Presynaptic α2δ subunits are key organizers of glutamatergic synapses

Abstract

In nerve cells the genes encoding for α2δ subunits of voltage-gated calcium channels have been linked to synaptic functions and neurological disease. Here we show that α2δ subunits are essential for the formation and organization of glutamatergic synapses. Using a cellular α2δ subunit triple-knockout/knockdown model, we demonstrate a failure in presynaptic differentiation evidenced by defective presynaptic calcium channel clustering and calcium influx, smaller presynaptic active zones, and a strongly reduced accumulation of presynaptic vesicle-associated proteins (synapsin and vGLUT). The presynaptic defect is associated with the downscaling of postsynaptic AMPA receptors and the postsynaptic density. The role of α2δ isoforms as synaptic organizers is highly redundant, as each individual α2δ isoform can rescue presynaptic calcium channel trafficking and expression of synaptic proteins. Moreover, α2δ-2 and α2δ-3 with mutated metal ion-dependent adhesion sites can fully rescue presynaptic synapsin expression but only partially calcium channel trafficking, suggesting that the regulatory role of α2δ subunits is independent from its role as a calcium channel subunit. Our findings influence the current view on excitatory synapse formation. First, our study suggests that postsynaptic differentiation is secondary to presynaptic differentiation. Second, the dependence of presynaptic differentiation on α2δ implicates α2δ subunits as potential nucleation points for the organization of synapses. Finally, our results suggest that α2δ subunits act as transsynaptic organizers of glutamatergic synapses, thereby aligning the synaptic active zone with the postsynaptic density.

Keywords: cultured hippocampal neurons; synapse formation; synaptic calcium channels; transsynaptic.

Conflict of interest statement

The authors declare no competing interest.

Copyright © 2021 the Author(s). Published by PNAS.

Figures

Fig. 1.

α2δ subunits are essential for survival, activity-induced synaptic recycling, and normal calcium current densities. (A) The Kaplan–Meier survival curves show an increased mortality in the distinct α2δ double-knockout mouse models (n = 9 to 21). (B) Mean life span was significantly reduced in α2δ-1/-2 and α2δ-2/-3 double-knockout mice when compared with α2δ-1/-3 or ducky mice [ANOVA, F(3,47) = 4.7, P = 0.006, with Holm–Sidak post hoc test, *P < 0.05, **P < 0.01]. (C) Putative synaptic varicosities from α2δ TKO/KD neurons failed to load FM4-64 dye upon 60 mM KCl depolarization (outline/triple KO). In contrast, control boutons transfected with eGFP only and nontransfected double-knockout boutons (asterisks) showed robust uptake of the FM-dye. [ANOVA on ranks, H(2)= 96.6, P < 0.001, with Dunn’s post hoc test, ***P < 0.001, 26 to 110 synapses from two to four culture preparations]. (Scale bar, 1 µm.) (D) Current properties of α2δ subunit single, double, and TKO/KD cultured hippocampal neurons. Representative Ba2+ whole-cell currents at Imax (Upper) and I/V-curves (Lower) recorded from hippocampal neurons. (Left) I/V curves reveal a strong reduction of calcium currents in TKO/KD neurons (triple KO), when compared with untransfected wild-type neurons or TKO/KD neurons transfected with α2δ-2 (rescue α2δ-2). (Right) Current densities in α2δ-2/-3 double but not in α2δ-3 single knockout were also reduced. For I/V curve properties see SI Appendix, Table S1. (E) Current densities at Imax for individual cells [ANOVA, F(4,71) = 11.3, P < 0.001, with Holm–Sidak post hoc test, **P < 0.01, ***P < 0.001, 8 to 26 cells from five culture preparations]. Horizontal lines represent means and error bars SEM.

Fig. 2.

α2δ subunits are essential for activity-induced presynaptic calcium transients. (A) Putative synaptic varicosities from neurons cotransfected with SynGCaMP6f and mCherry were selected in Fiji/ImageJ using the ROI tool (yellow circles). For quantification of the presynaptic calcium transients, regions were transferred to the corresponding recordings of SynGCaMP6f fluorescence, shown here as fluorescence change by subtracting an averaged control image from an image averaged around the maximal response. (Scale bar, 10 µm.) (B) SynGCaMP6f fluorescence (ΔF/F0) for stimulations with 1 AP, 3 APs, and 10 APs at 50 Hz. Lines show the mean fluorescence traces for control (50 cells from three independent cultures), TKO/KD (triple KO, 50 cells from three independent cultures), and the rescue condition expressing α2δ-1 together with SynGCaMP6f and mCherry (19 cells from two independent cultures). (C) Cumulative frequency distribution histograms of peak fluorescent responses (ΔF/F0) from all recorded putative synaptic varicosities of α2δ TKO/KD (light green), double-heterozygous control (dark green), and α2δ-1 overexpressing TKO/KD neurons (rescue α2δ-1, purple) in response to stimulations with 1, 3, or 10 APs (number of synapses: control, 1,100; triple KO, 1,100; rescue α2δ-1, 418). (D) Quantification of peak fluorescent amplitudes in response to stimulations with 1, 3, or 10 APs. Each dot represents the mean of 22 synapses from one neuron [Kruskal–Wallis ANOVA with Dunn’s multiple comparison test: 1 AP: H(3, 119) = 81, P < 0.0001; 3 AP: H(3, 119) = 48, P < 0.0001; 10 AP: H(3, 119) = 26, P < 0.0001; post hoc test: ***P ≤ 0.001; 50 (control, triple KO) and 19 (rescue α2δ-1) cells from three and two independent culture preparations].

Fig. 3.

Failure of presynaptic calcium channel clustering and synapsin accumulation in α2δ subunit triple-knockout/knockdown neurons. (A and B) Immunofluoresence analysis of axonal varicosities from wild-type neurons (control, neurons transfected with eGFP only), TKO/KD neurons (triple KO, α2δ-2/-3 double-knockout neurons transfected with shRNA-α2δ-1 plus eGFP), and TKO/KD neurons expressing α2δ-2 (rescue, α2δ-2/-3 double-knockout neurons transfected with shRNA-α2δ-1 plus eGFP and α2δ-2). Putative presynaptic boutons were identified as eGFP-filled axonal varicosities along dendrites of untransfected neurons (SI Appendix, Fig. S4) and outlined with a dashed line. Immunolabeling revealed a failure in the clustering of presynaptic P/Q- (A, CaV2.1) and N-type (B, CaV2.2) channels as well as in the accumulation of presynaptic synapsin in varicosities from α2δ TKO/KD neurons (Middle). In contrast, wild-type control neurons (Left) displayed a clear colocalization of the calcium channel clusters with synapsin in the eGFP-filled boutons. The linescan patterns recorded along the indicated line support these observations. Note that the sole expression of α2δ-2 (Right) or the sole presence of α2δ-1 in synapses from neighboring α2δ-2/-3 double-knockout neurons (asterisks in A and B) suffices to fully rescue presynaptic calcium channel clustering and synapsin accumulation. (C–E) Quantification of the fluorescence intensities of presynaptic CaV2.1 (C), CaV2.2 (D), and synapsin (E) clustering in control, TKO/KD, and α2δ-2–expressing (rescue) TKO/KD neurons [ANOVA with Holm–Sidak post hoc test, ***P < 0.001; CaV2.1: F(2, 58) = 10.8, P < 0.001, 16 to 25 cells from four to six culture preparations; CaV2.2: F(2, 37) = 13.7, P < 0.001, 11 to 16, two to four; synapsin: F(2, 99) = 15.5, P < 0.001, 30 to 36, five to eight; horizontal lines represent means and error bars SEM]. (Scale bars, 1 µm.)

Fig. 4.

Presynaptic α2δ subunits mediate glutamatergic synapse formation and transsynaptic differentiation. (A and B) Immunofluoresence micrographs of axonal varicosities from presynaptic α2δ TKO/KD neurons (eGFP-positive axonal varicosities, Left, arrows) as well as dendrites from postsynaptic TKO/KD neurons (eGFP-positive dendrites, Right, arrows). Axonal varicosities and dendrites are outlined by a dashed line and arrowheads mark exemplary innervating triple-knockout axons. The sketches summarize the observed labeling patterns. (A) α2δ TKO/KD neurons display a failure of presynaptic Cav2.1 channel and synapsin clustering exclusively in presynaptic axonal varicosities (arrows and sketch, Left). In contrast, postsynaptic TKO/KD neurons developed dendritic spines opposite presynaptic boutons containing Cav2.1 and synapsin clusters (arrows and sketch, Right) formed by axons from α2δ-2/-3 double-knockout neurons still containing α2δ-1. (B) Presynaptic α2δ TKO/KD induces a failure of the postsynaptic PSD-95 clustering indicating a transsynaptic action of α2δ subunits (arrows and sketch, Left). Conversely, postsynaptic TKO/KD neurons still receive proper synaptic input from neighboring α2δ-1 containing neurons as indicated by presynaptic synapsin and postsynaptic PSD-95 colocalized on TKO/KD dendritic spines (arrows and sketch, Right). (Scale bars, 2 µm and 8 µm.) (C and D) mEPSC recordings and analysis from control (double heterozygous), double-knockout (double KO, α2δ-2/-3 double-knockout), and postsynaptic TKO/KD neurons (triple KO) receiving synaptic input from neighboring double-knockout neurons (see A, Right). (C) Quantification of mEPSC amplitues (Left) and frequencies (Right). Amplitudes and frequencies of mEPSCs for each condition were normalized to the mean value of control condition for each individual experiment. [Amplitude: one-way ANOVA, F(2,75) = 1.56, P = 0.22; frequency: ANOVA, F(2,75) = 2.48, P = 0.09; n = 37 (control]), 26 (double KO), and 15 (postsynaptic triple KO) from four, four, and two culture preparations, respectively]. (D) Representative traces of mEPSC for condition described in C. (E and F) Failure of postsynaptic PSD-95 labeling opposite α2δ TKO/KD boutons. Similar to the presynaptic proteins (see Fig. 3) the sole expression of α2δ-2 (rescue, right column) or the sole presence of α2δ-1 in synapses from neighboring α2δ-2/-3 double-knockout neurons (asterisks in E, middle column, linescans) fully rescued postsynaptic PSD-95 clustering [ANOVA, F(2, 49) = 11.7, P < 0.001, with Holm–Sidak post hoc test, **P < 0.01, ***P < 0.001; 14 to 20 cells from three to four culture preparations). (G and I) The defect in synaptogenesis caused by loss of α2δ subunits specifically affects glutamatergic synapses, indicated by reduced fluorescent intensity of vGLUT1/AMPAR labeling (outline/linescan; t test, t(15) = 3.1, **P < 0.01; 7 and 10 cells from two and three culture preparations). (H and J) In contrast, vGAT/GABAAR labeling in GABAergic synapses did not seem to be reduced in α2δ TKO/KD neurons (outline/linescan; two and five cells from one and two culture preparations). Error bars indicate SEM. (Scale bars, 1 µm.)

Fig. 5.

Presynaptic α2δ subunit triple-knockout/knockdown does not affect pre- and postsynaptic differentiation in GABAergic synapses. (A and E) Representative immunofluoresence micrographs of axonal varicosities from presynaptic α2δ-3 knockout (control) or TKO/KD (triple KO) cultured GABAergic MSNs. Transfected neurons (22 to 24 DIV) were immunolabeled for vGAT and the GABAAR (A) and CaV2.1 and synapsin (E). Colocalization of fluorescence signals within eGFP-filled axonal varicosities (axons are outlined with dashed lines) was analyzed using line scans. (B and F) Sketches depicting the expected staining patterns in A and E, respectively. (C, D, G, and H) Quantification of the respective fluorescence intensities in control and TKO/KD neurons (t test, GABAAR: t(38) = 0.8, P = 0.41, 13 to 27 cells from three culture preparations; vGAT: t(38) = 1.7, P = 0.10, 13 to 27 cells from three culture preparations; CaV2.1: t(32) = 1.6, P = 0.13, 16 to 18 cells from two culture preparations; synapsin: t(32) = 0.7, P = 0.51, 16 to 18 cells from two culture preparations). Values for individual cells (dots) and means (lines) ± SEM are shown. Values were normalized to control (α2δ-3 knockout) within each culture preparation. (Scale bars, 1 µm.)

Fig. 6.

Ultrastructural analysis of pre- and postsynaptic specializations in excitatory α2δ subunit triple-knockout/knockdown synapses. (A) Exemplary EM micrographs of synaptic structures show similar presynaptic and postsynaptic differentiation in wild-type control (Left) and α2δ-2/-3 double-knockout (α2δ-2/-3 KO) cultured hippocampal neurons (for statistics see text). (B) Exemplary EM micrographs from silver-amplified eGFP-immunogold-stained presynaptic boutons and the corresponding postsynaptic region from double (α2δ-2/-3) and TKO/KD (triple KO) synapses. (C) Quantitative analyses showing that both the length of the AZ (Left) and the PSD (Middle) were significantly reduced in TKO/KD compared with eGFP-transfected α2δ-2/-3 double-knockout synapses in separate culture preparations (control eGFP) and nontransfected neighboring synapses within the same coverglass (control nt). In addition to the AZ and PSD length also the thickness, particularly the extension of the PSD from the membrane into the cytosol, was strongly reduced in TKO/KD compared with the respective control synapses [ANOVA with Tukey post hoc test, **P < 0.01, ***P < 0.001; AZ length: F(2,147) = 11.3, P < 0.001; PSD length: F(2,147) = 7.5, P < 0.001; PSD extension: F(2,147) = 44.6, P < 0.001. Horizontal lines represent means and error bars SEM]. Abbreviations in EM micrographs: b, presynaptic bouton; s, dendritic spine; ps, postsynaptic compartment. (Scale bars, 200 nm.)

Fig. 7.

Rescuing triple-knockout/knockdown synapses with α2δ-2-ΔMIDAS or α2δ-3-ΔMIDAS dissociates synapse differentiation from presynaptic calcium channel trafficking. (A and D) Immunofluorescence micrographs of axonal varicosities from presynaptic α2δ TKO/KD neurons (triple KO, eGFP-positive axonal varicosities, Left) and neurons expressing α2δ-2-ΔMIDAS or α2δ-2 (A) and α2δ-3-ΔMIDAS or α2δ-3 (D). Axonal varicosities are outlined by a dashed line. Immunolabeling for CaV2.1 and synapsin (syn) revealed that, unlike α2δ-2 or α2δ-3, expression α2δ-2-ΔMIDAS or α2δ-3-ΔMIDAS in TKO/KD neurons fully rescued presynaptic synapsin but not CaV2.1 clustering. The relative fluorescence of each signal was recorded along the indicated line to support these observations. (B, C, E, and F) Quantification of the relative synaptic area covered by the respective immunofluorescence of presynaptic CaV2.1 (B and E) and synapsin (C and F) [ANOVA, α2δ-2: CaV2.1, F(2, 78) = 18.9, P < 0.001, n = 41 (triple KO), 23 (MIDAS), and 17 (rescue) from 3 to 10 culture preparations; synapsin, F(2, 56) = 18.7, P < 0.001, n = 19 (triple KO), 23 (MIDAS), and 17 (rescue) from three to four culture preparations; α2δ-3: CaV2.1, F(2, 23) = 4.7, P < 0.019, n = 6 (triple KO), 6 (MIDAS), and 14 (rescue) from two to three culture preparations; synapsin: F(2, 41) = 17.3, P < 0.001, n = 12 (triple KO), 11 (MIDAS), and 21 (rescue) from three to four culture preparations; Tukey post hoc test, *P = 0.016, ***P < 0.001; horizontal lines represent means and error bars SEM]. (Scale bars, 1 µm.)

Fig. 8.

Model summarizing the putative roles of presynaptic α2δ subunits in glutamatergic synapse formation and differentiation. Our findings identified α2δ subunits as key organizers of glutamatergic synapses and propose their involvement in at least three critical steps during synapse maturation. By interacting with the α1 subunit they mediate the incorporation of VGCCs into the presynaptic AZ (1). α2δ subunits are involved in presynaptic differentiation and may, directly and/or indirectly via the entire VGCC complex, mediate the accumulation of synaptic vesicles (SV) to the synaptic terminal (2). Finally, α2δ subunits align the presynaptic AZ with the postsynaptic membrane and postsynaptic AMPARs. This may be mediated by a direct interaction with AMPARs (3a) or by interacting with classical synaptic cell adhesion molecules (SCAMs, 3b), as for example neurexins.