Establishing synthetic ribbon-type active zones in a heterologous expression system

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 toward 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 (Figure 1Aii), in contrast to a diffuse cytosolic distribution when merely expressing EGFP (Figure 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, appear unlikely to reflect RIBEYE entrapped in protein trafficking pathways or to represent degradation products of overexpressed RIBEYE (Figure 1—figure supplement 1A, B, and C).

Membrane targeting of Bassoon using a palmitoylation consensus sequence.

(A) Representative confocal images of human embryonic kidney 293 (HEK293) cells transfected with (i) GFP only (CtBP2/RIBEYE in magenta labeling nuclei, GFP in cyan), (ii) RIBEYE-GFP (CtBP2/RIBEYE in magenta, GFP in cyan), (iii) Bassoon (green), and (iv) RIBEYE-GFP and Bassoon. Note that antibodies against RIBEYE B-domain and the nuclear transcription factor CtBP2 result in similar staining patterns as CtBP2 is transcribed from the same gene as RIBEYE and is identical to the RIBEYE-B domain except for the first 20 N-terminal amino acids. The upper panel shows maximum projections, and the lower panel shows exemplary single sections from confocal stacks. Scale bar = 5 µm. (B) Schematic of construct for membrane targeting of Bassoon. The first 95 amino acids from full-length Bassoon were replaced with a palmitoylation consensus sequence from GAP43. Constructs without and with a C-terminal GFP tag were used, as depicted. (C) Sample confocal image (single section) showing membrane-targeted palm-Bassoon (green) expressed in HEK293 cells appears as puncta distributed along the periphery of the cell, marked by Na, K-ATPase α1 (blue). The inset shows a maximal projection of the confocal section. Scale bar = 5 µm. (D) Zoom-in from (C) shows colocalization of palm-Bassoon with membrane marker Na, K-ATPase α1. Scale bar = 1 µm. Schematic of construct for membrane targeting of Bassoon. (E) Quantification of Bassoon signal intensity at periphery vs inside of cell. Cells expressing palm Bassoon (N = 10 cells) clearly show a higher peripheral distribution compared to cells expressing full-length Bassoon (N = 10 cells), ****p<0.0001, Mann-Whitney-Wilcoxon test. Overlaid data points represent individual cells, crosses represent the mean values, central band indicates the median, whiskers represent 90/10 percentiles, and boxes represent 75/25 percentiles.

Next, for membrane targeting of these cytosolic RIBEYE clusters, we co-expressed the multidomain, cytomatrix of the AZ protein Bassoon (Figure 1Aiii). Prior work on the molecular underpinnings of ribbon synapses had identified 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 (Figure 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 (Figure 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 plasma membrane-standing Na, K-ATPase α1 (data representative of 3 transfections, Figure 1C and D). Palm-Bassoon does not appear to localize in the ER, Golgi apparatus, or lysosomes (Figure 1—figure supplement 1D, E, and F). Comparing the ratio of Bassoon signal intensity at the periphery vs 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, Figure 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; Figure 2Ai). Closer inspection of these immunofluorescent puncta with stimulated emission depletion (STED) nanoscopy (Figure 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. 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; Michanski et al., 2023; Figure 2Bi and ii; data from Michanski et al., 2023).

Co-expression of RIBEYE with palm-Bassoon results in ribbon-type active zone (AZ)-like structures.

(A) (i) Representative confocal image (single section) of a human embryonic kidney 293 (HEK293) cell transfected with RIBEYE-GFP (magenta) and palm-Bassoon (green). Co-expression of RIBEYE with palm-Bassoon targets RIBEYE to the cell membrane. Inset shows maximum projection. Scale bar = 5 µm. (ii) Exemplary 2D stimulated emission depletion (STED) images for RIBEYE – palm-Bassoon juxtapositions acquired from cells as shown in (i). Scale bar = 500 nm; individual channels have been depicted with an intensity-coded look-up table with warmer colors indicating higher intensity. (i) Representative maximum projection of confocal sections of apical organ of Corti from a Wistar rat (postnatal day 18); data as published in Michanski et al., 2023, stainings were for CtBP2/RIBEYE (magenta; labeling synaptic ribbons and inner hair cell [IHC] nuclei) and Bassoon (green; spots juxtaposing with ribbons represent IHC AZs, spots not juxtaposing with ribbons represent efferent synapses formed by lateral olivocochlear neurons onto SGN boutons). Scale bar = 5 µm. (B) (ii) Juxtaposing RIBEYE and Bassoon spots imaged in 2D STED and confocal mode, respectively. Scale bar = 500 nm. Note the striking resemblance to reconstituted RIBEYE+palm-Bassoon structures in HEK293 cells as shown in (Aii). STED images are from 3 sample transfections, representative of 7 total transfections; individual channels have been depicted with an intensity-coded look-up table with warmer colors indicating higher intensity. (C) Quantification of RIBEYE and Bassoon signal intensity at periphery vs inside of cell shows a higher peripheral distribution of RIBEYE and Bassoon in HEK293+RIBEYE+palm-Bassoon cells (N=9 cells) as compared to HEK293+RIBEYE+Bassoon cells (N=9 cells); ****p<0.0001, Mann-Whitney-Wilcoxon test. Overlaid data points represent individual cells, crosses represent mean values, central band indicates the median, whiskers represent 90/10 percentiles, and boxes represent 75/25 percentiles. Distribution of volumes of RIBEYE and palm-Bassoon puncta from HEK293 cells expressing RIBEYE and palm-Bassoon (n = 20 cells, quantifications from 5 sample transfections). Volumes of synaptic ribbons from rat IHCs (n = 29 cells, 3 rats) have been plotted for comparison. (E) Box plot depicting data from (D). Volumes of RIBEYE and palm-Bassoon puncta are comparable to each other (p>0.99, Kruskal-Wallis test with post hoc Dunn’s multiple comparison), but on average, they are much smaller and considerably more variable when compared to volumes of RIBEYE and Bassoon puncta from rat IHCs, respectively (****p<0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple comparison). Overlaid plus signs represent individual spots, crosses represent mean values, central band indicates the median, whiskers represent 90/10 percentiles, and boxes represent 75/25 percentiles. Quantification of RIBEYE and Bassoon signal intensity at periphery vs inside of cell shows a higher peripheral distribution of RIBEYE and Bassoon in HEK293+RIBEYE+palm-Bassoon cells (N=9 cells) as compared to HEK293+RIBEYE+Bassoon cells (N=9 cells); ****p<0.0001, Mann-Whitney-Wilcoxon test. Overlaid data points represent individual cells, crosses represent mean values, central band indicates the median, whiskers represent 90/10 percentiles, and boxes represent 75/25 percentiles. (F) RIBEYE puncta in HEK cells expressing RIBEYE and palm-Bassoon (nspots = 864, N = 20 cells) appear more spherical than Bassoon puncta in the same cells (nspots = 961, N = 20 cells; ****p<0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple-comparison test). Note the similar trend in IHC synaptic ribbons where RIBEYE puncta are more spherical than Bassoon puncta (nspots = 290 for RIBEYE and nspots = 314 for Bassoon, N = 29 cells, 3 rats; ****p<0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple-comparison test). Overlaid plus signs represent individual spots, crosses represent mean values, central band indicates the median, whiskers represent 90/10 percentiles, and boxes represent 75/25 percentiles.

We compared the RIBEYE and Bassoon signal intensities at the periphery vs 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 (Figure 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 Figure 2—figure supplement 1). 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 2—figure supplement 1B) 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 3D 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 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; nspots = 864 and volume = 0.17 ± 0.18 µm3 with CV = 1.10 for Bassoon; nspots = 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; nspots = 290, and volume = 0.27 ± 0.18 µm3, CV = 0.67 for Bassoon; nspots = 314, data from N=29 cells) (Figure 2D and E). Moreover, volumes of RIBEYE surfaces show a high positive correlation to volumes of corresponding palm-Bassoon surfaces (Pr = 0.778, ****p<0.0001), implying that larger palm-Bassoon structures may recruit bigger RIBEYE structures to the plasma membrane. We also note that the 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, nspots = 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, nspots = 2217, N=13 cells, ****p<0.001, Mann-Whitney-Wilcoxon test). RIBEYE clusters constituting SyRibbons appeared more spherical than the plate-like Bassoon structures in the same cells (sphericity Ψ=0.91 ± 0.05, nspots = 864 for RIBEYE vs Ψ=0.86 ± 0.07, nspots = 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, nspots = 290) compared to Bassoon spots (Ψ=0.76 ± 0.10, nspots = 314); ****p<0.0001, Kruskal-Wallis test with post hoc Dunn’s multiple-comparison test, data from N=29 cells (Figure 2F). Nonetheless, the fact that next to structures with volumes comparable to IHC ribbons, we also encountered smaller and larger structures likely reflect poorer regulation of RIBEYE and palm-Bassoon expression in the heterologous system.

Next, we analyzed SyRibbons in situ using cryo-electron tomography (cryo-ET), which capitalizes on cell vitrification to obtain near-native preservation. Plunge-frozen HEK cells transfected with RIBEYE-GFP and untagged palm-Bassoon were vitrified and subsequently milled using a cryo-focused ion beam (cryo-FIB; Figure 3—figure supplement 1A) to produce 150-nm-thick lamellae as previously described (Rigort et al., 2010; Pierson et al., 2024, see Materials and methods). Using a light microscope integrated into the cryo-FIB chamber, we targeted cell areas showing peripheral GFP fluorescence corresponding to SyRibbons (Figure 3—figure supplement 1B). Lamellae were subsequently transferred to a cryo-transmission electron microscope, and tomographic tilt series were acquired at fluorescence locations. The tomograms showed that GFP-positive RIBEYE spots corresponded to electron-dense structures (Figure 3A and B), as is characteristic of the synaptic ribbon (De Robertis and Franchi, 1956; Smith and Sjöstrand, 1961). These electron-dense SyRibbons appeared to be ~300–800 nm in size and were ovoid or ellipsoidal in shape (Figure 3B and C, Figure 3—figure supplement 1C). Some SyRibbons displayed a hollow core (Figure 3—figure supplement 1Ci and Ciii) and on closer inspection, some appeared to have a multi-lamellar ultrastructure (Figure 3—figure supplement 1Cii), both of which have been reported previously for natively expressed synaptic ribbons in the inner ear and the retina (Michanski et al., 2023; Michanski et al., 2019; Sobkowicz et al., 1982; Liberman, 1980; Stamataki et al., 2006; Wichmann and Moser, 2015). We captured a SyRibbon positioned within 100 nm from the plasma membrane (Figure 3C). Although bona fide ribbons display a halo of SVs tethered on their surface, we did not observe any obvious accumulation of vesicular structures around these SyRibbons, which is not unexpected for synapse-naïve HEK cells. This is in contrast to a previous report in R28 retinal progenitor cells, where heterologously expressed RIBEYE was shown to recruit vesicles (Magupalli et al., 2008). Intriguingly, in some tomograms, we observed membrane-bound SyRibbons (Figure 3—figure supplement 1Ci–iv), which may indicate an autophagic engulfment of some of these overexpressed structures.

Cryo-correlative microscopy captures membrane-localized SyRibbons.

(A) RIBEYE-GFP signal on a 150-nm-thick lamella, revealed by fluorescent light microscopy within the cryo-focused ion beam (cryo-FIB) chamber (cryo-LM). Dotted lines delineate cell membranes. The orange arrow indicates a plasma membrane-proximal GFP fluorescence, whereas purple arrowheads point to cytosolic RIBEYE aggregates. The tomogram shown in (C) was acquired at the boxed region. The image underwent background subtraction for better visualization. (B) Transmission electron microscopy (TEM) image of the lamella in (A). Orange arrow and purple arrowheads locate the respective regions in (A), showing that GFP-positive spots correlated to electron-dense bodies. Dotted lines delineate plasma membranes. The tomogram shown in (C) was taken at the boxed region. (C) Tomogram slice showing a SyRibbon acquired at the boxed region in (B). PM: plasma membrane, ECS: extracellular space, ER: endoplasmic reticulum.