Introduction

Synapses of sensory receptor cells of the inner ear and the retina are hallmarked by the presence of an electron-dense structure called the synaptic ribbon, which tethers synaptic vesicles (SVs) at the presynaptic active zone (AZ, [Moser et al, 2019; Matthews & Fuchs, 2010]). Proposed functional roles of the ribbon include (i) its role as a SV replenishment machine for tireless neurotransmission (Bunt, 1971; Holt et al, 2004; Khimich et al, 2005; Frank et al, 2010; Snellman et al, 2011; Graydon et al, 2011; Vaithianathan et al, 2016), (ii) “super-scaffold” regulating the abundance, topography, and function of AZ players such Ca²⁺ channels and release sites and their tight coupling (Khimich et al, 2005; Frank et al, 2010; Wong et al, 2014; Maxeiner et al, 2016; Jean et al, 2018; Grabner & Moser, 2021), and iii) coordination of SV release (Heidelberger et al, 1994; Glowatzki & Fuchs, 2002; Singer et al, 2004; Edmonds, 2004; Mehta et al, 2013). Yet, our molecular understanding of how these postulated functions are executed by synaptic ribbons is still limited.

Likewise, the clustering of Ca²⁺ channels at the AZ, how it is varied among different or even the same synapse types and how the ribbon contributes towards this, remain active areas of research. For example, ribbon synapses of mature inner hair cells (IHC) assemble Ca_V1.3 voltage-gated Ca²⁺ channels at the base of the ribbon, but vary in shape, size and channel complement (20-330) of the Ca_V1.3 clusters (Frank et al, 2010; Wong et al, 2014; Neef et al, 2018) partially dependent on the position within the IHC (Ohn et al, 2016; Özçete & Moser, 2021). IHC synapses lacking AZ-anchored ribbons in Bassoon mutant mice showed fewer Ca²⁺ channels with altered topography (Frank et al, 2010; Neef et al, 2018). While genetic disruption of Ca²⁺ channel-tethering AZ proteins RIM2 (Jung et al, 2015) and RBP2 (Krinner et al, 2017) reduced the abundance of Ca²⁺ channels at IHC synapses, their topography retained the typical stripe-like shape of Ca²⁺ clusters. Constitutive deletion of RIBEYE, the core scaffold protein of the synaptic ribbon, provided, so far, the most direct test of the scaffolding role of the ribbon (Maxeiner et al, 2016; Jean et al, 2018; Becker et al, 2018; Grabner & Moser, 2021). Here, ribbon-less synapses of IHCs showed an altered Ca²⁺ channel topography with a more spatially widespread presynaptic Ca²⁺ signal but intact Ca²⁺ channel complement (Jean et al, 2018). Moreover, the voltage-dependence of activation as well as the inactivation of Ca²⁺ channels appeared to be altered, suggesting a possible role of the synaptic ribbon in also regulating Ca²⁺ channel physiology (Jean et al, 2018). Changes in Ca²⁺ channel physiology upon deletion of RIBEYE were also found for Ca_V1.4 at ribbon-less rod photoreceptor AZs (Grabner & Moser, 2021). Finally, IHCs lacking piccolino, a ribbon-synapse specific isoform of the AZ protein piccolo, show AZs with smaller synaptic ribbons and altered clustering yet normal complement of Ca²⁺ channels (Michanski et al, 2023).

So how does the synaptic ribbon tune into the orchestra of AZ proteins that collectively organize Ca²⁺ channel abundance, topography, and function? To what extent can the effects observed upon disruption of Bassoon be attributed to the concomitant loss of the ribbon? These questions seem even more relevant given that, so far to our knowledge, no direct interactions between Bassoon (Frank et al, 2010) or RIBEYE and the Ca²⁺ channel have been described. Here, we adopted a bottom-up synthetic biology approach to assemble a minimal presynaptic protein machinery required to bring together RIBEYE and Ca²⁺ channels in a heterologous expression system to assess the role of the synaptic ribbon in regulating Ca²⁺ channel clustering and physiology. Similar approaches have previously been employed for reconstituting aspects of conventional synapses, for instance in “hemisynapses” between co-cultured neurons and non-neuronal cells overexpressing presynaptic or postsynaptic components (for review see Craig et al, 2006). More recently, Munc13-1 supra-molecular assemblies which recruit release machinery proteins were reconstituted (Sakamoto et al, 2018).

Expression of RIBEYE in synapse-naïve cell lines such as COS-7 cells led to large cytosolic assemblies (Schmitz et al, 2000). Membrane-proximal RIBEYE clusters and ribbon-like electron dense structures with vesicles could be found in a retinal progenitor cell-line (R28, [Magupalli et al, 2008]) yet, these cells might not be considered synapse-naïve. Regardless, the presence and function of Ca²⁺ channels at synthetic ribbon synapses have not yet been studied. The need for co-expressing at least three Ca²⁺ channel subunits (Ca_Vα_x, Ca_Vβ_x and Ca_Vα₂δ_x) in synapse-naïve cell lines renders acute co-expression of AZ multidomain proteins cumbersome. Here we took advantage of synapse-naïve Human Embryonic Kidney 293 (HEK293) cells stably expressing inducible Ca_V1.3α₁, which is the predominant subtype of voltage-gated Ca²⁺ channels at IHC ribbon synapses (Brandt et al, 2003; Platzer et al, 2000; Dou et al, 2004), as well as constitutively expressed Ca_Vβ₃ and Ca_Vα₂δ₁. We co-expressed RIBEYE along with membrane targeted Bassoon and observed structures with striking resemblance to IHC synaptic ribbon-type AZs. We characterized the structure and function of these synthetic ribbon-type AZs and identified Bassoon, RBP2, RIBEYE and Ca_V1.3 channels as the minimal components required for assembling a basic ribbon-type AZ. We observed that synthetic ribbons recruit Ca_V1.3 channels in large clusters and show localized Ca²⁺ influx upon stimulation. Our results support the role of synaptic ribbons in promoting Ca_V1.3 channel clustering. Synthetic ribbon-type AZs offer a novel approach for functionally studying protein-protein interaction and will likely complement, refine and reduce experiments on native ribbon synapses.

Results

Membrane-targeted Bassoon recruits RIBEYE to the cell membrane in HEK293 cells where structures alike inner hair cell synaptic ribbons are formed

Human epithelial HEK293 cells have the advantage of not expressing the synaptic machinery components studied here. This provides a clean background for reconstituting synthetic synapses from a minimal set of synaptic proteins. Our first step towards assembling a ribbon-type AZ in a heterologous expression system was to express RIBEYE, the core scaffold protein of the synaptic ribbon, and target it to the cell membrane. We performed transient transfection of a RIBEYE construct with a C-terminal EGFP tag in HEK293 cells and observed large spherical clusters of RIBEYE that appeared largely cytosolic (Fig 1Aii), in contrast to a diffuse cytosolic distribution when merely expressing EGFP (Fig 1Ai). These RIBEYE clusters form due to self-assembling properties of RIBEYE via multiple sites of homophilic interaction as have been demonstrated before in several cell lines. They do not colocalize with the endoplasmic reticulum (ER), Golgi apparatus, or lysosomes and hence, are unlikely to reflect RIBEYE entrapped in protein trafficking pathways or to represent degradation products of overexpressed RIBEYE (Fig EV1 A, B and C).

Figure 1-

Figure EV1-

Next, for membrane targeting of these cytosolic RIBEYE clusters, we co-expressed Bassoon (Fig 1Aiii). Prior work on the molecular underpinnings of ribbon synapses had identified the multidomain cytomatrix of the active zone protein Bassoon (tom Dieck et al, 1998) to critically contribute to anchoring the synaptic ribbon to the AZ membrane (Khimich et al, 2005; Dick et al, 2003; tom Dieck et al, 2005). Co-expression of full-length Bassoon along with RIBEYE in HEK293 cells showed colocalizing clusters of the two proteins that, however, remained largely cytosolic (Fig 1Aiv).

Next, for plasma membrane-targeting of Bassoon, we generated a construct by removing the first 95 N-terminal amino acids of Bassoon and replacing these with a palmitoylation consensus sequence of the neuronal Growth Associated Protein 43 (GAP43). We refer to this as palm-Bassoon throughout, and we used constructs with and without a C-terminal EGFP tag (Fig 1B). Expression of either of these palm-Bassoon constructs in HEK293 cells showed comparable immunofluorescence patterns with Bassoon puncta spread across the periphery of the cell, largely colocalizing with the endogenously expressed membrane protein Na, K-ATPase α₁ (data representative of 3 transfections, Fig 1C, D). Comparing the ratio of Bassoon signal intensity at the periphery versus inside of the cell in randomly selected single sections from confocal stacks of palm-Bassoon and Bassoon transfected cells (N = 10 cells, 3 transfections per group, Fig 1E) demonstrated a higher Bassoon signal intensity at the periphery of cells expressing the palm-Bassoon construct (****P < 0.0001, Mann-Whitney-Wilcoxon test), implying successful plasma membrane targeting of Bassoon.

Next, we co-expressed palm-Bassoon and RIBEYE in HEK293 cells and observed colocalizing RIBEYE and Bassoon immunofluorescent puncta at the periphery of cells (7 transfections; Fig 2Ai). Closer inspection of these immunofluorescent puncta with STimulated Emission Depletion (STED) nanoscopy (Fig 2Aii) revealed discrete structures typically consisting of ellipsoidal RIBEYE clusters juxtaposing on top of plate-like palm-Bassoon structures which seemingly anchor the RIBEYE clusters to the plasma membrane. Interestingly, we found the morphology of the RIBEYE + palm-Bassoon structures to be strikingly reminiscent of the arrangement of the two proteins in IHC ribbon synapses, where an ellipsoid/spherical synaptic ribbon composed of RIBEYE is found seated on a Bassoon plate that anchors it to the presynaptic AZ membrane (e.g. Wong et al, 2014; Michanski et al, 2019, 2023) (Fig 2B, i and ii; data from Michanski et al, 2023).

Figure 2-

We compared the RIBEYE and Bassoon signal intensities at the periphery versus inside of the cell in randomly selected single sections from confocal stacks of HEK cells expressing RIBEYE and palm-Bassoon (N = 9 cells) and HEK cells expressing RIBEYE and full-length Bassoon (N=9 cells). We found an increased peripheral distribution of both RIBEYE and Bassoon when using palm-Bassoon (Fig 2C, ****P < 0.0001, Mann-Whitney-Wilcoxon test), implying successful membrane targeting of RIBEYE by palm-Bassoon. In each transfection, ∼10% cells showed co-expression of both RIBEYE and palm-Bassoon (for representative sample overview see Fig EV2). Of those, cells expressing discrete RIBEYE-palm Bassoon structures were discernible by the characteristic RIBEYE distribution along the periphery of the cell. This peripheral distribution in turn seemingly depends upon RIBEYE and palm-Bassoon expression ratios (shown in Figure EV2B) as cells with little to no palm-Bassoon expression show predominantly cytosolic RIBEYE puncta and were not used for analysis. For simplicity, we henceforth refer to the structures composed of RIBEYE and palm-Bassoon in HEK293 cells as SyRibbons (for synthetic ribbons). We next performed three-dimensional surface renderings using Imaris 9.6 (Oxford Instruments) to assess these structures. In a given cell, only structures with colocalizing RIBEYE and palm-Bassoon immunofluorescence were considered for analysis to exclude the occasional non-membrane localized spots (on average 51.75 ± 40.84 RIBEYE surfaces colocalizing with Bassoon surfaces per cell; N = 20 cells). The volumes of RIBEYE and palm-Bassoon surfaces of SyRibbons were smaller on average and more variable (average volume ± standard deviation (SD) = 0.19 ± 0.23 µm³ with a coefficient of variation (CV) = 1.23 for RIBEYE; n_spots = 864 and volume = 0.17 ± 0.18 µm³ with CV = 1.10 for Bassoon; n_spots = 951; data from N = 20 cells, quantifications from 5 sample transfections) than volumes of synaptic ribbons and Bassoon immunofluorescent puncta from rat IHCs (volume = 0.29 ± 0.17 µm³, CV = 0.59 for RIBEYE; n_spots = 290, and volume = 0.27 ± 0.18 µm³, CV = 0.67 for Bassoon; n_spots = 314, data from N = 29 cells), (Figure 2D, E). Moreover, volumes of RIBEYE surfaces show a high positive correlation to volumes of corresponding palm-Bassoon surfaces (P_r = 0.778, ****P < 0.0001), implying larger palm-Bassoon structures may recruit bigger RIBEYE structures to the plasma membrane. We also note that volume of RIBEYE surfaces in SyRibbons appears smaller and well-regulated in contrast to the predominantly large, cytosolic RIBEYE assemblies in cells co-expressing RIBEYE and full-length Bassoon (volume = 0.30 ± 0.36 µm³, n_spots = 460, N = 10 cells, ****P < 0.001, Mann-Whitney-Wilcoxon Test). In turn, Bassoon clusters seemed regulated by co-expressed RIBEYE: Bassoon surfaces at the plasma membrane were larger at SyRibbons than in the absence of RIBEYE in cells only expressing palm-Bassoon (volume = 0.14 ± 0.21 µm³, n_spots = 2217, N = 13 cells, ****P < 0.001, Mann-Whitney-Wilcoxon Test).

Nonetheless, the fact that next to structures with volumes comparable to IHC ribbons, we also encountered smaller and larger structures, likely reflects poorer regulation of RIBEYE and palm-Bassoon expression in the heterologous system. RIBEYE clusters constituting SyRibbons appeared more spherical than the plate-like Bassoon structures in the same cells (sphericity Ψ = 0.91 ± 0.05, n_spots = 864 for RIBEYE versus Ψ = 0.86 ± 0.07, n_spots= 951 for Bassoon, data from N = 20 cells; ****P < 0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test). This follows the same trend as in rat IHCs where RIBEYE spots indeed appear more spherical (Ψ = 0.90 ± 0.05, n_spots = 290) compared to Bassoon spots (Ψ = 0.76 ± 0.10, n_spots = 314); ****P < 0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test, data from N = 29 cells (Figure 2F).

Figure EV2-

To study the potential of SyRibbons to cluster Ca²⁺ channels, we employed HEK293 cells stably expressing an inducible transgene of Ca_V1.3α₁ along with constitutive transgenes for Ca_Vβ₃ and Ca_Vα₂δ₁. We either used tetracycline for inducible Ca_V1.3α₁ channel expression (Fig 3A) or performed transient transfections with Ca_V1.3 constructs containing an N-terminal EGFP- (Fig 3B) or Halo-tag (Fig 3C) for direct fluorescence imaging. Expression of either Ca²⁺ channel complex resulted in small clusters at the plasma membrane as represented in Fig 3D (average volume 0.08 ± 0.13 µm³, n_spots = 949, N = 12 cells for untagged inducible Ca_V1.3α₁; 0.07 ± 0.08 µm³, n_spots = 392, N = 11 cells for EGFP-Ca_V1.3; and 0.07 ± 0.09 µm³, n_spots = 1281, N = 16 cells for Halo-Ca_V1.3α; P_{untagged/EGFP} = 0.547, P_{untagged/Halo} > 0.999, P_Halo/EGFP = 0.582, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test).

Figure 3-

Co-expression of palm-Bassoon did not show colocalization with Ca_V1.3, which is in line with observations with full-length Bassoon (Frank et al, 2010) and the same applied when co-expressing RIBEYE and Ca_V1.3, arguing against a direct interaction for both (Figure EV3). Rab-interacting molecule-binding protein (RIM-BP or RBP) links Bassoon to Ca²⁺ channels (Davydova et al, 2014) and hence was an interesting candidate for tethering Ca_V1.3 to Bassoon clusters. RBP has been shown to interact with Ca_V1.3 (Hibino et al, 2002) and to be required for normal Ca_V1.3 Ca²⁺ channel clustering at the IHC ribbon synapse (Krinner et al, 2017, 2021). Indeed, when co-expressing RBP2 with palm-Bassoon and Ca_V1.3, we found juxtaposed palm-Bassoon and Ca_V1.3 both with immunolabeled (Fig. 4A) and live-labeled (Fig. 4B, C) proteins indicating successful clustering of Ca_V1.3 at synthetic AZs.

Figure EV3-

Figure 4-

We next performed ruptured whole-cell patch-clamp to assess if RBP2 and/or palm-Bassoon or RIBEYE co-expression results in changes in Ca_V1.3 Ca²⁺ currents. After tetracycline treatment for Ca_V1.3α₁ expression (18-24 hours), cells were transfected with both RBP2-p2A-mKATE2 and palm-Bassoon-GFP, with only palm-Bassoon-GFP, only RBP2-p2A-mKATE2 or RIBEYE-GFP constructs (Figure 5A). After 18-24 hours, the cell culture media was changed, and cells were recorded 24-36 hours afterwards. We used 10 mM [Ca²⁺]_e and recorded Ca²⁺ current (density) - voltage relations (IVs) ∼1 minute after establishing the whole-cell configuration by applying step depolarizations of 20 ms from −86.2 mV to 58.8 mV in 5 mV increments. Ca²⁺ current density was on an average approximately twice as large in cells expressing both RBP2 and palm-Bassoon when compared with induced-only controls (Figure 5B, D; **P = 0.0041, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test). Ca²⁺ current densities also tended to be larger for cells expressing only palm-Bassoon or RBP2, but this difference was not statistically significant (P = 0.374 and 0.164 respectively, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test), while for cells expressing RIBEYE there was no trend to differ (P > 0.999, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test). We did not observe any noticeable differences in voltage-dependence, activation kinetics or inactivation kinetics of Ca_V1.3 Ca²⁺ current upon expression of any of these AZ proteins (Figure 5C, E, F; Figure EV4A, B).

Figure 5-

Figure EV4-

Larger synthetic ribbon-type AZs establish larger Ca_V1.3 Ca²⁺ channel clusters

Next, we tetra-transfected HEK293 cells with palm-Bassoon, Halo-tagged Ca_V1.3α₁, RBP2 and RIBEYE to explore the impact of co-expressing of RIBEYE on the clustering of Ca_V1.3 Ca²⁺ channels (Figure 6A for transfection scheme). Using confocal imaging of immunofluorescently labelled RIBEYE, Halo-Ca_V1.3 and Bassoon (RBP2 immunofluorescence was imaged using epifluorescence to select tetra-transfected HEK293 cells), we observed colocalization of Ca_V1.3α1, Bassoon and RIBEYE (Fig. 6B). Line profiles drawn tangentially along the plasma membrane showed that beyond the general distribution of a Ca_V1.3α₁ signal in the plasma membrane, Ca_V1.3α₁ signal hotspots occurred underneath the SyRibbons (Fig. 6C). The signal intensity of RIBEYE immunofluorescence positively correlated with the Ca_V1.3 signal immunofluorescence, (P_r = 0.839 and 0.886 from two sample line scans, ****P < 0.0001), which has previously been reported for immunolabeled native ribbon-type AZs of mouse IHCs (Ohn et al, 2016). We next turned to 2-color STED imaging which provided improved spatial resolution and highlighted the confined localization of Ca_V1.3 Ca²⁺ -channel clusters underneath SyRibbons as shown in Figure 6D.

Figure 6-

We performed surface renderings of confocally imaged Halo-tagged Ca_V1.3α₁, RIBEYE and palm-Bassoon immunofluorescence spots as described above for estimating the volume of SyRibbons and the Ca_V1.3 Ca²⁺ channel clusters (Fig.6E, F). This revealed significantly larger Ca_V1.3 Ca²⁺ channel clusters when colocalizing with SyRibbons in HEK293 + Halo-Ca_V1.3 + RBP2 + SyRibbons cells (average volume ± SD = 0.18 ± 0.28 µm³, n_spots= 634, N = 13 cells, data representative of 5 transfections) compared to non-colocalized Ca_V1.3 Ca²⁺ channel clusters from the same cells (volume = 0.06 ± 0.10 µm³, n_spots= 1354, N = 13 cells; ****P < 0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test) and Ca_V1.3 Ca²⁺ clusters in HEK293 cells solely expressing Halo-Ca_V1.3 Ca²⁺ channels (volume = 0.07 ± 0.09µm³, n_spots= 1281, N = 16 cells, data representative of 4 transfections; ****P < 0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test). Ca_V1.3 Ca²⁺ channel cluster volumes in the latter two did not differ significantly (P = 0.099, Kruskal-Wallis test with post hoc Dunn’s multiple comparison test). Moreover, volumes of Ca_V1.3 cluster surfaces show a moderate positive correlation with volumes of corresponding RIBEYE surfaces (P_r = 0.385, ****P < 0.0001), similar to previous observations at IHC ribbon synapses (Frank et al, 2009; Michanski et al, 2023; Ohn et al, 2016).

Ca²⁺ imaging reveals increased Ca²⁺ signal intensity at synthetic ribbon-type active zones

Finally, we set-out to functionally characterize synthetic ribbon-type AZs. We combined patch-clamp recordings of Ca_V1.3 Ca²⁺ influx with spinning-disk confocal microscopy to visualize Ca²⁺ signals at the synthetic ribbon-type AZs in HEK293 cells. Following tetracycline-induction for Ca_V1.3 expression, HEK293 cells were co-transfected with constructs expressing RIBEYE-GFP, untagged palm-Bassoon and RBP2-p2A-mKATE2, (Figure 7A). We recorded mKATE2 positive cells that expressed peripheral GFP puncta, indicative of RIBEYE and palm-Bassoon co-expression. We employed peripheral GFP expression as a proxy of SyRibbons for functional analysis (Figure 7B). Ruptured patch-clamp recordings were performed with 10 mM [Ca²⁺]_e and 10 mM intracellular EGTA to enhance the signal to background ratio for visualizing Ca²⁺ influx using the low-affinity red-shifted Ca²⁺ indicator Calbryte590 (100 µM, k_d = 1.4 µM). After loading the cell for approximately a minute, we first applied step depolarizations of 20 ms from −83 to +62 mV in 5-mV step increments to measure Ca²⁺ current (density)–voltage relations.

Figure 7-

We did not observe any changes in either the Ca²⁺ current density (P = 0.293, t-test) or the voltage-dependence of Ca²⁺ current (P = 0.780, t-test when comparing V_half) when comparing cells expressing Ca_V1.3 + RBP2 + SyRibbons with cells expressing only Ca_V1.3 (Figure 7C and 7D). Next, we applied a depolarizing pulse to +2 mV for 500 ms and acquired images at a frame rate of 20 Hz (Figure 7E). The intracellular Ca²⁺ signals mediated by Ca_V1.3 Ca²⁺ influx were visualized as an increase in Calbryte590 fluorescence. The increase of Calbryte590 fluorescence appeared to spread throughout the HEK293 cell membrane with discrete regions of higher signal intensity at sites underneath the SyRibbons (exemplary cell shown in Figure 7F with zoom-in at site with SyRibbon shown in 7G). We note that we did not find noticeable red fluorescence at SyRibbons in the absence of depolarization, which is similar to findings with Fluo-4FF in IHCs (Frank et al, 2009), but different from the situation for the large spherical presynaptic bodies in bullfrog hair cells that are stained by Fluo-3 (Issa & Hudspeth, 1996). Line profiles drawn tangentially to the membrane in composite ΔF image of Calbryte590 signal and RIBEYE-GFP showed a high correlation between the localization of SyRibbons and peak intensity of Calbryte590 fluorescence increase, indicating preferential Ca²⁺ signaling at the SyRibbons (two representative line scans shown in Figure 7H, Pearson correlation coefficients of 0.783 and 0.757). Recordings were made from 24 cells (7 transfections) and ROIs (diameter = 2 µm) at regions with and without SyRibbons were analyzed as shown in Figure 7I. Calbryte590 ΔF_max/F₀ values were calculated for each ROI which on an average showed higher ΔF_max/F₀ values for ROIs with SyRibbons (on average ∼26% higher) than for ROIs without SyRibbons (*P = 0.018, paired t-test, Figure 7J). We note that occasionally in some cells we observed a higher Calbryte590 signal increase at sites without SyRibbons, and we attribute this to the likely presence of AZ-like clusters composed of RBP2 and palm-Bassoon in these regions which we could not visualize.

Discussion

In this study, we reconstituted and characterized a minimal ribbon-type AZ model system using heterologous expression. Co-expressing Ca_V1.3 Ca²⁺ channels with membrane targeted Bassoon for RIBEYE anchorage to the plasma membrane and attracting exogenous RBP2 for clustering Ca²⁺ channels in HEK293 cells led to structures recapitulating basic aspects of IHC ribbon synapses. Despite the poor regulation of transgene expression in the synapse-naïve HEK293 cells, the AZ-like structures, in the subset of cells were similar in morphology to the native IHC AZs. This applied to the synthetic ribbons as well as to the AZ-like clusters of Bassoon, RBP2 and Ca_V1.3 Ca²⁺ channels. However, all three components exhibited variability beyond the substantial natural heterogeneity found among the IHC AZs (Moser et al, 2023). Functionally, Bassoon and RBP2 co-expression resulted in larger whole-cell Ca_V1.3 currents. While additional expression of RIBEYE does not seem to enhance whole-cell Ca²⁺ currents, we demonstrate a stronger, localized Ca²⁺ influx at SyRibbons. As SyRibbons partially resemble native hair cell AZs, we expect this easily available and experimentally accessible system to serve as an advanced testbed for functional interactions of AZ proteins and Ca²⁺ channels. Further efforts towards reconstituting synaptic vesicles and their release sites will help to enhance the utility of these synthetic AZs.

Synthetic active zone reconstitution model to study ribbon synapse assembly

By demonstrating that membrane-targeted Bassoon suffices to anchor ribbon-like RIBEYE assemblies, this work adds to the top-down and bottom-up evidence for a key role of Bassoon in attracting RIBEYE to the presynaptic density at the AZ in photoreceptors and hair cells (Dick et al, 2003; Khimich et al, 2005; Jing et al, 2013; tom Dieck et al, 2005). Additional expression of multi-domain proteins of the presynaptic density such as CAST (Ohtsuka et al, 2002; Inoue et al, 2006) might replace the need for artificial palmitoylation of Bassoon for membrane targeting. Furthermore, our results support a model of a bidirectional control of AZ size and shape by RIBEYE and proteins of the presynaptic density. The amount of RIBEYE assembled in membrane localized SyRibbons appeared regulated strongly contrasting the seemingly unregulated cytosolic RIBEYE assemblies dominating HEK293 cells in the absence of membrane-targeted Bassoon. Vice versa, clusters of Bassoon at the plasma membrane were larger at SyRibbons than in the absence of RIBEYE. Finally, the size of SyRibbons seemingly scaled with the size of Ca_V1.3 and Bassoon clusters similar to what is observed in IHCs (Ohn et al, 2016). Such a model is consistent with results from genetic perturbation of native ribbon synapses: i) disruption of RIBEYE leads to a disintegration of Bassoon into smaller subclusters in IHCs (Jean et al, 2018), and ii) disruption of CAST and ELKS reduces the size of the ribbon-type AZs in rod photoreceptors (tom Dieck et al, 2012; Hagiwara et al, 2018). We speculate that achieving the full extent of large ribbon-type AZs such as in rod photoreceptors as well as regulating the size of ribbon-type AZs at a specific synapse according to the precise functional demands requires such a functional interplay between proteins of the presynaptic density and RIBEYE. We note that other ribbon-resident proteins such as piccolino likely contribute to this fine-tuning of the size and shape of ribbon-type AZs. Indeed, disruption of piccolino in IHCs reduced the average ribbon size (Müller et al, 2019; Michanski et al, 2023). One exciting example of ribbon synapse heterogeneity manifests itself in IHCs where synapses show a spatial size gradient that likely relates to the diverse molecular and functional properties of the postsynaptic SGNs (reviewed in Moser et al, 2023).

Insights into presynaptic Ca²⁺ channel clustering and function

Current discussion of the clustering of Ca_V Ca²⁺ channels offers two different models: low affinity protein-protein interactions (e.g. Hibino et al, 2002; Kaeser et al, 2011) via specific domains and liquid-liquid phase separation involving intrinsically disordered domains (e.g. Heck et al, 2019). Work on ribbon synapses has considered several layers of organizing Ca_V Ca²⁺ channels at the AZ that seem more compatible with the former model: i) macro-clustering to which the synaptic ribbon contributes as a “super-scaffold” (Jean et al, 2018; Neef et al, 2018; Frank et al, 2010; Maxeiner et al, 2016), ii) tethering and micro-clustering by RBPs, RIMs, CAST and ELKS (Liu et al, 2011; Grabner et al, 2015; Jung et al, 2015; Krinner et al, 2017; Luo et al, 2017; Hagiwara et al, 2018) and iii) nanoscale coupling of Ca_V Ca²⁺ channels and SV release sites (Brandt, 2005; Jarsky et al, 2010; Wong et al, 2014; Maxeiner et al, 2016; Özçete & Moser, 2021; Grabner & Moser, 2021; Jaime Tobón & Moser, 2023) for which a role of the ribbon has been more controversial (e.g. Maxeiner et al, 2016; Grabner & Moser, 2021). MINFLUX nanoscopy of rod photoreceptor AZs recently revealed a well-ordered double-line array topography of Ca_V Ca²⁺ channels, RIM, Bassoon and ubMunc13-2 at the presynaptic membrane on both sides of the ribbon (Grabner et al, 2022). Clearly, we could not fully reconstitute this complex organization with the minimal set of molecular players at syRibbons in HEK293 cells. Yet, we demonstrate augmentation of Ca_V1.3 Ca²⁺ currents and clusters by co-expression of RBP and Bassoon and a positive effect of syRibbons on local Ca²⁺ signaling, recapitulating micro- and macro-clustering of Ca_V1.3 Ca²⁺ channels. However, we note that the functional impact of ribbon-mediated macro-clustering of Ca_V1.3 Ca²⁺ channels remains unclear at the moment. Our Ca²⁺ imaging data seems more reminiscent of previous work on ribbon-type AZs of immature IHCs (Wong et al, 2014) where a broader distribution of Ca_V1.3 and ribbons go along with an immature state of the IHC AZs. One clear limitation of the synthetic AZ system in HEK293 cells is the lack of SVs and the molecular machinery required for SV release. Nonetheless, our work paves the way for future studies including stable co-expression of AZ proteins including of the release site marker Munc13-1 (Sakamoto et al, 2018; Böhme et al, 2016) and potential delivery of SV machinery via co-expression (Park et al, 2021) or from synaptosomes. Alternatively, using neurosecretory cells such as pheochromocytoma cells (PC12) or adrenal chromaffin cells in primary culture, offer the advantage of providing synaptic-like microvesicles and a large set of components of the presynaptic machinery such that expression of exogenous proteins might be limited to RIBEYE. On the other hand, the complex molecular background involving various types of Ca_V channels will likely complicate the interpretation. Regardless, SyRibbons in cultured cells offer great availability and as adherent cells, are well accessible to sophisticated analysis by techniques such MINFLUX nanoscopy and cryo-electron tomography. Future use of polycistronic gene expression or stable cell lines expressing all components could help make the system become even more widely applicable.

Materials and Methods

HEK cell culture and transfections

In this study, we used human embryonic kidney (HEK293) cells stably expressing a tetracycline-inducible human Ca_V1.3 pore forming α₁-D subunit transgene (CACNA1D, NM_000720.2), and showing constitutive expression of Ca_Vβ₃ (CACNB3, NM_000725.2) and Ca_Vα₂δ-1 (CACNA2D1, NM_000722.2). The stably expressing cell line was acquired from Charles River Laboratories, Cleveland, Ohio, USA (Cat. No. CT6232). The cells were cultured as per product protocols in Dulbecco’s Modified Eagle Medium (DMEM) containing GlutaMax, high glucose and pyruvate (Gibco, Life Tech., 31966047), supplemented with 10% Fetal Bovine Serum (FBS; Gibco, Life Tech., A5256701) and 100 units/ml of Penicillin-Streptomycin (Gibco, Life Tech., 15070063). Additionally, the media was supplemented with 0.6 µM of Isradipine (Sigma Aldrich, I6658) and the following selection antibiotics (in mg/ml): 0.005 Blasticidin (Invivogen, ant-bl-05), 0.50 Geneticin (G418 Sulfate; Gibco, Life Tech., 10131027), 0.10 Zeocin (InvivoGen, ant-zn-05) and 0.04 Hygromycin (Thermo Fisher, 10687010). Cells were cultured in a humidified incubator at 37°C with 5% (v/v) CO₂ saturation. Cells were split at ∼70% confluency every 3-4 days to prevent adverse effects on cell growth and channel expression, by dissociating the cell using Accutase (Sigma-Aldrich, A6964). The passage number did not exceed more than 26 passages. For experiments, cells were grown in media lacking selection antibiotics. For induction of α₁-D subunit expression, cells were treated with selection antibiotic-free medium containing 3 µg/ml tetracycline. For transfection, 6.8 µg polyethylenimine (PEI, 25kDa linear, Polysciences, 23966) was added along with a total of 2 µg of DNA to a final volume of 100 µl of DMEM (without FBS). The PEI/DNA mixture was thoroughly mixed and allowed to incubate for 30 min at room temperature before being added to the cells (30 µl/well in a 24-well-plate containing 500 µl media, cell confluency ∼50-70%). For all experiments with co-transfections, we used an equimolar ratio of DNA (total amount of DNA was kept 2 µg in a 100 µl transfection mix). After 18-24 hours, transfection media was replaced with fresh media devoid of selection antibiotics. Cells were used for experiments (immunocytochemistry, patch clamp and Ca²⁺ imaging) 24-48 hours after changing the media or as specified. If cell confluency was too high, cells were reseeded to an appropriate confluency and allowed to settle for at least 12 hours before commencement of experiments.

Expression vectors for RIBEYE, Bassoon, Ca_V1.3 and RBP2

Table EV1:

The RIBEYE-GFP construct used consisted of a human RIBEYE cDNA cloned inframe into a pAAV-GFP vector driven by a hybrid CMV-enhancer-human-beta-actin promoter (CMV-HBA) and containing a downstream Woodchuck Hepatitis Virus (WHV) Posttranscriptional Regulatory Element (WPRE) for mRNA stabilization. All Bassoon constructs encode rat Bassoon. The untagged full-length constructs were cloned in the pCS2+ vector. For the generation of palm-Bassoon constructs, we used a pEGFP-C1 vector backbone with a CMV promoter and kanamycin resistance cassette. The insert comprised of a palmitoylation consensus sequence of GAP43 (ATGCTGTGCTGTATGAGAAGAACCAAACAGGTTGAAAAGAATGATGAGGACCAAAAGATTTCCGGACTCAGA TCTCGAG), followed by a cDNA sequence encoding amino acids 95-3938 of rat Bassoon, and a C-terminal monomeric GFP. For the untagged version, the GFP tag was replaced by a STOP codon. To generate the Halo-Ca_V1.3 plasmid (Schwenzer et al, 2024), human Ca_V1.3 cDNA (accession number NM_001128840.2) was de-novo synthesized and assembled into a HaloTag vector (Promega G7721) using restriction cloning, thus encoding an N-terminal fusion of Ca_V1.3 to HaloTag linked by a ‘GGS’ sequence. The analogous GFP-Ca_V1.3 plasmid was generated by exchanging the HaloTag sequence for mEGFP using restriction cloning. The RBP2 construct with an mKATE2 tag comprised of an insert coding for mouse RBP2 (accession ID NP_001074857.1) with a p2A cleavage site followed by a C-terminal mKATE2 cloned inframe into a f(syn)w-mKATE2-p2A vector driven by a CMV promoter-enhancer and containing downstream WPRE sequence. The untagged version was made by cloning the RBP2 cDNA from this plasmid and replacing RIBEYE-GFP in the pAAV-CMV-HBA-RIBEYE-GFP-WPRE plasmid by in-fusion cloning (Takara Bio USA, Inc.).

Immunocytochemistry and Imaging

HEK293 cells plated on poly-L-lysine coated coverslips were fixed with 99% chilled methanol at −20°C for 2 minutes as previously described (Picher et al, 2017b). The coverslips were washed thoroughly three times with PBS at room temperature (5 – 10 min). Blocking and permeabilization was performed with GSDB (goat serum dilution buffer: 16% normal goat serum, 450 mM NaCl, 0.3% Triton X-100, 20mM phosphate buffer, pH ∼7.4) for 45-60 minutes at room temperature. Samples were incubated with respective primary antibodies (diluted in GSDB, refer to Table EV2) overnight at 4°C or for 2 hours at RT. The samples were then washed three times (5–10 minutes each) with wash buffer (450 mM NaCl, 0.3% Triton X-100, 20mM phosphate buffer, pH ∼ 7.4). Incubation with appropriate secondary antibodies (also diluted in GSDB refer to Table EV2) was performed for 1 hour at room temperature in a light-protected wet chamber. Lastly, the coverslips were washed three times with wash buffer (5-10 minutes each) and one final time with PBS, before mounting onto glass slides with a drop of fluorescence mounting medium (Mowiol 4-88, Carl Roth). For live cell imaging of cells transfected with Halo-Ca_V1.3 (Figure 4B, C), cells plated on glass-bottomed dishes were treated with Janelia Fluor 646 HaloTag Ligand (Promega, Cat. No. GA1120) at a final labelling concentration of 200nM (in cell culture media). Cells were incubated with the ligand for ∼60 min at 37°C with 5% (v/v) CO₂ saturation, after which the media was replaced with an equal volume of fresh warm culture media.

Images from fixed and live samples were acquired in confocal/ STED mode using an Olympus IX83 inverted microscope combined with an Abberior Instruments Expert Line STED microscope (Abberior Instruments GmbH). We used lasers at 488, 561, and 633 nm for excitation and a 775 nm (1.2W) laser for stimulated emission depletion. 1.4 NA 100X or 20X oil immersion objectives were used for fixed samples, and a water immersion 60X objective was used for live samples. Confocal stacks or single sections were acquired using Imspector Software (pixel size = 80 X 80 nm along xy, 200 nm along z). For 2D-STED images, a pixel size of 15 X 15 nm (in xy) was used. Images in figure EV2A were acquired using a Leica SP8 confocal microscope (Leica Microsystems, Germany). All acquired images were visualized using NIH ImageJ software and adjusted for brightness and contrast. Samples and their corresponding controls were processed in parallel using identical staining protocols, laser excitation powers and microscope settings.

Table EV2:

Patch clamp and calcium imaging

Ca_V1.3 stably expressing HEK293 cells were plated on coverslips till about 40-50% confluency was achieved, following which they were induced for Ca_V1.3α1-D expression by incubating them for 24-48 hours in induction media supplemented with tetracycline. For transfected cells, transfection was performed after 24hrs of induction in the induction media as explained above. Cells were recorded in extracellular solution containing (in mM): 150 choline-Cl, 10 HEPES, 1 MgCl₂ and 10 CaCl₂; pH was adjusted to 7.3 with CsOH and osmolality was 300-310 mOsm/kg. For electrophysiological recordings in Figure 5 and Figure S4, pipette solution contained (in mM): 140 NMDG, 10 NaCl, 10 HEPES, 5 EGTA, 3 Mg-ATP and 1 MgCl₂. pH was adjusted to 7.3 with methanesulfonic acid and osmolality was around 290 mOsm/kg. For Ca²⁺ imaging in Figure 7, the pipette solution contained (in mM): 111 Cs-glutamate, 1 MgCl₂, 1 CaCl₂, 10 EGTA, 13 TEA-Cl, 20 HEPES, 4 Mg-ATP, 0.3 Na-GTP, 1 L-glutathione, 0.1 Calbryte590 (AAT Bioquest, Cat. No. 20706); pH was adjusted to 7.3, osmolality to 290 mOsm/kg. Resistance of patch pipettes was 3–7 MΩ. HEK293 cells were rupture patch clamped with EPC-10 amplifier (HEKA electronics, Germany) controlled by Patchmaster software at room temperature, as described previously (Picher et al, 2017b, 2017a). Cells were kept at a holding potential of −91.2 or −88 mV. All voltages were corrected for liquid junction potential offline (21.2 mV or 18 mV). Currents were leak corrected using a p/10 protocol. Recordings were discarded when leak current exceeded −50 pA, R_s exceeded 15 MΩ or offset potential fluctuated more than 5 mV. Current voltage (IV) relations were recorded around 1 minute after rupturing the cell, by applying increasing step-depolarisation pulses of voltage ranging from −86.2 mV to 58.8 mV, in steps of 5 mV. Depolarisation pulses of 500 ms were used for measure Ca_V1.3 inactivation in Figure EV4A.

For Ca²⁺ imaging experiments, we used a Yokogawa CSU-X1A spinning-disk confocal scanner, mounted on an upright microscope (Zeiss Examiner) with a 63X, 1.0 NA objective (W Plan-Apochromat, Zeiss). Images were acquired using an Andor Zyla sCMOS 4.2 camera, controlled by VisiView 5.0 software (Visitron Systems GmbH). GFP, mKATE2 and the low-affinity Ca²⁺ indicator Calbryte590 were excited by a VS Laser Merge System (Visitron Systems) with 488 and 561 nm laser lines. Spinning disk was set to 5000 rpm. mKATE2 positive cells showing peripheral GFP expression (indicative of RIBEYE and palm-Bassoon co-expression) were identified and used for recordings. The cell was loaded for approximately a minute with Calbryte590 dye, after which we applied step depolarizations of 20 ms from −83 to +62 mV in 5-mV step increments to measure Ca²⁺ current (density)–voltage relations. Next, a central plane of the cell was selected and the GFP fluorescence (green channel) of the cell showing SyRibbons was imaged. This was immediately followed by imaging of the increments of Calbryte590 fluorescence (red channel) in the same plane triggered by a depolarizing pulse to +2 mV for 500 ms (frame rate = 20 Hz). For Ca²⁺ imaging experiments, we also included occasionally occurring HEK293 cells which were electrically coupled with other neighbouring HEK293 cells. For such cells, we could not fully compensate for the slow capacitive currents and we did not use them for electrophysiological analysis.

Data analysis

Analysis of confocal images

All images were processed using NIH ImageJ software to make z-projections and multi-channel composites, and figures were created using Adobe Illustrator. For analysis of membrane distribution, a region of interest (ROI) of 1 µm thickness was drawn along the periphery of the cell which was identified either using Na, K ATPase α₁ immunofluorescence or by enhancing the contrast in images of transfected cells so that the cell boundary becomes distinguishable. Mean pixel intensity was measured along this ROI and for the area enclosed within the ROI. The ratio of the mean pixel intensity (periphery: inside) per cell was reported as an average from at least three central planes; we did not use the basal and top plane. For volumetric fits of SyRibbons, IHC ribbons and Ca²⁺ channel clusters, we used the inbuilt surface detection algorithm from Imaris 9.6 (Oxford Instruments). The surface detail was set to 0.16 µm and largest sphere diameters for background subtraction were set as follows: for RIBEYE/CtBP2 0.3 µm, for Bassoon 0.1 µm and for Ca_V1.3 0.08 µm. Thresholding was performed based on the quality of immunofluorescence to ensure all discernible surfaces were detected. All surfaces with less than 10 voxels were filtered out. A manual check was done to include undetected surfaces, remove surfaces not localizing within cell of interest, and split surfaces that would occasionally be clubbed together. For analysis of SyRibbons, only RIBEYE surfaces colocalizing with palm-Bassoon surfaces were used for quantifications (at 0.06 µm or less). Occasionally we would detect surfaces with volumes larger than 2 µm³ (usually cytosolic) or smaller than 0.02 µm³, which were not considered for analysis. For Ca_V1.3 cluster volumetric analysis, clusters classified as +SyRibbons were at 0.045 µm or less from palm-Bassoon positive RIBEYE surfaces. All other clusters were classified as –SyRibbons. Only clusters localized at the periphery of the cell were used for analysis.

Analysis of patch clamp and Ca²⁺ imaging data

Analysis of electrophysiology data and figure preparation was performed using Igor Pro 6 and 7 (WaveMetrics Inc.) and final figures were compiled using Adobe Illustrator. For analyzing current density-voltage relations (IVs), the evoked Ca²⁺ current was averaged from 5 to 10 ms after depolarization start. Cells expressing current amplitudes less than −49 pA and with current density greater than −60 pA/pF were not taken into account for analysis.

Ca²⁺ imaging data analysis was performed using a customized script in ImageJ. Briefly, we performed background subtraction in time series with Calbryte590 signal and then averaged 5 frames before stimulation and subtracted these from an average of three frames during stimulation to obtain ΔF images. For visualization, the images were smoothened (3 X 3 unweighted smoothening) and a composite was created by overlaying the ΔF images and the RIBEYE-GFP channel. For analyzing ΔF/F₀, we drew circular regions of interest (ROIs) of 2µm diameter at sites with and without SyRibbons in the background subtracted time series. The ΔF_max/F₀ was calculated as the of average ΔF/F₀ in first three frames of peak Ca²⁺ signal intensity at stimulation onset in each ROI. To minimize bias while assigning regions as with or without SyRibbons, ROIs were assigned in the RIBEYE-GFP channel, without looking at the corresponding Calbryte590 signal.

Statistics

Data sorting and statistics were performed using MS Excel, Igor Pro 7 and/or GraphPad Prism 10. Numerical data is represented mainly as mean ± standard deviation (SD). The normality of data was assessed using Jarque-Bera test and Kolmogorov-Smirnov test, and the equality of variances was checked using F-test. When comparing two samples, a two-tailed unpaired Student’s t-test (or paired t-test for Figure 7J) was performed for normally distributed samples with equal variances. If the conditions for normality and equality of variances were not met, an unpaired Mann-Whitney-Wilcoxon test was performed. When comparing multiple samples, a one-way ANOVA with post hoc Tukey’s test was used for normally distributed data with equal variances. For samples which were not normally distributed, we used Kruskal-Wallis test with post hoc Dunn’s multi-comparison correction. All box plots are depicted with crosses representing mean values, central bands indicating median, whiskers for 90/10 percentiles, boxes for 75/25 percentiles and individual data points overlaid. Non-significant differences have been indicated as n.s., while significant P values have been depicted as *P for < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Acknowledgements

We would like to thank Dr. Kathrin Kusch for advice on cloning, Dr. Jakob Neef for critical feedback on the manuscript, and Nare Karagulyan for valuable discussions regarding Ca²⁺ imaging. We are grateful to Sandra Gerke, Christiane Senger-Freitag, Sina Langer and Ina Preuss for expert technical assistance and Patricia Räke-Kügler for the administrative support during this study. We would also like to thank Prof. Erwin Neher and Prof. Silvio Rizzoli for their feedback throughout the course of this study. Lastly, we would like to thank lab rotation students Gantavya Arora and Ugur Coskun for their initial contributions to the project.

Funding

RK was supported by funding from the Studienstiftung des Deutschen Volkes. This work was further supported by funding of the European Union (ERC, “DynaHear”, grant agreement No. 101054467), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) via the EXC 2067/1 (MBExC) to SEL and TM, and Fondation Pour l’Audition (FPA RD-2020-10). Open access funding provided by Max Planck Society.

Author Contributions

TM and RK designed the study. RK performed the experiments and the analysis under supervision of TM. TD, NS and SEL prepared and characterized expression plasmids for membrane targeted Bassoon (TD) and EGFP/Halo-tagged Ca_V1.3α1 (NS, SEL). RK and TM prepared the manuscript with contribution of all authors.

Disclosure and Competing Interests Statement

The authors declare that no competing interests exist.