Multiplex imaging relates quantal glutamate release to presynaptic Ca2+ homeostasis at multiple synapses in situ

Abstract

Information processing by brain circuits depends on Ca²⁺-dependent, stochastic release of the excitatory neurotransmitter glutamate. Whilst optical glutamate sensors have enabled detection of synaptic discharges, understanding presynaptic machinery requires simultaneous readout of glutamate release and nanomolar presynaptic Ca²⁺ in situ. Here, we find that the fluorescence lifetime of the red-shifted Ca²⁺ indicator Cal-590 is Ca²⁺-sensitive in the nanomolar range, and employ it in combination with green glutamate sensors to relate quantal neurotransmission to presynaptic Ca²⁺ kinetics. Multiplexed imaging of individual and multiple synapses in identified axonal circuits reveals that glutamate release efficacy, but not its short-term plasticity, varies with time-dependent fluctuations in presynaptic resting Ca²⁺ or spike-evoked Ca²⁺ entry. Within individual presynaptic boutons, we find no nanoscopic co-localisation of evoked presynaptic Ca²⁺ entry with the prevalent glutamate release site, suggesting loose coupling between the two. The approach enables a better understanding of release machinery at central synapses.

Fig. 1

Fluorescence lifetime of Cal-590 provides readout of low intracellular Ca²⁺. a Fluorescence lifetime decay curves of Cal-590 in a series of calibrated [Ca²⁺]-clamped solutions finely adjusted to include appropriate intracellular ingredients^, (Methods). Fluorescence lifetime traces are normalised to their peak values (at ~1.3 ns post-pulse); the area under the curve over the time interval of 3 ns (tan shade) was measured and related to the peak value thus providing normalised total count (NTC) readout; λ_x^2p = 910 nm; temperature 33 °C. b The Cal-590 fluorescence lifetime imaging microscopy (FLIM) [Ca²⁺] sensitivity estimator, NTC, fitted with sigmoid-type function (χ² = 2.11 × 10⁻⁵, R² = 0.996). c Top: Example of a CA3 pyramidal cell (SF-iGluSnFR channel); arrowheads, proximal part of the traced axon; patch pipette is seen. Bottom: Example of an axonal bouton (Cal-590 channel) traced from the cell soma; spiral line and arrow, tornado linescan applied in the middle of the bouton; scale bar, 50 µm (top) and 1 µm (bottom). d Example of a single-bouton Cal-590 signal during a 500-ms 100-Hz burst of spike-inducing somatic 1 ms current pulses: recorded as a tornado linescan (top; ordinate, spiral rotation angle 0–10π reflects five concentric spiral circles, 2π radian/360° each), fluorescence intensity (integrated over the 10π spiral scan) time course (middle), and fluorescence decay (FLIM, bottom; ordinate, decay time; grey arrowhead, laser pulse onset). Scale bars, 300 ms (top, horizontal), 2.0 ΔF/F (middle, vertical), 5 ns (bottom, vertical). e Intra-bouton Cal-590 fluorescence decay time course (normalised to peak) representing basal [Ca²⁺] (blue) and peak [Ca²⁺] (red), in the experiment shown in d

Fig. 2

Multi-synapse imaging of quantal glutamate release with SF-iGluSnFR.A184S. a CA3 pyramidal cell axon fragment in area CA1 showing four presynaptic boutons (b1–b4); the scanning dwell points in the bouton centres (red dots, dwell-delay time ~1.5 ms per bouton) and laser scan trajectory (dotted yellow line) illustrated; scale bar, 5 µm. b A pseudo-linescan image of SF-iGluSnFR.A184S signals recorded simultaneously (one sweep example) at four boutons shown in a as indicated, during somatic generation of two action potentials 50 ms apart (top trace, current clamp). Arrow diagram relates displayed pixels to the scanning cycle: pixels at each displayed ith line are recorded sequentially among boutons (small arrows), and the next cycle fills the (i+1)th line of the display. Thus a pseudo-linescan image is generated showing brightness dynamics at individual boutons, with ~1.5 ms resolution; glutamate releases and failures can be seen; scale bars, 60 mV (vertical) and 100 ms (horizontal). c A summary of 22 trials in the experiment shown in a, b; green traces, single-sweep SF-iGluSnFR.A184S intensity readout at the four bouton centres; black traces, all-sweep-average; P₁:P₂, average probability P_r (release success rate) of the first (red) and second (blue) release events; scale bars, 50% ΔF/F (vertical) and 100 ms (horizontal). d Amplitude histograms (SF-iGluSnFR.A184S ΔF/F signal, first and second response counts combined; prepulse baseline subtracted) with a semi-unconstrained multi-Gaussian fit (blue line, Methods) indicating peaks that correspond to estimated quantal amplitudes; the leftmost peak corresponds to zero signal (failure; yellow shade); dotted lines, individual Gaussians; arrows, average amplitudes (including failures) of the first (red) and second (blue) glutamate responses

Fig. 3

Multiplex imaging of quantal glutamate release and presynaptic Ca²⁺ dynamics. a CA3 pyramidal cell. Left: SF-iGluSnFR.A184V channel (Methods); arrows, two main axonal branches; ~60 µm z-stack, λ_x^2p = 910 nm. Right, Cal-590 channel (~20 µm z-stack fragment), whole-cell (300 µM Cal-590); patch pipette seen; scale bars, 20 µm. b Axonal bouton traced into CA1 from CA3 pyramid^, : left, SF-iGluSnFR.A184V channel; tornado scan position shown; right, tornado linescan examples (rotation angle 0–5π rad reflects 2.5 spiral circles) during four evoked action potentials (APs; 20 Hz, arrows) shown in SF-iGluSnFR.A184V (green, note release failures) and Cal-590 (magenta) channels; scale bars, 2 µm (left), 200 ms (right). c Four characteristic sequential one-sweep recordings (SF-iGluSnFR channel) depicting individual quantal releases and failures. d Summary, glutamate release kinetics in bouton shown in b (SF-iGluSnFR.A184V channel; green, 20 trials 1 min apart; black, average); scale bars, 50% ΔF/F (vertical) and 100 ms (horizontal). e Summary, Cal-590 intensity signal ΔF/F (red channel) recorded as in d (magenta, individual trials; black, average); arrow, spontaneous Ca²⁺ entry with no glutamate release (see d); scale bar, 30% ΔF/F. f Dynamics of free presynaptic [Ca²⁺] averaged over 20 trials: Cal-590 fluorescence lifetime imaging microscopic readout (normalised photon count), converted to [Ca²⁺]. Note: data reflect free ion concentration volume-equilibrated and time-averaged over 5–10 ms. g Amplitude histograms (SF-iGluSnFR.A184V ΔF/F signal, first–to-fourth response counts combined; 8 ms pre-AP baseline subtracted); blue line, semi-constrained multi-Gaussian fit (Methods); peaks correspond to quantal amplitudes; leftmost peak, zero signal (release failure noise; yellow shade); false-positive cut-off at ~0.12 ΔF/F. h Trial-to-trial glutamate release signal amplitude (bouton in b) shown as ΔF/F SF-iGluSnFR.A184V readout (grey dots) and as quantal content (magenta bars, right ordinate; calculated from histogram in g) plotted against evoked Ca²⁺ entry signal (ΔF/F Cal-590, as in e) for all recorded APs (8 ms pre-AP baseline subtracted). Black and magenta lines, linear regressions (r, Pearson’s correlation; p, regression slope significance) for ΔF/F and quantal content data, respectively; yellow shade, failure response cut-off (as in g)

Fig. 4

Intra-axonal Ca²⁺ buffering by Cal-590 has little effect on [Ca²⁺] fluorescence lifetime imaging microscopic (FLIM) readout or average release probability. a Total photon count for Cal-590 (fluorescence intensity, measured within 300 ms before the evoked four-action potential (four-AP) train at 20 Hz, as in Figs. 3 and 4), normalised to the initial value, as recorded in individual axonal boutons (N = 7), plotted against the trial number (time). In these experiments, the intra-axonal Cal-590 concentration, hence Ca²⁺ buffering capacity, continues to rise with time, before eventual equilibration; solid line, linear regression (r, Pearson correlation; p < 0.001, slope significance). b Presynaptic resting [Ca²⁺] measured within 300 ms before the evoked four-AP train using Cal-590 FLIM readout, at axonal boutons shown in a, plotted against the trial number (time); other notation as in a (no correlation). c An increment in presynaptic [Ca²⁺] during evoked four-AP train measured with Cal-590 FLIM, at axonal boutons shown in a, b, plotted against the trial number (time); other notation as in a, b (no correlation). d Probability of evoked glutamate release (in response to a single action potential) in axonal boutons of CA3 pyramidal cells (organotypic hippocampal slices) expressing SF-iGluSnFR, with and without Cal-590 (300 µM) being loaded and equilibrated in whole-cell mode, as indicated. Dots, individual bouton recordings; bar graph, mean ± s.e.m. (0.41 ± 0.07 and 0.42 ± 0.06, n = 20 and n = 26, without and with Cal-590, respectively)

Fig. 5

Resting presynaptic Ca²⁺ and evoked Ca²⁺ entry control glutamate release but not its short-term plasticity. a Collage, CA3 pyramidal cell with axon traced into area CA1 (example; SF-iGluSnFR channel; xy projections of 10–15 µm z-stacks); inset, zoomed-in axonal fragment (orange dotted rectangle) depicting four axonal boutons B1–B4 (P_r, average release probability); scanning mode: tornado. Traces, summary recordings of glutamate release (green) and Ca²⁺ dynamics (red), as indicated; scale bars, 20 µm (left image), 100 ms (horizontal middle), 50% ΔF/F (green, SF-iGluSnFR.A184V), and 15% ΔF/F (red, Cal-590), 5 µm (right image). b Quantal content of glutamate release upon first action potential (AP), plotted against resting [Ca²⁺] (Cal-590 fluorescence lifetime imaging microscopic (FLIM) readout, averaged over 100 ms prepulse). Blue bars, the number of released vesicles estimated from the frequency histogram of the SF-iGluSnFR.A184V ΔF/F signal (as in Fig. 3g, h). Solid blue line, linear regression (r, Pearson's correlation; p, regression slope significance; n, number of events, N = 26 boutons recorded). See Supplementary Figure 4a for raw ΔF/F data. c Quantal content of cumulative glutamate release upon four APs, plotted against presynaptic resting [Ca²⁺] (Cal-590 FLIM readout); other notations as in b. See Supplementary Figure 4b for raw ΔF/F data. d Cumulative glutamate release upon four APs (as in c; tests with two APs excluded), plotted against cumulative [Ca²⁺] increment (Cal-590 FLIM readout during four APs); other notations as in b (N = 24 boutons). See Supplementary Figure 4c for raw ΔF/F data. e Paired-pulse ratio (PPR, ratio between second and first ΔF/F SF-iGluSnFR signal amplitudes, pre-AP 8 ms baselines subtracted) plotted against resting [Ca²⁺]. Other notations as in b. f Amplitude of ΔF/F SF-iGluSnFR signal evoked by 1–4 APs, relative to the first-AP ΔF/F signal, plotted against the AP number (1–4; N, number of recorded boutons). g Short-term facilitation/depression of AP burst-evoked glutamate release (linear regression slope over 1–4 ΔF/F SF- iGluSnFR signal change shown in f; tests with two APs were excluded) plotted against resting [Ca²⁺] (blue dots) and AP-evoked [Ca²⁺] increment (red circles). See Supplementary Figure 5 for trial-to-trial short-term plasticity readout, plotted against [Ca²⁺] parameters

Fig. 6

Nanoscale geometry of presynaptic bouton imaging with two-photon excitation. a CA3–CA1 axonal bouton (example), two-channel image (SF-iGluSnFR.A184V green, Cal-590 magenta); scale bar, 1 µm. b Axonal bouton diagram illustrating laser scanning settings, x–y (left) and z (right) planes, with membrane-bound SF-iGluSnFR (green) and cytosolic Cal-590 (magenta); black spiral (left) and straight line (right), tornado linescan trajectory; yellow shapes, characteristic point-spread function (PSF). c Diagram, typical experimental arrangement, with a tornado linescan (spiral) over an ellipsoidal axonal bouton (a and b, major and minor axes, respectively, in the elliptical x–y projection of the rotational ellipsoid). d Trigonometric diagrams explaining geodesic corrections for curved-surface distances projected onto the x–y plane. Left, for a circle or cylinder of radius b, the projected distance x_i from the centre corresponds to the curvilinear (geodesic) distance bφ where angle φ = arcsin (x_i/b) (green segment, geodesic distance ‘straightened’ in plane of view); thus the projected distance from the centre to the edge, b, corresponds to the geodesic distance πb/2. Right, for an elliptical section with major and minor axes a and b, respectively, the projected distance x_i from the centre corresponds to the geodesic distance (L_e, green segment), which is smaller than that for a circular correction (L_c, blue segment). The difference between L_e and L_c depends on the expected a/b ratio (see Supplementary Figure 4d–f for detail). e Example of a recorded bouton (SF-iGluSnFR.A184V channel), with a tornado scan shown; dotted oval, an estimated outline of the axonal bouton projection; scale bar, 1 µm. f Characteristic heat map of the ΔF/F SF-iGluSnFR.A184V fluorescence signal generated by action potential-evoked glutamate release (one-bouton example; imaging arrangement as in e, circular shape follows the tornado scan); average of 21 trials 1 min apart; dotted oval, bouton outline as shown in e; image projected on to the focal x–y plane; scale bar, 0.5 µm. g Heat map of ΔF/F SF-iGluSnFR as in f corrected for geodesic distances as opposed to the projected (visible) distances, in accord with the geometry of the bouton outline and the tornado scan position; scale bar, 0.5 µm

Fig. 7

Evaluating sub-microscopic glutamate signal spread and co-localisation of glutamate release and presynaptic Ca²⁺ entry in presynaptic boutons. a Examples of unimodal (left, 20/23 boutons) and bi-modal (right, 3/23 boutons) spatial profiles of the SF-iGluSnFR signal intensity (20-22 trial average) upon glutamate release, as seen in the focal plane; averaging spatial filtering (~100 nm range) applied; peaks point to the most likely release site location. b Spatial decay of the fluorescence signal (ΔF/F SF-iGluSnFR.A184V) measured in geodesic-corrected heat maps (as in Fig. 6g) for N = 23 individual boutons (green lines); circles, average; solid red line, best-fit exponent: spatial decay constant σ = 0.547 ± 0.016 µm (mean ± standard error). c Examples of simultaneously recorded heat maps showing glutamate signal (green, ΔF/F SF-iGluSnFR) and [Ca²⁺] increment profiles (magenta, Cal-590 fluorescence lifetime imaging microscopic readout upon four action potentials, magenta); dotted shapes, outlines of the recorded axonal bouton profiles; red dots, estimated glutamate release site locations; arrows, consistent hotspots of Cal-590 [Ca²⁺] transients; scale bar, 0.5 µm. d Diagrams illustrating the measurement of distances between recorded hotspots of glutamate and Ca²⁺ signals, as shown in c (left), and between points randomly scattered within a similar circular area (right). e The average distance between recorded hotspots of glutamate and Ca²⁺ signals (mean ± s.e.m., 0.308 ± 0.073 µm, n = 10; grey dots) is significantly lower than that between randomly scattered points (0.474 ± 0.025, n = 80, orange dots; *p < 0.028, t -test)