EAAT5 glutamate transporter rapidly binds glutamate with micromolar affinity in mouse rods

Continual maintenance of synaptic transmission requires continual resupply of neurotransmitters. Glutamatergic synapses rely on two sources of resupply (reviewed by Marx et al. [2015]). One is the supply of glutamine from the surrounding glia. Glutamine exported from glia is taken into neurons where it is converted to glutamate by glutaminase in the presynaptic terminal. The other major source is the reuptake of glutamate by the presynaptic neuron. The amount of glutamate generated by de novo synthesis is thought to be negligible. In rod photoreceptor cells of the vertebrate retina, the major reuptake molecules are type 2 and 5 excitatory amino acid transporters (EAAT2 and EAAT5; Eliasof et al., 1998; Lukasiewcz et al., 2021; Tang et al., 2022). EAAT5 transporters are clustered just beneath synaptic ribbons in the presynaptic active zone of rods and cones (Gehlen et al., 2021; Pow and Barnett, 2000; Rauen et al., 1996, 2004), perfectly positioned to capture recently released glutamate. EAAT2 transporters are also present in photoreceptor terminals (Tang et al., 2022). Nevertheless, it has been suggested that most of the glutamate released from photoreceptors is retrieved by EAAT1 in the surrounding Muller cells (Harada et al., 1998; Niklaus et al., 2017; Pow et al., 2000; Rauen et al., 1998). Genetic elimination of EAAT2 or blocking it with the selective inhibitor DHKA did not inhibit ERG b-waves in mouse retina (Tse et al., 2014) and caused only modest changes to light responses of horizontal cells and OFF bipolar cells in salamander retina (Rowan et al., 2010; Yang and Wu, 1997). Genetic elimination of EAAT5 from mouse retina also did not reduce ERG b-waves but did slow the kinetics of rod-driven light responses in bipolar and ganglion cells (Gehlen et al., 2021). EAAT5 and the retinal EAAT2 isoform (mGLT-1c) exhibit large thermodynamically uncoupled anion currents (Schneider et al., 2014). Effects of presynaptic EAAT2 and EAAT5 in photoreceptors on the kinetics of downstream neurons may therefore be due more to the anion currents that accompany glutamate binding than to glutamate uptake by these transporters (Arriza et al., 1997; Bligard et al., 2020; Fairman et al., 1995; Kovermann et al., 2022; Veruki et al., 2006; Wadiche et al., 1995a). Consistent with this proposal, heterologously expressed human EAAT5 showed activation kinetics that are far too slow (fast τ_on = 30 ms; slow τ_on = 200 ms) to support rapid glutamate retrieval (Gameiro et al., 2011). On the other hand, light response kinetics of rod bipolar cells were also unchanged by genetic elimination or pharmacological inhibition of Muller cell EAAT1 transporters (Hasegawa et al., 2006). Computer simulations of rod bipolar cell responses also show that a dense concentration of EAAT5 transporters would allow rods to recapture almost all of the glutamate that they release (Hasegawa et al., 2006).

To determine whether EAAT5 in rods exhibits the slow kinetics seen with heterologous expression, we recorded glutamate transporter currents in rod photoreceptor cells from mouse retina. We performed whole-cell patch clamp recordings using a pipette solution that contained the anion thiocayanate (SCN⁻) to enhance EAAT5 anion currents. We analyzed both spontaneous vesicle release events and glutamate transporter anion currents (I_A(glu)) evoked by rapid elevation of glutamate via photolytic uncaging of MNI-glutamate. Our results showed that, in contrast with heterologously expressed transporters, EAAT5 transporters in rods activate rapidly with time constants of 2–3 ms. Rapid-response onsets following vesicle release events and direct application of glutamate were previously reported in ground squirrel cones that also express EAAT5 at their terminals (Grabner et al., 2023; Szmajda and Devries, 2011). Evidence for rapid activation and an EC₅₀ for glutamate of ∼10 μM suggest that EAAT5 would be capable of retrieving synaptic glutamate released by rods. However, our results also suggest that EAAT5 transporters are saturated by the simultaneous release of only a few vesicles. Transporter saturation also occurs during glutamate release at the cone synapse (Grabner et al., 2023; Szmajda and Devries, 2011). This suggests that while uptake by rods may be able to recycle much of the glutamate released at the slow rates normally seen in darkness (Hays et al., 2020, 2021), other mechanisms (e.g., Muller cell uptake) likely contribute with stronger stimulation and release of a larger fraction of the synaptic vesicle pool.

We used mice of both sexes aged between 4 and 8 wk. Euthanasia was performed by CO₂ asphyxiation by cervical dislocation in accordance with AVMA Guidelines for the Euthanasia of Animals. Animal care and handling protocols were approved by the University of Nebraska Medical Center Institutional Animal Care and Use Committee. Most experiments were conducted using C56BL6J mice. A few experiments were conducted with mice in which the exocytotic Ca²⁺ sensor, synaptotagmin 1 (Syt1), was selectively removed from the rods. As described elsewhere (Grassmeyer et al., 2019; Quadros et al., 2017), Syt1^fl/fl mice have loxP sites flanking exon 6 of Syt1 and were crossed with Rho-iCre mice that selectively express cre-recombinase in rods (Li et al., 2005).

Whole-cell recordings of rods were obtained using flat-mount preparations of the isolated retina. Eyes were enucleated after euthanizing the mouse and placed in Ames’ medium (RRID: SCR_013653; US Biological) bubbled with 95% O₂/5% CO₂. The cornea was punctured with a scalpel and the anterior segment was removed. The retina was isolated after cutting optic nerve attachments. We then made three or four fine cuts at opposite poles and flattened the retina onto a glass slide in the perfusion chamber with photoreceptors facing up. The retina was anchored in place with a brain slice harp (cat. no. 64-0250; Warner Instruments). To expose rod inner segments in flat-mount retina, we gently touched the photoreceptors with a piece of nitrocellulose filter paper and then removed it to pull away adherent outer segments. The perfusion chamber was placed on an upright fixed-stage microscope (E600FN; Nikon) equipped with a 60× water immersion, long-working distance fluorescence objective (1.0 NA) suitable for UV. Tissue was superfused with room temperature Ames’ solution bubbled with 95% O₂/5% CO₂ at ∼1 ml/min. In a few experiments, we used an in-line heater to elevate the temperature to 35°C, monitored by a temperature probe (BAT-12; Physitemp) placed in the chamber.

Patch-recording electrodes were pulled on a Narishige PP-830 vertical puller using borosilicate glass pipettes (1.2 mm outer diameter, 0.9 mm inner diameter with internal filament; World Precision Instruments). Pipettes had tip diameters of 1–2 μm and resistances of 10–15 MΩ. Rod inner segments were targeted with positive pressure using recording electrodes mounted on Huxley-Wall or motorized micromanipulators (MP225; Sutter Instruments). Cones were distinguished from rods by their larger membrane capacitance (6 vs. 3 pF), greater spontaneous release rate, and larger Ca²⁺ currents (I_Ca; Grassmeyer et al., 2019).

Rod ribbons are surrounded by the glutamate transporter EAAT5 (Arriza et al., 1997; Eliasof et al., 1998; Hasegawa et al., 2006), and glutamate reuptake into rods by these transporters activates a large, anion current (I_A(glu); Arriza et al., 1997; Grant and Werblin, 1996; Schneider et al., 2014). I_A(glu) is thermodynamically uncoupled from the transport process (Machtens et al., 2015). Glutamate transporter anion currents can be observed in rods using Cl⁻ as the principal anion (Hays et al., 2020) but are enhanced by replacing Cl⁻ with a more permeable anion like thiocyanate (SCN⁻) in the patch pipette (Eliasof and Jahr, 1996). The intracellular pipette solution for these experiments contained (in mM) 120 KSCN, 10 TEA-Cl, 10 HEPES, 1 CaCl₂, 1 MgCl₂, 0.5 Na-GTP, 5 Mg-ATP, 5 EGTA, and 5 phospho-creatine, pH 7.3.

Whole-cell recordings were performed using either an Axopatch 200B amplifier (Molecular Devices) with signals digitized by a DigiData 1550 interface (Molecular Devices) using PClamp 10 software (RRID: SCR_011323) or Heka EPC-10 amplifier and Patchmaster software (RRID: SCR_018399; Lambrecht). Currents were acquired with filtering at 3 kHz. Responses shown in the figures were subsequently filtered by Clampfit at 1 kHz. Voltages were not corrected for a liquid junction potential of 3.9 mV.

For experiments with caged glutamate, 4-methoxy-7-nitroindolinyl- L-glutamate (MNI glutamate; 0.5–3 mM; Tocris Bioscience, and HelloBio) was bath applied and photolyzed by flashes of UV light derived from a Xenon arc flash lamp (JML-C2 Flash Lamp System, Rapp OptoElectric). Light flashes (220-μm diameter) were delivered through a 1,250-μm diameter quartz fiber optic via the epifluorescence port of the microscope and centered over the rod. Flash intensities were measured by the manufacturer using a joulemeter. Flash durations were ∼0.6 ms for intensities up to 6.7 mJ, ∼0.8 ms for 12.5 mJ, and ∼1 ms for the two brightest flashes. Photolytic uncaging of MNI-glutamate occurs with time constants of <1 ms (Canepari et al., 2001). Other chemical reagents were obtained from Sigma-Aldrich unless otherwise indicated.

To measure glutamate changes evoked by flash photolysis with different intensity flashes, we used a cell line of human lens epithelial cells (hLECs). In this cell line (SRA01/04), hLECs were immortalized with SV40 and maintained as described previously (Chhunchha et al., 2022). 24 h after plating in a 12-well plate, hLECs were treated with AAV2.2 virus coding for the fluorescent glutamate sensor iGluSnfR (Marvin et al., 2013, 2018; pAAV.CAG.SF-iGluSnFR.A184S; RRID: Addgene_106198) for 24 h. 2 d later, cells were replated onto coverslips for imaging. Infection efficiency varied among cells. We imaged one or two cells at a time. To measure iGluSnFr fluorescence as a function of glutamate levels, we bath applied different concentrations of glutamate. Glutamate-free solutions prepared in an Ames’ salt solution were used for 0 glutamate measurements. In six cells, we supplemented the existing glutamate in Ames’ medium (7 μM) with 30, 100, 300, and 1,000 mM glutamate. We also tested 0, 7, 37, 107, 307, and 1,007 mM glutamate prepared in an Ames’ salts solution (n = 9 cells). We did not see any significant difference in results obtained with these two approaches, so we combined them for analysis. We bath applied each solution for at least 4 min to ensure a stable bath concentration. We then acquired a z-stack through the entire cell thickness, using dim laser illumination along with camera binning of 4 × 4 or 8 × 8 to minimize bleaching. After subtracting baseline fluorescence measured in a cell-free region of the image, we calculated the change in cell fluorescence (ΔF) relative to the fluorescence F measured in 0 glutamate in the same cell. We fit the rise in ΔF/F as a function of glutamate concentration with a sigmoidal dose/response function. The maximum value for ΔF/F averaged 0.8 ± 0.38 (SD; n = 15).

To measure the rise in glutamate levels evoked by flash photolysis of MNI-glutamate, we measured fluorescence in hLECs transfected with iGluSnFr using a 35-ms acquisition with 8 × 8 camera binning to minimize confocal laser intensity. Three frames were acquired before flash uncaging to measure baseline fluorescence (F) in the cell in comparison to the change in fluorescence measured immediately after uncaging (ΔF/F).

Confocal imaging was performed using NIS Elements software (RRID: SCR_014329; Nikon) and a spinning disk confocal microscope that consisted of a laser confocal scan head (Perkin Elmer Ultraview LCI) equipped with a cooled CCD camera (RRID: SCR_017105; Hamamatsu Orca ER) mounted on a Nikon E600FN microscope. Fluorescent excitation was delivered from an argon/krypton laser at 488 nm, and emission was collected at 520 nm. Filters were controlled using a Sutter Lambda 10–2 filter wheel and controller. The objective (water immersion, 60×, 1.2 NA) was controlled with an E662 z-axis controller (Physik Instrumente). Images were analyzed using Nikon NIS Elements.

Analysis and data visualization were done using Clampfit 10 and GraphPad Prism 9 (RRID: SCR_002798; GraphPad Prism) software. Roughly equal numbers of male and female mice were used for these experiments. The criterion for statistical significance was set at α = 0.05. Unless otherwise noted, values in the text are reported as mean ± SD. Error bars in the figures show 95% confidence intervals.

Using whole-cell patch clamp recordings from rod photoreceptor cells in wholemount preparations of mouse retina, we recorded anion currents activated by glutamate binding to presynaptic glutamate transporters in the same rod. We voltage-clamped rods at −70 mV and recorded I_A(glu) evoked by the spontaneous release of glutamate as well as I_A(glu) evoked by photolytic uncaging of glutamate. In uncaging experiments, we elevated glutamate rapidly and evenly throughout the synaptic cleft using a flash UV lamp to uncage glutamate from MNI-glutamate (0.5–3 mM). Applying a bright flash in the absence of MNI-glutamate evoked no response other than a flash artifact (Fig. 1 A; n = 4 rods). As illustrated by the example in Fig. 1, B and C, photolytic uncaging of MNI-glutamate evoked inward glutamate transporter currents that showed a rapid increase and slower decline. We characterized the time course by fitting both the rise and decline in I_A(glu) with single exponentials.

The amplitude of I_A(glu) evoked by photolytic uncaging of glutamate increased as a function of flash lamp intensity. Fig. 2 A shows a series of responses in a single rod to flashes of differing intensity. Spontaneous I_A(glu) release events (asterisks, Fig. 2 A) can also be seen. Flash intensity is given at the left and the sequence of flashes is given at the right. Bright 17.6-mJ flashes evoked similar responses at the beginning and end of this sequence. We excluded measurements if there was evidence of significant rundown. When possible, we repeated the sequence in both directions. We saw no differences in response amplitude using 0.5 or 1 mM MNI-glutamate and so combined these data. The brightest flash of 19 mJ evoked inward currents that averaged 12.5 ± 4.00 pA (n = 15 events in 12 rods). The increase in amplitude with flash intensity diminished at higher intensities, showing evidence for response saturation. Consistent with this, photolytic uncaging with bright 17.6-mJ flashes using a higher concentration of 3 mM MNI-glutamate evoked similar responses as 0.5 and 1 mM MNI-glutamate (open triangles; Fig. 2, B–D).

To ensure that responses to UV flashes were due to glutamate uncaging and not incidental glutamate release by rods, we tested rods from mice in which the principal exocytotic Ca²⁺ sensor, synaptotagmin 1 (Syt1; n = 6 rods) was selectively eliminated (Grassmeyer et al., 2019). Genetic elimination of Syt1 from rods abolishes fast, synchronous release of the readily releasable pool of vesicles in rods (Grassmeyer et al., 2019; Mesnard et al., 2022). Responses to flash photolysis of MNI-glutamate in these rods did not differ significantly in amplitude from control rods with intact synaptic release capability, indicating that I_A(glu) responses were principally due to glutamate uncaging, not incidental presynaptic release.

As illustrated in Fig. 1, we fit the rise time in the currents evoked by glutamate uncaging with a single exponential. Fig. 2 B plots time constants for the rise in I_A(glu) as a function of flash intensity. Rise times averaged ∼3 ms when evoked by weak flashes and shortened to ∼2 ms at the highest flash intensities (Fig. 2 B). Rise times for I_A(glu) from rods in situ are thus 10-fold faster than those reported for heterologously expressed EAAT5 (fast time constant = 30 ms; Gameiro et al., 2011).

Exponential fits to the decline in I_A(glu) yielded time constants ranging from 50 to 150 ms (Fig. 2 C). While rise times shortened at higher glutamate levels, decay times lengthened with increasing flash intensity (Fig. 2 D). Like the rise in amplitude, the increase in decay times showed evidence for saturation at higher flash intensities.

I_A(glu) events evoked by the spontaneous release of glutamate-filled vesicles in rods showed similar rise and decline kinetics as I_A(glu) events evoked by uncaging of MNI-glutamate with weak flashes. This can be seen by comparing spontaneous events (asterisks) with small flash-evoked responses in Fig. 2 A. Fig. 3 A shows an example waveform averaged from eight spontaneous events in a single rod. The best fit time constant to the initial inward current in this example was 3.2 ms and the best fit to the decay was obtained with a time constant of 45.7 ms. Overall, spontaneous events averaged 5.2 ± 2.02 pA in amplitude (n = 407 events from 20 rods). When the amplitude histogram was fit with a two-component Gaussian function, the mean amplitude of the first component was slightly smaller: 4.83 ± 1.51 pA, with 90% of events contained in the first component and 10% in the second component (Fig. 3 A). Simultaneous multivesicular release occurs less often in mouse rods than salamander rods where multivesicular release accounts for ∼30% of the vesicles during spontaneous release (Hays et al., 2021).

Time constants for rise times of spontaneous events averaged 2.75 ± 1.00 ms (n = 173 events, 13 rods) and decay time constants averaged 44.4 ± 16.2 ms (n = 334 events, 18 rods). Rise times did not vary significantly with event amplitude (Fig. 3 B). In the overall sample, decay time constants lengthened slightly with larger events (slope significance = 0.045, Fig. 3 C) but none of the regression line slopes for individual cells differed significantly from zero.

Evoked and spontaneous glutamate transporter anion currents were both completely blocked by bath application of the transport inhibitor, TBOA (0.3 mM; Grassmeyer et al., 2019; Hays et al., 2021). A more potent inhibitor TFB-TBOA (3 μM; Tocris) also completely eliminated both spontaneous and evoked I_A(glu) responses (n = 5 rods). Rods also possess EAAT2 (Tang et al., 2022), and we blocked this transporter with the selective inhibitor DHKA (0.2 mM; Arriza et al., 1994). With DHKA, we saw no significant changes in decay time constants for spontaneous events or in the amplitude of either spontaneous or depolarization-evoked currents (Fig. 4 A; n = 6 rods). Rod terminals are enveloped by Muller glia that possesses EAAT1 (Burris et al., 2002; Pow et al., 2000; Rauen et al., 1998). We tested a selective inhibitor of this transporter (UCPH101, 25 μM; n = 5; Jensen et al., 2009) and saw no significant changes in the amplitude of either spontaneous or evoked release events or in the decay kinetics of spontaneous events (Fig. 4 B). Thus, while EAAT1 and EAAT2 transporters may assist with glutamate retrieval, the anion currents in rods arise solely from EAAT5.

Most recordings were done at room temperature, but we recorded spontaneous events in some cells at 35°C. Fig. 5 A shows example traces from a rod recorded at room temperature (23°C) and 35°. As illustrated by the average waveforms from this cell in Fig. 5 B, spontaneous I_A(glu) events showed a much more rapid rise and decline at 35°C. Summary data show that rise and decay kinetics of spontaneous events were roughly twice as fast at 35°C compared with room temperature, with Q₁₀ values for rise and decline times being 1.8 and 2.4, respectively (Fig. 5 C), similar to the Q₁₀ for Ca²⁺ current amplitude in rods (Hays et al., 2020). This sensitivity to temperature suggests that response kinetics are not strongly shaped by diffusion (which would vary little between these temperatures) but instead represents intrinsic properties of EAAT5.

To measure glutamate changes produced by uncaging MNI-glutamate, we used an AAV vector to express iGluSnFr in cultured human lens epithelial cells. Cells were replated onto coverslips 48 h after transfection and imaged 1 d later on a spinning disk confocal microscope. As illustrated by the fluorescence image in Fig. 6 A, iGluSnFr fluorescence varied among cells. We calibrated the glutamate sensitivity of iGluSnfr in individual hLECs by measuring cell fluorescence at different bath concentrations of glutamate from 0 to 1.07 mM. We converted these fluorescence changes to ΔF/F, where F was defined as the fluorescence measured in a glutamate-free solution in the same cell. Fitting the rise in ΔF/F as a function of bath glutamate concentration using a logistic function with a slope of 1 yielded an EC₅₀ for iGluSnFr in this preparation of 9 μM (Fig. 6 B).

We next used hLECs transfected with iGluSnfr to measure fluorescence changes evoked by flash photolysis of MNI-glutamate. We acquired data at 35 ms/frame and calculated the change in fluorescence evoked by flashes of varying intensity. Responses were evoked over a range of flash intensities using 0.5 and 1 mM MNI glutamate. We calculated ΔF/F as a function of flash voltage (Fig. 6 C), with F defined as the baseline fluorescence intensity immediately preceding the flash. We then used the best fit function that described glutamate sensitivity of iGluSnfr (Fig. 6 B) to convert ΔF/F values to glutamate levels (Fig. 6 D).

Replacing ΔF/F with [Glu] allowed us to estimate the glutamate concentration evoked by uncaging glutamate at different flash intensities (Fig. 6 D). Using this relationship, we fit the amplitude of flash-evoked I_A(glu) as a function of glutamate and found an EC₅₀ of 10.5 μM (Fig. 7 A). We also plotted rise and decay time constants as a function of glutamate levels evoked by uncaging of MNI-glutamate. These yielded EC₅₀ values of 16.4 and 8.0 μM, respectively (Fig. 7, B and C). The best fits were obtained with a slope of 1. Fits to the increase in amplitude showed the smallest standard error in best fit values for EC₅₀, but all three were consistent with an EC₅₀ of ∼10 μM.

To assess contributions from diffusion to these measurements, we measured the rate at which iGluSnFr fluorescence declined within a 20-μm diameter circular region of interest following photolytic uncaging of glutamate using a bright 17.5-mJ flash. In addition to cultured lens epithelial cells, we also made measurements in the inner plexiform layer of retinal slices 2 wk after intravitreal injection of the viral construct for iGluSnFr. In both preparations, iGluSnFr fluorescence declined within this small region with a time constant of about 1 s (hLECs: n = 4, 0.92 ± 0.11 s; retina: n = 7, 1.1 ± 0.099 s), much slower than the decay in I_A(glu) evoked by glutamate uncaging.

The present results show that EAAT5 glutamate transporters in rods can be activated by glutamate 10-fold faster (tau_on = 2–3 ms) than EAAT5 transporters expressed in HEK293 cells (fast tau_on = 30 ms, slow tau_on = 200 ms; Gameiro et al., 2011). These differences may be due to interactions with neighboring proteins as well as differences in the local lipid and ionic environment that can shape transporter performance (Robinson, 2006). The rapid kinetics suggest that EAAT5 is capable of rapidly binding glutamate to help keep synaptic glutamate levels near the EC₅₀ for the rod bipolar cell metabotropic glutamate receptor, mGluR6. As discussed below, the anion current associated with glutamate binding to this transporter may also impact synaptic function.

Vesicle fusion results in a rapid release of glutamate, so the EC₅₀ determined from fast uncaging provides insight into how local transporters will respond to fast vesicular release. However, it should be noted that the EC₅₀ measured from a 1-ms pulse of glutamate may differ from the steady state EC₅₀. Transporters have complex kinetic properties, and differences could result if the glutamate binding rates are slow, especially at subsaturating glutamate concentrations. Onset time constants as long as 25–50 ms can be seen at submicromolar glutamate concentrations when glutamate is rapidly applied to cone transporters at low concentrations during long steps (Grabner et al., 2023).

UV flash photolysis releases glutamate from MNI-glutamate in <1 ms (Canepari et al., 2001). Glutamate would be uncaged evenly throughout the region illuminated by the UV flash, and so diffusion should have little role in limiting activation kinetics following UV uncaging. I_A(glu) events that arise from the spontaneous release of glutamate-filled vesicles showed a similar rise time as events evoked by glutamate uncaging. Confocal images of EAAT5 immunolabeling show that it densely clustered on rod synaptic membranes adjacent to dendritic tips of rod bipolar cells (Gehlen et al., 2021; Tang et al., 2022). Simulations of glutamate diffusion within an invaginating rod synapse suggest that glutamate can reach its peak concentration at rod bipolar cell dendrites in <0.1 ms following release of a synaptic vesicle (Rao-Mirotznik et al., 1998). This suggests that diffusion of glutamate within the synapse is not likely to significantly limit the rise time for spontaneous I_A(glu) events.

The decay in I_A(glu) after glutamate uncaging is also not likely to be shaped significantly by glutamate diffusion. Using cultured epithelial cells and retinal slices, iGluSnFr fluorescence was measured in a 20-μm diameter region of interest at the center of the 220-mm diameter UV uncaging spot. Following a bright uncaging flash, glutamate measured in this region declined with time constants of ∼1 s, much slower than the decline in I_A(glu) in rods (soma diameter of 4–7 μm). Spontaneous I_A(glu) events also declined twice as fast when the temperature was increased from room temperature (∼296°K) to 35°C (308°K). This small increase in absolute temperature predicts an increase in the diffusion coefficient of 4%. Also consistent with the conclusion that decay time constants are more a function of transporter kinetics than diffusion, the cycle time for glutamate transport by EAAT2 has been estimated at ∼70 ms (Wadiche et al., 1995b).

We measured glutamate levels produced by flash photolysis of MNI glutamate using cultured hLECs. The concentration-dependent changes in amplitude, rise time, and decay time yielded EC₅₀ values for EAAT5 anion currents of ∼10 μM. By comparison, the EC50 found with heterologous expression of EAAT5 from mice and humans wase 25 and 61 μM, respectively (Gameiro et al., 2011; Schneider et al., 2014). This affinity of EAAT5 would help keep cleft glutamate levels in the low micromolar range The metabotropic glutamate receptors used by rod bipolar cells, mGluR6, have an EC₅₀ of 15–20 μM (Schoepp et al., 1999), close to that of EAAT5. Keeping glutamate levels in this range may help keep mGluR6 transduction machinery close to saturation. This has been shown to be important for the non-linear thresholding mechanism at rod synapses whereby small noisy fluctuations in rod membrane potential fail to generate detectable post-synaptic responses but larger rod light responses can evoke reliable responses in rod bipolar cells (Field and Rieke, 2002; Sampath and Rieke, 2004).

Deleting EAAT5 from the retina impairs temporal resolution, particularly at mesopic light levels (Gehlen et al., 2021). These effects involve rod bipolar cell synapses where EAAT5 can provide feedback inhibition and regulate output (Bligard et al., 2020; Tang et al., 2022; Veruki et al., 2006; Wersinger et al., 2006) but may also involve rod synapses. E_Cl is positive for the resting potential in amphibian rods (Thoreson et al., 2003) and mammalian cones (Szmajda and Devries, 2011). If this is also true in mammalian rods, then EAAT5 anion currents would be expected to stimulate an efflux of chloride. The consequent reduction in intracellular chloride would depolarize rods but could also directly inhibit CaV1.4 calcium currents (Rabl et al., 2003).

The maximum amplitude of I_A(glu) evoked by glutamate uncaging was 12.5 pA, similar to the maximum amplitude of depolarization-evoked I_A(glu) in rods (Mesnard et al., 2022). By comparison, single vesicle I_A(glu) events averaged 4.8 pA, suggesting that EAAT5 can be saturated by glutamate from only a few vesicles. Membrane capacitance measurements indicate there are ∼100 vesicles in the readily releasable pool of mouse rods that can be released within a few milliseconds by strong depolarizing stimulation (Grabner and Moser, 2021; Mesnard et al., 2022). However, rods rarely release this entire pool at once. At their normal resting potential of −40 mV in darkness, rods release vesicles at a rate of ∼12 vesicles/s (at 35°C; Hays et al., 2020). If we assume that the rate of I_A(glu) decline for single-vesicle release events is limited by the rate of glutamate uptake, then a decay time constant of 22 ms at 35°C suggests a maximum possible turnover rate of 45 glutamate molecules/sec. The inner diameter of synaptic vesicles in rods predicts 2,000–3,000 glutamate molecules/vesicle (Rao-Mirotznik et al., 1998). Retrieval of glutamate released at 12 vesicles/s might therefore require only 500–800 EAAT5 transporters. Thus, relatively small numbers of transporters may be sufficient to recycle much of the glutamate released at normally modest rates in darkness. This is consistent with evidence that EAAT5 plays a significant role in shaping rod bipolar cell light responses (Gehlen et al., 2021; Hasegawa et al., 2006; Tang et al., 2022). However, evidence for transporter saturation also supports findings that Muller glia provide an important supplemental mechanism for retrieving additional glutamate released with stronger stimulation (Burris et al., 2002; Pow et al., 2000; Rauen et al., 1998).

When expressed in cultured cells, EAAT5 transporters showed slow activation kinetics inconsistent with meaningful glutamate uptake on a physiological time scale. Our data show that EAAT5 transporters in rods activate with time constants of a few milliseconds and an EC₅₀ of ∼10 μm. The fast kinetics of EAAT5 anion currents in intact rods are well-matched to the kinetics of rod light responses. Consistent with earlier studies suggesting a key role for these transporters (Hasegawa et al., 2006), our results suggest that EAAT5 transporters near ribbon-release sites in rod terminals (Gehlen et al., 2021) would be capable of buffering and retrieving synaptic glutamate at the slow release rates normally encountered in darkness. However, evidence that EAAT5 currents can be saturated by the synchronous release of a few vesicles suggests that other transporters, including EAAT2 in rods (Tang et al., 2022) and EAAT1 in Muller glia (Harada et al., 1998; Niklaus et al., 2017; Pow et al., 2000; Rauen et al., 1998), would need to be recruited under conditions of especially strong release.

Data are available from the corresponding author upon reasonable request.

Joseph A. Mindell served as editor.

The authors thank Cody Barta for laboratory support.

Funding was provided by National Institutes of Health grants EY10542 and EY32396 to W.B. Thoreson.

Author contributions: B. Chhunchha cultured and transfected human LECs. W.B. Thoreson designed, conducted, and analyzed all other experiments. W.B. Thoreson wrote the manuscript. Both authors approved the final submission.