Nanoscale distribution of presynaptic Ca(2+) channels and its impact on vesicular release during development

Abstract

Synaptic efficacy and precision are influenced by the coupling of voltage-gated Ca(2+) channels (VGCCs) to vesicles. But because the topography of VGCCs and their proximity to vesicles is unknown, a quantitative understanding of the determinants of vesicular release at nanometer scale is lacking. To investigate this, we combined freeze-fracture replica immunogold labeling of Cav2.1 channels, local [Ca(2+)] imaging, and patch pipette perfusion of EGTA at the calyx of Held. Between postnatal day 7 and 21, VGCCs formed variable sized clusters and vesicular release became less sensitive to EGTA, whereas fixed Ca(2+) buffer properties remained constant. Experimentally constrained reaction-diffusion simulations suggest that Ca(2+) sensors for vesicular release are located at the perimeter of VGCC clusters (<30 nm) and predict that VGCC number per cluster determines vesicular release probability without altering release time course. This "perimeter release model" provides a unifying framework accounting for developmental changes in both synaptic efficacy and time course.

Figure 1

Distribution of Ca_v2.1 Immunoparticles at the Developing Calyx of Held Presynaptic Terminal Revealed with SDS-FRL (A1) Freeze-fracture replica image of a calyx of Held of a P21 rat. Presynaptic P-face classification is confirmed by the presence of convex structures in the cross-fractured face that are likely to reflect synaptic vesicles. Round, electron-dense particles (5 nm diameter) indicate immunogold-labeled Ca_v2.1 antibodies, which are often seen in characteristic concaved surface with dimples (Figures S2A and S2B). The higher magnification of this image is shown in Figure S2A. (A2) Zoom of box region in (A1). Arrows indicate 5 nm immunogold particles. (A3) Clusters of two or more particles were identified by the overlap of 100 nm radius circles (green) centered on the particles (blue). (B) Typical immunogold particle cluster at P7. (C) Typical immunogold particle cluster at P14. (D) Low magnification of immunogold particle clusters (green) in the presynaptic P-face (pink) of calyces of Held at different ages. (E) Cumulative histograms of NND between cluster centers comparing different ages in La1 (upper panel, p = 0.03 for P7 versus, P14, p = 0.27 for P7 versus, P21, p = 0.008 for P14 versus, P21) and La2 (lower panel, p = 0.26 for P7 versus, P14). Statistical comparisons were performed using a Kolmogorov-Smirnov test. La1 and La2 indicate samples reacted with different antibody batches resulting in a higher efficiency labeling for La2 (62% versus 19% for La1). All images are taken from La1 samples.

Figure 2

Ca_v2.1 Immunogold Particle Distribution within Clusters as a Function of Age (A) Overlaid gold particle clusters (La1) aligned at their center of gravity (92 clusters per age). (B) Spatial distribution profiles of immunogold particles showing particle distances from the center of gravity of its corresponding cluster. Ordinate indicates number of particles in concentric bins (8 nm). (C) Mean cluster area (± SEM) for La1. Only clusters having three or more particles were included in this analysis (^∗p < 0.01, one way ANOVA followed by Tukey's post hoc). (D) Cumulative histograms of immunogold particle number per cluster from La1 (upper) and La2 (lower). (E) Mean number (± SEM) of immunogold particles within a cluster for La1 (filled bars, ^∗p < 0.01, one-way ANOVA followed by Tukey's post hoc) and La2 (open bars, ^∗p < 0.01, t test). (F) Cumulative histograms of intracluster particle NND values for La1.

Figure 3

Developmental Reduction of AP-Induced Ca²⁺ Transients Measured with Confocal Spot Detection (A) Red channel confocal image of a P7 calyx of Held loaded with Alexa 594 and Oregon green BAPTA-5N. Single AP-induced Ca²⁺ transients are presented as ΔF/F and were recorded at various locations within the terminal (red circles) using confocal spot detection. Yellow and green traces show single Ca²⁺ transients recorded from confocal spots on the synaptic face and nonsynaptic regions, respectively. (B) Temporal relationship between the presynaptic AP (black trace) and Ca²⁺ transients recorded from spot locations along the synaptic face (yellow trace, average of seven locations in [A]) or at nonsynaptic regions (green trace, average of three locations in [A]) in a P7 calyx. Dashed line is a fit with Equation 2 (Supplemental Experimental Procedures). (C) Same as (A) but for a P14 calyx. Traces are averages of five trials. (D) Cumulative amplitude histograms of Ca²⁺ transient amplitudes at P7 and P14 (^∗∗p < 0.01, Kolmogorov-Smirnov test). Inset, the population average at P7 (green, 21 spots from 8 calyces) and P14 (21 spots from 6 calyces). (E) Left: Averaged AP (V_m) and synaptic-face Ca²⁺ transient in the absence (black; n = 29 traces) or presence (green; n = 21 traces) of 1 to 2 mM TEA (shaded areas denote 2× SEM). Right: Mean peak amplitude (± SEM) of Ca²⁺ transients in control and TEA (P14, n = 7 calyces, ^∗p < 0.05, paired t test). (F) Left: Whole-terminal I_Ca and averaged Ca²⁺ transient (n = 16 traces) evoked in voltage clamp using an AP-waveform recorded previously from P7 (green, AP₇, inset) and P14 (black, AP₁₄, inset) calyces. EGTA concentration was 2 mM. Right: Mean peak amplitudes (± SEM) of Ca²⁺ transients for AP₇ and AP₁₄ at P7 calyces (^∗∗p < 0.01, paired t test, n = 9 calyces).

Figure 4

Quantification of Endogenous Fixed Ca²⁺ Buffer Properties at P7 and P14 (A) Averaged Ca²⁺ transient at the synaptic face of P7 calyces in the presence of 0.1 mM (gray, n = 25 calyces) or 2 mM (black, n = 33) EGTA in presynaptic pipettes. (B) Same as (A) but for P14 calyces (n = 23 and 33 respectively). Inset shows normalized Ca²⁺ transients for 2 mM EGTA at P7 (green) and P14 (black) for kinetic comparison. (C) Average weighted mean time constant (± SEM) of the Ca²⁺ transients at P7 and P14. (D) A normalized simulated Ca²⁺ transient computed with a fast EFB k_off (1.0 × 10⁴ s⁻¹, K_d = 100 μM, red trace) matched the experimental Ca²⁺ transient (gray) better than when computed with a slow EFB k_off (1.0 × 10³ s⁻¹, K_d = 2 μM, blue). The EFB concentration was adjusted to keep binding capacity κ = 40. (E) Normalized simulated Ca²⁺ transients with a low-affinity EFB (K_d = 100 μM) of κ = 15 (green), 26 (blue), and 40 (black) in the presence of 0.1 mM (left) and 2 mM EGTA (right). Normalized experimental Ca²⁺ transients (P14; gray) recorded in 0.1 and 2 mM EGTA are plotted on left and right panels, respectively. Residuals (Res.) were calculated from the difference between the normalized simulation traces and experimental traces.

Figure 5

Effects of Intraterminal EGTA Perfusion on the EPSC Amplitude (A) EPSCs evoked by single presynaptic APs (elicited in current clamp) at P7, before (i) and after (ii) presynaptic internal pipette perfusion of 10 mM EGTA. Time 0 is defined by the time at which 10 mM EGTA was infused (horizontal solid gray bar). (B) EPSCs evoked from a P7 calyx before (i) and after (ii) bath application of 2 μM ω-conotoxin and 0.5 μM SNX, and then after presynaptic internal pipette perfusion from 0.1 to 10 mM EGTA (in the presence of VGCC blockers, iii). The reduction in EPSC amplitude due to blockers was 46% ± 5% (n = 5 calyces). (C) Same as (A) but for P14 calyces. (D) Same as (A) but for P21 calyces. (E) Effects of extracellular application of EGTA-AM (10 μM) on EPSCs evoked by extracellular fiber stimulation of an unperturbed P18 calyx. (F) Summary of the EPSC amplitude reduction (EGTA-inhibition; mean ± SEM) due to intraterminal perfusion of 10 mM EGTA (gray bars) at P7–P21 calyces after application of CgTX and SNX (p = 0.86, unpaired t test) or extracellular application of EGTA-AM (open bar). ^∗p < 0.01; one-way ANOVA followed by Tukey's post hoc.

Figure 6

Estimation of the VGCC-Sensor Distance with 3D Reaction-Diffusion Simulations of [Ca²⁺] and Vesicular Release (A) Cartoon of the “grid array” of 4 and 12 open VGCCs (top and bottom; circles) on the terminal membrane for a 3D reaction-diffusion simulation. The NND of the VGCCs was 35 nm, and each voxel was 5 × 5 × 5 nm. Colored voxels correspond to location of simulation traces shown in (C). Arrows indicate the location of line profiles in (D). (B) Spatial distribution of the [Ca²⁺] in single voxels at the terminal membrane generated by the open VGCC clusters in (A), displayed at the time of peak Ca²⁺ entry, 0.1 ms and 0.2 ms after the peak. We set t = 0 at the time of a 50% rise time of the presynaptic AP. (C) Time course of [Ca²⁺], vesicular release rate, and cumulative vesicular release probability at colored voxel locations in (A) for control conditions (0.1 mM EGTA, black) and 10 mM EGTA (green). (D) Spatial profile of [Ca²⁺], vesicular release probability (P_v), and the fractional reduction of P_v by 10 mM EGTA (EGTA-inhibition). Dashed line and arrow indicates distance at which 55% EGTA-inhibition was observed. (E) Contour plots for isovalue lines of EGTA-inhibition (35%, 55%, and 75%) around the open VGCC clusters shown in (A). (F) EGTA-inhibition as a function of distance between the vesicular Ca²⁺ sensor and VGCC cluster perimeter. The AP waveform and release sensor parameters were set for either P7 or P14 calyces. Horizontal lines indicate the average experimental values of EGTA-inhibition of 56% for P14 (gray) and 69% for P7 (blue). The vertical shaded regions indicate the range of distances between the sensor location and nearest open VGCC matching experimental EGTA-inhibition for differing number of VGCC per cluster for P14 (solid color lines) and P7 (dashed color lines) simulations. The locations where experimental EGTA-inhibition was observed, called PCDs, were 18–20 nm for P14 and 25–32 nm for P7. (G) NND of open VGCCs versus PCD for different number of open VGCCs. (H) P14 simulations showing EGTA-inhibition isovalues as in (E), but using five representative gold particle clusters (black dots) observed from SDS-FRL EM samples (La1). (I) EGTA-inhibition as a function of distance for VGCC locations corresponding to real gold particle patterns containing different numbers of open VGCCs. The PCD ranged between 18 and 20 nm.

Figure 7

Estimation of the PCD and Vesicular Release Probability in a Model with Randomly Generated Patterns of VGCC Openings (A) Left: Spatial line profiles (50 trials, gray lines) showing the fraction of sensors in the release state (release fraction) for voxels at the membrane. Release fraction of each trial was simulated from a randomly generated open channel pattern with open probability of 0.175 and a grid arrangement of 16 VGCCs with NND = 20 nm and [EGTA] = 0.1 mM. Black trace is the average across all trials and thus represents P_v. Green trace is the trial average for [EGTA] = 10 mM. Pairs of solid black ovals denote the VGCC locations. Right: 2D plot of the average P_v (0.1 mM EGTA). (B) EGTA-inhibition and P_v predicted for different sensor locations and numbers of channels using a P14 model (AP₁₄ and P14 Ca²⁺ sensor model). Horizontal line indicates the mean EGTA-inhibition of P_v observed in experiments. Gray region indicates PCD range for different number of VGCCs per cluster. (C) Same as (B) but for a P7 simulation (AP₇ and P7 Ca²⁺ sensor). (D) PCD plotted against the total number of VGCCs within each cluster. (E) P_v for AP₇ (green) or AP₁₄ (black) plotted against the numbers of VGCCs within each cluster. P7 and P14 Ca²⁺ sensors were used, respectively. Arrows indicate predicted total number of VGCCs per cluster: 29 and 26 for P7 and P14, respectively.

Figure 8

Developmental Changes in Vesicular Release Time Course Are Predicted by Perimeter Release Model (A) Representative examples of experimentally measured presynaptic APs, EPSCs, and vesicular release rates before (black) and after (gray) presynaptic perfusion of 10 mM EGTA in P7 and P14 calyces. Dashed lines indicate peak scaked release rate in 10 mM EGTA. (B) Mean synaptic delay (from the 50% rise time of the APs to the 20% rise time of the EPSCs, ± SEM) in the presence of 0.1 mM (open bars) or 10 mM EGTA (filled bars) in the presynaptic pipette solution (n = 10 calyces for P7, n = 8 for P14 and P21). (C) Release half duration (the width at half maximal of the release rate, ± SEM) in the presence of 0.1 mM and 10 mM EGTA, estimated by deconvolution at P7, P14, and P21 (^∗∗p < 0.01, one-way ANOVA). Internal perfusion of 10 mM EGTA reduced the release duration by ∼15% in P14 and P21 (^∗∗p < 0.05, paired t test), but not in P7 calyces. (D) Dependence of release duration on the PCD for AP₇ (green) and AP₁₄ (black) waveforms, simulated with the perimeter release model. Red circles indicate values predicted by experimental results, and an arrow indicates the direction of developmental change. (E) Same as (D) but for synaptic delay. (F) Temporally aligned simulated traces of the [Ca²⁺] and vesicular release rate for [EGTA] = 0.1 mM (black) and 10 mM (gray). For P7 and P14 simulations, the timing and duration of Ca²⁺ entry, number of open VGCCs, Ca²⁺ sensor affinity, and PCD for were adjusted specifically for each age. (G) The simulated effect of 10 mM EGTA on synaptic delay and vesicular release duration for the perimeter release model. (H) Cartoons showing possible AZ topographies for VGCCs and synaptic vesicles released by a single AP. Model 1, random placement of vesicles and VGCCs within AZ. Model 2, vesicles surrounded by rings of VGCCs. Model 3, random placement of vesicles and VGCC clusters, including within vesicle clusters. Model 4, perimeter release model, where releasable synaptic vesicles are positioned at the perimeter of a VGCC cluster. Whether there are more than one releasable vesicle is only speculative.