1. Neuroscience

GABABR silencing of nerve terminals

  1. Daniel C Cook
  2. Timothy A Ryan  Is a corresponding author
  1. Department of Anesthesiology, Weill Cornell Medical College, United States
  2. Department of Biochemistry, Weill Cornell Medical College, United States
https://doi.org/10.7554/eLife.83530
Share this article
Cite this article
  1. Daniel C Cook
  2. Timothy A Ryan
(2023)
GABABR silencing of nerve terminals
eLife 12:e83530.
https://doi.org/10.7554/eLife.83530
  2. Download BibTeX
  3. Download .RIS

Abstract

Control of neurotransmission efficacy is central to theories of how the brain computes and stores information. Presynaptic G-protein coupled receptors (GPCRs) are critical in this problem as they locally influence synaptic strength and can operate on a wide range of time scales. Among the mechanisms by which GPCRs impact neurotransmission is by inhibiting voltage-gated calcium (Ca2+) influx in the active zone. Here, using quantitative analysis of both single bouton Ca2+ influx and exocytosis, we uncovered an unexpected non-linear relationship between the magnitude of action potential driven Ca2+ influx and the concentration of external Ca2+ ([Ca2+]e). We find that this unexpected relationship is leveraged by GPCR signaling when operating at the nominal physiological set point for [Ca2+]e, 1.2 mM, to achieve complete silencing of nerve terminals. These data imply that the information throughput in neural circuits can be readily modulated in an all-or-none fashion at the single synapse level when operating at the physiological set point.

Editor's evaluation

In this important study Cook and Ryan find that at physiological temperatures, small changes in extracellular calcium concentrations shift individual presynaptic terminals between active and inactive states. This discovery is extended to the action of inhibitory synaptic modulation via GABA, emphasizing potentially broad relevance for our understanding of how synaptic transmission is stabilized and modulated throughout the brain. The evidence for the observed phenomenon is compelling, given the use of multiple functional reporters and rigorous analyses.

https://doi.org/10.7554/eLife.83530.sa0

Introduction

GPCRs serve a key control function in the brain, modifying the efficacy or time course of neurotransmission through pre- and postsynaptic mechanisms (Nadim and Bucher, 2014; Lovinger et al., 2022). Ligands of GPCRs are often secreted near synapses and exert their effects locally (van den Pol, 2012; van Westen et al., 2021; Ding et al., 2019). They can thereby modify region-specific synaptic properties and change the weighting of targeted synapses, in turn altering local computation. A canonical mode of presynaptic modulation occurs via the binding of voltage-gated calcium channels (VGCCs) by G-protein βγ subunits after they dissociate from the trimeric G-protein complex following GPCR activation (Herlitze et al., 1996). The impact of GPCR-mediated inhibition of VGCCs on neurotransmission (Kreitzer and Regehr, 2000; Yamada et al., 1999) is amplified by the exquisite sensitivity of exocytosis to the magnitude of Ca2+ influx, which underlies the ability to rapidly change exocytotic rates at sites of neurotransmitter release. This sensitivity is driven by the cooperative binding of multiple Ca2+ ions to one or more Ca2+ sensors that form core elements of the exocytic machinery (Heidelberger et al., 1994; Schneggenburger and Neher, 2000; Bollmann et al., 2000; Ariel and Ryan, 2010). Modifying the open probability of VGCCs, therefore, serves as a potent means of presynaptic control, as small changes in Ca2+ influx can lead to large changes in neurotransmitter release.

The impact of Ca2+ on exocytosis depends on local changes in Ca2+ concentration near the active zone, which in turn depend on the external Ca2+ concentration bathing the nerve terminal (Dittman and Ryan, 2019). Thus, the steepness of the relationship between exocytosis rates and Ca2+ influx depends crucially on resting [Ca2+]e. In vivo, plasma [Ca2+] is strictly regulated, and the CNS is very sensitive to plasma [Ca2+] perturbations, with seizures precipitated by hypocalcemia and psychiatric symptoms progressing to lethargy or coma with worsening hypercalcemia (Riggs, 2002). Feedback loops controlled by Ca2+ sensors of the parathyroid gland act to maintain plasma [Ca2+] at a set point of 1.2 mM in most mammals (Hannan et al., 2019). Here, we used quantitative optical tools to investigate the sensitivity of exocytosis rates at the single-synapse level when operating at the physiological set point for [Ca2+]e. In a narrow range of [Ca2+]e near the physiologic concentration, our experiments revealed an unexpected sub-proportionality of Ca2+ influx relative to [Ca2+]e, indicating that a minimal Ca2+ entry is needed to sustain robust Ca2+ influx during AP firing. We show that three different approaches to modestly decrease Ca2+ influx, (1) blocking a fraction of Ca2+ channels, (2) lowering [Ca2+]e, or (3) application of a GPCR agonist, baclofen, known to reduce Ca2+ influx lead to complete silencing of a subset of nerve terminals. Thus, in this operating regime, decreasing presynaptic Ca2+ current may act as a digital switch to eliminate single bouton synaptic throughput.

Results

Ca2+ influx into nerve terminals is not 1:1 proportional to [Ca2+]e near the physiologic set point

In order to examine how Ca2+ influx driven by action potentials (APs) at nerve terminals is impacted by changes in [Ca2+]e, we used the genetically-encoded Ca2+ indicator physin-GCaMP6f transfected sparsely into primary cultures of dissociated rat hippocampal neurons (Figure 1). We measured the changes in the GCaMP6f signal in response to a 20 AP burst stimulus (20 Hz) at 0.8, 1.2, and 2.0 mM [Ca2+]e (Figure 1A and B). Signals were corrected for the local abundance of the protein by determining the GCaMP6f fluorescence under saturating Ca2+ binding (determined following ionomycin application) and converted to absolute intracellular Ca2+ concentrations ([Ca2+]i, see methods). Previous studies investigating the dependency of Δ[Ca2+]i to [Ca2+]e in nerve terminals of the CNS found the relationship obeyed a simple saturable pore model with a Kd of [Ca2+]e ~2.5 mM (Ariel and Ryan, 2010; Schneggenburger et al., 1999). This model predicts that changes in Δ[Ca2+]i will exhibit 1-to-1 proportionality to [Ca2+]e at concentrations well below Kd, which is substantially higher than the physiologic set point ([Ca2+]e = 1.2 mM). One important confound, however, is that previous measurements were not carried out at physiological temperatures. Here, where the temperature was set to 37 °C, we observed an unexpected sub-proportional relationship of Δ[Ca2+]i to [Ca2+]e as the concentration was lowered below 2.0 mM (Figure 1B and C). The magnitude of difference between the expected and experimentally measured Δ[Ca2+]i was greater at [Ca2+]e 0.8 mM compared to 1.2 mM, highlighting that this discrepancy manifests most prominently below the set point (Figure 1C). A linear extrapolation from the mean Δ[Ca2+]i measurements relative to [Ca2+]e predicts the cessation of Ca2+ entry into nerve terminals at [Ca2+]e of 0.47 mM (Figure 1C). Experiments carried out using [Ca2+]e = 0.4 mM confirm that there is almost no measurable residual Ca2+ influx under these conditions (estimated Δ[Ca2+]~1.6 nM; Figure 1—figure supplement 1), far below the value expected (10.4 nM) if Ca2+ influx obeyed a simple linear proportionality.

Figure 1 with 1 supplement see all
Download asset Open asset
Unexpected sub-proportionality in dependence of presynaptic calcium (Ca2+) influx on [Ca2+]e.

(A) Schematic of the experimental protocol. Neurons are stimulated with a brief action potential (AP) train (20 AP in 1 s) at three different [Ca2+]e centered on the physiologic set point of 1.2 mM. Ionomycin, a Ca2+ ionophore, is administered at the end of the experiment to convert fluorescence into absolute [Ca2+]i. (B) An axon from a neuron expressing GCaMP6f localized to nerve terminals by synaptophysin (physin-GCaMP) at rest (top) and a difference image showing the response to 20 AP (bottom). The difference image is the mean peak (five frames) subtracted by the mean baseline (49 frames) in [Ca2+]e 2 mM. For display, a representative subset of terminals was highlighted. Scale bar is 10 µm. (C) Traces of responses to 20 AP from the neuron shown in B, with [Ca2+]e color-coded as in A. Expected Δ[Ca2+]i at 0.8 and 1.2 mM is represented by dashed lines and calculated as a change in influx relative to 2.0 mM proportional to the ratios of [Ca2+]e. Arrow bars show the difference between expected and measured Δ[Ca2+]i. Traces are mean, with error (SEM) represented by gray lines in only pre-stimulus and peak frames for clarity (n=211 nerve terminals). (D) Summary of Δ[Ca2+]i as a function of [Ca2+]e (mean ± 95%CI, n=9 neurons). The x-axis intercept for measured changes in Δ[Ca2+]i is 0.47 mM. *p<0.05, one-sample t-test compared to the expected value.

Modest changes of [Ca2+]e near the physiologic set point controls the proportion of silent nerve terminals

The above measurements represent ensemble averages of all terminals with a measurable signal following the saturation of physin-GCaMP. To further probe the apparent non-zero threshold of [Ca2+]e that enables Ca2+ entry, we assessed physin-GCaMP responses at individual nerve terminals with varying [Ca2+]e (Figure 2). Surprisingly, we found that a subset of terminals exhibited an all-or-none response to lowering [Ca2+]e below the physiologic set point such that Δ[Ca2+]i is selectively abolished (Figure 2B). To rigorously quantify this behavior, we employed a modest cutoff of one standard deviation of the pre-stimulus noise above the mean fluorescence to separate responding terminals from those without measurable synaptic activity, termed silent (Moulder et al., 2008; Altrock et al., 2003; Kim and Ryan, 2010). This definition of presynaptic silencing has been used previously in neurons expressing GCaMP3, a less sensitive indicator than GCaMP6f, and compared to the low-affinity, synthetic Ca2+ dye Magnesium Green with an excellent agreement, indicating that GCaMP sensors can reliably distinguish responding and silent terminals despite their non-linear Ca2+ dependence (Kim and Ryan, 2013). To evaluate the effectiveness of our thresholding, we compared cumulative frequency curves of ΔF for terminals separated into silent and responding populations (Figure 2—figure supplement 1). Responding terminals exhibit a clear shift towards higher ΔF with increasing [Ca2+]e, as expected, but the distribution of silent terminals was invariant with respect to [Ca2+]e. This analysis demonstrates that silent terminals are appropriately classified as incapable of generating detectable Ca2+ influx that would otherwise scale with [Ca2+]e. We observed that silencing occurred despite robust Δ[Ca2+]i at higher [Ca2+]e and even if synaptic neighbors of a terminal exhibited persistent Ca2+ entry at a lower concentration (Figure 2A and B), suggesting the impact of [Ca2+]e on neuronal function is exerted at the single synapse level. To further evaluate whether silencing may be driven by changes in the excitability of individual branches of neurons, we expressed GCaMP6f in the cytosol to clearly label the axon and marked terminals by co-expression of mRuby-synapsin (Figure 2—figure supplement 1A, B). We limited our analysis to regions unambiguously representing a single branch that had at least 10 terminals to provide a robust sample. These experiments showed that individual axonal branches have both silent and responding terminals (Figure 2—figure supplement 1C, D), indicating that branch point failure cannot account for [Ca2+]e mediated silencing.

Figure 2 with 2 supplements see all
Download asset Open asset
Selective nerve terminal silencing is caused by changing [Ca2+]e around the physiologic set point.

(A) Pre-stimulus average (left) and difference images (right) of AP stimulation (20 APs in 1 s) of a neuron expressing physin-GCaMP. The difference image is the mean peak (five frames) subtracted by the mean baseline (49 frames) at [Ca2+]e 2.0 and 0.8 mM. For display, a subset of representative terminals was selected and a Gaussian convolution with the radius of 1 pixel was applied. (B) Traces of the terminals indicated in A, shown as ΔF, demonstrate selective silencing (upper row compared to lower row) of a subset of terminals. Dotted line is the pre-stimulus mean fluorescence for each terminal.

Across a population of neurons (Figure 3), we found substantial heterogeneity in the proportion of silent terminals (coefficient of variation was 38%, 56%, and 57% for [Ca2+]e 0.8, 1.2, and 2.0 mM, respectively) but a strikingly disproportionate increase in the mean proportion of silencing below 1.2 mM (Figure 3C). The change in percentage of silencing versus change in [Ca2+]e was 48.9% mM–1 transitioning from 1.2 to 0.8 mM compared to 19.0% mM–1 from 2.0 to 1.2 mM (Figure 3C). The change in percentage of silencing versus change in [Ca2+]e was 48.9% mM–1 transitioning from 1.2 to 0.8 mM compared to 19.0% mM–1 from 2.0 to 1.2 mM (Figure 3C). We separately tested the proportion of silencing when transitioning from 1.2 mM to 0.4 mM, observing that nearly all terminals (93%) are silent in [Ca2+]e 0.4 mM (Figure 3—figure supplement 1). Here, the coefficient of variation, 5%, was substantially lower (32% in paired measurements in [Ca2+]e 1.2 mM), demonstrating this level of [Ca2+]e potently and consistently shuts down presynaptic function. The change in percentage of silencing versus change in [Ca2+]e transition between 0.8 mM to 0.4 mM is 73.3% mM–1, again highlighting the steep dependence of silencing on [Ca2+]e below the physiologic [Ca2+]e. As expected, responding terminals demonstrated Δ[Ca2+]i that scaled with [Ca2+]e (Figure 3D). Nerve terminal silencing was reversible as the proportion of silent terminals before and after switching [Ca2+]e from 2.0 mM to 0.8 mM was unchanged (Figure 3—figure supplement 2A, B).

Figure 3 with 2 supplements see all
Download asset Open asset
Silencing of Ca2+ influx into nerve terminals is potently modulated by changes in [Ca2+]e about the physiologic set point.

(A) Diagram of the experimental protocol. (B) Example traces of single nerve terminals expressing physin-GCaMP, demonstrating persistent responsiveness (1) or silencing (2) with lowering of Ca2+ below physiologic extracellular levels. Dotted line is pre-stimulus mean fluorescence (ΔF=0). (C) The percentage of silent nerve terminals. Black dots are mean, error bars are SEM, and gray dots are individual cells. (D) Summary of Δ[Ca2+]i of responding terminals. Dots are individual cells, bars are mean, and error bars are SEM. (C) and (D) were analyzed with one-way ANOVA and Tukey’s post-test for multiple comparisons, *p<0.05, **p<0.01, ***p<0.001, n=9.

The impact of [Ca2+]e on synaptic silencing is exerted through the global reduction in Ca2+ entry irrespective of VGCC subtype

Because silencing of Ca2+ influx occurred in a variable proportion of terminals in each neuron, we next investigated whether this selectivity is mediated by the relative expression of VGCC subtypes. Ca2+ influx at hippocampal nerve terminals occurs predominantly via N- and P/Q-type VGCCs, with substantial heterogeneity between neurons in their proportional contribution to Ca2+ influx (Ariel et al., 2012; Ermolyuk et al., 2013; Hoppa et al., 2012). Previously, cyclin-dependent kinase 5 (CDK5) was shown to silence a subpopulation of nerve terminals, an effect mediated by N-type channels only (Kim and Ryan, 2013). We, therefore, hypothesized that N-type channels may exhibit a greater propensity to [Ca2+]e-driven silencing as compared to P/Q-type channels, accounting for neuronal variability in the proportion of silent terminals. To examine whether [Ca2+]e-mediated silencing exhibits subtype selectivity, we utilized ω-conotoxin-GVIA and ω-agatoxin IVA, potent toxins that specifically inhibit N- and P/Q-type channels, respectively (Figure 4A). The R-type toxin, SNX-482, was employed in these experiments to exclude the minor but possibly confounding contribution of these channels to overall Ca2+ influx (Ermolyuk et al., 2013). As previously demonstrated, the contribution of presynaptic VGCC subtypes to Δ[Ca2+]i is variable between hippocampal neurons (Figure 4B and C; Ariel et al., 2012). A subset of terminals is silenced with acute inhibition of either subtype, but the isolation of N-type channels leads to greater proportions of silent terminals (Figure 4B and D). To further investigate whether a difference exists between subtypes in the likelihood of silencing, we compared Ca2+ entry in responding terminals to the proportion of silent terminals at each tested [Ca2+]e (Figure 4C and D). This analysis reveals a non-linear relationship, such that lowering of absolute Ca2+ influx steeply increases the fraction of silent terminals. The comparison of Δ[Ca2+]i to silencing was well described with a Hill equation (r2 = 0.99, Kd = 25 nM, n = –1.79) and closely matched the relationship observed without toxin application (r2 = 0.77, Kd = 23 nM, n = –1.46; Figure 4E). This analysis indicates that the proportion of silent nerve terminals is critically related to global absolute Ca2+ influx and that differences between N- and P/Q-type channels are driven primarily by the impact that acute inhibition has on Ca2+ entry.

Figure 4
Download asset Open asset
The proportion of nerve terminal silencing in neurons is related to absolute calcium (Ca2+) influx irrespective of CaV2 subtype.

(A) Diagram of the experimental protocol applied to neurons expressing physin-GCaMP. (B) The percentage of silent nerve terminals before and after toxin application to isolate a CaV2 subtype at [Ca2+]e 2 mM. (C) Summary of Δ[Ca2+]i in responding terminals after toxicologic isolation of N-type or P/Q-type channels. (D) Comparison of the effects of toxin application to isolate CaV2 subtypes on silencing (upper graph, shown on the left axis) and Ca2+ influx (lower graph, shown on the right axis). For B-D, symbols are the mean of individual neurons, lines and bars are the mean of all cells, and error bars are SEM. (B) and (C) were analyzed with a paired t-test, and D was analyzed with unpaired t-test. *p<0.05, ***p<0.001, n=6 for N-type and n=7 for P/Q-type. (E) The percentage of silent terminals plotted against Δ[Ca2+]i in responding terminals in control (black symbols) and following toxin application at varying [Ca2+]e. Symbols are mean and error bars are SEM. Data are fit with a Hill model, with the maximum at Δ[Ca2+]i = 0 constrained to 100%. (F) Relationship of the percentage of silent terminals to Δ[Ca2+]i in responding terminals with varying [Ca2+]e in neurons not treated with CaV2 toxins. Dots are the mean of individual cells from neurons presented in Figure 2. Data are fit as in E.

Changing [Ca2+]e near the physiologic set point potently controls the silencing of SV recycling and glutamate exocytosis at neurotransmitter release sites

Our observation of steep changes in the proportion of nerve terminals in which Ca2+ influx was silenced by lowering [Ca2+]e below the physiologic set point led us to investigate the impact of [Ca2+]e on neurotransmitter release. We first addressed silencing of SV recycling with pHluorin, a pH-sensitive fluorescent protein (Miesenböck et al., 1998; Sankaranarayanan et al., 2000), tagged to vGLUT1 (vGpH), which is quenched in the acidic lumen of synaptic vesicles but fluoresces with exocytosis and exposure to the relatively alkaline extracellular environment (Voglmaier et al., 2006; Figure 5A–C). We employed a robust stimulus of 200 AP (20 Hz) to distinguish responding from silent terminals (Kim and Ryan, 2013; Kim and Ryan, 2009; Figure 5A and C). Similar to synaptic Ca2+ influx, silencing of SV recycling occurred at a substantial subset of nerve terminals (~25 to 55%, Figure 5D). The proportion of silent terminals was heterogeneous across neurons but the mean and range at each [Ca2+]e were similar to the degree of silencing observed with Δ[Ca2+]i (Figure 3C). As expected, the increase in SV recycling in responding terminals scaled with [Ca2+]e (Figure 5E). Thus, as with Ca2+ influx, SV recycling is modulated in an all-or-none manner by modest changes in [Ca2+]e around the physiological set point ([Ca2+]e = 1.2 mM).

Figure 5
Download asset Open asset
Exocytosis of SVs exhibits [Ca2+]e-driven silencing.

(A) Diagram of the experimental protocol. (B) Segment of an axon expressing vGpH, showing a difference in fluorescence in nerve terminals revealed by the application of NH4Cl 50 mM. Scale bar is 10 µm, and the calibration is maximum-normalized fluorescence intensity. (C) SV recycling of individual nerve terminals indicated in B demonstrating selective silencing at [Ca2+]e 0.8 mM in (1) compared to the persistent response in (2). Traces of different [Ca2+]e are color-coded as in A. (D) Percentage of silent nerve terminals. Black dots are mean, error bars are SEM, and gray dots are individual cells. (E) SV exocytosis of responding terminals in different [Ca2+]e. Dots are individual neurons, bars are the mean, and error bars are SEM. (D) and (E) were analyzed with one-way ANOVA and Tukey’s post-test for multiple comparisons, *p<0.05, **p<0.01, ***p<0.01, n=9.

We next sought to directly assess how [Ca2+]e regulates the silencing of glutamate release in the regime of stimulation with a single AP. To do so, we utilized iGluSnFR3 v857 (iGluSnFR), a genetically-encoded, fluorescent biosensor of extracellular glutamate concentrations with an excellent signal-to-noise ratio and temporal resolution enabling quantification of glutamate release from one AP at individual nerve terminals (Aggarwal et al., 2022). Co-expression of mRuby-synapsin allowed measurements to be localized to neurotransmitter release sites (Figure 6B); however, because iGluSnFR is anchored by glycophosphatidylinositol (GPI) to the cell surface, the indicator does not necessarily discriminate between glutamate released by the transfected cell or neighboring terminals from non-transfected neurons (Aggarwal et al., 2022). Accordingly, while fluorescent signals at mRuby-synapsin puncta may receive contributions from additional terminals, silent terminals are definitively assigned as inactive. With this approach, we found glutamate exocytosis evoked by a single AP exhibited higher proportions of silencing at neurotransmitter release sites as [Ca2+]e was lowered, again demonstrating a marked increase in silencing as the concentration decreased below the physiologic set point (Figure 6C and D). Thus, using these two fluorescent biosensors quantifying neurotransmitter handling under different stimulation regimes demonstrates that [Ca2+]e is an important regulator of presynaptic function, differentially dictating across synapses and neurons whether neurotransmitter will be released.

Figure 6
Download asset Open asset
Glutamate release elicited by 1 AP demonstrates [Ca2+]e-driven silencing at neurotransmitter release sites.

(A) Diagram of the experimental protocol. Responses of iGluSnFR-expressing neurons were averaged over 15 single AP trials delivered in ~20 s intervals. (B) Difference image (peak minus baseline fluorescence) of an axon expressing iGluSnFR (cyan, upper) stimulated with 1 AP in [Ca2+]e 1.2 mM, with terminals marked by mRuby-synapsin (magenta, middle; merge lower). Scale bar 20 µm. (B) Traces of iGluSnFR measured at nerve terminals indicated in B, illustrating differential silencing as [Ca2+]e is lowered. Traces of different [Ca2+]e are color-coded as in A. (D) Percentage of silent nerve terminals. Black dots are mean, error bars are SEM, and gray dots are individual cells. (E) Glutamate release, quantified as ΔF/Fmax, in responding terminals as [Ca2+]e is decreased. Dots are the mean of individual cells, bars are mean, and error bars ± SEM. (D) and (E) were analyzed with one-way ANOVA and Tukey’s post-test for multiple comparisons. *p<0.05, **p<0.01, ***p<0.001, n=8.

GPCRs that lower synaptic Ca2+ currents similarly modulate the proportion of silent nerve terminals

The selective silencing of Ca2+ entry and neurotransmitter release in subpopulations of nerve terminals by globally reducing Ca2+ influx may be an important mechanism determining the behavior of synapses to biological processes affecting Ca2+ currents (ICa). For instance, a subset of important presynaptic GPCRs affects neurotransmission by lowering synaptic Ca2+ influx (Dolphin, 2003). One such canonical mechanism is the agonism of the GABAB receptor (GABABR), which lowers ICa by approximately 50% via interactions of the Gβγ subunit with presynaptic VGCCs (Mintz and Bean, 1993; Herlitze et al., 1996; Isaacson, 1998; Laviv et al., 2011). Thus, we hypothesized that GABABR-mediated decreases in ICa will significantly increase the degree of nerve terminal silencing at the physiologic set point of [Ca2+]e. We compared the effect of baclofen, a GABABR agonist, on the silencing of Ca2+ influx and SV recycling using physin-GCaMP and vGpH, respectively (Figure 7A–C). Consistent with our proposed mechanism, agonizing GABABR at [Ca2+]e 1.2 mM caused an additional ~20% of terminals to become silent in both Ca2+ influx and SV recycling (Figure 7D), similar to the proportional increase we observed decreasing [Ca2+]e to 0.8 mM. However, agonism with baclofen in [Ca2+]e 2.0 mM led to a substantially lower increase in the proportion of silencing (7.6 ± 3.4% vs 20.5 ± 4.8%, p=0.017; Figure 7—figure supplement 1A, B). Thus, GPCR-mediated silencing was dependent on [Ca2+]e such that the impact was far greater operating at the physiologic [Ca2+]e. These results confirm that GPCR-mediated modulation acting presynaptically to decrease ICa can silence a substantial proportion of nerve terminals.

Figure 7 with 1 supplement see all
Download asset Open asset
Agonism of GABABR with baclofen leads to silencing of calcium (Ca2+) influx and SV exocytosis.

(A) Diagram of the experimental protocol. (B) Representative traces of silencing of Ca2+ influx and SV exocytosis resulting from agonism of GABABR with baclofen in single nerve terminals expressing physin-GCaMP or vGpH, respectively. (C) Δ[Ca2+]i and ΔF/ΔFNH4Cl before and following the application of baclofen in responding terminals. (D) The percentage of silent terminals before and following treatment with baclofen was measured in physin-GCaMP and vGpH-expressing neurons (see labels above). For (C) and (D), dots are individual neurons, bars are mean of all neurons, and error bars are SEM. (C) and (D) were analyzed with a paired t-test, *p<0.05, **p<0.01, n=8 for physin-GCaMP, and n=9 for vGpH.

Discussion

The flux of Ca2+ in neurons plays a central role in neurotransmission by triggering the exocytosis of SVs, thereby releasing neurotransmitters into the synaptic cleft (Dittman and Ryan, 2019). Uncovering the mechanisms regulating Ca2+ entry into nerve terminals is therefore critical to understanding the molecular underpinnings of CNS function. The relationship between Ca2+ influx and the exocytotic rate is highly cooperative (Heidelberger et al., 1994; Schneggenburger and Neher, 2000; Bollmann et al., 2000; Ariel and Ryan, 2010), highlighting the amplification of neurotransmitter release caused by changes in Δ[Ca2+]i. Fewer studies have sought to determine the effect of [Ca2+]e on Δ[Ca2+]i and, to our knowledge, none have examined this in the range encompassing the physiological set point for [Ca2+]e at physiological temperature. In general, the narrowing of the AP at warmer temperatures (Hodgkin and Katz, 1949; Yu et al., 2012) results in lowering of Ca2+ influx during AP firing, so additionally studying lower [Ca2+]e results in a much lower regime of Δ[Ca2+]i. Previous attempts (carried out at sub-physiological temperatures) modeled this relationship with Michaelis-Menten kinetics, where Δ[Ca2+]i obeyed a simple saturation of VGCCs, but predicted a simple 1:1 relationship extrapolating to lower [Ca2+]e (Ariel and Ryan, 2010; Schneggenburger et al., 1999).

Here, we show an unexpected and impactful behavior of Ca2+ influx in the lower Δ[Ca2+]i regime, which is that small modulation of Ca2+ influx leads to dramatic silencing of individual nerve terminals. We discovered that a substantial subset of terminals (~40%) exhibited silencing of Ca2+ influx and SV exocytosis at physiologic [Ca2+]e, and this proportion increased steeply as [Ca2+]e was lowered to 0.8 mM and 0.4 mM (to ~60% and >90%, respectively). In comparison, the change in the proportion of silent terminals in the transition from [Ca2+]e 2.0 mM to 1.2 mM, a difference of ~15%, was smaller despite a larger step in [Ca2+]e. Previous estimates of synaptic silencing were drawn from experiments conducted at [Ca2+]e 2.0 mM and agree with our findings at this concentration (Altrock et al., 2003; Kim and Ryan, 2013; Moulder et al., 2004; Moulder et al., 2006). However, our results highlight that the proportion of silent terminals in cultured neurons is substantially higher under conditions of physiologic [Ca2+]e. Thus, [Ca2+]e is a crucial variable to consider in future studies examining synaptic silencing. The agreement of our findings in the silencing of both Δ[Ca2+]i and SV exocytosis indicate that silencing of Δ[Ca2+]i is driving the effect as opposed to an independent mechanism operating on SV exocytosis alone.

The proportion of silencing across different neurons was steeply related to absolute Δ[Ca2+]i, such that lowering of Δ[Ca2+]i was associated with increased proportions of silent terminals. Interestingly, this relationship persisted despite the selective blockade of VGCC subtypes by potent toxins, demonstrating the propensity for silencing is not related to their relative distribution at nerve terminals. Due to this effect, our analysis suggests that the minimal [Ca2+]e needed to sustain any neurotransmission is ~0.4–0.5 mM. Moreover, the degree of silencing caused by acutely blocking VGCC subtypes followed this relationship. Taken together, our results suggest the operation of a feedback mechanism causing the selective shutdown of terminals not meeting a threshold for Ca2+ entry. This mechanism has important implications for the impact of neuromodulators that regulate ICa because relatively small changes in Δ[Ca2+]i could cause a substantial proportion of terminals to become silent (Dolphin, 2003). As an example, we investigated the impact on synaptic function resulting from the inhibition of presynaptic VGCCs by Gβγ subunits using agonism of GABABR by baclofen. Previous studies have demonstrated that the Gβγ interacts with presynaptic VGCCs to lower ICa (Isaacson, 1998; Laviv et al., 2011; Kajikawa et al., 2001). We observed that, in responding terminals, baclofen was associated with ~40% decrease in Δ[Ca2+]i. This decrement in Ca2+ influx led to an increase in the proportion of silent terminals from ~40 to 60%. Notably, this difference in the degree of silencing and Δ[Ca2+]i agreed excellently with results from decreasing [Ca2+]e 1.2–0.8 mM, supporting the conclusion that the effect of baclofen is mediated by decreasing presynaptic Ca2+ influx. In addition, baclofen caused substantially fewer terminals to become silent in [Ca2+]e 2.0 mM, highlighting the importance of physiologic [Ca2+]e in dictating synaptic function.

While our findings have shown that [Ca2+]e contributes importantly to presynaptic neuromodulation, we have not identified the precise molecular mechanism. Previous investigations indicate that silencing is a drastic form of synaptic homeostasis, as chronic suppression of neuronal activity with tetrodotoxin (TTX) decreases the proportion of silent terminals (Moulder et al., 2006) and increasing activity with chronic depolarization increases this proportion (Moulder et al., 2004; Moulder et al., 2006). Several intracellular signaling pathways have been identified to set the fraction of silent terminals, including those involving adenylyl cyclase (Moulder et al., 2008), calcium/calmodulin-dependent protein kinase II (Ninan and Arancio, 2004), and calcineurin/cyclin-dependent kinase 5 (Kim and Ryan, 2010; Kim and Ryan, 2013). The disparate molecular pathways that converge on the regulation of synaptic silencing suggest that it is an important mode of plasticity for neurons in the CNS. Notably, in experiments investigating the impact of GPCR modulation, only a subset of GPCRs tested was able to alter silencing (Crawford et al., 2011). The silencing caused by GABABR agonism was dependent on proteasome function, and the authors concluded that protein degradation was necessary (Crawford et al., 2011). However, these experiments utilized 4 hr incubation with baclofen, as opposed to 5 min here, so an acute effect of GABABR agonism independent of proteasomal activity may have been missed. Indeed, silencing may have both proteasome-dependent and independent pathways, operating on slow (hours) or acute (<1 hr) time scales, respectively (Crawford et al., 2012). Moreover, pharmacologic inhibition of proteasome function prevents depolarization-induced silencing, so the effect of proteasome inhibition may not be specific to GABABR agonism (Jiang et al., 2010). Thus, our results suggest a separate, rapid mechanism by which nerve terminals can be silenced, adding to the diversity of pathways by which this form of plasticity is achieved.

We classified terminals for which the ΔF of Ca2+ entry falls below one standard deviation of pre-stimulus fluorescence as silent. This previously used definition agrees well with the other approaches used (Moulder et al., 2008; Moulder et al., 2004; Moulder et al., 2006; Crawford et al., 2011; Crawford et al., 2012; Hogins et al., 2011). Given that Ca2+ influx is itself non-linearly related to SV exocytosis, even if Ca2+ entry has only been inhibited to undetectable levels, this level would lead to even rarer neurotransmitter release. The sub-proportional relationship of Δ[Ca2+]i to [Ca2+]e and the heterogeneous changes of individual nerve terminals to changes in [Ca2+]e refutes the simple biophysical model of the relationship of Ca2+ influx to [Ca2+]e. Rather, our results support nerve terminal specificity in the propensity to have presynaptic function diminished by perturbations that lower the driving force for Ca2+ entry.

Our experiments have also not assessed whether silencing occurs independently at individual release sites. The sensor iGluSnFr3 can resolve spontaneous glutamate exocytosis that represents a single vesicle and, therefore, the output of a single release site (Aggarwal et al., 2022). However, quantifying the proportion of release sites that are silenced is limited by the variable number of active zones per nerve terminal in a population of CA1-CA3 hippocampal neurons (Schikorski and Stevens, 1997; Rigby et al., 2022). Moreover, iGluSnFr can detect glutamate released from neighboring but non-transfected neurons (Aggarwal et al., 2022). Thus, it would be difficult with our approach to unambiguously identify a release site and, therefore, to quantify the proportion of sites silenced. We speculate that because Ca2+ influx at the terminal is prevented, all active zones within a terminal are equally impacted by [Ca2+]e mediated silencing.

Our findings highlight that the physiologic set point of [Ca2+]e in the CNS functions as an important lever in the modulation of synaptic function because further lowering of Δ[Ca2+]i substantially increases the proportion of silent terminals. Although [Ca2+] of the CSF is tightly regulated globally (Forsberg et al., 2019; Jones and Keep, 1988; Barkai and Meltzer, 1982), models of the synapse that account for its limited volume and restriction of diffusion have predicted that drastic reductions of local [Ca2+]e may occur during repetitive presynaptic firing (Egelman and Montague, 1999; King et al., 2001). Thus, the phenomenon we observed may be important in vivo as a homeostatic mechanism to prevent excessive or runaway synaptic activity (Stringer et al., 2007). Our results suggest that the magnitude of the decrease in [Ca2+]e does not need to be large to shut down a sizable proportion of terminals.

In summary, we have utilized highly sensitive, genetically-encoded fluorescent biosensors to dissect the effects of [Ca2+]e near the physiologic set point on presynaptic function. We discovered that [Ca2+]e has a potent effect in setting the proportion of silent terminals which is driven by Δ[Ca2+]i. These findings provide evidence that [Ca2+]e is an important lever contributing to neuromodulation. Future studies will address the intracellular molecular targets responsible for this acute mechanism of synaptic silencing.

Materials and methods

Animals

All experiments involving animals were performed in accordance with protocols approved by the Weill Cornell Medicine Institutional Animal Care and Use Committee. Neurons were derived from Sprague-Dawley rats (Charles River Laboratories strain code: 001, RRID: RGD_734476) of either sex on postnatal days 0–2.

Neuronal culture

Request a detailed protocol

Primary neuronal cultures were prepared as previously described (Ryan, 1999). Hippocampal CA1 to CA3 regions were dissected, dissociated, and plated onto poly-L-ornithine-coated coverslips. Plating media consisted of the minimal essential medium, 0.5% glucose, insulin (0.024 g/l), transferrin (0.1 µg/l), GlutaMAX 1%, N-21 (2%), and fetal bovine serum (10%). After 1–3 days in vitro (DIV), cells were fed and maintained in media with the following modifications: cytosine β-D-arabinofuranoside (4 µM) and FBS 5%. Cultures were incubated at 37 °C in a 95% air/5% CO2 incubator. Calcium phosphate-mediated gene transfer was performed on DIV 6–8, and neurons were used for experiments on DIV 14–21.

Plasmid constructs

Request a detailed protocol

The following published DNA constructs were used: VGLUT1-pHluorin (vGpH) (Voglmaier et al., 2006), synaptophysin-GCaMP6f (physin-GCaMP) (de Juan-Sanz et al., 2017), cytosolic GCaMP6f (Chen et al., 2013), and GPI iGluSnFR3 v857(iGluSnFR) (Aggarwal et al., 2022), which was a gift from Kasper Podgorski. mRuby3-synapsin1a (Addgene plasmid #187896) was generated by removing GFP from GFP-synapsin (Chi et al., 2001) using restriction sites AgeI and BGIII, and substituting it in frame with mRuby obtained from pKanCMV-mRuby3-18aa-actin, which was a gift from Michael Lin (Addgene plasmid #74255). Cytosolic GCaMP6f under the CaMKII promoter was generated by cloning GCaMP6f (Chen et al., 2013) (Addgene plasmid #40755) into a CaMKII promoter vector (Chow et al., 2010) (Addgene plasmid #22217) as previously described (de Juan-Sanz et al., 2017).

Live-cell imaging

Request a detailed protocol

Coverslips were loaded onto a custom chamber and perfused at 100 µl min–1 via a syringe pump (Fusion 4000, Chemyx) with Tyrode’s solution containing (in mM, except if noted otherwise): NaCl 119, KCl 2.5, glucose 30, HEPES 25, CaCl2 0.8–2.0 with MgCl2 adjusted to maintain divalence of 4, D, L-2-amino-5-phosphonovaleric acid (APV) 50 µM, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) 10 µM, adjusted to pH 7.4. Temperature was maintained at 37 °C with a custom-built objective heater under feedback control (Minco). Fluorescence was stimulated with OBIS 488 nm LX or OBIS 561 nm LS lasers (Coherent) passing through a laser speckle reducer (LSR 3005 at 12° diffusion angle, Optotune). Live-cell imaging was performed with a custom-built, epifluorescence microscope. Emission was acquired with a 40 x, 1.3 numerical aperture objective (Fluar, Zeiss) and Andor iXon + Ultra 897 electron-multiplying charge-coupled device camera. APs were evoked with platinum-iridium electrodes generating 1 ms pulses of 10 V cm–1 field potentials via a current stimulus isolator (A385, World Precision Instruments). For physin-GCaMP and vGpH, neurons were stimulated with 20 and 200 AP, respectively, delivered at 20 Hz. For iGluSnFR, neurons were stimulated with a single AP, with averaging performed over 15 trials delivered every 20 s. A custom-designed Arduino board coordinated AP and laser stimulation with frame acquisition. Frame rates for AP recordings with vGpH, GCaMP6f, and iGluSnFR were 5, 50, and 100 Hz, respectively. To achieve frame rates of 50 and 100 Hz, a subregion of the EMCCD chip was used (347 pixels and 169 pixels, respectively, compared to 512 pixels). In experiments with neurons expressing vGpH, Tyrode’s solution with NH4Cl 50 mM replacing an equimolar concentration of NaCl was perfused at the end of the experiment to alkalinize intra-vesicular pHluorin molecules (Sankaranarayanan et al., 2000) and enable normalization of fluorescence to the total pool of internal vesicles (Kim and Ryan, 2010). Because vGpH has low surface accumulation and fluorescence is therefore largely quenched before AP stimulation (Balaji and Ryan, 2007), a brief (<1 min) exposure to NH4Cl was performed before experiments to identify transfected neurons and then washed out for at least 5 min. In experiments with physin-GCaMP, the chamber perfusate was exchanged for Tyrode’s solution with Ca2+ 4 mM, pH 6.9, and ionomycin 500 µM to saturate the fluorophore and allow conversion of fluorescence to absolute [Ca2+]i (de Juan-Sanz et al., 2017). In experiments with iGluSnFR, glutamate 100 mM was applied to saturate the sensor and normalize to maximum fluorescence (Aggarwal et al., 2022). Recordings of physin-GCaMP6f and vGpH included ~200 nerve terminals on average while recordings of iGluSnFR included ~50 nerve terminals due to the limited EMCCD dimensions at this frame rate.

Key pharmacological reagents

Request a detailed protocol

Ionomycin, ω-conotoxin GVIA, ω-agatoxin IVA, and SNX-482 were purchased from Alomone labs (catalog numbers I-700, C-300, STA-500, and RTS-500 respectively). The toxins were applied for 3 min in Tyrode’s solution at concentrations of 1 µM, 400 nM, and 500 nM for ω-conotoxin GVIA, ω-agatoxin IVA, and SNX-482, respectively (Ariel et al., 2012; Ermolyuk et al., 2013; Hoppa et al., 2012). Baclofen was purchased from Sigma-Aldrich (catalog number B 5399) and continuously perfused at 10 µM beginning 5 min before AP stimulation (Laviv et al., 2011).

Image analysis

Request a detailed protocol

Images were analyzed with ImageJ (Rasband, 1997). Circular regions of interest (ROIs) were semi-automatically placed around puncta corresponding to nerve terminals using a custom-written macro. Fluorescence within ROIs was background corrected by subtraction of signal from adjacent, non-synaptic ROIs using a custom-written macro. Images of individual 1 AP trials of iGluSnFr were averaged using a custom-written macro.

Experimental design and statistical analyses

Request a detailed protocol

Quantification of peak fluorescence responses and the percentage of silent terminals was performed with Excel. Terminals were classified as silent if the peak (ΔF) above the baseline (mean of 49 frames) was less than the standard deviation of the baseline (σbaseline, 49 frames) (Kim and Ryan, 2010). For physin-GCaMP and vGpH, the peak was calculated as the mean of five frames at the end of the AP train. For iGluSnFR, the peak was calculated as the mean of 3 frames, with the first frame in the range coinciding with the AP. For vGpH, fluorescence was normalized to the total internal pool of vesicles following perfusion of NH4Cl (Kim and Ryan, 2010). For iGluSnFR, fluorescence was normalized to the maximum following perfusion of a saturating concentration of glutamate (Aggarwal et al., 2022). For physin-GCaMP, fluorescence was converted to absolute [Ca2+]i using (de Juan-Sanz et al., 2017):

[Ca2+]e=KdFFmax-1Rf1-FFmax1n

where Fmax is the peak following saturation of GCaMP6f with ionomycin and Kd (0.38 µM), Rf (51.8), and n (2.3) are the in vitro dissociation constant, dynamic range, and Hill coefficient, respectively (Chen et al., 2013). Statistical analysis was performed with Prism version 9. Comparisons were made with the t-test except for Figure 7—figure supplement 1B which was analyzed with the Mann-Whitney U test. For more than two groups, ANOVA with Tukey’s post-test for multiple comparisons was used. Statistical significance was defined as p<0.05. The relationship of Δ[Ca2+]i in responding terminals to the proportion of silent terminals was fit with a Hill equation using least squares regression and constraining the maximum silencing at Δ[Ca2+]i = 0 to 100% (Figure 4E and F). Data are presented as mean ± SEM except in Figure 1D and Figure 1—figure supplement 1B, which demonstrates a mean ± 95% confidence interval to illustrate the experimental difference from the theoretical value.

Materials availability

Request a detailed protocol

Fluorescent biosensors are available through Addgene (iGluSnFR plasmid #187896, mRuby3-synapsin1a plasmid #74255) or from the corresponding author by request (physin-GCaMP, CKII cytosolic GCaMP6f, vGpH). Custom-written macros for ImageJ analysis are available through GitHub (Cook and Ryan, 2022). Supporting data is available from Dryad (doi:10.5061/dryad.1zcrjdfw0).

Data availability

All data generated or analyzed during this study is included in the manuscript and supporting file; Source Data has been uploaded onto Dryad (doi:10.5061/dryad.1zcrjdfw0) and customized code has been uploaded to Github (https://github.com/taryan2020/ImageJ, (Cook and Ryan, 2022 copy archived at swh:1:rev:a33d53ca21ac4bd2fbe1afc52eb9162b683ce0a0)).

The following data sets were generated

References

  45. Software
    1. Rasband WS
    (1997) ImageJ, version 0.1
    National Institutes of Health.
    https://imagej.nih.gov/ij/

Decision letter

  1. Graeme W Davis
    Reviewing Editor; University of California, San Francisco, United States
  2. John R Huguenard
    Senior Editor; Stanford University School of Medicine, United States
  3. Graeme W Davis
    Reviewer; University of California, San Francisco, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Neuromodulatory silencing of nerve terminals" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Graeme W Davis as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by John Huguenard as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

The essential revisions are representative of conversation among reviewers, attempting to synthesize the individual reviews.

1. Two reviewers focused on a specific issue: The obvious challenge is distinguishing between a very small response and a non-responder. The data, throughout the paper, are categorized (yes/no) and presented as a percentage. Thus, the crux of the manuscript relies upon thoroughly demonstrating that the discrimination between a responder and non-responder is not confounded. Suggestions for how this might be achieved are presented, but these are not intended to be the only solution to addressing this issue – it is entirely up to the authors.

2. The data set could be broadly strengthened by including data from individual release sites, rather than simply presenting averages. This is particularly the case with the GluSnFr data where the probabilistic release of single vesicles could be resolved, inclusive of amplitudes associated with spontaneous release events (when possible).

3. Two reviewers highlight a circular argument regarding the classification of silent sites, that are then shown to be silent. This should be addressed either through changes to the figures or the text. Reviewer 3 (comment 3). Reviewer 2 (final comment).

4. It would be important, in at least one experimental paradigm, to see that release returns to baseline following restoration of external calcium.

5. The issue of neuronal excitation can and should be evaluated. It is understood that these are sparse cultures. None-the-less, it remains possible that fields of view contain axons originating in more than one neuron, or in the case that they reside on different branches of the same axon, that branch point failures could 'cause' boutons to appear to function differently. A solution would be to systematically track whether sets of ROI unambiguously residing on single branches still function differently with respect to release (some being silenced and others not a lower calcium).'

6. A core feature of the manuscript is the quantification of calcium entry via the genetically encoded calcium sensor. There was a healthy debate among reviewers regarding the necessity of utilizing synthetic dyes. On the one hand, the superior properties of the dyes allow for highly quantitative validation of existing transgenic reporters and central conclusions. On the other hand, this requires loading (likely through patch pipettes). Addressing the robustness of the existing measures, either through dyes or validation with additional variants of transgenic reporters would strengthen the work. Again, however, the precise nature of the experiment is entirely up to the authors.

7. Two reviewers, while agreeing that the GABA data are interesting, commented that the data are somewhat under-developed. On the one hand, the title emphasizes "neuromodulation" and this could be modified to better reflect the single experiment shown regarding GABA. At another level, reviewer #3 argues that "The interesting issue would be to reveal how the effect of GABAB activation depends on the [Ca2+]e. This information is essential to see whether there is indeed a shoulder in its effectiveness curve." The title likely should be changed, at a minimum.

8. There was discussion about the main conclusion of the manuscript. 'Thus, as with Ca2+ influx, SV recycling is modulated in an all-or-none manner by modest changes in [Ca2+]e around the physiological set point.' It seems equally likely that authors have identified a steeper continuum rather than 'all or none' control. As this is a central tenant of the manuscript, it could perhaps be addressed through changes in the text to shore up the argument, differently emphasized through data presentation or through the addition of new data.

Reviewer #1 (Recommendations for the authors):

While fascinating, the conclusions are most powerful when the data can be distilled to direct observation of single release sites. Yet throughout, the data are presented as all-or-none discriminations at regions of interest (ROI) that may or may not represent single release sites. Specific criticisms are listed below.

1. One obvious challenge is distinguishing between a very small response and a non-responder. The data, throughout the paper, are categorized (yes/no) and presented as a percentage. Thus, the crux of the manuscript relies upon thoroughly demonstrating that the discrimination between a responder and non-responder is not confounded. For example, in several figures, nice example traces are provided showing a large initial response that results in a failure upon lowering extracellular calcium. But, beyond these examples, there are no quantitative data demonstrating that this is, indeed, routinely the case. Could the authors bin their data into quartiles based on the amplitude of the initial GCamp signal at 2.0 external calcium. Then, for each quartile, calculate the fractional silencing that occurs. If the fraction of silencing is constant across quartiles, then the authors have a good argument that the method of detection (GCamp) is not contributing to the observed effects.

2. Figure 6 uses GluSnFr to directly measure vesicle release, testing synapse silencing. Data are presented as the average of 15 single action potentials. Given that release is probabilistic, a silent synapse is a zero responder over some number of these trials, but we do not know how many. Two issues arise. First, individual successes could be averaged out to appear as a failure in the averaged trace. Second, a true release probability depends on the number of active zones that are present at a single R01. It seems that the true power of GluSnFr is not being deployed. Can the authors utilize spontaneous miniature amplitudes (or sucrose evoked miniatures) to achieve an estimate of single vesicle release event amplitude. Then, if sufficiently above the noise, define the release probability at high and low calcium.

3. In general, data from single RO1 would be nice to see, as opposed to averages and percentages per culture dish.

4. It would be important, in at least one experimental paradigm, to see that release returns to baseline following restoration of external calcium.

Reviewer #2 (Recommendations for the authors):

The authors illustrate two terminals in which the [Ca2+]e around 0.8 mM results in zero Ca2+ entry. They also show that at higher [Ca2+]e the terminal has detectable Ca2+ entry. In the following figure, they demonstrate that approximately 25% of the boutons are 'silent' even in 2 mM [Ca2+]e. What does this mean? Is there any evidence that they are functional boutons? This is inconsistent with the literature. In cortical pyramidal neuron axons, each AP results in detectable Ca2+ entry and the Ca2+ entry is 100% reliable from AP to AP. If there is no Ca2+ entry, it is always the consequence of the lack of an active zone (i.e. not functional bouton).

Keeping the superior properties of the synthetic dyes in mind, all [Ca2+] measurements should be repeated with high affinity synthetic dyes.

An interesting finding of the MS is that N and P/Q type Ca channels have no special role in Ca entry and in creating silent synapses. This should be emphasized.

Reviewer #3 (Recommendations for the authors):

1) Membrane excitability – Two observations are relevant here: The qualitative preservation of the phenomenon when two types of voltage gated calcium channels are blocked separately, and the large heterogeneity of the % of silenced boutons among neurons at a given extracellular calcium concentrations, which is at least as great as the range of modulation of the % of silenced synapses by extracellular calcium concentrations at single neurons. One then wonders if the findings might be attributed to a) the fidelity of the field potential-based stimulation system, that is, the degree to which neurons track the stimuli trains; b) the heterogeneity of neurons in this regard, c) this fidelity at different extracellular calcium concentrations for different neurons, and d) the identity of presynaptic sites analyzed in one run (are they all part of the same axon?). Along these lines, there is an assumption that the field potential-based stimulation system is the sole driver of excitation in these networks, which is reasonable given that excitatory synaptic transmission is mostly blocked pharmacologically (by CNQX and APV). Inhibitory transmission, however, was not blocked and thus, there is no guarantee that the inhibitory input neurons receive and its modulation by extracellular calcium does affect the degree to which neurons fire precisely and reliably at 20 Hz at all conditions. If it could be shown, at least for a substantial subset of the data, that all terminals analyzed for a particular neuron are part of an unambiguously identified axon stretch, with no branches (potential conduction failure points) and still demonstrate the claimed heterogeneity, this potential confound would be less of an issue.

2) It would seem that the x-axis intercept in Figure 1D (0.47 mM) predicts that at this concentration of extracellular calcium, pretty much all presynaptic terminals will be silent. Why not test this directly? This would nicely strengthen the findings.

3) Page 4, line 23. For the silenced population it is indeed expected, because, by definition, they did not respond and were consequently classified as silenced. As this is circular, the inclusion of these data in Figure 3D (and 6E) is a bit confusing and should probably be removed.

4) Figure 4 – end of legend – Data are fit as in D (should probably be E)

https://doi.org/10.7554/eLife.83530.sa1

Author response

The essential revisions are representative of conversation among reviewers, attempting to synthesize the individual reviews.

1. Two reviewers focused on a specific issue: The obvious challenge is distinguishing between a very small response and a non-responder. The data, throughout the paper, are categorized (yes/no) and presented as a percentage. Thus, the crux of the manuscript relies upon thoroughly demonstrating that the discrimination between a responder and non-responder is not confounded. Suggestions for how this might be achieved are presented, but these are not intended to be the only solution to addressing this issue – it is entirely up to the authors.

We thank the reviewers for raising this excellent point regarding the challenge of identifying nonresponding nerve terminals. To address this issue, we compared the frequency distributions of fluorescence responses (ΔF) across all nerve terminals separated into silent or responding populations (Figure 2 —figure supplement 1). This analysis reveals two key aspects: (1) the distribution shape of ΔF signals in the silent population is unchanged as [Ca2+]e varies from 0.8 to 2.0 mM, while the distribution of responders shifts, as expected, to higher ΔF with increasing [Ca2+]e. This indicates that our cutoff criterion for defining silencing was not artificially being biased by the absolute size of the signal at these different concentrations. (2) We observe a marked, clear separation in these frequency distribution of the silent versus responding populations reinforcing that our cutoff criterium yields distinct populations. Last, our results for silencing at [Ca2+]e 2 mM are consistent with previous publications using the same cutoff criterium or other definitions of presynaptic silencing, bolstering our observations and quantification in this manuscript. Conversely, for the first time to our knowledge, we have extended this methodology to investigate silencing at near-physiologic [Ca2+]e, discovering substantially different proportions of silent terminals within a physiologic range.

2. The data set could be broadly strengthened by including data from individual release sites, rather than simply presenting averages. This is particularly the case with the GluSnFr data where the probabilistic release of single vesicles could be resolved, inclusive of amplitudes associated with spontaneous release events (when possible).

We agree that extending our analysis to the level of single release sites would be interesting to establish whether they act independently within nerve terminals. With our approach, analysis of individual release sites is complicated by the possible activation of iGluSnFr by glutamate exocytosed from neighboring, non-transfected terminals and the variable number of release sites per terminal in CA1-CA3 hippocampal neurons. Thus, we would be limited in our ability to discriminate between neighboring release sites and characterize the proportion of silent release sites between different [Ca2+]e concentrations. We have added this consideration as a limitation in the discussion of the manuscript (page 12, lines 7-15). However, the abolishment of Ca2+ influx at terminals despite a robust train of action potentials suggests that all release sites within a terminal are simultaneously silenced by this mechanism. Furthermore, the central importance of nerve terminals to neurotransmission, the ability to readily identify them through various targeted fluorescent proteins, and the widespread use of this marker in other studies of silencing underscore its biologic and experimental relevance.

3. Two reviewers highlight a circular argument regarding the classification of silent sites, that are then shown to be silent. This should be addressed either through changes to the figures or the text. Reviewer 3 (comment 3). Reviewer 2 (final comment).

As suggested by the reviewers, figures 3, 5, and 6 have been updated to eliminate the silent bars in these graphs.

4. It would be important, in at least one experimental paradigm, to see that release returns to baseline following restoration of external calcium.

We thank the reviewers for recommending inclusion of this important control experiment. We quantified the percentage of silent terminals and Ca2+ influx in neurons that were stimulated initially in [Ca2+]e 2.0 mM, then 0.8 mM, and finally returned to 2.0 mM. We observed that both the proportion of silent terminals and the net influx of Ca2+ recovered to their initial values as [Ca2+]e was reverted to its starting level (Figure 3 —figure supplement 2). Importantly, these results demonstrate that presynaptic silencing is a reversible process that is under the control of [a2+]e.

5. The issue of neuronal excitation can and should be evaluated. It is understood that these are sparse cultures. None-the-less, it remains possible that fields of view contain axons originating in more than one neuron, or in the case that they reside on different branches of the same axon, that branch point failures could 'cause' boutons to appear to function differently. A solution would be to systematically track whether sets of ROI unambiguously residing on single branches still function differently with respect to release (some being silenced and others not a lower calcium).'

We thank the reviewers for this suggestion to exclude the possibility that branch point failures are responsible for nerve terminal silencing. We examined this possibility with neurons expressing GCaMP6f in the cytosol to provide a good morphological fill of the axonal volume and co-expressing mRuby-synapsin to mark terminals. We performed the analysis for [Ca2+]e 0.8 mM, the concentration we hypothesized would most likely cause branch point failure. To provide robust quantification, we limited our analysis to branches with at least 10 nerve terminals. This analysis demonstrates silent and responding terminals on the same branches of an axon, thereby excluding the possibility that these effects are driven by selective changes in excitability (Figure 2 —figure supplement 2). The overall percentage of silent terminals, though, is similar in this population of branches compared to the results of cell-wide percentages.

6. A core feature of the manuscript is the quantification of calcium entry via the genetically encoded calcium sensor. There was a healthy debate among reviewers regarding the necessity of utilizing synthetic dyes. On the one hand, the superior properties of the dyes allow for highly quantitative validation of existing transgenic reporters and central conclusions. On the other hand, this requires loading (likely through patch pipettes). Addressing the robustness of the existing measures, either through dyes or validation with additional variants of transgenic reporters would strengthen the work. Again, however, the precise nature of the experiment is entirely up to the authors.

We are grateful to the reviewers for highlighting this concern. Previous studies investigating presynaptic silencing (Kim and Ryan, 2013) utilized a less sensitive genetically-encoded calcium indicator, GCaMP3, and compared these measurements to the low-affinity, synthetic calcium dye, Magnesium Green. There was excellent agreement between the dye and GCaMP3, arguing against a limitation of GCaMP in detecting responses. Given the superiority of GCaMP6f to GCaMP3, we would expect a lower proportion of terminals classified as silent in our current experiments compared to the prior publication. However, we found a modest increase in the percentage of silent terminals (~15-20% compared to ~20-30%). We attribute the higher proportion of silent terminals in [Ca2+]e 2.0 mM in the current studies to performing experiments at physiologic temperature as opposed to 30°C, which narrows the action potential waveform and thereby reduces Ca2+ influx. However, the overall similarity in our current results with this publication argues against a limitation of GCaMP as responsible for the silencing ofa2+ entry we observed.

7. Two reviewers, while agreeing that the GABA data are interesting, commented that the data are somewhat under-developed. On the one hand, the title emphasizes "neuromodulation" and this could be modified to better reflect the single experiment shown regarding GABA. At another level, reviewer #3 argues that "The interesting issue would be to reveal how the effect of GABAB activation depends on the [Ca2+]e. This information is essential to see whether there is indeed a shoulder in its effectiveness curve." The title likely should be changed, at a minimum.

We thank the reviewers for suggesting how to strengthen our conclusions of the impact of [Ca2+]e on silencing mediated by the GABAB receptor, a GPCR that causes reduced Ca2+ entry via interaction of the Gβγ subunit with voltage-gated Ca2+ channels. As suggested, we investigated the effect of GABABR agonism with baclofen on silencing at [Ca2+]e 2.0 mM to compare to our previous results at 1.2 mM. Interestingly, the net change in percentage of silent terminals is substantially lower at 2.0 mM compared to 1.2 mM (7.6 ± 3.4% vs 20.5 ± 4.8%, respectively; Figure 7 —figure supplement 1). These results demonstrate that [Ca2+]e substantially influences the synaptic response to GABABR agonism, a GPCR that reduces Δ[Ca2+]i through a canonical G-protein subunit pathway.

Additionally, throughout the text and in the title, we have replaced the term “neuromodulator” to “GPCR” or “GABABR.”

8. There was discussion about the main conclusion of the manuscript. 'Thus, as with Ca2+ influx, SV recycling is modulated in an all-or-none manner by modest changes in [Ca2+]e around the physiological set point.' It seems equally likely that authors have identified a steeper continuum rather than 'all or none' control. As this is a central tenant of the manuscript, it could perhaps be addressed through changes in the text to shore up the argument, differently emphasized through data presentation or through the addition of new data.

Thank you for this insight. We have modified the text to point out that even if the Ca2+ influx is not completely silenced, the non-linear nature of the release process would effectively amplify the functional silencing effect (page 11, line 24 – page 12, line 6).

Reviewer #1 (Recommendations for the authors):

While fascinating, the conclusions are most powerful when the data can be distilled to direct observation of single release sites. Yet throughout, the data are presented as all-or-none discriminations at regions of interest (ROI) that may or may not represent single release sites. Specific criticisms are listed below.

1. One obvious challenge is distinguishing between a very small response and a non-responder. The data, throughout the paper, are categorized (yes/no) and presented as a percentage. Thus, the crux of the manuscript relies upon thoroughly demonstrating that the discrimination between a responder and non-responder is not confounded. For example, in several figures, nice example traces are provided showing a large initial response that results in a failure upon lowering extracellular calcium. But, beyond these examples, there are no quantitative data demonstrating that this is, indeed, routinely the case. Could the authors bin their data into quartiles based on the amplitude of the initial GCamp signal at 2.0 external calcium. Then, for each quartile, calculate the fractional silencing that occurs. If the fraction of silencing is constant across quartiles, then the authors have a good argument that the method of detection (GCamp) is not contributing to the observed effects.

Please see the response to #1 from the Essential revisions.

2. Figure 6 uses GluSnFr to directly measure vesicle release, testing synapse silencing. Data are presented as the average of 15 single action potentials. Given that release is probabilistic, a silent synapse is a zero responder over some number of these trials, but we do not know how many. Two issues arise. First, individual successes could be averaged out to appear as a failure in the averaged trace. Second, a true release probability depends on the number of active zones that are present at a single R01. It seems that the true power of GluSnFr is not being deployed. Can the authors utilize spontaneous miniature amplitudes (or sucrose evoked miniatures) to achieve an estimate of single vesicle release event amplitude. Then, if sufficiently above the noise, define the release probability at high and low calcium.

Please see the response to #2 from the Essential revisions.

3. In general, data from single RO1 would be nice to see, as opposed to averages and percentages per culture dish.

In addition to the inclusion of representative traces for the different biosensors used, we believe that the cumulative frequency distributions (Figure 2 —figure supplement 1, see also response to #1 from the Essential revisions) provide the best overall representation of the distribution of ΔF values from individual ROIs.

4. It would be important, in at least one experimental paradigm, to see that release returns to baseline following restoration of external calcium.

Please see the response to #4 from the Essential revisions.

Reviewer #2 (Recommendations for the authors):

The authors illustrate two terminals in which the [Ca2+]e around 0.8 mM results in zero Ca2+ entry. They also show that at higher [Ca2+]e the terminal has detectable Ca2+ entry. In the following figure, they demonstrate that approximately 25% of the boutons are 'silent' even in 2 mM [Ca2+]e. What does this mean? Is there any evidence that they are functional boutons? This is inconsistent with the literature. In cortical pyramidal neuron axons, each AP results in detectable Ca2+ entry and the Ca2+ entry is 100% reliable from AP to AP. If there is no Ca2+ entry, it is always the consequence of the lack of an active zone (i.e. not functional bouton).

We thank the reviewer for this question. We presume the reviewer is referring to the article by Koester and Sakmann (2000), in which Ca2+ influx was measured in terminals following patching and loading of synthetic Ca2+ dyes. One important limitation of this study is that terminals were identified only morphologically as visible varicosities in the axon. This approach may have biased their measurements to particularly large nerve terminals that may not be representative. Moreover, a study performed by Frenguelli and Malinow (1996) in neocortical slices using a similar technique and post-hoc staining to verify that Ca2+ measurements were from boutons did observe failures of Ca2+ influx not attributable to failed AP propagation. Our results are consistent with these observations and previously published studies from the Ryan lab (Kim and Ryan, 2013).

Keeping the superior properties of the synthetic dyes in mind, all [Ca2+] measurements should be repeated with high affinity synthetic dyes.

Please see the response to #6 from the Essential revisions.

Reviewer #3 (Recommendations for the authors):

1) Membrane excitability – Two observations are relevant here: The qualitative preservation of the phenomenon when two types of voltage gated calcium channels are blocked separately, and the large heterogeneity of the % of silenced boutons among neurons at a given extracellular calcium concentrations, which is at least as great as the range of modulation of the % of silenced synapses by extracellular calcium concentrations at single neurons. One then wonders if the findings might be attributed to a) the fidelity of the field potential-based stimulation system, that is, the degree to which neurons track the stimuli trains; b) the heterogeneity of neurons in this regard, c) this fidelity at different extracellular calcium concentrations for different neurons, and d) the identity of presynaptic sites analyzed in one run (are they all part of the same axon?). Along these lines, there is an assumption that the field potential-based stimulation system is the sole driver of excitation in these networks, which is reasonable given that excitatory synaptic transmission is mostly blocked pharmacologically (by CNQX and APV). Inhibitory transmission, however, was not blocked and thus, there is no guarantee that the inhibitory input neurons receive and its modulation by extracellular calcium does affect the degree to which neurons fire precisely and reliably at 20 Hz at all conditions. If it could be shown, at least for a substantial subset of the data, that all terminals analyzed for a particular neuron are part of an unambiguously identified axon stretch, with no branches (potential conduction failure points) and still demonstrate the claimed heterogeneity, this potential confound would be less of an issue.

Please see the response to #5 from the Essential revisions.

2) It would seem that the x-axis intercept in Figure 1D (0.47 mM) predicts that at this concentration of extracellular calcium, pretty much all presynaptic terminals will be silent. Why not test this directly? This would nicely strengthen the findings.

We thank the reviewer for this suggestion and we have conducted additional experiments comparing Δ[Ca2+]i and the proportion of silencing in [Ca2+]e 0.4 mM and 1.2 mM (Figure 1 —figure supplement 1 and Figure 3 —figure supplement 1). These experiments demonstrate that the overwhelming majority of terminals (93 ± 2%) are silent and the mean Δ[Ca2+]i is nearly 0. Thus, these experiments support that the overwhelming majority of terminals have a non-zero threshold of [Ca2+]e for [Δa2+]i.

3) Page 4, line 23. For the silenced population it is indeed expected, because, by definition, they did not respond and were consequently classified as silenced. As this is circular, the inclusion of these data in Figure 3D (and 6E) is a bit confusing and should probably be removed.

As suggested, we have removed the silent bars from these graphs.

4) Figure 4 – end of legend – Data are fit as in D (should probably be E)

Thank you for identifying this typographical error. It has been corrected.

https://doi.org/10.7554/eLife.83530.sa2

Article and author information

Author details

  1. Daniel C Cook

    Department of Anesthesiology, Weill Cornell Medical College, New York, United States
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0714-4050

  2. Timothy A Ryan

    1. Department of Anesthesiology, Weill Cornell Medical College, New York, United States
    2. Department of Biochemistry, Weill Cornell Medical College, New York, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    taryan@med.cornell.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2533-9548

Funding

NINDS (NS036942)

  • Timothy A Ryan

National Institute of General Medical Sciences (GM148935)

  • Daniel C Cook

National Institute of Neurological Disorders and Stroke (NS117139)

  • Timothy A Ryan

Foundation for Anesthesia Education and Research (MRTG-02-15-2020-Cook)

  • Daniel C Cook

Burroughs Wellcome Weill Cornell Physician-Scientist Academy

  • Daniel C Cook

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank members of the Ryan lab for their valuable discussion of the manuscript. We give special thanks to Andrew Nelson for developing the custom Arduino board and its graphical user interface. We thank Kasper Podgorski for kindly providing iGluSnFR3 v857. This work was funded by NIH grants to TAR (NS036942 and NS117139), NIH grant to DCC (K08GM148935), a Mentored Research Training Grant (MRTG) from the Foundation for Anesthesia Education and Research (FAER) to DCC (MRTG-02-15-2020-Cook), and a grant from the Burroughs Wellcome Weill Cornell Physician-Scientist Academy to DCC.

Ethics

All experiments involving animals were performed in accordance with protocols approved by the Weill Cornell Medicine Institutional Animal Care and Use Committee (IACUC protocol 0601-450A).

Senior Editor

  1. John R Huguenard, Stanford University School of Medicine, United States

Reviewing Editor

  1. Graeme W Davis, University of California, San Francisco, United States

Reviewer

  1. Graeme W Davis, University of California, San Francisco, United States

Version history

  1. Preprint posted: September 17, 2022 (view preprint)
  2. Received: September 17, 2022
  3. Accepted: April 3, 2023
  4. Accepted Manuscript published: April 4, 2023 (version 1)
  5. Version of Record published: April 19, 2023 (version 2)

Copyright

© 2023, Cook and Ryan

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,096
    Page views
  • 225
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Daniel C Cook
  2. Timothy A Ryan
(2023)
GABABR silencing of nerve terminals
eLife 12:e83530.
https://doi.org/10.7554/eLife.83530

Categories and tags

Research organism

Further reading

    1. Neuroscience

    ACC neural ensemble dynamics are structured by strategy prevalence

    Mikhail Proskurin, Maxim Manakov, Alla Karpova
    Research Article Updated

    Medial frontal cortical areas are thought to play a critical role in the brain’s ability to flexibly deploy strategies that are effective in complex settings, yet the underlying circuit computations remain unclear. Here, by examining neural ensemble activity in male rats that sample different strategies in a self-guided search for latent task structure, we observe robust tracking during strategy execution of a summary statistic for that strategy in recent behavioral history by the anterior cingulate cortex (ACC), especially by an area homologous to primate area 32D. Using the simplest summary statistic – strategy prevalence in the last 20 choices – we find that its encoding in the ACC during strategy execution is wide-scale, independent of reward delivery, and persists through a substantial ensemble reorganization that accompanies changes in global context. We further demonstrate that the tracking of reward by the ACC ensemble is also strategy-specific, but that reward prevalence is insufficient to explain the observed activity modulation during strategy execution. Our findings argue that ACC ensemble dynamics is structured by a summary statistic of recent behavioral choices, raising the possibility that ACC plays a role in estimating – through statistical learning – which actions promote the occurrence of events in the environment.

    1. Neuroscience

    Multiple NTS neuron populations cumulatively suppress food intake

    Weiwei Qui, Chelsea R Hutch ... Darleen Sandoval
    Research Article

    Several discrete groups of feeding-regulated neurons in the nucleus of the solitary tract (nucleus tractus solitarius; NTS) suppress food intake, including avoidance-promoting neurons that express Cck (NTSCck cells) and distinct Lepr- and Calcr-expressing neurons (NTSLepr and NTSCalcr cells, respectively) that suppress food intake without promoting avoidance. To test potential synergies among these cell groups we manipulated multiple NTS cell populations simultaneously. We found that activating multiple sets of NTS neurons (e.g., NTSLepr plus NTSCalcr (NTSLC), or NTSLC plus NTSCck (NTSLCK)) suppressed feeding more robustly than activating single populations. While activating groups of cells that include NTSCck neurons promoted conditioned taste avoidance (CTA), NTSLC activation produced no CTA despite abrogating feeding. Thus, the ability to promote CTA formation represents a dominant effect but activating multiple non-aversive populations augments the suppression of food intake without provoking avoidance. Furthermore, silencing multiple NTS neuron groups augmented food intake and body weight to a greater extent than silencing single populations, consistent with the notion that each of these NTS neuron populations plays crucial and cumulative roles in the control of energy balance. We found that silencing NTSLCK neurons failed to blunt the weight-loss response to vertical sleeve gastrectomy (VSG) and that feeding activated many non-NTSLCK neurons, however, suggesting that as-yet undefined NTS cell types must make additional contributions to the restraint of feeding.

    1. Genetics and Genomics
    2. Neuroscience

    Opposing chemosensory functions of closely related gustatory receptors

    Ji-Eun Ahn, Hubert Amrein
    Research Article

    In the fruit fly Drosophila melanogaster, gustatory sensory neurons express taste receptors that are tuned to distinct groups of chemicals, thereby activating neural ensembles that elicit either feeding or avoidance behavior. Members of a family of ligand -gated receptor channels, the Gustatory receptors (Grs), play a central role in these behaviors. In general, closely related, evolutionarily conserved Gr proteins are co-expressed in the same type of taste neurons, tuned to chemically related compounds, and therefore triggering the same behavioral response. Here, we report that members of the Gr28 subfamily are expressed in largely non-overlapping sets of taste neurons in Drosophila larvae, detect chemicals of different valence, and trigger opposing feeding behaviors. We determined the intrinsic properties of Gr28 neurons by expressing the mammalian Vanilloid Receptor 1 (VR1), which is activated by capsaicin, a chemical to which wild-type Drosophila larvae do not respond. When VR1 is expressed in Gr28a neurons, larvae become attracted to capsaicin, consistent with reports showing that Gr28a itself encodes a receptor for nutritious RNA. In contrast, expression of VR1 in two pairs of Gr28b.c neurons triggers avoidance to capsaicin. Moreover, neuronal inactivation experiments show that the Gr28b.c neurons are necessary for avoidance of several bitter compounds. Lastly, behavioral experiments of Gr28 deficient larvae and live Ca2+ imaging studies of Gr28b.c neurons revealed that denatonium benzoate, a synthetic bitter compound that shares structural similarities with natural bitter chemicals, is a ligand for a receptor complex containing a Gr28b.c or Gr28b.a subunit. Thus, the Gr28 proteins, which have been evolutionarily conserved over 260 million years in insects, represent the first taste receptor subfamily in which specific members mediate behavior with opposite valence.