Abstract
G-protein βγ subunits (Gβγ) interact with presynaptic proteins and regulate neurotransmitter release downstream of Ca2+ influx. To accomplish their roles in sensory signaling, photoreceptor synapses use specialized presynaptic proteins that support neurotransmission at active zone structures known as ribbons. While several G-protein coupled receptors (GPCRs) influence synaptic transmission at ribbon synapses of cones and other retinal neurons, it is unknown whether Gβγ contributes to these effects. We tested whether activation of one particular GPCR, a metabotropic glutamate receptor (mGluR), can reduce cone synaptic transmission via Gβγ in tiger salamander retinas. In recordings from horizontal cells, we found that an mGluR agonist (L-AP4) reduced cone-driven light responses and mEPSC frequency. In paired recordings of cones and horizontal cells, L-AP4 slightly reduced cone ICa (∼10%) and caused a larger reduction in cone-driven EPSCs (∼30%). Proximity ligation assay revealed direct interactions between SNAP-25 and Gβγ subunits in retinal synaptic layers. Pretreatment with the SNAP-25 cleaving protease BoNT/A inhibited L-AP4 effects on synaptic transmission, as did introduction of a peptide derived from the SNAP-25 C terminus. Introducing Gβγ subunits directly into cones reduced EPSC amplitude. This effect was inhibited by BoNT/A, supporting a role for Gβγ/SNAP-25 interactions. However, the mGluR-dependent reduction in ICa was not mimicked by Gβγ, indicating that this effect was independent of Gβγ. The finding that synaptic transmission at cone ribbon synapses is regulated by Gβγ/SNAP-25 interactions indicates that these mechanisms are shared by conventional and ribbon-type synapses. Gβγ liberated from other photoreceptor GPCRs is also likely to regulate synaptic transmission.
SIGNIFICANCE STATEMENT Dynamic regulation of synaptic transmission by presynaptic G-protein coupled receptors shapes information flow through neural circuits. At the first synapse in the visual system, presynaptic metabotropic glutamate receptors (mGluRs) regulate cone photoreceptor synaptic transmission, although the mechanisms and functional impact of this are unclear. We show that mGluRs regulate light response encoding across the cone synapse, accomplished in part by triggering G-protein βγ subunits (Gβγ) interactions with SNAP-25, a core component of the synaptic vesicle fusion machinery. In addition to revealing a role in visual processing, this provides the first demonstration that Gβγ/SNAP-25 interactions regulate synaptic function at a ribbon-type synapse, contributing to an emerging picture of the ubiquity of Gβγ/SNARE interactions in regulating synaptic transmission throughout the nervous system.
Introduction
Early-stage visual processing in the vertebrate retina relies on the unique synaptic signaling capabilities of rod and cone photoreceptors. Changes in membrane voltage due to photon absorption in photoreceptor outer segments regulate tonic glutamate release at the synaptic terminal. Photoreceptor synapses differ from their conventional counterparts in a number of ways. For instance, exocytosis depends on large planar protein structures known as ribbons that tether vesicles near the presynaptic membrane and ready them for exocytosis (Heidelberger et al., 2005). Rather than relying on the N- and P-type channels common at conventional synapses, photoreceptors use minimally inactivating L-type Ca2+ channels to provide a tonic Ca2+ signal that reflects photoreceptor membrane voltage (Corey et al., 1984; Morgans, 2001). Additionally, they rely on Syntaxin3 and Complexin3/4 rather than Syntaxin1 and Complexin1/2 isoforms used at many other synapses (Morgans et al., 1996; Reim et al., 2005). Photoreceptors also use a Ca2+ sensor of unknown molecular identity with unusually high Ca2+ affinity and low cooperativity that differs from synaptotagmin1/2 found at conventional synapses (Rieke and Schwartz, 1996; Thoreson et al., 2004; Duncan et al., 2010).
Photoreceptor synaptic transmission is under dynamic regulation by a variety of G-protein-coupled receptors (GPCRs) sensitive to ligands, including dopamine, adenosine, endocannabinoids, glutamate, and somatostatin (Straiker et al., 1999; Akopian et al., 2000; Stella and Thoreson, 2000; Hirasawa et al., 2002; Stella et al., 2002, 2007, 2009; Fan and Yazulla, 2004; Hosoi et al., 2005; Struik et al., 2006). These can influence synaptic output by enhancing or reducing Ca2+ influx (Stella and Thoreson, 2000; Stella et al., 2002, 2007, 2009), as well as by actions on other presynaptic currents (Akopian and Witkovsky, 1996; Akopian et al., 2000; Straiker and Sullivan, 2003; Fan and Yazulla, 2004). GPCRs signal to downstream effectors via heterotrimeric G-proteins comprised of an α subunit (Gα) and beta-gamma subunit complex (Gβγ). At synapses, Gβγ can interact directly with P/Q and N-type Ca2+ channels and inwardly rectifying K+ channels to modulate presynaptic Ca2+ signals (Logothetis et al., 1987; Tedford and Zamponi, 2006; Currie, 2010). In addition, Gβγ can also inhibit synaptic vesicle exocytosis downstream of Ca2+ influx by interacting directly with SNARE proteins, most notably SNAP-25, where it competes with Ca2+-dependent synaptotagmin interactions (Blackmer et al., 2001, 2005; Gerachshenko et al., 2005; Yoon et al., 2007; Betke et al., 2012; Wells et al., 2012).
A role for Gβγ/SNARE interactions in regulating synaptic transmission is well established in several systems (Blackmer et al., 2005; Gerachshenko et al., 2005; Delaney et al., 2007; Zhang et al., 2011; Hamid et al., 2014), yet it is unclear whether this occurs at retinal ribbon synapses with their specialized complement of synaptic proteins and distinct signaling properties. In this study, we leverage a powerful set of approaches, including paired whole-cell recordings, whole-cell capacitance recordings, mEPSC and quantal glutamate transporter current analyses, and proximity ligation assay (PLA) to test whether activation of Group III metabotropic glutamate receptors (mGluRs) recruits Gβγ/SNARE interactions to regulate glutamate release at cone ribbon synapses. This was motivated, in part, by work showing that mGluR activation caused an ∼15% reduction in Ca2+ influx yet a 30%–50% decrease in synaptic transmission (Hosoi et al., 2005). Because photoreceptor exocytosis depends linearly on Ca2+ influx (Thoreson et al., 2004; Duncan et al., 2010), a 15% in ICa reduction should also cause a 15% reduction in exocytosis. The mismatch hinted at processes regulating exocytosis downstream of Ca2+ influx. Here, we show that mGluRs modulate cone synaptic transmission by two distinct mechanisms: (1) direct interaction of Gβγ with SNAP-25 and (2) reduction in Ca2+ influx independently of Gβγ. This reveals an important mechanism for modulating output at photoreceptor ribbon synapses, and shows that Gβγ/SNARE interactions are a widespread mechanism capable of regulating transmission at both ribbon-style and conventional synapses. Liberation of Gβγ by activation of other GPCRs, perhaps even including opsins, are also likely to regulate release from photoreceptors in a similar manner.
Materials and Methods
Animals.
We performed experiments using retinas of aquatic tiger salamanders (Ambystoma tigrinum; Charles Sullivan) of either sex. Care and handling protocols were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center. Salamanders were housed on a 12 h light/dark cycle at 4°C–8°C.
Retinal slices.
A video of our retinal slice and electrophysiology protocols has been published previously (Van Hook and Thoreson, 2013). At 1–2 h after the beginning of the animals' dark cycle, salamanders were immersed in MS222 (0.25 g/L) for 15 min. Under room light, they were then decapitated and quickly pithed and the heads were hemisected. Under fiber optic illumination (∼1000 lux), an eye was enucleated and nested on a saline-soaked wad of cotton on a linoleum block. The anterior portion of the eye was removed, and the resulting eyecup was cut into 2–4 pieces, which were placed vitreal side down on a piece of nitrocellulose membrane (5 × 10 mm; type AAWP, 0.8 μm pores; EMD Millipore). The filter paper was submerged in cold amphibian saline, and the sclera was peeled away, leaving the retina adhering to the membrane. The retina was cut into 125 μm slices using a razor blade tissue slicer (Stoelting). The resulting slices were rotated 90 degrees to view the retinal layers and anchored in the recording chamber by embedding the ends of the nitrocellulose membrane in beads of vacuum grease.
Patch-clamp electrophysiology.
For whole-cell patch-clamp recordings, the recording chamber with slices was transferred to the stage of an upright fixed-stage microscope (Nikon E600FN or Olympus BX51WI). The slices were continuously superfused with an oxygenated amphibian saline solution containing the following (in mm): 116 NaCl, 2.5 KCl, 1.8 CaCl2, 0.5 MgCl2 5 glucose, and 10 HEPES. For experiments testing the Ca2+ sensitivity of release following BoNT/A treatment, we prepared an amphibian saline solution in which [CaCl2] was reduced to 1.0 mm and [MgCl2] was increased to 1.3 mm to maintain overall divalent concentration. The pH of amphibian saline solutions was adjusted to 7.8 with NaOH and the osmolality was 240–245 mOsm, as measured with a vapor pressure osmometer (Wescor). All reagents were obtained from Sigma-Aldrich unless specified otherwise.
Patch pipettes were pulled from borosilicate glass (1.2 mm OD, 0.9 mm ID, with an internal filament; World Precision Instruments) using PC-10 or PP-830 vertical pullers (Narishige) or a Sutter P-1000 Flaming/Brown style puller. Patch pipettes had tip diameters of ∼1 micron and resistances of 15–20 mΩ. Synaptic transmission was monitored using paired whole-cell recordings from cones and horizontal cells (HCs).
The standard pipette solution for cones contained the following (in mm): 50 Cs-gluconate, 40 Cs-glutamate, 10 TEA-Cl, 3.5 NaCl, 1 CaCl2, 1 MgCl2, 10 ATP-Mg, 0.5 GTP-Na, 10 HEPES, and 5 EGTA. The addition of glutamate to the cone pipette solution enhances synaptic currents and helps to minimize rundown in whole-cell recordings by supplying glutamate for synaptic vesicles (Bartoletti and Thoreson, 2011). The pipette solution for HC pipettes was identical, except that the 40 mm Cs-glutamate was replaced with an additional 40 mm Cs-gluconate. For perforated patch recordings, we included a combination of gramicidin (5 μg/ml) and amphotericin-B (200 μm) in the pipette solution. Gramicidin was diluted from a 5 mg/ml stock in 95% ethanol and amphotericin-B from a 200 mm stock in DMSO. Pipette solutions with perforating agents were protected from light and prepared fresh every 3 h. Reported voltages were corrected for the measured liquid junction potential of 9 mV.
Synaptic transmission was monitored using paired whole-cell recordings from cones and HCs. The identity of HCs was confirmed by their morphology, position in the slice, and electrophysiological characteristics (Van Hook and Thoreson, 2013). HCs were voltage-clamped at −69 mV, and cones were voltage-clamped at −79 mV. Synaptic vesicle exocytosis from cones was evoked by a 100 ms step depolarization to −19 mV, which is sufficient to deplete the readily releasable pool of vesicles (Bartoletti et al., 2010). Leak and capacitative currents were subtracted using a P/8 leak subtraction protocol. We measured glutamate transporter currents in cones (Picaud et al., 1995a, b), both spontaneous quantal currents and those evoked by depolarization, using the following pipette solution (in mm): 90 potassium thiocyanate, 10 TEA-Cl, 3.5 NaCl, 1 MgCl2, 10 HEPES, 0.05 EGTA, 10 ATP-Mg, 0.5 GTP-Na. Thiocyanate (SCN−) is a more permeant anion through glutamate transporters (Szmajda and Devries, 2011; Grabner et al., 2015) and thereby boosts the amplitude of individual events.
HC responses to the 2 Hz square wave light stimulus (1100 Lux) were recorded in the presence of a rod-adapting background delivered from a blue LED (470 nm; 3.3 × 1012 photons/cm2/s). For white light stimuli, we controlled voltage across a white light LED using an isolated pulse stimulator (A-M Systems, model 2100). Light intensity was calibrated by measuring power at a variety of voltages and converted to Lux by assuming a luminous efficacy of 100 Lumens/W.
Whole-cell membrane capacitance (Cm) recordings were performed as described previously (Van Hook and Thoreson, 2012) using the “track-in” mode of an Optopatch patch-clamp amplifier (Cairn Research). CsCl (3 mm) and dl-threo-b-benzyloxyaspartic acid (TBOA, 100 μm, Tocris Bioscience) were added to the bath solution to minimize conductance changes due to Ih and glutamate transporters, respectively. Amplifier output of membrane current, Cm, and access resistance were digitized with a Digidata 1322A AD/DA board and acquired with pClamp 10.4 software (Molecular Devices; RRID:SCR_011323). The shafts of patch pipettes were coated with dental wax to reduce stray capacitance, and whole-cell recordings were established by positioning the pipette tip on the ellipsoid. The holding potential was varied sinusoidally at ∼500 Hz, 30 mV peak-to-peak around a holding potential of −79 mV. Synaptic vesicle exocytosis was evoked with a depolarizing step (−79 to −19 mV, 25 ms), and output from the phase-lock amplifier was blanked for 3 ms after the step to avoid the influence of gating charges and to allow time for the phase angle feedback circuitry to settle. Cm measurements were made ∼30 ms following the end of the step. We excluded cells with sizeable poststimulus changes in series resistance.
ct-SNAP-25 peptide synthesis.
We synthesized SNAP-25 193–206 peptides using the ResPep SL peptide synthesizer (Intavis AG) method published previously (Wells et al., 2012). Samples were further purified on a Gilson preparative reversed-phase HPLC system with a reverse-phase C18 column (Luna, 30 × 50 mm; Phenomenex) at 50 ml/min with a 5%–95% acetonitrile in water gradient with 0.1% TFA over 7 min. Samples were then subjected to liquid chromatography/mass spectrometry (6120 LCMS; Agilent Technologies) for purity and mass spectrometric identification of the peptide. HPLC fractions were evaporated to retrieve the peptide of interest.
Drugs and pharmacology.
G-protein β1γ1 subunits were purified from bovine retina as described previously (Mazzoni et al., 1991). Gβ1γ1 was diluted in the cone pipette solution (0.3–2 μm) from a stock (5.5 mg/ml). As a control for Gβ1γ1 experiments, we incubated an aliquot of the Gβ1γ1 stock in a boiling water bath for 20–30 min before diluting it in pipette solution. In other experiments, we included ct-SNAP-25 peptide (D-E-A-N-Q-R-A-T-K-M-L-G-S-G) (Gerachshenko et al., 2005) in the cone pipette solution (1 μm) from a freshly made stock (500 μm in DMSO). We used the selective Group III mGluR agonist L-AP4 (L-(+)-2-amino-4-phosphonobutyric acid) to probe the involvement of mGluRs in regulating cone synaptic transmission. Because the Kd of Group III mGluR's for L-AP4 is reported to vary from ∼1 μm for mGluR4, 6, and 8 at 160–1300 μm for mGluR7 (Schoepp et al., 1999), L-AP4 was used at a working concentration of 100 μm. L-AP4 was diluted in amphibian saline from a 10 mm stock solution made in H2O and bath-applied to the slices. For “vehicle controls,” water was diluted into the amphibian saline and applied to the tissue. Botulinum neurotoxin A (BoNT/A, List Biological Laboratories) was diluted to 20 nm in amphibian saline from a 1 μm stock. Slices were incubated in BoNT/A for 30–60 min before being washed three times with amphibian saline and used for electrophysiological recordings.
Electrophysiology analysis.
Electrophysiology data were analyzed with Clampfit 10.4 (Molecular Devices; RRID:SCR_011323). Evoked EPSC amplitude was measured at the peak of the EPSC. Ca2+ influx was measured as the total Ca2+ charge transfer over the 100 ms step (QCa). ICa voltage dependence was measured by correcting QCa for driving force and fitting with a Boltzman function in GraphPad Prism 4 (RRID:SCR_002798). For measuring the Ca2+ dependence of vesicle exocytosis, we measured EPSCs evoked by steps to a series of test potentials from a holding potential of 61 mV. The average Ca2+ current was extrapolated from a fit of responses to a voltage ramp (−99 to 51 mV, 0.5 mV/ms) by fitting with a Boltzmann function adjusted for the ICa reversal potential (Vrev = 50 mV) to compensate for lingering voltage-gated K+ current at depolarized potentials as follows:
EPSC amplitude was plotted against ICa amplitude, and the data were fit with a linear regression to test for a linear correlation of synaptic vesicle release and Ca2+ influx. Gβ1γ1 dose dependence was assessed by fitting with a sigmoidal dose–response curve. mEPSCs and spontaneous quantal transporter currents were detected and analyzed using Mini Analysis 6.0.7 (Synaptosoft; RRID:SCR_002184). Mini Analysis uses a multiparameter sequential algorithm for detecting mEPSC waveforms rather than a simple amplitude threshold to minimize false-positive detections arising from noise and baseline current fluctuations. Mini Analysis first finds a local maximum and baseline, then, in sequence, calculates amplitude, time to peak, time to decay, and area. After automatic detection, we evaluated detected events by eye. Double peaks were analyzed using an algorithm within Mini Analysis that adjusts the baseline of the second peak by extrapolating the exponential decay of the first peak. Event frequency was calculated for each cell as an average of the instantaneous frequency calculated from interevent intervals of 100–200 individual events. Event amplitude for each cell was calculated as the mean amplitude of all the detected events for each cell. Event kinetics (rise and decay time constants) was calculated from the average of all detected events in a given cell. To quantify the kinetics of endocytosis, we measured the t(50), which was the time to 50% recovery of the measured response from the end of the depolarizing step. Data are presented as mean ± SEM, and statistical significance was evaluated with two-tailed unpaired or paired Student's t tests with p < 0.05 being considered statistically significant. In some instances, the nonparametric Wilcoxon sign-rank test (paired samples) was also used, as indicated.
Immunofluorescence and PLA.
For immunohistochemistry, salamander eyes were enucleated and the anterior portion, including the lens and vitreous were removed. The remaining eyecups were fixed in a 4% PFA solution in PBS for 1 h at room temperature. After a brief wash with PBS, they were cryoprotected in 30% sucrose overnight, and then embedded in Tissue-Tek OCT compound, frozen, and cut into 30 μm sections that were mounted on Superfrost Plus microscope slides (Fisher). Sections were blocked and permeabilized with a solution of 0.5% Triton X-100 (Sigma) and 5.5% donkey serum before being incubated with primary antibodies overnight at 4°C. The following day, sections were washed in PBS, blocked, and incubated with AlexaFluor-488 or 568-conjugated secondary antibodies (1:200 dilution, AB_2534104, RRID:AB_2534102, RRID:AB_2535792, RRID:AB_2535788, RRID:AB_2534017). They were then washed again and coverslipped with Fluoro-Gel (Electron Microscopy Sciences). For immunocytochemistry, a whole salamander retina was isolated in Ca2+-free amphibian saline (in mm as follows: 116 NaCl, 2.5 KCl, 5 MgCl2, 5 glucose, 10 HEPES, pH 7.4) before being incubated in Ca2+-free saline with 0.2 mg/ml cysteine and 30 units/ml papain (Worthington Biochemical) for 30 min. Retinas were washed successively in Ca2+-free saline with 1% BSA (Sigma), Ca2+-free saline alone, and Ca2+-free saline containing 1 mg/ml DNase (Worthington). Retinas were then gently triturated in amphibian saline solution, and cells were allowed to settle onto concanavalin A-coated Superfrost Plus microscope slides for 60 min at 4°C. Cells were then fixed with 4% PFA for 15 min before being washed with PBS. Antibody staining was performed as for retinal sections. SNAP-25 was probed using a mouse monoclonal antibody that is specific for SNAP-25 (EMD Millipore, MAB331, RRID:AB_94805 1:100 dilution). The immunogen was a crude human brain synaptic immunoprecipitated, and the antibody detects a single ∼25–27 kDa band on Western blots (Honer et al., 1993; Thompson et al., 1998; Gottschall et al., 2010). Gβ was detected with a rabbit polyclonal antibody (Santa Cruz Biotechnology T-20, RRID:AB_631542, 1:500 dilution). A goat polyclonal antibody raised against CtBP2 (Santa Cruz Biotechnology E-16, RRID:AB_2086774, 1:150 dilution) was used to detect synaptic ribbons.
To probe for protein–protein interactions of SNAP-25 and Gβγ, we used the Duolink in situ PLA (Sigma) according to the manufacturer's protocol. Briefly, after overnight incubation with the rabbit-anti-Gβ (1:2500 dilution), mouse-anti-SNAP-25 (1:1000 dilution), and goat-anti-CtBP2 (1:150 dilution), sections were processed for CtBP2 immunohistochemistry, as above. They were then washed and incubated with PLA probes diluted 1:5 at 37°C in a humidified chamber. After another wash, the tissue was incubated with the ligation solution and ligase for 30 min at 37°C in a humidified chamber. They were then washed again and incubated with an amplification solution containing polymerase and fluorescently labeled oligonucleotides for 100 min at 37°C in a humidified chamber. Samples were washed a final time, allowed to dry, and coverslipped with the Duolink In Situ Mounting Medium. As negative controls for PLA, the SNAP-25 antibody was omitted from samples run simultaneously as experimental samples.
Images were acquired on a Nikon E600FN microscope equipped with a Yokogawa spinning disk laser confocal scan head (PerkinElmer Ultraview LCA) and a cooled CCD camera (Orca ER) with NIS Elements software (Nikon; RRID:SCR_014329). AlexaFluor-488 was excited with a 488 nm laser and AlexaFluor-568, and the fluorescently labeled PLA oligonucleotides were excited with a 561 nm laser. Displayed images are maximum-intensity projections from a series of z-sections. Images were acquired using identical settings from experimental and control samples. Brightness and contrast were adjusted uniformly within and between the images for experimental and control samples. PLA puncta were counted in thresholded images using the “analyze particles” feature of ImageJ (RRID:SCR_003070), which detects isolated continuous objects in the image. Neither PLA puncta size nor circularity was constrained for detection. Puncta were detected in regions of interest drawn around each of the retinal layers, as identified in bright-field images and by immunohistochemical staining for CtBP2.
Results
To test the ability of mGluR to regulate synaptic transmission from photoreceptors and study the mechanisms underlying this regulation, we made paired whole-cell recordings of cones and second-order HCs in vertical slices of salamander retina (Fig. 1A–C). In these recordings, we measured leak-subtracted Ca2+ currents (ICa) in cones in response to a depolarizing step (−79 to −19 mV) while simultaneously measuring EPSCs in HCs (Vhold = −69 mV). This depolarizing step simulates the depolarization occurring at termination of a bright flash of light. Following application of the Group III mGluR agonist L-AP4 (100 μm), EPSC amplitude was reduced by 28 ± 5% (n = 9; p = 0.007, paired t test). In contrast to effects on the EPSC, ICa was not significantly changed following application of L-AP4 (7 ± 3% increase; p = 0.11, paired t test). In vehicle controls, however, we saw a 5 ± 5% increase in EPSC amplitude and 19 ± 4% increase in QCa. The increase in both EPSC amplitude and Ca2+ influx reflects a gradual runup over the course of these recordings. The runup in ICa is likely due to gradual diffusion of Cs+ and TEA into the terminal, which block outward potassium currents, whereas the runup in EPSC amplitude is due to an increase in quantal amplitude that results from inclusion of glutamate in the cone patch pipette solution (Bartoletti and Thoreson, 2011). Correcting for this runup by subtracting 5% from the relative EPSC amplitude change and 19% from the QCa change gives an estimated ∼33% reduction in EPSC amplitude and an estimated ∼12% reduction in Ca2+ influx following application of L-AP4.
Group III mGluR activation reduces cone synaptic transmission. A, Paired whole-cell recording from a cone and a postsynaptic HC in a vertical slice from salamander retina. In control conditions (black traces), depolarization of the cone (−79 to −19 mV, 100 ms) evoked an inward Ca2+ current (ICa) in the cone and an EPSC in the HC. After addition of vehicle (gray traces), both ICa and the EPSC amplitude increased slightly due to runup. B, In a different cone-HC pair, the Group III mGluR agonist L-AP4 (100 μm) caused a decrease in the cone-driven EPSC. C, Group data (mean ± SEM) showing that, in the presence of L-AP4 (n = 9), EPSC amplitude was decreased and the runup of QCa was reduced relative to vehicle controls (n = 7). *p < 0.05 (unpaired t tests). **p < 0.005 (unpaired t tests). D, Similar experiments conducted with amphotericin/gramicidin perforated patch recordings from cones to minimize runup of either QCa or quantal amplitude showing that L-AP4 caused a reduction in both EPSC and QCa (n = 6) relative to vehicle controls (n = 5). *p < 0.05 (unpaired t test). **p < 0.005 (unpaired t test). E, L-AP4 caused a small outward current in a voltage-clamped HC (Vhold = −69 mV), reduced the amplitude of the outward current evoked at light onset, and reduced the inward current at light onset.
To verify this result under conditions that prevent whole-cell dialysis-dependent runup, we made perforated patch recordings from cones while making conventional whole-cell recordings from postsynaptic HCs (Fig. 1D). In these experiments, L-AP4 caused a 14 ± 5% decrease in QCa (n = 6; p = 0.036, paired t test) and a 28 ± 4% decrease in EPSC amplitude (n = 6; p = 0.00029, paired t test). With vehicle controls, in contrast, there was a nonsignificant 2 ± 3% increase in EPSC amplitude (n = 5, p = 0.25, paired t test) and a nonsignificant 3 ± 6% increase in QCa (n = 4, p = 0.45, paired t test).
Although several studies have pointed to decreases in photoreceptor synaptic transmission following L-AP4 application (Nawy et al., 1989; Hirasawa et al., 2002; Hosoi et al., 2005), others have shown either no effect (Slaughter and Miller, 1981) or examples of L-AP4 increasing the responses of Off-responding retinal neurons (Arkin and Miller, 1987). In darkness, photoreceptors are relatively depolarized and tonically release glutamate onto postsynaptic neurons. Light-evoked hyperpolarization reduces the rate of glutamate release. Consistent with a reduced synaptic drive from cones due to activation of presynaptic mGluRs, the outward current evoked in HCs at light onset was reduced to 79 ± 16% of control in the presence of L-AP4 (n = 10, p = 0.014, Wilcoxon; Fig. 1E). The phasic inward current at light offset, which reflects exocytosis of ribbon-associated synaptic vesicles from cones (Jackman et al., 2009), was reduced (69 ± 11% of control; n = 13; p = 0.022, Wilcoxon).
In darkness, tonic synaptic drive from photoreceptors can be detected as miniature quantal EPSCs in HCs (Fig. 2A–C). Consistent with a role for mGluR activation in reducing photoreceptor synaptic output, L-AP4 also caused a reduction in HC mEPSC frequency, from 125 ± 11 Hz to 97 ± 13 Hz (n = 7, p = 0.0073, paired t test). Unlike a previous study of L-AP4 effects on carp HCs, where L-AP4 application was accompanied by a small reduction in mEPSC amplitude (Hirasawa et al., 2002), we did not observe any significant change in mEPSC amplitude in our experiments (control: 8.0 ± 0.7 pA; L-AP4: 7.6 ± 0.5 pA, n = 7, p = 0.42, paired t test). L-AP4 also did not affect mEPSC kinetics; the 10%–90% rise time was nearly identical in both control and L-AP4 (control: 1.4 ± 0.11 ms; L-AP4: 1.4 ± 0.12 ms; n = 7, p = 0.9979, paired t test), as was the decay time constant (control: 2.27 ± 0.25 ms; L-AP4: 2.25 ± 0.20 ms; n = 7, p = 0.90, paired t test).
mGluR activation reduces Ca2+-dependent quantal release from cones. A, Voltage-clamp recording of mEPSCs from an HC (Vhold = −69 mV) in a vertical slice of salamander retina in control conditions and in the presence of L-AP4 (100 μm). L-AP4 caused a reduction in mEPSC frequency. B, Group data (mean + SEM) of mEPSC frequency showing that L-AP4 caused a reduction in event frequency without affecting amplitude or event kinetics (n = 7 HCs). *p < 0.05 (paired t test). C, Voltage-clamp recordings from cones showing glutamate transporter anion currents evoked presynaptically by spontaneous exocytosis of synaptic vesicles. A potassium thiocyanate intracellular solution was used to enhance the amplitude of anion transporter currents. Traces show events recorded from different cones in vehicle controls and L-AP4 (100 μm). D, Group data (mean + SEM) showing that L-AP4 had no significant effect on event frequency. Not significant, p > 0.05 (unpaired t test). Vehicle controls, n = 9 cones; L-AP4 (100 μm), n = 11 cones. E, Glutamate transporter currents evoked by a 25 ms depolarizing step (−79 to −19 mV, P/8 leak-subtracted) in two different cones in the presence of vehicle or 100 μm L-AP4. Despite causing no change in quantal amplitude or frequency of Ca2+-independent spontaneous release events (C), L-AP4 caused a significant reduction in evoked transporter currents. *p < 0.05 (unpaired t test). F, Voltage-clamp recording of mEPSCs from an HC (Vhold = −69 mV control conditions and in the presence of 50 μm CPPG). G, CPPG caused an increase in mEPSC frequency (n = 11 HC). *p < 0.05 (paired t test).
We next recorded presynaptic glutamate transporter currents arising from fusion of single vesicles using a pipette solution with thiocyanate (SCN−) as the major anion (Fig. 2C,D) (Picaud et al., 1995a, b; Szmajda and Devries, 2011; Grabner et al., 2015). SCN− is a more permeant anion for glutamate transporters and thereby enhances the amplitude of quantal transporter currents (Szmajda and Devries, 2011; Grabner et al., 2015). In darkness, AMPA receptor-mediated mEPSCs are the result of glutamate release from both rods and cones that is both Ca2+-dependent (due to opening of voltage-gated Ca2+ channels and Ca2+-induced Ca2+ release) and Ca2+-independent (Maple et al., 1994; Cork et al., 2016). When cones are voltage-clamped to minimize Ca2+ channel opening, spontaneous release appears to be entirely Ca2+-independent (Cork et al., 2016). Measuring glutamate transporter currents in voltage-clamped cones allowed us to examine these Ca2+-independent vesicle fusion events from the voltage-clamped cone. In vehicle controls, we detected events with an average amplitude of 8.5 ± 0.9 pA and a frequency of 17.7 ± 1.4 Hz (n = 9). In cones treated with L-AP4, the frequency was 16.4 ± 1.0 Hz, which was not significantly different from vehicle (p = 0.46, unpaired t test, n = 11). The event amplitude was also not significantly different in L-AP4-treated cones (9.0 ± 0.8 pA, n = 11, p = 0.68, unpaired t test).
Although L-AP4 did not reduce Ca2+-independent spontaneous glutamate release, application of L-AP4 did reduce the amplitude of glutamate transporter currents evoked by a depolarizing stimulus (Fig. 2E). In L-AP4-treated cones, a 20 ms depolarization (with a P/8 leak subtraction protocol) evoked a 174 ± 35 pA current (n = 9), which was significantly smaller than the 313 ± 37 pA current evoked in vehicle controls (n = 8, p = 0.016, unpaired t test). Thus, our results point to a role of Group III mGluRs in regulating Ca2+-dependent, but not Ca2+-independent, exocytosis from cones. The ability of mGluRs to inhibit release from cones is consistent with previous work in fish (Nawy et al., 1989; Hirasawa et al., 2002) and amphibian retina (Hosoi et al., 2005).
The ability to modulate release with the exogenously applied glutamate agonist L-AP4 suggests that glutamate released into the synaptic cleft during darkness does not saturate presynaptic mGluRs on cones. To test whether cleft glutamate has a meaningful impact on cone synaptic output in darkness, we next blocked mGluR signaling using the Group III mGluR antagonist CPPG (50 μm; Fig. 2F,G). Consistent with mGluR-dependent tonic inhibition of synaptic vesicle exocytosis in darkness, CPPG caused a small but significant increase in the frequency of mEPSCs observed in HCs (from 98 ± 3 to 111 ± 3 Hz; n = 11; p = 0.0078, paired t test) without affecting mEPSC amplitude (control = 8.5 ± 0.5 pA; CPPG = 8.6 ± 0.6 pA; n = 11; p = 0.82, paired t test).
To explore the functional impact of mGluR signaling during a more complex time-varying light stimulus, we next recorded HC light responses to a 2 Hz square-wave white light stimulus train (1100 Lux, 20 s duration; Fig. 3). Because mGluR activation with L-AP4 reduced light responses, one might expect CPPG to have the opposite effect. Instead, in these experiments, CPPG significantly reduced the light response power measured at 2 Hz from 4.3 ± 0.2 to 3.5 ± 0.3 log(pA2/Hz) (p = 0.00028, paired t test; n = 16). This result supports a role for mGluRs in shaping light-response encoding across the cone synapse, which might be attributable to mGluRs preventing vesicle pool depletion, as proposed previously (Hosoi et al., 2005).
Inhibition of mGluRs affects HC responses to a square-wave light stimulus. HC responses to a 2 Hz square wave light stimulus (white LED, 1100 Lux, 20 s duration) in control conditions and following application of CPPG (50 μm). The response was reduced in the presence of CPPG and recovered after washing to control solution.
A previous study on the role of Group III mGluRs in cones showed that application of L-AP4 caused a modest decrease in QCa (∼15%) at the peak of the I–V curve (−20 mV) and a 30%–50% decrease in synaptic transmission to second-order neurons (Hosoi et al., 2005). The mismatch in L-AP4 effects on synaptic transmission and Ca2+ influx is notable, as studies of the Ca2+ dependence of synaptic vesicle exocytosis from cones have indicated that the exocytotic Ca2+ sensor has a cooperativity of n = 1–2 (Thoreson et al., 2004; Duncan et al., 2010). Moreover, our previous work studying the effects of intracellular Ca2+ buffering in cones (Bartoletti et al., 2011; Mercer et al., 2011; Van Hook and Thoreson, 2014, 2015) has suggested that phasic exocytosis is triggered by Ca2+ nanodomains, which could also give rise to linear Ca2+ dependence of synaptic vesicle exocytosis (Eggermann et al., 2011). The unconventionally low Ca2+ cooperativity of the cone Ca2+ sensor and evidence for nanodomain control of exocytosis suggest that a reduction in Ca2+ influx should lead to a similar reduction in exocytosis. This contrasts with both our results and those of Hosoi et al. (2005), where the reduction in exocytosis appeared to be twofold to fourfold greater than the reduction in ICa.
To verify that cone exocytosis is indeed linearly correlated with Ca2+ influx, we used an approach that has previously been used to study the Ca2+ dependence of exocytosis by inner hair cells (Goutman and Glowatzki, 2007). In these experiments (Fig. 4A), we recorded EPSCs in HCs while evoking synaptic vesicle exocytosis from cones with steps to −29, −19, −9, 1, 11, or 21 mV from a holding potential of 61 mV so that the Ca2+ conductance was maximally activated and the only effect of voltage was to change the Ca2+ driving force. In these recordings (n = 8 cone-HC pairs), EPSC amplitude changed in proportion to ICa (measured using a P/8 leak-subtracted ramp protocol from −99 to 51 mV at 0.5 mV/ms). When we plotted the EPSC amplitude as a function of ICa (Fig. 4B), the two exhibited a strong linear correlation (r2 = 0.98, p = 0.0002, Pearson correlation). This confirms that exocytosis in cones is linearly correlated with the local Ca2+ influx, a result stemming from low cooperativity of the Ca2+ sensor (Thoreson et al., 2004; Duncan et al., 2010; Bartoletti et al., 2011; Van Hook and Thoreson, 2015).
Synaptic vesicle exocytosis is linearly correlated with Ca2+ influx at the cone synapse. A, Overlaid plots of ICa and EPSCs. ICa (gray line) is plotted as the reversal potential-corrected Boltzmann fit of the average cone membrane current recorded from n = 8 cones with a P/8 leak subtraction protocol (−99 to +51 mV ramp, 0.5 mV/ms). EPSC amplitudes (black square) are mean ± SEM of normalized EPSCs recorded in cone-HC pairs in which the cone was held near the Ca2+ reversal potential (61 mV) before being stepped down to a variety of test potentials (−29 to 21 mV) to evoke an EPSC (n = 8 cone-HC pairs). B, Replotting of data from A showing normalized EPSCs plotted against the ICa. The relationship was well fit with a straight line (r2 = 0.98, p = 0.0002, Pearson correlation), showing that exocytosis is linearly related to the amount of Ca2+ influx.
Thus, our data support a role for a Group III mGluR in regulating synaptic transmission by cone photoreceptors, in concert with previous work (Nawy et al., 1989; Hirasawa et al., 2002; Hosoi et al., 2005). The linear dependence of exocytosis on Ca2+ influx suggests that a mechanism downstream of Ca2+ entry contributes to the decrease in synaptic transmission. Gβγ is a likely candidate to mediate this effect, as previous studies in lamprey and rodent neurons have revealed that Gβγ can bind directly to components of the SNARE complex to inhibit exocytosis downstream of Ca2+ influx (Blackmer et al., 2001, 2005; Gerachshenko et al., 2005; Yoon et al., 2007; Wells et al., 2012). Current evidence indicates that this is largely the result of interactions of the Gβγ complex with amino acids located on the C terminus of SNAP-25 (Blackmer et al., 2001, 2005; Gerachshenko et al., 2005; Yoon et al., 2007; Wells et al., 2012), although other components such as syntaxin also appear to bind Gβγ (Yoon et al., 2007; Wells et al., 2012; Zurawski et al., 2016).
To examine localization of the Gβγ complex and SNAP-25 in salamander retina, we stained 30 μm retinal sections with antibodies against Gβ and SNAP-25 (Fig. 5A). We found strong Gβ labeling in photoreceptor outer segments as expected from its association with opsin, as well as labeling in the outer plexiform layer (OPL). SNAP-25 labeling was detected strongly in the IPL and OPL, as expected from its function in synaptic vesicle fusion. There are conflicting reports about whether SNAP-25 is expressed at cone photoreceptor synapses (Catsicas et al., 1992; Ullrich and Südhof, 1994; Brandstätter et al., 1996; Grabs et al., 1996; Greenlee et al., 2001; Hirano et al., 2011). Therefore, we sought to confirm its presence in cone synaptic terminals by staining isolated cone photoreceptors with an antibody sensitive to SNAP-25 (Fig. 5B). In these experiments, robust SNAP-25 staining was detected at cone terminals. Terminals were identified by staining for synaptic ribbons with an antibody raised against CtBP2. Although cone outer segments were lost in these isolated cells, we did detect some SNAP-25 labeling throughout inner segments, which is consistent with expression of SNAP-25 on the Golgi and a role for SNARE proteins in intracellular transport (Morgans and Brandstätter, 2000; Mazelova et al., 2009; Zulliger et al., 2015). We also performed double immunostaining for SNAP-25 and Gβ in isolated cones (Fig. 5C). In these cells, Gβ was detected throughout the cell, and strongly at the synaptic terminal, along with SNAP-25. To test for protein–protein interactions between the Gβγ complex and SNAP-25, we used the PLA after incubating 30 μm retinal sections with both Gβ and SNAP-25 antibodies. In retinal sections (Fig. 5D–G), there was PLA signal (red puncta) present in the OPL, the location of photoreceptor synaptic terminals (indicated by CtBP2 staining), as well as in the ganglion cell layer, inner plexiform layer, and photoreceptor outer segments. The density of PLA puncta varied with retinal layers (Fig. 5E; n = 4 sections), with the highest density in the IPL (11.0 ± 2.3/100 μm2). There was also a band of PLA puncta in the OPL (4.1 ± 0.8/100 μm2), higher than in the neighboring INL (2.9 ± 0.7/100 μm2; p = 0.008) or photoreceptor inner segments (1.8 ± 0.7/100 μm2; p = 0.012). When we omitted the SNAP-25 antibody as a negative control (n = 3 sections; Fig. 5G), the PLA signal was nearly abolished. This confirms that Gβγ and SNAP-25 directly interact with one another in the OPL, as well as other regions of retina.
Localization and interactions of the Gβγ complex and SNAP-25. A, The 30 μm vertical section of a salamander retina showing staining for SNAP-25 (green) and Gβ (red). SNAP-25 staining is especially apparent in the OPL and IPL, and Gβ staining is present in photoreceptor outer and inner segments and in the OPL. B, Staining of an isolated cone photoreceptor for synaptic ribbons (CtBP2, red) and SNAP-25 (green) showing strong signal for both in the cone synaptic terminal (arrowhead). C, Staining of an isolated cone for Gβ (red) and SNAP-25 (green), showing labeling throughout the cell, including in the synaptic terminal (arrowhead). D, PLA for Gβ and SNAP-25 interactions shows strong PLA signal (red puncta) throughout the retina, including in the OPL. E, Density of PLA puncta (n = 4 retinal sections) measured in different retinal layers (black bars) and in negative control (n = 3 retinal sections) experiments in which the SNAP-25 antibody was omitted (white bars). F, Cropped image of the OPL showing immunohistochemical staining for synaptic ribbons (CtBP2, green) and PLA signal for SNAP-25/Gβ interactions (red). G, The PLA signal was nearly abolished when the SNAP-25 antibody was omitted.
Having found Gβγ-SNAP-25 interactions in the OPL, we next used two complementary approaches to test whether this specific interaction contributes to the Group III mGluR-dependent reduction in synaptic transmission by cones (Fig. 6). In the first approach (Fig. 6A), we incubated retinal slices in botulinum neurotoxin A (BoNT/A; 20 nm) for 60 min before beginning electrophysiological recordings. BoNT/A cleaves 9 amino acids from the C terminus of SNAP-25 at a fairly well-conserved pair of residues (Q197 and R198) (Schiavo et al., 1993). The surrounding residues are also important for BoNT/A recognition in mammals (Schiavo et al., 1993; Washbourne et al., 1997). Although the tiger salamander SNAP-25 sequence is unknown, this site and the rest of the SNAP-25 C terminus are largely conserved in the Japanese fire belly newt (Cynops pyrrhogaster, GenBank accession #BAE47569.1), another amphibian related to the tiger salamander. The actions of BoNT/A at this site have been shown to inhibit Gβγ-SNAP-25 interactions and Gβγ-mediated reductions in synaptic transmission in other neurons (Gerachshenko et al., 2005; Delaney et al., 2007; Hamid et al., 2014). In recordings from tissue treated with BoNT/A (n = 5 cone-HC pairs), L-AP4 application resulted in a nonsignificant 8 ± 4% increase in ICa (p = 0.14, paired t test; n = 5) and a nonsignificant 7 ± 4% reduction in the EPSC amplitude (p = 0.17, paired t test; n = 5). After correcting for runup, this corresponded to an estimated ∼11% reduction in Ca2+ influx and a similar ∼12% estimated reduction in EPSC amplitude. Thus, although L-AP4 still caused a reduction in both exocytosis and ICa, the two were reduced by a similar fraction, in contrast to L-AP4 treatment in non-BoNT/A-treated tissue.
mGluR-mediated inhibition of exocytosis involves the SNAP-25 C terminus. A, Paired voltage-clamp recording of a cone and a HC in which retinal slices were pretreated with 20 nm BoNT/A for 1 h before beginning recordings. ICa runs up slightly following L-AP4 application and the EPSC is slightly reduced. B, Left, Slices were incubated in amphibian saline solution for 1 h before beginning recording (control). Right, Slices were incubated with 20 nm BoNT/A for 1 h before beginning recording (BoNT/A). In paired voltage-clamp recordings of a cone and a HC, switching from the standard amphibian saline solution containing 1.8 mm Ca2+ to a solution containing 1.0 mm Ca2+ reduced the EPSC. When slices were treated with BoNT/A, the reduction in EPSC amplitude was greater following application of the 1.0 mm Ca2+ solution. C, Paired cone/HC voltage-clamp recording in which the cone patch pipette solution contained the ct-SNAP-25 peptide (1 μm) in control conditions (black) and in the presence of L-AP4 (100 μm, gray). ICa ran up slightly following L-AP4 application, and the EPSC was slightly reduced. D, Group data (mean + SEM) showing the effects of L-AP4 (normalized to preL-AP4 controls) when the cone patch pipette contained the ct-SNAP-25 peptide (n = 7 cone-HC pairs) or when slices were pretreated with BoNT/A (n = 5 cone-HC pairs). The L-AP4 data (n = 9 cone-HC pairs) are replotted from Figure 1 for comparison with ct-SNAP-25 and BoNT/A experiments. The effect of L-AP4 was significantly reduced in ct-SNAP-25 or BoNT/A conditions, whereas Ca2+ influx was unaffected. **p < 0.005 (unpaired t tests). Not significant, p > 0.05 (unpaired t tests).
Surprisingly, EPSC amplitude in BoNT/A-treated cone-HC pairs (196 ± 22 pA; n = 17) did not differ significantly from EPSCs recorded in untreated tissue (231 ± 28 pA, n = 24, p = 0.32, unpaired t test). There are two factors that likely contribute to this finding. The first is that the amplitude of cone-driven EPSCs across individual cones is highly variable (range = 48–728 pA in this sample) and scales with the number of ribbon contacts the cone makes with any given HC (47 pA per ribbon-contact) (Bartoletti et al., 2010). Second, the SNAP-25 C terminus is a site of Ca2+-dependent interaction with synaptotagmin and BoNT/A-mediated cleavage of the last 9 residues appears to cause a shift in the Ca2+ dependence of that interaction (Otto et al., 1995; Gerona et al., 2000; Humeau et al., 2000). Because our stimulus reaches the peak of the cone ICa, the nanodomain of elevated intracellular [Ca2+] might be sufficient to overcome the effect of BoNT/A (Gerona et al., 2000; Humeau et al., 2000). We tested this possibility by recording cone-driven EPSCs when [Ca2+]o was reduced from 1.8 mm to 1.0 mm to reduce the concentration of Ca2+ attained in nanodomains beneath open Ca2+ channels (Fig. 6B). In slices treated with BoNT/A, switching from 1.8 to 1.0 mm [Ca2+]o reduced EPSC amplitude by 41 ± 7% (n = 6), a significantly greater reduction than occurred in control (non-BoNT/A) slices (19 ± 5%, n = 6; p = 0.03). This verifies that BoNT/A is effective at salamander cone synapses and is consistent with previous reports indicating that high [Ca2+] can overcome the BoNT/A-mediated reduction in exocytosis (Gerona et al., 2000; Humeau et al., 2000).
In another set of experiments, we introduced a 14 amino acid peptide matching the C terminus of SNAP-25 into the cone through the patch pipette (ct-SNAP-25; Fig. 6C). This region of SNAP-25 was shown to be important for Gβγ binding (Blackmer et al., 2005; Yoon et al., 2007; Wells et al., 2012), and a peptide corresponding to this region prevents Gβγ-mediated inhibition of exocytosis (Gerachshenko et al., 2005). In recordings in which cones were dialyzed with 1 μm ct-SNAP-25, ICa was increased by 9 ± 3% (n = 7; p = 0.019, paired t test) and the EPSC was not significantly changed (1 ± 3% decrease; p = 0.4, paired t test) following L-AP4 application (100 μm) compared with a 28 ± 5% decrease in control. However, if we account for runup of both the EPSC and ICa, these values correspond to an estimated ∼6% decrease in EPSC amplitude and an estimated ∼10% decrease in ICa. Unlike L-AP4 treatment in control, application of L-AP4 to cones dialyzed with the inhibitory ct-SNAP-25 peptide produced similar decreases in both ICa and EPSC amplitude. Like BoNT/A experiments, this suggests that the ct-SNAP-25 peptide effectively blocks the interaction of Gβγ with the S + or − NARE complex and removes the mGluR effect on EPSCs and that Group III mGluR activation reduces exocytosis both by reducing Ca2+ influx and by direct Gβγ-SNAP-25 interactions downstream of Ca2+ influx.
L-AP4 application also caused a small reduction in Ca2+ influx (Fig. 1). Gβγ is able to reduce ICa by directly interacting with voltage-gated Ca2+ channels, including some L-type channels (Ivanina et al., 2000; Tedford and Zamponi, 2006; Currie, 2010; Farrell et al., 2014). To verify that the Gβγ complex is able to inhibit exocytosis and test for a direct influence of Gβγ on ICa, we made paired whole-cell recordings of cones and HCs and introduced Gβ1γ1 complex isolated from bovine rods (0.1–2 μm) into the cone through the patch pipette (Fig. 7). In these experiments, we simultaneously measured leak-subtracted ICa in cones and EPSCs in HCs by depolarizing the cone shortly after establishing a whole-cell configuration and every 3 min for >20 min thereafter. As a negative control, we boiled the Gβ1γ1 stock solution for 30 min before diluting it in the cone pipette solution. As shown in Figure 7C–F, although Gβ1γ1 caused a dose-dependent reduction in EPSC amplitude (IC50 = 80 nm), there was no effect on ICa amplitude. With boiled Gβ1γ1, the EPSCs ran up, reaching 131 ± 17% of the first EPSC 15 min into the recording (n = 7 cone/HC pairs). The EPSC was reduced to 88 ± 14% (n = 4) with 100 nm Gβ1γ1, 56 ± 10% with 300 nm Gβ1γ1 (n = 8), and 48 ± 14% with 2 μm Gβ1γ1. ICa ran up to 138 ± 12% for the boiled control (n = 5), 142 ± 12% for 100 nm, 134 ± 5% for 300 nm, and 124 ± 14% for 2 μm Gβ1γ1 (n = 7). Because these measurements were taken with a step to the peak of the ICa activation curve (−19 mV), which might not have revealed Gβ1γ1-mediated changes in ICa voltage dependence, we next measured leak-subtracted ICa in cones dialyzed with boiled Gβ1γ1 or 300 nm Gβ1γ1 using a series of voltage steps (−69 to 31 mV, 10 mV increments). In these experiments (Fig. 7G,H), neither the maximum QCa nor the voltage dependence was affected by the presence of Gβ1γ1 (boiled peak QCa: 18 ± 1 pC, n = 6; Gβ1γ1 peak QCa: 18 ± 1 pC at −19 mV, n = 5; p = 0.46; boiled V50 = −35 ± 2 mV, n = 6; Gβ1γ1 V50 = −36 ± 1 mV, n = 5; p = 0.5). The finding that exogenous Gβ1γ1 did not alter Ca2+ channel function suggests that the effects of Group III mGluR activation on ICa are more likely the result of Gα-dependent or other signaling pathways downstream of Group III mGluRs.
Cone synaptic transmission is inhibited by direct introduction of Gβ1γ1 subunits. A, Paired cone-HC voltage-clamp recordings in which the cone patch pipette contained boiled Gβ1γ1 subunits. The cone was depolarized (−79 to −19 mV, 100 ms) every 3 min, triggering an EPSC in the HC. Cone ICa was recorded using a P/8 leak subtraction protocol. Over the course of the recording, both cone Ca2+ influx and the HC EPSC increased slightly. B, Similar experiment in which the cone patch pipette solution contained 300 nm Gβ1γ1. Similar to the boiled control (A), Ca2+ influx ran up over the recording. Unlike the boiled control, the EPSC amplitude was inhibited after several minutes of whole-cell recording. C, Group EPSC time series showing that EPSC amplitude (mean ± SEM) gradually ran up over the course of the recording when the cone was dialyzed with boiled Gβ1γ1 (solid black, n = 7) but was inhibited when the pipette contained functional Gβ1γ1 subunits (solid gray, 100 nm, n = 4; dashed black, 300 nm, n = 8; dashed gray, 2 μm, n = 5). D, Time series measurements of integrated ICa (QCa, mean ± SEM) showing that Ca2+ influx was similar over the course of the recording whether the cone pipette contained boiled or functional Gβ1γ1 subunits. E, Group data (mean + SEM) showing the EPSC and QCa measured at 15 min of recording relative to the first response (at time = 0). The EPSC amplitude was significantly reduced (*p < 0.05, **p < 0.01, unpaired t tests) relative to the boiled control, but QCa was not affected (p > 0.05, unpaired t test). F, Dose–response plot of data normalized for runup in the boiled condition and fit with a sigmoidal dose–response curve showing that Gβ1γ1 effects on EPSCs had an IC50 of 80 nm. G, Cone Ca2+ influx (mean ± SEM) recorded at a series of test potentials using a P/8 leak subtraction protocol (100 ms steps) showing that the current amplitude was similar in both Gβ1γ1 boiled controls (solid black line, n = 5 cones) and when cones were dialyzed with 300 nm Gβ1γ1 (dashed gray line, n = 6 cones). H, Plot of normalized Ca2+ conductance (GCa, mean ± SEM) at a series of test potentials and fit with Boltzmann functions showing that the ICa voltage dependence was similar when cones were dialyzed with either Gβ1γ1 or boiled Gβ1γ1. The Gβ1γ1 V50 was −35.8 ± 1.3 mV, and the boiled control V50 was −35.3 ± 2.0 mV (p > 0.05, unpaired t test).
In a parallel set of experiments (Fig. 8), we used whole-cell Cm recordings to measure synaptic vesicle exocytosis directly from cones dialyzed with either functional or boiled Gβ1γ1 (300 nm). Because Cm recordings provide a measure of changes in membrane surface area resulting from addition of vesicular membrane, they are not prone to runup due to the inclusion of glutamate in the presynaptic patch pipette, which affects quantal content (Bartoletti and Thoreson, 2011). Additionally, because we needed to introduce the large Gβ1γ1 protein complex through the patch pipette, we could not use perforated patch recordings for this purpose. When cones were dialyzed with Gβ1γ1, the capacitance transient evoked by a 25 ms depolarizing step (from −79 to −19 mV) was reduced by 53 ± 5% (n = 6) relative to the first response within 8–12 min of whole-cell recording (p = 0.0029, paired t test). In contrast, when the cone was dialyzed with boiled Gβ1γ1, the depolarization-evoked capacitance response was relatively unchanged after 8–12 min of whole-cell recording (2 ± 10% increase, n = 6, p = 0.97, paired t test). The use of Cm recordings confirms results obtained using paired whole-cell measurements of synaptic transmission. Gβγ has been linked to changes in vesicle fusion mode, shifting to a greater proportion of kiss-and-run exo/endocytosis (Chen et al., 2005; Photowala et al., 2006; Schwartz et al., 2007; Yoon et al., 2008), possibly due to interactions of the endocytic protein dynamin with Gβγ (Lin and Gilman, 1996). These Cm recordings also allowed us to assess the influence of Gβ1γ1 on endocytic vesicle retrieval. Despite the reduction in exocytotic Cm increases in the presence of Gβ1γ1, the kinetics of endocytic membrane retrieval appeared unchanged (t(50) = 688 ± 100 ms with boiled Gβ1γ1 controls and 632 ± 80 ms with unboiled Gβ1γ1, p = 0.68). As noted in Materials and Methods, we included the glutamate transporter inhibitor TBOA in the bath during Cm recordings to minimize artifacts from activation of the transporter Cl− conductance. At calyx of Held synapses, raising cleft glutamate levels following glutamate transporter inhibition leads to an mGluR-dependent presynaptic inhibition of synaptic release (Renden et al., 2005). In our Cm recordings, the presynaptic cone was voltage-clamped to inhibit glutamate release and so cleft glutamate was unlikely to be similarly elevated. Consistent with this, a prior study from our laboratory showed that TBOA had no significant effect on cone-driven EPSCs recorded in a paired voltage-clamp recording configuration (Cadetti et al., 2008).
Exocytotic capacitance responses are inhibited by Gβ1γ1. A, Whole-cell Cm recording of a cone in which the patch pipette contained 300 nm Gβ1γ1. After 10 min of recording (gray trace), the exocytotic Cm increase evoked by a depolarizing stimulus (−79 to −19 mV, 25 ms) was reduced relative to a response recorded early in the recording (black trace). The access resistance (Ra) was unchanged by the depolarizing stimulus. B, A similar experiment to that illustrated in A, except that the cone patch pipette contained boiled Gβ1γ1. In this recording, the Cm response recorded late in the recording (12 min, gray trace) was similar to the response recorded early in the recording (3 min, black trace). C, Group data (mean ± SEM) of exocytotic Cm responses recorded late (9–12 min of whole-cell recording) normalized to early responses (0–3 min recording) showing that, when cones were dialyzed with functional Gβ1γ1 (n = 6 cones), the responses were significantly reduced (***p < 0.005, paired t test), whereas those with the boiled Gβ1γ1 were not affected (not significant, p > 0.05, paired t test). D, Normalized whole-cell Cm traces in which the patch pipette contained either 300 nm Gβ1γ1 or boiled Gβ1γ1 displayed on an expanded time base to illustrate that the rapid endocytic decline in Cm was unaffected by Gβ1γ1. Displayed traces are from two different cones, and responses were evoked ∼10 min into whole-cell recording with a 25 ms depolarizing step (to −19 mV from −79 mV).
To test that the inhibition in synaptic transmission accompanying dialysis of cones with exogenous Gβ1γ1 is the result of Gβγ-SNARE interactions, we incubated retinal slices with BoNT/A (20 nm) for 30 or 60 min before beginning experiments (Fig. 9). For retinal tissue treated for 30 min with BoNT/A, the EPSC amplitude was reduced to 89 ± 8% of the first response by ∼10 min after establishing whole-cell recording configuration with a pipette solution containing 300 nm Gβ1γ1 (n = 5 cone/HC pairs). This is a less dramatic reduction in EPSC amplitude than seen with 300 nm Gβ1γ1 in retinas not pretreated with BoNT/A (reduced to 56% of first EPSC, above). Because longer incubation with BoNT/A will lead to a greater quantity of SNAP-25 lacking the C terminus, we next doubled the BoNT/A incubation time to 60 min before beginning recordings. In these slices, the effect of Gβ1γ1 on the EPSC was largely reversed, with the EPSC amplitude being 117 ± 21% of the first EPSC recorded at break-in (n = 6 cone-HC pairs). This approached the ∼131% increase in EPSC amplitude seen with boiled Gβ1γ1, above (Fig. 6). Thus, Gβγ-dependent inhibition of exocytosis in cones appears to depend on an intact SNAP-25 C terminus, supporting a role for Gβγ-SNAP-25-mediated inhibition of exocytosis following Group III mGluR activation.
Gβ1γ1 effects on synaptic transmission depend on the SNAP-25 C terminus. A, Paired whole-cell recording of a cone and a postsynaptic HC in which the cone patch pipette solution contained 300 nm Gβ1γ1 and the slices were incubated with BoNT/A (20 nm) for 30 min before beginning experiments. Whereas ICa showed typical runup, the EPSC was slightly reduced compared with the earlier response after 9 min of recording. B, Similar experiment as in A, except that the slices were pretreated with BoNT/A for a longer period of 60 min before beginning recordings. In this condition, both the EPSC and ICa amplitude were increased over an early response after 9 min of recording. C, Group data (mean + SEM) showing the effects of BoNT/A pretreatment on the Gβ1γ1-dependent effects on synaptic transmission. Boiled and 300 nm Gβ1γ1 data are replotted from Figure 5 for comparison with BoNT/A-treated experiments. With 30 min BoNT/A pretreatment, the effect of 300 nm Gβ1γ1 on EPSC amplitude was significantly inhibited relative to 300 nm Gβ1γ1 alone (p < 0.05). The 60 min of BoNT/A treatment also significantly inhibited the effects of 300 nm Gβ1γ1 relative to Gβ1γ1 alone (p < 0.05), and the EPSCs appeared to run up, nearly reaching the boiled Gβ1γ1 control values.
Discussion
Presynaptic mGluRs in retina and throughout the CNS suppress synaptic transmission by several mechanisms including inhibition of voltage-gated Ca2+ channels, enhancement of K+ conductances, and direct G-protein interactions with the vesicle fusion machinery (Sladeczek et al., 1993; Scanziani et al., 1995; Takahashi et al., 1996, 2001; von Gersdorff et al., 1997; Cochilla and Alford, 1998; Koulen et al., 1999, 2005; Awatramani and Slaughter, 2001; Higgs and Lukasiewicz, 2002; Higgs et al., 2002; Lorez et al., 2003; Hosoi et al., 2005; Mateo and Porter, 2007; Quraishi et al., 2007; Erdmann et al., 2012). Here, we show that mGluR-dependent suppression of synaptic transmission in cones occurs, in part, due to Gβγ interacting with the SNAP-25 C terminus. mGluR activation also reduced ICa, although this was not mimicked by direct introduction of Gβγ through the cone patch pipette, implying that it depends on Gα pathways. Although HCs also reportedly express mGluRs (Linn and Gafka, 1999), the effects of L-AP4 here were clearly presynaptic in origin, as they were blocked by introduction of the ct-SNAP25 peptide into the cone during paired recordings. The effect of Gβγ on exocytosis was confirmed to be the result of Gβγ interacting with SNAP-25 by the inhibition of L-AP4 effects by either pretreatment with BoNT/A or by introduction of the ct-SNAP-25 peptide. Moreover, direct introduction of Gβγ subunits dose-dependently inhibited exocytosis, and this effect was inhibited by BoNT/A pretreatment. Gβγ-SNARE interactions triggered by Gi/o-coupled GPCRs have been identified in a handful of other synapses: in mammals, the hippocampal CA1-subicular synapse (Hamid et al., 2014), Schaeffer collateral-CA1 synapses (Zhang et al., 2011), and inputs to the amygdala (Delaney et al., 2007), and in lamprey, reticulospinal axon synapses (Takahashi et al., 2001; Gerachshenko et al., 2005). Ours is the first study demonstrating the existence of Gβγ-SNAP-25 interaction-dependent modulation of synaptic transmission in the retina and at a ribbon-type synapse.
Gβγ regulates synaptic transmission by several mechanisms. Among these, direct interactions of Gβγ with the pore-forming α1 subunit of N- and P/Q-type Ca2+ channels have been shown to reduce synaptic transmission via a voltage-dependent mechanism for inhibition of ICa (Bean, 1989; Tedford and Zamponi, 2006; Currie, 2010). A comparable inhibition does not seem to occur at CaV1.4 (α1f) L-type channels that regulate synaptic release from photoreceptors due to the absence of a key Gβγ-interacting sequence in the α1 subunits (Herlitze et al., 1997; Furukawa et al., 1998a, b; Currie, 2010). However, a handful of studies have revealed a that Gβγ is capable of distinct voltage-independent direct inhibition of CaV1.2 (α1C) L-type channels in vitro (Ivanina et al., 2000) and endogenous L-type channels in retinal ganglion cells (Farrell et al., 2014). In our study, although mGluR activation reduced Ca2+ influx (notable after correction for runup), two results indicate that this occurred independently of Gβγ. First, although the ct-SNAP-25 peptide prevented a portion of the mGluR-dependent inhibition, it had no effect on ICa effects. Second, direct introduction of Gβγ subunits into cones had no effect on ICa amplitude or voltage dependence.
Gβγ can also indirectly regulate synaptic transmission by gating G-protein activated inwardly rectifying K+ channels expressed somatically or at presynaptic active zones (Ladera et al., 2008; Fernández-Alacid et al., 2009; Michaeli and Yaka, 2010; Betke et al., 2012). This increases K+ conductance, providing a voltage shunt and hyperpolarizing the neuron. Although G-protein activated inwardly rectifying K+ channels might influence the cone voltage responses, we voltage-clamped cones and used Cs+-based pipette solutions and therefore did not test for such a contribution.
Thus, our results do not support Gβγ-dependent regulation of ICa or IK underlying the effects on exocytosis, although they are consistent with a small contribution from Gα-dependent ICa effects. Instead, our major finding is that mGluR activation reduces cone synaptic transmission via direct Gβγ interactions with SNAP-25, a core SNARE protein required for Ca2+-triggered exocytosis (Banerjee et al., 1996; Bronk et al., 2007). Gβγ/SNARE interactions have been extensively characterized in lamprey reticulospinal-motor neuron synapses and also demonstrated to regulate exocytosis in hippocampus, amygdala, and pancreatic β cells (Chen et al., 2005; Photowala et al., 2006; Schwartz et al., 2007; Yoon et al., 2008). Other work has shown that mGluRs inhibit synaptic inputs onto hippocampal granule cells downstream of Ca2+ entry and independently of stereotypical GPCR/Gα-dependent pathway blockers (Erdmann et al., 2012), a result likely attributable to Gβγ-SNARE interactions.
By interacting with the SNAP-25 C terminus, Gβγ competes with binding of Ca2+-bound synaptotagmin1 (syt1) (Gerachshenko et al., 2005; Yoon et al., 2007; Wells et al., 2012; Zurawski et al., 2016). Gβγ also interacts with the SNAP-25 N terminus (Wells et al., 2012), although the major influence on exocytosis is due to C-terminal interactions. The selective influence of Gβγ on Ca2+-dependent exocytosis is a likely explanation for why L-AP4 reduces HC mEPSCs, which are the result of both Ca2+-dependent and Ca2+-independent exocytosis, without any apparent effect on cone quantal glutamate transporter currents, which are entirely due to Ca2+-independent exocytosis in cones (Maple et al., 1994; Cork et al., 2016). While syt1/2 are the major exocytotic Ca2+ sensors in most neurons, it is unclear which synaptotagmin isoform is responsible for photoreceptor exocytosis (Berntson and Morgans, 2003; Heidelberger et al., 2003; Kreft et al., 2003; Fox and Sanes, 2007). Different isoforms might result in subtly different Gβγ-SNAP-25-mediated effects on exocytosis. Binding studies also indicate that Gβγ interacts with syntaxin1a where it also competes with Ca2+-dependent synaptotagmin interactions (Yoon et al., 2007). This interaction occurs at the syntaxin1a H3 domain, a region important for interaction with synaptobrevin and SNAP-25 (Hayashi et al., 1994; McMahon and Südhof, 1995; Ramakrishnan et al., 2009). Of note, retinal ribbon synapses, such as those found in cones, contain synataxin3 rather than syntaxin1a (Brandstätter et al., 1996; Morgans et al., 1996), which shares a 61% homology with syntaxin1a (Bennett et al., 1993).
It is unclear exactly how Gβγ/SNARE interactions change the nature of exocytosis. The competition of Gβγ with Ca2+-synaptotagmin binding to SNARE proteins points toward a reduced Ca2+ sensitivity of release (Gerachshenko et al., 2005; Delaney et al., 2007; Yoon et al., 2007; Erdmann et al., 2012). Gβγ interactions also change the vesicle fusion mode so that release shifts to being dominated by kiss-and-run rather full-collapse fusion (Chen et al., 2005; Photowala et al., 2006; Schwartz et al., 2007; Yoon et al., 2008), possibly via interactions with the endocytic protein dynamin (Lin and Gilman, 1996). In kiss-and-run, transmitter is released through a narrow fusion pore formed at the mouth of the synaptic vesicle, reducing the amount of exocytosed transmitter and slowing mEPSC kinetics (He et al., 2006; He and Wu, 2007). In our recordings of postsynaptic mEPSCs in HCs, however, neither the mEPSC kinetics nor amplitude were altered by L-AP4 application, and the time course of endocytic membrane retrieval measured with Cm recordings was unaffected by Gβγ subunits, pointing instead to a change in effective vesicle pool size or release probability. The lack of an effect on fusion mode might reflect the minimal reliance of cone endocytosis on dynamin (Van Hook and Thoreson, 2012).
What functional role(s) do Gβγ/SNARE interactions play in cones? 5-HT leads to Gβγ-dependent synaptic depression on a very rapid time scale (∼20 ms) at the lamprey reticulospinal synapse (Gerachshenko et al., 2005). mGluRs can function as autoreceptors that rapidly influence synaptic transmission, making them contributors to short-term synaptic plasticity (von Gersdorff et al., 1997; Awatramani and Slaughter, 2001; Chen et al., 2002; Lorez et al., 2003; Mateo and Porter, 2007; Kyuyoung and Huguenard, 2014). Rapid synaptic depression is key for shaping the frequency response and filtering properties of synaptic transmission, and this can be influenced by several factors at the photoreceptor synapse (DeVries, 2000; Armstrong-Gold and Rieke, 2003; Zhang and Wu, 2005; DeVries et al., 2006; Van Hook et al., 2014). Indeed, blockade of mGluRs using CPPG reduced the amplitude of HC responses to a 2 Hz light stimulus. One interpretation of these data is that inhibition of presynaptic mGluRs leads to synaptic vesicle pool depletion by removing a mGluR-mediated “brake” on exocytosis, consistent with the increase in mEPSC frequency, as has been proposed previously (Hosoi et al., 2005). Increased cleft glutamate concentration with CPPG might also reduce synaptic responses via AMPAR desensitization and saturation. However, we did not see any change in mEPSC amplitude with CPPG application, arguing against a substantial effect of desensitization/saturation. Gβγ-dependent synaptic inhibition of release might also support adaptation at local spatial scales, with glutamate spillover to neighboring cones potentially influencing color opponent and center-surround receptive fields via mGluR (Luo and Liang, 2003; Szmajda and Devries, 2011).
Photoreceptors contain a variety of GPCRs beyond mGluRs, such as dopamine, adenosine, somatostatin, and endocannabinoid receptors that might inhibit photoreceptor synaptic transmission via Gβγ-SNARE interactions. Finally, phototransduction itself is a GPCR-dependent process, and phototransduction proteins translocate from outer segments and regulate synaptic transmission (Sokolov et al., 2004; Yamamoto et al., 2007; Herrmann et al., 2010; Huang et al., 2010; Majumder et al., 2013). It is a tantalizing possibility that Gβγ translocating from outer segments might contribute to synaptic light adaptation by interacting with SNARE proteins at the terminal.
Footnotes
This work was supported by National Institutes of Health Grants EY10542 to W.B.T., F32EY023864 to M.J.V.H., EY010291 to H.E.H., MH101679 to H.E.H., and DK109204 to H.E.H., and Research to Prevent Blindness Senior Scientific Investigator Award to W.B.T.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Matthew J. Van Hook, Department of Ophthalmology and Visual Sciences, University of Nebraska Medical Center, Durham Research Center 1, Room 4009, 985840 Nebraska Medical Center, Omaha, NE 68198-5840. matt.vanhook{at}unmc.edu
