT molecule. The model of EAAT was composed of four states: Tout (before glutamate binding), GluTout (after glutamate binding), GluTin (after translocation), and Tin (after glutamate unbinding in cytoplasm) (Diamond, 2001). The binding and unbinding rate constants were 2 3 107 M21 $ s21 and 300 s21, respectively. The forward rate constant for each step (GluTout > GluTin > Tin > Tout) was 500 s21, 2000 s21, and 40 s21, respectively. Reverse transport was not assumed. Each vesicle contained 2,000 glutamate molecules (Rao-Mirotznik et al., 1998). Vesicle fusion occurred instantaneously at the central point of the presynaptic active zone (the release site). The area of synaptic cleft was unknown, and thus we assumed a fairly large area, 8 mm 3 8 mm, taking into account the long decay time constant of the ‘‘response rate’’ in the absence of functional EAATs. Taking advantage of the spatial symmetry of our three-dimensional model, we performed simulations only in a quarter part of the geometry. All traces representing the time course of glutamate concentration changes in figures were Gaussian filtered (s = 1 ms) for better view.

Acknowledgments This work was supported by Grant-in-Aid for Scientific Research on Priority Areas (12053212) and the Special Coordination Funds for Promoting Science and Technology (the Neuroinformatics Research in Vision Project) from The Ministry of Education, Culture, Sports, Science and Technology to M.T. J.H. is a Research Fellow of the Japan Society for the Promotion of Science. Received: December 26, 2005 Revised: January 24, 2006 Accepted: February 17, 2006 Published: April 5, 2006

References Arriza, J.L., Eliasof, S., Kavanaugh, M.P., and Amara, S.G. (1997). Excitatory amino acid transporter 5, a retinal glutamate transporter coupled to a chloride conductance. Proc. Natl. Acad. Sci. USA 94, 4155–4160. Barbour, B. (2001). An evaluation of synapse independence. J. Neurosci. 21, 7969–7984. Barbour, B., and Ha¨usser, M. (1997). Intersynaptic diffusion of neurotransmitter. Trends Neurosci. 20, 377–384. Bartol, T.M., Jr., Land, B.R., Salpeter, E.E., and Salpeter, M.M. (1991). Monte Carlo simulation of miniature endplate current generation in the vertebrate neuromuscular junction. Biophys. J. 59, 1290– 1307. Baylor, D.A., and Fuortes, M.G. (1970). Electrical responses of single cones in the retina of the turtle. J. Physiol. 207, 77–92. Bergles, D.E., and Jahr, C.E. (1997). Synaptic activation of glutamate transporters in hippocampal astrocytes. Neuron 19, 1297–1308. Berntson, A., and Taylor, W.R. (2000). Response characteristics and receptive field widths of on-bipolar cells in the mouse retina. J. Physiol. 524, 879–889. Berntson, A., Smith, R.G., and Taylor, W.R. (2004). Postsynaptic calcium feedback between rods and rod bipolar cells in the mouse retina. Vis. Neurosci. 21, 913–924. Clark, B.A., and Barbour, B. (1997). Currents evoked in Bergmann glial cells by parallel fibre stimulation in rat cerebellar slices. J. Physiol. 502, 335–350. Danbolt, N.C. (2001). Glutamate uptake. Prog. Neurobiol. 65, 1–105. de la Villa, P., Kurahashi, T., and Kaneko, A. (1995). L-glutamate-induced responses and cGMP-activated channels in three subtypes of retinal bipolar cells dissociated from the cat. J. Neurosci. 15, 3571–3582.

Presynaptic Glutamate Uptake by Rod Photoreceptors 73

Derouiche, A. (1996). Possible role of the Mu¨ller cell in uptake and metabolism of glutamate in the mammalian outer retina. Vision Res. 36, 3875–3878. DeVries, S.H. (2001). Exocytosed protons feedback to suppress the Ca2+ current in mammalian cone photoreceptors. Neuron 32, 1107– 1117. Dhingra, A., Faurobert, E., Dascal, N., Sterling, P., and Vardi, N. (2004). A retinal-specific regulator of G-protein signaling interacts with Gao and accelerates an expressed metabotropic glutamate receptor 6 cascade. J. Neurosci. 24, 5684–5693.

Masu, M., Iwakabe, H., Tagawa, Y., Miyoshi, T., Yamashita, M., Fukuda, Y., Sasaki, H., Hiroi, K., Nakamura, Y., Shigemoto, R., et al. (1995). Specific deficit of the ON response in visual transmission by targeted disruption of the mGluR6 gene. Cell 80, 757–765. Matsui, K., Hasegawa, J., and Tachibana, M. (2001). Modulation of excitatory synaptic transmission by GABAC receptor-mediated feedback in the mouse inner retina. J. Neurophysiol. 86, 2285–2298. Moser, T., and Beutner, D. (2000). Kinetics of exocytosis and endocytosis at the cochlear inner hair cell afferent synapse of the mouse. Proc. Natl. Acad. Sci. USA 97, 883–888.

Diamond, J.S. (2001). Neuronal glutamate transporters limit activation of NMDA receptors by neurotransmitter spillover on CA1 pyramidal cells. J. Neurosci. 21, 8328–8338.

Nawy, S. (1999). The metabotropic receptor mGluR6 may signal through Go, but not phosphodiesterase, in retinal bipolar cells. J. Neurosci. 19, 2938–2944.

Diamond, J.S. (2002). A broad view of glutamate spillover. Nat. Neurosci. 5, 291–292.

Nawy, S. (2004). Desensitization of the mGluR6 transduction current in tiger salamander On bipolar cells. J. Physiol. 558, 137–146.

Diamond, J.S., and Jahr, C.E. (1997). Transporters buffer synaptically released glutamate on a submillisecond time scale. J. Neurosci. 17, 4672–4687.

Nawy, S., and Jahr, C.E. (1990). Suppression by glutamate of cGMPactivated conductance in retinal bipolar cells. Nature 346, 269–271.

Dowling, J.E., and Ripps, H. (1973). Effect of magnesium on horizontal cell activity in the skate retina. Nature 242, 101–103. Eskandari, S., Kreman, M., Kavanaugh, M.P., Wright, E.M., and Zampighi, G.A. (2000). Pentameric assembly of a neuronal glutamate transporter. Proc. Natl. Acad. Sci. USA 97, 8641–8646. Field, G.D., and Rieke, F. (2002). Nonlinear signal transfer from mouse rods to bipolar cells and implications for visual sensitivity. Neuron 34, 773–785. Franks, K.M., Bartol, T.M., Jr., and Sefnowdki, T.J. (2002). A Monte Carlo model reveals independent signaling at central glutamatergic synapses. Biophys. J. 83, 2333–2348. Gaal, L., Roska, B., Picaud, S.A., Wu, S.M., Marc, R., and Werblin, F.S. (1998). Postsynaptic response kinetics are controlled by a glutamate transporter at cone photoreceptors. J. Neurophysiol. 79, 190– 196. Glowatzki, E., and Fuchs, P.A. (2002). Transmitter release at the hair cell ribbon synapse. Nat. Neurosci. 5, 147–154. Hack, I., Frech, M., Dick, O., Peichl, L., and Brandsta¨tter, J.H. (2001). Heterogeneous distribution of AMPA glutamate receptor subunits at the photoreceptor synapses of rodent retina. Eur. J. Neurosci. 13, 15–24. Harada, T., Harada, C., Watanabe, M., Inoue, Y., Sakagawa, T., Nakayama, N., Sasaki, S., Okuyama, S., Watase, K., Wada, K., and Tanaka, K. (1998). Functions of the two glutamate transporters GLAST and GLT-1 in the retina. Proc. Natl. Acad. Sci. USA 95, 4663–4666. Haverkamp, S., and Wa¨ssle, H. (2000). Immunocytochemical analysis of the mouse retina. J. Comp. Neurol. 424, 1–23. Hertz, L., Dringen, R., Schousboe, A., and Robinson, S.R. (1999). Astrocytes: glutamate producers for neurons. J. Neurosci. Res. 57, 417–428. Hestrin, S. (1992). Activation and desensitization of glutamate-activated channels mediating fast excitatory synaptic currents in the visual cortex. Neuron 9, 991–999. Higgs, M.H., and Lukasiewicz, P.D. (1999). Glutamate uptake limits synaptic excitation of retinal ganglion cells. J. Neurosci. 19, 3691– 3700. Hosoi, N., Arai, I., and Tachibana, M. (2005). Group III metabotropic glutamate receptors and exocytosed protons inhibit L-type calcium currents in cones but not in rods. J. Neurosci. 25, 4062–4072. Kamphuis, W., Dijk, F., and O’Brien, B.J. (2003). Gene expression of AMPA-type glutamate receptor subunits in rod-type ON bipolar cells of rat retina. Eur. J. Neurosci. 18, 1085–1092. Koulen, P., Kuhn, R., Wa¨ssle, H., and Brandsta¨tter, J.H. (1999). Modulation of the intracellular calcium concentration in photoreceptor terminals by a presynaptic metabotropic glutamate receptor. Proc. Natl. Acad. Sci. USA 96, 9909–9914. Lehre, K.P., and Danbolt, N.C. (1998). The number of glutamate transporter subtype molecules at glutamatergic synapses: chemical and stereological quantification in young adult rat brain. J. Neurosci. 18, 8751–8757.

Nielsen, T.A., DiGregorio, D.A., and Silver, R.A. (2004). Modulation of glutamate mobility reveals the mechanism underlying slow-rising AMPAR EPSCs and the diffusion coefficient in the synaptic cleft. Neuron 42, 757–771. Otis, T.S., and Jahr, C.E. (1998). Anion currents and predicted glutamate flux through a neuronal glutamate transporter. J. Neurosci. 18, 7099–7110. Otis, T.S., Kavanaugh, M.P., and Jahr, C.E. (1997). Postsynaptic glutamate transport at the climbing fiber-Purkinje cell synapse. Science 277, 1515–1518. Palmer, M.J., Taschenberger, H., Hull, C., Tremere, L., and von Gersdorff, H. (2003). Synaptic activation of presynaptic glutamate transporter currents in nerve terminals. J. Neurosci. 23, 4831–4841. Peghini, P., Janzen, J., and Stoffel, W. (1997). Glutamate transporter EAAC-1-deficient mice develop dicarboxylic aminoaciduria and behavioral abnormalities but no neurodegeneration. EMBO J. 16, 3822–3832. Pow, D.V., and Barnett, N.L. (2000). Developmental expression of excitatory amino acid transporter 5: a photoreceptor and bipolar cell glutamate transporter in rat retina. Neurosci. Lett. 280, 21–24. Rao, R., Buchsbaum, G., and Sterling, P. (1994). Rate of quantal transmitter release at the mammalian rod synapse. Biophys. J. 67, 57–63. Rao-Mirotznik, R., Harkins, A.B., Buchsbaum, G., and Sterling, P. (1995). Mammalian rod terminal: architecture of a binary synapse. Neuron 14, 561–569. Rao-Mirotznik, R., Buchsbaum, G., and Sterling, P. (1998). Transmitter concentration at a three-dimensional synapse. J. Neurophysiol. 80, 3163–3172. Rauen, T. (2000). Diversity of glutamate transporter expression and function in the mammalian retina. Amino Acids 19, 53–62. Rauen, T., Rothstein, J.D., and Wa¨ssle, H. (1996). Differential expression of three glutamate transporter subtypes in the rat retina. Cell Tissue Res. 286, 325–336. Rauen, T., Taylor, W.R., Kuhlbrodt, K., and Wiessner, M. (1998). High-affinity glutamate transporters in the rat retina: a major role of the glial glutamate transporter GLAST-1 in transmitter clearance. Cell Tissue Res. 291, 19–31. Reichelt, W., and Kno¨pfel, T. (2002). Glutamate uptake controls expression of a slow postsynaptic current mediated by mGluRs in cerebellar Purkinje cells. J. Neurophysiol. 87, 1974–1980. Rossi, D.J., and Hamann, M. (1998). Spillover-mediated transmission at inhibitory synapses promoted by high affinity a6 subunit GABAA receptors and glomerular geometry. Neuron 20, 783–795. Rothstein, J.D., Martin, L., Levey, A.I., Dykes-Hoberg, M., Jin, L., Wu, D., Nash, N., and Kuncl, R.W. (1994). Localization of neuronal and glial glutamate transporters. Neuron 13, 713–725. Rusakov, D.A. (2001). The role of perisynaptic glial sheaths in glutamate spillover and extracellular Ca2+ depletion. Biophys. J. 81, 1947–1959.

Neuron 74

Rusakov, D.A., and Kullmann, D.M. (1998). Extrasynaptic glutamate diffusion in the hippocampus: ultrastructual constraints, uptake, and receptor activation. J. Neurosci. 18, 3158–3170. Sampath, A.P., and Rieke, F. (2004). Selective transmission of single photon responses by saturation at the rod-to-rod bipolar synapse. Neuron 41, 431–443. Shiells, R.A., and Falk, G. (1990). Glutamate receptors of rod bipolar cells are linked to a cyclic GMP cascade via a G-protein. Proc. Biol. Sci. 242, 91–94. Shiells, R.A., and Falk, G. (2000). Activation of Ca2+-calmodulin kinase II induces desensitization by background light in dogfish retinal ‘on’ bipolar cells. J. Physiol. 528, 327–338. Shimamoto, K., Lebrun, B., Yasuda-Kamatani, Y., Sakaitani, M., Shigeri, Y., Yumoto, N., and Nakajima, T. (1998). DL-threo-b-benzyloxyaspartate, a potent blocker of excitatory amino acid transporters. Mol. Pharmacol. 53, 195–201. Slaughter, M.M., and Miller, R.F. (1981). 2-amino-4-phosphonobutyric acid: a new pharmacological tool for retina research. Science 211, 182–185. Stryer, L. (1986). Cyclic GMP cascade of vision. Annu. Rev. Neurosci. 9, 87–119. Thoreson, W.B., Rabl, K., Townes-Anderson, E., and Heidelberger, R. (2004). A highly Ca2+-sensitive pool of vesicles contributes to linearity at the rod photoreceptor ribbon synapse. Neuron 42, 595–605. Tsukamoto, Y., Morigiwa, K., Ueda, M., and Sterling, P. (2001). Microcircuits for night vision in mouse retina. J. Neurosci. 21, 8616–8623. Vandenbranden, C.A., Verweij, J., Kamermans, M., Muller, L.J., Ruiter, J.M., Vrensen, G.F., and Spekreijse, H. (1996). Clearance of neurotransmitter from the cone synaptic cleft in goldfish retina. Vision Res. 36, 3859–3874. Vandenbranden, C.A., Yazulla, S., Studholme, K.M., Kamphuis, W., and Mamermans, M. (2000). Immunocytochemical localization of the glutamate transporter GLT-1 in goldfish (Carassius auratus) retina. J. Comp. Neurol. 423, 440–451. Vardi, N., Duvoisin, R., Wu, G., and Sterling, P. (2000). Localization of mGluR6 to dendrites of ON bipolar cells in primate retina. J. Comp. Neurol. 423, 402–412. von Gersdorff, H. (2001). Synaptic ribbons: versatile signal transducers. Neuron 29, 7–10. von Gersdorff, H., and Matthews, G. (1994). Dynamics of synaptic vesicle fusion and membrane retrieval in synaptic terminals. Nature 367, 735–739. von Gersdorff, H., Vardi, E., Matthews, G., and Sterling, P. (1996). Evidence that vesicles on the synaptic ribbon of retinal bipolar neurons can be rapidly released. Neuron 16, 1221–1227. Wadiche, J.I., and Kavanaugh, M.P. (1998). Macroscopic and microscopic properties of a cloned glutamate transporter/chloride channel. J. Neurosci. 18, 7650–7661. Wadiche, J.I., Arriza, J.L., Amara, S.G., and Kavanaugh, M.P. (1995). Kinetics of a human glutamate transporter. Neuron 14, 1019–1027. Watase, K., Hashimoto, K., Kano, M., Yamada, K., Watanabe, M., Inoue, Y., Okuyama, S., Sakagawa, T., Ogawa, S., Kawashima, N., et al. (1998). Motor discoordination and increased susceptibility to cerebellar injury in GLAST mutant mice. Eur. J. Neurosci. 10, 976–988. Wiessner, M., Fletcher, E.L., Fischer, F., and Rauen, T. (2002). Localization and possible function of the glutamate transporter, EAAC1, in the rat retina. Cell Tissue Res. 310, 31–40. Wilson, M., Tessier-Lavigne, M., and Attwell, D. (1987). Noise analysis predicts at least four states for channels closed by glutamate. Biophys. J. 52, 955–960. Winkler, B.S., Kapusta-Bruneau, N., Arnold, M.J., and Green, D.G. (1999). Effects of inhibiting glutamine synthetase and blocking glutamate uptake on b-wave generation in the isolated rat retina. Vis. Neurosci. 16, 345–353.