Abstract
Retinal bipolar cells (BCs) compose the canonical vertical excitatory pathway that conveys photoreceptor output to inner retinal neurons. Although synaptic transmission from BC terminals is thought to rely almost exclusively on Ca2+ influx through voltage-gated Ca2+ (CaV) channels mediating L-type currents, the molecular identity of CaV channels in BCs is uncertain. Therefore, we combined molecular and functional analyses to determine the expression profiles of CaV α1, β, and α2δ subunits in mouse rod bipolar (RB) cells, BCs from which the dynamics of synaptic transmission are relatively well-characterized. We found significant heterogeneity in CaV subunit expression within the RB population from mice of either sex, and significantly, we discovered that transmission from RB synapses was mediated by Ca2+ influx through P/Q-type (CaV2.1) and N-type (CaV2.2) conductances as well as the previously-described L-type (CaV1) and T-type (CaV3) conductances. Furthermore, we found both CaV1.3 and CaV1.4 proteins located near presynaptic ribbon-type active zones in RB axon terminals, indicating that the L-type conductance is mediated by multiple CaV1 subtypes. Similarly, CaV3 α1, β, and α2δ subunits also appear to obey a “multisubtype” rule, i.e., we observed a combination of multiple subtypes, rather than a single subtype as previously thought, for each CaV subunit in individual cells.
SIGNIFICANCE STATEMENT Bipolar cells (BCs) transmit photoreceptor output to inner retinal neurons. Although synaptic transmission from BC terminals is thought to rely almost exclusively on Ca2+ influx through L-type voltage-gated Ca2+ (CaV) channels, the molecular identity of CaV channels in BCs is uncertain. Here, we report unexpectedly high molecular diversity of CaV subunits in BCs. Transmission from rod bipolar (RB) cell synapses can be mediated by Ca2+ influx through P/Q-type (CaV2.1) and N-type (CaV2.2) conductances as well as the previously-described L-type (CaV1) and T-type (CaV3) conductances. Furthermore, CaV1, CaV3, β, and α2δ subunits appear to obey a “multisubtype” rule, i.e., a combination of multiple subtypes for each subunit in individual cells, rather than a single subtype as previously thought.
Introduction
Bipolar cells (BCs) relay visual information from primary sensory neurons, i.e., rod and cone photoreceptors, to amacrine and ganglion cells of the inner retina in the canonical “vertical excitatory” pathway (Masland, 2012; Euler et al., 2014). Neurotransmission between BCs and postsynaptic neurons occurs at ribbon synapses, named for the appearance of the electron-dense ribbon organelle in presynaptic active zones; such ribbon synapses also are found in auditory, vestibular, and lateral line hair cells as well as in retinal photoreceptors (Matthews and Fuchs, 2010; Cho and von Gersdorff, 2012; Lagnado and Schmitz, 2015). Release sites and voltage-gated Ca2+ (CaV) channels are tightly-coupled at most ribbon-type synapses, at which exocytosis is thought to result largely or exclusively from Ca2+ influx mediated by an L-type Ca conductance (Pangrsic et al., 2018), although there are reports of T-type CaV conductances contributing to transmission from some BC types, including rod bipolar (RB) cells in the rat retina (Pan et al., 2001; Singer and Diamond, 2003). This is in contrast to conventional synapses, at which neurotransmission is mediated primarily by P/Q-type and/or N-type Ca conductances (Dolphin and Lee, 2020).
All CaV channels are large protein complexes composed of multiple subunits: a pore-forming α1 subunit and auxiliary β and α2δ subunits. To date, ten unique α1 subunit-encoding genes, four genes encoding β1-4 subunits, and four genes encoding α2δ1-4 subunits have been identified (Zamponi et al., 2015; Dolphin and Lee, 2020). Functionally, Ca currents are classified by their voltage dependence, kinetics, and sensitivity to channel blockers. L-type Ca currents, high-voltage-activated currents sensitive to dihydropyridines, are mediated by channels containing CaV1.1, CaV1.2, CaV1.3, or CaV1.4 α1 subunits (Zamponi et al., 2015; Pangrsic et al., 2018; Dolphin and Lee, 2020). High-voltage-activated, P/Q-type, N-type, and R-type Ca currents are mediated by channels containing α1 subunits in the CaV2 family (CaV2.1, CaV2.2, and CaV2.3; Zamponi et al., 2015; Dolphin and Lee, 2020). Low-voltage-activated T-type Ca currents are mediated by CaV3.1, CaV3.2, or CaV3.3 α1 subunit-containing channels (Perez-Reyes, 2003; Zamponi et al., 2015; Dolphin and Lee, 2020).
Studies of neurotransmission at ribbon synapses in the retina and inner ear indicate that multiple CaV channel types may be involved, although the focus has been on CaV1-containing channels that mediate L-type conductance (Pangrsic et al., 2018). It is well established that photoreceptors express CaV1.4 α1 in addition to β2 and α2δ4 subunits (Morgans, 2001; Mansergh et al., 2005; Lee et al., 2015), but the molecular identity of CaV channel subunits in BCs remains elusive: immunohistochemistry, fluorescence in situ hybridization (FISH), and RT-PCR assays suggest that all the four CaV1 subtypes might be expressed in BCs (Pangrsic et al., 2018; Van Hook et al., 2019). T-type conductances also have been recorded in BCs, indicating the presence of CaV3-containing channels in these cells, although whether such channels are consistently expressed at ribbon synapses is uncertain (Kaneko et al., 1989; de la Villa et al., 1998; Pan, 2000; Ma and Pan, 2003; Hu et al., 2009).
This work was motivated by a recent study in which single-cell RNA sequencing (scRNA-seq) was used for molecular classification of mouse retinal BCs; resultant “transcriptomes” of 15 molecularly defined BC types, including the RB, are available in the Gene Expression Omnibus Database (GEO accession number: GSE81905; Shekhar et al., 2016). We re-analyzed this dataset with particular attention to transcripts of CaV-related genes. We found unexpectedly complex expression patterns of CaV subunits in all mouse retinal BCs.
Using this finding as a guide, we explored the molecular identity, subcellular localization, and function in synaptic transmission of CaV channels in RBs specifically. Single-cell RT-PCR (scRT-PCR) demonstrated heterogeneous expression of CaV subunits in a population of mouse RBs. Immunohistochemical observation of the location of CaV subunit expression in the mouse retina, coupled with electrophysiological and pharmacological analyses of synaptic transmission from RBs, demonstrated that P/Q-type, N-type, T-type, and L-type Ca conductances all combine to control glutamate release at RB ribbon synapses.
Materials and Methods
Animals
All animal procedures were approved by the Institutional Animal Care and Use Committees (IACUC) at Sun Yat-sen University. Two transgenic mouse lines, a cre driver line and a reporter line, were used. RBs were targeted for transgene expression in the BAC-Pcp2-IRES-Cre [B6.Cg-Tg(Pcp2-cre)3555Jdhu/J; Jax 010536] line, which expresses cre recombinase primarily in RBs under the control of mouse Purkinje cell protein (Pcp2; Zhang et al., 2005; Ivanova et al., 2010). Channelrhodopsin-2 (ChR2) was expressed in RBs by cre-dependent recombination after crossing the Pcp2-cre mouse line with Ai32 [B6.Cg-Gt(ROSA)26Sortm32(CAG-COP4*H134R/EYFP)Hze/J; Jax 012569; Madisen et al., 2012]. The wild-type C57BL/6J and Pcp2-cre::Ai32 mice of either sex at ages between 4 and 12 weeks were used in electrophysiological and immunocytochemical experiments, and wild-type mice of either sex at the ages of postnatal day (P)17 and P28–P84 were used for RT-PCR and immunohistochemical studies.
Electrophysiology
Retinal slices at the thickness of 200 µm were prepared from light-adapted retinas isolated from both adult Pcp2-cre::Ai32 and wild-type mice. Retinas were isolated in sodium bicarbonate buffered Ames' medium (Sigma) equilibrated with 95% O2 and 5% CO2 (carbogen) at room temperature, and then embedded in low-melting temperature agarose (Sigma Type VIIA, 2% in a HEPES buffered saline). Slices were cut on a vibrating microtome (Leica VT1200s) and stored in carbogen-bubbled Ames' medium at room temperature until use.
Recordings were performed at near-physiological temperature (30–35°C). Retinal slices were superfused continuously at a rate of 1–2 ml/min with carbogen-bubbled artificial CSF (ACSF) containing the following (in mm): 119 NaCl, 23 NaHCO3, 10 glucose, 1.25 NaH2PO4, 2.5 KCl, 1.15 CaCl2, 1.5 MgCl2, 2 NaLactate, and 2 NaPyruvate. 2-amino-4-phosphonobutyrate (L-AP4, 2 µm), and (S)−1-(2-amino-2-carboxyethyl)−3-(2-carboxy-5-phenylthiophene-3-yl-methyl)−5-methylpyrimidine-2,4-dione (ACET, 1 µm) were included in the ACSF to block synaptic transmission between photoreceptors and BCs. Picrotoxin (50 µm), (1,2,5,6-tetrahydropyridin-4-yl) methylphosphinic acid (TPMPA, 50 µm) or 4-imidazoleacetic acid (I4AA, 10 µm), strychnine (0.5 µm), and tetrodotoxin (TTX, 0.5 µm) were added to the ACSF to block GABAAR-mediated, GABACR-mediated, GlyR-mediated, and voltage-gated Na channel-mediated currents, respectively.
During voltage-clamp recordings, pipettes were filled with the following (in mm): 95 Cs-methanesulfonate, 20 TEA-Cl, 1 4-AP, 10 HEPES, 8 PO4-creatine, 4 ATP-Mg, 0.4 GTP-Na3, and 1 BAPTA. The pH value was adjusted to 7.2 with CsOH and osmolarity to ∼285 mOsm with sucrose. During current-clamp recordings, pipettes were filled with the following (in mm): 110 K-gluconate, 5 NaCl, 10 HEPES, 8 PO4-creatine, 4 ATP-Mg, 0.4 GTP-Na3, and 1 BAPTA. The pH value was adjusted to 7.2 with KOH and osmolarity to ∼285 mOsm with sucrose. To visualize the cell morphology after recordings, Alexa Fluor 488 or 594 was included in the pipette solutions. Generally, RB or AII holding potential was −80 mV, and membrane potentials were corrected for junction potentials of approximately −10 mV. Access resistances were <25 ΜΩ for RBs and <20 ΜΩ for AII amacrine cells and were compensated by 50–90%. Recordings were made using MultiClamp 700B amplifiers. Recorded currents were digitized at 10–20 kHz and low-pass filtered at 2 kHz by an ITC-18 A/D board (Heka/Instrutech) controlled by software written in Igor Pro 6 (WaveMetrics). Recorded RB Ca currents were leak-subtracted off-line (P/4 protocol). Analysis was performed in Igor Pro 6.
Optogenetics
ChR2 was activated by a high-power blue LED (Thor Laboratories; λpeak ∼470 nm) directed through a 60× lens to create a light spot ∼125-µm diameter. The light intensities and durations were controlled by the data acquisition routines (Igor Pro 6).
Analysis of scRNA-seq datasets
The existing scRNA-seq dataset (GEO accession number: GSE81905) that contains transcriptomes of all mouse retinal BCs (Shekhar et al., 2016) was re-analyzed to examine CaV channel subunits in BCs specifically. To reproduce clustering analysis, the clustering algorithm for the retinal BC data achieved with Vsx2-GFP mice at P17 was implemented and performed using a published R markdown script (Shekhar et al., 2016). The single cell libraries that contained >10% mitochondrially-derived transcripts were filtered. Batch correction and principal component analysis (PCA) were performed on the cells for which over 500 genes were detected. Among the selected cells, only genes that were present in at least 30 cells and those having over 60 transcripts were considered. Based on the Louvain–Jaccard method (Blondel et al., 2008; Levine et al., 2015), PC scores were used to embed the single cells on a 2D map using t-distributed stochastic neighbor embedding (t-SNE; van der Maaten and Hinton, 2008). The gene transcription patterns of CaV channel subunits across BC clusters were shown in dotplots, which depicted the proportion of different BC clusters (row) that expressed CaV subunit (column) using dot size and the average number of CaV subunit transcripts in specific cluster using dot color.
Another existing dataset (GEO accession number: GSE63473), which contained the information about transcriptomes of individual rod and cone photoreceptors isolated from the mouse retinas, was used to examine transcription of CaV channel subunits in photoreceptors (Macosko et al., 2015).
RT-PCR analysis of single-cell and whole-retina mRNAs
The cytoplasm from individual RBs was harvested by patch pipettes to identify expressed CaV channel subunits by scRT-PCR. The patch pipettes (6–8 MΩ) were autoclaved to inactivate RNases and filled with the internal solution as follows (in mm): 110 K-gluconate, 5 NaCl, 10 HEPES, 1 BAPTA, 8 PO4-creatine, 4 ATP-Mg, and 0.4 GTP-Na3, pH was adjusted with KOH to 7.4. Recombinant ribonuclease inhibitor (RRI; Clontech, catalog #2313A) was included in the internal solution to better preserve mRNA. RBs in freshly made retinal slices were identified by their morphology with the aid of Alexa Fluor 488 in the internal solution. Cellular cytoplasm and nucleus were harvested into the pipette by applying gentle negative pressure under visual control. Harvesting was immediately terminated on loss of gigaseal. Finally, RB identity was confirmed by detection of Prkca gene expression during scRT-PCR analysis. Control experiments were conducted by using the internal solution without any harvested cell content.
Specific primer sets (total of 26; Table 1) were pooled with a final concentration of 0.1 µm for each primer. Reverse transcription and sequence-specific amplification of the harvested cell contents were performed using the Single Cell Sequence Specific Amplification kit (Vazyme Biotech, catalog #P621-A). PCR tubes were immediately frozen at −80°C. After brief centrifugation at 25°C, the tubes were placed on C1000 Touch Thermal Cycler (Bio-Rad). Reverse transcription was performed at 50°C for 60 min, followed by reverse transcriptase inactivation and Taq polymerase activation by heating to 95°C for 3 min. Subsequently, going through 20 cycles of CaV-sequence-specific amplification by denaturing at 95°C for 15 s, cDNA was annealed and elongated at 60°C for 15 min. After preamplification, PCR tubes were stored at –20°C to avoid evaporation. The preamplified products were then diluted to 50-fold and re-amplified by another round of PCR with specific primer pair for each subtype of CaV channel subunits using Phanta Max Super-Fidelity DNA Polymerase (Vazyme Biotech, catalog #P505-d1/d2/d3). PCR experiments were performed following manufacturer's instructions with an appropriate annealing temperature (56–58°C) for 30 or 35 cycles. The PCR products (10 µl for each sample) were electrophoresed through 2% or 3% agarose gel, stained with SYBR Safe DNA Gel Stain (Invitrogen, catalog #S33102) and imaged under UV light excitation.
Key resources
Whole-retina mRNAs were extracted from retinal lysates of adult wild-type mice by using the SteadyPure Universal RNA Extraction kit (Accurate Biotechnology, catalog #AG21017). RNase-free water was used as a negative control. The mRNAs were reversely transcribed into cDNAs using the PrimeScriptTM RT Master Mix (TaKaRa, catalog #RR036A). PCR was performed with CaV-specific primers at 20 cycles of 94°C for 20 s, 58°C for 30 s, and 72°C for 1 min. The products of the first-round PCR were then amplified by a second-round PCR (35 cycles) with the same parameters and reaction solution to meet the detectable level. The second-round PCR products (10 µl for each sample) were electrophoresed through a 2% agarose gel and imaged under UV light.
Immunohistochemistry and immunocytochemistry
Retinas from both P17 and adult wild-type mice were dissected and fixed in 4% paraformaldehyde (PFA) for 20 min. After fixation, retinas were infiltrated with graded (10%, 20%, and 30%) sucrose in PBS, embedded in OCT, and sectioned vertically at 14 µm on a cryostat (Leica). Immunohistochemical staining was conducted by using the indirect fluorescence method. Retinal sections were blocked in PBS with 0.1% Triton X-100 (PBST) plus 6% bovine serum albumin (BSA) for 2 h at room temperature. Following removal of the blocking solution, retinal sections were incubated with primary antibodies overnight at 4°C. After rinsing, appropriate fluorescence-tagged secondary antibodies were applied for 2 h in darkness at room temperature. The commercially available primary and secondary antibodies used in this study were listed in Table 1.
The anti-CaV2.1, anti-CaV3.1, and anti-CaV3.2 antibodies have been verified in corresponding knock-out mice in previous studies (Nishimune et al., 2016; Gilbert et al., 2017; Cheng et al., 2019). The anti-CaV2.2 antibody used in this study (catalog #152313, Synaptic Systems) was generated using the same immunogen as another anti-CaV2.2 antibody (catalog #152303, Synaptic Systems) that has been verified in CaV2.2 knock-out mice (Lenkey et al., 2015). The staining patterns generated with two anti-CaV3.3 antibodies (catalog #ACC-009, Alomone Labs; catalog #C4616, Sigma) were quite similar. All the antibodies were diluted in PBST with 3% BSA. After rinsing, DAPI staining (1:1000, catalog #C1002, Invitrogen) was applied for 20 min at room temperature. Between incubations, sections were washed three times for 10 min each using PBST.
For the immunocytochemical experiments, retinas from adult wild-type mice were dissected in Hanks' solution containing the following (in mm): 137 NaCl, 0.5 NaHCO3, 1 NaH2PO4, three KCl, 2 CaCl2, 1 MgSO4, 20 HEPES, 1 NaPY, and 16 glucose, adjusted to pH 7.4 with NaOH. The retinas were then digested with 8.3 mg/ml papain, 0.75 mg/ml DL-cysteine and 0.75 mg/ml BSA in Hanks' solution for 30 min at 36°C. After several rinses with PBS, the retinas were mechanically dissociated with fire-polished glass pipettes and isolated retinal cells were placed on coverslips in PBS for ∼35 min at room temperature. The retinal cells were fixed with 4% PFA in PBS for 30 min and, after several rinses with PBS, blocked for 1 h in PBST with 6% BSA. The cells were then incubated with primary antibodies for 2 h and further incubated with secondary antibodies in darkness for 30 min at room temperature (for the detailed information about the antibodies, see Table 1).
For experiments with both retinal sections or isolated cells, controls were conducted either by omission of primary antibodies or by preincubation of primary antibodies with the corresponding immunopeptides. All labeled sections or cells were examined with a confocal laser scanning microscope (LSM 880 or 980, Carl Zeiss) with a Plan-Apochromat 63×/1.4 or 100×/1.4 oil-immersion objective. Because no significant differences in the expression pattern of CaV channel subunits were observed in the mouse peripheral and foveal retina, images of single optical sections with 1024 × 1024 resolution and a thickness of ∼0.6 µm were arbitrarily taken in both retinal regions and adjusted for size, colors, contrast and brightness using ZEN software (Carl Zeiss) and Photoshop software (Adobe Systems).
Colocalization analysis of either CaV1.3 or CaV1.4 with PKCα in RB axon terminals was quantified as the Pearson's coefficient using the JACoP plugin of the extended ImageJ version Fiji. The Pearson's coefficient is a correlation coefficient representing the relationship between two variables (e.g., CaV and PKCα signals in this study; Bolte and Cordelières, 2006). As a control, the image with the CaV signal was rotated 180° out of phase, and the same analysis was performed. The Pearson's coefficients achieved with the original and rotated images were then compared with estimate colocalization.
Statistical analysis
Prism 6 (GraphPad software) was used to run the statistical analysis. For better comparison among different groups, data acquired in each cell were normalized to the value under control condition where appropriate. The Kolmogorov–Smirnov (KS) test for normality was used to compare the cumulative distributions of data. Differences between experimental samples were assessed for significance using two-tailed Student's t test, Wilcoxon signed-rank test or Welch's t test. Significance was taken as p < 0.05. All data were represented as mean ± SEM.
Results
Analysis of scRNA-seq data reveals unexpectedly complex gene transcription patterns of CaV α1, β, and α2δ subunits in mouse retinal BCs
We began by analyzing CaV α1, β, and α2δ gene expression in different BC types of the mouse retina using the existing scRNA-seq dataset (GEO accession number: GSE81905; Shekhar et al., 2016): the dotplots in Figure 1 illustrate the percentage of CaV subunit expressing cells (the size of each circle) and the average expression level (color) for each BC type. Transcription of CaV genes was more complex than expected, particularly as follows: besides CaV1 and CaV3 transcripts, which mediate the L-type and T-type currents observed in mammalian BCs (Van Hook et al., 2019), CaV2.1 and CaV2.2 transcripts, which generate P/Q-type and N-type Ca currents respectively, also were expressed at reasonable levels in almost all BCs. CaV2.3 transcripts (giving rise to R-type currents) generally were expressed at extremely low levels, except for in a few BC types such as BC5A (type 5A BC; Fig. 1A).
scRNA-seq analysis reveals complex gene expression patterns of CaV α1, β, and α2δ subunits in mouse retinal BCs. A, Gene expression patterns of CaV α1 subunits in different types of BCs. The protein that each gene encodes is given in parentheses, and the type of calcium currents that each α1 subunit mediates is shown on the top. The size of each circle represents the percentage of expressing cells in the group (PercExp) in which the gene expression is detected. The color represents the average expression level in expressing cells (AvgExp). BC, bipolar cells; RB, rod bipolar cells. B, Gene expression patterns of CaV β and α2δ subunits in BCs.
The vast majority of BCs expressed distinct combinations of 2 or 3 CaV1 subtypes (including CaV1.1, CaV1.3, and CaV1.4), which mediate L-type Ca currents, rather than a single CaV1 subtype as commonly thought and reported previously (Pangrsic et al., 2018; Van Hook et al., 2019); it was noteworthy that the transcription of the Cacna1c gene encoding CaV1.2 was not detected in any BC type (Fig. 1A). Almost all BCs expressed distinct combinations of 2 or 3 CaV3 subtypes (CaV3.1, CaV3.2, and CaV3.3), which mediate T-type Ca currents (Fig. 1A). Unlike photoreceptors, in which β2 is the only CaV β subunit interacting with CaV1 channels, specifically CaV1.4 (Lee et al., 2015), multiple β subunits, predominantly β2 and β4 subunits (encoded by the Cacnb2 and Cacnb4 genes, respectively), were observed in most BCs (Fig. 1B). This diversity of β subunit transcript expression is consistent with our observation of CaV2 channel transcripts in BCs (Fig. 1A) because β4 has been reported to be the predominant CaV β subunit that is associated with CaV2 channels in the brain (Müller et al., 2010; Dolphin, 2012). Most of BC types expressed distinct combinations of two to four α2δ subunits (encoded by the Cacna2d1-4 genes), predominantly α2δ3 and α2δ4, with some BC types such as BC3B and BC4 (types 3B and 4 BCs, respectively) having high levels of all four transcripts (Fig. 1B). These findings were quite striking, since it is well accepted that photoreceptors, which also have ribbon-type presynaptic active zones, express exclusively CaV1.4, β2, and α2δ4 subunits (Morgans, 2001; Mansergh et al., 2005; Lee et al., 2015); the expectation of a relatively simple expression pattern of CaV channel subunits in BCs is therefore common. As a control for this unexpected finding, we analyzed another existing dataset (GEO accession number: GSE63473), which contains single-cell transcriptomes for mouse retinal rod and cone photoreceptors (Macosko et al., 2015). We found that rod and cone photoreceptors exhibited much higher levels of mRNAs encoding CaV1.4, β2, and α2δ4 subunits than other CaV-related transcripts (data not shown), consistent with published observations (Morgans, 2001; Mansergh et al., 2005; Lee et al., 2015).
We therefore expect significant molecular diversity of CaV channel subunits in mouse retinal BCs and believe that the transcription pattern for each of CaV1, CaV3, β, and α2δ subunits in BCs generally adheres to a “multisubtype” rule, i.e., a combination of multiple subtypes rather than a single subtype for each subunit in a certain cell type.
Unique gene transcription patterns of CaV channel subunits in RBs confirmed by scRT-PCR
We undertook a targeted examination of CaV transcripts in mouse RBs using scRT-PCR to examine potential heterogeneity of CaV subunit expression in BCs in greater detail. Because the existing scRNA-seq dataset was acquired from mouse retinas at P17 (Shekhar et al., 2016), we performed scRT-PCR experiments using RBs from P17 mice as well as from adult (4- to 12-week-old) mice to determine whether there are developmental changes in CaV subunit expression. As well, we used multiple primer pairs for each subunit to detect known splice variants (see Materials and Methods).
Consistent with our analysis of the scRNA-seq dataset (above), scRT-PCR detected multiple CaV α1, β, and α2δ subunit mRNAs in individual RBs from both P17 and adult mice (n = 38 cells from P17 animals; n = 40 cells from adult animals; Fig. 2A–E). CaV2.3, however, was not detected in any RB, and Cav1.2 was found in only one RB from a P17 mouse (Fig. 2A,B,D; Table 2). The efficacy and specificity of the primer pairs for these two CaV α1 subunits was validated by RT-PCR analysis of whole-retina mRNAs (data not shown), confirming that CaV1.2 and CaV2.3 were rarely expressed in RBs.
Expression of CaV channel subunits in P17 and adult mouse RBs
scRT-PCR confirms the unique expression patterns of CaV subunits in RBs. A, Co-expression of multiple CaV1 subtypes in a single RB cell from an adult mouse retina. CaV1.1, CaV1.3, and CaV1.4 but not CaV1.2 exist in the same RB. B, Co-expression of CaV2.1 and CaV2.2 but not CaV2.3 channels in another RB. C, Co-expression of all three CaV3 subtypes in a single RB. D, E, The percentages of CaV α1, β, and α2δ subunit expression in individual RBs of both P17 and adult mice. P17, postnatal day 17. See also Table 2. F–I, Schematic diagrams illustrating the high heterogeneity for each of CaV1, CaV3, β, and α2δ subunits in individual RBs from adult mice. The percentages of distinct combinations of subtypes are shown, and the exact cell numbers are given in parentheses. See also Table 3.
Differential expression of CaV channel subunits in individual RBs
Individual RBs were more likely to express any given CaV transcript when analyzed with scRT-PCR than with scRNA-seq, owing to the higher sensitivity of the scRT-PCR method (Fig. 2D,E; compare with Fig. 1). As a rule, mRNA expression assessed by scRT-PCR was stable from P17 to adulthood, but some developmental changes were observed: both CaV1.1 and CaV2.2 mRNAs were detected more frequently in adult RBs, whereas β1, β4, and α2δ3 subunit mRNAs were less frequently detected in adult RBs (Fig. 2D,E; Table 2).
It is worth emphasizing that a high degree of heterogeneity in CaV subunit mRNA expression was observed within the population of RBs, i.e., expression profiles varied between individual cells (Fig. 2F–I; Table 3), with most RBs co-expressing multiple α1, β, and α2δ subunits. In adult mice, for example, the specific combination of CaV1.3 and CaV1.4 was found in 62% (n = 23) of CaV1-expressing cells (n = 37; Fig. 2F); this is in keeping with the scRNA-seq analysis (Fig. 1A). Similarly, CaV3.1, CaV3.2, and CaV3.3 were co-expressed in the large majority (74%; n = 21) of CaV3-expressing cells (n = 28; Fig. 2G), and the combinations of β1-4 and α2δ1-4 subtypes accounted for 68% and 58% (Fig. 2H,I), respectively, of accessory CaV subunit expression in RBs. In conclusion, it would seem that the majority of RBs co-express multiple subtypes of α1 subunits mediating L-type and T-type Ca conductances.
Functional analysis of CaV subunit diversity
To begin to understand the functional significance of the varied CaV subunit expression observed in RBs, we performed three sets of experiments:
We examined the subcellular localization of CaV1, CaV2, and CaV3 subunits in RBs in sections of P17 and adult mouse retinas using fluorescence immunohistochemistry. Additionally, we examined the expression of these CaV subunits in the axon terminals of RBs dissociated from adult mouse retinas using fluorescence immunocytochemistry. Consistent with our scRT-PCR analysis, no significant differences in the expression pattern of CaV channels in the mouse retina were observed between P17 and adulthood (data not shown). Therefore, for simplicity, we present only results from adult retinas here.
We made whole-cell voltage-clamp recordings of Ca currents from identified RBs and characterized the underlying channel types using well-established pharmacological tools.
We determined whether various Ca conductances could contribute to synaptic transmission from RBs by expressing the light-gated ion channel ChR2 in RBs using cre-mediated recombination and recording optogenetically-evoked EPSCs in AII amacrine cells (Liang et al., 2021) under conditions in which various Ca conductances were manipulated pharmacologically.
In summary, we found that CaV1, CaV2, and CaV3 subunits were found in RB terminals and that the L-type, P/Q-type, N-type, and T-type conductances they mediate all contribute to synaptic transmission from RB ribbon synapses. Below, we consider these channel types individually, beginning with the CaV2 subtypes which, to our knowledge, have not been shown in published work to mediate transmission at ribbon synapses generally and at RB synapses specifically.
CaV2 subunits in RB axon terminals mediate Ca2+ influx through P/Q-type and N-type conductances
Although P/Q-type and N-type Ca conductances mediate transmission at most conventional synapses, a role for them at ribbon synapses has not been proposed (Dolphin and Lee, 2020). Therefore, it was remarkable that both scRNA-seq and scRT-PCR analyses revealed the existence of CaV2.1 and CaV2.2 in RBs (Figs. 1A, 2B,D). To examine the subcellular localization of these channels, we performed fluorescent immunohistochemical labeling of CaV2.1, PKCα (a specific marker of RBs), and RIBEYE (a ribbon-specific protein) in mouse retinal sections. We found that CaV2.1 was expressed in the axon terminals of RBs at sites both near and away from ribbons (Fig. 3A,B). CaV2.1 was also found to be expressed strongly in the ganglion cell layer (GCL) and throughout the inner plexiform layer (IPL; Fig. 3A), the latter of which was consistent with our observations of CaV2.1 mRNA expression in almost all BCs (Fig. 1A) and with published observations of CaV2.1 channel expression in many types of amacrine and ganglion cells (Van Hook et al., 2019).
P/Q-type CaV channels mediate Ca2+ influx into RB axon terminals. A, Confocal images showing triple labeling of CaV2.1 (green), PKCα (a specific cell marker of RB cells; magenta), and RIBEYE (a ribbon-specific protein; blue) in a mouse retinal section. CaV2.1 channels were predominantly expressed in the IPL and the GCL. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar: 20 µm. B, Magnification of the signals in the frames of A. CaV2.1 channels were expressed in RB axon terminals both near and away from ribbons (arrows). Scale bar: 2.5 µm. C, With 10 μm nimodipine (Nimo; an L-type CaV channel blocker) in the extracellular solution (ES), the Ca current (ICa) evoked by a voltage step from −80 to −20 mV was only partially blocked by 10 μm mibefradil (Mibe; a T-type CaV channel blocker), and the residual current was almost completely blocked by 200 nm ω-agatoxin IVA (Aga; a P/Q-type CaV channel blocker). D, Magnification of the traces shown in C. E, With 10 μm Nimo in the ES, the Ba current (IBa) evoked by a voltage step from −80 to −20 mV was partially blocked by 10 μm Mibe, and the residual current was further reduced by 200 nm Aga. Note that a small component of IBa persists in the presence of Nimo, Mibe, and Aga. F, Average traces showing the Ba currents mediated by L-type (black; n = 18), T-type (blue; n = 21), and P/Q-type (orange; n = 7) CaV channels in RBs, which exhibit apparently different current kinetics. The IBa mediated by each type of CaV channels in a single RB was calculated by subtracting the current recorded after application of a specific channel blocker from the current recorded before application of that drug. The subtracted currents from different RBs were then averaged across cells. G, With 10 μm Nimo and 10 μm Mibe in the ES, RB ICa (green) evoked by a voltage ramp from −80 to +40 mV was strongly reduced by 200 nm Aga (magenta). The subtracted current (orange) represents the ICa mediated by P/Q-type channels.
To determine whether CaV2.1 creates functional Ca channels mediating P/Q-type Ca currents in RBs, we recorded membrane currents from RBs under conditions in which other Ca channel types were blocked pharmacologically. In the presence of 10 µm nimodipine in the external solution to block L-type conductance (Pan and Lipton, 1995; Pan et al., 2001), depolarization of an RB from −80 to −20 mV for 100 ms generated a transient, inward Ca current with rapid and strong inactivation (Fig. 3C,D). This current was partially attenuated by the T-type conductance antagonist, mibefradil (10 µm; n = 9), and the residual current was blocked almost completely by subsequent application of 200 nm ω-agatoxin-IVA (agatoxin hereafter), a specific P/Q-type current antagonist (Fig. 3C,D; n = 5). We made the same observation when 10 mm Ba2+ was used as the charge carrier, replacing Ca2+, to increase the amplitude of the recorded currents (compare Fig. 3C and E). L-type, T-type, and P/Q-type currents recorded in RBs showed distinct waveforms, consistent with the published literature in the field (Fig. 3F). Interestingly, in the presence of nimodipine, mibefradil, and agatoxin, we could record a residual current (Fig. 3E,G) which presumably was mediated by N-type Ca channels according to our scRNA-seq and scRT-PCR analyses. Therefore, we repeated the experiments above to assess CaV2.2 expression and function in RBs.
Fluorescence immunohistochemistry, using antibodies recognizing CaV2.2, PKCα, and RIBEYE, was performed: CaV2.2 was found to be expressed throughout the IPL, including in the axon terminals of RBs (Fig. 4A,B). In the presence of 10 µm nimodipine and 10 µm mibefradil, a membrane current sensitive to 2 µm ω-conotoxin-GVIA, a specific N-type current blocker, was recorded in response to a voltage ramp from −80 to +40 mV (Fig. 4C; n = 4). Notably, a similar current was also detected in three out of five RBs under the experimental condition without nimodipine and mibefradil preincubation (Fig. 4D).
N-type CaV channels mediate Ca currents in RBs. A, Confocal images showing triple labeling of CaV2.2 (green), PKCα (magenta) and RIBEYE (blue) in a mouse retinal section. Immunostaining for CaV2.2 was mainly observed in the IPL, and sparse labeling was also found in the OPL. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar: 20 µm. B, Magnification of the signals in the frames of A. Most of CaV2.2 channels were located relatively away from ribbons in the axon terminals of RBs. Scale bar: 2.5 µm. C, Averaged traces from multiple RBs (n = 4) showing that, with Nimo (10 μm) and Mibe (10 μm) in the extracellular solution (ES), Ca current (ICa) evoked by a voltage ramp from −80 to +40 mV (control, green) was mildly reduced by 2 μm ω-conotoxin-GVIA (Ctx, brown). The ICa mediated by N-type channels (magenta) was shown by subtracting the Ctx trace from the control trace. D, With normal ES, RB ICa evoked by a voltage ramp from −80 to +40 mV (control, black; n = 3) was slightly reduced by 2 μm Ctx (brown). The subtracted trace showing ICa mediated by N-type channels (magenta) was similar to that shown in C.
L-type Ca currents in RBs are mediated by CaV1.3-containing and CaV1.4-containing channels
It is well established that BCs generally and RBs specifically exhibit L-type Ca currents that mediate transmission from the ribbon synapses of these cells (Pan et al., 2001; Jarsky et al., 2010; Pangrsic et al., 2018; Van Hook et al., 2019). The molecular identity of the α1 subunits in the underlying channels, however, is uncertain, and our scRNA-seq and scRT-PCR analyses revealed CaV1.1, CaV1.3 and CaV1.4 in RBs (Figs. 1A, 2A,D,F).
There are several reports of CaV1.1 expression exclusively in the dendritic tips of RBs based on fluorescence immunohistochemistry (Specht et al., 2009; Soto et al., 2012; Tummala et al., 2014), but the specificity of the anti-CaV1.1 antibody used in those studies is questionable (Hasan et al., 2016). Here, then, we restricted our attention to CaV1.3 and CaV1.4. To examine the subcellular localization of these proteins, we performed fluorescent immunohistochemical labeling of CaV1.3 or CaV1.4, PKCα, and RIBEYE in mouse retinal sections. Remarkably, we found that both CaV1.3 and CaV1.4 were expressed in the axon terminals of RBs (Fig. 5). CaV1.4 was found at sites both proximate to and distant from presynaptic ribbons (Fig. 5A,B), whereas CaV1.3 appeared to be restricted to a position proximate to the presynaptic ribbons (Fig. 5C,D). Additionally, CaV1.3 also was found to be expressed in the somata of RBs (Fig. 5C). Thus, it would seem that L-type Ca currents in RB terminals are mediated by both CaV1.3-containing and CaV1.4-containing channels.
Both CaV1.3 and CaV1.4 channels are located near ribbons in RB axon terminals. A, Confocal images showing triple labeling of CaV1.4 (green), PKCα (magenta), and RIBEYE (blue) in a mouse retinal section. CaV1.4 channels were predominantly expressed in the OPL and the IPL, and, to a lesser extent, in the GCL. The lower panels show magnification of the images in the frames of upper panels. CaV1.4 was expressed both near and away from ribbons (arrows) in RB axon terminals. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bars: 20 µm (upper panels) and 2.5 µm (lower panels). B, Colocalization analysis of CaV1.4 and PKCα in RB axon terminals. The Pearson's coefficient is used to estimate the correlation of green and magenta signals (i.e., CaV1.4 and PKCα) in the frames of A. As a control, the green signal in the frame was rotated 180° out of phase, and then the same analysis was performed. The Pearson's coefficient dropped significantly after rotation of the green signal, which confirms that colocalization of CaV1.4 and PKCα in RB axon terminals is real. C, Triple labeling of CaV1.3 (green), PKCα (magenta), and RIBEYE (blue) in a mouse retinal section. CaV1.3 channels were expressed throughout the retina except for the ONL. Colocalization of CaV1.3 with RIBEYE in the axon terminals of RBs (arrows) could be clearly seen. Labeling of CaV1.3 also was observed on the cell membrane of RB somata (asterisks). Scale bars: 20 µm (upper panels) and 2.5 µm (lower panels). D, Colocalization analysis of CaV1.3 and PKCα in RB axon terminals. The Pearson's coefficient is used to estimate the correlation of green and magenta signals (i.e., CaV1.3 and PKCα) in the frames of C. As a control, the green signal in the frame was rotated 180° out of phase, and then the same analysis was performed. This analysis confirms the colocalization of CaV1.3 and PKCα in RB axon terminals.
CaV3.1, CaV3.2, and CaV3.3 generate T-type Ca currents in RBs
Our scRNA-seq and scRT-PCR analyses revealed in RBs transcripts of three genes (Cacna1g, Cacna1h, and Cacna1i) that encode the three known CaV3 subunits (Figs. 1A, 2C,D,G). To examine the subcellular localization of these channels, we performed fluorescent immunohistochemical labeling of CaV3.1, CaV3.2, or CaV3.3 in addition to both PKCα and RIBEYE in mouse retinal sections.
All three CaV3 subunits were found throughout the retina, including in RB terminals near presynaptic ribbons, although CaV3.1 immunoreactivity was weaker than that of CaV3.2 or CaV3.3 (Fig. 6A–C). Specifically, CaV3.3 immunoreactivity was found throughout the IPL. It was stronger in ON than in OFF sublamina and greatest in RB terminals (Fig. 6A). This observation is consistent with scRNA-seq data analysis that revealed higher Cacna1i transcription in ON BCs (including RBs and ON cone BCs) than in OFF BCs (Fig. 1A).
All three types of CaV3 channels are expressed in RB axon terminals. A, Confocal images showing triple labeling of CaV3.3 (green), PKCα (magenta), and RIBEYE (blue) in a mouse retinal section. CaV3.3 channels were predominantly expressed in the IPL, with the strongest expression observed in the innermost of the IPL. The lower panels show magnification of the images in the frames of upper panels. Expression of CaV3.3 could be clearly seen near ribbon sites (arrows) in an RB axon terminal (oval). B, Triple labeling of CaV3.2 (green), PKCα (magenta), and RIBEYE (blue) in a mouse retinal section. CaV3.2 channels were expressed throughout the retina except for the ONL. In an RB axon terminal (oval), CaV3.2 was expressed at sites both near and away from ribbons (arrows). C, Triple labeling of CaV3.1 (green), PKCα (magenta), and RIBEYE (blue) in a mouse retinal slice. CaV3.1 channels were expressed moderately in the INL and the IPL, and strongly in the GCL. Relatively weak and puncta labeling of CaV3.1 could be observed near ribbons (arrows) in RB axon terminals (ovals). Note that CaV3.1 channels also were expressed in the nucleus of cells with somata located in the INL and the GCL, but not in the ONL. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bars: 20 µm (upper panels) and 2.5 µm (lower panels).
CaV3.2 was expressed throughout the IPL, including in RB terminals, and in the somata of cells located in the inner nuclear layer (INL) and the GCL (Fig. 6B). CaV3.1 also was observed in the IPL and in the axon terminals of RBs (Fig. 6C). Additionally, CaV3.1 channels were expressed in the nucleus of cells located in the INL and the GCL, but not in those located in the outer plexiform layer (Fig. 6C).
T-type Ca currents have been described in RBs of the rat and mouse retinas (Kaneko et al., 1989; de la Villa et al., 1998; Pan, 2000). Additionally, in the rat retina, activation of T-type CaV channels mediates neurotransmitter release from RBs (Pan et al., 2001). In line with this, here we found in the mouse retina that RB Ca and Ba currents recorded in the presence of 10 µm nimodipine were reduced by the T-type CaV channel blocker mibefradil (10 µm; Fig. 3B–D). To exclude the possibility that mibefradil might exert some off-target effects such as R-type CaV channels or ORAI channels, as reported previously (Randall and Tsien, 1997; P. Li et al., 2019), we tested the effects of other T-type CaV channel blockers, including NNC 55-0396 (20 µm; structurally similar to mibefradil) and ML218 (6 µm; structurally different from mibefradil; M. Li et al., 2005; Xiang et al., 2011), on Ca currents recorded in RBs when 10 µm nimodipine was included in the external solution. Each drug inhibited RB Ca currents (Fig. 7).
T-type CaV channels mediate Ca2+ currents in mouse retinal RBs. A, Representative traces showing that, with 10 μm nimodipine (Nimo) in the extracellular solution (ES), Ca current (ICa) evoked by a voltage step from −80 to −20 mV in an RB was partially blocked by 6 μm ML218, a T-type Ca channel blocker. B, Magnification of the traces shown in A. C, The ICa recorded in another RB was reduced by 20 μm NNC 55-0396 (NNC), another T-type Ca channel blocker. D, Magnification of the traces shown in C. E, Summary data showing the effects of mibefradil (Mibe), ML218 and NNC on RB ICa. All three blockers exhibited significant inhibition. The data were represented as mean ± SEM. Paired t tests (for Mibe and NNC) and Wilcoxon signed-rank test (for ML218) were used (control 1 vs Mibe, t(8) = 13.64, p < 0.0001; control 2 vs ML218, n = 7, p = 0.0156; control 3 vs NNC, t(11) = 4.196, p = 0.0015); *p < 0.05, **p < 0.01, ****p < 0.0001. F, Summary data showing the relative effects of three different T-type Ca channel blockers on RB ICa. The peak amplitudes were normalized to the amplitude under control condition in each cell before averaging across cells. The data were represented as mean ± SEM. Wilcoxon signed-rank tests were used (control 1 vs Mibe, n = 9, p = 0.0039; control 2 vs ML218, n = 7, p = 0.0156; control 3 vs NNC, n = 12, p = 0.0010); *p < 0.05, **p < 0.01, ***p < 0.001.
CaV channel subunits are located in the axon terminals of RBs but not on the processes from other retinal cells contacting RB terminals
To exclude the possibility that the limited spatial resolution of conventional light microscopy might have led us to confuse CaV expression on neurites apposed to RB terminals with CaV expression in the terminals themselves, we performed two sets of experiments. First, as a negative control experiment, we demonstrated that synapsin 1, a synaptic vesicle-associated protein absent from ribbon-type synapses (Mandell et al., 1990), is not colocalized with PKCα in the axon terminals of RBs (Fig. 8A,B); rather, RB terminals were wrapped by processes labeled with synapsin 1 (Fig. 8A).
CaV channel subunits are located in the axon terminals of RBs but not on the processes from other retinal cells contacting RB terminals. A, Confocal images showing expression of synapsin 1 (green) and PKCα (magenta) in an adult mouse retinal section. In the merged image (green + magenta), synapsin 1, which is not expressed at ribbon-type synapses, was rarely observed in PKCα-positive RB terminals. IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar: 5 µm. B, Confocal images showing expression of synapsin 1 (green) and PKCα (magenta) in an RB dissociated from an adult mouse retina. Synapsin 1 was not observed in RB terminal. Scale bar: 5 µm. C–J, Confocal images showing expression of a CaV subunit (green) and PKCα (magenta) in RBs dissociated from adult mouse retinas: all CaV channel subunits examined were found in RB terminals. As a control, no CaV subunits were detected after omission of primary antibodies against CaV subunits. Scale bar: 5 µm (C–J).
Second, as a positive control, we examined by immunocytochemistry the subcellular localization of CaV1, CaV2, and CaV3 subunits in RBs dissociated from adult mouse retinas. Consistent with our observations in RBs from retinal sections, CaV1 (CaV1.3 and CaV1.4), CaV2 (CaV2.1 and CaV2.2), and CaV3 (CaV3.1, CaV3.2, and CaV3.3) subunits were all found in the axon terminals of isolated RBs, a preparation in which processes from other retinal cells contacting RB terminals are absent (Fig. 8C–J).
P/Q-type, N-type, T-type, and L-type conductances control evoked release at RB→AII ribbon synapses
We have shown recently (Liang et al., 2021) that cre-mediated recombination in the adult Pcp2-cre::Ai32 mouse retina can be used to express ChR2 predominantly in RBs and that optogenetic stimulation of neurotransmission at RB→AII amacrine cell synapses is stable over long periods (>20 min). To determine whether multiple types of Ca channels contribute to synaptic transmission at the RB→AII synapse, we blocked transmission from photoreceptors to bipolar and horizontal cells with the mGluR6 agonist L-AP4 (2–5 µm) and the kainate receptor antagonist ACET (1 µm) and used brief flashes (2–10 ms) of 470-nm LED light to activate RBs; the EPSCs recorded in AIIs under this condition reflect evoked release from RBs (Fig. 9A).
P/Q-type and N-type as well as the previously known L-type and T-type CaV channels control neurotransmitter release from RBs. A, A schematic diagram showing the optogenetic study of synaptic transmission between RBs and AII amacrine cells. ChR2 was expressed predominantly in RB cells and a few ON cone bipolar (CB) cells by cre-dependent recombination in adult Pcp2-cre::Ai32 mouse retinas. With all the synaptic transmission between photoreceptors and BCs is blocked pharmacologically, brief flashes of 470-nm LED can directly activate ChR2+ RBs and induce postsynaptic responses in AIIs, which mainly reflect neurotransmitter release from RBs. It has been determined that the electrical coupling between ON CBs and AIIs is negligible under this experimental condition (Liang et al., 2021). C, cone; R, rod. B–E, With normal extracellular solution (ES), the EPSCs recorded in AIIs, which were evoked by 470-nm LED light stimulation, were reduced substantially by 10 μm nimodipine (Nimo), and inhibited more strongly by additional application of 10 μm mibefradil (Mibe), 200 nm ω-agatoxin IVA (Aga), and 2 μm ω-conotoxin-GVIA (Ctx). In the presence of four different CaV channel blockers, the residual responses were blocked completely by DNQX (20 μm), the antagonist of non-NMDA glutamate receptors. Vhold = −80 mV. F, The effects of different drugs on ChR2-evoked EPSCs in AIIs. For better comparison among groups, the current integrals were normalized to the integral under control condition in each cell before averaging across cells. The data were represented as mean ± SEM. Welch's t tests were used (Nimo, n = 10; Nimo + Mibe, n = 7; Nimo + Mibe + Aga, n = 10; Nimo + Mibe + Aga + Ctx, n = 9; Nimo vs Nimo + Mibe, p = 0.0022; Nimo + Mibe vs Nimo + Mibe + Aga, p = 0.0049; Nimo + Mibe + Aga vs Nimo + Mibe + Aga + Ctx, p = 0.0167); *p < 0.05, **p < 0.01. G, With Nimo and Mibe in the ES, ChR2-evoked EPSCs were dramatically inhibited by application of 200 nm Aga (n = 5). H, Nimo, Mibe, and Aga only had little effects on ChR2-evoked membrane depolarization of RBs (n = 5). I, With Nimo, Mibe, and Aga in the ES, ChR2-evoked EPSCs were further inhibited by application of 2 μm Ctx (n = 4). Tiny responses, likely mEPSCs, persisted in the presence of Ctx and were blocked completely by DNQX (20 μm). The residual responses after application of DNQX were mediated by gap junctions between AIIs and ChR2+ ON CBs. J, With normal ES, ChR2-evoked EPSCs were slightly reduced by 2 μm Ctx. K, The effect of 2 μm Ctx on ChR2-evoked EPSCs in AIIs. The data were represented as mean ± SEM. Paired Student's t test was used (control, −354.80 ± 87.74 pA; Ctx, −304.30 ± 72.57 pA; n = 5, p = 0.0163); *p < 0.05. L, Ctx had little effect on ChR2-evoked membrane depolarization of RBs (n = 3). M, A schematic diagram showing that, with the mGluR6 agonist L-AP4 in the ES, EPSCs in AIIs can be recorded by puffing LY 341495 (LY, group II metabotropic glutamate receptor antagonist; 5–10 μm) onto the dendrites of RBs and ON CBs located at the outer plexiform layer in wild-type mice. N, The LY-evoked EPSCs recorded in an AII were strongly reduced by co-application of Nimo and Mibe, and further suppressed by Aga. Currents mediated by gap junctions between AIIs and ON CBs could be seen after application of DNQX (20 μm). O, Magnification of the traces shown in N. P, The effects of different drugs on LY-evoked EPSCs in AIIs. The current integrals were normalized to the integral under control condition in each cell before averaging across cells. The data were represented as mean ± SEM. Welch's t tests were used (Nimo, n = 8; Nimo + Mibe, n = 6; Nimo + Mibe + Aga, n = 6; Nimo vs Nimo + Mibe, p = 0.0070; Nimo vs Nimo + Mibe + Aga, p = 0.0001; Nimo + Mibe vs Nimo + Mibe + Aga, p = 0.0120); *p < 0.05, **p < 0.01, ***p < 0.001.
To avoid complications arising from rundown of ChR2-evoked EPSCs that might occur over a period sufficient to sequential application of four different CaV channel blockers (i.e., nimodipine, mibefradil, agatoxin, and conotoxin), we compared the relative effects of these drugs on AII EPSCs by performing four independent experiments. Evoked EPSCs were reduced to ∼44% of control by bath application of 10 µm nimodipine (Fig. 9B,F; for better comparison among groups, all the data were normalized; nimodipine, 0.443 ± 0.043, n = 10). Co-application of 10 µm nimodipine and 10 µm mibefradil resulted in a larger attenuation (Fig. 9C,F; nimodipine + mibefradil, 0.244 ± 0.032, n = 7; nimodipine vs nimodipine + mibefradil, t(14.83) = 3.699, p = 0.0022, Welch's t test). The EPSCs were reduced further by application of 10 µm nimodipine, 10 µm mibefradil and 200 nm agatoxin (Fig. 9D,F; nimodipine + mibefradil + agatoxin, 0.113 ± 0.014, n = 10; nimodipine + mibefradil vs nimodipine + mibefradil + agatoxin, t(8.46) = 3.776, p < 0.0049, Welch's t test) and almost entirely abolished by application of 10 µm nimodipine, 10 µm mibefradil, 200 nm agatoxin and 2 µm conotoxin (Fig. 9E,F; nimodipine + mibefradil + agatoxin + conotoxin, 0.063 ± 0.012, n = 9; nimodipine + mibefradil + agatoxin vs nimodipine + mibefradil + agatoxin + conotoxin, t(16.78) = 2.659, p = 0.0167, Welch's t test). In the presence of these CaV channel antagonists, any residual EPSC was blocked completely by 20 µm DNQX, a non-NMDA glutamate receptor antagonist (Fig. 9E). Thus, we conclude that P/Q-type and N-type Ca conductances, in addition to the well-characterized L-type and T-type conductances, mediate transmission from RBs.
To support this conclusion, we recorded EPSCs after the retinal slices were preincubated with 10 µm nimodipine and 10 µm mibefradil to block all L-type and T-type conductances. Under this experimental condition, synaptic transmission persisted and was blocked almost completely by application of 200 nm agatoxin (Fig. 9G). Additionally, nimodipine, mibefradil, and agatoxin had minimal effects on the ChR2-evoked membrane potential change in RBs (Fig. 9H; n = 5), indicating that Ca2+ influx through L-type, T-type, and P/Q-type CaV channels contributed to exocytosis rather than depolarization of RBs. Note that DMSO alone, at the same concentration as was present with 200 nm agatoxin, did not significantly change the EPSC amplitude (n = 5; data not shown).
Consistent with this result, application of 2 µm conotoxin further suppressed the small EPSCs recorded in AIIs in the presence of nimodipine, mibefradil, and agatoxin (Fig. 9I). Under the experimental condition without any CaV channel blocker preincubation, application of 2 µm conotoxin slightly reduced the peak amplitude of EPSCs recorded in AIIs (Fig. 9J,K; control, −354.80 ± 87.74 pA; conotoxin, −304.30 ± 72.57 pA; control vs conotoxin, t(4) = 3.988, p = 0.0163, Student's t test). In addition, conotoxin alone had little effect on ChR2-evoked membrane potential change in RBs (Fig. 9L; n = 3). These results indicate that N-type CaV conductances, albeit to a lesser extent than other types, also mediate Ca2+ influx into RB axon terminals and subsequent neurotransmitter release.
To exclude the possibility that the observed effects of CaV channel antagonists on evoked release might arise from exogeneous expression of ChR2 in RBs, we repeated our experiments while assessing EPSCs in wild-type mice. Here, EPSCs were evoked in AIIs by pressure ejection of LY 341495 (the mGluR6 antagonist; 5–10 µm) onto the dendrites of RBs and ON cone BCs located at the outer plexiform layer (Fig. 9M; Snellman et al., 2009). As for ChR2-evoked EPSCs (Fig. 9B–F), LY-evoked EPSCs were reduced substantially by co-application of 10 µm nimodipine and 10 µm mibefradil and further suppressed by additional application of 200 nm agatoxin (Fig. 9N–P; for better comparison among groups, all the data were normalized; nimodipine, 0.324 ± 0.041, n = 8; nimodipine + mibefradil, 0.160 ± 0.029, n = 6; nimodipine + mibefradil + agatoxin, 0.052 ± 0.018; nimodipine vs nimodipine + mibefradil, t(11.64) = 3.269, p = 0.0070, Welch's t test; nimodipine + mibefradil vs nimodipine + mibefradil + agatoxin, t(8.439) = 3.168, p = 0.0120, Welch's t test; nimodipine vs nimodipine + mibefradil + agatoxin, t(9.501) = 6.067, p = 0.0001, Welch's t test). Note that in these experiments, small residual postsynaptic currents were observed in the presence of 20 µm DNQX: these are gap-junction mediated and arise from depolarization of electrically-coupled ON cone BCs (Fig. 9M–O). These coupling currents were subtracted when measuring the integrals of the LY-evoked EPSCs under different experimental conditions (Fig. 9P). The unique expression patterns of CaV channels that control glutamate release at BC ribbon synapses are summarized in a schematic diagram shown in Figure 10.
A schematic diagram showing the unique expression patterns of CaV channels that control neurotransmitter release from mouse retinal BCs. While rod and cone photoreceptors exclusively express L-type CaV1.4, β2, and α2δ4 subunits, BCs show much more complex expression patterns of CaV channels: P/Q-type (CaV2.1), N-type (CaV2.2), T-type (CaV3), and L-type (CaV1) channels all combine to control glutamate release from BC ribbon synapses; furthermore, CaV1, CaV3, β, and α2δ subunits adhere to a “multisubtype” rule, i.e., a combination of multiple subtypes for each subunit in individual BCs. Note that the scale of the various cellular components is distorted to show the expression patterns of CaV channels in the axon terminals more clearly. CaV, voltage-gated Ca2+ channels.
Discussion
We conducted integrative analyses to unveil the molecular composition, subcellular localization, and functional roles of various CaV channels in RBs. We found high molecular diversity of CaV channel subunits at RB ribbon synapses and thereby clarified the poorly understood molecular identities of CaV channels in RBs. Interestingly, we demonstrated, to our knowledge, for the first time, that CaV2 channels function at mammalian ribbon synapses and contribute to synaptic transmission. As well, our work indicates that expression patterns of CaV1, CaV3, β, and α2δ subunits adhere to a “multisubtype” rule, i.e., a combination of multiple subtypes rather than a single subtype for each CaV channel type in an individual RB. Since scRNA-seq analysis accurately predicts the expression of CaV channel subtypes in RBs, our study will be helpful as a guide for future studies of CaV channel expression and functions in other BCs. Finally, as the “multisubtype” rule may apply to other central and peripheral synapses, we suggest that it should be taken into account in future studies of CaV channels.
P/Q-type and N-type CaV channels function at ribbon synapses
Ca2+ influx through P/Q-type and/or N-type CaV channels underlies neurotransmission at most synapses (Dolphin and Lee, 2020), including inhibitory synapses in the retina (Van Hook et al., 2019), but we have not found it reported in studies of mammalian ribbon synapses. Rather, there is a general consensus that exocytosis at ribbon synapses mainly results from Ca2+ entry via L-type CaV channels (Pangrsic et al., 2018; Dolphin and Lee, 2020). Interestingly, knock-out of CaV1.3, the major α1 subunit in mouse cochlear inner hair cells (IHCs) and vestibular hair cells (VHCs), decreases Ca currents by ∼85% and 50%, respectively, suggesting that other types of CaV channels may also be present (Platzer et al., 2000; Dou et al., 2004). Indeed, previous studies have reported the presence of T-type CaV channels in chick basilar hair cells, mouse IHCs, and rat outer hair cells (OHCs) as well as in retinal BCs of different species (Kaneko et al., 1989; de la Villa et al., 1998; Pan, 2000, 2001; Levic et al., 2007; Inagaki et al., 2008; Nie et al., 2008). Additionally, the non-L-type Ca currents in amphibian hair cells from saccules and semicircular canals may be mediated by N-type or R-type CaV channels (Su et al., 1995; Martini et al., 2000; Rodriguez-Contreras and Yamoah, 2001).
To date, however, no published studies confirm the expression and function of P/Q-type CaV channels at ribbon synapses. Therefore, we were surprised that scRNA-seq data analysis indicated the expression of both P/Q-type and N-type CaV channels in almost all BCs (Fig. 1). The presence of P/Q-type and N-type CaV channels in RBs was further confirmed by scRT-PCR (Fig. 2), and the subcellular localization of these two types of CaV channels, especially their expression in the axon terminals, was then determined by our immunohistochemical and immunocytochemical studies (Figs. 3, 4, 8). Functional analysis of neurotransmission at RB→AII synapses further proved that both P/Q-type and N-type CaV channels mediated Ca2+ influx into RB terminals (Figs. 3, 4) and controlled neurotransmitter release (Fig. 9). Given the high consistency of observations among scRNA-seq, scRT-PCR, immunohistochemical, immunocytochemical, and electrophysiological analyses, we could imagine that P/Q-type and N-type CaV channels also exist and function in other BC terminals (Fig. 1). Thus, in the present study, we demonstrate, for the first time, that P/Q-type and N-type CaV channels function at a mammalian ribbon synapse.
Molecular identity of L-type CaV channels in BCs
L-type CaV channels are the major conduits for fast and sustained Ca2+ influx essential for neurotransmission at ribbon synapses (Pangrsic et al., 2018; Dolphin and Lee, 2020). These CaV1-containing channels, including CaV1.2–1.4, are thought to be well-suited to mediate the fast and continuous transmitter release which is necessary for encoding sensory information at ribbon synapses (Pangrsic et al., 2018; Dolphin and Lee, 2020). Previous studies have shown that CaV1.3 channels are expressed predominantly in cochlear IHCs and OHCs and VHCs, whereas CaV1.4 channels are expressed almost exclusively in the retina (Pangrsic et al., 2018; Van Hook et al., 2019; Dolphin and Lee, 2020). Indeed, rod and cone photoreceptors in the retina predominantly express L-type CaV channels, and CaV1.4 is the only pore-forming α1 subunit detected in them (Mansergh et al., 2005; Zeitz et al., 2015).
Despite years of effort from many researchers, the molecular identity of CaV1 channels in BCs remains elusive. According to the published literature, all CaV1 channel subtypes may be present in these cells (Pangrsic et al., 2018). Based on their relatively fast activation kinetics at relatively hyperpolarized potentials, both CaV1.3 and CaV1.4 channels are thought to be best-suited to encode graded potential changes in BCs and therefore are considered the best candidates for the presynaptic CaV1 channels in BC terminals (Pangrsic et al., 2018; Dolphin and Lee, 2020). Indeed, CaV1.4 expression has been reported in the IPL of rodent retinas and more specifically in the axon terminals of isolated mouse RBs (Morgans, 2001; Berntson et al., 2003; Mansergh et al., 2005). Confusing this picture, however, is the fact that Ca currents recorded from BCs show Ca2+-dependent inactivation whereas the inactivation of CaV1.4-containing channels is Ca2+-independent (Koschak et al., 2003). By contrast, CaV1.3 channels exhibit relatively weak and slow Ca2+-dependent inactivation (Xu and Lipscombe, 2001). RT-PCR analysis has revealed that ON-type BCs in goldfish retina express CaV1.3 channels (LoGiudice et al., 2006). Additionally, knock-out of CaV1.3 in mice reduces the b-wave amplitude of dark-adapted ERG, which reflects primarily the activity of RBs, suggesting the expression of CaV1.3 in mouse RBs (Shi et al., 2017). Thus, the kinetics of macroscopic Ca currents recorded from BCs appear to reflect a combination of CaV1.3 and CaV1.4 properties.
In the present work we found that RBs expressed three subtypes of CaV1 channels, including CaV1.1, CaV1.3 and CaV1.4, rather than a single CaV1 subtype as expected (Figs. 1, 2, 5, 8). Most importantly, both CaV1.3 and CaV1.4 channels were located near ribbons in RB axon terminals (Figs. 5, 8) and thus are likely to underlie the L-type conductance-mediated Ca2+ influx driving neurotransmitter release. Notably, expression of CaV1.2 was rarely observed by scRNA-seq and scRT-PCR analyses (Figs. 1, 2), suggesting that the presence of CaV1.2 in RB terminals detected by immunohistochemistry in previous studies might be because of nonspecific labeling by the anti-CaV1.2 antibodies used (Satoh et al., 1998; Rieke et al., 2008).
Also, it has been reported that CaV1.1 channels are expressed exclusively at the dendritic tips of RBs (Specht et al., 2009; Soto et al., 2012; Tummala et al., 2014). However, there is a more recent study arguing that the CaV1.1 antibody used in previous studies may cross-react with another synaptic protein, GPR179 (Hasan et al., 2016). The subcellular localization and functions of any CaV1.1 channels in RBs thus remain uncertain. Nevertheless, expression of different combinations of CaV1 subtypes rather than a single CaV1 subtype likely is a general rule for almost all BCs (Fig. 1). Consequently, studies of single CaV1 channel-type knock-out animals are likely to be inconclusive, and double-knock-out or even triple-knock-out animals may be required for future investigations.
Multiple types of CaV channels function at a single synapse
Co-existence of multiple types of CaV channels, at either conventional or ribbon synapses, has been commonly observed in many brain regions. At ribbon synapses, L-type and T-type Ca currents have been detected simultaneously in IHCs, OHCs, VHCs in the inner ear, and BCs in the retina (Kaneko et al., 1989; de la Villa et al., 1998; Pan, 2000, 2001; Dou et al., 2004; Levic et al., 2007; Inagaki et al., 2008; Nie et al., 2008).
Different types of CaV channels at a single synapse have been found to couple differentially to transmitter release. For example, at the calyx of Held synapses, P/Q-type CaV channels are more tightly coupled to transmitter release than N-type and R-type channels, probably because of their distinct presynaptic localization (Wu et al., 1999). Immunoelectron microscopy analysis of parallel fiber→Purkinje cell synapses in the cerebellum reveals that CaV2.1 channels are clustered within the active zone and exhibit nanodomain coupling, while CaV2.2 channels are located relatively distant from release sites (Eggermann et al., 2011; Kusch et al., 2018). Interestingly, nanodomain coupling between L-type CaV channels and release sites has also been found to exist at most ribbon synapses, including the RB→AII synapse (Jarsky et al., 2010; Pangrsic et al., 2018).
Co-expression of CaV1 and CaV2 channels at synapses, however, rarely has been reported. Therefore, it was quite striking to find in our present work that P/Q-type, N-type, L-type, and T-type CaV channels were co-expressed in a single type of neurons, i.e., BCs, and all involved in neurotransmitter release from RBs. It will be interesting, then, to explore in the future whether these different types of CaV channels function cooperatively or independently to support various Ca2+-dependent physiological processes. As well, future studies could address the hypothesis that distinct combinations of CaV β and α2δ subunits in various BC types diversify BC outputs and underlie differences in inner retinal processing channels.
Footnotes
↵†J.H.S. and J.-B.K. are co-senior authors.
This work was supported by the National Institutes of Health Grant EY017836 (to J.H.S.) and the fundamental research funds of State Key Laboratory of Ophthalmology, Sun Yat-sen University (to J.-B.K.). We thank Ting-Ting Zhang and Min Gao for technical assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Jiang-Bin Ke at kejiangbin{at}mail.sysu.edu.cn or jbke99{at}hotmail.com
