Spontaneous action potentials in retinal horizontal cells of goldfish (Carassius auratus) are dependent upon L-type Ca²⁺ channels and ryanodine receptors

Abstract

Horizontal cells (HCs) are interneurons of the outer retina that undergo graded changes in membrane potential during the light response and provide feedback to photoreceptors. We characterized spontaneous Ca²⁺-based action potentials (APs) in isolated goldfish (Carassius auratus) HCs with electrophysiological and intracellular imaging techniques. Transient changes in intracellular Ca²⁺ concentration ([Ca²⁺]_i) were observed with fura-2 and were abolished by removal of extracellular Ca²⁺ or by inhibition of Ca²⁺ channels by 50 µM Cd²⁺ or 100 µM nifedipine. Inhibition of Ca²⁺ release from stores with 20 µM ryanodine or 50 µM dantrolene abolished Ca²⁺ transients and increased baseline [Ca²⁺]_i. This increased baseline was prevented by blocking L-type Ca²⁺ channels with nifedipine, suggesting that Ca²⁺-induced Ca²⁺ release from stores may be needed to inactivate membrane Ca²⁺ channels. Caffeine (3 mM) increased the frequency of Ca²⁺ transients, and the store-operated channel antagonist 2-aminoethyldiphenylborinate (100 μM) counteracted this effect. APs were detected with voltage-sensitive dye imaging (FluoVolt) and current-clamp electrophysiology. In current-clamp recordings, regenerative APs were abolished by removal of extracellular Ca²⁺ or in the presence of 5 mM Co²⁺ or 100 µM nifedipine, and APs were amplified with 15 mM Ba²⁺. Collectively, our data suggest that during APs Ca²⁺ enters through L-type Ca²⁺ channels and that Ca²⁺ stores (gated by ryanodine receptors) contribute to the rise in [Ca²⁺]_i. This work may lead to further understanding of the possible role APs have in vision, such as transitioning from light to darkness or modulating feedback from HCs to photoreceptors.

NEW & NOTEWORTHY Horizontal cells (HCs) are interneurons of the outer retina that provide inhibitory feedback onto photoreceptors. HCs respond to light via graded changes in membrane potential. We characterized spontaneous action potentials in HCs from goldfish and linked action potential generation to a rise in intracellular Ca²⁺ via plasma membrane channels and ryanodine receptors. Action potentials may play a role in vision, such as transitioning from light to darkness, or in modulating feedback from HCs to photoreceptors.

INTRODUCTION

Horizontal cells (HCs) are second-order interneurons in the inner nuclear layer of the retina. They facilitate visual processing, including lateral feedback inhibition onto photoreceptors to mediate edge detection and color opponency (Thoreson and Mangel 2012; Twig et al. 2003). HCs receive glutamatergic input from photoreceptors and characteristically respond with graded changes in membrane potential (Baylor et al. 1971; Perlman et al. 2011).

Less frequently, action potentials (APs) have been reported in HCs of fish, including the catfish (Ictalurus punctatus; Dixon et al. 1993; Johnston and Lam 1981; Shingai and Christensen 1983, 1986; Takahashi et al. 1993), goldfish (Carassius auratus; Tachibana 1981), common carp (Cyprinus carpio; Murakami and Takahashi 1987), and skate (Raja spp.; Lasater et al. 1984). APs have been observed in isolated HCs and in the intact retina (Murakami and Takahashi 1987; Tachibana 1981). They are dependent on Ca²⁺ influx through membrane ion channels and persist in the absence of Na⁺ (Shingai and Christensen 1983; Tachibana 1981). In addition, Kreitzer et al. (2012) observed spontaneous increases in intracellular Ca²⁺ concentration ([Ca²⁺]_i) with fluo-4 imaging. Neither APs nor spontaneous Ca²⁺ events are well characterized in HCs, and their role in vision (if any) is unknown.

HCs provide inhibitory feedback to photoreceptors by release of protons into the synaptic cleft (Barnes 2003; Hirasawa and Kaneko 2003; Vessey et al. 2005; Warren et al. 2016), by an ephaptic mechanism involving hemichannels (Gardner et al. 2015; Kamermans et al. 2001), or by release of GABA (Hirano et al. 2005, 2016; Liu et al. 2013). APs in HCs involve changes in [Ca²⁺]_i and membrane voltage (V_m) that would be likely to impact any of these mechanisms of feedback.

In the present study, we observed spontaneous changes in [Ca²⁺]_i and membrane potential in goldfish HCs. We designed experiments to quantitatively characterize these events and to identify the source of Ca²⁺. We compared the parameters of Ca²⁺ transients with fura-2-based imaging to those of voltage-sensitive dye (FluoVolt) and current-clamp recordings. For the first time, we show that Ca²⁺-based APs are driven by plasma membrane ion channels and intracellular stores and that AP frequency and amplitude are modulated by membrane potential in HCs.

METHODS

Ethical approval.

Procedures for animal care and use were reviewed and approved by the University of Ottawa Animal Care and Veterinary Services (protocol BL-1760) and were implemented in accordance with regulations of the Canadian Council on Animal Care. Adult common goldfish were obtained from a commercial supplier (AQUAlity Tropical Fish Supply Inc., Mississauga, ON, Canada) and were housed in the aquatic facilities at the University of Ottawa. Fish were maintained in 170-liter tanks fitted with a flow-through system of fresh, aerated, and dechloraminated water at a constant temperature of 18°C. Tank photoperiod was kept at a constant 12:12-h light-dark cycle. Goldfish were dark adapted for ~1 h, euthanized by rapid decapitation, and pithed.

Isolated cell preparation.

HC isolation closely followed that of Jonz and Barnes (2007). Unless otherwise stated, all reagents and chemicals were sourced from Sigma-Aldrich (Oakville, ON, Canada). Eyes were removed and placed in cold Ca²⁺-free Ringer solution (Table 1). The sclera and lens were dissected, and the whole retina was removed from the eyecup and placed in hyaluronidase (100 U/mL; catalog no. H-3506) in L-15 solution for 20 min at room temperature. L-15 solution was composed of 70% L-15 (Leibovitz’s medium) and 30% Ca²⁺-free Ringer. Retinas were washed three times for 3 min each in fresh L-15 solution and then placed in L-15 solution containing 7 U/mL papain (catalog no. 3126; Worthington Biochemical Corp., Lakewood, NJ) for 40 min. Papain was previously activated with 2.5 mM l-cysteine. Retinas were again rinsed three times in fresh L-15 solution. Small (~4 mm²) sections of retina were removed and mechanically dissociated by repeated, gentle trituration in L-15 solution. The resulting cell suspension was plated onto uncoated plastic-bottomed dishes (Corning Inc., Bedford, MA) and allowed to settle and adhere for ~20 min before use.

Table 1.

Composition of Ringer and extracellular recording solutions

Reagent	Ca2+-free Ringer, mM	Extracellular Recording Solutions, mM
		Control solution	Ca2+ free	Cd2+	Ba2+	Co2+
NaCl	120	120	120	120	97.5	112.5
KCl	2.6	5	5	5	5	5
CaCl2		2.5		2.5	2.5	2.5
MgCl2		2	3.5	2	2	2
Glucose	10	10	10	10	10	10
HEPES	10	10	10	10	10	10
NaH2PO4	0.5
EGTA			1
CdCl2				0.05
BaCl2					15
CoCl2						5

In all solutions, pH was adjusted to 7.8 with NaOH.

Relative [Ca²⁺]_i measurements.

Relative changes in free [Ca²⁺]_i were assessed by microspectrofluorometric imaging with the membrane-permeant form of the Ca²⁺ indicator fura-2 (Fura-2-LeakRes AM; Teflabs, Austin, TX). Isolated cell preparations were protected from light and incubated in normal extracellular solution (see Table 1) containing 5 µM fura-2 and 0.1% (vol/vol) of a 10% (wt/vol) Pluronic F-127 solution for 30 min at room temperature to facilitate dye loading. Cells were then washed three times in extracellular solution to remove remaining esterified products. Isolated HCs were identified from other isolated cell types by their distinguished large, flat bodies and thick dendrites (Tachibana 1983) under low-intensity bright-field illumination with an upright compound microscope (FN-1; Nikon, Tokyo, Japan). Fluorescence imaging was performed with a Lambda DG-5 wavelength changer (Sutter Instruments, Novato, CA) and a Chroma 79001 filter set (340-nm and 380-nm band-pass filters for excitation, 510-nm bandpass for emission; Chroma Technology, Bellows Falls, VT). Excitation and emission light was passed through a Nikon ×40 water-immersion objective lens (numerical aperture 0.8). Images were collected with a CCD camera (QImaging, Surrey, BC, Canada) by focusing on a region of interest that encompassed an HC soma. Excitation wavelength was iteratively changed between 340 and 380 nm, and emission intensity was recorded for both excitation wavelengths every 2 s with NIS Elements software (Nikon). For some experiments, Northern Eclipse software (Empix Imaging Inc., Mississauga, ON, Canada) was used for data collection. Imaging data were logged in Excel (Microsoft Corp., Redmond, WA).

Relative V_m measurements.

Relative changes in V_m were monitored with a FluoVolt voltage-sensitive dye kit (ThermoFisher Scientific, Waltham, MA). FluoVolt accurately measures relative changes in voltage (see also McPheeters et al. 2017), and product information states that a 25% change in fluorescence represents a change of up to 100 mV. Two microliters of FluoVolt dye and 20 µL of PowerLoad (a dye loading agent) were added to 2 mL of extracellular solution (Table 1). Cells were incubated for 30 min at room temperature, washed three times in extracellular solution, and imaged within 4 h. HCs were identified under low-intensity bright-field illumination with a compound microscope (FN-1; Nikon). A Chroma 41012 filter set was used (480-nm band-pass filter for excitation; 510-nm long pass for emission). Light passed through a Nikon ×40 water-immersion objective lens. Exposure time was 100 ms. Recordings were taken once every second with NIS Elements software (Nikon).

Electrophysiology.

Patch-clamp recordings were taken with the whole cell configuration. Capillary glass (PG52151-4; World Precision Instruments, Sarasota, FL) was pulled on a vertical micropipette puller (PC-10; Narishige International Inc., East Meadow, NY) to make patch electrodes with tip resistances of 5–8 MΩ. Electrodes were filled with intracellular solution consisting of (in mM) 10 NaCl, 120 KCl, 2 CaCl₂, 2 MgATP, 5 EGTA, and 10 HEPES and adjusted to pH 7.4 with KOH. Only cells for which >1-GΩ membrane seals were achieved were used. Preliminary experiments in current-clamp mode indicated that resting V_m was often unstable at current (I) = 0, precluding further analysis. To establish stable recordings of V_m, we first held each cell briefly at a command potential in voltage-clamp mode and recorded the holding current (I_H). Cells were then switched to current clamp, and the value of I_H was injected. This effectively reduced rapid shifts in V_m. All protocols were implemented with pCLAMP 7 software, an Axopatch 1D amplifier (Axon Instruments, Sunnyvale, CA), and a Digidata 1440A (Axon Instruments). Signals were filtered at 5 kHz. All recordings of V_m are corrected for liquid junction potential (V_L), which was subtracted from the pipette potential (V_p) with the formula V_m = V_p – V_L.

Experimental procedures and solutions.

For all experiments, a perfusion chamber (Warner Instruments Inc., Hamden, CT) was inserted into 35-mm culture dishes (Corning), within which cells were plated and maintained. The chamber was continuously perfused at ~2 mL/min by gravity-fed recording solutions at room temperature. Recording solution was removed from the chamber at the same rate with a peristaltic pump (Fisher Scientific Canada, Ottawa, ON, Canada). The involvement of Ca²⁺ channels in the generation of spontaneous events was tested by perfusing the chamber with Ca²⁺-free solution containing 1 mM EGTA, solutions containing divalent cations (Ba²⁺, Cd²⁺, or Co²⁺), or solutions containing the L-type Ca²⁺ channel blocker nifedipine (100 µM; Table 1). The role of internal Ca²⁺ stores was tested with ryanodine (20 μM), dantrolene (50 μM), or caffeine (3 mM). A role for store-operated channels (SOCs) was tested with 2-aminoethyldiphenylborinate (2-APB; 100 μM).

Analysis.

Raw traces from imaging experiments are presented in units of the ratio of fluorescence emission intensity after excitation at 340 nm and 380 nm (F₃₄₀/F₃₈₀). These values are proportional to [Ca²⁺]_i. Parameters chosen to describe each spontaneous event included event frequency, duration, amplitude, area under the curve (AUC), and time to peak (TTP) from baseline level. Baseline [Ca²⁺]_i levels were also analyzed. Ca²⁺ events were analyzed with a peak analysis algorithm in OriginPro 2016 (OriginLab Corp., Northampton, MA). Frequency was measured as the number of events per sampling period (5 min) of observation. Duration was measured as the time between the start and end of each event with a local maximum function compared with baseline. Amplitude was measured as the difference between peak F₃₄₀/F₃₈₀ and baseline before the event, as determined by the Origin peak analysis gadget using a second-derivative algorithm for determining baseline values. AUC (F₃₄₀/F₃₈₀·s⁻¹) was calculated as the integral of amplitude over time of each event. This value approximates Ca²⁺ influx into the cytosol. TTP was the time between the start of the event and peak amplitude. Baseline F₃₄₀/F₃₈₀ measurements were determined as the average of the last 30 s of the algorithm-determined baseline for each period. For amplitude, AUC, and baseline measurements, values were scaled to the largest-magnitude event for each experiment (Lv et al. 2014) because of variability between cells.

All statistical tests for Ca²⁺ imaging were performed in Prism 8 (GraphPad Software Inc., San Diego, CA). The one-tailed Wilcoxon matched-pairs signed-rank test was used for ryanodine and dantrolene experiments. Mann–Whitney tests were used for experiments in which dantrolene was coapplied with 2-APB or nifedipine. Friedman’s and Dunn’s multiple-comparison tests were used for caffeine and 2-APB experiments.

Current-clamp recordings were analyzed with peak analysis algorithms in OriginPro 2016. Duration, amplitude, and AUC were calculated as indicated for Ca²⁺ imaging traces. Frequency of voltage events was reported as the number of events per final 5 min of each treatment. For experiments in Fig. 6, C and D, baseline values were established with OriginPro baseline and peak analysis tools to take an average of membrane potentials within 5-min sampling periods, excluding the duration of spontaneous events. Prism 8 was used for linear regressions. In Fig. 6C, three outliers were excluded from analysis, based on techniques in Motulsky and Brown (2006). In Fig. 7, outliers were omitted for presentation only, including one duration value (303 s) and three AUC values (5,130.97, 5,700.94, and 21,633.6 mV·s). Data sets for Fig. 8 were analyzed in Prism 8 software with a one-way ANOVA, with Bonferroni’s posttest (α = 0.05).

Fig. 6.

Spontaneous depolarizations of goldfish horizontal cells in current clamp. A and B: in whole cell current-clamp recordings, large depolarizations in membrane potential (V_m) are shown in cells from resting V_m of –82 mV (A) and –67 mV (B). Scale bars indicate V_m in millivolts and time in seconds. C: the number of events per 5-min sampling period was greater in more depolarized cells (n = 54, r² = 0.11, P = 0.0158). Line of best fit ± 95% confidence interval is shown. D: the amplitude of spontaneous depolarizations was greater in hyperpolarized cells (n = 49, r² = 0.45, P < 0.0001). Line of best fit ± 95% confidence interval is shown.

Fig. 7.

Frequency distributions of spontaneous depolarization parameters in isolated horizontal cells; 264 spontaneous depolarizations were measured in 57 cells, and the frequency distribution for the following parameters were plotted: frequency of spontaneous events within a 5-min sampling period (A), duration (B), amplitude (C), area under the curve (D), and the time of rise from baseline to peak amplitude (E).

Fig. 8.

Spontaneous depolarizations in current clamp were Ca²⁺ dependent. A: addition of 15 mM Ba²⁺ increased the amplitude, duration, and area under the curve of spontaneous events (n = 6). These effects were reversible upon washout. B: 100 µM nifedipine irreversibly abolished spontaneous events (n = 5). C: addition of 5 mM Co²⁺ also irreversibly abolished spontaneous events (n = 5). D: in 6 cells, Ca²⁺-free solutions abolished spontaneous activity. Scale bars indicate membrane potential in millivolts and time in minutes.

RESULTS

Characterization of spontaneous changes in [Ca²⁺]_i.

In the present study, spontaneous [Ca²⁺]_i events were defined as large, transient increases in whole cell [Ca²⁺]_i that occurred without stimulation in a glutamate-free solution and that recovered to baseline [Ca²⁺]_i levels (Fig. 1, A and B). Spontaneous Ca²⁺ activity of this type was recorded in 157 of 177 cells (89%).

Fig. 1.

Fura-2 measurement of spontaneous changes in intracellular Ca²⁺ concentration ([Ca²⁺]_i) in isolated horizontal cells (HCs). A: sequence of pseudocolor fluorescence images (1–5) indicates a spontaneous change in [Ca²⁺]_i during superfusion of a HC with normal extracellular solution, as determined by the fura-2 emission ratio. Blue/violet indicates lower [Ca²⁺]_i; red/white indicates higher [Ca²⁺]_i. Each frame is separated by 5 s; peak [Ca²⁺]_i was captured in frame 3. Scale bar, 25 µm. B: a representative trace illustrating transient [Ca²⁺]_i changes indicated by fura-2 fluorescence in HCs during superfusion with normal extracellular solution. F₃₄₀/F₃₈₀, fura-2 fluorescence emission ratio. C: representative trace shows a typical response of an HC to brief 100 µM glutamate (glu) application. An initial transient increase in [Ca²⁺]_i decays to an elevated [Ca²⁺]_i that persists for the duration of glutamate application (n = 41). D: superfusion of Ca²⁺-free solution during glutamate application eliminated increased [Ca²⁺]_i (n = 6). A spontaneous Ca²⁺ transient (arrow) was recorded before glutamate application. Scale bars indicate F₃₄₀/F₃₈₀ and time in seconds.

In the retina, HCs receive glutamatergic input from photoreceptors. The [Ca²⁺]_i response of isolated HCs to glutamate application was characterized by an initial increase in [Ca²⁺]_i to a peak amplitude, a decay to an elevated plateau phase that persisted for as long as glutamate was applied, and recovery to baseline levels after glutamate washout (Fig. 1C). This response could be repeated multiple times in the same cell. Removal of extracellular Ca²⁺ (${\text{Ca}}_{\text{e}}^{\text{2+}}$) eliminated glutamate-elicited increases in [Ca²⁺]_i (Fig. 1D). Spontaneous [Ca²⁺]_i events were not observed during glutamate application.

To assess the characteristics of spontaneous [Ca²⁺]_i events, data from event frequency, duration, amplitude, AUC, and TTP were pooled from 279 spontaneous events. Frequency distributions of these data are presented in Fig. 2. Spontaneous events were of low frequency and long duration. Within a 5-min sampling period a median of 7.0 events with an interquartile range (IQR) of 6.0 was calculated (Fig. 2A). Spontaneous events lasted for a median duration of 18.0 s (IQR 10.0 s; Fig. 2B). The amplitude (Fig. 2C) and AUC (Fig. 2D) were obtained from relative changes in the fura-2 emission ratio (F₃₄₀/F₃₈₀) and are proportional to [Ca²⁺]_i. The median amplitude was 0.67 (units of F₃₄₀/F₃₈₀; IQR 0.30) and the AUC was 0.21 (units of F₃₄₀/F₃₈₀·s⁻¹; IQR 0.17). The TTP of each event was 6.0 s (IQR 2.0 s).

Fig. 2.

Frequency distributions of spontaneous Ca²⁺ event parameters in isolated horizontal cells; 279 spontaneous intracellular Ca²⁺ concentration ([Ca²⁺]_i) events were measured in 5 cells, and the frequency distribution for the following parameters were plotted: the frequency of spontaneous events within a 5-min sampling period (A), the duration of spontaneous events (B), the amplitude of spontaneous events in units of the fura-2 fluorescence emission ratio (F₃₄₀/F₃₈₀) (C), the area under the curve of spontaneous events (units of F₃₄₀/F₃₈₀·s⁻¹) (D) and the time of rise from baseline to peak amplitude (s) (E).

Spontaneous [Ca²⁺]_i events required extracellular Ca²⁺.

Figure 3A shows that removal of ${\text{Ca}}_{\text{e}}^{\text{2+}}$ resulted in elimination of spontaneous [Ca²⁺]_i activity and could be achieved with full recovery upon reapplication of normal extracellular solution (n = 18). Reapplication of ${\text{Ca}}_{\text{e}}^{\text{2+}}$ resulted in an initial persistent elevation of [Ca²⁺]_i (Fig. 3A) that eventually recovered to preapplication baseline levels along with the return of spontaneous activity. L-type Ca²⁺ channels are the major type of voltage-gated Ca²⁺ channels (VGCCs) in goldfish HCs (Country and Jonz 2017; Huang and Liang 2005; Tachibana 1983). Blockade of Ca²⁺ influx through these channels was therefore performed with Cd²⁺ (50 µM) or nifedipine (100 µM) in the presence of 2.5 mM ${\text{Ca}}_{\text{e}}^{\text{2+}}$. Extracellular Cd²⁺ reversibly eliminated spontaneous activity (Fig. 3B; n = 5). Nifedipine (100 µM) also eliminated spontaneous activity, which did not recover upon removal of nifedipine solution (Fig. 3C; n = 5). These data suggest that the influx of ${\text{Ca}}_{\text{e}}^{\text{2+}}$ via L-type channels is required for generation of spontaneous Ca²⁺ activity.

Fig. 3.

Spontaneous activity required influx of Ca²⁺. A: removal of extracellular Ca²⁺ (bar) reversibly eliminated spontaneous events (n = 18). Note the persistent elevation of intracellular Ca²⁺ concentration upon replacement of extracellular Ca²⁺ in the superfusate (arrows). B: 50 µM Cd²⁺ (bar) reversibly eliminated spontaneous events (n = 5). C: 100 µM nifedipine (bar) eliminated spontaneous events with no recovery (n = 5). Scale bars indicate the fura-2 fluorescence emission ratio (F₃₄₀/F₃₈₀) and time in seconds.

Ca²⁺ stores are implicated in spontaneous [Ca²⁺]_i events.

Goldfish HCs are known to have both ryanodine receptors (RyRs) and SOCs (Lv et al. 2014). To test for the role of RyRs in spontaneous activity, ryanodine or dantrolene (specific RyR antagonists) was applied in Ca²⁺ imaging experiments. Ryanodine (20 μM) abolished spontaneous activity and increased baseline (n = 8; Fig. 4A). Similarly, 50 μM dantrolene abolished activity and increased posttreatment baseline (n = 7; Fig. 4B). We tested for the source of this increased Ca²⁺ baseline by coapplying 50 µM dantrolene with 100 µM 2-APB (a SOC antagonist) or with 100 µM nifedipine. Baseline Ca²⁺ increased with dantrolene and 2-APB (n = 13, P < 0.05; Fig. 4C), but there was no significant increase with dantrolene and nifedipine compared with controls (n = 9; Fig. 4D).

Fig. 4.

Intracellular Ca²⁺ stores contribute to spontaneous activity. A and B: both ryanodine (20 µM, n = 8; A) and dantrolene (50 µM, n = 7; B) irreversibly abolished spontaneous activity and increased baseline intracellular Ca²⁺ concentration ([Ca²⁺]_i). C: [Ca²⁺]_i baseline increased in response to coapplication of dantrolene and 2-aminoethyldiphenylborinate (2-APB) (100 µM). D: coapplication of dantrolene and nifedipine (100 µM) prevented the dantrolene-induced increase in [Ca²⁺]_i baseline. E: caffeine (3 mM; n = 10; black bar) increased the frequency of spontaneous activity. Coapplication with 100 µM 2-APB (gray bar) prevented this increase. F: summary of the effect of caffeine on frequency. Means ± SD are presented. *Significance (P < 0.05) compared with preapplication values. Scale bars indicate the fura-2 fluorescence emission ratio (F₃₄₀/F₃₈₀) and time in seconds.

At 3 mM, the RyR agonist caffeine significantly increased the frequency of spontaneous [Ca²⁺]_i events (n = 10, P < 0.001; Fig. 4, E and F). In contrast, caffeine’s effects on frequency were negated in the presence of 100 µM 2-APB (n = 10). In fact, coapplication of caffeine and 2-APB reduced frequency to zero in several cells. In contrast, APs persisted in the presence of 2-APB alone, even after 20 min (n = 3). Caffeine also hastened spontaneous events: median duration was reduced from 29.3 s to 18.7 s (n = 10, P < 0.01), and TTP was reduced from 12.3 s to 7.8 s (n = 10, P < 0.05). AUC was reduced from 0.47 to 0.19 (units of F₃₄₀/F₃₈₀·s; n = 10, P < 0.05), but amplitude was not significantly reduced.

Characterization of spontaneous changes in V_m.

We tested whether isolated HCs showed spontaneous depolarizations with FluoVolt, a voltage-sensitive dye. FluoVolt permitted observation of spontaneous activity in cells with an intact cytosol and membrane as in fura-2 experiments. HCs displayed spontaneous increases in fluorescence, consistent with membrane depolarization (n = 6; Fig. 5). During each depolarizing event there was a sharp rise to peak fluorescence, followed by a slow decline and then a steep return to baseline. Mean change in V_m was 10.2 ± 0.6%, and events lasted 8–20 s.

Fig. 5.

FluoVolt voltage-sensitive dye imaging in goldfish horizontal cells. Relative changes in membrane potential were tracked over time (n = 6). Increases in fluorescence represent depolarizing events. Scale bars indicate fluorescence normalized to the first 10 s of recording (F/F₀) and time in seconds; a 5% change in fluorescence corresponds to a change of up to 20 mV (see methods).

In current-clamp experiments, spontaneous changes in V_m, similar to those in FluoVolt experiments, were observed. Most often, however, current injection was required to stabilize and record regenerative changes in V_m. Regenerative APs were observed in 48 of 57 cells (84.2%). These events decayed slowly for several seconds and returned sharply to baseline (Fig. 6, A and B). Both the frequency and amplitude of APs measured under current clamp were voltage dependent. HCs with more negative values of baseline V_m had fewer spontaneous events compared with cells with a more depolarized baseline (n = 54, r² = 0.11, P < 0.05; Fig. 6C). Events reached higher amplitudes in cells at relatively hyperpolarized V_m (n = 49, r² = 0.45, P < 0.01; Fig. 6D), likely because of a greater driving force.

Frequency, duration, amplitude, AUC, and TTP were pooled from 264 spontaneous events. Frequency distributions of these data are presented in Fig. 7. Spontaneous events were of low frequency and long duration. Within a 5-min sampling period, a median of 3.0 (IQR 4.0) events were recorded (Fig. 7A). Spontaneous events lasted for a duration of 11.5 s (IQR 13.9 s; Fig. 7B). Median amplitude was 61.9 mV (IQR 26.6 mV; Fig. 7C), and median AUC was 442.8 mV·s (IQR 704.8 mV·s; Fig. 7D). The median TTP was 1.8 s (IQR 1.7 s; Fig. 7E).

Action potentials required extracellular Ca²⁺.

To test for Ca²⁺ channel activity, Ba²⁺ was applied during current-clamp experiments. In L-type and other high-voltage-activated (HVA) Ca²⁺ channels, Ba²⁺ increases current as it is more permeant through HVA channels than Ca²⁺ (Budde et al. 2002). Compared with Ca²⁺, 15 mM Ba²⁺ increased AUC 3.8-fold and increased duration 3.0-fold (n = 6, P < 0.05; Fig. 8A). In separate experiments, 5 mM Co²⁺ (a nonspecific Ca²⁺ channel blocker) or 100 μM nifedipine was applied to test for L-type channel involvement. Both drugs irreversibly abolished activity in all cells (n = 5; Fig. 8, B and C). In experiments in which Ca²⁺ was removed, spontaneous activity was abolished irreversibly or returned with a much smaller amplitude (Fig. 8D).

DISCUSSION

This study has demonstrated the spontaneous generation of APs in isolated retinal HCs of the goldfish by imaging changes in [Ca²⁺]_i and V_m and by using electrophysiological techniques. We have shown that APs are dependent upon Ca²⁺ and are driven by activity of plasma membrane Ca²⁺ channels and intracellular RyRs. We have demonstrated that AP activity in HCs is maintained by prolonged membrane currents and is modulated by changes in membrane potential.

In most previous reports, APs recorded from HCs were not spontaneous but were instead elicited with a depolarizing stimulus (e.g., Johnston and Lam 1981; Shingai and Christensen 1986) or were potentiated with K⁺ channel blockers and Ba²⁺ solutions (Murakami and Takahashi 1987). Tachibana (1981) was able to elicit APs with depolarizing current injections but also found that ~50% of cells exhibited spontaneous APs, which he attributed to natural fluctuations in resting V_m. Additionally, spontaneous Ca²⁺ transients have been observed in isolated goldfish HCs (Kreitzer et al. 2012).

In all three techniques used in the present study (Ca²⁺ imaging, whole cell current-clamp recording, and FluoVolt imaging) spontaneous events possessed similar characteristics. These included a steep rise in [Ca²⁺]_i (or depolarization of V_m), a slow decay period, and a steep return to baseline. However, Ca²⁺ transients were slower than changes in V_m (duration: 18.0 s vs. 11.5 s), took longer to reach their peak (TTP: 6.0 s vs. 1.8 s), and were more frequent (7.0 vs. 3.0 per 5-min sampling period). One possible explanation is that during the recovery phase of APs V_m returns to resting levels before [Ca²⁺]_i. The slower characteristics of Ca²⁺ transients, compared with changes in V_m, may also be attributed to the effects of fura-2 during imaging. Fura-2 is a Ca²⁺ buffer, so it is expected to reduce free [Ca²⁺]_i and delay Ca²⁺-dependent inactivation (CDI) of VGCCs (Budde et al. 2002; Tachibana 1983; Tse et al. 1994).

Ca²⁺ dynamics during action potentials.

Previous reports, in which APs were elicited by depolarization, found that APs depended on extracellular Ca²⁺. Removing extracellular Ca²⁺ or blocking HVA Ca²⁺ channels with divalent ions (e.g., Cd²⁺ or Co²⁺) abolished APs, whereas HVA channel-permeant ions (e.g., Ba²⁺, Sr²⁺) or high extracellular [Ca²⁺] amplified APs (Johnston and Lam 1981; Shingai and Christensen 1986; Tachibana 1981; Takahashi et al. 1993). Just as these depolarization-induced events were dependent on Ca²⁺, we found that spontaneous Ca²⁺ transients and changes in V_m were abolished in the presence of divalent VGCC blockers and nifedipine and that they were amplified and prolonged in the presence of Ba²⁺. Our work links spontaneous and induced APs and extends previous results by confirming that L-type channels are responsible for Ca²⁺ influx.

Ca²⁺ stores in goldfish HCs are gated by RyRs (Huang et al. 2004; Lv et al. 2014) and are refilled, at least in part, by SOCs (Lv et al. 2014). RyRs mediate Ca²⁺-induced Ca²⁺ release (CICR) in HCs by releasing Ca²⁺ from the endoplasmic reticulum after cytosolic [Ca²⁺] increases (Country and Jonz 2017). In the present study, RyR blockade abolished spontaneous events, followed by an irreversible increase of baseline [Ca²⁺]_i. This rise in [Ca²⁺]_i was likely due to L-type Ca²⁺ currents, because coapplication of nifedipine with dantrolene prevented the increase in baseline [Ca²⁺]_i. In contrast, inhibition of SOCs with 2-APB in the presence of dantrolene did not prevent the rise in [Ca²⁺]_i.

This [Ca²⁺]_i increase during RyR blockade might be explained by the relationship between VGCCs and CICR, as follows. During APs, Ca²⁺ enters via L-type Ca²⁺ channels and triggers CICR through RyR-gated stores. Then, CICR inhibits L-type channels through CDI (Linn and Gafka 2001). But during ryanodine or dantrolene blockade, CICR would not occur and L-type channels would not be inhibited, letting baseline [Ca²⁺]_i rise.

Caffeine (a RyR agonist) reversibly increased the frequency of spontaneous events and also reduced event duration and TTP. Because caffeine and elevations in [Ca²⁺]_i increase the probability that RyRs will open (Masumiya et al. 2001), CICR may occur faster in the presence of caffeine, increasing [Ca²⁺]_i enough to inactivate L-type channels through CDI (Budde et al. 2002; Linn and Gafka 2001) and rapidly returning [Ca²⁺]_i to baseline levels. A previous report found that 2-APB, a SOC inhibitor, prevents Ca²⁺ stores from refilling and prevents caffeine-induced CICR (Lv et al. 2014). If Ca²⁺ stores are depleted, SOCs in the plasma membrane can open, allowing Ca²⁺ entry to refill stores (Hogan and Rao 2007). In the present study, removing and reapplying ${\text{Ca}}_{\text{e}}^{\text{2+}}$ led to a prolonged elevation of [Ca²⁺]_i (Fig. 3A). In separate experiments, when caffeine was coapplied with 2-APB the frequency of Ca²⁺ transients was not affected. These experiments are consistent with findings from Lv et al. (2014) that suggest that SOCs are present in goldfish HCs and confirm that Ca²⁺ stores contribute to spontaneous Ca²⁺ transients. Stored Ca²⁺ may be important for generating APs in HCs because of the limited availability of Ca²⁺ in the synaptic cleft. The volume of the invaginating synaptic cleft has been calculated to be miniscule (3 × 10⁻¹⁸ L; Raviola and Gilula 1975; Vessey et al. 2005), so that an extracellular [Ca²⁺] of 2 mM would give ~3,600 ions of free Ca²⁺. Ca²⁺ influx during an AP could therefore greatly decrease extracellular [Ca²⁺]. A similar argument has been made for the importance of Ca²⁺ stores and CICR in maintaining tonic glutamate release from photoreceptor terminals (Rabl and Thoreson 2002; Suryanarayanan and Slaughter 2006). In addition, CICR synchronizes Ca²⁺ oscillations among cardiomyocytes (Dougoud et al. 2016; Plummer et al. 2011). This raises the possibility that in the retina CICR may also synchronize spontaneous Ca²⁺ activity throughout HC syncytia by triggering Ca²⁺ flux into nearby cells through gap junctions.

What is the role of action potentials in horizontal cells?

HCs receive glutamatergic input from photoreceptors, which depolarizes HCs and triggers Ca²⁺ influx (Country and Jonz 2017; Thoreson and Mangel 2012). We observed spontaneous events in vitro in the absence of glutamate. Tachibana (1981) suggested that APs are the result of natural fluctuations in resting V_m. He proposed that V_m would fluctuate into the activation range of VGCCs, opening them and triggering an AP. In the present study, we found that cells with a more depolarized resting V_m had a higher frequency of APs and that all cells observed had a resting V_m (–90 to –50 mV) that was more negative than the activation threshold for HC L-type channels (–45 to –30 mV; Country and Jonz 2017; Shingai and Christensen 1983; Sullivan and Lasater 1992; Tachibana 1983). By extension of Tachibana’s model, our findings show that subthreshold resting V_m in HCs may indeed fluctuate until it reaches the L-type activation threshold, increasing the probability that L-type channels will open and initiate an AP. We observed fluctuations in resting V_m of up to 30 mV in current-clamp and voltage-sensitive dye experiments, and these fluctuations would put V_m well within range of L-type channel activation. Although our experiments were performed in vitro, our data suggest that in the intact retina light-dependent changes in glutamate release from photoreceptor terminals will modulate HC V_m, and these changes will accordingly control AP frequency.

Furthermore, since glutamate seems to suppress or occlude spontaneous APs, this suggests that APs may be more prevalent (though at lower frequency) in the retina during light stimulation, when glutamate release from photoreceptors is reduced and V_m in HCs is hyperpolarized.

In the intact retina, HCs are hyperpolarized by light (approximately –60 mV) and are depolarized in darkness (approximately –25 mV; Sun et al. 2017; Yang et al. 1988). APs may be important in the light-to-dark transition by rapidly shifting V_m to the more depolarized potential. This may be the case in the turtle. In sharp electrode recordings of HCs in red-eared turtle (Pseudemys scripta elegans) eyecups, off-responses (i.e., the response to removal of a light stimulus) included regenerative changes in V_m, similar to an AP (Akopian et al. 1991). Applying Co²⁺ or removing extracellular Ca²⁺ abolished regenerative off-responses, and it took longer for HCs to resettle to a depolarized baseline. However, to our knowledge, this has not yet been tested in fish.

Understanding Ca²⁺-based APs may offer novel insights into HC physiology, including their role in processing visual stimuli. HCs provide inhibitory feedback to photoreceptors (Baylor et al. 1971; Thoreson and Mangel 2012) via one of three mechanisms: 1) during light stimulation, HCs may reduce GABA release onto photoreceptors (GABA disinhibition) (Hirano et al. 2016; Liu et al. 2013; Wu and Dowling 1980); 2) hemichannels at HC dendrites may affect photoreceptor membrane potential via an ephaptic mechanism (Byzov and Shura-Bura 1986; Kamermans et al. 2001); and/or 3) HCs may affect pH or proton buffering (Barnes 2003; Vroman et al. 2014; Warren et al. 2016) in the synaptic cleft that may, in turn, affect the conductance and activation of photoreceptor VGCCs (Barnes 2003; Hirasawa and Kaneko 2003; Vessey et al. 2005). Generation of Ca²⁺-based APs might affect any of these mechanisms. For example, GABA release is partially Ca²⁺ dependent in nonmammalian species (Ayoub and Lam 1985; Cunningham and Neal 1985; Lasater and Lam 1984; Thoreson and Mangel 2012). In addition, during APs Ca²⁺ influx would likely diminish extracellular [Ca²⁺]. Low extracellular [Ca²⁺], in turn, is known to increase hemichannel open probability (DeVries and Schwartz 1992; Jonz and Barnes 2007; Sáez et al. 2005) and increase gap junctional coupling among HCs (McMahon and Mattson 1996). In isolated HCs spontaneous Ca²⁺ transients cause extracellular acidification (Kreitzer et al. 2007, 2012), so that Ca²⁺ APs may lead to photoreceptor inhibition.

Conclusions.

We have characterized spontaneous, Ca²⁺-based APs in isolated HCs of the goldfish retina. We propose a model in which V_m fluctuations are large enough to open L-type VGCCs, which further depolarize the cell. The resulting Ca²⁺ influx triggers CICR through RyR-gated intracellular Ca²⁺ stores, which may in turn inactivate L-type channels and allow for repolarization. We propose that APs in HCs (Murakami and Takahashi 1987; Shingai and Christensen 1983; Tachibana 1981) and spontaneous Ca²⁺ events (Kreitzer et al. 2012) represent the same phenomenon, by comparing event characteristics, Ca²⁺ dependence, and the involvement of L-type Ca²⁺ channels. This work will be instrumental in exploring the role that spontaneous activity in HCs may have in vision.

GRANTS

This research was supported by the Natural Sciences and Engineering Research Council of Canada (grant nos. 342303 and 05571, M. G. Jonz), the Canadian Foundation for Innovation, and the Ontario Research Fund (grant no. 16589, M. G. Jonz).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.W.C., B.F.N.C., and M.G.J. conceived and designed research; M.W.C. and B.F.N.C. performed experiments; M.W.C., B.F.N.C., and M.G.J. analyzed data; M.W.C., B.F.N.C., and M.G.J. interpreted results of experiments; M.W.C., B.F.N.C., and M.G.J. prepared figures; M.W.C., B.F.N.C., and M.G.J. drafted manuscript; M.W.C., B.F.N.C., and M.G.J. edited and revised manuscript; M.W.C., B.F.N.C., and M.G.J. approved final version of manuscript.