Evoked release of presynaptic terminals is heterogeneous in short axon segments. A, top: representative images of synaptophysin-pHluorin⁺ synaptic vesicle clusters (green) along an local axon segment from a 14-day in vitro (DIV) neuron after live imaging followed by fixation and immunostaining for MAP2 (red). All of the terminals analyzed in this segment were apposed to neuronal dendrites from multiple postsynaptic neurons. Arrows and arrowheads indicate all presynaptic terminals in this axon segment (defined using criteria described in materials and methods): arrows indicate releasing terminals, and arrowheads indicate nonreleasing terminals. Terminals 1–4 indicate the releasing or nonreleasing presynaptic terminals shown in the pseudocolored time-lapse image series (bottom). The cluster labeled with an asterisk was trafficked to that location in the interval between when the NH₄Cl image was taken and fixation and was excluded from the analysis. Scale bar, 10 μm. Bottom: pseudocolored time-lapse images of presynaptic terminals 1–4 taken before, during (thick horizontal bar, 3.5-s duration), and after electrical stimulation. The changes in fluorescence (ΔF) of evoked vesicle exocytosis (evoked release) at terminals 1–4 are measured as fluorescence by the end of stimulation minus baseline fluorescence (also refer to methods and materials). The color scale is in arbitrary fluorescence units (AFU). B: total fluorescence of individual presynaptic terminals plotted against time to show ΔF during stimulation. The stimulation duration is indicated by the thick horizontal bar below the x-axis from 3.5 to 7 s. The ΔF values of presynaptic terminals 1–4 are heterogeneous, varying over a >10-fold range; e.g., ΔF for terminal 2 was 30.9 × 10⁴ AFU, whereas ΔF for terminal 3 was 3.4 × 10⁴ AFU. Fluorescence returned to baseline within 1–2 min after stimulation ceased, as synaptophysin-pHluorin⁺ vesicles were endocytosed and reacidified (τ = 16.1 ± 0.3 s, 11 presynaptic terminals; comparable to Sankaranarayanan and Ryan 2000). C–E: comparison of pair ratios. C: the pair ratios of evoked release for terminals innervating the same dendrite were not significantly different from the pair ratios of evoked release between 2 consecutive trials of stimulation at the same terminals (Mann-Whitney U-test, P = 0.15) but were significantly less than the average pair ratio of 2 random presynaptic terminals in a short axon segment (Mann-Whitney U-test, P = 0.02). D: the pair ratios for terminals innervating different dendrites from the same postsynaptic neuron were significantly higher than the pair ratios of 2 consecutive trials (Mann-Whitney U-test, P = 0.04) but were not significantly different from the pair ratio of 2 random presynaptic terminals (Mann-Whitney U-test, P = 0.13). E: the pair ratios for terminals innervating different postsynaptic neurons were significantly higher than the pair ratios of 2 consecutive trials (Mann-Whitney U-test, P = 0.002) but were not significantly different from the average pair ratio of 2 random presynaptic terminals (Mann-Whitney U-test, P = 0.314). The pair ratios of evoked release between 2 consecutive trials of stimulation at the same terminals were significantly less than the pair ratios of 2 random presynaptic terminals in a short axon segment (Mann-Whitney U-test, P < 0.001). Values are means ± SE; two-repeats control: 6 neurons, 49 terminals, and 49 pairs of trials; one axon, one dendrite: 9 neurons, 33 pairs of terminals; one axon, two dendrites: 8 neurons, 42 pairs of terminals; one axon, two neurons: 5 neurons, 30 pairs of terminals. Random pairs: 6 neurons, 261 pairs of terminals.

Presynaptic heterogeneity is correlated with the amount of postsynaptic NMDA receptors and PSD-95. Local axon segments were stimulated with 70 pulses delivered at 20 Hz to measure evoked release, followed by NH₄Cl saline perfusion to measure total vesicle pool sizes. Post hoc immunostaining was performed with PSD-95, NR1, GluR1, or MAP2 antibodies to reveal postsynaptic specializations or dendrites. We examined 6–8 neurons for each marker. A: representative images of evoked ΔF (1st panel), total vesicle pool size (2nd panel), post hoc immunostaining of NR1 clusters (3rd panel; green channel is pHluorin, red channel is NR1), and colocalized pixels between NR1 and synaptophysin-pHluorin (4th panel). Arrowheads indicate the terminals plotted in B and C. Scale bar, 5 μm. B: correlation plot of total vesicle pool size and evoked ΔF of terminals shown in A (correlation coefficient r = 0.98, Pearson correlation test, P = 0.02). C: correlation plot of NR1 intensity and evoked ΔF of terminals shown in A (r = 0.99, Pearson correlation test, P = 0.006). D: correlation plot of NR1 intensity and total vesicle pool size of terminals shown in A (r = 0.97, Pearson correlation test, P = 0.03). E: the average correlation coefficient between evoked release and NR1 (9 segments; r = 0.8 ± 0.04) or PSD-95 (11 segments; r = 0.80 ± 0.04) was significantly different from the MAP2 correlation coefficient (Mann-Whitney U-test: NR1, *P = 0.003; PSD-95, *P = 0.003) but not that of GluR1 (r = 0.6 ± 0.1, P = 0.27). Evoked release was not significantly correlated with the amount of MAP2 for terminals in 7/8 axon segments measured (r = 0.4 ± 0.1). F: the average correlation coefficients between total vesicle pool size and NR1 (r = 0.8 ± 0.03) or PSD-95 (r = 0.80 ± 0.04) were significantly different from the MAP2 correlation coefficient (Mann-Whitney U-test: NR1, *P = 0.008; PSD-95, *P = 0.01) but not that of GluR1 (r = 0.7 ± 0.1, P = 0.17). Total vesicle pool size was not significantly correlated with the amount of MAP2 for terminals in 6/8 segments measured (r = 0.5 ± 0.1). G: cumulative distribution of total vesicle pool size in presynaptic terminals colocalized with PSD-95 increased significantly from 10 to 21 DIV (Kruskal-Wallis test, *P < 0.02). H: cumulative distribution of total vesicle pool size in presynaptic terminals not colocalized with PSD-95 was not significantly different from 10 to 21 DIV (Kruskal-Wallis test, P = 0.13). Cc, correlation coefficient.

Spatial distribution of evoked release and total vesicle pool size across single axon arbors. A: Neurolucida tracing of a neuron transfected with synaptophysin-pHluorin and mCherry (not shown) to enable long-distance axon tracking. Colors indicate axon segments that were stimulated and imaged. Images taken after stimulation show the distribution of releasing presynaptic terminals (arrows). Scale bar, 100 μm. B: evoked release of individual presynaptic terminals, color coded by segment. C: the average random pair ratios of evoked release for terminals in color-coded axon segments shown in A (range: 2.9 ± 0.5 to 4.4 ± 0.5; average of all 5 segments from this neuron, white bar), as well as 22/24 axon segments from 6 neurons (average of terminals in all segments: 3.4 ± 0.2), are significantly higher than that for the two-repeats control (indicated by dashed line, 1.9 ± 0.2; Mann-Whitney U-test, P < 0.05). The random pair ratios of evoked release for terminals in color-coded segments and all segments from all axons are not significantly different (1-way ANOVA, P = 0.66). D: total vesicle pool size of individual presynaptic terminals, color coded by segment. E: the average random pair ratios of total vesicle pool size for terminals in short axon segments shown in A (range: 3.3 ± 0.6 to 5.0 ± 0.9; average of all 5 segments from this neuron, white bar), as well as 18/24 axon segments from 6 neurons (average of terminals in all segments: 3.2 ± 0.2), are significantly higher than the two-repeats control (indicated by dashed line, 1.9 ± 0.1; Mann-Whitney U-test, P < 0.05). The random pair ratios of total vesicle pool size for terminals in color-coded segments and all segments from all axons are not significantly different (1-way ANOVA, P = 0.92). F: relationship between coefficient of variance (CV) of evoked release and total vesicle pool size of presynaptic terminals from the most proximal (white circles; r = 0.65, Pearson correlation test, P < 0.05) and the most distal segments (black circles; r = 0.82; Pearson correlation test, P < 0.05) from 13 neurons (slope = 0.96 ± 0.05; Pearson correlation test, P < 0.0001).

Evoked release and release fraction are higher in distal compared with proximal segments of individual axon arbors. A: average evoked release in proximal axon segments (black symbols; 1.3 ± 0.2 × 10⁴ AFU) is slightly but significantly smaller than distal segments (1.9 ± 0.3 × 10⁴ AFU; 16 neurons; Wilcoxon matched-pairs test, P < 0.001). The average evoked release in proximal compared with distal segments is indicated by red symbols. B: average total vesicle pool size in proximal axon segments (black symbols; 15.2 ± 3.9 × 10⁴ AFU) of individual neurons is not significantly different from that in distal segments (10.1 ± 2.0 × 10⁴ AFU; 16 neurons; Wilcoxon matched-pairs test, P = 0.17). The average of total vesicle pool size in proximal compared with distal segments is indicated by red symbols. C: average release fraction in proximal axon segments (black symbols; 0.15 ± 0.02) is significantly smaller than that in distal segments (0.30 ± 0.04; 16 neurons; Wilcoxon matched-pairs test, P = 0.003). The average release fraction of proximal compared with distal segments is indicated by red symbols. D: relationship between evoked release and total vesicle pool size of presynaptic terminals from axon segments I-V from the representative neuron shown and color coded in Fig. 3A. Correlation coefficient and Pearson correlation test results for each segment: segment I, r = 0.67, P = 0.15; segment II, r = 0.86, P = 0.006; segment III, r = 0.81, P = 0.002; segment IV, not available; segment V, r = 0.88, P = 0.004. In this example, as for all axons measured (24 segments, 6 neurons), some segments have a similar regression slope, indicative of similar release fraction (segments I and II, segments III and IV), whereas release fraction is significantly different among other segments (1-way ANOVA, P = 0.007). E: relationship between evoked release and total vesicle pool size of all terminals in the most proximal segments (gray) and the most distal segments (black) from 13 neurons (evoked release: r = 0.90, Spearman correlation test, P < 0.0001; total vesicle pool size: r = 0.85, Spearman correlation test, P < 0.0001). The linear regression lines are significantly different (average proximal slope = 0.05 ± 0.005, distal slope = 0.14 ± 0.009; F-test, P < 0.001), indicating that distal terminals have a significantly higher release fraction compared with proximal terminals.

Higher distal compared with proximal evoked release can be accounted for by higher distal individual vesicle release probability and/or readily releasable pool size. A simple model of vesicle exocytosis in hippocampal nerve terminals was capable of recapitulating the experimental data and provides some insight in the possible mechanisms underlying spatial differences in release properties. A: the number of vesicles releasing neurotransmitter (ER, evoked release) was given by the number of vesicles in the readily releasable pool (RRP), the release probability of individual vesicles (Pv), and the number of times the axon was stimulated (N). The size of the RRP was proportional to total vesicle pool size (TVP), defined as the sum of the RRP, the reserve pool, and the resting pool. The fraction of vesicles released (RF) was defined as the ratio of ER to TVP. Thus ER = RRP × Pv × N; RRP = k × TVP; and RF = ER/TVP, where k = RRP/TVP. Based on an anatomical model of vesicle distribution in terminals (Südhof 2000), TVP was set at 200 vesicles and RRP at 5 vesicles, which yields an initial value of k = 0.025. After substitution and rearrangement, individual vesicle release probability is Pv = RF/(k × N); the measured value of RF, 0.15 for proximal segments, can be used to determine an initial value for Pv in the model. B: TVP was systematically varied over a 10-fold range to generate heterogeneity in TVP and yield values of ER consistent with the experimental data from proximal segments (black line). The measured increase in RF for distal segments from 0.15 to 0.30 could be achieved in the model by either doubling the Pv (red line) or increasing the size of the RRP independent of the size of the TVP, i.e., doubling k (blue line). This holds true for any combination of Pv and RRP such that Pv × k = RF/N. These model findings are consistent with the notion that differences in ER within short axon segments (either proximal or distal) can be attributed to the changes in terminal size, or at least TVP size, whereas proximal/distal differences arise from either changes in RRP size or Pv.