Co-packaging of opposing neurotransmitters in individual synaptic vesicles in the central nervous system

Abstract

Many mammalian neurons release multiple neurotransmitters to activate diverse classes of postsynaptic ionotropic receptors. Entopeduncular nucleus somatostatin (EP Sst+) projection neurons to the lateral habenula (LHb) release both glutamate and GABA, but it is unclear whether these are packaged into the same or segregated pools of synaptic vesicles. Here, we describe a method combining electrophysiology, spatially patterned optogenetics, and computational modeling designed to analyze the mechanism of glutamate/GABA co-release in mouse brain. We find that the properties of postsynaptic currents elicited in LHb neurons by optogenetically activating EP Sst+ terminals are only consistent with co-packaging of glutamate/GABA into individual vesicles. Furthermore, presynaptic neuromodulators that weaken EP Sst+ to LHb synapses maintain the co-packaging of glutamate/GABA while reducing vesicular release probability. Our approach is applicable to the study of multi-transmitter neurons throughout the brain, and our results constrain the mechanisms of neuromodulation and synaptic integration in LHb.

Keywords: GABA; basal ganglia; computational modeling; digital micromirror device; entopeduncular nucleus; glutamate; lateral habenula; neurotransmitter co-release.

Fig. 1.. Electrophysiological and molecular evidence for glutamate/GABA co-release from EP Sst+ axons in LHb.

A) left, Injection of Cre-dependent AAV encoding the optogenetic activator oChIEF into the EP of Sst-Cre mice. right, Expression of tdTom in soma at the injection site (top) and in axons of EP Sst+ neurons in the LHb (bottom). Scale bars=500 µm. B) PSCs recorded from a LHb neuron (V_h=−35 mV) following optogenetic activation of EP Sst+ axons using minimal stimulation in an acute brain slice. Some trials result in failures whereas others evoke both inward and outward PSCs as seen in the biphasic PSCs. Blue: timing and duration of the laser pulse. C) Serial sections of brain tissue containing EP Sst+ terminals expressing Synaptophysin-YFP were sequentially immunostained for multiplex fluorescence imaging. Field of view with antibodies against the pre- (Synapsin 1 (white), Vglut2 (magenta), and Vgat (yellow)) and post-(PSD95 (cyan), Gephyrin (red)) synaptic markers. D) Enlargement of the inset in C) demonstrating colocalization in Synapsin-1-expressing YFP-labelled Sst+ terminals (top) of proteins necessary for GABA (Vgat) and glutamate (Vglut2) release (bottom). E) Z-scored enrichment of immunopuncta within YFP+ boutons relative to that expected at random. Colors indicate data from the same image stack. Dashed lines: ±5 Z-scores. F) Average cross-correlations of Z-scored fluorescence signals for all pairs of antibodies. G) Average co-variances of Z-scored fluorescence signals for all pairs of antibodies within the YFP+ terminals.

Fig. 2.. Statistical features of PSCs predicted by two models of co-release.

A) top, Potential modes of glutamate and GABA co-release from individual synaptic terminals in which each class of vesicle is released independently (left) or the neurotransmitters are co-packaged and released together in the same vesicle (right). bottom, PSCs predicted by the independent (left) and co-packaging (right) models at low p_r. B) i_max and i_min for trials in A) for the independent (left) and co-packaging (right) models. C) Scatterplots of i_max and -i_min of 200 PSCs generated by simulations of independent (p_r=0.5, left) and co-packaging (p_r=0.75, right) models with the same rate of synaptic failures (0.25). Amplitudes are normalized to the average i_max (y-axis) and -i_min (x-axis) of success trials. Histograms (in counts) of the normalized i_max and -i_min with successes of release shown on the right (blue) and top (red) and failures of release in each shown in gray. Successes of release (i_max or -i_min exceed the thresholds indicated by red dotted lines) trials are shown by black filled circles whereas failures are in gray. D) left, Schematics of the areas within the scatterplots used to count events and calculate the probabilities of detecting inhibitory (p(I)) or excitatory (p(E)) currents as well as of biphasic currents (p(E∩I)). center and right, The statistical independence of the probabilities of detecting inhibitory (p(I)) and excitatory (p(E)) PSCs for the two models; the observed probability of excitatory and inhibitory PSCs (p(E∩I), purple) was compared to that expected by chance (p(E)p(I), gray). Results for independent (center) and co-packaging (right) models are shown with p_r=0.5 and were used in E)-F) as well. E) left, Schematics of the areas within the scatterplots used to determine presence or absence of excitatory (top) and inhibitory (bottom) PSCs in each trial. center and right, Simulated cdfs of i_max (blue) given the presence (i_max(E), solid) or absence (i_max(no E), dashed) of EPSC in the independent (center) and co-packaging (right) models. Similar analyses were performed for the -i_min (red) given the presence (-i_min(I), solid) or absence (-i_min(no I), dashed) of an IPSC. F) left, Schematics of the areas of the scatterplots that contain all (top) or success (bottom) trials. center and right, Analysis of the trial-by-trial correlation of i_max and -i_min of all trials (dark green), success trials (light green), and after shuffling (gray).

Fig. 3.. Approach to measure PSCs evoked by optogenetic stimulation of EP Sst+ axons in LHb.

A) DMOS setup. S: mechanical shutter; HD: holographic diffuser (10° diffusing angle); DMD: digital micromirror device; L1–2: lens; OBJ: objective lens. B) Workflow schematic of Cre-dependent AAV encoding the optogenetic activator oChIEF injection into the EP of Sst-Cre mice, followed by recordings in acute-brain slices of LHb. C) Optically-evoked average E- and IPSCs acquired at V_h=−70 (red) or 0 (dark blue) mV, respectively. Light blue vertical bars: timing of the laser stimulation pulses with each delivered to a different location in the slice. PSCs are the average of 5 trials. D) The number of stimulation spots triggering PSCs (x-axis) in individual (top, y-axis) or across all (bottom) cells grouped by EPSC only (orange), IPSC only (blue), or both (purple). E) Optically-evoked average biphasic, compound PSCs recorded at V_h=−35 mV, in the same neuron as in C). PSCs are the average of 5 trials. Inset shows the expanded PSC inside the gray shaded box. F) Fitted IPSC/EPSC amplitude relationships for data from 6 LHb cells (left) and corresponding R² values (right). Colors indicate cell identity as in D).

Fig. 4.. DMOS evoked unitary responses from EP Sst+ axons in LHb.

A) Spatial heatmaps showing the effects of sequential addition of TTX and 4-AP on total charge of EPSCs (V_h=−64mV) of all stimulation spots using DMOS under high photo-stimulation intensity. The cell was located approximately at the center. B) Spatial heatmaps comparing total charge of EPSCs (V_h=−64mV) using DMOS under high (top) and minimal (bottom) photo-stimulation intensity. C) Average (top) and individual (bottom) unitary PSCs recorded at an intermediate V_h. Repetitive stimulation at 3 spots consistently evoked EPSC-only (red), IPSC-only (blue), or biphasic (purple) PSCs. D) The proportions of minimal stimulation spots that triggered PSCs at V_h=−27 or −35 mV, as indicated (top), or across all cells (bottom) grouped as EPSC-only (orange), IPSC-only (blue), or biphasic (purple). Asterisks indicate statistical significance (Fisher’s exact test) of differences in proportions of each group at −27 and −35 mV.

Fig. 5.. Unitary responses from glutamate/GABA co-releasing boutons.

A) Optically-evoked PSCs from a hotspot consistent with the independent model. top, Example traces aligned to stimulus onset (blue region). i_max (blue dot) and i_min (red dot) were extracted from within the gray region. bottom, Histogram of i_max (blue) and i_min (red) timing. B) Scatterplot of -i_min vs. i_max for all trials at the spot shown in A). C) Comparison of the p(E∩I) (purple) to p(E)*p(I) (gray). left, Histograms of probabilities generated from bootstrap analysis of actual (top) and shuffled (bottom) data. right, Simulated histograms of probabilities generated by independent (top) and co-packaging (bottom) models using synaptic parameters extracted from the data in B). D) Cdfs of i_max(E) (blue solid), i_max(no E) (blue dashed), -i_min(I), (red solid), and -i_min(no I) (red dashed) for data in B). E) Correlation of i_max and -i_min across all trials (dark green), success trials (light green), and across all trials after shuffling (gray). Bootstrapped correlation coefficients for data from B) (left) and for results of simulations (parameters as in C)) of independent (middle) and co-packaging (right) models. F-J) As in panels A)-E) but for a spot with properties consistent with the co-packaging model.

Fig. 6.. Statistical analyses of all co-releasing terminals support the co-packaging model.

A) Parametrization of a model feature indicator calculated by subtracting the medians of the cdfs (Δcdf_0.5) representing the distributions of p(E∩I) (purple) and p(E)*p(I) (gray). B) Histograms of the model feature indicators for the 5 statistical features. The data represent “both” group in Fig 4F. Bin width is 0.05. C) Heatmap of transformed model feature indicators from B (y-axis) across spots exhibiting both PSCs (x-axis). Color intensity represents increasing support for the co-packaging model. D) Correlation heatmap of model feature indicators. E) Average model feature indicators of all unitary co-releasing spots from C). Each dot is data from one spot, with color indicating the cell identity. Larger values indicate greater support for the co-packaging model. Data from the black outlined spots are shown in detail in G. F) Three noise sources in the recordings. G) Scatterplots of i_max and -i_min for ambiguous (left, green dot from E) and co-packaging (right, orange dot from E) hotspots. Histograms of the i_max and -i_min of evoked (right, blue; top, red) and spontaneous (brown) PSCs. H) Average model feature indicators for individual spots are correlated with the fraction of outlier current values during baseline periods (left) and the average i_max and -i_min SNR (right). Colors indicate cells identities as in E). Pearson correlation coefficients are given.

Fig. 7.. 5-HT reduces pr of glutamate and GABA while maintaining their co-packaging.

A) DMD ring stimulation used to generate propagating action potentials in EP Sst+ axons. B) Average EPSC (V_h=−64 mV) and IPSC (V_h=10 mV) using ring stimulation (blue box) recorded in a LHb neuron before (gray dashed) and after (black) application of 5-HT (1 µM). C-D) Schematic of DMOS to activate individual presynaptic boutons (C) and average biphasic PSCs (V_h=−35 mV, black line) before (top, n=141 trials) and after (bottom, n=147 trials) application of 5-HT (250 nM) (D). 5-HT proportionally reduced the PSC – the average PSC before 5-HT application is shown scaled and overlaid (gray) on the bottom. E) Optically-evoked PSCs from a hotspot consistent with the co-packaging model depicted as in Fig. 5F. F) Scatterplot of i_max and -i_min for the spot shown in E) depicted as in Fig. 5G). The probabilities of detecting an EPSC, IPSC, and both are given. G) left, Optically-evoked biphasic unitary PSCs (V_h=−35 mV) of the “both” success trials. right, Cdfs of i_max (blue) and -i_min (red) of these trials. H) Analysis of statistical features for the data in F): left, Comparison of the p(E∩I) (purple) to p(E)*p(I) (gray). middle, Cdfs i_max (blue) given the presence (solid) or absence (dashed) of an EPSC and vice versa. right, Correlation of i_max and -i_min across all (dark green), success (light green), and shuffled (gray) trials. I-L) As in panels E-H after 5-HT (250 nM) bath application for the same site.

Fig. 8.. Summary of 5-HT effects on glutamate/GABA co-packaging.

A) 5-HT effect on quantiles of i_max and -i_min for the data shown in 7E-L. Dots show the amplitudes of the average trace for quantile of the data before (gray) and after (black) 5-HT application. B) As in A) for showing the predictions of the independent (orange) and co-packaging (black) models. C) 5-HT effects on average model feature indicators of unitary co-releasing spots consistent with co-packaging model based on the statistical features shown in Fig. 6C. Arrows indicate the direction of the change due to 5-HT application. D) Average ratio between PSC and noise of individual spots vs. average model feature indicators. Pearson correlation coefficient is given. E) Comparison of observed model axis change due to 5-HT and that predicted by simulation with considering changes in pr, noise, and amplitudes of biphasic PSCs. Linear fit (solid gray), the 95% prediction interval (dashed), and the Pearson correlation coefficient are shown.