Abstract
It is a commonly accepted view that light stimulation of mammalian photoreceptors causes a graded change in membrane potential instead of developing a spike. The presynaptic Ca2+ channels serve as a crucial link for the coding of membrane potential variations into neurotransmitter release. Cav1.4 L-type Ca2+ channels are expressed in photoreceptor terminals, but the complete pool of Ca2+ channels in cone photoreceptors appears to be more diverse. Here, we discovered, employing whole-cell patch-clamp recording from cone photoreceptor terminals in both sexes of mice, that their Ca2+ currents are composed of low- (T-type Ca2+ channels) and high- (L-type Ca2+ channels) voltage-activated components. Furthermore, Ca2+ channels exerted self-generated spike behavior in dark membrane potentials, and spikes were generated in response to light/dark transition. The application of fast and slow Ca2+ chelators revealed that T-type Ca2+ channels are located close to the release machinery. Furthermore, capacitance measurements indicated that they are involved in evoked vesicle release. Additionally, RT-PCR experiments showed the presence of Cav3.2 T-type Ca2+ channels in cone photoreceptors but not in rod photoreceptors. Altogether, we found several crucial functions of T-type Ca2+ channels, which increase the functional repertoire of cone photoreceptors. Namely, they extend cone photoreceptor light-responsive membrane potential range, amplify dark responses, generate spikes, increase intracellular Ca2+ levels, and boost synaptic transmission.
SIGNIFICANCE STATEMENT Photoreceptors provide the first synapse for coding light information. The key elements in synaptic transmission are the voltage-sensitive Ca2+ channels. Here, we provide evidence that mouse cone photoreceptors express low-voltage-activated Cav3.2 T-type Ca2+ channels in addition to high-voltage-activated L-type Ca2+ channels. The presence of T-type Ca2+ channels in cone photoreceptors appears to extend their light-responsive membrane potential range, amplify dark response, generate spikes, increase intracellular Ca2+ levels, and boost synaptic transmission. By these functions, Cav3.2 T-type Ca2+ channels increase the functional repertoire of cone photoreceptors.
Introduction
In the outer segment of cone photoreceptors, incident light is absorbed by photopigments, which trigger phototransduction and generate a photocurrent (Baylor et al., 1987; Schnapf et al., 1987). The photocurrent is further processed at the inner segment by voltage-sensitive channels (Barnes and Hille, 1989; Barnes, 1994) to generate photovoltage at the cone photoreceptor synaptic terminal. Previously identified voltage-gated channels of cone photoreceptors are the delayed rectifier K+ channels, the fast transient K+ channels, the hyperpolarization-activated cation channels, and the high-voltage-activated L-type Ca2+ channels (Barnes and Hille, 1989; Maricq and Korenbrot, 1990; Yagi and Macleish, 1994; Schneeweis and Schnapf, 1995; Gayet-Primo et al., 2018). Based on the repertoire of the expressed voltage-sensitive ion channels, it is generally believed that depolarization of the cone photoreceptor membrane does not result in spiking behavior, but rather in a gradual potential change (Barnes and Hille, 1989; Yagi and Macleish, 1994; Schneeweis and Schnapf, 1995; Masland, 2012). On the other hand, there is physiological evidence for fast, spike-like currents in monkey rod photoreceptors (Schnapf et al., 1990). Additionally, Na+ action potentials were also described in human cone and rod photoreceptors from detached retinal tissue (Kawai et al., 2001, 2005), but there is no evidence for Na+ action potentials in healthy, uncompromised photoreceptors.
Electrophysiological studies have shown the presence of high-voltage-activated (HVA) L-type Ca2+ channels, but not of other types of Ca2+ channels, in cone photoreceptors of multiple species (Barnes and Hille, 1989; Lasater and Witkovsky, 1991; Wilkinson and Barnes, 1996). Immunocytochemical labeling of the outer plexiform layer of the retina with antibodies against the pore-forming subunits of the HVA Cav1.3 (α1D) and Cav1.4 (α1F) channels indicated the existence of both Ca2+ channel types in photoreceptors (Nachman-Clewner et al., 1999; Morgans, 2001; Morgans et al., 2001; Specht et al., 2009; Kersten et al., 2010). Antibodies against Cav1.3 labeled a subtype of cone photoreceptors in the tree shrew retina, which were probably the middle wavelength-sensitive ones (Taylor and Morgans, 1998; Morgans, 1999). Immunoelectron microscopy also suggested the presence of Cav1.3 channels in mouse cone photoreceptor synaptic terminals (Kersten et al., 2010), but the ERG b-wave was not altered in a mutant Cav1.3 mouse model (Wu et al., 2007). By contrast, experiments using a mutant Cav1.4 mouse model show a dramatically reduced ERG b-wave, and in humans, the mutation of the Cav1.4 channel causes congenital stationary night blindness. This demonstrates that the main Ca2+ channel of photoreceptors is Cav1.4 (Strom et al., 1998; Mansergh et al., 2005). However, a Cav1.4 mutation in mouse cone photoreceptors did not significantly change tonic synaptic vesicle release (Zanetti et al., 2021), indicating the presence of another, possibly non-L-type voltage-sensitive Ca2+ channel, in cone photoreceptors.
The voltage dependence of Ca2+ channels in cone photoreceptors of nonmammalian vertebrates is considered to be similar to that in rod photoreceptors (Bader et al., 1982; Kaneko and Tachibana, 1986). Additionally, previous electrophysiological measurements of cone photoreceptor Ca2+ currents (ICa) in mice showed a similar voltage dependence to rod photoreceptors (Hirano et al., 2016; Babai et al., 2019; Grassmeyer et al., 2019). Whole-cell patch-clamp recordings of cone photoreceptors usually target the cell bodies or inner segments, which are separated from the synaptic terminals by a relatively long axon. In previous works, the signal flow through cone photoreceptor axons showed a short (Lasater et al., 1989) and also a long (Bryman et al., 2020) length constant in turtle and macaque retina, respectively. As the length constant is not determined in mouse cone photoreceptors, the anatomic arrangement may limit electrical access to voltage-gated channels at the synapse where voltage-gated Ca2+ channels are clustered (Xu and Slaughter, 2005; Choi et al., 2008; Mercer et al., 2011). In this study, we characterized the voltage dependence of mouse cone photoreceptor ICa by recording directly from their synaptic terminals using low-resistance (∼10 MΩ) patch pipettes.
Materials and Methods
Mice.
Adult (age, 2–5 months) male and female C57BL/6 mice (The Jackson Laboratory) and Tg(Rac3-EGFP)JZ58Gsat/Mmcd (Rac3-eGFP; Landgraf et al., 2012; Regus-Leidig et al., 2013) mice were used for the electrophysiological and fluorescence-activated cell sorting (FACS) experiments. Rac3-eGFP mice, expressing enhanced green fluorescent protein (eGFP) in all cone photoreceptor cells were used. These mice were obtained from the Mutant Mouse Regional Resource Center (MMRRC), a National Center for Research Resources-NIH-funded strain repository and were donated to the MMRRC by the National Institute of Neurological Disorders and Stroke-funded GENSAT (Gene Expression Nervous System Atlas) BAC Transgenics Project. The Rac3-eGFP construct was generated by inserting an eGFP reporter gene, followed by a polyadenylation sequence, into the bacterial artificial chromosome (BAC) clone RP23-62A17 at the initiating ATG codon of the first coding exon of the Rac3 gene. Consequently, eGFP expression was driven by the regulatory sequence of the Rac3 gene. The resulting modified BAC (BX1967) was used to isolate the transgene. The animals were kept in a 12 h light/dark cycle. All experiments were performed in compliance with the guidelines for the welfare of experimental animals issued by the Federal Government of Germany and the University of Erlangen-Nürnberg.
Fluorescence-activated cell sorting of photoreceptors.
For sorting of photoreceptors, retinae of Rac3-eGFP mice were dissociated by papain digestion (10 U/ml; Worthington Biochemical) at 37°C for 20 min and subsequent trituration. After washing with FACS buffer (5 mm EDTA in 0.1 m PBS), pH 7.4, the dissociated cells were treated with RNase A (0.9 U; Carl Roth) in FACS buffer at 37°C for 5 min, for degrading the RNA of damaged cells. After DAPI wash, the cells were resuspended in FACS buffer with RiboLock RNase inhibitor (20 U; Thermo Fisher Scientific) for inhibiting remaining RNase activity. Cells were sorted in a FACS Aria III (BD Biosciences) with an 85 µm nozzle at the Chair of Genetics, Friedrich-Alexander University Erlangen-Nürnberg and directly collected in RLT Buffer (Qiagen) containing 1% β-mercaptoethanol. Cone photoreceptors were sorted by cone-specific fluorescence of Rac3-eGFP mice. Rod photoreceptors were sorted by forward and sideward scatter in an approach adapted from Feodorova et al. (2015). The population identified by forward scatter (FSC)/sideward scatter (SSC) was positive for CD73, a surface marker of cone/rod photoreceptor common precursors and mature rod photoreceptors (Koso et al., 2009), corroborating this sorting strategy. For the figure, images were created using FlowJo Software version 10.6.2 for Mac (BD).
RT-PCR.
Total RNA of sorted photoreceptors and whole retinae were isolated with the RNeasy Micro and Mini Kit (Qiagen), respectively. RNA was reverse transcribed to cDNA using the iScript cDNA Synthesis Kit (BIO-RAD). To exclude amplification of genomic DNA, control experiments were performed in which reverse transcriptase was omitted. Possible contamination artifacts were excluded by replacing the template with water. Both controls were negative. Table 1 summarizes the primers used for amplification. The low-voltage-activated (LVA) primers have been successfully applied for the investigation of mouse islet cells (Vignali et al., 2006). For RT-PCR, degenerated primers which bind to every known LVA channel (Cav3.1, Cav3.2, Cav3.3) were used. Subsequently, specific primer pairs for each LVA Ca2+ channel were used for a nested PCR. For all PCRs, a standard PCR protocol with Taq DNA Polymerase (Qiagen) was used. Amplicon sizes were verified on 1.5% agarose gels stained with EtBr (Carl Roth). Images were arranged using CorelDRAW 2019 (Corel).
Primer pairs used for RT-PCR
Electrophysiology and slice preparation.
First, mice were sedated with isoflurane (3% inhalant) and killed by cervical dislocation. The preparation of horizontal retina slices has been described in detail in a previous article (Feigenspan and Babai, 2017). Eyes were enucleated, then the retina was removed from the eyeball and cut into six to eight pieces. Retina pieces were embedded into 1.8% low-melting agarose dissolved in Ames' Medium (Sigma-Aldrich). Horizontal slices, 160–180 µm thick, were cut with a vibratome (Leica Microsystems), and kept in Ames' Medium at 37°C in an incubator containing 5% CO2 and 55% O2 for 20–30 min. For light-response experiments, mice were dark adapted for at least 1 h. Then, retinal slices and patch-clamp experiments were performed under dim red light (background illumination, 1 mW/m2).
Retinal slices were observed with a 63× water-immersion objective (Zeiss) using a fixed-stage microscope (Axio Examiner, Zeiss) equipped with Dodt contrast. Whole-cell currents were recorded from cone photoreceptor terminals in a horizontal slice of the retina with an EPC-10 Patch-Clamp Amplifier (Heka Elektronik), low-pass filtered at 2.9 kHz using a built-in Bessel filter, and digitized at 10 kHz with Patchmaster software (Heka Elektronik). Voltages were corrected for liquid junction potentials (9 mV). Recordings were conducted at room temperature (22–24°C) and near body temperature (∼33°C). Patch electrodes were pulled from borosilicate glass (Sutter Instrument) to a final resistance of 8–12 MΩ. Electrode tips were coated with Sylgard 184 (Dow Corning), and their series resistance (10–30 MΩ) was compensated up to 66%. Electrodes were positioned with NMN-21 micromanipulators (NARISHIGE Group). Retinae were superfused in every physiological experiment with 8 µm 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) to block the activation of postsynaptic AMPA and kainate receptors. We used the VM8 Perfusion System (ALA Scientific Instruments) for the local application of agonist and antagonist substances. The puffing solution, similar to the bath solution, always contained 8 µm CNQX. Preparations were continuously superfused at ∼1 ml/min with extracellular solution containing the following (in mmol/L–1): 116 NaCl, 22.6 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 2 CaCl2, 1 MgCl2, 10 glucose, and 5 HEPES, at pH 7.4, bubbled with 95% O2/5% CO2. The intracellular solution contained the following (in mmol/L–1): 136.6 Cs-gluconate, 5 EGTA, 13 tetraethylammonium chloride (TEA-Cl), 15 HEPES, 4 Mg-ATP, and 0.4 GTP, at pH 7.2. K-based intracellular solution contained the following (in mmol/L–1): 134.5 K-gluconate, 10 KCl, 10 HEPES, 5 EGTA, 1 CaCl2, 1 MgCl2, 0.5 Na-GTP, and 4 Mg-ATP, at pH 7.2. All chemical reagents were obtained from Sigma-Aldrich except when specified otherwise.
Capacitance measurements were made using the Sine + DC technique, the lockin extension of the EPC-10 amplifier (sine wave frequency, 800 Hz; peak amplitude, 30 mV). We blanked output from the phase-lock amplifier for 10 ms after the step and began measurements 10 ms later to circumvent any influence of gating charges and let time for the phase angle feedback circuitry to settle. Recordings were only evaluated if access resistance was <38 MΩ and holding current was under −30 pA at holding potential (Vh) = −60 mV. Average resting membrane capacitance (Cm), membrane resistance, and access resistance values for cone photoreceptors were 3.53 ± 0.18 pF, 1148.16 ± 125.80 MΩ, and 32.47 ± 2.76 MΩ, respectively. Full-field light stimulation (∼130 W/cm2 irradiance) was generated by a xenon arc lamp of the lambda DG4 illumination system (Sutter Instrument). Light signals were conveyed to the sample via fiber optics.
To distinguish cone from rod photoreceptors we performed whole-cell recordings and compared the resting Cm and ICa amplitude of the two photoreceptor types. We found no overlap in the distribution of Cm and ICa (Cm_rod, 1.32 ± 0.21 pF; Cm_cone, 3.53 ± 0.18 pF; ICa_rod, −8.78 ± 0.43 pA; ICa_cone, −36.04 ± 2.31 pA; n = 10 and 26, respectively). Additionally, we compared HVA and LVA ICa values recorded from fluorescently labeled cone photoreceptor terminals of Rac3-eGFP mice and from cone photoreceptor terminals of C57BL/6 mice. We found nonsignificant differences [LVA current (ILVA)_Rac: −11.37 ± 0.86 pA; ILVA_WT: −11.40 ± 0.80 pA; p = 0.9846; HVA current (IHVA)_Rac: −35.52 ± 1.55 pA; IHVA_WT: −36.04 ± 2.31 pA; p = 0.8681; n = 9 and 13, respectively, unpaired t test]. These experiments strongly indicate that the identification of cone photoreceptors in our experiments was successful.
Statistical analysis.
Statistical analysis and data visualization were done using GraphPad Prism (GraphPad Software) and Igor Pro 6.3 (WaveMetrics, Inc.). Data are reported as the mean ± SEM, and n indicates the number of cells measured. For each experiment, at least three mice were used. If the statement of normality was met by the Kolmogorov–Smirnov test (p > 0.05) to data points, mean values were compared using a paired two-sample t test. The Mann–Whitney U test was used when the normal distribution assumption was not encountered. For clearness of illustration, current traces were low-pass filtered at 1 kHz. In all figures, significance was identified as follows: *p < 0.05, **p < 0.01, and ***p < 0.001.
Results
Mouse cone photoreceptors exhibit ILVA and IHVA
First, we examined the voltage dependence of ICa in mouse cone photoreceptors. We prepared horizontal slices of the retina (Feigenspan and Babai, 2017) in which a horizontal cut was made through the outer plexiform layer to reveal cone photoreceptor terminals (Fig. 1A, white arrowhead). Cone photoreceptor terminals were discriminated from rod photoreceptor terminals (Fig. 1A, black arrowhead) by their size, anatomic localization, and ICa amplitude (see Materials and Methods). Cone photoreceptor terminals are approximately threefold bigger and localize closer to the inner retina than the relatively small rod photoreceptor terminals. Additionally, cone photoreceptors display ICa amplitudes that are six to eight times bigger than that of rod photoreceptors. To isolate ICa, we used a cesium and TEA-Cl-based intracellular solution with 2 mm CaCl2 in the bath solution as the charge carrier. To exclude feedback from horizontal cells, the extracellular solution contained 8 µm CNQX, which is a potent AMPA and kainate receptor antagonist (Feigenspan and Babai, 2015; Babai et al., 2016). For the recording, we used pipettes with ∼10 MΩ resistance pulled from borosilicate glass (Sutter Instrument), and we achieved a final access resistance of 20–30 MΩ. Series resistance was compensated up to 66%. We rejected recordings with a larger than −30 pA leak current at −69 mV Vh. Cone photoreceptors were held at −69 mV, and ICa was measured by using a ramp voltage protocol from −89 mV (800 ms) to +49 mV with a speed of 0.1875 mV/ms (Fig. 1B1,B2). Surprisingly, the ICa trajectory showed two peaks, one at a lower (Fig. 1B2, LVA) and one at a higher (Fig. 1B2, HVA) membrane potential (Vm) range. The double peak seen in the current–voltage profile of cone photoreceptor ICa is comparable to ICa in certain types of retinal bipolar and horizontal cells where both HVA (L-type) and LVA (T-type) Ca2+ channels are present (Hu et al., 2009; Feigenspan et al., 2020). Therefore, we next measured ICa using the same voltage ramp protocol but starting from a more depolarized Vh value of −39 mV (800 ms). At this slightly depolarized Vm, LVA T-type Ca2+ channels typically show strong inactivation (Perez-Reyes, 2003), while HVA L-type Ca2+ channels only inactivate moderately (Taylor and Morgans, 1998). Consequently, only IHVA should appear in the current trajectory. As a result, we found that the amplitude of ILVA significantly decreased, but that of IHVA did not change (Fig. 1B2,C1,C2). The peak of ILVA and IHVA localized at approximately −40 mV and approximately −20 mV, respectively (Fig. 1C3), and the presence of the two current types yielded a wide range of responsive Vm. The first peak suggested the presence of T-type Ca2+ channels because the typical T-type ICa reaches its maximal amplitude between −40 and −10 mV (Ertel et al., 1997). However, unlike T-type ICa in other tissues, ILVA was not fully inactivated at −39 mV, a potential that is close to the physiological Vm in darkness (−40 mV). In summary, patch-clamp recordings from the terminal of cone photoreceptors indicate the unexpected presence of an ILVA, suggesting the expression of T-type Ca2+ channels in cone photoreceptors additionally to L-type Ca2+ channels.
The presence of ILVA and IHVA in cone photoreceptors. A, Example transmission microscope image of a patch-clamp recording from a cone photoreceptor terminal in a mouse horizontal retinal slice. White arrowhead, Cone photoreceptor terminal; black arrowhead, rod photoreceptor terminal. B1, Example whole-cell patch-clamp recording from mouse cone photoreceptor terminals. Currents were monitored in response to a voltage ramp protocol from −89 to +49 mV, with a speed of 0.1875 mV/ms. Prepulse duration was 800 ms at different Vh values (black, −89 mV; gray, −39 mV). B2, Example trace shows the current–voltage relationship of ICa. Arrowheads indicated ILVA and IHVA components. C1–C3, Peak amplitudes of ILVA: Vh = −89 mV: −12.72 ± 2.15 pA, n = 12; Vh = −39 mV: −6.90 ± 1.53 pA, n = 12; p < 0.0001, paired t test. Peak amplitudes of IHVA: Vh = −89 mV: −19.66 ± 3.33 pA, n = 12; Vh = −39 mV: −20.19 ± 3.45 pA, n = 12; p = 0.4511, paired t test. ICa peak position: LVA: −42.48 ± 0.46 mV, n = 12; HVA: −20.75 ± 0.71 mV, n = 12; p < 0.0001, Mann–Whitney test.
Pharmacological isolation of LVA and HVA Ica
The pore-forming subunits of the T-type Ca2+ channel family consist of three members: Cav3.1 (α1G), Cav3.2 (α1H), and Cav3.3 (α1I), which show similar activation and inactivation kinetics (Klöckner et al., 1999; Perez-Reyes, 2003). Nickel blocks all three T-type Ca2+ channel members at higher concentrations, but it selectively blocks Cav3.2 channels at lower concentrations (Lee et al., 1999, 2002; Kang et al., 2006). We applied 100 µm nickel and stimulated cone photoreceptors with a ramp voltage protocol described in Figure 1. Example ICa responses are illustrated in Figure 2A1. We used a pressure-controlled puffing system (model ALA-VM8, ALA Scientific Instruments), which allowed the fast and local application of nickel and CNQX (8 µm) containing solution through a 75-µm-diameter tube directly onto cone photoreceptor terminals. Control ICa was measured during the puff application of the bath solution to exclude changes caused by mechanical effects. We found that the application of nickel caused a ∼70% and a ∼50% reduction of IHVA and ILVA components, respectively (Fig. 2A1,A2). Nickel is a commonly described ion that blocks LVA Ca2+ channels, but HVA Ca2+ channels, especially L-type Ca2+ channels, also show nickel sensitivity (Zamponi et al., 1996; Bradley et al., 2004; To et al., 2020). Therefore, the reduction of the IHVA component is unsurprising. To further analyze the origin of the ILVA component, we puffed 1 µm mibefradil, a T-type Ca2+ channel blocker (Martin et al., 2000), onto cone photoreceptor terminals. However, we only found a slight tendency for reduced amplitudes for both ILVA and IHVA (Fig. 2B1,B2). We did not use a higher concentration of mibefradil to isolate T-type Ca2+ channels because it has been shown that mibefradil is also an antagonist for L-type Ca2+ channels (Leuranguer et al., 2001; Lee et al., 2006; To et al., 2020), and Cl–, K+, and Na+ channels (Heady et al., 2001). Next, we tested 5 µm Z944, a selective T-type Ca2+ channel antagonist (Casillas-Espinosa et al., 2015), which significantly decreased ILVA amplitude but unexpectedly increased the IHVA component (Fig. 2C1,C2), revealing Z944 as a possible agonist for L-type Ca2+ channels in cone photoreceptors. In general, the sensitivity of ILVA to nickel and Z944 suggests the presence of T-type Ca2+ channels in cone photoreceptors, and the sensitivity of IHVA to Z944 suggests that the L-type Ca2+ channel in cone photoreceptors may have unique molecular properties.
Pharmacological separation of ILVA from IHVA. A1, Example current–voltage relationship of ramp-evoked ICa in mouse cone photoreceptors in the presence and absence of 100 μm NiCl2. Arrowheads indicate ILVA and IHVA component (LVA, HVA). A2, Peak ILVA amplitude: control: −11.03 ± 1.80 pA, n = 5; nickel (Ni2+): −6.22 ± 1.13 pA, n = 5; p = 0.0030, paired t test. Peak IHVA amplitude: control: −37.45 ± 5.25 pA, n = 5; nickel (Ni2+): −11.01 ± 1.52 pA, n = 5; p = 0.0022, paired t test. B1, Example current–voltage relationship of evoked ICa in the presence and absence of 1 μm mibefradil (Mibe). B2, Peak ILVA amplitude: control: −11.15 ± 0.94 pA, n = 6; mibefradil: −9.81 ± 0.93 pA, n = 6; p = 0.0641, paired t test. Peak IHVA amplitude: control: −37.04 ± 3.48 pA, n = 6; mibefradil: −35.06 ± 3.27 pA, n = 6; p = 0.1712, paired t test. C1, Example current–voltage relationship of evoked ICa in the presence and absence of 5 μm Z944. C2, Peak ILVA amplitude: control: −12.51 ± 1.58 pA, n = 7; Z944: −6.05 ± 0.82 pA, n = 7; p = 0.0003, paired t test. Peak IHVA amplitude: control: −36.61 ± 4.06 pA, n = 7; Z944: −49.25 ± 6.04 pA, n = 7; p = 0.0371, paired t test. D1, Current response to a voltage ramp protocol from −89 to +61 mV, with a speed of 0.1875 mV/ms in the presence (gray) and absence (black) of 3 mm CsCl. Prepulse duration was 800 ms. D2, Example trace shows current–voltage relationship of ICa during 3 mm CsCl application. D3, Peak ILVA amplitude: control: −15.72 ± 0.87 pA, n = 7; CsCl: −15.71 ± 1.25 pA, n = 7; p = 0.9786, paired t test. Peak IHVA amplitude: control: −29.42 ± 3.71 pA, n = 7; CsCl: −32.09 ± 4.05 pA, n = 7; p = 0.0933, paired t test. E1, Example current–voltage relationship of evoked ICa in the presence and absence of 10 μm nifedipine (Nife). E2, Peak IHVA amplitude: control: −37.45 ± 5.25 pA, n = 5; nifedipine: −19.73 ± 6.42 pA, n = 5; p = 0.0022, paired t test. Peak ILVA amplitude: control: −10.83 ± 0.87 pA, n = 5; nifedipine: −10.86 ± 1.00 pA, n = 5; p = 0.9082, paired t test. F1, Example current–voltage relationship of evoked ICa in the presence and absence of 1 μm TTX. F2, Peak ILVA amplitude: control: −12.80 ± 0.26 pA, n = 5; TTX: −12.92 ± 0.47 pA, n = 5; p = 0.7628, paired t test. Peak IHVA amplitude: control: −34.51 ± 0.81 pA, n = 5; TTX: −33.61 ± 1.02 pA, n = 5; p = 0.2789, paired t test. G1, Example current–voltage relationship of evoked ICa in the presence and absence of 10 μm ω-conotoxin GVIA (Cono). G2, Peak ILVA amplitude: control: −12.05 ± 1.77 pA, n = 5; Cono: −11.81 ± 2.06 pA, n = 5; p = 0.7978, paired t test. Peak IHVA amplitude: control: −31.37 ± 2.92 pA, n = 5; Cono: −31.03 ± 3.60 pA, n = 5, p = 0.7205; paired t test. H1, Example current–voltage relationship of evoked ICa in the presence and absence of (3 mm CsCl, 10 μm nifedipine, 5 μm Z944 and 100 μm nickel). H2, Peak ILVA amplitude (LVA), −10.72 ± 1.53 pA; peak IHVA amplitude (HVA), −40.16 ± 5.78 pA; peak ICa amplitude in the presence of L-type and T-type Ca2+ channel blockers (Mix), −3.0 ± 0.33 pA. LVA versus mix, p = 0.005; HVA versus mix, p = 0.0028; n =5, unpaired t test.
Next, we confirmed whether other non-T-type Ca2+ channels contribute to ILVA. We tested whether the hyperpolarization-activated cation current (Ih), which is responsible for shaping the photoreceptor light response (Akopian and Witkovsky, 1996), affects Ca2+ currents by puff application of 3 mm CsCl (Fig. 2D1,D2). We found no significant change in ILVA and IHVA amplitudes (Fig. 2D3), suggesting that the measured ILVA was because of T-type Ca2+ channels and not hyperpolarization-activated cation channels. To inhibit the IHVA component, we used 10 µm nifedipine, an L-type Ca2+ channel blocker. Nifedipine has a high affinity to L-type and a very low affinity to T-type Ca2+ channels (Stengel et al., 1998). Nifedipine significantly blocked the IHVA component, but left the ILVA component untouched (Fig. 2E1,E2). This indicates that the ILVA component was not because of the activation of HVA L-type Ca2+ channels, further providing evidence for the presence of LVA T-type Ca2+ channels in mouse cone photoreceptors. Finally, we verified whether tetrodotoxin (TTX)-sensitive Na+ channels or ω-conotoxin GVIA-sensitive N-type Ca2+ channels influenced ILVA or IHVA. During puff application of 1 µm TTX (Fig. 2F1,F2) or 1 µm ω-conotoxin GVIA (Fig. 2G1,G2), we found no significant difference in IHVA and ILVA Ca2+ current amplitudes (Fig. 2F2,G2). Measuring ICa in the presence of a mixture of antagonistic compounds, including T-type and L-type Ca2+ channel blockers (10 µm nifedipine, 5 µm Z944, and 100 µm nickel additionally to 8 µm CNQX and 3 mm CsCl), resulted in a rather small residual current (Fig. 2H1,H2), which indicates a minimal nonspecific effect. In conclusion, the results from our pharmacological experiments strongly indicate the presence of T-type Ca2+ channels in addition to L-type Ca2+ channels in mouse cone photoreceptors.
Biophysiological characterization of ILVA and IHVA
In the next set of experiments, we measured the electrophysiological properties of ILVA and IHVA components. ILVA was isolated by puff application of 8 µm CNQX, 3 mm CsCl, and 10 µm nifedipine containing extracellular solution, thereby removing any postsynaptic feedback, Ih, and IHVA components. In this condition, ILVA of cone photoreceptors was recorded in response to a series of voltage step stimulations (50 ms) from Vh = −69 mV (Fig. 3A). Furthermore, based on the results of our pharmacological experiments (Fig. 2), we chose 8 µm CNQX, 3 mm CsCl, and 5 µm Z944 to isolate IHVA. In this condition, IHVA produced typical high-voltage-activated current trajectories in response to a series of voltage steps (Fig. 3B). The average IHVA and ILVA amplitude relationships to Vm (I–V) in cone photoreceptors are illustrated in Figure 3C1. Relative conductance values of IHVA and ILVA were fitted with a Boltzmann function to estimate the voltage dependency of the activation (Fig. 3C2). As expected, the half-activation (V50) of IHVA shifted to the right by ∼10 mV compared with V50 of ILVA, but slope values were comparable to each other (Fig. 3C3). Additionally, the time course of activation and inactivation of ILVA was measured on maximal current amplitude traces, which were chosen from the step series. Maximal ILVA amplitudes were typically evoked by a voltage step from −69 to −41 mV (50 ms; Fig. 3D1). Fast inactivation was estimated by fitting the decay phase of the current with an exponential function (Fig. 3D1) and found to be <30 ms (Fig. 3D2). Time to ILVA peak values stayed at ∼13 ms (Fig. 3D2). In general, the time courses of activation and inactivation kinetics of cone photoreceptor ILVA were very similar to currents produced by other neuronal T-type Ca2+ channels (Klöckner et al., 1999). The time course of IHVA inactivation was measured as the ratio of IHVA amplitude at the beginning and the end of the current trace (Fig. 3E1). IHVA showed an ∼15% reduction during a voltage step to −17 mV (50 ms) from Vh = −69 mV (Fig. 3E2). HVA L-type Ca2+ currents typically show little fast inactivation in photoreceptors (Barnes and Hille, 1989; Yagi and Macleish, 1994). To determine the steady-state inactivation properties of ILVA and IHVA in cone photoreceptors, currents were elicited by test pulses either to −39 mV (ILVA) or −19 mV (IHVA) preceded by a series of conditioning potential steps ranging from −100 to 0 mV (duration, 300 ms; Fig. 3F,G). The inactivation was calculated by dividing the test pulse-evoked current amplitudes with maximal current amplitudes (I/Imax; Fig. 3H1). Data points were fitted by a Boltzmann function. Results indicated that V50 parameters of IHVA inactivation were shifted to the right by ∼15 mV, while slope values stayed similar (Fig. 3H2). Interestingly, the activation and inactivation curves of both ILVA and IHVA revealed a noninactivating current window in the range of the physiological Vm (−60 to −30 mV) where channel inactivation is incomplete and thus regulates steady-state Ca2+ levels (Fig. 3H1).
Physiological properties of ILVA and IHVA. A, Representative examples of ILVA in cone photoreceptors evoked by 50 ms voltage steps to different potentials (Vh = −62, −38, −32, −20, −8, and 52 mV) in the presence of 8 μm CNQX, 3 mm CsCl, and 10 μm nifedipine. B, Representative examples of IHVA evoked by 50 ms voltage steps to different potentials (Vh = −62, −38, −32, −20, −8, and 52 mV) in the presence of 8 μm CNQX, 3 mm CsCl, and 5 μm Z944. C1, Current–voltage relationship of ILVA and IHVA currents evoked by voltage steps from −77 to 43 mV by 6 mV steps. C2, Voltage dependence of steady-state activation of ILVA and IHVA (g/gmax). Data points were fitted with the Boltzmann function (dashed lines). C3, Mean V50: ILVA, −44.17 ± 1.98 mV; IHVA, −34.38 ± 1.78 mV; n = 6; p = 0.0043, unpaired t test. Mean activation slope factor (slope): ILVA, −4.77 ± 0.70; IHVA, 4.99 ± 0.83; n = 6; p = 0.8417, unpaired t test. D1, Typical maximal ILVA in cone photoreceptors evoked by a voltage step from −69 to 41 mV (50 ms). The dashed red line indicates ILVA decay fitted with an exponential function. D2, Corresponding time to ILVA peak, 13.18 ± 0.81 ms. Decay, tau of single exponential function: 28.82 ± 2.38 ms. E1, Typical maximal IHVA in cone photoreceptors evoked by a voltage step from −69 to −17 mV (50 ms). E2, Corresponding fast inactivation ratio, 0.85 ± 0.03. F, Example traces showing the steady-state inactivation of ILVA. A 300-ms-long conditioning pulse over a series of Vh, from −100 to −20 mV (20 mV increments), was applied before the test pulse (−39 mV, 50 ms). G, Example traces showing the steady-state inactivation of IHVA. The 300-ms-long conditioning pulse over a series of Vh values from −100 to −20 mV (20 mV increments) was applied before the test pulse (−19 mV, 50 ms). H1, Steady-state inactivation of ILVA and IHVA were calculated by normalizing the test pulse-evoked current amplitude to the maximum current amplitude and were plotted over the prepulse Vm. Data points were fitted with Boltzmann function and illustrated as solid black (IHVA) and gray (ILVA) lines. Dashed lines are the Boltzmann fits of the ILVA and IHVA activation kinetics from C2. H2, Mean V50: ILVA, −54.82 ± 0.81 mV; IHVA, −40.11 ± 2.76 mV; n = 8 and n = 5; p < 0.0001, unpaired t test. Mean inactivation slope factor (slope): ILVA, 5.06 ± 0.49; IHVA, 5.45 ± 0.87; n = 8 and n = 5; p = 0.6773, unpaired t test.
Ca2+ spikes are present in cone photoreceptors
LVA T-type Ca2+ channels are suggested to play a crucial role in the activity of several types of neurons in the CNS by generating low-threshold spikes and activating higher-threshold ion channels, which induce a burst of firing and oscillatory behavior (Perez-Reyes et al., 1998; Perez-Reyes, 2003). To investigate the impact of T-type Ca2+ channels on cone photoreceptor Vm changes, we performed whole-cell current-clamp recordings. Because the Cs+-based and TEA-Cl-based pipette solution used in the previous experiments blocks K+ currents, thereby increasing the input resistance of the cells and the probability of spiking, we examined whether cone photoreceptors generate spikes in the presence of a K+-gluconate-based intracellular solution. In the current-clamp configuration, we first injected steady currents to keep Vm hyperpolarized (approximately −60 mV) and then depolarized cone photoreceptors stepwise by steady current injection to mimic different levels of light exposure. Figure 4A1 shows an example of the Vm changes of a cone photoreceptor in response to a current injection series from −24 to +6 pA (Δ5 pA, 1 s). When cone photoreceptor Vm was held at a hyperpolarized level of approximately −60, and approximately −50 mV by steady current injection, we did not observe any spike behavior. When Vm reached a threshold of approximately −40 mV, Vm showed a fast rise, which generated an immediate large spike (arrowhead) and additional smaller spikes with some delay (Fig. 4A1). The amplitude of a spike is classically dependent on how long the responsible ion channels are open at the rising phase. Therefore, for the higher spike peaks L-type Ca2+ channels could be responsible as they do not inactivate strongly (Fig. 3E). To determine the ion channel involvement in the observed spikes, we performed puff application of several pharmacological agents during the current injections described above. We could not evoke spikes during the puff application of 100 µm nickel onto the cone photoreceptor terminal (Fig. 4A2). In this condition, the Vm at the beginning of the stimuli stayed well below that seen in control conditions, and the injected current–Vm plot stayed rather linear in the range of physiological Vm (Fig. 4A3). We also tested the effect of Z944 on cone photoreceptor spike generation. The 5 µm Z944 blocked ILVA but increased IHVA in the previous experiment (Fig. 2C1,C2). Here, spikes could still be generated by current injections; however, peak Vm stayed slightly, but significantly, below control values (Fig. 4B1–B3). This suggests that an increased level of HVA L-type Ca2+ current could partially compensate for the reduced level of LVA T-type Ca2+ current considering spike generation. This assumption is also supported by the results of the next experiment, where the application of the L-type Ca2+ channel blocker nifedipine (10 µm) blunted the initial Vm rise (Fig. 4C1,C2). However, peak Vm values (Fig. 4C3, gray trace) were less linear than what was observed in the presence of nickel (Fig. 4A3, gray trace). The slight depolarization remaining in the presence of nifedipine at more than −40 mV Vm suggests that the activation of T-type Ca2+ channels helps to boost the rising phase of the depolarization to then activate L-type Ca2+ channels. Next, we tested for the possible role of Na+ channels in spike generation. During puff application of 1 µm TTX, the depolarization evoked by a series of current injections was not altered (Fig. 4D1–D3). Additionally, by puffing 3 mm CsCl, we also found that Ih did not significantly influence cone photoreceptor spiking behavior (Fig. 4E1–E3). Altogether, the experiments indicate that the spiking behavior in cone photoreceptors at depolarized Vm is because of the cooperative activation of both T-type and L-type Ca2+ channels.
Ca2+ spikes present at cone photoreceptors. A1, Example of cone photoreceptor Vm change to current injections (6, 1, −4, −9, −14, and −24 pA) in current-clamp mode using K+-gluconate-based intracellular pipette solution. Arrowhead labels the initial fast rise of the Vm (spike) at the beginning of the stimulus. A2, Example trace of cone photoreceptor Vm change to the same level of current injections in the presence of 100 μm nickel (Ni2+). A3, Vm amplitude versus stimulus intensity plot. Significance was tested with a paired t test among each stimulus intensity (−22 pA, p = 0.7346; −18 pA, p = 0.1548; −14 pA, p = 0.2918; −10 pA, p = 0.3324; −6 pA, p = 0.0389; −2 pA, p < 0.0001; 2 pA, p < 0.0001; 6 pA, p = 0.2151; n = 6). B1, B2, Example traces of cone photoreceptor Vm change to current pulses (0, −5, −10, −15, −20, and −25 pA), in the absence (control, B1) and presence (gray, B2) of 5 μm Z944. B3, Vm amplitude versus stimulus intensity plot. Significance was tested with a paired t test in each stimulus intensity (−25 pA, p = 0.9846; −20 pA, p = 0.9583; −15 pA, p = 0.9467; −10 pA, p = 0.1408; −5 pA, p = 0.3731; 0 pA, p = 0.0485; 5 pA, p = 0.0498; 10 pA, p = 0.2916; n = 7). C1, C2, Example traces of cone photoreceptor Vm change to current pulses (2, −4, −10, −16, and −22 pA), in the absence (control; C1) and presence (Nife; gray; C2) of 10 μm nifedipine. C3, Vm amplitude versus stimulus intensity plot. Significance was tested with a paired t test in each stimulus intensity (−35 pA, p = 0.1143; −30 pA, p = 0.1352; −25 pA, p = 0.1362; −20 pA, p = 0.9151; −15 pA, p = 0.5775; −10 pA, p = 0.0150; −5 pA, p = 0.0199; 0 pA, p = 0.050; n = 6). D1, D2, Example traces of cone photoreceptor Vm change to current pulses (0, −8, −12, −16, and −24 pA), in the absence (control; D1) and presence (TTX; gray; D2) of 1 μm tetrodotoxin. D3, Vm amplitude versus stimulus intensity plot. Significance was tested with paired t test in each stimulus intensity (−25 pA, p = 0.296; −20 pA, p = 0.1322; −15 pA, p = 0.2485; −10 pA, p = 0.1641; −5 pA, p = 0.6813; 0 pA, p = 0.1703; 5 pA, p = 0.3681; n = 4). E1, E2, Example traces of cone photoreceptor Vm change to current pulses (2, −4, −10, −13, and −19 pA), in the absence (control; E1) and presence (Cs+; gray; E2) of 1 μm CsCl. E3, Vm amplitude versus stimulus intensity plot. Significance was tested with paired t test in each stimulus intensity (−25 pA, p = 0.2575; −20 pA, p = 0.2498; −15 pA, p = 0.2172; −10 pA, p = 0.1939; −5 pA, p = 0.1366; 0 pA, p = 0.0712; 5 pA, p = 0.0512; n = 5).
Self-generating spike behavior at depolarized Vm
In the next set of experiments, we asked whether cone photoreceptors generate spikes spontaneously at dark Vm. Therefore, we injected a steady current for 10 s in the current-clamp configuration to keep the Vm of cone photoreceptors close to −40 mV, a typical dark Vm. As in the previous experiment, we used K+-gluconate-based intracellular pipette solution to keep input resistance more physiological. Remarkably, in this condition, cone photoreceptors repeatedly generated spikes. Figure 5A1 shows a representative example of cone photoreceptor spiking behavior when cone photoreceptor Vm was adjusted to approximately −40 mV by injecting −3 pA current. Puff application of 100 µm nickel onto the cone photoreceptor terminal blocked spikes completely in seven of seven cells (Fig. 5A2), and the effect could be washed out after a few minutes. The application of 10 µm nifedipine similarly blocked spike activity in six of six cells (Fig. 5B2), and the effect could be washed out after several minutes. Additionally, we tested the effect of 5 µm Z944 on spontaneous spike generation in cone photoreceptors. In this condition, spikes could still be observed with a frequency similar to control conditions (Fig. 5C1–C3), but the peak amplitude and the half-width of the spikes were significantly altered (Fig. 5C3). These results fit well with the examined effect of Z944 on ILVA and IHVA (Fig. 2). In summary, the effect of Z944 on spike kinetics and amplitude and the elimination of spontaneous spikes by nickel and nifedipine indicate that both L-type and T-type Ca2+ channels are responsible for the spontaneous spike generation in cone photoreceptors. Because spikes are typically associated with Na+ channels, we also examined whether 1 µm TTX inhibits spike generation. We found no significant difference in spike frequency (p = 0.4712, n = 4) and peak amplitude (p = 0.3820, n = 4) compared with control conditions. Examination of spontaneous spike behavior in the presence of 3 mm CsCl also showed no significant changes in spike frequency and peak amplitude (p = 0.2641, p = 0.2131, respectively, n = 4), suggesting no role of Ih in spike generation. In I = 0 recording mode, cone photoreceptors showed an average Vm of −37.01 ± −4.23 mV (n =11). At this potential, cone photoreceptors also consistently generated spontaneous spikes. In general, the analysis of control spikes indicated an average frequency of ∼1.5 Hz, a peak amplitude of approximately −13.5 mV, a baseline of approximately −40.5 mV, and a half-width of ∼32 ms (Fig. 5D). These parameters are different from Na+ action potential parameters measured in a typical mammalian central neuron (Bean, 2007). To further analyze the kinetics of the spontaneous spike events in cone photoreceptors, we picked representative, individual spikes from the recordings at depolarized Vm (approximately −40 mV), as illustrated in Figure 5E1. Initially, we created trajectory plots of spike rate of change (Fig. 5E2), and we measured the rise rate of Vm. We determined the spike threshold as 7.5% of the maximal rise rate at 0.22 mV/ms (Fig. 5E3, gray horizontal line). This value yielded an average of −36.2 ± 1.05 mV (Fig. 5E4) for the activation threshold of spontaneous spikes (gray horizontal line), which is very close to the typical dark Vm (Ingram et al., 2020). We found a maximal rise rate of ∼3.1 mV/ms (Fig. 5E4). This value is 50–150 times slower than was measured for Na+ action potentials (Hodgkin and Huxley, 1952), further strengthening the finding that spikes in cone photoreceptors are not generated by Na+ channels. Altogether, these experiments showed that mouse cone photoreceptors are able to develop self-generating spike activity around the dark Vm. A plausible explanation for this oscillatory behavior in spike generation is the periodic activation and inactivation of LVA T-type Ca2+ channels, which further activate L-type Ca2+ channels (Perez-Reyes, 2003).
Spontaneous spike activity at dark Vm. A1–C2, Example Vm changes in control condition and during 100 μm nickel (Ni2+), 10 μm nifedipine (Nife), and 5 μm Z944 application. C3, Peak Vm of spontaneous spike events: control, −13.3 ± 2.50 mV; Z944, −19.45 ± 3.13 mV; n = 7; p = 0.0337, paired t test. Spontaneous spike frequency: control, 1.45 ± 0.38 Hz; Z944, 1.1 ± 0.32 Hz; n = 7; p = 0.2462, paired t test. Half-width: control, 38.0 ± 2.50 ms; Z944, 50.68 ± 3.13 ms; p = 0.0145, paired t test; n = 7. D, Summary of spike parameters generated in control condition. Frequency, 1.47 ± 0.23; peak Vm, −13.56 ± 1.72 mV; baseline Vm, −40. 57 ± 0.50 mV; half-width, 31.96 ± 1.72 ms. E1, Example of a spontaneous spike in cone photoreceptors. E2, Time derivative of the Vm of the trace in E1 plotted against Vm. E3, Zoomed in region of the rising phase from E2. Dashed gray line, Exponential fit to estimate the average of the trace; straight gray lines, 7.5% of the maximal rise rate reveals the Vm threshold of the spike. E4, Vm at 0.22 dV/dt: −36.17 ± 0.29 mV; n = 14; maximal Vm rise rate: 3.17 ± 0.24 mV/ms; n = 14. F1, Example Vm changes at near body temperature (33°C). F2, Frequency of spikes, 1.05 ± 0.17 Hz; peak Vm, −10.65 ± 2.05 mV; baseline, −35.52 ± 1.1 mV; half-width, 27.77 ± 4.3 ms. Room temperature versus 33°C: frequency, p = 0.2610; peak Vm, p = 0.3458; baseline, p = 0.0002; half-width, p = 0.4844, unpaired t test; n = 12 and n = 5, respectively. G1, Example trace showing Vm changes evoked by full-field light flash (300 ms, ∼130 W/cm2 irradiance) in dark-adapted cone photoreceptor. G2, Vm before light stimulation (Dark Vm), −39.65 ± 0.99 mV; light response evoked Vm hyperpolarization (Vmin), −53.96 ± 2.46 mV; offset of the light response (Vmax), −3.61 ± 4.08 mV.
Another important question is whether these Ca2+ spikes have physiological relevance. Therefore, we investigated spontaneous spike behavior near body temperature (33°C; Fig. 5F1,F2). Spontaneous spikes were also evoked at higher temperature by current injection values that were used during the experiments in room temperature (room temperature, −1.41 ± 0.74 pA; 33°C, −2.8 ± 1.02 pA; p = 0.1349, n = 12 and n = 5, respectively, unpaired t test). We also tested whether spikes could be generated by natural stimuli. We performed current-clamp (I = 0) experiments in dark-adapted retinae using dim red light during slicing and recording. Full-field light flashes (irradiance, ∼130 W/cm2) evoked a typical light response in cone photoreceptors and an additional spike at light offset (Fig. 5G1). Cone photoreceptors initially stayed at Vm = approximately −39 mV (Fig. 5G1, Dark Vm), then the light flash induced a fast hyperpolarization to approximately −54 mV (Fig. 5G2, Vmin). Spikes evoked by light offset reached a peak approximately Vm = 0 mV (Vmax), as observed for spikes generated by current injections (Fig. 4, controls). These experiments illustrate that mouse cone photoreceptors generate Ca2+ spikes in dark and at light offset. Therefore, cone photoreceptors seem to process light information by both gradual Vm change and spike generation.
Spike activity increases intracellular Ca2+ levels in cone photoreceptor terminals
We showed that cone photoreceptors are able to generate spikes in response to a Vm jump and spontaneously at depolarized Vm and provided pharmacological evidence that spikes were produced by the activation of voltage-sensitive T-type and L-type Ca2+ channels. Here, we wanted to further investigate whether spikes stimulate the influx of Ca2+ into the cone photoreceptor terminal. Therefore, to detect spatially averaged intracellular Ca2+ changes, we used 100 µm Fluo-4 Ca2+ indicator (Thermo Fisher Scientific) in the pipette solution. Fluo-4 has been shown to be sufficiently sensitive (Kd = 345 nm) to detect low concentrations of free intracellular Ca2+ (Gee et al., 2000). In the experiments, we targeted cone photoreceptor terminals in the horizontal retina slice preparation with the patch electrodes and measured Ca2+ signals parallel to current-clamp measurements. First, we injected negative currents to keep the cells at hyperpolarized Vm, then we injected positive currents to depolarize the cells to a level that is close to the threshold of spike generation (approximately −40 mV). Figure 6A shows an example of a cone photoreceptor held at approximately −80 mV by injecting −13 pA current followed by a Vm step to more positive values for 5 s by injecting −4 pA current. In this example, the cone photoreceptor showed spikes at the start of the stimulation but shortly after (∼2 s), Vm remained steady at approximately −50 mV without spikes (Fig. 6A, black trace). The Vm of −50 mV is more negative than the activation threshold of the T-type Ca2+ channels (approximately −40 mV). Accordingly, intracellular Ca2+ concentration (ΔF/F0) rose at the start of the stimulus and then declined slowly to baseline values. This indicates that the current injection activated voltage-sensitive Ca2+ channels at the beginning of the stimulus, which resulted in an intracellular Ca2+ rise in the terminal, but because of the absence of further spike generation at −50 mV, Ca2+ was slowly removed from the terminal. In the next example, Vm was changed to a more depolarized level by a current step to −2 pA (Fig. 6B). During the presence of this depolarizing current, Vm of the cone photoreceptor showed continuous spike activity at ∼1 Hz, and Ca2+ signals were elevated to a steady level. Additionally, because of the spike activity, a clear concurrent rise of the Ca2+ signal was observed (Fig. 6B, arrows). The relationship between spikes and intracellular Ca2+ concentration was analyzed by measuring the linear correlation between spike timing and Ca2+ signal peaks (Fig. 6C). We found a significant correlation (p < 0.0001, R2 = 0.9992), indicating that spikes were generated by Ca2+. We found an average ΔF/F0 of 1, and the Ca2+ signal peak could be detected ∼30 ms after spike peaks (Fig. 6D). Altogether, these experiments show that during self-generating spike activity at around dark Vm, the intracellular Ca2+ concentration rises from baseline level and stays at a steady high level, while individual spikes can further increase Ca2+ levels temporarily in cone photoreceptor terminals. Additionally, the local Ca2+ concentration rise suggests that both T-type and L-type Ca2+ channels localize at the cone photoreceptor terminal membrane.
Spikes cause intracellular Ca2+ rise in cone photoreceptor terminals. A, Example of simultaneous current-clamp recording (from −13 to −4 pA, 5 s) and Ca2+ imaging with 100 μm Fluo-4 in the cone photoreceptor terminal. Pictures in the top panel were taken from the patched cone photoreceptor terminal at the time points labeled with arrows (t1–t3). B, Example of simultaneous current-clamp recording and Ca2+ imaging during stronger stimulation (from −13 to −2 pA, 5 s). Arrows indicate Ca2+ rise because of the individual spike events. C, Linear relationships between the timing of spike and Ca2+ signal peak (p < 0.0001, R2 = 0.9992). D, Relative Ca2+ change during spikes (ΔF/F0), 1.08 ± 0.33; Δt between spike peak amplitude and Ca2+ signal peak (Δt), 30.58 ± 8.36 ms.
Both LVA and HVA Ca2+ channels contribute to synaptic vesicle release
To investigate the role of LVA T-type Ca2+ channels in synaptic vesicle release, first, we took advantage of the biophysical properties of the channels. At depolarized Vm (−41 mV), ILVA showed partial (∼46%) inactivation, but IHVA inactivation was only ∼15% (Fig. 3). We therefore, compared synaptic vesicle release evoked by a test stimulus (−19 mV, 25 ms) preceded by an ∼15-s-long, either hyperpolarized (Vh = −69 mV; Fig. 7A1, black) or depolarized (Vh = −39 mV, Fig. 7A1, gray) potential. Evoked synaptic vesicle release in cone photoreceptors was monitored by Cm measurements with the Sine + DC method (HEKA 10 Amplifier, HEKA Elektronik). We found that ICa charge (QICa) was ∼1.5 pC in the presence of LVA and HVA Ca2+ channels (Vm = −69 mV). When Vm was held at −39 mV, QICa decreased to ∼0.8 pC, significantly diminishing evoked synaptic vesicle release (Fig. 7A2). The number of released synaptic vesicles in cone photoreceptors was estimated by taking into account the single synaptic vesicle capacitance contribution, which has a typical value of 43.7 aF in mouse photoreceptors based on a 37.3 nm synaptic vesicle diameter (Fuchs et al., 2014). This yields ∼1730 synaptic vesicles at Vh = −69 mV and ∼1030 synaptic vesicles at Vh = −39 mV; thus, ∼41% fewer synaptic vesicles were released by the inactivation of LVA T-type Ca2+ channels. This result indicates that LVA T-type Ca2+ channels contribute to Ca2+-dependent synaptic vesicle release at cone photoreceptors. Next, we examined the contribution of HVA L-type Ca2+ channels to evoked synaptic vesicle release by the application of 10 µm nifedipine that reduced IHVA by ∼68% and left ILVA unaltered (Fig. 2E1,E2). Synaptic vesicle release was triggered by a test pulse (−19 mV, 25 ms) from Vh = −69 mV (Fig. 7B1), and nifedipine was applied with a pressure-controlled puffing system. QICa showed a significant, ∼49%, reduction in the presence of 10 µm nifedipine, but evoked synaptic vesicle release was only reduced by ∼16% (Fig. 7B2). This result indicates that LVA T-type Ca2+ channels contribute more to synaptic vesicle release than HVA L-type Ca2+ channels (Fig. 7A1) and therefore have a boosting effect on synaptic vesicle release evoked by a strong stimulus (−19 mV, 25 ms) at mouse cone photoreceptors. As this is an unexpected finding, we further examined the effect of LVA T-type Ca2+ channels on synaptic vesicle release at cone photoreceptors. In the following experiments, we included 10 mm bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA) in the pipette solution. BAPTA is a Ca2+ buffer with a faster on-rate than EGTA has, intercepting Ca2+ influx at the active zone via Ca2+ channels to reveal nanodomain or microdomain organization (Adler et al., 1991). The presence of BAPTA in the pipette did not change QICa compared with EGTA (5 mm EGTA; Fig. 7B, black: QICa = 1.16 ± 0.19 pC, n = 6; 10 mm BAPTA; Fig. 7C, black: QICa = 1.28 ± 0.10 pC, n = 5, p = 0.6236, unpaired t test). However, 10 mm BAPTA caused a significant reduction in ΔCm (5 mm EGTA, Fig. 7B, black: 90.00 ± 10.73 pC, n = 6; 10 mm BAPTA, Fig. 7C, black: 35.16 ± 5.18 pC, n = 5, p = 0.0018, unpaired t test). The ∼61% reduction of Cm indicates that most of the evoked synaptic vesicle release was triggered by Ca2+ influx through nanodomain-organized Ca2+ channels. Moreover, 39% of the evoked synaptic vesicle release in cone photoreceptors is either not sensitive to BAPTA or is Ca2+ independent. Nifedipine application, in the presence of BAPTA, caused a reduction of QICa (Fig. 7C2) similar to what we observed with the EGTA-containing intracellular solution (Fig. 7B2). However, ΔCm values decreased by ∼48% in the presence of the fast Ca2+ buffer (Fig. 7C2), which exceeded the degree of reduction that we measured in the presence of 5 mm EGTA (∼16%; Fig. 7B2). The stronger nifedipine sensitivity of the release in the presence of BAPTA indicates that most of the HVA L-type Ca2+ channels are organized into nanodomains at the active zone (Bartoletti et al., 2011). Moreover, the smaller nifedipine sensitivity in the presence of EGTA (Fig. 7B,C, gray) suggests that LVA T-type Ca2+ channels provide an additional Ca2+ source to trigger and boost synaptic vesicle release and that they locate close to the release sites. The mild stimuli evoked QICa in the presence of 5 mm EGTA and 10 mm BAPTA reached similar values, but the synaptic vesicle release was significantly different (Fig. 7D). The reduced ΔCm levels with BAPTA and the maintained ΔCm levels with EGTA further strengthen the finding that LVA T-type channels are present close to the release sites and are able to boost synaptic vesicle release. To further examine the boosting effect of LVA T-type Ca2+ channels, we measured cone photoreceptor synaptic vesicle release in response to mild (−39 mV, 25 ms) and strong (−19 mV, 25 ms) depolarizing steps. Stimuli were applied pairwise to the same cone photoreceptor using a pipette solution containing either 5 mm EGTA or 10 mm BAPTA (Fig. 7E,F). At Vm = approximately −39 mV, the LVA T-type Ca2+ current has a maximal amplitude, while HVA L-type Ca2+ channels are activated only slightly (Figs. 1, 3). Consequently, mild stimulation should activate mainly LVA T-type Ca2+ channels, and strong stimulation should activate both Ca2+ channel types. We found that a mild stimulus significantly diminished QICa compared with the strong stimulus in both Ca2+ buffering conditions (Fig. 7E,F). However, in the presence of 5 mm EGTA, the mild stimulus reduced synaptic vesicle release by only ∼12% (Fig. 7E), which is consistent with the result shown in Figure 7B, where mainly LVA T-type Ca2+ channels dominated evoked synaptic vesicle release. This shows that LVA T-type Ca2+ channels are capable of triggering synaptic vesicle release with high efficiency (∼88%) when photoreceptors are quickly depolarized from the light to the dark membrane potential (−39 mV). On the other hand, the comparison of mild and strong depolarization steps in the presence of 10 mm BAPTA revealed an ∼70% difference in evoked synaptic vesicle release (Fig. 7F). This demonstrates that LVA T-type Ca2+ channels are less effective in boosting synaptic vesicle release when the fast Ca2+ buffer is present; hence, they likely locate close to HVA L-type Ca2+ channels.
LVA Ca2+ channels boost synaptic vesicle release. A1, Black, Example traces showing ICa and Cm induced by a voltage step (from −69 to −19 mV, 25 ms) at cone photoreceptors. Stim, Stimulus; Gm, membrane conductance; Gs, series conductance; gray, ICa and Cm induced by a voltage step (from −39 to −19 mV, 25 ms). Labeling similar to black traces. A2, QICa: Vh = −69 mV: 1.35 ± 0.19 pC; Vh = −39 mV: 0.83 ± 0.06 pC; n = 7; p = 0.0003, paired t test; Cm: Vh = −69 mV: 75.62 ± 4.69 fF; Vh = −39 mV: 45.02 ± 2.90 fC; n = 7; p = 0.0003, paired t test. B1, Black, Example traces of ICa and Cm jump induced by voltage stimulus (from −69 to −19 mV, 25 ms), Patch pipette contained 5 mm EGTA. Gray, Same as black but in the presence of 10 μm nifedipine. B2, QICa: control, 1.16 ± 0.19 pC; Nife, 0.59 ± 0.09 pC; n = 6; p = 0.0047, paired t test; Cm: control, 90.99 ± 10.73 fF; Nife, 76.23 ± 9.52 fF; n = 6; p = 0.0007, paired t test. C1, Black, Example traces of ICa and Cm change induced by a voltage step (from −69 to −19 mV, 25 ms). Pipette contained 10 mm BAPTA. Gray, Same as black but in the presence of 10 μm nifedipine. C2, QICa: control: 1.28 ± 0.10 pC, n = 5; Nife: 0.74 ± 0.04 pC, n = 5; p = 0.0021, paired t test; Cm: control: 35.16 ± 5.18 fF, n = 5; Nife: 18.45 ± 2.23 fF, n = 5; p = 0.0063, paired t test. D1, Black, Example traces of ICa and Cm jump induced by a voltage step (from −69 to −39 mV, 25 ms). Pipette contained 5 mm EGTA. Gray, Same as black but with 10 mm BAPTA-containing intracellular solution. D2, QICa: 5 mm EGTA, 0.36 ± 0.04 pC; 10 mm BAPTA, 0.34 ± 0.03 pC; n = 5 and n = 4; p = 0.7302, unpaired t test; Cm: 5 mm EGTA, 83.68 ± 13.76 fF; 10 mm BAPTA, 16.61 ± 3.99 fF; n = 5 and n = 4; p = 0.0159, unpaired t test. E, Paired comparison of (−19 mV, 25 ms; strong) and mild (−39 mV, 25 ms; mild) stimulus evoked QICa with 5 mm EGTA containing intracellular solution. Mild, QICa = 0.36 ± 0.04 pC; strong, QICa = 1.09 ± 0.1 pC; n = 5; p = 0.0005, paired t test. Mild, ΔCm = 83.68 ± 13.76 fF; strong, ΔCm = 95.24 + 12.42 fF; n = 5; p = 0.0021, paired t test. F, Paired comparison of strong (−19 mV, 25 ms) and mild (−39 mV, 25 ms) stimulus-evoked QICa with 10 mm BAPTA containing intracellular solution. Mild: QICa = 0.33 ± 0.03 pC; strong: QICa = 1.12 ± 0.04 pC; n = 4; p = 0.0005, paired t test; mild: ΔCm = 16.61 ± 3.99 fF; p = 0.0005; strong: ΔCm = 55.97 ± 9.44 fF; n = 4; p = 0.01, paired t test.
Cone photoreceptors express Cav3.2 channel mRNA
To investigate which LVA T-type Ca2+ channels are present in cone photoreceptors, we performed RT-PCR experiments on sorted rod and cone photoreceptors from dissociated retinae of Rac3-eGFP mice, which express eGFP in cone photoreceptors (Gong et al., 2003; Fuchs et al., 2014). For fluorescence-activated cell sorting, eGFP fluorescence and FSC/SSC were used to sort cone and rod photoreceptors, respectively (Fig. 8A1,A2). The unlabeled sorting of rod photoreceptors by FSC/SSC is possible because of the high backscatter of heterochromatin in adult rod photoreceptor somata (Solovei et al., 2009; Feodorova et al., 2015). This method was optimized and validated by the use of CD73, a cell surface marker of cone/rod photoreceptor common precursors and mature rod photoreceptors (Koso et al., 2009). We performed nested RT-PCR with primers that have been validated and successfully deployed on mouse islet cDNA (Vignali et al., 2006). The results of the nested RT-PCR of sorted photoreceptors indicate that cone photoreceptors express Cav3.2, whereas rod photoreceptors do not express any LVA T-type Ca2+ channel (Fig. 8B1). To verify our sorting strategy, we also checked for the expression of β-actin (Actb), the cone photoreceptor marker short-wave-sensitive opsin 1 (Opn1sw), and the rod photoreceptor marker rhodopsin (Rho; Fig. 8B2). While Actb was present in all analyzed samples, Opn1sw was enriched in cone photoreceptors and Rho in rod photoreceptors, corroborating our sorting strategy (Fig. 8B2). On whole-retina cDNA, application of the primers led to specific bands for all three LVA T-type Ca2+ channels Cav3.1 (271 bp), Cav3.2 (305 bp), and Cav3.3 (258 bp; Fig. 8C1), as well as for the markers Actb (196 bp), Opn1sw (175 bp), and Rho (81 bp; Fig. 8C2). No bands were detectable in the negative control, in which reverse transcriptase was omitted during cDNA synthesis of whole-retina mRNA, to exclude amplification of genomic DNA (Fig. 8C1,C2). This confirms the nested RT-PCR results and demonstrates the mRNA expression of Cav3.2 in cone photoreceptors. Furthermore, data of previous work using “Drop-Seq” analysis of individual cells also suggests the presence of Cav3.2 mRNA in cone photoreceptors (Macosko et al., 2015).
Analysis of T-type Ca2+ channels in sorted photoreceptors. A1, FACS strategy for isolating rod photoreceptors by FSC/SSC. A2, FACS strategy for isolating cone photoreceptors by GFP fluorescence of Rac3-eGFP mice. B1, Nested RT-PCR analysis of sorted cone and rod photoreceptor samples with LVA-specific primer pairs for Cav3.1, Cav3.2, and Cav3.3. A distinct band is present for the cone photoreceptor sample using Cav3.2 primers, whereas no bands can be observed for the sorted rod photoreceptor population. B2, Conventional RT-PCR with primers for Rho, Opn1sw, and Actb on sorted rod and cone photoreceptors. Strong bands are visible for Opn1sw and Rho in cone and rod photoreceptor samples, respectively, as well as for Actb. C1, C2, RT-PCR analysis of whole-retina and negative control samples, in which reverse transcriptase was omitted during cDNA synthesis (retina, –RT). For whole retina, the following expected bands are present: Cav3.1 (271 bp), Cav3.2 (305 bp), Cav3.3 (258 bp), Rho (81 bp), Opn1sw (175 bp), and Actb (196 bp). No bands are visible for retina –RT samples.
Discussion
LVA and HVA Ica in cone photoreceptors
There is good anatomic and functional evidence for the presence of Cav1.3 and/or Cav1.4 L-type Ca2+ channels in mammalian photoreceptors (Dieck et al., 2005; Mansergh et al., 2005; Morgans et al., 2005; Wu et al., 2007; Xiao et al., 2007; Regus-Leidig et al., 2014a). To the best of our knowledge, there is so far no evidence for the presence of non-L-type Ca2+ channels in photoreceptors. A previous study, which described the membrane conductances of mouse cone photoreceptors showed evidence for only L-type Ca2+ currents (Ingram et al., 2020). In this work, Ames' Medium was used as an extracellular solution. Ames' Medium contains among others ascorbic acid, which blocks T-type Ca2+ currents (Nelson et al., 2007), and the intracellular solution contained QX-314-Br, which blocks LVA Ca2+ currents (Talbot and Sayer, 1996).
Here, we provide functional and structural evidence by directly targeting mouse cone photoreceptor terminals by patch pipettes of ∼10 MΩ resistance that the measured ILVA is most likely produced by T-type Ca2+ channels. First, T-type Ca2+ channel antagonists such as nickel and Z944 significantly blocked ILVA. Second, physiological parameters of pharmacologically isolated ILVA analyzed in Figure 3 were similar to reported values for T-type Ca2+ channels in heterologous expression systems (Klöckner et al., 1999; Perez-Reyes, 2003). Third, spikes coincide with increased Ca2+ levels in the synaptic terminal. Moreover, RT-PCR experiments revealed the presence of Cav3.2 but not Cav3.1 or Cav3.3 T-type Ca2+ channel mRNA in FACS sorted cone photoreceptors. Nevertheless, and despite the use of a wide variety of pharmacological tools to isolate Ca2+ currents in cone photoreceptors, we cannot fully exclude the contribution of other ion conductances to our measurements. For example, the presence of TTX-resistant Nav1.9 channels in photoreceptors was shown by immunolabeling (O'Brien et al., 2008). However, Nav1.9 channels start to activate at hyperpolarized potentials (approximately −80 mV) and show ultraslow inactivation kinetics (Dib-Hajj et al., 2015; Lin et al., 2016), which is different from the ILVA kinetics reported in this work. In our experiments, ILVA of cone photoreceptors was not sensitive to blockers of TTX-sensitive Na+ channels, hyperpolarization-activated channels, or N-type Ca2+ channels, excluding these ion channels as contributors to ILVA.
LVA Ca2+ channels are also called T-type Ca2+ channels referring to their physiological properties, displaying a transient and tiny current. Based on their physiological properties, the different isoforms of T-type Ca2+ channels are indistinguishable because their activation and inactivation parameters taken from the Boltzmann equation are similar (Klöckner et al., 1999). In our experiment, IHVA peaked at approximately −20 mV, which is relatively far from the dark-induced Vm change in photoreceptors but ILVA started to activate at approximately −50 mV and showed a peak current at approximately −40 mV, which is in the range of the typical light-generated Vm change (from −70 to −40 mV). This indicates that the light-induced Vm change first activates LVA T-type Ca2+ channels followed by HVA L-type Ca2+ channels; thereby, synaptic vesicle release with light-induced Vm change could be connected. Another mechanism to overcome the misalignment between the L-type Ca2+ channel activation range and the light-evoked Vm change is the Ca2+ influx through cGMP channels in salamander cone photoreceptors, which contributes to neurotransmitter release (Rieke and Schwartz, 1994). ILVA would appear to lessen the need for such a mechanism in mouse cone photoreceptors. The pharmacological identification of LVA T-type Ca2+ channels is not straightforward in the presence of HVA L-type Ca2+ channels. At concentrations used to block LVA T-type Ca2+ channels, compounds such as ethosuximide (Coulter et al., 1989) and amiloride (Tang et al., 1988) also affect other types of Ca2+ channels. Mibefradil, a commonly used LVA T-type Ca2+ channel blocker also inhibits HVA L-type Ca2+ channels (Leuranguer et al., 2001; Lee et al., 2006; To et al., 2020). The small but not significant effect of 1 µm mibefradil on ILVA and IHVA and the opposing effect of 5 µm Z944 on ILVA and IHVA (Fig. 2) suggest that HVA L- and LVA T-type Ca2+ channels in mouse cone photoreceptors have unique molecular properties. Several unique isoforms of synaptic proteins have already been reported for photoreceptors such as Piccolino (Regus-Leidig et al., 2013, 2014b), complexin 4 (Reim et al., 2005, 2009), and syntaxin 3B (Morgans et al., 1996; Curtis et al., 2008). Nickel is a divalent cation that preferentially inhibits the LVA T-type Ca2+ channel Cav3.2 at low micromolar concentrations (Lee et al., 1999). It significantly blocked ILVA in our experiments (Fig. 2). The partial block of IHVA by nickel is consistent with the findings that nickel also blocks HVA L-type Ca2+ channels (McFarlane and Gilly, 1998; Hobai et al., 2000; To et al., 2020). In general, the results of the pharmacological experiments indicate that the LVA Ca2+ channels in mouse cone photoreceptors are LVA T-type Ca2+ channels and not hyperpolarization-activated channels or TTX-sensitive Na+ or N-type Ca2+ channels (Fig. 2). RT-PCR experiments on sorted cone photoreceptors propose that T-type Ca2+ channels, present at cone photoreceptors, are the Cav3.2 channel and not the Cav3.1 or Cav3.3 channel (Fig. 8).
Spiking photoreceptors
Mammalian photoreceptors are generally characterized as nonspiking neurons (Yagi and Macleish, 1994; Schneeweis and Schnapf, 1995; Ingram et al., 2020). However, toad rod, turtle cone, and lizard cone photoreceptors can generate Ca2+ spikes (Fain et al., 1980; Gerschenfeld et al., 1980; Maricq and Korenbrot, 1988). Cone photoreceptors showed fast spike-like currents occasionally in premortem and postmortem monkey retina (Bryman et al., 2020). Furthermore, spikes and biphasic light responses were also observed in monkey photoreceptors in another study (Schnapf et al., 1990). However, spike-like currents and biphasic light responses were not observed in a more recent work, suggesting that spikes at monkey photoreceptors might be caused by tissue degradation (Cao et al., 2014). Cultured human rod and cone photoreceptors have been shown to express voltage-gated Na+ channels and generate action potentials sensitive to TTX (Kawai et al., 2001, 2005). However, spike generation in human photoreceptors required that they be held at a highly hyperpolarized Vm (approximately −80 mV), which is more negative than the Vm normally attained in bright light (approximately −70 mV), and spikes were observed only in unhealthy tissue. This indicates that action potentials in human photoreceptors might be because of the experimental conditions and may not appear under physiological conditions (Van Hook et al., 2019). Light-evoked Ca2+ spikes in dark-adapted goldfish ON bipolar cells have also been observed (Protti et al., 2000); however, here the block of L-type Ca2+ channels caused periodic spiking different from our observation when nifedipine blocked spike generation (Fig. 5). The spikes emerging at light offset might enhance the contrast detection at cone photoreceptors between dark and light. This is similar to retinal bipolar cell behavior where small signals are accelerated and amplified by T-type Ca2+ channels (Protti et al., 2000). In our experiment, the observed continuous spiking activity at approximately −40 mV did not require a preceding period of hyperpolarization of the cone photoreceptor membrane (Fig. 5), indicating that the spike generation could play a role in the information transfer at constant light levels. Furthermore, photoreceptor light responses could be converted to both sustained and transient components, which led to increased intracellular Ca2+ levels in the synaptic terminal (Fig. 6). This could trigger several Ca2+-dependent processes such as the activation of Ca2+-induced Ca2+ release, the activation of Ca2+-dependent Cl– and K+ channels, the acceleration of synaptic vesicle replenishment, alterations in the local driving force for Ca2+, as well as controlling synaptic vesicle release kinetics.
Action potentials are generally associated with the spread of depolarizing signals along the axon. Mouse cone photoreceptors have a relatively short, ∼50-µm-long axon. The impact of axon properties showed little effect on the passive signal weakening between the soma and terminal of a bipolar cell, which has an axon length that is 1.5 times longer than that of cone photoreceptors (Oltedal et al., 2009). Consequently, a shorter cone photoreceptor axon should have an even smaller effect on the passive signal spread. In numerous neurons, T-type Ca2+ channels generate low-threshold spikes, which in turn trigger the activation of Na+ channels and thereby a burst of action potentials and oscillatory behavior (Huguenard, 1996; Perez-Reyes, 2003). We measured the sensitivity of Ca2+ currents and spikes to 1 µm TTX, which did not affect Ca2+ current or spike amplitude. Nav1.9 channels usually do not generate spikes; they rather act as a threshold channel (Herzog et al., 2001; Osorio et al., 2014). The results of our experiments suggest that spikes at cone photoreceptors are generated by T- and L-type Ca2+ channels. Ca2+ spikes are not uncommon in biological systems. For example, L-type Ca2+ channels trigger spikes in dopamine neurons with a 0.5 mV/ms rise rate (Iyer et al., 2017) and T-type Ca2+ channels trigger spikes in several neuron types reviewed in the study by Perez-Reyes (2003).
Inferences of the function of LVA T-type Ca2+ channels in cone photoreceptors
Neurotransmitter release at ribbon synapses has been generally attributed to only HVA L-type Ca2+ channels. For example, the ribbon containing auditory hair cells in the bullfrog amphibian papilla, which also express both T- and L-type Ca2+ channels, did not show any synaptic vesicle release during the pharmacological block of L-type Ca2+ channels (Cho and von Gersdorff, 2014). However, several studies have reported a key role for LVA Ca2+ channels in neurotransmitter release and hormone secretion in neuronal tissues (Angstadt and Calabrese, 1991; Tang et al., 2011), and also for retinal neurons other than photoreceptors, such as salamander ganglion cells (Henderson and Miller, 2007), mouse horizontal cells (Feigenspan et al., 2020), and rat bipolar cells (Pan, 2000; Pan et al., 2001). Furthermore, LVA Ca2+ channels directly triggered exocytosis in ribbon-containing synapses such as retinal bipolar cells (Kaneko et al., 1989; Pan et al., 2001). In addition, in developing chicken basilar papilla auditory hair cells, the temporal characteristics of Ca2+ entry through LVA T-type and HVA L-type Ca2+ channels greatly influenced synaptic release (Levic and Dulon, 2012). Interestingly, neurotransmission at photoreceptor ribbon synapses is much more sensitive to Ca2+ than at conventional synapses (Mercer and Thoreson, 2011); however, both synapse types use synaptotagmin-1 as a Ca2+ sensor (Grassmeyer et al., 2019). We show evidence that cone photoreceptors use two Ca2+ sources, namely LVA L-type and HVA T-type Ca2+ channels, which activate at different voltage ranges, have different amplitudes, and localize close to the release site. Therefore, the use of both Ca2+ channel types could contribute to the adjustment of high Ca2+ sensitivity of the release machinery.
Because the detection of bright, high-frequency visual signals requires both presynaptic and postsynaptic mechanisms (Grabner et al., 2016), the use of LVA and HVA Ca2+ channels in both presynaptic (cone photoreceptors) and postsynaptic neurons (horizontal and bipolar cells) might increase the visual signal processing capacity of the retina by giving greater dimension in the gain and kinetics of synaptic signaling. At the postsynaptic site, each horizontal and bipolar cell type could individually further widen or reduce the cone photoreceptor provided dynamic range depending on postsynaptic mechanisms (e.g., at the ground squirrel cone photoreceptor to c2b OFF bipolar cell synapse, the dynamic range of the signaling is reduced by AMPA receptor saturation to improve temporal performance; Grabner et al., 2016). Our results indicate that cone photoreceptor HVA L-type channels “piggyback” on LVA T-type Ca2+ channels, meaning that depolarization at light offset activates first LVA T-type Ca2+ channels, which further activates HVA L-type Ca2+ channels. Therefore, the presence of LVA T-type Ca2+ channels moves the Ca2+ current activation level to more negative potentials where the dark-produced Vm change could already evoke Ca2+ influx and synaptic vesicle release. As a result, LVA T-type Ca2+ channels extend the dynamic range of cone photoreceptor signaling by boosting synaptic vesicle release near the dark membrane potential.
Footnotes
This work was supported by the “Deutsche Forschungsgemeinschaft” (Grant BA 6688/1-1). We thank Dr. Andreas Gießl for helpful advice in performing PCR experiments on sorted cells. Furthermore, we thank Andrea Nerz and Beata Schmidt for excellent technical assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Norbert Babai at norbert.babai{at}fau.de